Download - Research Article Comprehensive Analysis of Transcriptome ...downloads.hindawi.com/journals/ijg/2015/206937.pdf · Comprehensive Analysis of Transcriptome Sequencing Data in ... mentary

Research ArticleComprehensive Analysis of Transcriptome Sequencing Data inthe Lung Tissues of COPD Subjects

Woo Jin Kim1 Jae Hyun Lim2 Jae Seung Lee3 Sang-Do Lee3

Ju Han Kim2 and Yeon-Mok Oh3

1Department of Internal Medicine and Environmental Health Center Kangwon National University Hospital School of MedicineKangwon National University Chuncheon 200-722 Republic of Korea2Seoul National University Biomedical Informatics and Systems Biomedical Informatics Research Center Division of BiomedicalInformatics Seoul National University College of Medicine Seoul 110-799 Republic of Korea3Department of Pulmonary and Critical Care Medicine and Clinical Research Center for Chronic Obstructive Airway DiseasesAsan Medical Center University of Ulsan College of Medicine Seoul 138-736 Republic of Korea

Correspondence should be addressed to Yeon-Mok Oh ymoh55amcseoulkr

Received 30 May 2014 Accepted 12 January 2015

Academic Editor Brian Wigdahl

Copyright copy 2015 Woo Jin Kim et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Background and Objectives Chronic obstructive pulmonary disease (COPD) is a complex disease characterized by airflowlimitation Although airway inflammation and oxidative stress are known to be important in the pathogenesis of COPD themechanismunderlying airflow obstruction is not fully understood Gene expression profiling of lung tissuewas performed to definethe molecular pathways that are dysregulated in COPD Methods RNA was isolated from lung tissues obtained from 98 subjectswith COPD and 91 control subjects with normal spirometry The RNA samples were processed with RNA-seq using the HiSeq2000 system Genes expressed differentially between the two groups were identified using Studentrsquos t-test Results After filteringfor genes with zero counts and noncoding genes 16676 genes were evaluated A total of 2312 genes were differentially expressedbetween the lung tissues of COPD and control subjects (false discovery rate corrected 119902 lt 001) The expression of genes related tooxidative phosphorylation and protein catabolism was reduced and genes related to chromatin modification were dysregulated inlung tissues of COPD subjects Conclusions Oxidative phosphorylation protein degradation and chromatin modification were themost dysregulated pathways in the lung tissues of COPD subjects These findings may have clinical and mechanistic implicationsin COPD

1 Introduction

Chronic obstructive pulmonary disease (COPD) is character-ized by chronic airflow limitation that is not fully reversibleand is highly prevalent worldwide [1] Although cigarettesmoking is a major risk factor for COPD the mechanism bywhich inhaled smoke contributes to airflow obstruction is notfully understood Current theories for this include persistentairway inflammation that is modified by oxidative stressexcess proteinases autoimmunity and apoptosis of alveolarcells [2]

Gene expression studies of diseased lungs can generatehigh-throughput results to shed light on the molecular

processes underlying COPD pathogenesis Whole-genomeexpression of COPD in humans has been studied usingairway epithelium and resected lung tissue in several groups[3ndash9] These studies mostly used microarrays to profile geneexpression patterns in COPD Next-generation sequencingtechnology was recently applied to transcriptomics RNA-seqtechnology provides read counts of RNA fragments in eachgene [10] Background and cross-hybridization are not issuesin RNA-seq and the technology can quantify both lowly andhighly abundant transcripts [11] In addition this methodcan provide information about splice variants allele-specificexpression and fusion transcripts [12] RNA-seq data of

Hindawi Publishing CorporationInternational Journal of GenomicsVolume 2015 Article ID 206937 9 pageshttpdxdoiorg1011552015206937

2 International Journal of Genomics

airways was recently published [13 14] however the numberof subjects in these studies was relatively small

We performed gene expression profiling using RNA-seqof lung tissues resected from large numbers of subjects withCOPD or with control subjects to better understand themolecular mechanisms responsible for the pathogenesis ofCOPD

2 Methods

21 Study Populations Subjects were patients who requiredresection for lung cancer and who were registered in theAsan Biobank from January 2008 to November 2011 Theinclusion criteria were a postbronchodilator FEV

1FVC ratio

(ratio of forced expiratory volume in the first second to forcedvital capacity) of less than 07 for the COPD group andnormal spirometry for the control group in accordance withAmerican Thoracic SocietyEuropean Respiratory Societycriteria [15] This study was approved by the institutionalreview board of Asan Medical Center (2011-0711) and writteninformed consent was obtained from all patients

22 RNA Preparation and Sequencing Total RNA was iso-lated from apparently normal fresh frozen lung tissue thatwas remote from the lung cancer RNA integrity was assessedusing an Agilent Bioanalyzer and RNA purity was assessedusing a NanoDrop spectrophotometer

One 120583g of total RNAwas used to generate cDNA librariesusing the TruSeq RNA library kit The protocol consistedof poly A-selected RNA extraction RNA fragmentationreverse transcription using random hexamer primers and100 bp paired-end sequencing using the Illumina HiSeq 2000system All data have been deposited in the NCBI GeneExpression Omnibus (GEO) public repository and can beaccessed through the accession number GSE57148

23 Quality Control and Data Management For quality con-trol read quality was verified using FastQC and read align-ment was verified using Picard Differential gene expression(DEG) analysis was performed using TopHat and Cufflinkssoftware [16] To estimate expression levels the RNA-seqreads were mapped to the human genome using TopHat(version 141) [17] and quantified using Cufflinks software200 [18] Cufflinks software was run with the UCSC hg19human genome and transcriptome references The numbersof isoform and gene transcripts were calculated and therelative abundance of transcripts was measured in fragmentsper kilobase of exon per million fragments mapped (FPKM)

Expression levels were extracted as a FPKMvalue for eachgene of each sample using Cufflinks software Genes withFPKM values of 0 across all samples were excluded Filtereddata were subject to upper quantile normalization Statisticalsignificance was determined using Studentrsquos 119905-test The falsediscovery rate (FDR) was controlled by adjusting 119875 valuesusing the Benjamini-Hochberg algorithm The analysis stepsused are summarized in Figure 1

To investigate whether DEGs are related to clinicalphenotypes we performed a linear regression analysis for

Reconstructed transcriptome

FPKM table

Expression data

Result

Read annotation

Read mapping

Further analysis

Normalization and verification

Read quantification

FPKM fragments per kilobase of exon per million fragments mapped

Aligned BAM file

Raw data(fastq file) Quality control

(fast QC)

Figure 1 Schematic overview of the transcript analysis of RNA-seqexperiment Briefly we used TopHat to align raw fastq files and usedCufflinks to read annotation and quantification FastQCwas used tocheck read quality

5 clinical phenotypes with respect to its gene expressionWe considered each clinical phenotype as the responder forregression and each gene expression as the predictor

24 Quantitative Real-Time PCR (qRT-PCR) Several geneswhose expression level was found to be related to COPDstatus by RNA-seq were validated using TaqMan real-timePCR The results were normalized to GAPDH Ct valuesPrimer sequences for the genes of interest are given in Table 2

25 Pathway Analysis Functional enrichment analysis wasperformed using gene set enrichment analysis (version 208)which combines information from previously defined genesets obtained from the Molecular Signature Database (ver-sion 31) Biological gene functional annotation analysis wasperformed using DAVID (version 67) with a list of DEGsBioLattice (version 11)was used to annotate coexpressed genegroups to GO biological process terms and visualize theirrelations [19]

26 Differential Alternative Splicing To detect differentialalternative splicing between the two groups subjects fromeach group were evaluated using a multivariate Bayesianalgorithm called ldquomultivariate analysis of transcript splicingrdquo[20] Differential alternative splicing including exon skip-ping mutually exclusive exons alternative 51015840 or 31015840 splice siteusage and intron retention was investigated Exon usage ofVIM between two groups was visualized using DEXSeq [21]

27 Connectivity Map A connectivity map [22] was usedto identify potential drugs that might reverse the gene

International Journal of Genomics 3

Table 1 Demographics of COPD subjects and control subjects withnormal lung function

COPD subjects Control subjects 119875 valueMale 119899 () 98 1000 91 1000Age years 675 plusmn 64 609 plusmn 95 lt00001Smoking (py) 480 plusmn 220 352 plusmn 172 lt00001FEV1 719 plusmn 134 910 plusmn 124 lt00001FEV1FVC 571 plusmn 78 748 plusmn 43 lt00001DLCO 774 plusmn 138 928 plusmn 132 lt00001COPD chronic obstructive pulmonary disease DLCO diffusing capacity ofthe lung for CO2 pack-years FEV1 forced expiratory volume in 1 secondUnless otherwise stated the mean plusmn standard deviation is shown

expression pattern associated with the pathogenesis of earlyCOPDThe connectivity map is a collection of genome-widetranscriptional expression data from cultured human cellstreated with bioactive small moleculesThe basic assumptionof the connectivity map is that transcriptional perturbationcan occur or be treated by certain drugs that intrigue similarchanges

3 Results

31 Demographic Characteristics The demographic charac-teristics of 98 subjects with COPD and 91 control subjects areshown inTable 1 All subjects weremale and themean age andthe mean pack years of cigarette smoking history were higherin the COPD group than in the control group As expectedpulmonary function was significantly lower in the COPDgroup than in the control subjects Most of COPD subjectswere in early stage In COPD group 28 subjects took inhaledcorticosteroid 40 subjects took tiotropium and 22 subjectstook short-acting beta-agonist None in control group tookbronchodilator

32 Quality Control and DEGs The total number of readsproduced from each sample was 38742474 plusmn 7332014reads (mean plusmn standard deviation) The difference in thenumber of reads betweenCOPD samples and control sampleswas not statistically significant After read alignment withTopHat and read quantification with Cufflinks using UCSChg19 transcriptome reference a total of 189 samples and23146 genes were analyzed A total of 248 genes had zeroFPKM values in all samples After filtering for genes withzero counts in whole samples noncoding genes and lowvariance genes 16676 genes were analyzed Out of thesegenes 2312 genes were differentially expressed between thetwo groups (FDR corrected 119902 lt 001) (Table 3) Therewere many overlaps between 119905-test and EdgeR (see supple-mentary Table 1 in the Supplementary Material availableonline at httpdxdoiorg1011552015206937) Regressionanalysis of the DEGs with clinical phenotypes revealed thatgenes were more associated with FEV

1(forced expiratory

volume in the first second) and FEV1FVC ratio than with

age and smoking history (Figure 2)

Age Cigar FVC

000

005

010

015

020

025

030

Adju

sted R

val

ues

FEV1 FEV1FVC

Figure 2 Regression analysis of the differentially expressed geneswith clinical phenotypes Genes were more associated with FEV

1

(forced expiratory volume in 1 second) and FEV1FVC ratio (ratio of

forced expiratory volume in 1 second to forced vital capacity) thanwith age and smoking history

RNASeqqPCR

FGG MCL1 PDE4A S100A6 SERPINE1 SFTPC TMSB4X

3

25

2

1

15

0

minus05

minus1

minus15

05

minus2Log 2

(fold

chan

ge C

OPD

nor

mal

)

Figure 3 RNA-seq and quantitative real-time PCR results of sevengenes Results of fold change by two methods are shown

33 Validation To validate the RNA-seq results we per-formed qRT-PCR of the FGG MCL1 PDE4A S100A6 SER-PINE1 SFTPC and TMSB4X genes using the same RNAsamples that were used for RNA-seq We used a subset of 156samples out of 189 samples for validation The RNA-seq andqRT-PCR results were in good agreement (Figure 3)

34 Clustering and Gene Functions A heat map shows thegenes that were differentially expressed between the twogroups (Figure 4) DAVID revealed that the expression ofgenes involved in protein catabolism oxidative phospho-rylation and chromatin modification differed most signif-icantly between the two groups (Table 4) Expression ofgenes encoding proteasome components including PSMA2PSMB1 PSMC5 and PSMD4 was lower in the COPD groupthan in the control group Gene set enrichment analysisrevealed that the most significantly downregulated pathways


Table 2 Primers for mRNA expression profiling

Gene symbol Assay ID Context sequenceTMSB4X Hs03407480 gH GGTGAAGGAAGAAGTGGGGTGGAAGMCL1 Hs01050896 m1 TAAACAAAGAGGCTGGGATGGGTTTSFTPC Hs00161628 m1 AGAGCCCGCCGGACTACTCCGCAGCS100A6 Hs00170953 m1 CCTCCCTACCGCTCCAAGCCCAGCCFGG Hs00241037 m1 TGGAGTTTATTACCAAGGTGGCACTSERPINE1 Hs01126607 g1 CAACCCCACAGGAACAGTCCTTTTCPDE4A Hs01102342 mH ACCGCATCCAGGTCCTCCGGAACAT

Table 3 Top 20 genes differentially expressed between COPD subjects and subjects with normal lung function

Gene symbol Gene function Fold change119875 value 119902 value Expression levels

log2(COPDNormal) (log

2(FPKM))

RAD54L2 Androgen receptor-interacting protein 070 123E minus 24 205E minus 20 413

UBR4 Ubiquitin protein ligase E3 componentn-recognin 4 046 624E minus 24 104E minus 19 971

KPNA6 Karyopherin alpha 6 027 225E minus 21 375E minus 17 1103VPS28 Vacuolar protein minus051 111E minus 20 186E minus 16 6090STRA13 Stimulated by retinoic acid minus078 229E minus 20 382E minus 16 2196SPEN Spen family transcriptional repressor 047 268E minus 20 447E minus 16 999

HERC2 HECT and RLD domain containing E3ubiquitin protein ligase 2 042 373E minus 20 622E minus 16 605

GTF3C3 General transcription factor IIIC 058 653E minus 20 109E minus 15 952TCF20 Transcription factor 20 (AR1) 035 722E minus 20 120E minus 15 649MRPL21 Mitochondrial ribosomal protein L21 minus055 191E minus 19 319E minus 15 2224

COX6A1 Cytochrome c oxidase subunit VIapolypeptide 1 minus066 217E minus 19 363E minus 15 23771

ZZEF1 Zinc finger ZZ-type with EF-hand domain 1 030 236E minus 19 394E minus 15 756STX8 Syntaxin 8 minus074 300E minus 19 500E minus 15 2976UBAP2L Ubiquitin associated protein 2-like 027 348E minus 19 580E minus 15 2999

TRRAP Transformationtranscriptiondomain-associated protein 040 363E minus 19 605E minus 15 496

ENTPD4 Ectonucleoside triphosphatediphosphohydrolase 4 047 374E minus 19 624E minus 15 1188

TRIM56 Tripartite motif containing 56 045 437E minus 19 730E minus 15 1009NHSL2 NHS-like 2 121 488E minus 19 814E minus 15 144SETD5 SET domain containing 5 023 517E minus 19 862E minus 15 1695

PRKAR2A Protein kinase cAMP-dependentregulatory type II alpha 079 539E minus 19 899E minus 15 833


in the COPD group were oxidative phosphorylation andbiosynthetic process (FDR 119902 lt 025)

We used 119896-means clustering to construct 30 groupsof coexpressed genes for BioLattice After matching withhypergeometric distributions to annotate those gene sets toGO terms concept lattice for coexpressed gene clusters withGO biological process terms was visualized showing thatgenes related to transcription and oxidative phosphorylationare enriched in the major clusters (Figure 5)

In the COPD group the heat map shows hierarchi-cal clustering of two subgroups using Euclidean distance

(Figure 6) Four hundred and four genes that were differ-entially expressed between these two subgroups in COPDsubjects (119902 lt 005) are enriched in the mitochondrial andsteroid biosynthesis pathway Group 1 showed tendency forlower FEV

1and lower DLCO (carbon monoxide diffusing

capacity) When comparing each group with the controlDEGs between group 1 and the controls only consisted of 18genes and their 119875 values were negligible Meanwhile DEGsbetween group 2 and the controls consisted of 4072 genes andtheir 119875 values were even higher than those of DEGs betweenthe COPD group and the controls


Table 4 Representative DAVID results of pathway that was differentially expressed between COPD and control groups

Term Count of genes involved Fold enrichment FDRGO0030529simribonucleoprotein complex 171 279 276E minus 35GO0070013simintracellular organelle lumen 346 163 216E minus 20GO0005739simmitochondrion 239 185 808E minus 20GO0006119simoxidative phosphorylation 36 305 188E minus 06GO0030163simprotein catabolic process 135 180 898E minus 09GO0006396simRNA processing 125 190 152E minus 09GO0006351simtranscription DNA-dependent 67 190 425E minus 04GO0015031simprotein transport 138 150 107E minus 03GO0016568simchromatin modification 61 185 450E minus 03GO0006511simubiquitin-dependent protein catabolic process 55 189 786E minus 03

Color keyand histogram

Value

Coun

t

0

15000

0 5

Figure 4 Heat map of RNA-seq results of lung tissues from COPD and control subjects Color column sidebar indicated status of subjectswhere red means COPD subjects and blue means control group

There was no difference in medication history betweenthe two groups

35 Isoform and Alternative Splicing Isoforms that weredifferentially expressed between the COPD and controlgroups were evaluated by Cufflinks Pathway analysis resultsof the DEIs were similar to those of the DEGs Howeveramong the DEIs 310 were not in the DEG list which areenriched in genes encoding proteins that function in celljunctions and focal adhesions The multivariate analysis oftranscript splicing program (MATS) revealed that specificalternative splicing events were significantly more commonin COPD subjects than in control subjects Five categories ofdifferentially expressed isoforms were identified by MATSskipped exon alternative 51015840 splice site alternative 31015840 splice

site mutually exclusive exons and retained introns Signif-icantly different events between the COPD group and thecontrol group are shown in the supplementary Table 2 (FDR119902 lt 001) Intron retention of HNRNPH1 and VIM occurredsignificantly more in the COPD group than in the controlgroup Mutually exclusive exons occurred more frequentlyin the COPD group than in the control group in 78 genesFigure 7 shows exon usage of VIM between the two groupsvisualized using DEXSeq [21]

36 Connectivity Map A connectivity map [22] was used toidentify potential drugs that might reverse the gene expres-sion pattern associated with the pathogenesis of early COPDGene expression changes arising from treatment with severaldrugs were negatively correlated with the expression patterns


Top

Bottom

C1819

C195

C169

27

C1724

C1415

C24

C37

C426

C816

C106

C123

C13 C11

C929

C1520

C713

C517

C623

P oxidative phosphorylationP protein targetingP transportP intracellular transportP localizationP establishment of localizationP cellular localizationP establishment of cellular localization

P electron transport

P generation of precursor metabolites and energy

P regulation of transcription DNA-dependentP regulation of metabolismP regulation of cellular metabolismP regulation of biological processP regulation of physiological process

P regulation of cellular process

P regulation of cellular physiological process

Figure 5 BioLattice analysis of RNA-seq data of lung tissues from COPD and control subjects Each node indicates significantly enrichedGO terms with 119875 values lt 001 Lines indicate significant sharing of genes within two nodes which indicates similarity of two enriched GOterms with respect to differentially expressed genes

that differ in the COPD group compared to the control groupGene expression changes arising from treatment with MG-262 (119875 = 000004) and puromycin (119875 lt 000001) were mostsignificantly negatively correlated

4 Discussion

In this study we used RNA-seq to identify genes whoseexpression differs between COPD and control subjects Intotal 2312 genes were identified with FDR 119902 lt 001 Wevalidated a subset of RNA-seq data with qRT-PCR and theresults were in good agreement

Previous studies have investigated the gene expressionprofiles of COPD patients using microarray serial analysis ofgene expression andRNA-seqMicroarray data indicates thatMICAL2 and NOTCH2 are upregulated in the resected lungtissue of COPDpatients [5] the expression of these genes washigher in the COPD group than in the control group in thepresent study The expression of genes encoding ribosomalproteins and S100A6 was lower in the COPD group than inthe control group and RNA-seq previously indicated thatthe expression of these genes is reduced in the small airwayepithelium of smokers [14] However there is little overlap

between studies in terms of the genes that are differentiallyexpressed between people with reduced lung function andthose with normal lung function [23] This may be becausedifferent methods were used or because different phenotypeswere examined Of interest 582 differentially expressed geneswere not in the Affymetrix microarray (U133a)

Previous studies of the COPD transcriptome reported ahigh degree of overlap in the biological processes affectedIn the current study the most altered pathway in COPDpatients was mitochondrial oxidative phosphorylation Theexpression of mitochondrial genes was previously shown tobe reduced in the lung tissues of COPD subjects using serialanalysis of gene expression [3] A recent report showed thatexpression of the mitochondrial membrane protein PHB1 isdownregulated in lung tissue of COPD patients suggestingthat mitochondrial stability is reduced [24] Another reportrevealed thatmitochondrialmass is reduced in airway epithe-lial cells exposed to particulate matter and this defect is alsoobserved in the daughter cells [25] One possible explanationmay be related to the fact that an increase in the CO

2level can

reduce oxidative phosphorylation [26]However consideringthat our COPD subjects were in relatively early stages of thedisease they are not expected to have a clinically meaningfulincrease in CO

2levels


minus4 0 2 4 6Value

0

10000

Color Keyand histogram

Cou

nt

Figure 6 Heat map of RNA-seq results of lung tissues from COPD subjects Hierarchical clustering of two subgroups in COPD subjects isshown

VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9

Figure 7 Exon usage of vimentin according to COPD status COPDsubjects showed more exon usage of exon 1 to exon 4

In the current study the protein catabolism pathwaywas dysregulated in the lung tissues of COPD subjectsMany genes related to the 20S proteasome including PSMA2PSMB1 PSMC5 PSMD4 and PSMD13 as well as ubiquitin

ligase complex genes including STUB1 SELS and DERL2were downregulated in COPD subjects The ubiquitination-proteasome pathway is dysregulated in COPD but the mech-anism by which this occurs is not fully understood [27] Theexpression of 26S proteasome-associated genes is lower inlung tissues of moderate COPD subjects than in those ofthe smoker control subjects with normal lung function [3]The expression and activity of the proteasome are reduced inlung tissues of COPD subjects due to dysregulation of Nrf2[28] Impairment of proteasomal activityexpression may beimportant in the pathogenesis of COPD Interestingly theproteasome inhibitor MG-262 was on top of a list of drugsthat reverse the gene expression pattern of COPD In recentexperiments inmice a proteasome inhibitor was suggested tobe a therapeutic agent for pulmonary arterial hypertensionvia inhibition of pulmonary vascular smooth muscle cellproliferation and correction of endothelial dysfunction [29]Inhibition of proteasome inhibitors can reverse diaphrag-matic function in a COPD mouse model [30]

In the current study chromatin modification genes wereupregulated in the COPD group An epigenetic mechanism isreportedly important in the pathophysiology of COPD [31]Chromatin modification is an important mechanism in epi-genetics In the current study the expression of MLL whichplays an important role in H3K4methylation [32] and CHDwhich is important for chromatin modification and openingof chromatin to allow transcription [33] was increased inCOPD subjects These changes may lead to the upregulationof transcription The expression of HDAC10 was decreased


in the lung tissue of COPD subjects while the expression ofHDAC2 was previously reported to be decreased in COPD[34] Several studies report the mechanism by which theexpression ofHDAC2 is reduced inCOPD but further studiesare required to explain how the expression of several genesrelated to chromatin modification is increased in COPD

In a recent study investigating genes associated withthe severity of emphysema the major pathways affectedwere inflammation and tissue repair [35] However theinflammation pathway was not majorly affected in COPDsubjects in the current study probably because the subjectswere at relatively early stages of the disease

One of the advantages of RNA-seq is that DEIs can beidentified Pathway analysis results of the DEIs were similarto those of the DEGs however genes encoding proteins thatfunction in cell junctions and cell migration were foundspecifically in the DEIs and not in the DEGs This suggeststhat specific isoforms of these genes may function in thelung Alternative splicing of genes is tissue-specific and hasimportant roles in development physiological responses andthe pathogenesis of diseases [36] Interestingly the geneencoding vimentin which is a structural protein had moreretained introns in the COPD group than in the controlgroup which may alter the sequence of the protein

There are several limitations of this study COPD subjectswere older and had more pack years of smoking than thecontrol group However regression analysis results showedthat most of DEGs had stronger correlation with lungfunction than age or smoking amount All subjects weresmoking or ex-smoking men The results of this study maynot be applicable to nonsmoking or female COPD subjectsNormal looking tissue adjacent to the lung cancer tissuewas used for analysis Lung tissue consists of many celltypes including macrophages epithelium and endotheliumMicrodissection of lung tissue or single cell sequencingwouldbe required to determine whether the differential expressionis altered in all lung cells or only in a specific subset of cellsFinally it is difficult to determine whether the dysregulatedpathways identified in this study are a cause or a consequenceof the pathogenesis of COPD Experiments in which theincreasedecrease of the DEGs is reversed and shown toslow disease progression are needed to confirm that thesepathways are causally involved in the pathogenesis of COPD

In conclusion reduced oxidative phosphorylation mod-ulation of protein catabolism and dysregulation of tran-scription are important molecular features of early stages ofCOPD Genes and splicing variants were identified that weredifferentially expressed between COPD subjects and controlsubjects RNA-seq was useful tool to increase understandingof the pathophysiology of COPD

Abbreviations

COPD Chronic obstructive pulmonary diseaseFPKM Fragments per kilobase per millionDEG Differentially expressed geneDEI Differentially expressed isoform

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

This study was conceived by Woo Jin Kim Ju Han Kim andYeon-Mok Oh Data were collected by Jae Seung Lee Yeon-Mok Oh and Sang Do Lee Statistical analyses were carriedout by Woo Jin Kim Jae Hyun Lim and Ju Han Kim Allauthors contributed to and approved the final draft of thepaper Woo Jin Kim and Jae Hyun Lim contributed equally tothis work Ju Han Kim and Yeon-Mok Oh acted as equivalentco-senior authors

Acknowledgments

This study was supported by the National Project for Per-sonalized Genomic Medicine (A111218-11-GM02) the KoreaHealthcare TechnologyRampDProjectMinistry forHealth andWelfare (HI13C1634) and Basic Science Research Programthrough the National Research Foundation of Korea (NRF)funded by theMinistry of Education Science andTechnology(2012-0000994)

References

[1] J Vestbo S S Hurd A G Agustı et al ldquoGlobal strategy forthe diagnosis management and prevention of chronic obstruc-tive pulmonary disease GOLD executive summaryrdquo AmericanJournal of Respiratory and Critical Care Medicine vol 187 no 4pp 347ndash365 2013

[2] M G Cosio M Saetta and A Agusti ldquoImmunologic aspectsof chronic obstructive pulmonary diseaserdquo The New EnglandJournal of Medicine vol 360 no 23 pp 2445ndash2454 2009

[3] W Ning C J Li N Kaminski et al ldquoComprehensive geneexpression profiles reveal pathways related to the pathogenesisof chronic obstructive pulmonary diseaserdquo Proceedings of theNational Academy of Sciences of the United States of Americavol 101 no 41 pp 14895ndash14900 2004

[4] H A Golpon C D Coldren M R Zamora et al ldquoEmphysemalung tissue gene expression profilingrdquo The American Journal ofRespiratory Cell and Molecular Biology vol 31 no 6 pp 595ndash600 2004

[5] A Spira J Beane V Pinto-Plata et al ldquoGene expression pro-filing of human lung tissue from smokers with severe emphy-semardquo American Journal of Respiratory Cell and Molecular Bio-logy vol 31 no 6 pp 601ndash610 2004

[6] I-MWang S Stepaniants Y Boie et al ldquoGene expression pro-filing in patients with chronic obstructive pulmonary diseaseand lung cancerrdquo The American Journal of Respiratory andCritical Care Medicine vol 177 no 4 pp 402ndash411 2008

[7] S Bhattacharya S Srisuma D L Demeo et al ldquoMolecularbiomarkers for quantitative and discrete COPD phenotypesrdquoAmerican Journal of Respiratory Cell andMolecular Biology vol40 no 3 pp 359ndash367 2009

[8] S M S Francis J E Larsen S J Pavey et al ldquoExpression pro-filing identifies genes involved in emphysema severityrdquo Respira-tory Research vol 10 article 81 2009


[9] K Steiling M van den Berge K Hijazi et al ldquoA dynamic bron-chial airway gene expression signature of chronic obstructivepulmonary disease and lung function impairmentrdquo AmericanJournal of Respiratory and Critical Care Medicine vol 187 no 9pp 933ndash942 2013

[10] A Mortazavi B A Williams K McCue L Schaeffer and BWold ldquoMapping and quantifying mammalian transcriptomesby RNA-Seqrdquo Nature Methods vol 5 no 7 pp 621ndash628 2008

[11] F Ozsolak and P M Milos ldquoRNA sequencing advances chal-lenges and opportunitiesrdquo Nature Reviews Genetics vol 12 no2 pp 87ndash98 2011

[12] V Costa M Aprile R Esposito and A Ciccodicola ldquoRNA-Seq and human complex diseases recent accomplishments andfuture perspectivesrdquo European Journal of Human Genetics vol21 no 2 pp 134ndash142 2013

[13] J Beane J Vick F Schembri et al ldquoCharacterizing the impactof smoking and lung cancer on the airway transcriptome usingRNA-SeqrdquoCancer Prevention Research vol 4 no 6 pp 803ndash8172011

[14] N RHackettMWButler R Shaykhiev et al ldquoRNA-Seq quan-tification of the human small airway epithelium transcriptomerdquoBMC Genomics vol 13 no 1 article 82 2012

[15] M R Miller J Hankinson V Brusasco et al ldquoStandardisationof spirometryrdquo European Respiratory Journal vol 26 no 2 pp319ndash338 2005

[16] C Trapnell A Roberts L Goff et al ldquoDifferential gene andtranscript expression analysis of RNA-seq experiments withTopHat and Cufflinksrdquo Nature Protocols vol 7 no 3 pp 562ndash578 2012

[17] C Trapnell L Pachter and S L Salzberg ldquoTopHat discoveringsplice junctions with RNA-Seqrdquo Bioinformatics vol 25 no 9pp 1105ndash1111 2009

[18] C Trapnell B A Williams G Pertea et al ldquoTranscript assem-bly and quantification by RNA-Seq reveals unannotated tran-scripts and isoform switching during cell differentiationrdquoNature Biotechnology vol 28 no 5 pp 511ndash515 2010

[19] J H Kim ldquoChapter 8 biological knowledge assembly andinterpretationrdquo PLoS Computational Biology vol 8 no 12Article ID e1002858 2012

[20] S Shen JW Park JHuang et al ldquoMATS a Bayesian frameworkfor flexible detection of differential alternative splicing fromRNA-Seq datardquo Nucleic Acids Research vol 40 no 8 p e612012

[21] S Anders A Reyes andWHuber ldquoDetecting differential usageof exons from RNA-seq datardquo Genome Research vol 22 no 10pp 2008ndash2017 2012

[22] J Lamb E D Crawford D Peck et al ldquoThe connectivity mapusing gene-expression signatures to connect small moleculesgenes and diseaserdquo Science vol 313 no 5795 pp 1929ndash19352006

[23] J E Zeskind M E Lenburg and A Spira ldquoTranslating theCOPD transcriptome insights into pathogenesis and tools forclinical managementrdquo Proceedings of the American ThoracicSociety vol 5 no 8 pp 834ndash841 2008

[24] N Soulitzis E Neofytou M Psarrou et al ldquoDownregulation oflung mitochondrial prohibitin in COPDrdquo Respiratory Medicinevol 106 no 7 pp 954ndash961 2012

[25] ADi Pietro G Visalli B Baluce et al ldquoMultigenerationalmito-chondrial alterations in pneumocytes exposed to oil fly ashmetalsrdquo International Journal of Hygiene and EnvironmentalHealth vol 214 no 2 pp 138ndash144 2011

[26] C U Vohwinkel E Lecuona H Sun et al ldquoElevated CO2levels

cause mitochondrial dysfunction and impair cell proliferationrdquoThe Journal of Biological Chemistry vol 286 no 43 pp 37067ndash37076 2011

[27] S Meiners and O Eickelberg ldquoWhat shall we do with the dam-aged proteins in lung disease Ask the proteasomerdquo EuropeanRespiratory Journal vol 40 no 5 pp 1260ndash1268 2012

[28] D Malhotra R Thimmulappa N Vij et al ldquoHeightened endo-plasmic reticulum stress in the lungs of patients with chronicobstructive pulmonary disease the role of Nrf2-regulatedproteasomal activityrdquo The American Journal of Respiratory andCritical Care Medicine vol 180 no 12 pp 1197ndash1207 2009

[29] S-Y Kim J-H Lee J W Huh et al ldquoBortezomib alleviatesexperimental pulmonary arterial hypertensionrdquoAmerican Jour-nal of Respiratory Cell and Molecular Biology vol 47 no 5 pp698ndash708 2012

[30] H van Hees C Ottenheijm L Ennen M Linkels RDekhuijzen and L Heunks ldquoProteasome inhibition improvesdiaphragm function in an animal model for COPDrdquo AmericanJournal of Physiology Lung Cellular and Molecular Physiologyvol 301 no 1 pp 110ndash116 2011

[31] I V Yang and D A Schwartz ldquoEpigenetic control of geneexpression in the lungrdquo American Journal of Respiratory andCritical Care Medicine vol 183 no 10 pp 1295ndash1301 2011

[32] P Wang C Lin E R Smith et al ldquoGlobal analysis of H3K4methylation defines MLL family member targets and points toa role for MLL1-mediated H3K4 methylation in the regulationof transcriptional initiation by RNA polymerase IIrdquo Molecularand Cellular Biology vol 29 no 22 pp 6074ndash6085 2009

[33] M Murawska and A Brehm ldquoCHD chromatin remodelers andthe transcription cyclerdquo Transcription vol 2 no 6 pp 244ndash2532011

[34] K ItoM ItoWM Elliott et al ldquoDecreased histone deacetylaseactivity in chronic obstructive pulmonary diseaserdquo The NewEngland Journal of Medicine vol 352 no 19 pp 1967ndash19762005

[35] J D Campbell J E McDonough J E Zeskind et al ldquoA geneexpression signature of emphysema-related lung destructionand its reversal by the tripeptide GHKrdquo Genome Medicine vol4 no 8 p 67 2012

[36] A Kalsotra and T A Cooper ldquoFunctional consequences ofdevelopmentally regulated alternative splicingrdquoNature ReviewsGenetics vol 12 no 10 pp 715ndash729 2011

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Anatomy Research International

PeptidesInternational Journal of


Hindawi Publishing Corporation httpwwwhindawicom

International Journal of

Volume 2014

Zoology


Molecular Biology International

GenomicsInternational Journal of


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


BioinformaticsAdvances in

Marine BiologyJournal of



Signal TransductionJournal of


BioMed Research International

Evolutionary BiologyInternational Journal of



Biochemistry Research International

ArchaeaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Genetics Research International


Advances in

Virolog y

Hindawi Publishing Corporationhttpwwwhindawicom

Nucleic AcidsJournal of

Volume 2014

Stem CellsInternational



Enzyme Research



Microbiology


airways was recently published [13 14] however the numberof subjects in these studies was relatively small

We performed gene expression profiling using RNA-seqof lung tissues resected from large numbers of subjects withCOPD or with control subjects to better understand themolecular mechanisms responsible for the pathogenesis ofCOPD

2 Methods

21 Study Populations Subjects were patients who requiredresection for lung cancer and who were registered in theAsan Biobank from January 2008 to November 2011 Theinclusion criteria were a postbronchodilator FEV

1FVC ratio

(ratio of forced expiratory volume in the first second to forcedvital capacity) of less than 07 for the COPD group andnormal spirometry for the control group in accordance withAmerican Thoracic SocietyEuropean Respiratory Societycriteria [15] This study was approved by the institutionalreview board of Asan Medical Center (2011-0711) and writteninformed consent was obtained from all patients

22 RNA Preparation and Sequencing Total RNA was iso-lated from apparently normal fresh frozen lung tissue thatwas remote from the lung cancer RNA integrity was assessedusing an Agilent Bioanalyzer and RNA purity was assessedusing a NanoDrop spectrophotometer

One 120583g of total RNAwas used to generate cDNA librariesusing the TruSeq RNA library kit The protocol consistedof poly A-selected RNA extraction RNA fragmentationreverse transcription using random hexamer primers and100 bp paired-end sequencing using the Illumina HiSeq 2000system All data have been deposited in the NCBI GeneExpression Omnibus (GEO) public repository and can beaccessed through the accession number GSE57148

23 Quality Control and Data Management For quality con-trol read quality was verified using FastQC and read align-ment was verified using Picard Differential gene expression(DEG) analysis was performed using TopHat and Cufflinkssoftware [16] To estimate expression levels the RNA-seqreads were mapped to the human genome using TopHat(version 141) [17] and quantified using Cufflinks software200 [18] Cufflinks software was run with the UCSC hg19human genome and transcriptome references The numbersof isoform and gene transcripts were calculated and therelative abundance of transcripts was measured in fragmentsper kilobase of exon per million fragments mapped (FPKM)

Expression levels were extracted as a FPKMvalue for eachgene of each sample using Cufflinks software Genes withFPKM values of 0 across all samples were excluded Filtereddata were subject to upper quantile normalization Statisticalsignificance was determined using Studentrsquos 119905-test The falsediscovery rate (FDR) was controlled by adjusting 119875 valuesusing the Benjamini-Hochberg algorithm The analysis stepsused are summarized in Figure 1

To investigate whether DEGs are related to clinicalphenotypes we performed a linear regression analysis for

Reconstructed transcriptome

FPKM table

Expression data

Result

Read annotation

Read mapping

Further analysis

Normalization and verification

Read quantification


Aligned BAM file

Raw data(fastq file) Quality control

(fast QC)

Figure 1 Schematic overview of the transcript analysis of RNA-seqexperiment Briefly we used TopHat to align raw fastq files and usedCufflinks to read annotation and quantification FastQCwas used tocheck read quality

5 clinical phenotypes with respect to its gene expressionWe considered each clinical phenotype as the responder forregression and each gene expression as the predictor

24 Quantitative Real-Time PCR (qRT-PCR) Several geneswhose expression level was found to be related to COPDstatus by RNA-seq were validated using TaqMan real-timePCR The results were normalized to GAPDH Ct valuesPrimer sequences for the genes of interest are given in Table 2

25 Pathway Analysis Functional enrichment analysis wasperformed using gene set enrichment analysis (version 208)which combines information from previously defined genesets obtained from the Molecular Signature Database (ver-sion 31) Biological gene functional annotation analysis wasperformed using DAVID (version 67) with a list of DEGsBioLattice (version 11)was used to annotate coexpressed genegroups to GO biological process terms and visualize theirrelations [19]

26 Differential Alternative Splicing To detect differentialalternative splicing between the two groups subjects fromeach group were evaluated using a multivariate Bayesianalgorithm called ldquomultivariate analysis of transcript splicingrdquo[20] Differential alternative splicing including exon skip-ping mutually exclusive exons alternative 51015840 or 31015840 splice siteusage and intron retention was investigated Exon usage ofVIM between two groups was visualized using DEXSeq [21]

27 Connectivity Map A connectivity map [22] was usedto identify potential drugs that might reverse the gene





3 Results



1(forced expiratory



Age Cigar FVC

000

005

010

015

020

025

030

Adju

sted R

val

ues

FEV1 FEV1FVC


1



RNASeqqPCR


3

25

2

1

15

0

minus05

minus1

minus15

05

minus2Log 2

(fold

chan

ge C

OPD

nor

mal

)










2(FPKM))























Value

Coun

t

0

15000

0 5







Top

Bottom

C1819

C195

C169

27

C1724

C1415

C24

C37

C426

C816

C106

C123

C13 C11

C929

C1520

C713

C517

C623









4 Discussion





2level can


2levels


minus4 0 2 4 6Value

0

10000


Cou

nt


VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9











Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology





3 Results



1(forced expiratory



Age Cigar FVC

000

005

010

015

020

025

030

Adju

sted R

val

ues

FEV1 FEV1FVC


1



RNASeqqPCR


3

25

2

1

15

0

minus05

minus1

minus15

05

minus2Log 2

(fold

chan

ge C

OPD

nor

mal

)










2(FPKM))























Value

Coun

t

0

15000

0 5







Top

Bottom

C1819

C195

C169

27

C1724

C1415

C24

C37

C426

C816

C106

C123

C13 C11

C929

C1520

C713

C517

C623









4 Discussion





2level can


2levels


minus4 0 2 4 6Value

0

10000


Cou

nt


VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9











Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology







2(FPKM))























Value

Coun

t

0

15000

0 5







Top

Bottom

C1819

C195

C169

27

C1724

C1415

C24

C37

C426

C816

C106

C123

C13 C11

C929

C1520

C713

C517

C623









4 Discussion





2level can


2levels


minus4 0 2 4 6Value

0

10000


Cou

nt


VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9











Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology





Value

Coun

t

0

15000

0 5







Top

Bottom

C1819

C195

C169

27

C1724

C1415

C24

C37

C426

C816

C106

C123

C13 C11

C929

C1520

C713

C517

C623









4 Discussion





2level can


2levels


minus4 0 2 4 6Value

0

10000


Cou

nt


VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9











Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


Top

Bottom

C1819

C195

C169

27

C1724

C1415

C24

C37

C426

C816

C106

C123

C13 C11

C929

C1520

C713

C517

C623









4 Discussion





2level can


2levels


minus4 0 2 4 6Value

0

10000


Cou

nt


VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9











Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology


minus4 0 2 4 6Value

0

10000


Cou

nt


VIM NA COPD NOR

10000

5000

3000

1000

2000

100

10

Fitte

d ex

pres

sion

Exon

1

Exon

2

Exon

4

Exon

3

Exon

5

Exon

6

Exon

7

Exon

8

Exon

10

Exon

9











Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology







Abbreviations






Acknowledgments


References














































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology






































Volume 2014

Zoology






















Advances in

Virolog y



Volume 2014




Enzyme Research



Microbiology