Comparative Expression Profiling in Pulmonary Fibrosis Suggests a Role of Hypoxia-inducible Factor-1α in Disease Pathogenesis

Comparative Expression Profiling in Pulmonary FibrosisSuggests a Role of Hypoxia-inducible Factor-1a inDisease Pathogenesis

Argyris Tzouvelekis1*, Vaggelis Harokopos2*, Triantafillos Paparountas2*, Nikos Oikonomou2,Aristotelis Chatziioannou3, George Vilaras4, Evangelos Tsiambas4, Andreas Karameris4, Demosthenes Bouros1,and Vassilis Aidinis2

1Department of Pneumonology, Medical School, Democritus University of Thrace, and University Hospital of Alexandroupolis, Alexandroupolis,

Greece; 2Institute of Immunology, Biomedical Sciences Research Center ‘‘Alexander Fleming’’, Athens, Greece; 3Institute of Biological Researchand Biotechnology, National Hellenic Research Foundation, Athens, Greece; and 4Department of Pathology, Veterans Administration Hospital

(N.I.M.T.S), Athens, Greece

Rationale: Despite intense research efforts, the etiology and patho-genesis of idiopathic pulmonary fibrosis remain poorly understood.Objectives: To discover novel genes and/or cellular pathways in-volved in the pathogenesis of the disease.Methods: We performed expression profiling of disease progressionin a well-characterized animal model of the disease. Differentiallyexpressed genes that were identified were compared with allpublicly available expression profiles both from human patientsand animal models. The role of hypoxia-inducible factor (HIF)-1a indisease pathogenesis was examined with a series of immunostain-ings, both in the animal model as well as in tissue microarrayscontaining tissue samples of human patients, followed by comput-erized image analysis.Measurements and Main Results: Comparative expression profilingproduced a prioritized gene list of high statistical significance, whichconsisted of the most likely disease modifiers identified so far inpulmonary fibrosis. Extending beyond target identification, a seriesof meta-analyses produced a number of biological hypotheses ondisease pathogenesis. Among them, the role of HIF-1 signaling wasfurther explored to reveal HIF-1a overexpression in the hyperplasticepithelium of fibrotic lungs, colocalized with its target genes p53and Vegf.Conclusions: Comparative expression profiling was shown to bea highly efficient method in identifying deregulated genes andpathways. Moreover, tissue microarrays and computerized imageanalysis allowed for the high-throughput and unbiased assessmentof histopathologic sections, adding substantial confidence in path-ologic evaluations. More importantly, our results suggest an earlyprimary role of HIF-1 in alveolar epithelial cell homeostasis anddisease pathogenesis, provide insights on the pathophysiologicdifferences of different interstitial pneumonias, and indicate theimportance of assessing the efficacy of pharmacologic inhibitors ofHIF-1 activity in the treatment of pulmonary fibrosis.

Keywords: idiopathic pulmonary fibrosis (IPF); expression profiling;

tissue microarrays; hypoxia-inducible factor-1a (HIF-1a)

Idiopathic interstitial pneumonias (IIPs) are a heterogeneousgroup of diseases comprising seven distinct clinical and patho-logic entities. Idiopathic pulmonary fibrosis (IPF) and crypto-genic organizing pneumonia (COP) represent two of the mostprevalent members of the disease group, with major differencesin pathogenesis, clinical course, and prognosis (1). IPF is a re-fractory and lethal IIP characterized by fibroblast proliferation,extracellular matrix deposition, and progressive lung scarring,and comprises the histopathologic pattern of usual interstitialpneumonia (UIP). The incidence of IPF is estimated at 6.8 to16.3 cases per 100,000 per year in the United States, and themean survival from the time of diagnosis is 3 to 5 yearsregardless of treatment (2–4). Although the etiology and path-ogenesis of IPF remain poorly understood, current researchsuggests that the mechanisms driving IPF reflect abnormal,deregulated wound healing in response to multiple sites ofongoing alveolar epithelial injury, involving increased activityand possibly exaggerated responses by a spectrum of proin-flammatory and profibrogenic factors (3, 5).

Expression profiling, the estimation of the expression level ofthousands of genes by DNA microarrays, is a powerful tool forbiologists, bioinformaticians, and statisticians in their attempt todecipher the complex organization of biological phenomena. Inthis context, and to identify genes and/or cellular pathways in-volved in the initiation and progression of IPF, we used thebleomycin (BLM)-induced animal model, the closest equivalentof the human disease. RNA lung samples were isolated at dif-ferent endpoints in the development of the disease and hybrid-ized to cDNA microarrays. After robust statistical selection ofdifferentially expressed genes (DEGs), results were compared

AT A GLANCE COMMENTARY

Scientific Knowledge on the Subject

Despite intense research efforts, the etiology and patho-genesis of idiopathic pulmonary fibrosis remain poorlyunderstood, which is reflected in the lack of effectivetreatment.

What This Study Adds to the Field

Our results suggest an early primary role of HIF-1 inalveolar epithelial cell homeostasis and disease pathogen-esis, provide insights on the pathophysiologic differences ofdifferent interstitial pneumonias, and indicate the impor-tance of assessing the efficacy of pharmacologic inhibitorsof HIF-1 activity in the treatment of pulmonary fibrosis.

(Received in original form May 8, 2007; accepted in final form August 29, 2007)

Supported by the Society for Respiratory Research and Treatment of Eastern

Macedonia and Thrace (D.B.), European Commission Network of Excellence

grant QLRT-CT-2001-01407 (V.A.), and Hellenic Ministry for Development grant

GSRT-PENED-136 (V.A.). A.T. is a recipient of an annual research grant in

respiratory medicine provided by GlaxoSmithKline.

*These authors contributed equally to this article.

Correspondence and requests for reprints should be addressed to Vassilis Aidinis,

Ph.D., Institute of Immunology, B.S.R.C. Alexander Fleming, 34 Fleming Street,

16672, Athens, Greece. E-mail: [email protected]

This article has an online supplement, which is accessible from this issue’s table of

contents at www.atsjournals.org

Am J Respir Crit Care Med Vol 176. pp 1108–1119, 2007

Originally Published in Press as DOI: 10.1164/rccm.200705-683OC on August 30, 2007

Internet address: www.atsjournals.org

with all publicly available microarray datasets in IPF (6–15),both from mice and humans, thus creating a unique list of likelydisease modifiers. Furthermore, gene ontology and pathwayanalysis revealed hypoxia signaling among the most statisticallyimportant deregulated pathways. Prompted by the meta-analysisresults, we investigated the role of hypoxia-inducible factor(HIF)-1a in disease pathogenesis, in the animal model as well asin human patients, to reveal an early primary role of HIF-1a inIPF development. Some of the results of these studies havebeen previously reported in the form of abstracts (16, 17).

METHODS

Animals

All mouse strains were bred and maintained in the C57/Bl6 backgroundfor over 20 generations in the animal facilities of the Biomedical Sci-ences Research Center ‘‘Alexander Fleming’’ (Athens, Greece) underspecific pathogen–free conditions, in compliance with the Declarationof Helsinki principles. Mice were housed at 20–228C, with 55 6 5%humidity, and a 12-hour light:dark cycle; food and water was givenad libitum. All experimentation was approved by an internal institu-tional review board, as well as by the Veterinary Service and FisheryDepartment of the local governmental prefecture. Pulmonary fibrosiswas induced by a single tail vein injection of BLM hydrogen chloride(100 mg/kg body weight; 1/3 of lethal dose, 50% [1/3LD50]; NipponKayaku Co. Ltd., Tokyo, Japan) to 6- to 8-week-old mice as previouslyreported in detail (18).

Expression Profiling

Total RNA from the right lobe of lung specimens was isolated byhomogenization in ice-cold TRIzol reagent (Invitrogen Life Sciences,Carlsbad, CA) followed by a single passage through an RNeasy column(QIAGEN GmbH, Hilden, Germany). Isolated total RNA was reversetranscribed with Superscript Reverse Transcriptase II (Invitrogen), andthe cDNA was indirectly labeled using the amino-allyl cDNA labelingmethod. Experimental samples were mixed with equimolar amounts ofthe baseline sample (which was used as a common reference samplethroughout) and hybridized in quadruplicates to cDNA glass micro-array slides (Riken, Yokohama, Japan) interrogating 18,816 genes.After image analysis, all microarray data were subjected to preprocess-ing, lowess normalization, centering, and/or averaging. To selectstatistically significant DEGs, and because there is no internationalconsensus on the most appropriate method for statistical selection, weused simultaneously the two most widely used methods: a parametricand a nonparametric analysis of variance (Kruskal-Wallis), using pro-prietary algorithms implemented in MATLAB (version 7.1, release 14;The MathWorks, Inc., Natick, MA). Reverse transcriptase–polymerasechain reaction (RT-PCR) gene validation was performed using MMLVreverse transcriptase and an oligo-dT(15) primer (Promega, Madison,WI). Detailed information on expression profiling, including gene on-tology and pathway analysis, are provided in the online supplement.

Human Subjects

In total, 45 newly diagnosed patients with IIPs of two differenthistopathologic patterns (IPF/UIP, and COP/organizing pneumonia[COP/OP]) were recruited in our study. The diagnosis of IIPs was basedon the consensus statement of the American Thoracic Society/Euro-pean Respiratory Society in 2002 (1, 19). Subjects were separatedaccording to the histopathologic pattern of the IIPs as shown in Table1. All patients were treatment naive at study inclusion. Paraffin-embedded surgical lung specimens (open lung biopsy or by video-assisted thoracoscopic surgery) from two different fibrotic regions ofeach individual were sampled. All patients were fully informed andsigned an informed consent form in which they agreed to the anony-mous usage of their lung samples for research purposes.

Tissue Microarrays, Immunohistochemistry, and

Computerized Image Analysis

Tissue microarrays (TMAs) were constructed from 85 tissue samplesconsisting of 45 lung specimens from two different histopathologic

patterns of IIPs and 40 control tissues derived from the normal part oflungs removed for benign lesions. After epitope demasking, TMAs wereimmunostained with a number of antibodies against HIF-1a, surfactantprotein A (SP-A), vascular endothelial growth factor (VEGF), p53, andDNA fragmentation factor (DFF). Signal intensities were quantifiedwith computerized image analysis using a semiautomated system.Statistical analysis was performed using SPSS 13.0 software (SPSS,Inc., Chicago, IL). Details on these methodologies can be found in theonline supplement.

RESULTS

Expression Profiling

To identify genes and/or cellular pathways involved in theinitiation and progression of IPF, we performed expressionprofiling of disease progression in the animal model of BLM-induced pulmonary inflammation and fibrosis (18, 20). In thismodel, and as reported previously (18), BLM administrationresults in progressive subpleural/peribronchial pulmonary in-flammation, which subsequently diffuses into the parenchyma.Inflammation is followed by the development of mainly sub-pleural and peribronchial fibrotic patches, characterized byalveolar septa thickening and focal dilation of respiratory bron-chioles and alveolar ducts. Concomitantly, collagen accumulationpeaks 23 days post–BLM injection (Figure E4 of the onlinesupplement). The model is very reproducible, using standardizedprocedures and dedicated functional readouts exhibiting minimalvariation (18). RNA lung samples were isolated at 7, 15, and 23days post–BLM administration, corresponding to the inflamma-tory, intermediate, and fibrotic phases of the disease (Figure E4).Similarly, as a baseline control, RNA lung samples were isolatedfrom littermate mice 23 days after administration of saline alone.

Equimolar amounts of purified RNA from five mice perendpoint were pooled, to minimize biological diversity, andfluorescently labeled using the amino-allyl indirect labelingmethod as described in METHODS. Identical labeled samplesfrom the same pool were mixed with the labeled commonreference sample (wt/saline) and hybridized in (technical)quadruplicates to cDNA glass microarray slides, interrogating18,816 genes. After image acquisition and analysis, microarraydata were analyzed as outlined in Figure E1, using proprietaryalgorithms implemented in MATLAB. Briefly, and as describedin detail in METHODS, after preprocessing, lowess normalizationand quality control (Figure E2), centering was applied eitherbefore or after averaging, thus producing two gene matrices.These two matrices were further analyzed with two differentstatistical selection methods, one parametric and one nonpara-metric, thus ending up with four different lists of likely DEGs.The 1,172 genes identified as differentially expressed from allmethods (having therefore a very high statistical significance

TABLE 1. DEMOGRAPHIC AND SPIROMETRIC CHARACTERISTICSOF PATIENTS WITH IPF/UIP, PATIENTS WITH COP/OP, ANDCONTROL SUBJECTS

Characteristics IPF/UIP COP/OP Normal

Number 25 20 40

Sex, male/female 19/6 14/6 30/10

Age, median yr 68 (40–75) 50 (35–65) 39 (26–60)

Smokers/nonsmokers 20/5 7/13 25/15

FVC, % pred 75 6 3 81 6 4 106 6 14

FEV1, % pred 84 6 4 86 6 3 104 6 17

KCO, % pred 63 6 5 70 6 6 92 6 8

Definition of abbreviations: COP/OP 5 cryptogenic organizing pneumonia/

organizing pneumonia; IPF/UIP 5 idiopathic pulmonary fibrosis/usual interstitial

pneumonia.

Values are expressed as mean 6 SD or median (range).

Tzouvelekis, Harokopos, Paparountas, et al.: HIF-1a in IPF Pathogenesis 1109

and a very low false discovery rate) are shown in Table E1. Thedifferential expression of a small number of genes (clu, Hba-a1,spp1, slc6a6, nish, mt1) was further confirmed with semiquan-titative RT-PCR (at three different RNA concentrations in thelinear range of the reaction) in separate pools of five experi-mental animals and their controls (Figure E3).

Comparative Expression Profiling and Meta-analysis

To validate our list of DEGs (Table E1) in a high-throughputmode, and to compare results from different animal models aswell as from human patients, we collected (through databasesearching and personal communications) all publicly availableinformation from published expression profiling datasets on IPF(6–15), each one with different levels of data quality, annotation,and availability. Mouse and human Entrez-Gene IDs for allreported DEGs from the different datasets/studies were retrievedusing the Ingenuity Pathways Analysis (IPA version 5.0; In-genuity Systems, Redwood City, CA) software. Comparisonswere performed separately for both human and mouse Entrez-Gene IDs and results were fused together to avoid exclusions dueto species nonconcordance. Strikingly, and although the com-pared data were obtained from various models and organisms(which conceptually are governed by different pathogeneticmechanisms), using different microarray platforms (containingdifferent genes) and statistical methods, we identified a largenumber of genes in common between our dataset and thepublished genes in pairwise comparisons (Table 2 and TableE2). Therefore, the combined gene list (Table E3) containing 296(nonredundant) genes (common DEGs [cDEGs]) identified asdifferentially expressed from at least two independent studies(our own and a published one) is self-validated, has a highstatistical significance, and therefore is a valuable resource oflikely disease-modifying genes. Among them, 35 genes wereidentified from three different datasets and 6 genes from fourdatasets (as highlighted in Table E3), prioritizing these geneseven further.

To prioritize the cDEGs systematically, an in-depth meta-analysis was conducted. Initially, a very extensive literaturesearch with automated text mining using the Biolab ExperimentAssistant software (Version 2.7; Biovista, Athens, Greece) andmanual (PubMed) search revealed a total of 81 genes that havebeen found to play a direct role in the development of the disease(Table E4). This set of genes was used as a training set for thesoftware application Endeavour (SymBioSys, Center for Compu-tational Systems Biology, Katholieke Universiteit Leuven, Leuven,The Netherlands), which performs computational prioritization of‘‘test genes,’’ based on a set of ‘‘training genes’’ (21). Endeavouruses nine different data sources including both vocabulary-based(e.g.,GeneOntology[GO])aswellasotherdatasources(e.g.,BLASTand microarray databases). The ranking of a test gene for a givendata source is calculated based on its similarity with the traininggenes, whereas the final prioritization is calculated based on orderstatistics of the individual rankings (21). The statistically moresignificant (according to Endeavour, P , 0.05) cDEGs are shownin Figure 1.

To get additional functional insights on a potential role ofDEGs in the development of the disease, we then examinedwhether any of these genes are tumor necrosis factor (TNF) ortransforming growth factor (TGF) targets, the major proin-flammatory or profibrogenic factor, respectively, with a definiterole in disease induction (18, 22, 23). TNF or TGF DEGs wereidentified from published microarray datasets (24–28) andcompared with IPF DEGs (Table E1; 1,172 genes). As shownin Table 2 (and Table E2 in detail), 19 and 37 DEGs,respectively, were found to be TNF or TGF targets. Remark-

ably, 14 (of 19) and 21 (of 37) TNF/TGF targets that were foundas IPF DEGs were also included in the cDEG list (Table E3),highlighting the role of TNF and TGF in disease development.Five and seven of them, respectively, are also included in thestatistically significant Endeavour-prioritized cDEG list (high-lighted in Figure 1).

Finally, in an attempt to combine expression profiling withgenetic linkage studies, IPF DEGs (Table E1) were comparedwith possible susceptibility genes from identified quantitativetrait loci for BLM-induced pulmonary fibrosis (Blmpf1 and 2;References 29–31). Eleven of 22 genes from the blmpf1/2 loci,respectively, have been identified as DEGs (highlighted inTable E1) and three of five of these were also included in thecDEG list (highlighted in Table E3).

GO and Pathway Analysis

In parallel with the statistical identification of DEGs and theirprioritization, and to (1) prove the validity and extend the utilityof the expression data analysis even further, (2) infer deregu-lated biological functions from the gene expression data, and (3)define functional criteria for further gene selection, the selectedgenes (Table E1) were annotated in the form of the GO terms,in the categories Molecular Function and Biological Process.GO term frequencies in the selected gene list were thencalculated and their statistical significance (expressed as a Pvalue) were estimated (through their hypergeometric distribu-tion) as reported previously, and as described in detail inMETHODS. As shown in Table 3, a number of well-expectedfunctions and processes were found to be deregulated duringthe pathogenesis of BLM-induced pulmonary inflammation andfibrosis, such as inflammatory response, and chemokine, cyto-kine, and growth factor activity. As anticipated, GO analysisindicated multiple levels of gene expression regulation duringpathogenesis (RNA helicase activity, transcription corepressoractivity, transcription factor binding, magnesium ion binding;RNA processing, nuclear mRNA splicing, mRNA processing).The adhesion–cytoskeleton axis was also highlighted from theanalysis, as indicated (directly or indirectly) from a number ofderegulated functions and processes (respectively: GTPaseactivity, actin binding; actin filament severing, cell matrixadhesion). Notably, oxygen transport was indicated as the mostsignificant deregulated GO function as well as GO process,indicating hypoxia as a pathogenic insult that could lead to (orexacerbate) pulmonary fibrosis.

In a similar, complementary effort, the software program IPA(Ingenuity Systems) was used for automated gene expressiondata integration in cellular canonical pathways, as these are(pre)defined and curated by IPA. DEGs (Table E1) wereexamined for their participation in IPA canonical pathways,followed by a statistical test to examine if the pathway associationcould be observed by chance alone. The statistically significant(P , 0.05) deregulated canonical pathways are shown in Table 3.Remarkably, integrin and hypoxia signaling were ranked first inthe list of statistically significant deregulated pathways, furthersupporting the GO analysis results.

Early HIF-1a Overexpression in BLM-induced

Pulmonary Inflammation and Fibrosis

To examine the role of hypoxia in the pathogenesis of pulmonaryfibrosis, as indicated by the GO/IPA analysis, we then focused onthe role of the HIF-1a, the major transcription factor thatmediates cellular responses to hypoxia (32). SemiquantitativeRT-PCR analysis indicated that the mRNA levels of Hif-1a arefound to be up-regulated upon administration of BLM and thedevelopment of pulmonary inflammation and fibrosis (Figure

1110 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 176 2007

TABLE 2. LIST OF COMPARED DATASETS

First

Author/

Reference Organism

Bleomycin

Model

Microarray Type Statistical Selection

P Value Summary

DEGs

Reported

Common

DEGs

P < 0.1

Type of

Array

Chip

Design

Sample

Size Normalization

Analysis/

Clustering

IPF

Kaminski (8) Mouse Intratracheal

instillation of

bleomycin

Oligonucleotide

/6,000 genes

Mu6500

GeneChip

array

(Affymetrix)

30 Lung

tissue

specimens

/4 replicates

Mean

hybridization

intensities

of all probe

sets on each

array were

scaled to an

arbitrary,

fixed level.

Gene

Cluster and

TreeView

programs/2

clusters

NA Global analysis

of gene expression

in PF reveals distinct

programs regulating

lung inflammation

and fibrosis.

468 46

Lemay (11) Mouse Administered

bleomycin

(osmotic

minipumps)

Oligonucleotide

/22,690 probe

sets,

12,422 genes

MOE430A

GeneChip

arrays

(Affymetrix)

94 Lung

tissue

specimens

/11 arrays

Routines from

Bioconductor

within the

R statistical

language.

Robust probe

level model.

NA <0.1 Combination of

genomic approaches

was used to identify

candidate genes for

susceptibility to

bleomycin-induced

pulmonary fibrosis.

3,304 237

Zuo (15) Human NA Oligonucleotide

/8,400 genes

Genechip

Hugene FL

array

5 IPF lung

tissue

specimens

/2

replicates

Gene expression

levels

normalized

by a scaling

factor

multiplied

to the average

of differences

of probe pairs.

Cluster and

TreeView

programs

NA Gene expression

analysis reveals

matrylisin as a

key regulator of

PF in mice and

humans.

163 6

Cosgrove (6) Human NA Oligonucleotide /

7,068 transcripts

Hu6800

GeneChips

arrays

(Affymetrix)

5 IPF lung

tissue

specimens

and 5

controls

NA Paired,

2-tailed

t tests,

nonparametric

Mann-Whitney

analyses

<0.05 Gene expression

analysis reveals

PEDF as a target

gene for TGF-b1

and assigns it a

role as a regulator

of pulmonary

angiogenesis

and an important

mediator in IPF.

11 2

Pardo (12) Human NA Oligonucleotide

/�20,000 genes

Codelink

Uniset I

slides

13 IPF

lung tissue

specimens

Codelink

expression II

analysis suite

Weakest

link models/

logistic

regression

model

<0.05 Osteopontin found

highly expressed

in IPF human

lungs using

gene expression

analysis.

Considered

a potential

target for

therapeutic

intervention

in this incurable

disease.

1,170 43

Katsuma (10) Mouse Intratracheal

instillation

of bleomycin

cDNA

/4.224 clones

Custom Made

cDNA chip

22 Lung

tissue

specimens/4

replicates

Classical

linear

regression

techniques

k-Means

cluster

analysis/

hierarchical

clustering

<0.05 Molecular monitoring

of bleomycin-induced

pulmonary fibrosis

by cDNA

microarray–based

gene expression

profiling.

87 9

TNF-related

Banno (24) Human NA Oligonucleotide

/ 12,000 genes

HGU95Av2

arrays

(Affymetrix)

Human

RNA from

TNF-a–

treated

cells and

controls/2

replicates

Array data

normalized

against

control

samples,

Genes with

both absolute

expression

levels and

P values

among all

four time

points

considered

regulated.

Paired, 2-tailed

t tests,

nonparametric

Mann-Whitney

analyses

/hierarchical

clustering

NA Expression profiling

of TNF-treated

cells identified

targets responsible

for the harmful

and beneficial

effects of TNF-a

as well as new

significant and

novel actions

of TNF-a.

219 19

(Continued )

Tzouvelekis, Harokopos, Paparountas, et al.: HIF-1a in IPF Pathogenesis 1111

E5). Of the 70 well-known HIF-1a targets (identified fromReferences 33 and 34) and the OMIM [Online MendelianInheritance in Man] and Transfac databases), 42 were includedin the microarray we used (containing 18,816 genes) and 6 ofthose were found to be statistically significantly overexpressed(Bnip3l, Flt1, Siah1, Bhlhb2, Vegfa, and Vegfc; Table E1, 1,172genes), a number much higher than the one expected by chancealone. To examine if the observed enrichment of HIF-1a targetsin the DEG list was statistically significant, we tested the nullhypothesis as described in online METHODS, where the enrich-ment of HIF-1a targets was found to be statistically significant(P 5 0.0153). The overexpression of VEGFa was furtherconfirmed with RT-PCR analysis together with two more majorHIF-1a targets (cxcl12, pgk1; which are not included in themicroarray) (Figure E5). Moreover, immunohistochemistry forHIF-1a on lung paraffin sections from BLM-treated mice (7, 15and 23 d postadministration) confirmed overexpression ofHIF-1a during the development of the disease, localizing it mainlyat the epithelium and the endothelium (Figure 2). Remarkably,both mRNA and protein levels of Hif-1a were found to be up-

regulated as early as 7 days post–BLM treatment, before anydestruction of lung architecture and consequent gas exchange prob-lems, indicating an early role of Hif-1a in disease pathogenesis.

HIF-1a and HIF Target Genes Overexpression in the

Pulmonary Epithelium of Patients with IPF/COP

To confirm the observed overexpression of HIF-1a in the animalmodel of pulmonary fibrosis, we then examined HIF-1a expres-sion in lung sections of human patients with IPF/UIP and COP/OP (Table 1). To expedite and standardize experimental proce-dures, we created two TMA blocks consisting of 125 tissue coreseach, derived from 25 IPF, 20 COP, and 40 normal lung samples.TMA blocks were immunostained with anti-HIF-1a antibodiesand analyzed quantitatively/statistically by computerized imageanalysis as described in METHODS. As shown in Figure 3 (andFigure E6), a significant expression of HIF-1a expression was ob-served in IPF and COP samples, which was almost missing fromnormal lung control samples. In IPF samples, the expression ofHIF-1a was localized almost exclusively in hyperplastic type II

TABLE 2. (CONTINUED)

First

Author/

Reference Organism

Bleomycin

Model

Microarray Type Statistical Selection

P Value Summary

DEGs

Reported

Common

DEGs

P < 0.1

Type of

Array

Chip

Design

Sample

Size Normalization

Analysis/

Clustering

TGF-related

Chambers (25) Human NA Oligonucleotide

/ 6,000 genes

Hu-GeneFL

Array

GeneChip

(Affymetrix)

Human

lung

fibroblasts/2

replicates

Gene

expression

levels

normalized

by a scaling

factor

multiplied

to the average

of differences

of probe pairs.

The data

are arranged

according to

groupings

based on

current known

biological

functions.

<0.05 Global expression

profiling of

fibroblast

responses to

transforming

growth factor-bI

reveals the

induction of ID1.

124 12

Verrecchia (27) Human NA cDNA

/ 265 genes

Atlas human

cell interaction

cDNA

expression

arrays

(Clontech)

Human

total RNA

from control

and TGF-b–

treated

fibroblasts

Array data

normalized

against control

samples and

significant

change was

set at more

than 2-fold

above

background,

4 different

time points.

NA NA Expression profiling

of TGF-b fibroblast

and application

of stringent criteria

revealed novel Smad

targets as well as

early-induced

TGF-b/Smad targets.

98 6

Zavadil (28) Human NA cDNA / 7,873

and 8,707

probes

Unique human

expressed

sequence tags

(ESTs)

5H and H1

arrays

Human RNA

samples

from

TGF-b–

stimulated

cells/3

replicates

Calibration

of the signal

and background

intensities by

a correlation

factor (0.9–1.1).

Subtracting

background

from local

signal intensities

for each channel

determined NSI.

Hierarchical

clustering

was performed

using Pearson

correlation

metric on

Cy5, Cy3

ratios for

each

transcript.

NA Gene expression

analysis of TGF-b

stimulated cells on

human ESTs spotted

on glass slides

reveals a coordinated

regulation of genetic

programs induced by

TGF-b during the

initial phase of EMT.

141 13

Renzoni (26) Human NA Oligonucleotide

/12,000 genes

Human U95Av2

chips

(Affymetrix)

6 Lung tissue

specimens

(UIP, NSIP),

3 controls

Scaling to an

average intensity

of 150 per gene.

Reproducibility

was assessed

using two

pairs of RNA

samples from

the same

control line.

NA <0.05 Identification of several

novel TGF-b targets

in lung fibroblasts by

oligonucleotide

microarray–based

gene expression

profiling. Confirmation

of the induction

of angiotensin

II receptor type 1.

151 11

Definition of abbreviations: EMT 5 endothelial mesenchymal transition; IPF 5 idiopathic pulmonary fibrosis; NSIP 5 nonspecific interstitial pneumonia; PF 5

pulmonary fibrosis; TGF 5 transforming growth factor; TNF 5 tumor necrosis factor; UIP 5 usual interstitial pneumonia.

1112 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 176 2007

alveolar epithelial cells (AECs) (as shown with double immu-nostaining with SP-A; Figure E7) overlying areas of highly pro-liferative fibroblasts, called fibroblast foci. Interestingly, and inaccordance with results from the animal model, HIF-1a was alsopresent in the alveolar epithelium lying within areas of normallung (Figure E6), implicating Hif-1a activation as an early eventin the fibrogenic cascade. In COP samples, the expression ofHIF-1a was localized both in the alveolar epithelium overlyingareas of active fibrosis called Masson bodies (MBs) as well in thefibrotic interstitium, indicating differences in the pathogeneticmechanisms of IPF and COP.

To examine if the observed HIF-1a overexpression wasfollowed by the subsequent activation of the most prominent

HIF target genes, we then examined the expression patterns ofVEGF (also identified and confirmed as cDEG; Table 3 andFigure E5, respectively) and p53. Remarkably, both VEGF andp53 were found to be overexpressed in IPF/COP, following theexact expression pattern of HIF-1a, exclusively in AECs over-lying fibroblast foci in IPF and in both AECs and MBs in COP(Figure 4 and Figure E6), as also shown with double HIF-1a/VEGF, HIF-1a/p53 immunostainings (Figure E7).

Increased Apoptosis in the Epithelium of Fibrotic Lungs

The caspase-mediated pulmonary epithelial apoptosis has beenproposed as one of the initiating insults in the pathogenesis of

Figure 1. Endeavour-

prioritized differentiallyexpressed genes, identi-

fied from at least two

independent datasets.

One and eight geneswere identified in three

and four different data-

sets, respectively. Que-

ried data source, inorder of headings on left

side of figure: Du,

Dummy model Ouzu-nis; Du, Dummy model

propector; BI, BLAST;

En; ENSEMBL EST

model; IP, INTERPRO;KE, KEGG; EX, microar-

ray expression database

1; EX, microarray ex-

pression database 2;Go, GO analysis; Mo,

transcriptional motifs

and cis-regulatory ele-ments; pV, Endeavour

P value.

Tzouvelekis, Harokopos, Paparountas, et al.: HIF-1a in IPF Pathogenesis 1113

pulmonary fibrosis, whereas the relative resistance of (myo)fi-broblasts to apoptosis has been suggested as one of the perpet-uating stimuli of the fibrotic response. To confirm whether theincreased expression of the proapoptotic tumor suppressor genep53 in the hyperplastic epithelium results in increased epithelialapoptosis, the expression pattern of the caspase-activated DFF(or caspase-activated DNase) (35) was determined (Figure 5). In-creased apoptosis was clearly noticed at the alveolar epitheliumof patients with IPF and COP compared with control subjects,confirming previous results (36). Further analyzing the regionalapoptotic profiles, we observed minimal apoptosis in areas of ac-tive fibrosis, compared with the surrounding hyperplastic epithe-lium. In addition, increased DFF expression was found withinthe fibromyxoid lesions of COP lung compared with fibroblasticfoci seen in the histopathologic pattern of UIP, indicating dis-tinct apoptotic profiles between these two disease entities.

To further analyze the distinct apoptotic profiles exhibited byAECs and fibroblasts in IPF lung, we assessed the expression ofthe antiapoptotic protein BCL2. Prominent staining of BCL2 wasprimarily observed within the fibrotic interstitium in IPF lung andespecially within areas of accumulated fibroblast-like cells (fibro-blasts and myofibroblasts) compared with the overlying hyper-plastic epithelium (Figure 5). In addition, bcl2 expression wasalmost absent in MBs of COP lung, indicating that fibroblastsderived from IPF lung, in contrast to those seen in COP lung, areresistant to apoptosis through enhanced expression of antiapop-

totic mechanisms, including bcl2, confirming previous results (36).The latter may explain differences between patients with IPF andthose with COP in disease progressiveness and treatment respon-siveness. Finally, BCL2 staining was almost absent within areas ofnormal lung and in the alveolar epithelium of control lung samples.

TABLE 3. GENE ONTOLOGY AND PATHWAY ANALYSIS OFDIFFERENTIALLY EXPRESSED GENES

Annotation P Value

GO: molecular function

Oxygen transporter activity 0.00000

Growth factor activity 0.00106

Calmodulin binding 0.00203

RNA helicase activity 0.00484

Chemokine activity 0.00705

GTPase activity 0.00707

Cytokine activity 0.01373

Nucleotide binding 0.01689

Transcription corepressor activity 0.01710

Pre-mRNA splicing factor activity 0.01772

NADH dehydrogenase activity 0.02065

Transcription factor binding 0.02974

Magnesium ion binding 0.03630

Actin binding 0.04515

GO: biological process

Oxygen transport 0.00000

Spermine metabolism 0.00025

Actin filament severing 0.00082

Inflammatory response 0.00108

Transport 0.00520

RNA processing 0.00991

Cell cycle arrest 0.00994

Nuclear mRNA splicing. via spliceosome 0.01274

Cell-matrix adhesion 0.01440

Chemotaxis 0.03215

Endocytosis 0.03790

mRNA processing 0.04773

IPA: pathway

Integrin signaling 0.00525

Hypoxia signaling 0.01698

VEGF signaling 0.01950

Citrate cycle 0.02630

p38 MAPK signaling 0.03236

ERK/MAPK signaling 0.03802

PI3K/AKT signaling 0.04786

Definition of abbreviations: GO 5 Gene Ontology; IPA 5 Ingenuity Pathways

Analysis; VEGF 5 vascular endothelial growth factor.

From Table E1 of the online supplement.

Figure 2. Increased hypoxia inducible factor (HIF)-1a expression inbleomycin (BLM)-induced pulmonary inflammation and fibrosis. (A)

Representative immunohistochemistry with an anti-HIF-1a antibody on

lung paraffin sections from BLM-treated mice (7, 15, and 23 d postadmi-nistration). (B) Computerized image analysis of immunostained sections.

1114 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 176 2007

DISCUSSION

Comparative Expression Profiling, Gene Prioritization,

and Meta-analysis

Expression profiling, the relative quantification of the expressionlevel of thousands of genes simultaneously, is one of the mostpromising approaches for understanding mechanisms of differ-entiation, development, and disease. Especially in the case ofpolygenic diseases, such as interstitial pneumonias, expressionprofiling is a prerequisite for the generation of new workinghypotheses on pathogenetic mechanisms. In this context, and toidentify genes and/or cellular pathways involved in the initiationand progression of IPF, we performed expression profiling ina well-characterized animal model. After robust data analysisand statistical selection, a large number of differential expressedgenes were identified. To extend beyond the subsequent confir-mation of differential expression for each of the reported genesand to validate our results in a high-throughput mode, the list ofDEGs was compared with all publicly available information ondifferential expression in either IPF or various animal models ofthe disease. The commonly identified genes (Table 2) have a highstatistical significance and take advantage of all accumulatedknowledge of differential expression in IPF. Identified DEGswere further prioritized according to their relationship withgenes well known to be associated with IPF, based on the

massive amount of information stored in different databases(21). The inherent statistics have identified and ranked 33 genesas highly statistically significant, representing the best candidatesidentified so far in IPF (Figure 1). Moreover, 10 of these geneswere identified as TNF or TGF targets, the major proinflamma-tory and profibrotic factors, respectively, with a definite role indisease induction (18, 22, 23).

Finally, and to follow a more ‘‘systemic’’ approach in dissect-ing IPF pathogenesis (14), we used GO and pathway analysis toidentify deregulated functions, processes, and pathways (Table 3).Beside the well-known or anticipated deregulated functions inIPF, a few biological hypotheses (Table 3) have emerged from ouranalysis. In this report, we have followed the generated hypoth-esis that hypoxia could be a pathogenic insult that could lead to(or exacerbate) pulmonary fibrosis.

A Primary Role for HIF-1a in the Pathogenesis of

Pulmonary Fibrosis

IPF is a prototype fibrotic disease involving abnormal woundhealing in response to multiple sites of ongoing alveolar epithelialinjury (36–38). Hypoxia, the lack of oxygen, can modulatealveolar epithelial cell homeostasis by promoting significant andadverse effects on epithelial function, including VEGF and sur-factant protein production, disruption of cytoskeleton integrity,

Figure 3. Increased hypoxia inducible factor

(HIF)-1a expression in the pulmonary epithe-lium of patients with idiopathic pulmonary

fibrosis (IPF) and those with cryptogenic orga-

nizing pneumonia (COP). (A) Representative

immunohistochemistry with an anti-HIF-1a

antibody on tissue microarrays containing 25

IPF, 20 COP, and 40 normal lung samples. (B)

Computerized image analysis of immunos-tained sections.

Tzouvelekis, Harokopos, Paparountas, et al.: HIF-1a in IPF Pathogenesis 1115

and the triggering of apoptosis (39). Alveolar hypoxia can alsopromote macrophage recruitment and enhanced expression ofinflammatory mediators (40). Thus, hypoxia could represent a po-tential fibrotic stimulus through induction of epithelial apoptosis,angiogenesis, and modulation of the inflammatory response.HIF-1 is recognized as a master regulator of hypoxic signalingby activating gene transcription of genes encoding proteins thatmediate the cellular adaptive response under hypoxic conditions(32, 41). The HIF-1b subunit is constitutively expressed, whereasthe HIF-1a subunit is subject to ubiquitination and proteosomaldegradation, a process that is inhibited under hypoxic conditions(42). Thus, we examined HIF-1a expression in BLM-inducedpulmonary fibrosis in serial time points following disease pro-gression. The mRNA levels of HIF-1a, as well as of HIF-1 targetgenes vegfa, cxcl12, pgk1, were found to be up-regulated uponadministration of BLM and the development of pulmonaryinflammation and fibrosis. Immunohistochemistry for HIF-1a

confirmed overexpression of HIF-1a, localizing it mainly at theepithelium. Notably, HIF-1a (and HIF-1 target genes) over-

expression was observed early in the pathologic cascade, beforeany deterioration of lung architecture and consequent gasexchange problems, indicating an early role for HIF-1 in thedevelopment of the modeled disease.

Because the BLM animal model is not fully representative ofIPF due to its self-limiting nature and rapidity of development,we then investigated HIF-1a immunolocalization in lung samplesfrom patients with IPF/UIP and COP/OP, two histopathologicpatterns of pulmonary fibrosis with different clinical course andprognosis. We used the pioneering technology of tissue micro-arrays, which allowed us the simultaneous analysis of up to 85samples in a single experiment under highly standardized con-ditions. Thus, all tissue samples were analyzed in an identical,unbiased fashion, with minimal tissue damage and precisepositioning of arrayed samples, which not only facilitates man-ual interpretation of the staining but also serves as an ideal basisfor automated analysis amenable to robust statistics. A significantexpression of HIF-1a was observed in IPF and COP samples,which was almost missing from normal lung control samples.

Figure 4. Increased expression of vascular en-

dothelial growth factor (VEGF) and p53 in

idiopathic pulmonary fibrosis (IPF) lungs. (A)Representative immunohistochemistry with an

anti-VEGF and an anti-p53 antibody on tissue

microarrays containing 25 IPF, 20 cryptogenicorganizing pneumonia (COP), and 40 normal

lung samples. (B) Computerized image analysis

of immunostained sections. AECs, alveolar epi-

thelial cells; FF, fibroblastic foci; MB, Masson’sbodies.

1116 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 176 2007

Surprisingly, HIF-1a was localized not only in the hyperplasticalveolar epithelium surrounding areas of active fibrosis but alsowithin areas of IPF lung that appear histologically normal,suggesting that HIF-1a induction is an early event in the path-ogenesis. On the contrary, HIF-1a was almost absent withinfibroblastic foci underlying hyperplastic epithelium, whereas itwas present within fibromyxoid lesions characterizing the patternof COP/OP. In addition, positive staining was localized in typeII AECs immediately adjacent to MBs in COP lung.

HIF-1a expression was colocalized with the up-regulatedexpression of VEGF, an HIF target gene and a potent inducerof angiogenesis (43). Our findings are in accordance withprevious studies showing vascular heterogeneity within thefibrotic lung and, more importantly, they implicate for the firsttime HIF-1a as a master regulator of VEGF expression infibrotic lungs. Although numerous studies have examined so farthe interplay between aberrant vascular and matrix remodeling,the relative role of angiogenesis in the initiation and/or pro-

gression of the fibrotic cascade still remains elusive and con-troversial (44). p53, a tumor suppressor gene and an HIF-1a

target, was also found to be up-regulated and colocalized withits transcription factor in areas of alveolar hyperplasia sur-rounding fibromyxoid lesions in IPF and COP biopsy samples.Upon exposure to stress, such as DNA damage or hypoxia, p53is stabilized to promote transcription of target genes regulatingcell cycle progression, apoptosis, and cellular homeostasis (45).p53 stabilization under hypoxia was shown to be HIF-1 de-pendent (46), whereas accumulated levels of p53 have beenshown to inhibit HIF-1 activity by targeting HIF-1a for murinedouble minute 2 (Mdm2)-mediated ubiquitination and protea-some degradation (47). Concerning hypoxia and apoptosis, aunifying picture is still lacking, and the impact of HIF-1a re-mains controversial (45). Nevertheless, HIF-1a has been shownto enhance apoptosis of AECs (48), whereas we showed that theexpression pattern of HIF-1a/p53 correlated with the expressionpattern of DFF, a direct indicator of DNA fragmentation and

Figure 5. Expression pattern of DNA fragmen-

tation factor (DFF) and BCL2 in idiopathic pul-monary fibrosis (IPF) lungs. (A) Representative

immunohistochemistry with an anti-DFF and an

anti-BCL2 antibody on tissue microarrays con-

taining 25 IPF, 20 cryptogenic organizing pneu-monia (COP), and 40 normal lung samples. (B)

Computerized image analysis of immunostained

sections. AECs, alveolar epithelial cells; FF, fibro-blastic foci; MB, Masson’s bodies.

Tzouvelekis, Harokopos, Paparountas, et al.: HIF-1a in IPF Pathogenesis 1117

apoptosis (35). Increased apoptosis was clearly noticed in the hy-perplastic epithelium, whereas it was almost absent within fi-broblastic foci, in the IPF lung. In line with this, fibroblastic fociexhibited enhanced expression of the antiapoptotic protein BCL2,which was almost absent in the alveolar epithelium. Our findingsfurther support the consensus notion of the ‘‘apoptotic paradox’’in IPF—apoptosis susceptibility in epithelial cells and apoptosisresistance in fibroblasts/myofibroblasts (49)—highlighting therole of the HIF-1a–p53 axis. Furthermore, the differential ex-pression of the HIF-1a–p53 apoptotic axis within fibroprolifer-ative areas in IPF and COP lung may explain the resolution oflesions in COP lung in response to corticosteroids and the main-tenance of fibroblastic foci, resulting in dismal prognosis despitetreatment, and provides a novel mechanism through whichhighly proliferative fibroblast-like cells exert their resistanceto apoptosis by showing a cell-type–specific HIF-1a activation.

The data presented clearly support for the first time a role ofHIF-1a in the pathogenesis of IPF. HIF-1a activation under thehypoxic conditions of a fibrotic lung is well expected and couldpromote perpetuation of the disease, most likely throughmodulation of preferential apoptosis of epithelial cells andsubsequent angiogenesis, as shown in this article. Most impor-tant, HIF-1a overexpression was found early in the pathogenesisof the modeled disease and it was noted at histologically normalareas of human fibrotic lungs, suggesting that HIF-1a activationcan represent an early event and a potential fibrotic stimulus.Animal model studies with epithelium-specific ablation of Hif-1a will most likely provide further mechanistic insights. HIF-1a

activation could potentially be triggered not only from thehypoxic conditions of the fibrotic lung (and perpetuate disease)but also in normoxic conditions under the influence of variousimmunomodulatory and/or inflammatory factors that have beenshown to play a role in the development of the disease. Insulingrowth factor (IGF) and TGF can synthesize HIF-1a independentof oxygen level via phosphoinositide 3-kinase (PI3K) or mitogen-activated protein kinase (MAPK) pathways (50). TNF has beenshown to induce HIF-1a protein levels (51), most likely throughnuclear factor-kB activation (52) and/or the production of re-active oxygen species (53). Therefore, pharmacologic HIF-1a

inactivation could prove to be beneficial for patients with IPF,both by inhibiting the perpetuating effects of fibrotic tissue hyp-oxia, as well as by targeting HIF-mediated primary pathogenicstimuli on alveolar epithelial cells.

Conflict of Interest Statement: A.T. is a recipient of a V15,000 respiratory researchaward provided by GlaxoSmithKline in 2005. V.H. does not have a financialrelationship with a commercial entity that has an interest in the subject of thismanuscript. T.P. does not have a financial relationship with a commercial entitythat has an interest in the subject of this manuscript. N.O. does not havea financial relationship with a commercial entity that has an interest in the subjectof this manuscript. A.C. does not have a financial relationship with a commercialentity that has an interest in the subject of this manuscript. G.V. does not havea financial relationship with a commercial entity that has an interest in the subjectof this manuscript. E.T. does not have a financial relationship with a commercialentity that has an interest in the subject of this manuscript. A.K. does not havea financial relationship with a commercial entity that has an interest in the subjectof this manuscript. D.B. does not have a financial relationship with a commercialentity that has an interest in the subject of this manuscript. V.A. does not havea financial relationship with a commercial entity that has an interest in the subjectof this manuscript.

Acknowledgment: The authors thank Dr. Y. Hayashizaki (Genome ExplorationResearch Group, RIKEN Genomic Sciences Center, RIKEN Yokohama Institute,Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa, Japan) for his generous gift ofRIKEN microarrays.

References

1. American Thoracic Society; European Respiratory Society. AmericanThoracic Society/European Respiratory Society international multi-disciplinary consensus classification of the idiopathic interstitialpneumonias. Am J Respir Crit Care Med 2002;165:277–304.

2. American Thoracic Society; European Respiratory Society. Idiopathicpulmonary fibrosis: diagnosis and treatment: international consensusstatement. Am J Respir Crit Care Med 2000;161:646–664.

3. Gross TJ, Hunninghake GW. Idiopathic pulmonary fibrosis. N Engl JMed 2001;345:517–525.

4. Raghu G, Weycker D, Edelsberg J, Bradford WZ, Oster G. Incidenceand prevalence of idiopathic pulmonary fibrosis. Am J Respir CritCare Med 2006;174:810–816.

5. Selman M, King TE, Pardo A. Idiopathic pulmonary fibrosis: prevailingand evolving hypotheses about its pathogenesis and implications fortherapy. Ann Intern Med 2001;134:136–151.

6. Cosgrove GP, Brown KK, Schiemann WP, Serls AE, Parr JE, GeraciMW, Schwarz MI, Cool CD, Worthen GS. Pigment epithelium-derived factor in idiopathic pulmonary fibrosis: a role in aberrantangiogenesis. Am J Respir Crit Care Med 2004;170:242–251.

7. Kaminski N. Microarray analysis of idiopathic pulmonary fibrosis. Am JRespir Cell Mol Biol 2003; 29(3, Suppl)S32–S36.

8. Kaminski N, Allard JD, Pittet JF, Zuo F, Griffiths MJ, Morris D,Huang X, Sheppard D, Heller RA. Global analysis of gene expressionin pulmonary fibrosis reveals distinct programs regulating lung in-flammation and fibrosis. Proc Natl Acad Sci USA 2000;97:1778–1783.

9. Kaminski N, Rosas IO. Gene expression profiling as a window intoidiopathic pulmonary fibrosis pathogenesis: can we identify the righttarget genes? Proc Am Thorac Soc 2006;3:339–344.

10. Katsuma S, Nishi K, Tanigawara K, Ikawa H, Shiojima S, Takagaki K,Kaminishi Y, Suzuki Y, Hirasawa A, Ohgi T, et al. Molecular monitoringof bleomycin-induced pulmonary fibrosis by cDNA microarray-basedgene expression profiling. Biochem Biophys Res Commun 2001;288:747–751.

11. Lemay AM, Haston CK. Bleomycin-induced pulmonary fibrosis suscep-tibility genes in AcB/BcA recombinant congenic mice. PhysiolGenomics 2005;23:54–61.

12. Pardo A, Gibson K, Cisneros J, Richards TJ, Yang Y, Becerril C,Yousem S, Herrera I, Ruiz V, Selman M, et al. Up-regulation andprofibrotic role of osteopontin in human idiopathic pulmonaryfibrosis. PLoS Med 2005;2:e251.

13. Selman M, Pardo A, Barrera L, Estrada A, Watson SR, Wilson K, Aziz N,Kaminski N, Zlotnik A. Gene expression profiles distinguish idio-pathic pulmonary fibrosis from hypersensitivity pneumonitis. AmJ Respir Crit Care Med 2006;173:188–198.

14. Studer SM, Kaminski N. Towards systems biology of human pulmonaryfibrosis. Proc Am Thorac Soc 2007;4:85–91.

15. Zuo F, Kaminski N, Eugui E, Allard J, Yakhini Z, Ben-Dor A, Lollini L,Morris D, Kim Y, DeLustro B, et al. Gene expression analysis revealsmatrilysin as a key regulator of pulmonary fibrosis in mice andhumans. Proc Natl Acad Sci USA 2002;99:6292–6297.

16. Tzouvelekis A, Karameris A, Koutsopoulos A, Triggidou R, Tsiambas E,Froudarakis M, Steiropoulos P, Pataka A, Kouliatsis G, Anevlavis S,et al. Morphometric study of HIF-1a-VEGF axis in patients withidiopathic pulmonary fibrosis with the use of tissue microarrays. EurRespir J 2006;28:338.

17. Tzouvelekis A, Tsiambas E, Karameris A, Koutsopoulos A, Triggidou R,Froudarakis M, Vilaras G, Tsoumakari D, Bouros D. Differentialexpression of hypoxia inducible factor-1a-vascular endothelial growthfactor (HIF-1a-VEGF) axis in patients with idiopathic pulmonaryfibrosis and cryptogenic organizing pneumonia: a tissue microarraystudy [abstract]. Am J Respir Crit Care Med 2007;175:A139.

18. Oikonomou N, Harokopos V, Zalevsky J, Valavanis C, Kotanidou A,Szymkowski DE, Kollias G, Aidinis V. Soluble TNF mediates the transitionfrom pulmonary inflammation to fibrosis. PLoS ONE 2006;1:e108.

19. Demedts M, Costabel U. ATS/ERS international multidisciplinary con-sensus classification of the idiopathic interstitial pneumonias. Eur RespirJ 2002;19:794–796.

20. Gharaee-Kermani M, Ullenbruch M, Phan SH. Animal models ofpulmonary fibrosis. Methods Mol Med 2005;117:251–259.

21. Aerts S, Lambrechts D, Maity S, Van Loo P, Coessens B, De Smet F,Tranchevent LC, De Moor B, Marynen P, Hassan B, et al. Gene pri-oritization through genomic data fusion. Nat Biotechnol 2006;24:537–544.

22. Kelly M, Kolb M, Bonniaud P, Gauldie J. Re-evaluation of fibrogeniccytokines in lung fibrosis. Curr Pharm Des 2003;9:39–49.

23. Sheppard D. Transforming growth factor b: a central modulator ofpulmonary and airway inflammation and fibrosis. Proc Am ThoracSoc 2006;3:413–417.

24. Banno T, Gazel A, Blumenberg M. Effects of tumor necrosis factor-alpha (TNF alpha) in epidermal keratinocytes revealed using globaltranscriptional profiling. J Biol Chem 2004;279:32633–32642.

1118 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 176 2007

25. Chambers RC, Leoni P, Kaminski N, Laurent GJ, Heller RA. Globalexpression profiling of fibroblast responses to transforming growthfactor-beta1 reveals the induction of inhibitor of differentiation-1 andprovides evidence of smooth muscle cell phenotypic switching. Am JPathol 2003;162:533–546.

26. Renzoni EA, Abraham DJ, Howat S, Shi-Wen X, Sestini P, Bou-Gharios G, Wells AU, Veeraraghavan S, Nicholson AG, DentonCP, et al. Gene expression profiling reveals novel TGFbeta targets inadult lung fibroblasts. Respir Res 2004;5:24.

27. Verrecchia F, Chu ML, Mauviel A. Identification of novel TGF-beta/Smad gene targets in dermal fibroblasts using a combined cDNAmicroarray/promoter transactivation approach. J Biol Chem 2001;276:17058–17062.

28. Zavadil J, Bitzer M, Liang D, Yang YC, Massimi A, Kneitz S, Piek E,Bottinger EP. Genetic programs of epithelial cell plasticity directedby transforming growth factor-beta. Proc Natl Acad Sci USA2001;98:6686–6691.

29. Bauer AK, Malkinson AM, Kleeberger SR. Susceptibility to neoplasticand non-neoplastic pulmonary diseases in mice: genetic similarities.Am J Physiol Lung Cell Mol Physiol 2004;287:L685–L703.

30. Haston CK, Tomko TG, Godin N, Kerckhoff L, Hallett MT. Murinecandidate bleomycin induced pulmonary fibrosis susceptibility genesidentified by gene expression and sequence analysis of linkageregions. J Med Genet 2005;42:464–473.

31. Haston CK, Wang M, Dejournett RE, Zhou X, Ni D, Gu X, King TM,Weil MM, Newman RA, Amos CI, et al. Bleomycin hydrolase anda genetic locus within the MHC affect risk for pulmonary fibrosis inmice. Hum Mol Genet 2002;11:1855–1863.

32. Semenza GL. Hypoxia-inducible factor 1: control of oxygen homeostasisin health and disease. Pediatr Res 2001;49:614–617.

33. Semenza GL. HIF-1 and human disease: one highly involved factor.Genes Dev 2000;14:1983–1991.

34. Semenza GL. Targeting HIF-1 for cancer therapy. Nat Rev Cancer2003;3:721–732.

35. Widlak P, Garrard WT. Discovery, regulation, and action of the majorapoptotic nucleases DFF40/CAD and endonuclease G. J Cell Bio-chem 2005;94:1078–1087.

36. Plataki M, Koutsopoulos AV, Darivianaki K, Delides G, Siafakas NM,Bouros D. Expression of apoptotic and antiapoptotic markers in epi-thelial cells in idiopathic pulmonary fibrosis. Chest 2005;127:266–274.

37. Selman M, Pardo A. Role of epithelial cells in idiopathic pulmonaryfibrosis: from innocent targets to serial killers. Proc Am Thorac Soc2006;3:364–372.

38. Uhal BD. Epithelial apoptosis in the initiation of lung fibrosis. EurRespir J Suppl 2003;44:7s–9s.

39. Jain M, Sznajder JI. Effects of hypoxia on the alveolar epithelium. ProcAm Thorac Soc 2005;2:202–205.

40. Madjdpour C, Jewell UR, Kneller S, Ziegler U, Schwendener R, BooyC, Klausli L, Pasch T, Schimmer RC, Beck-Schimmer B. Decreasedalveolar oxygen induces lung inflammation. Am J Physiol Lung CellMol Physiol 2003;284:L360–L367.

41. Semenza GL. Signal transduction to hypoxia-inducible factor 1. BiochemPharmacol 2002;64:993–998.

42. Semenza GL. Hypoxia-inducible factor 1: oxygen homeostasis anddisease pathophysiology. Trends Mol Med 2001;7:345–350.

43. Hirota K, Semenza GL. Regulation of angiogenesis by hypoxia-in-ducible factor 1. Crit Rev Oncol Hematol 2006;59:15–26.

44. Tzouvelekis A, Anevlavis S, Bouros D. Angiogenesis in interstitial lungdiseases: a pathogenetic hallmark or a bystander? Respir Res2006;7:82.

45. Schmid T, Zhou J, Brune B. HIF-1 and p53: communication oftranscription factors under hypoxia. J Cell Mol Med 2004;8:423–431.

46. An WG, Kanekal M, Simon MC, Maltepe E, Blagosklonny MV, NeckersLM. Stabilization of wild-type p53 by hypoxia-inducible factor 1alpha.Nature 1998;392:405–408.

47. Ravi R, Mookerjee B, Bhujwalla ZM, Sutter CH, Artemov D, Zeng Q,Dillehay LE, Madan A, Semenza GL, Bedi A. Regulation of tumorangiogenesis by p53-induced degradation of hypoxia-inducible factor1alpha. Genes Dev 2000;14:34–44.

48. Krick S, Eul BG, Hanze J, Savai R, Grimminger F, Seeger W, Rose F.Role of hypoxia-inducible factor-1a in hypoxia-induced apoptosis ofprimary alveolar epithelial type II cells. Am J Respir Cell Mol Biol2005;32:395–403.

49. Thannickal VJ, Horowitz JC. Evolving concepts of apoptosis in idio-pathic pulmonary fibrosis. Proc Am Thorac Soc 2006;3:350–356.

50. Hellwig-Burgel T, Stiehl DP, Wagner AE, Metzen E, Jelkmann W.Hyp-oxia-inducible factor-1 (HIF-1): a novel transcription factor inimmune reactions [review]. J Interferon Cytokine Res 2005;25:297–310.

51. Albina JE, Mastrofrancesco B, Vessella JA, Louis CA, Henry WL Jr,Reichner JS. HIF-1 expression in healing wounds: HIF-1alpha in-duction in primary inflammatory cells by TNF-alpha. Am J PhysiolCell Physiol 2001;281:C1971–C1977.

52. Zhou J, Schmid T, Brune B. Tumor necrosis factor-alpha causesaccumulation of a ubiquitinated form of hypoxia inducible factor-1alpha through a nuclear factor-kappaB-dependent pathway. MolBiol Cell 2003;14:2216–2225.

53. Kietzmann T, Gorlach A. Reactive oxygen species in the control ofhypoxia-inducible factor-mediated gene expression. Semin Cell DevBiol 2005;16:474–486.

Tzouvelekis, Harokopos, Paparountas, et al.: HIF-1a in IPF Pathogenesis 1119