Global Methylation Patterns in Idiopathic Pulmonary
Einat I. Rabinovich1, Maria G. Kapetanaki1, Israel Steinfeld2, Kevin F. Gibson1, Kusum V. Pandit1,
Guoying Yu1, Zohar Yakhini2,3, Naftali Kaminski1*
1Dorothy P. and Richard P. Simmons Center for Interstitial Lung Disease, Division of Pulmonary, Allergy and Critical Care Medicine, University of Pittsburgh, School of
Medicine, Pittsburgh, Pennsylvania, United States of America, 2Department of Computer Sciences, Technion – Israel Institute of Technology, Haifa, Israel, 3Agilent
Laboratories, Tel-Aviv, Israel
Background: Idiopathic Pulmonary Fibrosis (IPF) is characterized by profound changes in the lung phenotype including
excessive extracellular matrix deposition, myofibroblast foci, alveolar epithelial cell hyperplasia and extensive remodeling.
The role of epigenetic changes in determining the lung phenotype in IPF is unknown. In this study we determine whether
IPF lungs exhibit an altered global methylation profile.
Methodology/Principal Findings: Immunoprecipitated methylated DNA from 12 IPF lungs, 10 lung adenocarcinomas and
10 normal histology lungs was hybridized to Agilent human CpG Islands Microarrays and data analysis was performed using
BRB-Array Tools and DAVID Bioinformatics Resources software packages. Array results were validated using the EpiTYPER
MassARRAY platform for 3 CpG islands. 625 CpG islands were differentially methylated between IPF and control lungs with
an estimated False Discovery Rate less than 5%. The genes associated with the differentially methylated CpG islands are
involved in regulation of apoptosis, morphogenesis and cellular biosynthetic processes. The expression of three genes
(STK17B, STK3 and HIST1H2AH) with hypomethylated promoters was increased in IPF lungs. Comparison of IPF methylation
patterns to lung cancer or control samples, revealed that IPF lungs display an intermediate methylation profile, partly similar
to lung cancer and partly similar to control with 402 differentially methylated CpG islands overlapping between IPF and
cancer. Despite their similarity to cancer, IPF lungs did not exhibit hypomethylation of long interspersed nuclear element 1
(LINE-1) retrotransposon while lung cancer samples did, suggesting that the global hypomethylation observed in cancer was
not typical of IPF.
Conclusions/Significance: Our results provide evidence that epigenetic changes in IPF are widespread and potentially
important. The partial similarity to cancer may signify similar pathogenetic mechanisms while the differences constitute IPF
or cancer specific changes. Elucidating the role of these specific changes will potentially allow better understanding of the
pathogenesis of IPF.
Citation: Rabinovich EI, Kapetanaki MG, Steinfeld I, Gibson KF, Pandit KV, et al. (2012) Global Methylation Patterns in Idiopathic Pulmonary Fibrosis. PLoS ONE 7(4):
Editor: Oliver Eickelberg, Helmholtz Zentrum Mu ¨nchen/Ludwig-Maximilians-University Munich, Germany
Received June 30, 2011; Accepted February 16, 2012; Published April 10, 2012
Copyright: ? 2012 Rabinovich et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: NK, EIR, MK, KFG, KPV and GY were funded by NIH grants R01HL095397, R01LM009657, RC2HL101715 and the Dorothy P. and Richard P. Simmons Chair
for Pulmonary Research. EIR was funded by American Lung Association Senior Research Training Fellowship Grand. The above funders had no role in study
design, data collection and analysis, decision to publish or preparation of the manuscript.
Competing Interests: The authors have read the journal’s policy and have the following conflicts: Dr. Yakhini is employed at Agilent Laboratories and his work
on the manuscript was funded by Agilent. There are no patents, products in development or marketed products to declare. Dr Yakhini’s employment at Agilent
had no impact on the design of the study and does not alter the authors’ adherence to all the PLoS ONE policies on sharing data and materials.
* E-mail: email@example.com
Idiopathic pulmonary fibrosis (IPF) is a non-neoplastic pulmo-
nary disease, characterized by extracellular matrix deposition,
myofibroblasts foci formation and alveolar epithelial cell hyper-
plasia [1,2,3,4]. The disease is progressive and in most cases
unresponsive to corticosteroid and immunosuppressive therapy
. Although the exact etiology of the disease is still under
investigation, several studies suggest that a combination of genetic
and environmental factors may be the cause of IPF . Exposure
to wood, metal dust or stone/sand/silica as well as smoking,
farming and handling livestock are associated with IPF in several
independent studies . A unique feature to the lung phenotype in
IPF is the extent to which the lung is altered from normal. Alveolar
epithelial cells and fibroblasts exhibit distinct and profound
changes in their phenotypes with alveolar epithelial cells
undergoing hyperplasia and potentially epithelial mesenchymal
transdifferentiation and fibroblasts becoming activated and
exhibiting myofibroblast features. Multiple studies demonstrated
that the lung phenotype in IPF is dramatically different than that
of the healthy lung with globally different patterns of mRNA and
microRNA expression [8,9,10,11,12,13,14] and aberrations in
multiple pathways such as coagulation , apoptosis [16,17],
oxidative stress , epithelial mesenchymal transition [19,20,21]
and developmental pathways [21,22,23]. Usually it is assumed that
multiple cycles of injury lead to this phenotype, however these
PLoS ONE | www.plosone.org1April 2012 | Volume 7 | Issue 4 | e33770
injuries do not explain how those profound phenotypic changes
are sustained and even progress years after the initial injury.
Global epigenetic changes, traditionally defined in the context of
heritable changes that are not coded by changes in DNA
sequence, have rapidly emerged as a general mechanism by
which cellular molecular phenotypes are stably altered during
development, cellular differentiation, response to environmental
stress and disease pathogenesis [24,25,26]. It is well established
that nutritional, chemical and physical factors can have a
profound effect on gene expression . Not only can they cause
mutations in the promoter and coding regions of genes but they
can also orchestrate a variety of epigenetic changes . Two of
the best described mechanisms of epigenetic control are DNA
methylation and chromatin remodeling. DNA methylation
typically involves the addition of a methyl group to the 5 position
of the cytosine pyrimidine ring of a CpG dinucleotide .
Clustered CpGs form CpG islands whose state of methylation is
critical for the activity of transposable elements and the
transcriptional regulation of genes through direct blockage of
transcription factors or chromatin remodeling . Alterations of
CpG methylation have been implicated in many diseases where
the hypermethylation of the promoter associated CpG islands
results in transcriptional silencing  while the hypomethylation
results in loss of imprinting and transcriptional activation .
Aberrant methylation of CpG dinucleotides is a well-recognized
epigenetic hallmark of multiple diseases including lung cancers
[33,34,35,36]. So far, the extent and role of epigenetic changes has
not been studied in IPF.
In this study, we analyze global methylation patterns of IPF
using human CpG island microarrays. In addition to compiling a
DNA methylation profile that differentiates IPF patients from
normal individuals, we compared this profile to that of lung
adenocarcinoma patients. Our results reveal an altered DNA
methylation pattern in IPF which shows great similarity to the
methylation pattern of lung cancer. Our work is the first step in
understanding the role of DNA methylation in the pathogenesis of
IPF. Furthermore, the similarity of IPF with cancer may reveal
common underlying molecular mechanisms and offer therapeutic
options for IPF patients adopted from cancer biology .
Lung tissue samples were obtained through the University of
Pittsburgh Health Sciences Tissue Bank (Pittsburgh, PA). They
included 12 frozen lung tissue samples from IPF patients, 10
frozen lung tissue samples from adenocarcinoma patients and 10
histological normal lung samples obtained from the same group of
adenocarcinoma patients (Table 1). The diagnosis of IPF was
based on microscopic findings that were consistent with usual
interstitial pneumonia [1,3]. All adenocarcinoma tumors were
obtained from patients staged as T1b-T2bN0M0. All cancer
patients were smokers and older than IPF patients. The lung
samples that were removed from patients with lung cancer
contained both adenocarcinoma tissues and normal histology
tissues obtained from disease-free margins of the lung. The IPF
patients fulfilled the diagnostic criteria of the American Thoracic
Society and the European Respiratory Society . Patients
consented to the donation of the removed tissue to the tissue
bank and the use of the archived tissue was approved by the
Table 1. Human subjects.
Variable Control/Cancer IPF
Number of subjects 1012
Average age, yr.71(612.44)60(65.51)
Figure 1. CpG islands are differentially methylated in IPF and
control samples. (A) Human CpG Island Microarray data: the heatmap
on the left is the visual comparison of global methylation profiles
between the 10 Control and the 12 IPF samples. The heatmap on the
right consists only of the differentially methylated CpG probes
highlighted by the red rectangle (p-value , 0.05, FDR,5%). Methylated
CpG islands are shown in progressively brighter shades of yellow,
depending on the fold difference, and hypomethylated CpG islands are
shown in progressively brighter shades of purple. Grey stands for no
difference between the two sample groups being compared. (B)
EpiTYPER confirmation of differentially methylated CpG islands. Bars
represent the average methylation of all the samples per study group.
The X-axis shows the genomic location of each CpG island and Y-axis
shows the percentage of methylation.
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org2April 2012 | Volume 7 | Issue 4 | e33770
University of Pittsburgh Institutional Review Board (IRB020123,
IRB0506140, IRB0411036). (Pittsburgh, PA).
MeDIP and Hybridization to Human CpG Island
Genomic DNA was extracted from 10 mg of frozen lung tissues
using the DNeasy Blood & Tissue kit (Qiagen Sciences, CA)
following the manufacturer’s protocol. Five micrograms of
genomic DNA were sonicated (Sonic Dismembrator, Fisher
Scientific) to achieve fragment lengths between 300–600 bp.
Methylated DNA was immunoprecipitated (IP) using 5-Methylcy-
tidine monoclonal antibody (AbD Serotec, NC) as instructed in the
Agilent Microarray Analysis of Methylated DNA Immunoprecip-
itation protocol (Agilent Technologies, Santa Clara, CA) [39,40].
Immunoprecipitated DNA and total genomic DNA were labeled
with Cy3 and Cy5 respectively, using Agilent Genomic Labeling
kit PLUS (Agilent Technologies, Santa Clara, CA) and hybridized
to the human CGI oligonucleotide microarrays (Agilent Technol-
ogies). The arrays were designed according to the University of
California at Santa Cruz (UCSC) genome-browser CpG island list
and contained 237,000 probes covering more than 90% of the
27,639 human CpG islands at a density of 1 probe per 100 bp as
described in Straussman et al  along with quality control assays
that assess the platform’s performance. Further validation of
Agilent’s MeDIP microarray platform was achieved by Yamashita
et al . Following hybridization and washing, the arrays were
scanned in an Agilent G2509C microarray scanner and raw data
were obtained using Feature Extraction Ver.9.5.3.
Microarray Data Analysis
For our analysis we only included probes with a hybridization
Tm value between 79uC and 93uC as these show higher quality
signal . We subsequently divided the probes according to their
Tm into 14 groups/bins differing by 1uC. Probe signals in each bin
were standardized to have an average of 0 and a standard
deviation of 1. To work in a CpG island oriented manner we
scored each island for its likelihood to be methylated. For that
purpose, each probe was mapped to the genome and the signals of
the probes that were mapped to a single CpG island were
averaged to obtain the island’s methylation score . The
complete microarray data have been deposited in the Gene
Expression Omnibus (GSE29895), are MIAME compliant and
MassARRAY EpiTYPER Assay
CpG dinucleotide methylation was quantified by the MassArray
EpiTYPER platform (Sequenom Inc, CA) . The EpiTYPER
assay is a MALDI TOF mass spectrometry based quantitative
method for measuring CpG methylation down to a single
dinucleotide resolution. 500 ng of fragmented DNA from each
sample was modified by bisulfite treatment. Following PCR with
specific primers and Shrimp Alkaline Phosphatase treatment,
fragments were ligated to a T7 promoter segment, and then
transcribed into RNA. The synthesized RNA was cleaved with
RNase A and all cleavage products were analyzed by MassArray
in the Genomics and Proteomics Core Laboratory (GPCL,
University of Pittsburgh, Pittsburgh, PA) according to the
manufacturer’s instructions. Primers were designed using the
EpiDesigner Software (http://www.epidesigner.com/index.html)
(Table S1 in Supporting Information).
Quantitative Real-Time Polymerase Chain Reaction (qRT-
Total RNA was extracted from frozen lung tissue with
miRNeasy mini kit (Qiagen Sciences,CA) following the manufac-
turer’s protocol . 500 ng of the extracted RNA sample was
used as a template for the reverse transcriptase reaction. 25 ng of
the synthesized cDNA was amplified in a qPCR reaction using
TaqMan universal PCR master mix (Applied Biosystems, Foster
City,CA) and TaqMan gene expression assays for the following
Hs00169491_m1), HIST1H2AH (assay Hs00544732_s1) and
GUSB (assay Hs99999908_m1 ). All assays were done in triplicates
and appropriate Non-Transcriptase and Non-Template control
reactions were included. GUSB (encoding b-glucoronidase) was
used as a housekeeping gene for normalization and the results
were analyzed by the DDCT method  after averaging the
triplicates of each assay. Fold change was calculated by taking the
average of all the control samples as the baseline.
Differentially methylated CpG islands were identified by
analyzing the CpG Island Microarray data with the Class
Comparison feature of BRB-ArrayTools 3.7.0 (http://linus.nci.
nih.gov/BRB-ArrayTools.html). We controlled for multiple testing
by setting the significance level at a False Discovery Rate (FDR) of
less than 5% . Data visualization was accomplished using the
Genomica  and the JavaTreeView software packages. The
Student’s t test was applied to for the EpiTYPER MassArray and
qRT-PCR to test significance of the results. Significance of overlap
of differentially methylated islands (DMI) between IPF and Cancer
samples and enrichment of DMIs in promoter regions was
calculated using the hypergeometric distribution. Pathway analysis
was performed using DAVID Bioinformatics Resources 6.7 
and IPA Ingenuity Systems (http://www.ingenuity.com).
Table 2. Functional annotation clustering of differentially methylated CpG islands.
MCF2L,PROC,AKT1,IGF1R,NOTCH1,IGF2R, BIRC8, BCL6,DNAJB6, ARHGDIA,TERT
negative regulation of
regulation of cell
histone acetylation2.841.36E-03 KAT2A,BRD1,KAT2B,CREBBP
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org3April 2012 | Volume 7 | Issue 4 | e33770
Table 3. Differentially methylated promoters between IPF and control samples.
CpG locationGene Eponine Predicted TSS
chr7:148825072-148826208 ZNF 746chr7.53295-53308 CHR7_M10391.19E-030.52
chr19:2653532-2653987GNG7chr19.27595-27597CHR19_M0064,CHR19_P0 1.72E-03 0.53
chr22:37569114-37570371 NPTXRchr22.39508-39517 CHR22_M0327 3.77E-050.56
chr12:119290634-119291409MSI1 chr12.12663-12672 CHR12_M0815 4.79E-050.58
chr17:53515753-53516493 DYNLL2 chr17.23937CHR17_P06991.10E-040.58
chrY:2537106-2537697 NCRNA00103chrY.61288 CHRY_P00072.22E-040.59
chr13:31318630-31319274EEF1DP3chr13.13604-13608 CHR13_P00912.02E-04 0.6
chr6:27222201-27223160HIST1H2AH chr6.48519CHR6_M0196,CHR6_P02006.12E-04 0.61
chr20:46971570-46972079 ARFGEF2 chr20.36249-36250CHR20_P0396 1.65E-040.62
chr20:390530-391317TBC1D20 chr20.35209-35211CHR20_M0004 3.33E-040.62
chr3:20056462-20057430 KAT2Bchr3.40742-40745 CHR3_P01211.15E-040.63
chr8:1908966-1910279KBTBD11 chr8.53949-53953 CHR8_P0014,CHR8_M03.49E-04 0.63
chr9:21549133-21549816 LOC554202 chr9.56427CHR9_M01159.90E-040.63
chr7:16427303-16427790ISPD chr7.511054.83E-04 0.64
chr7:4889233-4890102RADIL chr7.50913-50914CHR7_M00424.83E-04 0.64
chr10:73203097-73203498C10orf54 chr10.6463-64641.31E-03 0.64
chr2:37237405-37237906EIF2AK2 CHR2_M0214 7.71E-040.64
chr9:138236486-138236814LHX3CHR9_M0813 9.64E-05 0.66
chr22:37481403-37482422SUN2 chr22.39495-39505CHR22_M0325.1 1.15E-039.68
chr17:77421925-77423424 ARHGDIAchr17.24948-24955 CHR17_M10491.00E-03 0.68
chr1:219026639-219027226 MOSC1chr1.4565 CHR1_P1706_R1 5.74E-040.69
chr4:77391611-77392084FAM47D chr4.44416-44418CHR4_P0385 2.44E-04 0.72
chr3:52714577-52715466GLT8D1-SPCS1 chr3.41472-41477CHR3_M0331,CHR3_P0373 3.65E-040.72
chr2:10747175-10747692NOL10 CHR2_M0064, CHR2_P00548.54E-041.51
chr22:29886124-29886466 RNF185 CHR22_P02272.63E-041.53
chr4:191142227-191143118 TUBB4Q chr4.454029.16E-051.53
chr3:50357893-50358314ZMYND10 chr3.41343CHR3_M03002.53E-04 1.56
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org 4 April 2012 | Volume 7 | Issue 4 | e33770
The patterns of DNA methylation in lung samples of IPF,
cancer patients and controls, were determined using Agilent
Human CpG Islands microarrays. Overall, 12 IPF, 10 lung
adenocarcinoma and 10 normal histology samples from the same
adenocarcinoma patients were included in our study (Table 1).
The analysis of the microarray data was divided into two parts. In
the first part, the IPF or the adenocarcinoma samples were
compared to the control samples to compile two separate lists of
differentially methylated CpG islands. In the second part, the two
lists were compared to assess for differences or similarities between
the methylation changes that are associated with each disease.
IPF Lung Samples Show a Different Methylation Profile
when Compared to Normal Histology Lung Samples.
The 25,406 out of 27,639 human CpG islands that had an
acceptable Tm (see methods) were analyzed using the Class
Comparison algorithm from BRB Array Tool software package.
625 CpG islands were found to be differentially methylated in IPF
lung tissue samples when compared to control lung tissue samples
(Figure 1A, Table S2 in Supporting information). 91.2% of the
625 differentially methylated CpG islands were located in intronic,
exonic or and intergenic areas and only 8.8% in promoters.
Considering that 10,923 of the 25,406 (43%) CpG islands in our
study localize to promoters, this result indicates that a significantly
larger than expected (p , 10–79) proportion of changes in
methylation, when comparing IPF and control samples, occurs in
regions that are not annotated as promoters in the current genome
To validate the microarray results, 3 differentially methylated
CpG islands showing various degrees of change in their
methylation levels were picked and analyzed with the Sequenom’s
MassArray EpiTYPER assay. The EpiTYPER assays showed
decreased CpG island methylation in the IPF lung samples which
was in agreement with the microarray data (Figure 1B). All
differentially methylated CpG islands were mapped to the genome
using the UCSC genome browser  and a list of genes that
contain CpG islands showing significantly hyper- or hypomethyla-
tion in IPF lung samples was compiled (Table S2 in Supporting
Information). A Functional Annotation Clustering of these genes
using DAVID Bioinformatics Resources 6.7 revealed that a
significant number of them are involved in apoptosis, cell
morphogenesis, the regulation of cellular biosynthetic processes
and histone acetylation (Table 2). The modified Fisher Exact p-
Value/EASE Score is calculated to measure gene-enrichment in
any given annotation term. It ranges from 0 to 1 with 0
representing perfect enrichment. ‘‘Score’’ stands for Group
Enrichment Score, which is calculated using the p-values of the
individual members of each Functional Annotation Cluster. The
higher the number is the higher the cluster ranks in biological
Decrease in Promoter CpG Island Methylation is
Associated with Increased Gene Expression
Typically, the methylation of promoter localized in CpG
islands affects gene expression of the downstream genes . All
625 differentially methylated CpG islands were checked for
promoter localization and presence of a Trascriptional Start Site
(TSS) using the UCSC Genome Browser . 55 CpG islands
were mapped in the promoter region of genes that had a well-
characterized TSS (Table 3). An IPA functional analysis showed
that the genes with differentially methylated CpG islands in their
promoters are associated with biological processes such as cellular
assembly and organization, cellular growth and proliferation, cell
morphology, cancer ,cell signaling, gene expression and cell
death (Table S3 in Supporting Information). We analyzed by
qRT-PCR three genes localized in differentially regulated
regions. Serine/Threonine Kinase 17b (STK17B) and Serine/
Threonine Kinase 3 (STK3) are involved in apoptosis while
histone cluster 1 H2ah (HIST1H2AH) is essential is nucleosome
formation. All three transcripts showed increased levels of
expression in IPF samples compared to controls but only
STK17B and HIST1H2AH have a p-value ,0.05 while in the
case of STK3 the p-value is 0.07 (Figure 2).
Table 3. Cont.
CpG locationGeneEponine Predicted TSS
Figure 2. Expression of genes with differentially methylated
promoters. qRT-PCR assay on 3 genes with hypomethylated
promoter-associated CpG islands showed increase in the expression
of the downstream gene. Y-axis shows fold change of detected
transcripts in IPF samples when the expression in controls is set to
baseline equal to 1. * denotes p-values ,0.05. Error bars are based on
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org5April 2012 | Volume 7 | Issue 4 | e33770
The Methylation Profiles of IPF and Adenocarcinoma
Lung Samples Overlap
To determine the similarity of IPF samples to lung cancer we
performed Principal Component Analysis (PCA) and Class
Comparison using BRB-Array Tools on microarray data from
12 IPF patients and the 10 lung cancer patients (normal histology
and adenocarcinoma samples included). PCA analysis demon-
strated that IPF samples were positioned between the control and
cancer samples suggesting that IPF samples had a methylation
profile with partial similarity to both groups (Figure 3A). It is
worth mentioning that IPF samples were more similar to cancer
than control samples despite the fact that cancer and control tissue
were obtained in pairs from the same patient. This observation
may suggest that the majority of differences between IPF and
controls may be related to differential environmental exposures or
smoking effect because these differences persisted in the compar-
ison of cancer to control despite the fact they came from the same
subject. Class comparison analysis revealed that 2428 CpG islands
were differentially methylated between cancer samples and normal
histology controls. When compared to the 625 that are
differentially methylated between IPF and Controls, 402 CpG
islands overlapped. In other words, 65% of the CpG islands that
have an altered methylation pattern in IPF lung samples are also
modified in lung cancer samples (Figure 3B and Table S4 in
Supporting Information). This overlap is highly significant, as the
probability of such an overlap to occur in random is very low
(p,102256). 45% of the 402 overlapped CpG islands are located in
intronic and intergenic areas, 6% in promoters and 49% in exons.
To determine whether similar methylation patterns in IPF and
cancer result from a global change in methylation we assessed
LINE-1 methylation. LINE-1 retrotransposons are abundantly
and equally distributed across the genome and their methylation
pattern is often used as an indicator of global methylation levels
. The methylation status of LINE-1 (GenBank: X58075.1) was
defined in all three study groups (IPF, Cancer and Control) using
the EpiTYPER MassArray assay. The PCR primers were
designed to encompass the 15 CpG sites or units including the
possible intrinsic LINE-1 promoter (Table S1 in Supporting
Information). Although LINE-1 elements were found to be
hypomethylated in the adenocarcinoma samples no significant
change of the methylation levels was detected the in IPF samples
(Figure 4) suggesting that methylation changes in IPF were
specific to regions.
In this study, we used human CpG island microarrays to
identify the differentially methylated CpG islands in the lung tissue
of IPF patients. Our results indicate that the CpG island
methylation profile of the IPF lung samples is very different from
that of control samples and greatly overlaps with methylation
changes observed in lung adenocarcinoma samples. Despite the
observed similarity in CpG methylation between IPF and lung
cancer, the lack of LINE-1 methylation in IPF suggests a more
specific DNA methylation, which is confined to certain regions of
One of the most impressive results of our study is the extent of
differentially methylated regions in the IPF lungs. Interestingly the
majority of the differentially methylated CpG islands rest in
promoter-distal sites or intragenic regions and only 8.8% of them
are localized in gene promoters. Whereas the methylation status of
promoter associated CpG islands can directly affect transcription,
the role of the CpG methylation outside the immediate promoter
region remains somewhat unclear. It is proposed that methylation
Figure 3. Comparison of IPF and Adenocarcinoma to control
samples. (A) 3-D plot representation of the results after Principal
Component Analysis of all 3 sample groups. Each color-coded dot
represents a sample (red-control, green-IPF and blue-cancer) and each
sample is positioned in the 3-D space according to its similarity or
difference to the others. (B) Comparison of differentially methylated
CpG islands that overlap between IPF and Lung Cancer. The heatmap
consists of 402 differentially methylated CpG islands that are found to
overlap between IPF and Lung Cancer. High methylation levels of a CpG
islands are shown in yellow while low methylation levels of methylation
are shown in purple. Grey stands for no difference between the two
groups being compared.
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org6April 2012 | Volume 7 | Issue 4 | e33770
of CpG island shores outside the promoter could also control
transcription of downstream genes  or lead to histone
modifications . Methylation changes that occur in intragenic
regions could impact RNA splicing . In addition, methylation
changes may affect the expression of non-coding RNAs  and
thus indirectly affect global changes in gene expression. The
biological impact of modest changes in the degree of CpG
methylation is in fact unpredictable. As an example in the case of
prostate cancer, a gradual increase in methylation from 12.6% to
19.3% or 21.8% signified a transition from a benign state to a
localized or metastatic cancer, respectively . However,
regardless of the direct downstream effects, the extent of the
methylation changes we found, supports previous observations
about the degree and profundity of molecular changes in the IPF
Naturally, the motivation to assemble methylation profiles is to
find the underlying mechanisms that drive changes in gene
expression. The detailed characterization of each one of the
differentially methylated CpG islands in IPF patients is beyond the
scope of this study. Globally, genes with differentially methylated
CpG islands in their promoters were involved in biological
processes such as cellular assembly and organization, cellular
growth and proliferation, cell morphology, cancer, cell signaling,
gene expression and cell death. All of these processes could be
implicated in IPF pathogenesis. In our validation we focused on
genes with differentially methylated promoters. We selected the
Serine/Threonine Kinase 17b (STK17B) and Serine/Threonine
Kinase 3 (STK3) because of their role in apoptosis  and the
histone cluster 1 H2ah (HIST1H2AH) because of the recent
interest in histone modifications in fibrosis. STK17B and
HIST1H2AH were significantly up-regulated in our IPF samples
which is in agreement with the hypo-methylated state of their
promoter associated CpG islands. Interestingly, the majority of the
differentially methylated islands that were within or close to known
genes were outside promoter regions. Some of these methylation
changes were in genes that were previously reported to be
increased in IPF such as COL18A1 , genes that are implicated
in myofibroblast differentiation such as NOTCH1  or markers
of progressive IPF like SMARCA4 . In addition, the promoter
of CXCL3, a gene which is found to be up-regulated in the lung of
bleomycin treated mice , was also hypomethylated in our IPF
samples. When we looked for the overlap of differentially
expressed genes in IPF in our previously published gene expression
datasets [13,14] we found that there were 46 genes that had both
differentially methylated gene related CpG islands and gene
expression changes. While a detailed analysis of methylation and
expression changes in the same tissue would be better suited to
address the correlation of methylation and gene expression
changes our findings suggest that at least some of the methylation
changes that we observed do have an effect on lung gene
expression and thus may contribute to the lung phenotype in IPF.
A remarkable finding of our study is the similarity in DNA
methylation patterns between IPF and lung adenocarcinoma.
Recently, Vancheri et al compared IPF to cancer and described the
pathogenic similarities between the two diseases. More specifically,
they referred to common genetic and epigenetic alterations,
uncontrolled proliferation, tissue invasion and perturbation of
signal transduction pathways . The similarity between cancer
and IPF spreads to microRNA expression such as in the case of let-
7d and hsa-miR-21, which are found to be down-regulated or up-
regulated respectively in both diseases [8,9]. All of these
observations are in accord with published studies reporting high
incidence of cancer in IPF patients when compared to healthy
individuals [62,63]. DNA hypomethylation is a hallmark of cancer
 and in many types of cancer including lung carcinomas it is
accompanied with lower levels of methylation in repetitive DNA
elements [34,65]. While the similarity in the differentially
methylated CpG islands suggests common epigenetic mechanisms
between IPF and cancer, our analysis of LINE-1 methylation
indicates that this similarity is limited. LINE-1 repeats comprise
about 20% of the human genome . LINE-1 elements are
usually methylated in somatic tissues but they are often
Figure 4. Methylation profile of LINE-1 retrotransposon. Global methylation levels of the LINE-1 retrotransposons were defined using the
EpiTYPER mass array assay. Each bar represents the methylation in one of the 15 CpG dinucleotides or CpG units that span the LINE-1 sequence.
Methylation levels are calculated as the average of all samples in each group (Control, IPF, Cancer) and standard error bars are included. The X axis
shows the CpG dinucleotide or the CpG unit and the Y axis shows the percentage of methylation. The LINE-1 promoter is indicated in purple. The red
arrowheads indicate units of 2-3 CpG dinucleotides.
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org7April 2012 | Volume 7 | Issue 4 | e33770
hypomethylated in tumors [65,67] resulting in increased mobility,
which in turn leads to gene disruptions  and chromosomal
instability . While LINE-1 retrotransposons were hypomethy-
lated in our cancer samples they were not in IPF samples leading
to the conclusion that CpG island methylation changes in IPF are
somewhat parallel to cancer but are not as extensive and do not
involve global changes in LINE-1 methylation. This suggests that
despite the similarities between the DNA methylation profiles of
IPF and cancer, there are different mechanisms that cause and
sustain these changes.
One of the major concerns in our global profiling approach is
tissue heterogeneity. The IPF lungs contain mixed areas of normal
tissue, myofibroblast foci and honeycombing . The IPF lung is
also highly variable in its cellular content as it contains normal cells
like epithelial, endothelial cells and fibroblasts as well as abnormal
ones like hyperplastic type II alveolar epithelial cells, myofibro-
blasts, potentially altered endothelial cells and varying degrees of
inflammatory cells. Thus it is possible that the signal we obtained is
only an under-estimation of the real epigenomic changes caused
by an admixture of normal and abnormal regions, microenviron-
ments and cell types. Naturally, it is impossible based on our
analysis to determine whether the observed DNA methylation
changes are cell type specific. In this context, our strategy of
averaging signals across an island could also lead to loss of
information and underestimation of epigenetic changes. However,
we chose this approach because although it is less sensitive, we felt
it provided us with global results, reduced the need to deal with
probe variability and provided a good approximation of
differentially methylated CpG islands. In the future it may make
sense to refine both the measurement approach and data analysis
to obtain more detailed results. The heterogeneity of our samples
as well as the different methodologies used to identify the
differences in CpG methylation could also explain the absence
of PTGER2 and Thy-1 from our list of significantly methylated
genes. The promoters of PTGER2 and Thy-1 were found
to be hypermethylated in fibrotic lung fibroblasts and fibrotic
tissue from IPF patients resulting in low levels of the coded
proteins. In fact Thy-1 it is shown that the downregulation occurs
only in areas of dense fibrosis and fibrotic foci while the rest of the
tissue remains unaffected [70,71]. However, the demonstration of
significant global methylation changes despite the limitations of
our methods, may be indicative of the importance of epigenomic
regulation in IPF and lead to many more detailed discoveries and
To the best of our knowledge our study is the first one to
describe global DNA methylation changes in IPF lungs. Taken
together with the extensive changes in gene histology, gene
expression and microRNA profiles our results highlight the
profundity and complexity of events underlying the phenotypic
changes in IPF and to some extent suggest that interfering with
one pathway may not be sufficient to reverse these changes. The
differentially methylated CpG islands we identified should be
further studied as their regulation could provide insights about
how genotype and the environment interact to determine the lung
phenotype in IPF. Based on our results, we believe that epigenetic
modifications play a key role in the pathogenesis of IPF and thus
could serve as disease biomarkers and therapeutic targets.
The sequence of EpiTYPER MassArray
guishing IPF from controls.
Differentially methylated CpG islands distin-
methylated promoters in IPF.
Functional Analysis of the 55 Differentialy
lapping between IPF and cancer.
Differentially methylated CpG islands over-
The authors would like to thank the members of Dr. Kaminski’s Lab for
their support and constructive criticism.
Conceived and designed the experiments: EIR NK KFG. Performed the
experiments: EIR. Analyzed the data: ZY IS EIR NK MGK. Contributed
reagents/materials/analysis tools: KVP GY KFG. Wrote the paper: EIR
MGK NK. Sample selection: KFG EIR NK.
1.American Thoracic Society (2000) Idiopathic pulmonary fibrosis: diagnosis and
treatment. International consensus statement. American Thoracic Society
(ATS), and the European Respiratory Society (ERS). Am J Respir Crit Care
Med 161: 646–664.
Gross TJ, Hunninghake GW (2001) Idiopathic pulmonary fibrosis. N Engl J Med
Katzenstein AL, Myers JL (1998) Idiopathic pulmonary fibrosis: clinical
relevance of pathologic classification. Am J Respir Crit Care Med 157:
Selman M, King TE, Pardo A (2001) Idiopathic pulmonary fibrosis: prevailing
and evolving hypotheses about its pathogenesis and implications for therapy.
Ann Intern Med 134: 136–151.
Collard HR, King TE Jr. (2001) Treatment of idiopathic pulmonary fibrosis: the
rise and fall of corticosteroids. Am J Med 110: 326–328.
Taskar VS, Coultas DB (2006) Is idiopathic pulmonary fibrosis an environ-
mental disease? Proc Am Thorac Soc 3: 293–298.
Taskar V, Coultas D (2008) Exposures and idiopathic lung disease. Semin
Respir Crit Care Med 29: 670–679.
Pandit KV, Corcoran D, Yousef H, Yarlagadda M, Tzouvelekis A, et al. (2010)
Inhibition and role of let-7d in idiopathic pulmonary fibrosis. Am J Respir Crit
Care Med 182: 220–229.
Liu G, Friggeri A, Yang Y, Milosevic J, Ding Q, et al. (2010) miR-21 mediates
fibrogenic activation of pulmonary fibroblasts and lung fibrosis. J Exp Med 207:
10. Rajkumar R, Konishi K, Richards TJ, Ishizawar DC, Wiechert AC, et al. (2010)
Genomewide RNA expression profiling in lung identifies distinct signatures in
idiopathic pulmonary arterial hypertension and secondary pulmonary hyper-
tension. Am J Physiol Heart Circ Physiol 298: H1235–1248.
11. Selman M, Pardo A, Barrera L, Estrada A, Watson SR, et al. (2006) Gene
expression profiles distinguish idiopathic pulmonary fibrosis from hypersensitiv-
ity pneumonitis. Am J Respir Crit Care Med 173: 188–198.
12. Selman M, Carrillo G, Estrada A, Mejia M, Becerril C, et al. (2007) Accelerated
variant of idiopathic pulmonary fibrosis: clinical behavior and gene expression
pattern. PLoS One 2: e482.
13. Rosas IO, Richards TJ, Konishi K, Zhang Y, Gibson K, et al. (2008) MMP1
and MMP7 as potential peripheral blood biomarkers in idiopathic pulmonary
fibrosis. PLoS Med 5: e93.
14. Konishi K, Gibson KF, Lindell KO, Richards TJ, Zhang Y, et al. (2009) Gene
expression profiles of acute exacerbations of idiopathic pulmonary fibrosis.
Am J Respir Crit Care Med 180: 167–175.
15. Scotton CJ, Krupiczojc MA, Konigshoff M, Mercer PF, Lee YC, et al. (2009)
Increased local expression of coagulation factor X contributes to the fibrotic
response in human and murine lung injury. J Clin Invest 119: 2550–2563.
16. Thannickal VJ, Horowitz JC (2006) Evolving concepts of apoptosis in idiopathic
pulmonary fibrosis. Proc Am Thorac Soc 3: 350–356.
17. Fattman CL (2008) Apoptosis in pulmonary fibrosis: too much or not enough?
Antioxid Redox Signal 10: 379–385.
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org8April 2012 | Volume 7 | Issue 4 | e33770
18. Hecker L, Vittal R, Jones T, Jagirdar R, Luckhardt TR, et al. (2009) NADPH
oxidase-4 mediates myofibroblast activation and fibrogenic responses to lung
injury. Nat Med 15: 1077–1081.
19. Kim KK, Kugler MC, Wolters PJ, Robillard L, Galvez MG, et al. (2006)
Alveolar epithelial cell mesenchymal transition develops in vivo during
pulmonary fibrosis and is regulated by the extracellular matrix. Proc Natl Acad
Sci U S A 103: 13180–13185.
20. Willis BC, Borok Z (2007) TGF-beta-induced EMT: mechanisms and
implications for fibrotic lung disease. Am J Physiol Lung Cell Mol Physiol
21. Kim KK, Wei Y, Szekeres C, Kugler MC, Wolters PJ, et al. (2009) Epithelial cell
alpha3beta1 integrin links beta-catenin and Smad signaling to promote
myofibroblast formation and pulmonary fibrosis. J Clin Invest 119: 213–224.
22. Selman M, Pardo A, Kaminski N (2008) Idiopathic pulmonary fibrosis: aberrant
recapitulation of developmental programs? PLoS Med 5: e62.
23. Vuga LJ, Ben-Yehudah A, Kovkarova-Naumovski E, Oriss T, Gibson KF, et al.
(2009) WNT5A is a regulator of fibroblast proliferation and resistance to
apoptosis. Am J Respir Cell Mol Biol 41: 583–589.
24. Schwartz DA (2010) Epigenetics and environmental lung disease. Proc Am
Thorac 7(2): 123–125.
25. Suzuki MM, Bird A (2008) DNA methylation landscapes: provocative insights
from epigenomics. Nat Rev Genet 9: 465–476.
26. Toyota M, Suzuki H, Yamashita T, Hirata K, Imai K, et al. (2009) Cancer
epigenomics: implications of DNA methylation in personalized cancer therapy.
Cancer Sci 100: 787–791.
27. Waterland RA, Jirtle RL (2004) Early nutrition, epigenetic changes at
transposons and imprinted genes, and enhanced susceptibility to adult chronic
diseases. Nutrition 20: 63–68.
28. Dolinoy DC, Jirtle RL (2008) Environmental epigenomics in human health and
disease. Environ Mol Mutagen 49: 4–8.
29. Jones PA, Takai D (2001) The role of DNA methylation in mammalian
epigenetics. Science 293: 1068–1070.
30. Slotkin RK, Martienssen R (2007) Transposable elements and the epigenetic
regulation of the genome. Nat Rev Genet 8: 272–285.
31. Jones PA, Baylin SB (2002) The fundamental role of epigenetic events in cancer.
Nat Rev Genet 3: 415–428.
32. Li E, Beard C, Jaenisch R (1993) Role for DNA methylation in genomic
imprinting. Nature 366: 362–365.
33. Ting AH, McGarvey KM, Baylin SB (2006) The cancer epigenome–
components and functional correlates. Genes Dev 20: 3215–3231.
34. Pfeifer GP, Rauch TA (2009) DNA methylation patterns in lung carcinomas.
Semin Cancer Biol 19: 181–187.
35. Rauch T, Wang Z, Zhang X, Zhong X, Wu X, et al. (2007) Homeobox gene
methylation in lung cancer studied by genome-wide analysis with a microarray-
based methylated CpG island recovery assay. Proc Natl Acad Sci U S A 104:
36. Rauch TA, Zhong X, Wu X, Wang M, Kernstine KH, et al. (2008) High-
resolution mapping of DNA hypermethylation and hypomethylation in lung
cancer. Proc Natl Acad Sci U S A 105: 252–257.
37. Vancheri C, Failla M, Crimi N, Raghu G (2010) Idiopathic pulmonary fibrosis:
a disease with similarities and links to cancer biology. Eur Respir J 35: 496–504.
38. Lababede O, Meziane MA, Rice TW (1999) TNM staging of lung cancer: a
quick reference chart. Chest 115: 233–235.
39. Mohn F, Weber M, Schubeler D, Roloff TC (2009) Methylated DNA
immunoprecipitation (MeDIP). Methods Mol Biol 507: 55–64.
40. Weber M, Davies JJ, Wittig D, Oakeley EJ, Haase M, et al. (2005)
Chromosome-wide and promoter-specific analyses identify sites of differential
DNA methylation in normal and transformed human cells. Nat Genet 37:
41. Straussman R, Nejman D, Roberts D, Steinfeld I, Blum B, et al. (2009)
Developmental programming of CpG island methylation profiles in the human
genome. Nat Struct Mol Biol 16: 564–571.
42. Yamashita S, Hosoya K, Gyobu K, Takeshima H, Ushijima T (2009)
Development of a novel output value for quantitative assessment in methylated
DNA immunoprecipitation-CpG island microarray analysis. DNA Res 16:
43. Ehrich M, Nelson MR, Stanssens P, Zabeau M, Liloglou T, et al. (2005)
Quantitative high-throughput analysis of DNA methylation patterns by base-
specific cleavage and mass spectrometry. Proc Natl Acad Sci U S A 102:
44. Chomczynski P, Sacchi N (1987) Single-step method of RNA isolation by acid
guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:
45. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using
real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25:
46. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I (2001) Controlling the false
discovery rate in behavior genetics research. Behav Brain Res 125: 279–284.
47. Segal E, Friedman N, Koller D, Regev A (2004) A module map showing
conditional activity of expression modules in cancer. Nat Genet 36: 1090–1098.
48. Huang da W, Sherman BT, Lempicki RA (2009) Systematic and integrative
analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 4:
49. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, et al. (2002) The
human genome browser at UCSC. Genome Res 12: 996–1006.
50. Jaenisch R, Bird A (2003) Epigenetic regulation of gene expression: how the
genome integrates intrinsic and environmental signals. Nat Genet 33 Suppl:
51. Weisenberger DJ, Campan M, Long TI, Kim M, Woods C, et al. (2005)
Analysis of repetitive element DNA methylation by MethyLight. Nucleic Acids
Res 33: 6823–6836.
52. Irizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, et al. (2009) The
human colon cancer methylome shows similar hypo- and hypermethylation at
conserved tissue-specific CpG island shores. Nat Genet 41: 178–186.
53. Klug M, Heinz S, Gebhard C, Schwarzfischer L, Krause SW, et al. (2010)
Active DNA demethylation in human postmitotic cells correlates with activating
histone modifications, but not transcription levels. Genome Biol 11: R63.
54. Choi JK (2010) Contrasting chromatin organization of CpG islands and exons in
the human genome. Genome Biol 11: R70.
55. Dudziec E, Miah S, Choudhry HM, Owen HC, Blizard S, et al. (2011)
Hypermethylation of CpG Islands and Shores around Specific MicroRNAs and
Mirtrons Is Associated with the Phenotype and Presence of Bladder Cancer.
Clin Cancer Res 17: 1287–1296.
56. Kim JH, Dhanasekaran SM, Prensner JR, Cao X, Robinson D, et al. (2011)
Deep sequencing reveals distinct patterns of DNA methylation in prostate
cancer. Genome Res 21: 1028–1041.
57. Sanjo H, Kawai T, Akira S (1998) DRAKs, novel serine/threonine kinases
related to death-associated protein kinase that trigger apoptosis. J Biol Chem
58. Pang M, Zhuang S (2010) Histone deacetylase: a potential therapeutic target for
fibrotic disorders. J Pharmacol Exp Ther 335: 266–272.
59. Liu T, Hu B, Choi YY, Chung M, Ullenbruch M, et al. (2009) Notch1 signaling
in FIZZ1 induction of myofibroblast differentiation. Am J Pathol 174:
60. Boon K, Bailey NW, Yang J, Steel MP, Groshong S, et al. (2009) Molecular
phenotypes distinguish patients with relatively stable from progressive idiopathic
pulmonary fibrosis (IPF). PLoS One 4: e5134.
61. Keane MP, Belperio JA, Moore TA, Moore BB, Arenberg DA, et al. (1999)
Neutralization of the CXC chemokine, macrophage inflammatory protein-2,
attenuates bleomycin-induced pulmonary fibrosis. J Immunol 162: 5511–5518.
62. Ozawa Y, Suda T, Naito T, Enomoto N, Hashimoto D, et al. (2009) Cumulative
incidence of and predictive factors for lung cancer in IPF. Respirology 14:
63. Le Jeune I, Gribbin J, West J, Smith C, Cullinan P, et al. (2007) The incidence of
cancer in patients with idiopathic pulmonary fibrosis and sarcoidosis in the UK.
Respiratory medicine 101: 2534–2540.
64. Wilson AS, Power BE, Molloy PL (2007) DNA hypomethylation and human
diseases. Biochim Biophys Acta 1775: 138–162.
65. Chalitchagorn K, Shuangshoti S, Hourpai N, Kongruttanachok N,
Tangkijvanich P, et al. (2004) Distinctive pattern of LINE-1 methylation level
in normal tissues and the association with carcinogenesis. Oncogene 23:
66. Faulkner GJ, Kimura Y, Daub CO, Wani S, Plessy C, et al. (2009) The
regulated retrotransposon transcriptome of mammalian cells. Nat Genet 41:
67. Estecio MR, Gharibyan V, Shen L, Ibrahim AE, Doshi K, et al. (2007) LINE-1
hypomethylation in cancer is highly variable and inversely correlated with
microsatellite instability. PLoS One 2: e399.
68. Han JS, Szak ST, Boeke JD (2004) Transcriptional disruption by the L1
retrotransposon and implications for mammalian transcriptomes. Nature 429:
69. Howard G, Eiges R, Gaudet F, Jaenisch R, Eden A (2008) Activation and
transposition of endogenous retroviral elements in hypomethylation induced
tumors in mice. Oncogene 27: 404–408.
70. Huang SK, Fisher AS, Scruggs AM, White ES, Hogaboam CM, et al. (2010)
Hypermethylation of PTGER2 confers prostaglandin E2 resistance in fibrotic
fibroblasts from humans and mice. Am J Pathol 177: 2245–2255.
71. Sanders YY, Pardo A, Selman M, Nuovo GJ, Tollefsbol TO, et al. (2008) Thy-1
promoter hypermethylation: a novel epigenetic pathogenic mechanism in
pulmonary fibrosis. Am J Respir Cell Mol Biol 39: 610–618.
Methylation in Lung Fibrosis
PLoS ONE | www.plosone.org9 April 2012 | Volume 7 | Issue 4 | e33770