Professional Documents
Culture Documents
T R A N S L AT I O N A L G E N E T I C S
CpG islands
CpG-rich regions of DNA that
are often associated with the
transcription start sites of
genes and that are also found
in gene bodies and intergenic
regions.
Over the past 25 years, an increasing amount of evidence has supported a role for epigenetic processes in
cell biology and tissue physiology. The potential role of
these epigenetic processes in human disease is exemplified by aberrant DNA methylation in cancer, both at
individual genes and on a genome-wide scale. Initially,
studies focused on candidate gene approaches to identify alterations occurring in distinct diseases; however,
new powerful technologies, such as comprehensive DNA
methylation microarrays1,2 and genome-wide bisulphite
sequencing3, have recently emerged (BOX1). These have
reinforced the notion of epigenetic disruption at least as
a signature of human diseases4.
An indication that alterations in epigenetic marks
could be used as biomarkers specifically DNA
methylation provided by the use of the hypermethylation events at the genes O6methylguanine-DNA
methyltransferase (MGMT)5 in glioma and glutathione
Stransferase pi 1 (GSTP1)6 in prostate cancer; these
hypermethylation events have been shown to be effective for determining an appropriate treatment choice in
glioma and in the diagnosis of prostate cancer, respectively. The detection of epigenetic alterations is therefore
a promising tool for the diagnosis and prognosis of disease and for the prediction of drug response.
This Review focuses on epigenetic profiling at the
level of DNA methylation7,8. There is an emphasis on
neoplasia, as this is the area in which using these epigenetic markers is best characterized, but many of these
concepts are applicable to other disorders, such as
neurological, metabolic and cardiovascular disorders
and autoimmune diseases. DNA methylation marking
repressed genes is a frequent event in development, differentiation and tissue homeostasis. Cancer cells present
a global loss of DNA methylation centred in hypomethylated blocks and regional hypermethylation, especially
in CpG-rich contexts8,9. For cancer, recent evidence has
given an indication of the altered DNA methylation
events beyond classical promoter hypermethylation of
CpG islands of tumour suppressor genes 10. The early
signs are that this knowledge will be useful in practical
clinical situations, such as in the diagnosis of cancers of
unknown primary origin, the screening for malignancies in high-risk populations or biomarker-based selection of the patients that should receive treatment with
epigenetic drugs. Furthermore, there is a growing list of
mutated chromatin remodeller genes that contribute to
the tumorigenesis process and that may be targeted with
drugs in the future for cancer treatment11 (BOX2). In the
future, studying chromatin structure may provide useful
insights for diagnosis and prognosis (but this requires
further research and is not discussed here).
This Review provides an insight into epigenetic
alterations implicated in diagnostics, prognostics and
prediction to drug response and their current use in
clinical settings. First, an overview of what makes a suitable DNA-methylation-based biomarker is presented,
including a consideration of the non-invasive tissues that
may be used to profile these. Then the applications of
DNA profiling in disease diagnosis followed by disease
prognosis and prediction to drug response are presented,
including a consideration of relevant examples at every
stage. Finally, future directions for the development of
relevant biomarkers and their applications are discussed.
REVIEWS
removed by a range of enzymes that may be targeted for
disease treatment (BOX2); this is reviewed elsewhere4,11,12.
Given their dynamic nature, epigenetic disease biomarker genes have to be determined by considering
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
Box 2 | Epigenetic modifiers: potential drug targets
Epigenetic drugs in clinical use and trials
Tumours that display global epigenomic alterations might also benefit from therapies that restore these global patterns.
Some of the modifying enzymes that put in place the mark are mutated in disease and can be targeted by specific
molecules, representing the first examples of the development of personalized medicine therapies that combine
genomic and epigenomic knowledge. Considering the dynamic and reversible nature of epigenetic marks, they
represent an attractive target for personalized therapy. However, treatment with currently approved epigenetic drugs
(DNA methyl transferase (DNMT) and histone deacetylase (HDAC) inhibitors) is rather broad, and yet to be defined
epigenetic cancer subtypes might respond differently. Here, high-resolution profiling might improve the treatment
efficiency by guiding the therapy decision. Currently approved epigenetic treatments are described below, although
readers are referred elsewhere for further detailed discussion of these11,12,109.
To date, four epigenetic drugs have been approved by the US Food and Drug Administration (FDA): two DNMT
inhibitors and two HDAC inhibitors. 5azacytidine (5azaCR; manufactured by Celgene as Vidaza) was approved in
2004 specifically to inhibit DNA methylation, and two years later its variant 5aza2deoxycytidine was approved
(5aza-CdR; manufactured by Eisai as Dacogen). Both were approved for the treatment of higher-risk myelodysplastic
syndrome. In addition, S110 (REF. 110), which is a dinucleotide containing 5aza-CdR and is suggested to have enhanced
stability and efficiency, complements the list of DNMT inhibitors that are in use or that are being tested in clinical trials.
Vorinostat (Merck) an FDA-approved HDAC inhibitor for the treatment of cutaneous T cell lymphoma (CTCL)111,112
was also confirmed to induce complete response or haematological improvement in AML patients113. Besides
Vorinostat, Romidepsin (Celgene), another HDAC inhibitor, revealed remarkable efficacies in the treatment of
cutaneous T cell lymphoma (CTCL) patients114. Two additional HDAC inhibitors Panobinostat (Novartis) and CI994
(Pfizer) are currently being tested in clinical phaseIII trials for the treatment of lymphomas and non-small-cell lung
cancer (NSCLC), respectively.
A new generation of targeted epigenetic drugs
Owing to an increased resolution of genomic and epigenetic profiling methods, targeted epigenetic treatments, such as
the specific inhibition of enzymes by small molecules, are being developed, adding to the drug arsenal to improve drug
performance further while reducing toxicity to healthy tissue. Specific treatments that target epigenetic modifiers are
the subject of current investigation, and the most recent examples are briefly considered in the following section.
Bromodomain and extra-terminal domains (BETs) are adaptor molecules that are involved in chromatin-dependent
signal transduction, resulting in transcriptional initiation and elongation. The bromodomain family member BRD4 has
been shown to be involved in gene fusions in squamous carcinomas and leukaemia. The small molecule JQ1 has been
established as a potent inhibitor of BRD4, and JQ1 treatment showed strong antiproliferative effects in cell line and
xenograft models that harboured BRD4 fusion115. Chromosomal translocation involving mixed lineage leukaemia (MLL),
which is a histone lysine methyl transferase, is an initiator of aggressive forms of leukaemia with extremely poor
prognosis. MLL fusion partners, especially members of the super elongation complex (SEC), such as AF4, AF6, AF9, AF10
and ENL, have been identified as interaction partners with BET proteins. Using the small-molecule inhibitor of the BET
family I-BET151 causes displacement of BRD3BRD4 and SEC from chromatin, followed by inhibition of coactivator
function and subsequent repression of key oncogenes, such as MYC, B cell CLL/lymphoma 2 (BCL2) and
cyclin-dependent kinase 6 (CDK6)116.
Fusion products of translocated MLL genes also initiate aberrant recruitment of DOT1L, which is a histone
methyltransferase that is involved in histone H3 lysine 79 (H3K79) methylation, to MLL target promoters, resulting in
mislocalized enzymatic activity117,118. In patients harbouring the translocated MLL genes, altered DOT1L activity
functions as a downstream event associated with the activation of leukaemogenic genes such as homeobox A9 (HOXA9)
and Meis homeobox 1 (MEIS1). Recently, Daigle and co-workers119 developed the small molecule, EPZ004777, which
specifically inhibits the H3K79 methylation activity of DOT1L. EPZ004777 treatment results in selective killing of
leukaemia cells that harbour the MLL translocation in mouse xenograft models.
Personalized medicine
Therapeutic decisions based
on genetic and epigenetic
information of individual
patients
fluoropyrimidines in colorectal cancer patients22. In addition, loss of imprinting at the insulin-like growth factor 2
(IGF2) gene locus is a frequent event in cancer and has
potential use for colon cancer diagnosis23,24.
Early tumour detection is one of the major factors in
successful cancer treatment. Using non-invasive methods, such as the analysis of blood, stool, saliva or urine
samples, has clear advantages over invasive techniques,
such as biopsies. Tumour DNA can enter biological
fluids in different ways: tumour cells can be directly
released from the tissue of origin; necrotic tumour cells
are engulfed by macrophages by phagocytosis and subsequently the tumour cell enters the blood or urethra; or
free tumour DNA directly reaches the fluids by cell lysis.
Analysis of epigenetic alterations can then be carried out
by these non-invasive methods25.
REVIEWS
Table 1 | Hypermethylated genes in cancer and their associated tissue types
Gene name
Gene function
Cancer type
APC
WNT signalling
AR
Prostate
BMAL1
AHR signalling
Leukaemia, lymphoma
BRCA1
Breast, ovarian
CDH1
Cellcell adhesion
Breast, prostate
CDH11
Cellcell adhesion
CDH13
Cellcell adhesion
CDKN2A
CDKN2B
Leukaemia
DAPK1
EMP3
Signal transduction
Glioma
ESR1
Breast
GSTP1
Detoxification
IGFBP3
Signal transduction
LGALS3
Prostate
MASPIN
Peptidase inhibitor
Pancreas
MGMT
DNA repair
miR148a
Metastasis suppression
Metastasis
Metastasis suppression
Metastasis
miR9
Metastasis suppression
Metastasis
miR200s
Epithelialmesenchymal transition
MLH1
DNA repair
NORE1A
NSD1
Nuclear receptor
Glioma, neuroblastoma
PYCARD
Apoptosis
RARB
Prostate-specific antigen
(PSA). A serine protease of the
kallikrein gene family that is
secreted into seminal fluid by
prostatic epithelial cells and is
found in the serum. As it is
almost exclusively a product of
prostate cells, measurement in
blood has proved to be useful
as a tumour marker for
diagnosis of prostate cancer
and monitoring the
effectiveness of treatment.
established GSTP1 as a promising biomarker of prostate cancer with a sensitivity of 82% and a specificity of
95%33. As GSTP1 hypermethylation was detected in 69%
of proliferative inflammatory atrophy lesions (precursor lesions for the development of prostate cancer and/
or high-grade prostatic intraepithelial neoplasia) in an
analysis of 27 patient samples, it is thought to be an early
event in tumorigenesis, underlining its importance for
diagnosis34.
Initially, Goessl and colleagues35 confirmed the presence of GSTP1 hypermethylation in urine (36%; 4 out
of 11 subjects), serum (72%; 23 out of 32 subjects) and
ejaculate (50%; 4 out of 8 subjects) from patients with
prostate cancer. Later, the GSTP1 hypermethylation was
confirmed in independent studies that detected 32%,
42% and 13% of subjects displaying hypermethylation
by analysing 62, 168 and 83 serum samples from patients
with prostate cancer, respectively3638. Here, a combination with testing of the serum-based biomarker prostatespecific antigen (PSA) could increase the sensitivity of
both methods in the diagnosis of prostate cancer38.
GSTP1 hypermethylation in urine was further confirmed with sensitivities of 27% and 50% in studies
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
Table 1 (cont.) | Hypermethylated genes in cancer and their associated tissue types
Gene name
Gene function
Cancer type
RASSF1A
RBP1
RIZ1
Histone methyltransferase
S100P
Pancreatic
SEPT9
Colon
SFRP1
WNT signalling
SFRP2
WNT signalling
Colon
SNCG
Breast, ovarian
SOCS1
Cytokine signalling
Liver
TFPI2
Colon
THBS1
Glioma
TIG1
Prostate
TIMP2
Metallopeptidase inhibitor
Prostate
TP73
Stress response
Lymphoma
TSHR
TSH receptor
Thyroid
VHL
Hypoxia response
Kidney
WIF1
WNT signalling
Colon
WRN
DNA helicase
Colon
For further details, please see REF. 16. APC, adenomatous polyposis coli; AR, androgen receptor; BMAL1, brain and muscle
ARNT-like 1; BRCA1, breast cancer 1, early onset; CDH1, cadherin 1, type1, Ecadherin; CDH11, cadherin 11, type2,
OBcadherin; CDH13, cadherin 13, Hcadherin (heart); CDKN2A, cyclin-dependent kinase inhibitor 2A; DAPK1,
death-associated protein kinase 1; EMP3, epithelial membrane protein 3; ESR1, oestrogen receptor 1; GSTP1, glutathione
Stransferase pi 1; IGFBP3, insulin-like growth factor binding protein 3; LGALS3, lectin, galactoside-binding, soluble, 3;
MASPIN, also known as SERPINB5; MGMT, O6methylguanine-DNA methyltransferase; MLH1, mutL homologue 1, colon cancer,
nonpolyposis type2; NORE1A, also known as RASSF5; NSD1, nuclear receptor binding SET domain protein 1; PYCARD, PYD
and CARD domain containing; RARB, retinoic acid receptor beta; RASSF1A, RAS association (RalGDS/AF6) domain family
member 1; RBP1, retinol-binding protein 1; RIZ1, also known as PRDM2; S100P, S100 calcium binding protein P; SEPT9,
septin9; SFRP1, secreted frizzled-related protein 1; SNCG, synuclein, gamma (breast cancer-specific protein 1); SOCS1,
suppressor of cytokine signalling 1; TFPI2, tissue factor pathway inhibitor 2; THBS1, thrombospondin 1; TIG1, retinoic acid
receptor responder (tazarotene-induced) 1; TIMP2, TIMP metallopeptidase inhibitor 2; TP73, tumour protein p73; TSHR, TSH
receptor; VHL, von Hippel-Lindau tumour suppressor, E3 ubiquitin protein ligase; WIF1, WNT inhibitory factor 1; WRN, Werner
syndrome, RecQ helicase-like.
that analysed 22 and 36 prostate cancer patient samples, respectively39,40, and diagnosis was technically
improved by prostatic massage41 and the combination
of analysis of hypermethylation at different markers,
such as cyclin-dependent kinase inhibitor 2A (CDKN2A,
which encodes the protein p16), ARF and MGMT42,43.
Meta-analysis of 22 studies that analysed the value of
GSTP1 as a biomarker for prostate cancer in biological
fluids, including plasma, serum and urine, revealed a
high specificity of 89% but showed modest sensitivity
(52%)44. Analysing GSTP1 hypermethylation in biological fluid presents a promising alternative to conventional
methods of detection of prostate cancer. In particular,
the combination of GSTP1 with additional biomarker
genes and PSA testing could highly improve the sensitivity of diagnosis.
In addition to GSTP1, other genes exhibit a high incidence of hypermethylation in prostate cancer and can be
detected in biological fluids37,38,42. In particular, the following genes were confirmed to gain DNA methylation
to similar extents as those detected for GSTP1 (promoter
hypermethylation levels given in brackets): the tumour
suppressor adenomatous polyposis coli (APC; 8390%),
REVIEWS
Healthy tissue
Cancer
MGMT
CpG island
hypermethylation
MGMT
MGMT
Transcription
Transcription
MGMT
Translation
DNA
repair
Translation
DNA
damage
Prediction
of drug
response
Brain biopsy
GSTP1
CpG island
CpG island
hypermethylation
Urine system biopsy
GSTP1
GSTP1
Transcription
Transcription
GSTP1
Translation
Tumour
detection
Translation
6
Figure 1 | GSTP1 and MGMT: case examples for epigenetic profiling in diagnosis and prognosis.
methylguanineNatureOReviews
| Genetics
DNA methyltransferase (MGMT) protects normal cells against transition mutations by removing alkyl groups (red
squares), which have been introduced by carcinogens such as nitrosamides, from guanine bases. Hypermethylation of
MGMT is correlated with blocked gene expression, leading to the accumulation of alkylated guanines and subsequent
DNA damage. However, hypermethylated MGMT in glioblastomas is a predictor of good response to DNA-alkylating
drugs, such as carmustine and temozolomide, as an active MGMT is able to repair modified bases introduced by the
therapy. Hypermethylation of MGMT is used as biomarker for drug response in glioblastoma biopsies but can also be
detected in serum. Glutathione Stransferase pi1 (GSTP1) has been established as biomarker for prostate cancer
diagnosis and prognosis. It is involved in cellular detoxification of xenobiotics and carcinogens (red circles). In a normal
context, GSTP1 actively removes damaging agents from the cell. Hypermethylation of GSTP1 in tumours is correlated
with gene repression and subsequent cellular accumulation of xenobiotics and carcinogens, thereby favouring
tumorigenesis. Hypermethylation of GSTP1 is used as a biomarker for malignant cells in prostate biopsies but can also be
detected in urine and serum.
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
blood-based CRC diagnosis was later confirmed in an
independent study that screened 50 CRC patients; cancer
was detected with a sensitivity of 90% and a specificity
of 88% using the methylation at this gene as a marker48.
Using stool as a non-invasive screening material for 84
CRC patients revealed hypermethylation of TFPI2 as
a CRC marker. The presence of hypermethylation at
TFPI2 could detect CRCs in samples from patients with
cancer with a sensitivity of 7689%49.
Gene promoter hypermethylation has also been established as a biomarker candidate for patients with glioblastoma50 and patients with non-small-cell lung cancer
(NSCLC)51. Glioblastomas could be detected with a sensitivity of 95% and a specificity of 60% by analysing MGMT
promoter methylation in the serum of 37 patients50. Here
the patient serum DNA hypermethylation levels of
MGMT were highly correlated with the primary tissue
levels and were able to predict a lack of cancer progression and overall survival of the patient. The detection of
MGMT hypermethylation in combination with CDKN2A
and GSTP1 hypermethylation allowed for the detection of
NSCLC with 73% sensitivity but analysed a small cohort
of 15 samples51. Using saliva as a source of biomarkers
in 30 patients suffering from head and neck cancer identified the combination of MGMT, CDKN2A and deathassociated protein kinase 1 (DAPK1) hypermethylation
as powerful biomarkers, detecting 65% of tumours52. In
followup studies, the number of potential biomarkers was
complemented by homeobox A9 (HOXA9) and nidogen 2
(NID2) hypermethylation, showing detection sensitivities
of 75% and 87%, respectively, and specificities of 53% and
21%, respectively, giving a higher sensitivity when tested
in combination53.
The detection of DNA methylation alteration in
biological fluids represents a promising tool for noninvasive cancer detection. However, many candidates
present a low specificity, which results in a false-positive
diagnosis. Reference data sets allow for the initial exclusion of genes that show high inter-individual variation
or high levels of basic methylation, making it difficult
to detect a gain in cancer patients. Improving candidate
selection by using reference DNA methylomes could
accelerate biomarker discovery and specificity by minimizing the number of unsuitable markers for prognostic DNA methylation approaches selected for further
investigation.
Genome-wide epigenetic technologies might also
facilitate the identification of cancers of unknown primary origin. Following the detection of metastatic diseases, it is crucial for the success of the treatment and, in
turn, for the prognosis to identify the primary origin of
the tumour. In a comprehensive study, analysing DNA
methylation of 1,505 CpG sites in 1,628 human samples, Fernandez etal.20 determined tissue- and cancerspecific fingerprints and were able to assign a tissue of
origin for 69% of the cancers of unknown primary origin analysed (FIG.2).
non-cancer context54 (TABLE2). Twin studies are a powerful approach as, despite their identical genetic background, twins have differences in DNA methylation55.
Three studies of monozygotic twins who were discordant
for disease development revealed surprising epigenetic
differences. Monozygotic twins who were discordant for
systemic lupus erythematosus (SLE), which is an autoimmune disorder, revealed differences in DNA methylation
of blood cells in a number of genes that are involved in
immune function18. A similar study setup using blood
from monozygotic twins who were discordant for childhood-onset type1 diabetes identified a disease-specific
DNA methylation signature, which was subsequently validated in a cohort outside the twin context56. Differential
DNA methylation markers may also help in the diagnosis
of diseases in which only a few genetic mutations have
been linked to the development of the disease, such as
Alzheimers disease57. Comparing DNA methylation levels
in neurons of monozygotic twins who were discordant for
the disease unravelled global loss of methylation levels in
patients compared with their healthy relatives, suggesting
a role for an altered epigenetic landscape in Alzheimers
disease58. In this regard, the incidence of neurodegenerative disorders increase with advanced age, and recent data
obtained by whole-genome bisulphite sequencing show
important DNA methylation changes in nonagenarians
and centenarians59. In all of these studies, how these
changes might contribute to future diagnosis has to be
clarified in followup studies.
DNA methylation in the diagnosis of non-cancer diseases. It is of note that DNA methylation profiling is
also able to identify differentially methylated genes in a
REVIEWS
a Disease diagnosis
Malignant tissue
c Therapy response
Poor response
Benign tissue
Good response
Patient sample
b Disease prognosis
Short-term survival
Long-term survival
Biopsy
Saliva
Blood
Stool
Urine
Cancers of
unknown
primary origin
Liver
Skin
Brain
Figure 2 | High-resolution screening of patient samples in disease diagnosis, prognosis, prediction of drug
Nature Reviews | Genetics
response and tumour-type identification. Future cancer diagnosis, prognosis and therapy will benefit from
epigenetic high-resolution screening technologies. Profiling and subsequent classification can be carried out on
primary tissue or biological fluids, such as saliva, blood, stool and urine. Profiles in panels a, b and c are representative
profiles that give an indication of what may be possible with genome-wide profiling in the future. a | The assessment of
reference DNA methylome data sets for normal tissues and cancer types enables us to classify subtypes on the basis
of their DNA methylation signatures. These newly defined methylation-profile-based subtypes are suitable for the
differentiation between malignant and benign tissue. b | Methylation signatures can also be used for the differentiation
between patients with short-term or long-term survival. c | Different methylation profiles can be used to predict the
different response to drug treatment. d | Furthermore, the unique and tissue-specific DNA methylation profiles of
normal cells can help in the identification of the tissue of origin of cancers with unknown primary origin. Panel d is
modified, with permission, from REF. 20 (2011) CSH Press.
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
Table 2 | Epigenetic alterations in non-cancer diseases
Epigenetic
modification
Disease
Alteration
DNA
methylation
Alzheimers disease
Hypermethylation (NEP)
Angelman syndrome
Atherosclerosis
Aberrant methylation
ATRX syndrome
Diabetes type1
Aberrant methylation
Friedrichs ataxia
Hypermethylation (FXN)
Multiple sclerosis
Hypomethylation (PADI2)
PraderWilli syndrome
Retts syndrome
Mutation (MECP2)
Rheumatoid arthritis
Alzheimers disease
Aberrant phosphorylation
Atherosclerosis
Diabetes type1
Fragile X syndrome
Friedrichs ataxia
Huntingtons disease
Parkinsons disease
Histone
modification
Nucleosome
positioning
RubensteinTaybi syndrome
Mispositioning
Diabetes type1
Rheumatoid arthritis
For further details, see REF.53. AIM2, absent in melanoma 2; CD154, also known as CD40LG; CLTA4, clathrin, light chain A;
CREBBP, CREB-binding protein; DNMTB, DNA (cytosine5)-methyltransferase 3 beta; DR3, also known as TNFRSF25; EP300,
E1A-binding protein p300; FMT1, fragile X mental retardation 1; FXN, frataxin; IL-6, interleukin 6; L1, also known as FAM49B;
MECP2, methyl-CpG-binding protein 2 (Retts syndrome); NEP, also known as MME; NFB1, nuclear factor- B p105 subunit;
PADI2, peptidyl arginine deiminase, typeII; PRF1, perforin 1 (pore-forming protein).
DNA methylation profiling for metastasis identification. Crucial to prognosis is the detection and treatment of tumour cells that have disseminated from the
point of origin. DNA methylation alterations in metastatic cancer cells were first reported at Ecadherin for
breast cancer61 and extended to an additional member
of the cadherin family (cadherin 11) in head and neck
cancer and melanoma samples62. Another study identified microRNA gene hypermethylation profiles that are
characteristic of human metastasis63.
DNA methylomes in prognosis. Cancer cells display a
global loss of DNA methylation; however, CpG-rich
gene promoters are frequently targets of DNA hypermethylation associated with tumour-suppressor gene
silencing and cancer formation and progression. Thus,
comprehensive DNA methylation profiling presents a
potent tool to profile regions with possible prognosis
predictive potential (BOX1; FIG.2). As high-resolution
data serve not only as a source for cancer-subtype-specific profiles but also for the identification of powerful
single epigenetic biomarker genes and gene combinations at various stages of disease, many types of clinical
applications can benefit from this advanced screening
strategy.
The prognostic potential of DNA methylation profiling was initially reported for childhood acute lymphoblastic leukaemia samples64. Determining the DNA
methylation level of 401 patients and 1,320CpG sites
(using GoldenGate from Illumina) allowed the classification of the patients into acute lymphoblastic leukaemia subtypes. Furthermore, DNA methylation profiling
allowed the stratification of patients with high hyperdiploidy and translocation t(12;21) into two subgroups
with different probabilities of relapse. In particular, DNA
methylation profiling of 40 genes permitted the identification of pathologically distinct disease subtypes and the
identification of 20 genes at which methylation predicted
REVIEWS
disease relapse64. Here, DNA methylation profiling
therefore resulted in subtype-specific and predictive
signatures with potential use as future prognostic biomarkers for childhoodacute lymphoblastic leukaemia.
A more comprehensive study that analysed ~14,000
genes (using Infinium 27K from Illumina) classified
154 normal and CRC samples according to their DNA
methylation profile65. Strikingly, this study identified
subgroups correlating with prognostic markers such as
mutL homologue 1, colon cancer, nonpolyposis type2
(MLH1) methylation and v-raf murine sarcoma viral
oncogene B1 (BRAF), TP53 or v-Ki-ras2 Kirsten rat
sarcoma viral oncogene (KRAS) mutation, suggesting
that DNA methylation analysis might contribute to
CRC classification, and a combination of epigenetic and
genetic analysis might further improve the accuracy of
prognosis.
Another study, focusing on the analysis of developmental regulated Polycomb target genes (PCTGs)
and loci heavily methylated in embryonic stem
cells (MESCs) in 1,475 samples using the Infinium
27k technology, identified both regions to be differentially methylated in various cancer types 66.
Hypermethylation of PCTGs was observed as an early
event in tumorigenesis even detectable before neoplastic changes occur, hence suggesting PCTGs as potential
diagnostic markers. In contrast, hypomethylation of
MESCs progressed concomitantly with tumour stage
and correlated with poor prognosis in breast, ovarian,
cervical and endometrial cancers.
The DNA methylation profiling of 248 breast cancer
samples identified a previously unrecognized subtype
associated with T lymphocyte infiltration. Importantly,
profiling immune genes determined a prognostic value
of the profile. In particular, the promoter hypermethylation of lymphocyte transmembrane adaptor 1(LAX1)
and CD3D significantly correlated with survival in certain breast cancer subtypes67.
Aiming to identify a DNA methylation signature
related to breast cancer metastasis, 39 primary breast
tumour specimens were analysed using the Infinium
27K platform68. Strikingly, DNA methylation profiling
determined two distinct groups with a substantially different association to metastasis-free survival. Assessing
the DNA methylation of only three genes of the methylation signature rho guanine nucleotide exchange
factor (ARHGEF7), ALX homeobox 4 (ALX4) and
RAS-protein-specific guanine nucleotide-releasing factor 2 (RASGRF2) allowed the prediction of metastasis-free cancer and overall survival. In particular, the
authors further suggest a common signature of alteration
in DNA methylation between gliomas, colon and breast
cancers.
Analysing 272 glioblastoma samples with the
GoldenGate and Infinium 27K technology, a subgroup
of cancers (8.8%) was identified, showing high levels of
CpG island methylation69. Furthermore, the authors
detected a high correlation between hypermethylation
and mutation of isocitrate dehydrogenase 1 (IDH1).
Mutant IDH1 protein converts ketoglutarate into
the oncometabolite (R)-2hydroxyglutarate (2HG),
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
Table 3 | Hypermethylated genes predict drug sensitivity
Gene name
Gene function
Therapeutical consequences
Tumour type
application
Example
refs
ABCB1
Protein transport
Sensitivity to doxorubicin
Breast
120
APAF1
Apoptotic activator
Resistance to adriamycin
Melanoma
121
BRCA1
Breast, ovary
CDK10
Resistance to anti-oestrogens
Breast
122
CHFR
Ovary, endometrium,
stomach
123
ESR1
ER signalling
Resistance to anti-oestrogens
Breast
124
FANCF
Sensitivity to cisplatin
Ovarian
125
GSTP1
Detoxification
Sensitivity to doxorubicin
126
IGFBP3
Signal transduction
Resistance to cisplatin
Lung
127
LINE1
Repetitive element
Resistance to fluoropyrimidines
Colon
22
MGMT
DNA repair
Sensitivity to temozolomide,
BCNU, ACNU, procarbazine
MLH1
DNA repair
Resistance to cisplatin
Colon, stomach,
endometrium, ovary
128
MT1E
Antioxidant
Sensitivity to cisplatin
Melanoma
129
PITX2
Transcriptional regulator
Resistence to tamoxifen
Breast
130
PLK2
Cell division
Ovary
131
PRKCDBP
Signal transduction
Resistance to TNF
Colon
132
SFN
Signal transduction
Lung
133
SLC19A1
Folate transporter
Resistance to methotrexate
Lymphomas
134
SULF2
Heparin signalling
Sensitivity to camptothecin
Lung
135
TFAP2E
Transcriptional regulator
Sensitivity to fluorouracil
Colon
136
TGM2
Apoptosis
137
TP73
Stress response
Sensitivity to cisplatin
Renal, melanoma
138
WRN
DNA helicases
Sensitivity to irinotecan
Colon
139
ERCC5
DNA repair
Resistance to nemorubicin
Ovary
140
81
ABCB1, ATP-binding cassette, subfamily B (MDR/TAP), member 1; ACNU, (1-4-amino-2-methyl-5-pyrimidinyl)-methyl-3-(2chloroethyl)-3-nitrosourea hydrochloride; APAF1, apoptotic activator; BCNU, bis-chloroethylnitrosourea; BRCA1, breast cancer 1,
early onset; CDK10, cyclin-dependent kinase 10; CHFR, checkpoint with forkhead and ring finger domains, E3 ubiquitin protein
ligase; ERCC5, excision repair cross-complementing rodent repair deficiency, complementation group 5; ESR1, oestrogen
receptor1; FANCF, Fanconi anaemia, complementation group F; GSTP1, glutathione Stransferase pi 1; IGFBP3, insulin-like growth
factor binding protein 3; MGMT, O6methylguanine-DNA methyltransferase; MLH1, mutL homologue 1, colon cancer, nonpolyposis
type2; MT1E, metallothionein 1E; PITX2, paired-like homeodomain 2; PLK2, polo-like kinase 2; PRKCDBP, protein kinase C, delta
binding protein; SFN, stratifin; SLC19A1, solute carrier family 19; SULF2, sulphatase 2; TFAP2E, transcription factor AP2 epsilon
(activating enhancer binding protein 2 epsilon); TGM2, transglutaminase 2 (C polypeptide, protein-glutamine-gamma-glutamyltransferase); TP73, tumour protein p73; TNF, tumour necrosis factor-; WRN, Werner syndrome, RecQ helicase-like.
carmustine. Subsequently, MGMT promoter hypermethylation has been associated with treatment
outcome of additional alkylating agents, such as
temozolomide71, and other chemotherapeutic drugs,
including cilengitide73 and cyclophosphamide74, as well
as radiation75.
A further example of a DNA methylation event
being indicative of successful chemotherapeutic intervention is that of hypermethylation at breast cancer 1,
early onset (BRCA1). Mutations in BRCA1, which is
a gene involved in DNA repair, have frequently been
observed in patients with inherited breast tumours76.
REVIEWS
Furthermore, hypermethylation and associated transcriptional repression of BRCA1 have also been observed
in sporadic breast and ovarian cancers77. Cancers that
are marked by altered BRCA1 expression can be treated
with two distinct treatment strategies: conventional
chemotherapy or specific inhibition of DNA repair proteins. Triple-negative breast tumours (a breast cancer
subtype that lacks the expression of v-erb-b2 erythroblastic leukaemia viral oncogene homologue 2 (ERBB2;
also known as HER2/neu) as well as the oestrogen and
progesterone receptors), showing frequent hypermethylation of BRCA1 (REF. 78), were associated with an
improved response to cytotoxic drugs such as cisplatin79.
In addition, BRCA1 gene mutation carriers benefit from
a targeted therapy using inhibitors of PARP, which area
proteins that are involved in base excision repair (that is,
a DNA repair mechanism that is distinct from homologous recombination, in which BRCA1 is involved) 80.
Two PhaseII trials are currently ongoing that aim to
establish the efficiency of PARP inhibitors in the treatment of BRCA1-mutated breast and ovarian cancers.
Consequently, it is tempting to speculate that PARP
inhibitors might also successfully target sporadic cases
that are positive for promoter BRCA1 methylation81.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved
REVIEWS
27. Berdasco, M. & Esteller, M. Aberrant epigenetic
landscape in cancer: how cellular identity goes awry.
Dev. Cell 19, 698711 (2010).
28. Lee, W.H. etal. Cytidine methylation of regulatory
sequences near the piclass glutathione Stransferase
gene accompanies human prostatic carcinogenesis.
Proc. Natl Acad. Sci. USA 91, 1173311737 (1994).
29. Yegnasubramanian, S. etal. Hypermethylation of CpG
islands in primary and metastatic human prostate
cancer. Cancer Res. 64, 19751986 (2004).
30. Jernimo, C. etal. Quantitation of GSTP1 methylation
in non-neoplastic prostatic tissue and organ-confined
prostate adenocarcinoma. J.Natl Cancer Inst. 93,
17471752 (2001).
31. Bastian, P.J. etal. Diagnostic and prognostic
information in prostate cancer with the help of a small
set of hypermethylated gene loci. Clin. Cancer Res. 11,
40974106 (2005).
32. Padar, A. etal. Inactivation of cyclin D2 gene in
prostate cancers by aberrant promoter methylation.
Clin. Cancer Res. 9, 47304734 (2003).
33. Van Neste, L. etal. The epigenetic promise for
prostate cancer diagnosis. Prostate 72, 12481261
(2011).
34. Nakayama, M. etal. Hypermethylation of the human
glutathione Stransferase gene (GSTP1) CpG island is
present in a subset of proliferative inflammatory
atrophy lesions but not in normal or hyperplastic
epithelium of the prostate: a detailed study using
laser-capture microdissection. Am. J.Pathol. 163,
923933 (2003).
35. Goessl, C. etal. Fluorescent methylation-specific
polymerase chain reaction for DNA-based detection of
prostate cancer in bodily fluids. Cancer Res. 60,
59415945 (2000).
36. Reibenwein, J. etal. Promoter hypermethylation of
GSTP1, AR, and 1433 in serum of prostate cancer
patients and its clinical relevance. Prostate 67,
427432 (2007).
37. Ellinger, J. etal. CpG island hypermethylation in
cell-free serum DNA identifies patients with localized
prostate cancer. Prostate 68, 4249 (2008).
38. Sunami, E. etal. Multimarker circulating DNA assay
for assessing blood of prostate cancer patients.
Clin. Chem. 55, 559567 (2009).
39. Gonzalgo, M.L., Pavlovich, C.P., Lee, S.M. &
Nelson, W.G. Prostate cancer detection by GSTP1
methylation analysis of postbiopsy urine specimens.
Clin. Cancer Res. 9, 26732677 (2003).
40. Cairns, P. etal. Molecular detection of prostate cancer
in urine by GSTP1 hypermethylation. Clin. Cancer Res.
7, 27272730 (2001).
41. Goessl, C. etal. DNA-based detection of prostate
cancer in urine after prostatic massage. Urology 58,
335338 (2001).
42. Hoque, M.O. etal. Quantitative methylation-specific
polymerase chain reaction gene patterns in urine
sediment distinguish prostate cancer patients from
control subjects. J.Clin. Oncol. 23, 65696575
(2005).
43. Rogers, C.G. etal. High concordance of gene
methylation in post-digital rectal examination and
post-biopsy urine samples for prostate cancer
detection. J.Urol. 176, 22802284 (2006).
44. Wu, T. etal. Measurement of GSTP1 promoter
methylation in body fluids may complement PSA
screening: a meta-analysis. Br. J.Cancer 105, 6573
(2011).
45. Ellinger, J. etal. CpG island hypermethylation at
multiple gene sites in diagnosis and prognosis of
prostate cancer. Urology 71, 161167 (2008).
46. Lee, B.B. etal. Aberrant methylation of APC, MGMT,
RASSF2A, and Wif1 genes in plasma as a biomarker
for early detection of colorectal cancer. Clin. Cancer
Res. 15, 61856191 (2009).
47. Lofton-Day, C. DNA methylation biomarkers for bloodbased colorectal cancer screening. Clin. Chem. 54,
414423 (2008).
48. Warren, J.D. etal. Septin 9 methylated DNA is a
sensitive and specific blood test for colorectal cancer.
BMC Med. 9, 133 (2011).
49. Glockner, S.C. Methylation of TFPI2 in stool DNA: a
potential novel biomarker for the detection of colorectal
cancer. Cancer Res. 69, 46914699 (2009).
50. Balana, C. etal. Tumour and serum MGMT promoter
methylation and protein expression in glioblastoma
patients. Clin. Transl. Oncol. 13, 677685 (2011).
51. Esteller, M. etal. Detection of aberrant promoter
hypermethylation of tumor suppressor genes in serum
DNA from non-small cell lung cancer patients. Cancer
Res. 59, 6770 (1999).
REVIEWS
104. Dedeurwaerder, S. etal. Evaluation of the Infinium
Methylation 450K technology. Epigenomics 3,
771784 (2011).
105. Tost, J., Schatz, P., Schuster, M., Berlin, K. & Gut, I.G.
Analysis and accurate quantification of CpG
methylation by MALDI mass spectrometry. Nucleic
Acids Res. 31, e50 (2003).
106. Cocklin, R.R. & Wang, M. Identification of methylation
and acetylation sites on mouse histone H3 using
matrix-assisted laser desorption/ionization time
offlight and nanoelectrospray ionization tandem
mass spectrometry. J.Protein Chem. 22, 327334
(2003).
107. Ehrich, M. etal. Quantitative high-throughput analysis
of DNA methylation patterns by base-specific cleavage
and mass spectrometry. Proc. Natl Acad. Sci. USA
102, 1578515790 (2005).
108. Schatz, P., Dietrich, D. & Schuster, M. Rapid analysis
of CpG methylation patterns using RNase T1
cleavage and MALDI-TOF. Nucleic Acids Res. 32,
e167 (2004).
109. Kelly, T.K., De Carvalho, D.D. & Jones, P.A.
Epigenetic modifications as therapeutic targets.
Nature Biotech. 28, 10691078 (2010).
110. Yoo, C.B. etal. Delivery of 5aza2deoxycytidine to
cells using oligodeoxynucleotides. Cancer Res. 67,
64006408 (2007).
111. Kelly, W.K. etal. Phase I study of an oral histone
deacetylase inhibitor, suberoylanilide hydroxamic acid,
in patients with advanced cancer. J.Clin. Oncol. 23,
39233931 (2005).
112. Duvic, M. etal. Phase 2 trial of oral vorinostat
(suberoylanilide hydroxamic acid, SAHA) for refractory
cutaneous Tcell lymphoma (CTCL). Blood 109, 3139
(2007).
113. Garcia-Manero, G. etal. Phase 1 study of the histone
deacetylase inhibitor vorinostat (suberoylanilide
hydroxamic acid [SAHA]) in patients with advanced
leukemias and myelodysplastic syndromes. Blood 111,
10601066 (2008).
114. Whittaker, S.J. etal. Final results from a multicenter,
international, pivotal study of romidepsin in refractory
cutaneous Tcell lymphoma. J.Clin. Oncol. 28,
44854491 (2010).
115. Filippakopoulos, P. etal. Selective inhibition of BET
bromodomains. Nature 468, 10671073 (2010).
116. Dawson, M.A. etal. Inhibition of BET recruitment to
chromatin as an effective treatment for MLL-fusion
leukaemia. Nature 478, 529533 (2011).
117. Bernt, K.M. etal. MLL-rearranged leukemia is
dependent on aberrant H3K79 methylation by
DOT1L. Cancer Cell 20, 6678 (2011).
118. Krivtsov, A.V. etal. H3K79 methylation profiles define
murine and human MLLAF4 leukemias. Cancer Cell
14, 355368 (2008).
119. Daigle, S.R. Selective killing of mixed lineage leukemia
cells by a potent small-molecule DOT1L inhibitor.
Cancer Cell 20, 5365 (2011).
Acknowledgements
FURTHER INFORMATION
Groups homepage: http://www.pebc.cat
BLUEPRINT project: http://www.blueprint-epigenome.eu
Computational Epigenetics: http://www.computationalepigenetics.de
CURELUNG project: http://www.curelung.eu
Epigenomics at UCL: http://www.ucl.ac.uk/cancer/medicalgenomics/medgenhome
International Human Epigenome Consortium: http://www.
ihec-epigenomes.org
Roadmap Epigenomics Project: http://www.
roadmapepigenomics.org
ALL LINKS ARE ACTIVE IN THE ONLINE PDF
www.nature.com/reviews/genetics
2012 Macmillan Publishers Limited. All rights reserved