UBC Faculty Research and Publications

A gene expression signature of emphysema-related lung destruction and its reversal by the tripeptide… Campbell, Joshua D; McDonough, John E; Zeskind, Julie E; Hackett, Tillie L; Pechkovsky, Dmitri V; Brandsma, Corry-Anke; Suzuki, Masaru; Gosselink, John V; Liu, Gang; Alekseyev, Yuriy O; Xiao, Ji; Zhang, Xiaohui; Hayashi, Shizu; Cooper, Joel D; Timens, Wim; Postma, Dirkje S; Knight, Darryl A; Lenburg, Marc E; Hogg, James C; Spira, Avrum Aug 31, 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13073_2012_Article_401.pdf [ 3.01MB ]
JSON: 52383-1.0223848.json
JSON-LD: 52383-1.0223848-ld.json
RDF/XML (Pretty): 52383-1.0223848-rdf.xml
RDF/JSON: 52383-1.0223848-rdf.json
Turtle: 52383-1.0223848-turtle.txt
N-Triples: 52383-1.0223848-rdf-ntriples.txt
Original Record: 52383-1.0223848-source.json
Full Text

Full Text

RESEARCH Open AccessA gene expression signature of emphysema-related lung destruction and its reversal by thetripeptide GHKJoshua D Campbell1,2, John E McDonough3, Julie E Zeskind1,2, Tillie L Hackett3, Dmitri V Pechkovsky3,Corry-Anke Brandsma4, Masaru Suzuki3, John V Gosselink3, Gang Liu1, Yuriy O Alekseyev5, Ji Xiao1, Xiaohui Zhang1,Shizu Hayashi3, Joel D Cooper6, Wim Timens4, Dirkje S Postma7, Darryl A Knight3, Marc E Lenburg1,2*,James C Hogg3† and Avrum Spira1,2*AbstractBackground: Chronic obstructive pulmonary disease (COPD) is a heterogeneous disease consisting of emphysema,small airway obstruction, and/or chronic bronchitis that results in significant loss of lung function over time.Methods: In order to gain insights into the molecular pathways underlying progression of emphysema andexplore computational strategies for identifying COPD therapeutics, we profiled gene expression in lung tissuesamples obtained from regions within the same lung with varying amounts of emphysematous destruction fromsmokers with COPD (8 regions × 8 lungs = 64 samples). Regional emphysema severity was quantified in eachtissue sample using the mean linear intercept (Lm) between alveolar walls from micro-CT scans.Results: We identified 127 genes whose expression levels were significantly associated with regional emphysemaseverity while controlling for gene expression differences between individuals. Genes increasing in expression withincreasing emphysematous destruction included those involved in inflammation, such as the B-cell receptorsignaling pathway, while genes decreasing in expression were enriched in tissue repair processes, including thetransforming growth factor beta (TGFb) pathway, actin organization, and integrin signaling. We found concordantdifferential expression of these emphysema severity-associated genes in four cross-sectional studies of COPD. Usingthe Connectivity Map, we identified GHK as a compound that can reverse the gene-expression signature associatedwith emphysematous destruction and induce expression patterns consistent with TGFb pathway activation.Treatment of human fibroblasts with GHK recapitulated TGFb-induced gene-expression patterns, led to theorganization of the actin cytoskeleton, and elevated the expression of integrin b1. Furthermore, addition of GHK orTGFb restored collagen I contraction and remodeling by fibroblasts derived from COPD lungs compared tofibroblasts from former smokers without COPD.Conclusions: These results demonstrate that gene-expression changes associated with regional emphysemaseverity within an individual’s lung can provide insights into emphysema pathogenesis and identify noveltherapeutic opportunities for this deadly disease. They also suggest the need for additional studies to examine themechanisms by which TGFb and GHK each reverse the gene-expression signature of emphysematous destructionand the effects of this reversal on disease progression.* Correspondence: mlenburg@bu.edu; aspira@bu.edu† Contributed equally1Division of Computational Biomedicine, Department of Medicine, BostonUniversity School of Medicine, 72 East Concord Street, Boston, MA 02118,USAFull list of author information is available at the end of the articleCampbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67© 2012 Campbell et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.BackgroundChronic obstructive pulmonary disease (COPD) is a sig-nificant public health problem worldwide and the thirdleading cause of death in the United States [1]. It is char-acterized by irreversible airflow limitation due to obstruc-tion in the small conducting airways and emphysematousdestruction of the gas exchanging tissue of the lung.Tobacco smoke is a significant risk factor for COPD andat least 25% of smokers develop this disease [2]. Currenttheories concerning disease pathogenesis include animbalance between protease and anti-protease activity,induced apoptosis of alveolar cells through deregulationof pathways involved in oxidative stress, chronic inflam-mation, and aberrant tissue remodeling that lead to thedestruction of the extracellular matrix (ECM) in the lung[3,4]. Lung repair and regeneration are potential pro-cesses to target with novel therapeutics in COPD asabnormal tissue repair by the epithelial-mesenchymaltrophic unit can result in either fibrosis or destruction ofthe ECM [5]. However, the molecular mechanismsresponsible for the pathogenesis of COPD remain poorlyunderstood.Several groups have profiled gene expression in patientswith and without COPD or between patients with varyinglevels of airflow obstruction in order to understand differ-ences in gene expression related to COPD [6-11]. Whilethese studies have provided an initial look into the COPDtranscriptome, their results primarily relied on the use oflung function tests to define the presence or degree ofCOPD. Lung function phenotypes can neither distinguishbetween obstruction in the small airways and emphysema-tous destruction of the lung parenchyma nor provideinformation about regional differences in disease severity.Recently, McDonough et al. used micro-CT scans toquantify the degree of emphysema in different regions oflungs from patients with severe COPD by measuring themean linear intercept (Lm), a morphological measurementof alveolar destruction [12]. In order to gain insights intobiological pathways associated with increasing emphysemaseverity within a patient and explore computational strate-gies for identifying COPD therapeutics, we obtained pairedsamples from eight regions at regular intervals betweenthe apex and base of each explanted lung from six patientswith severe COPD (Global Initiative for Chronic Obstruc-tive Lung Disease (GOLD) stage IV) and two donor lungs.The degree of emphysematous destruction was quantifiedin one tissue sample from each region by Lm, while geneexpression was profiled in the adjacent tissue sample fromthe same region.We identified a number of genes whose expression isassociated with increasing emphysematous destruction andfound that pathways enriched among these genes wereinvolved in the immune response and tissue remodeling.Using the Connectivity Map (CMap) [13], we found thatthe tripeptide Gly-His-Lys (GHK) was able to reverse theaberrant patterns of gene expression associated withincreasing emphysema severity and induce patterns of geneexpression consistent with transforming growth factor beta(TGFb) pathway activation. Furthermore, we showed thatby treating distal lung fibroblasts from COPD patients withGHK, we can restore normal contractile function throughre-organization of the actin cytoskeleton and up-regulationof integrin-b1. These data further support the potential ofGHK as a therapeutic in the treatment of emphysema.Materials and methodsSample acquisition and processingSingle lungs (n = 6) were removed from patients treatedfor severe COPD by double lung transplantation at theUniversity of Pennsylvania. Donor lungs (n = 2) forwhich no suitable recipient was identified were releasedfor research use from the Gift of Life Organ Procure-ment Organization in Philadelphia. This study wasapproved by the institutional review boards and con-forms to the Helsinki Declaration. Written informedconsent for use of these specimens and the relevantclinical and radiological data required for this researchwere obtained from each patient prior to surgery andfrom the next of kin of the persons whose donated lungwas released for research. Each lung was removed fromthe thorax, cooled to 1.6°C, and transported to thelaboratory where the bronchial stump was cannulated[14]. The lung was then inflated using a compressed airsource attached to an underwater seal to slowly increasetranspulmonary pressure (PL) from 0 to 30 cmH2O.The specimen was then held at a transpulmonary pres-sure of 10 cmH2 Owhile frozen by liquid nitrogen vapor(-130°C). The frozen specimen had a multidetector CTscan followed by being cut into 2-cm thick slices in thesame plane as the CT scan. Tissue samples were col-lected using a sharpened steel cylinder (cork bore dia-meter of 14 mm). One sample from a cluster of fourcore samples of lung obtained from each site was pro-cessed for micro-CT [12]. A companion core from thesame cluster was used for the gene profiling and valida-tion studies reported here. The representative nature ofthese samples with respect to the entire lung was estab-lished by comparing the densities of the sampled siteswith the frequency distribution of the densities in theentire lung on multidetector CT as reported in McDo-nough et al.[12].Measurement of mean linear interceptThe severity of emphysema within each core was esti-mated by measuring Lm. A micro-CT scan of each coreprovided approximately 1,000 contiguous 16-μm thickimages. Lm was measured at 20 regularly spaced intervalsof each of the micro-CT scans using a previouslyCampbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 2 of 16validated grid of test lines projected onto the image and acustom macro linked to specialized software (ImageProPlus; MediaCybernetics (Rockville, MD, USA). The num-ber of intercepts between these lines and tissue wascounted. Lm was calculated as the total length of the testlines divided by the number of cross-overs with tissue(equal to the number of intercepts divided by 2).Microarray sample processingHigh molecular weight (mRNA-containing fraction)RNA was isolated from tissue cores using the miRNeasyMini Kit (Qiagen, Valencia, CA, USA). RNA integritywas assessed using an Agilent 2100 Bioanalyzer andRNA purity was assessed using a NanoDrop spectro-photometer. RNA (1 μg) was processed and hybridizedonto the Human Exon 1.0 ST array (Affymetrix Inc.,Santa Clara, CA, USA) according to the manufacturer’sprotocol as previously described [15]. Expression Con-sole Version 1.1 (Affymetrix Inc.) was used to generatetranscript-level gene expression estimates for the ‘core’exon probesets via the robust multichip average (RMA)algorithm. Gene symbols of transcript IDs wereretrieved using DAVID [16]. These gene expression dataare available through the Gene Expression Omnibus(GEO) under the accession GSE27597.Microarray data analysisTwo linear mixed-effects models were used to identifygene expression profiles associated with the degree ofregional emphysema severity as measured by Lm:1. Geneij = β0 + βSlice × Sliceij + αj + εij2. Geneij = β0 + βSlice × Sliceij + βLm × Lmij + αj + εiji = 1, 2, ..., 8; j = 1, 2, ..., 8εij ∼ N(0, σ 2)αj ∼ N(0, σ 2aj)Geneij is the log2 expression value for sample i in patientj for a single gene. Slice is a fixed effect controlling for theposition within the lung from which the sample core wasobtained. The random term εij represents the randomerror, which was assumed to be normally distributed, ajrepresents the random effect for patient, and b0 representsthe intercept. Model 2 contains an additional fixed effectterm for emphysema severity measured by the natural logof Lm. A gene’s expression profile was considered asso-ciated with Lm if model 2 fit better than model 1 as deter-mined by a significant P-value from a likelihood ratio testbetween the two models after applying a false discoveryrate (FDR) correction. In the immunohistochemistryexperiments, these linear models were also used to exam-ine the relationship between Lm and the volume fractionof tissue with positive staining by substituting volume frac-tion (Vv) for gene expression as a dependent variable. Allstatistical analyses were conducted using R statisticalsoftware v2.9.2 and the nlme package in Bioconductorv2.4 [17].Functional enrichment analysisFunctional enrichment analysis was performed usingDAVID 2008 or Gene Set Enrichment Analysis (GSEA)v2.0.7 [16,18]. For DAVID, functional enrichment wasexamined among Gene Ontology categories, and KEGGand BIOCARTA pathways. All genes in the species Homosapiens were used as a reference set. For GSEA, geneswere ranked by the t-statistic of the bLm coefficient in thelinear mixed-effects model and then analyzed for theenrichment of canonical pathways and Gene Ontologyterm gene sets obtained from MSigDB v2.5.Connecting to other gene-expression datasetsUsing GSEA, sets of genes reported to change withCOPD-related phenotypes or with TGFb treatment inother gene-expression studies were examined in a rankedlist of genes ordered from most induced in severe emphy-sema to most repressed in severe emphysema by thet-statistic of the bLm coefficient in the linear mixed-effects model. Conversely, sets of genes we identified assignificantly positively or negatively associated with Lmwere examined within gene lists ranked by the degree ofdifferential expression as determined by re-analyzing pre-viously published COPD-or TGFb-related microarraystudies. See Additional file 1 for a description of the datanormalization procedures and statistical analyses used togenerate gene sets and/or ranked gene lists for each ofthe previously published gene-expression datasets.Connectivity MapIn order to find compounds that reverse gene-expressionpatterns associated with emphysema severity, we gener-ated separate signatures for each COPD or TGFb gene-expression dataset examined in this study. Signatures weregenerated by identifying the 50 genes most up-regulatedand the 50 genes most down-regulated with respect to aCOPD or TGFb-related phenotype. Each signature wasqueried against the CMap using the algorithm describedby Lamb et al. [13]. See Additional file 1 for a descriptionof the statistical analysis used to generate each query sig-nature for each phenotype within each dataset. The list ofall CMap query signatures used in this analysis include: 1)genes that change in expression as a function of regionalemphysema severity in this study; genes that change inexpression with 2) forced expiratory volume in 1 second(FEV1), 3) FEV1/forced vital capacity (FVC), or 4) betweencases versus controls in Bhattacharya et al. [7]; genes thatchange in expression between 5) controls versus emphy-sema patients or between 6) controls versus a1-antitrypsindisease in Golpon et al. [6]; genes that change in expres-sion with 7) FEV1 or 8) diffusing capacity of carbonCampbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 3 of 16monoxide (DLCO) in Spira et al.[9]; genes that change inexpression with 9) FEV1, 10) FEV1/FVC, 11) DLCO, 12)non-smokers versus GOLD2, or 13) non-smokers versusGOLD3 in Wang et al.[10]; and genes that change inexpression with TGFb treatment from 14) Qin et al.[19],15) Classen et al.[20], 16) Renzoni et al.[21], 17) Koinumaet al.[22], and 18) Malizia et al.[23]. For comparison of theCMap data to our in vitro studies of the effects of GHK inprimary lung fibroblasts, raw data for GHK-treated andcontrol samples were downloaded from the CMap websiteand normalized using MAS5.0 with the Affymetrix CDF.Genes were ranked by a paired t-test between treatmentand controls of different batches and compared to genesets of GHK and TGFb treatment using GSEA.Isolation and culture of lung fibroblastsLung tissue of former smokers (defined as quitting smok-ing for at least one year before surgery) with normal lungfunction or GOLD stage IV COPD was obtained frompatients undergoing surgery for resection for pulmonarycarcinoma or lung transplantation. Fibroblast cultureswere established from parenchymal lung tissue by anexplant technique as previously described [24]. Isolatedcells were characterized as fibroblasts by morphologicalappearance and expression pattern of specific proteins asdescribed previously [24,25]. Fibroblast cultures werestored into liquid nitrogen until use.ImmunofluorescenceFibroblast cultures at passage 3 were cultured in eight-well chamber slides (Gibco, Burlington, ON, Canada) ingrowth medium (DMEM, 10% fetal bovine serum (FBS),penicillin, and streptavidin from Invitrogen, Burlington,ON, Canada). After reaching 70% confluence, fibroblastswere cultured for 24 h in 1% FBS DMEM and then incu-bated with either TGFb1 10ng/ml (Peprotech, Dollarddes Ormeaux, Quebec, Canada), GHK 10 nM (Sigma,Markham, Ontario, Canada) or control media (1% FBSDMEM, penicillin, streptavidin) for a further 48 h. Afterstimulations, chamber slides were fixed with 4% parafor-maldehyde for 20 minutes, blocked in 10% goat serum inphosphate-buffered saline (PBS) with 0.1% saponin for1 h and then stained with integrin-b1 antibody (M-106,Santa Cruz Biotechnology, Santa Cruz, CA, USA) in 0.1%saponin in PBS for 2 h at room temperature. Followingwashing in PBS with 0.1% saponin and 0.1% Tween 20,secondary antibody conjugated with goat anti-Mouse IgGAlexa Fluor 488 and Phalloidin conjugated with AlexaFluor 594 were incubated for 2 h at room temperature.Following final washes, cultures were incubated withDAPI 1 ng/ml and then coverslipped with cytoseal.Confocal images were acquired with a Leica AOBS SP2laser scanning confocal microscope (Leica, Heidelberg,Germany). The images were overlaid and the contrastenhancements were performed on the images usingVolocity software™ (Improvisions Inc., Boston, MA,USA) as previously described [26].Collagen gel contraction assaysFibroblast cultures at passage 3 were cultured in six-welltissue culture plates (Gibco, Canada) in growth medium(DMEM, 10% FBS, penicillin, and streptavidin fromInvitrogen, Canada). After reaching 70% confluence,fibroblasts were cultured for 24 h in 1% FBS DMEMand then incubated with TGFb1 10 ng/ml (Peprotec,Canada), GHK 10 nM (Sigma, Canada) or growth mediacontrol for a further 48 h. Prior to the end of the treat-ment time point, a 12-well tissue culture plate was incu-bated with 1% bovine serum albumin in DMEM for 2 h.The medium was removed and then 500 µl of 0.4 mg/mltype I collagen (BD Biosciences, Mississauga, ON,Canada) was added and allowed to polymerize for 8 h at37°C. The treated fibroblasts were then trypsinized andseeded at 2 × 105 cells/500 µl of 1% FBS DMEM, penicil-lin, streptavidin in duplicate on the collagen gels and cul-tured for an additional 24 h at 37°C in 5% CO2. The gelswere imaged before and after and the extent of gel con-traction measured using Image Pro Software.Multi-photon and second harmonic generationmicroscopyThe collagen gels were fixed with 4% paraformaldehydefor 20 minutes and washed in PBS with 0.1% saponinand 0.1% Tween 20 before being incubated with phalloi-din conjugated with Alexa Fluor 594 for 1 h at roomtemp. Gels were then mounted on to a glass slide usingSecure-seal™ imaging spacers (size 20 mm; Sigma) andaqueous mounting media. The gels were then imagedusing second harmonic generation microscopy to deter-mine fibrilar collagen as previously described [27]. Foreach cell volume, Z-section images were compiled andthe three-dimensional image restoration was performedusing Volocity software (Improvisions, Inc.). A noise-removal filter with a kernel size of 3 × 3 was applied tothese three-dimensional images.ResultsStudy populationLm was quantified using micro-CT scans in eight sam-ples taken at regular intervals from apex to base of lungsfrom six subjects that required transplantation for COPDand two organ donors (Figure 1). Table 1 shows demo-graphic information and clinical characteristics of theeight subjects used in this study. As expected, samplesfrom subjects with COPD had a higher mean and agreater range of Lm values between samples compared tothose from donor lungs, indicating that there are regionsof severe emphysema in COPD subjects (Table 1).Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 4 of 16Donors(n=2)COPD(n=6)(a)(b)(c)(d)No Emphysema(Low Lm)Emphysema(High Lm)Micro-CT Gene expressionSum length of line segmentsTotal number of intercepts betweenline segments and the tissueLm  = Figure 1 Outline of study design. (a) Whole lungs were removed from patients with severe COPD and from donors, inflated with air, andrapidly frozen in liquid nitrogen vapor. (b) The frozen specimens were cut into 2-cm slices from apex to base of the lung. (c) Adjacent tissuecores were removed from 8 different slices of each lung (8 patients with 8 slices = 64 total regions). (d) Micro-CT was used to measure Lm at 20evenly spaced intervals throughout one core from each region.Table 1 Subject demographics for lung tissue samplesPatient ID Description Sex Age Pack years Smoking status Lm mean ± SD (μm) Lm range (μm)6965 COPD M 62 50 Former 716 ± 164 494-9826967 COPD F 61 25 Former 414 ± 82 334-5856968 COPD F 63 38 Former 724 ± 252 357-1,0136969 COPDa,b F 56 54 Former 1,822 ± 1270 521-4,6206970 COPDc M 55 15 Former 1,352 ± 599 647-2,5516971 COPD M 59 30 Former 1,097 ± 441 720-2,1016982 Donor M 59 - Never 384 ± 47 344-4736983 Donor M 62 24 Former 289 ± 41 231-352Subjects with COPD had FEV1/FVC <70% and FEV1 <25% predicted.a-cSome patients had other diseases: avon Willebrand disease; bhypertension; ca1-antitrypsindeficiency disease.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 5 of 16Subject 6967 was diagnosed with a pure airway obstruc-tion COPD phenotype without emphysema [28]. Consistentwith this diagnosis, the distribution of Lm measurementsfor this patient closely resembles the distribution of Lmmeasurements from the donor lungs. Subject 6970 wasdiagnosed with a1-antitrypsin deficiency. The remainingfour subjects with COPD had the centrilobular emphysema-tous phenotype commonly observed in smokers. The distri-bution of emphysematous destruction in tissue cores fromthese COPD patients range from little to no emphysema(Lm < 600) to very severe emphysema (Lm > 1,000) [12].Subject 6969 had one sample excluded from subsequentanalysis because its Lm measurement was an outlier (morethan three times the interquartile range of the distributionof Lm measurements in all cores from all lungs examined).Pathways associated with regional emphysema severityUsing linear mixed-effect models, the expression levels of127 genes were significantly associated with Lm and thusassociated with regional emphysema severity (Figure 2a;FDR <0.10; see Additional file 2 for the analytic resultsfor all genes). Using DAVID [16] or GSEA [18], we foundthat genes with functions in the B-cell receptor signalingpathway were over-represented among the up-regulatedgenes, while genes involved in cellular structure, integrinsignaling, extracellular matrix production, focal adhesion,blood vessel morphogenesis, and the vascular endothelialgrowth factor and TGFb pathways were enriched amongthe down-regulated genes (FDR <0.05; see Additionalfile 3 for a list of all significantly enriched pathways). Theexpression of CD79A, a component of the B-cell recep-tor, increased in expression with increasing emphysemaseverity (Figure 2b), and the expression of ACVRL1 (alsoknown as activin-like kinase I), a receptor in the TGFbpathway, decreased in expression with increasing emphy-sema severity (Figure 2c). These two genes are shown asexamples of the characteristic relationship between Lmand gene expression as observed in Figure 2a. To predicttranscription factors that might be responsible for theobserved patterns of differential expression, we inferred agene expression relevance network using the ContextLikelihood of Relatedness (CLR) algorithm [29]. Tran-scription factors with the most connections to othergenes included EPAS1 (also known as HIF-2a), KLF13,TAL1, TBX3, GATA2, and BCL11A (Additional file 4).Fourteen genes whose expression is significantly corre-lated with regional emphysema severity or transcriptionfactors that are highly connected to these genes in therelevance network were selected for quantitative RT-PCRvalidation in a subset of tissue cores from subjects withsevere emphysema (see Additional file 1 for methods).Twelve out of the fourteen genes had a significant corre-lation between the expression values derived from themicroarray and quantitative RT-PCR, showing that theassociation of gene expression with regional emphysemaseverity is reproducible across assays (Pearson correla-tion, P < 0.05; Additional file 5).In order to demonstrate that the 127 gene signature isrelated to regional emphysema severity within individualsand not to differences between donors and COPD patientsor to differences in levels of emphysema between COPDpatients, we repeated the same statistical analysis while onlyincluding the five COPD patients with emphysema andstandardizing the Lm measurements within each patientcore to a mean of zero and a standard deviation of one(Z-score). Using GSEA, the sets of up-and down-regulatedgenes in the 127-gene signature identified in the previousanalysis with all eight patients and unscaled Lm measure-ments were concordantly enriched among genes differen-tially expressed when only the five emphysema patientswere analyzed with Z-scored Lm measurements, indicatingthat this gene signature is associated with regional emphy-sema severity (FDR <0.001, GSEA; see Additional file 6 forthe enrichment plot).Validation of pathways up-regulated in regions of severeemphysemaIn order to investigate whether the up-regulation of com-ponents of the B cell receptor signaling pathway is asso-ciated with a change in the quantity of B cells in lungtissue, we quantified the Vv of CD79A protein, a markerfor B cells, in relation to Lm by immunohistochemistry(see Additional file 1 for methods). CD79A-positive B cellswere observed in the alveolar and small airway wall tissue(Figure 3). Vv was quantified in alveolar tissue for all 64samples and in small airway tissue for 43 samples that con-tained small airways and was found to be positively corre-lated to Lm in both the alveolar and small airway walltissue (P < 0.001), indicating that B cell abundanceincreases as emphysema severity increases.Validation of pathways down-regulated in regions ofsevere emphysemaSeveral members of the TGFb pathway were among thegenes that had decreased expression as a function of regio-nal emphysema severity. These genes included ACVRL1,ENG, TGFBR2, and SMAD6. Other components in thisfamily, including BMPR2 (FDR q-value = 0.125) andSMAD7 (FDR q-value = 0.223), also showed evidence ofmodest down-regulation while SMAD1 showed evidenceof modest up-regulation (FDR q-value = 0.101). To deter-mine whether the TGFb pathway might be affected byemphysema pathogenesis, we used seven previously pub-lished studies that had examined the effect of TGFbligands on gene expression to develop a collection of sig-natures of TGFb pathway activation [19-23,30,31]. Genesthat exhibited significantly decreased expression withincreasing emphysema severity were enriched amongCampbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 6 of 16genes induced in response to TGFb treatment in a total ofthree datasets (FDR <0.05, GSEA). Similarly, the sets ofgenes most induced by TGFb from each of the seven data-sets examined were enriched among genes whose expres-sion decreased as a function of emphysema severity (FDR<0.05, GSEA). As an example, the enrichment of genesassociated with emphysema severity among genes chan-ging with TGFb treatment in the dataset from Maliziaet al. [23] is shown in Figure 4a,b. See Additional file 7 forthe GSEA enrichment plots for all seven datasets.To further validate these findings, we cultured humanlung fibroblasts with and without TGFb1 and found thatthe set of genes most induced by TGFb1 were enrichedamong genes that decrease in expression with increasingregional emphysema severity (FDR <0.05, GSEA; see Addi-tional file 8 for the GSEA enrichment plot and Additionalfile 1 for the fibroblast culture methods). Immunostainingof lung tissue from the same regions on which weperformed gene expression analysis localized SMAD2, adown-stream signal transducer of TGFb, to the alveolarand airway walls while members of the bone morphoge-netic protein (BMP) pathway, including SMAD6 andSMAD1, were primarily seen in vascular endothelial cells(Additional file 9).Relationship to expression profiles in other COPD studiesIn order to show that the gene expression signature ofregional emphysema severity is present in larger cohortsof patients with earlier stages of disease, we used GSEA toexamine the relationship between genes associated withregional emphysema severity in this dataset and genesassociated with COPD phenotypes in other cross-sectionalstudies [6-11]. The genes that decreased in expressionwith increasing emphysema severity were significantlyenriched among genes down-regulated as a function ofCOPD-related phenotypes in four of the five previouslyNoEmphysemaSevereEmphysema5 6 7 8Natural Log of Lm(a) (b)(c)5.5 6.0 6.5 7.0 Log of LmExpression of CD79A5.5 6.0 6.5 7.0 Log of LmExpression of ACVRL169656967696869696970697169826983PatientFigure 2 Gene expression signature of regional emphysema severity. (a) Supervised heatmap of genes whose expression is associated withLm (FDR <0.10). Samples are organized from low to high Lm. Each row corresponds to a gene and each column corresponds to a sample.Green represents lower relative expression and red represents higher relative expression. (b,c) Expression of CD79A (b) and ACVRL1 (c) areplotted against the natural log of Lm with the color of each point indicating the subject from which the sample was derived.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 7 of 16published datasets that we examined (FDR <0.05, GSEA;Additional file 10). In addition, genes that increased inexpression with increasing emphysema severity wereenriched amongst genes up-regulated as a function ofCOPD-related phenotypes in three of the five datasets(FDR <0.05, GSEA). As an example, the enrichment ofgenes associated with regional emphysema severity amonggenes differentially expressed with the presence of emphy-sema in the dataset from Golpon et al.[6] is shown inFigure 4c,d. Conversely, sets of genes reported to be differ-entially expressed with COPD in four of the six othercross-sectional studies were enriched among the geneschanging in expression with increasing regional emphy-sema severity (FDR <0.05, GSEA; Additional file 10).Examples of genes validated by quantitative RT-PCR inthis study and concordantly differentially expressed inother studies are shown in Additional file 11. Many ofthese datasets, such as Bhattacharya et al.[7] and Wanget al.[10], contained larger numbers of patients with a vari-ety of stages of disease (for example, GOLD stage 0through GOLD stage IV). The enrichment of genes asso-ciated with regional emphysema severity with COPD-related phenotypes in these other datasets suggests thatthe biological processes associated with increasing emphy-sema severity within a patient with severe COPD also varyin individuals with earlier stages of disease.Prediction of novel therapeutics for emphysemaIn order to identify compounds that might reverse thegene-expression pattern associated with progression ofemphysema, we utilized the CMap [13], a compendium ofmicroarray experiments that measure the effect of thera-peutic compounds on gene expression in cancer cell lines.Signatures of genes that 1) change in expression withregional emphysema severity in this dataset, 2) change inexpression with lung function measures in other datasets[6,7,9,10], or 3) change in expression with TGFb treatmentin other datasets [19-23] were each used as separatequeries into the CMap data. We found that gene expres-sion changes resulting from treatment with the tripeptideGHK, a compound thought to accelerate wound healing[32,33], were negatively correlated with expression pat-terns associated with increasing regional emphysemaseverity (P = 0.006) and the COPD-related expression pat-terns observed in Bhattacharya et al.[7] and Golpon et al.[6] (P < 0.05). In addition, the gene expression effects ofLow emphysema(Low Lm)AlveolarTissueSmallAirwayWallHigh emphysema(High Lm)Figure 3 Validation of differential expression for CD79A by immunohistochemistry. Representative images of CD79A-positive cells (arrows)in the alveolar tissue and the small airway walls. Positive staining appears red. Scale = 200 µm; inset = 10 µm.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 8 of 16GHK are similar to the effects of TGFb treatmentobserved by Malizia et al.[23] (P = 0.004).As the CMap examined the effect of GHK in cancer celllines, we next sought to verify the effect of GHK treatmentin a cell type more relevant to emphysema pathogenesis.We utilized human lung fibroblasts because fibroblasts arethe major interstitial cell within the alveolar unit that cansynthesize and remodel the ECM and previous studieshave demonstrated that GHK can induce ECM productionin dermal fibroblasts [32-34]. Human lung fibroblast cul-tures were treated with two concentrations of GHK orwith TGFb1 (see Additional file 1 for methods). Geneexpression profiling of these cells demonstrated that the200 genes most induced by GHK at 1 μM in cancer celllines in the CMap dataset were enriched among genes thatincreased after treatment with GHK at 0.1 nM infibroblast cultures (FDR <0.05, GSEA). Furthermore,genes whose expression is decreased with increasingemphysema severity are enriched among genes induced byGHK at 10 nM (FDR < 0.05, GSEA; Figure 5a,b). Geneswhose expression is altered by GHK treatment at eitherconcentration are also enriched among genes that changewith TGFb1 treatment (FDR <0.05, GSEA; Figure 5c,d).See Additional file 8 for the GSEA enrichment plots show-ing the relationship between GHK, TGFb, and emphysemaseverity signatures.Reversal of COPD-related phenotypes in fibroblastsby GHKGenes induced in human lung fibroblasts after treat-ment with GHK were enriched in actin cytoskeletonorganization and focal adhesion pathways (FDR < 0.05,Down-regulated with regional emphysema severityUp-regulated with regional emphysema severityNoEmphysemaEmphysemaNoEmphysemaSevereEmphysemaGolpon et alGolpon et alControlTGFbTreatedDown-regulated with regional emphysema severityMalizia et alMalizia et al(c)(a) NoEmphysemaSevereEmphysema(b)(d)Up-regulatedwith TGFbDown-regulatedwith TGFbUp-regulatedwith emphysemaDown-regulatedwith emphysemaFigure 4 Relation between gene expression changes associated with regional emphysema severity (Lm) and other gene-expressionstudies by GSEA. (a) Relation between gene expression changes associated with regional emphysema severity and those induced by TGFbtreatment of A549 cells from Malizia et al.[23]. The color bar represents the fold change between the cell lines treated with and withoutTGFb1for 11,910 genes in Malizia et al.[23]. Red indicates a more positive fold change and green indicates a more negative fold change (inducedor repressed with TGFb, respectively). The vertical lines represent the position of genes associated with regional emphysema severity in theranked gene list. The height of the vertical lines corresponds to the magnitude of the running enrichment score from GSEA. Blue vertical linesindicate that the gene is part of the ‘core’ enrichment (that is, all the genes from the absolute maximum enrichment score to the end of theranking). (b) Supervised heatmaps of relative gene expression levels for the core enrichment genes in both the regional emphysema and Maliziaet al. datasets (11 genes down-regulated with emphysema severity but up-regulated with TGFb). Each gene is represented in the same rowacross heatmaps. (c) Genes changing in expression with increasing regional emphysema severity were enriched in the cross-sectional study ofCOPD-related gene expression from Golpon et al.[6]. The color bar represents the t-statistic from a t-test between five emphysema patients andfive non-smokers for 5,209 genes in Golpon et al.[6]. Blue and orange vertical lines indicate that the gene is part of the core enrichment.(d) Supervised heatmaps of relative gene expression levels for the core enrichment genes in both the regional emphysema and Golpon datasets(8 genes concordantly up-regulated; 19 genes concordantly down-regulated).Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 9 of 16DAVID). These included integrins involved in collagenattachment, such as ITGB1. ITGB1 gene expression wasalso down-regulated with increasing emphysema severityin lung tissue (P = 0.008). Resolution of damaged tissuerequires mesenchymal cells to attach to collagen fibersthrough integrin-dependent mechanisms and generatemechanical tension via the actin cytoskeleton to promotetissue contraction and wound size reduction. Using distallung fibroblasts isolated from former smokers with andwithout COPD, we found that GHK (10 nM), like TGFb1(10 ng/ml), induced alterations in integin-b1 localization(green staining in Figure 6a) and reorganized actin toform contractile filaments (red staining in Figure 6a). Wefurther demonstrated using a three-dimensional collagengel contraction bioassay that distal lung fibroblastsderived from former smokers with COPD (n = 5) wereunable to fully contract collagen I gels compared to fibro-blasts obtained from former smokers without COPD (n =5, P < 0.05; Figure 6b,c; see Additional file 12 for subjectdemographics), similar to what has been previouslydescribed [35]. However, fibroblasts derived from COPDpatients first treated for 48 h with either TGFb1 or GHKwere able to induce full collagen I gel contraction com-parable to that observed in fibroblasts from formersmokers without COPD (P < 0.01, Figure 6b,c). Using thesecond harmonic generation properties of fibrilar col-lagen and multi-photon microscopy, we confirmed thatfibroblasts from former smokers with COPD were unableto efficiently remodel collagen into fibrils (Figure 6d).Importantly, following 48 h of treatment with TGFb1 orGHK on lung fibroblasts from former smokers withCOPD, we were able to restore this intrinsic defect,which we propose is through organization of the actincytoskeleton to a contractile phenotype as demonstratedby the confocal images displayed in Figure 6a.DiscussionThe goal of this study was to identify gene expressionchanges associated with regional emphysema severity inorder to elucidate biological processes underlying the pro-gression of emphysema and to identify potential COPDtherapeutics. By measuring gene expression from regionsof varying emphysema severity within the same lung andby using a morphologic measurement of airspace size(Lm), which reflects the degree of alveolar destruction, wewere able to identify gene expression changes associatedspecifically with the emphysematous component ofCOPD.Down-regulated with regional emphysema severityControlGHKTreatedNoEmphysemaSevereEmphysemaControlGHKTreatedTGFbTreated$OWNåREGULATEDWITHTREATMENTOF'(+5PåREGULATEDWITHTREATMENTOF'(+(b)(a)(d)(c)Up-regulatedwith GHKDown-regulatedwith GHKUp-regulatedwith TGFbDown-regulatedwith TGFbFigure 5 Effect of GHK treatment on expression in human lung fibroblasts. (a) Genes decreasing in expression with increasing regionalemphysema severity were enriched among genes that are induced by GHK at 10 nM. (b) Supervised heatmaps of relative gene expressionlevels for the core enrichment genes in both datasets (18 genes down-regulated with Lm but up-regulated with GHK). Each gene is representedin the same row across heatmaps. (c) Genes differentially expressed with treatment of GHK at 0.1 nM were concordantly enriched among genesthat change with treatment of TGFb1. (d) Heatmap of relative gene expression levels for the core enrichment genes (118 genes up-regulatedand 124 down-regulated with both GHK and TGFb1).Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 10 of 16ControlTGFb1GHKF-actin DAPI MergeFibroblasts from former smoker Fibroblasts from former smoker with COPD(b)ControlTGFb1GHKFormer smokerPercent of collagengel contractionControl   TGFb  GHK5060708090100Former smokerCOPDP<0.05P<0.01P<0.01(c)(a)(d)F-actin DAPI MergeIntegrin-b1 Integrin-b1Control TGFb1Former smokerCOPDGHKCOPDFibrilar collagenActinFigure 6 Effect of GHK treatment on collagen contraction by fibroblasts from former smokers with COPD. (a) Representativeimmunofluorescent images of distal lung fibroblasts from former smokers with and without COPD treated with GHK (10 nM), TGFb1 (10 ng/ml),or media control for 48 h and stained with phalloidin to localize the actin cytoskeleton (red), integrin-b1antibody (green) and DAPI to localizenuclei (blue). (b) Representative images of collagen I gel bioassays at 24 h after being seeded with distal lung fibroblasts from former smokerswith and without COPD previously treated with GHK, TGFb1, or media control for 48 h. (c) The percentage of collagen I contraction wassignificantly decreased in fibroblasts derived from former smokers with COPD compared to former smokers without COPD (P < 0.05) but wassignificantly increased with addition of TGFb1 or GHK (P < 0.01). (d) Representative enface Z-stack slices of three-dimensional reconstructedcollagen I gel bioassays demonstrating actin in fibroblasts (green, phalloidin) and second harmonic signal originating from collagen fibrils(purple, 414 nM). Fibroblasts from former smokers with COPD were unable to efficiently remodel collagen into fibrils. However, this intrinsicdefect was restored with treatment of TGFb1 or GHK.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 11 of 16Interestingly, there was significant enrichment betweengenes differentially expressed in COPD or associatedwith worsening lung function in other datasets and thosewe found to be associated with regional emphysemaseverity. Importantly, this similarity supports the notionthat regional differences in emphysema severity reflectthe processes that occur with general COPD pathogen-esis and progression and are not only present in patientswith end-stage disease. Overall, these observations sug-gest a similarity in the gene expression alterations thataccompany airflow obstruction, gas exchange abnormal-ities, and alveolar destruction measured by Lm.A common characteristic in the pathology of COPD isprogressive lymphocyte infiltration of the small airwaysand alveolar walls [36]. In addition, the formation of ter-tiary lymphoid organs within this infiltration suggests thepresence of an adaptive immune response to persistentforeign or autoimmune antigens [37,38]. The presentstudy extends these observations by showing that theexpression patterns of several components of the B-cellreceptor signaling pathway have increased expression inregions of severe emphysema. Iga (CD79A) and Igb(CD79B) are proteins that associate with the B-cell recep-tor and transmit its signal upon stimulation. Immunohis-tochemistry showed a significant relationship between thevolume fraction of the airway wall and alveolar tissue posi-tively stained for CD79A and an increase in Lm. This rela-tionship supports an increased number of B cells in bothairway wall and alveolar tissues and is consistent with theinduction of CD79A during tissue destruction associatedwith the increase in Lm.The TGFb signaling pathway is involved in a variety ofcellular processes, including immune response, extracel-lular matrix remodeling, angiogenesis, and cell differen-tiation. This pathway has also been implicated in avariety of diseases such as cancer and fibrosis [39]. It hasbeen hypothesized that the TGFb pathway could play arole in COPD pathogenesis, but its role is not completelyunderstood [40]. Togo et al.[35] found that fibroblastsisolated from COPD patients exhibited reduced chemo-taxis, reduced nuclear to cytoplasmic ratios of phos-phorylated SMAD3, and decreased a-smooth muscleactin production compared to controls when treated withTGFb. Decreased mRNA expression or protein levels forTGFb1, TGFBR1 [41], SMAD3 [42], SMAD6 [43], andSMAD7 [41,43] have been reported in more advancedstages of COPD or fibroblasts from COPD patients. Inboth alveolar and bronchiolar epithelium of emphysema-tous lungs, a decrease in phosphorylated SMAD2 hasbeen shown by immunohistochemistry [44]. In normalhuman lung parenchyma, repair processes in response tomechanical injury are associated with increased TGFbsignaling, while a decrease in expression has beenobserved for TGFb-related genes with worsening lungfunction in patients with COPD [25,45]. Furthermore,association studies have identified both promoter andcoding region polymorphisms in the TGFb1 gene thatassociate with increased risk for COPD [46-48]. In thepresent study, we identified several components of theTGFb and BMP pathways that have decreasing expres-sion with increasing emphysema severity. In the BMPpathway, ACVRL1 and ENG are receptors involved inthe phosphorylation of SMAD1 and are expressed in themature lung vasculature. The changing expression ofSMAD6 and SMAD1, their localization predominantly tovascular endothelial cells, and the roles of ACVRL1 andENG in angiogenesis support the hypothesis of aberranttissue remodeling in the lung vasculature during emphy-sema pathogenesis. In the TGFb pathway, TGFBR2 is areceptor involved in the phosphorylation of SMAD2/3and is important for many tissue remodeling processes,including wound repair. Moreover, genes found to beinduced by TGFb in diverse studies were down-regulatedin regions of severe emphysema. The localization ofSMAD2 to alveolar and airway tissue and the decreasedTGFb pathway activity seen with increasing emphysemaseverity support the hypothesis that a decrease in TGFbpathway activity also contributes to emphysemapathogenesis.As COPD remains a major public health concern due tolack of effective therapeutic strategies, we sought to usecomputational methods to identify compounds that mightmodulate molecular processes associated with emphysemapathogenesis. The CMap is a large compendium of micro-array experiments that measures the effect of over 1,000compounds on gene expression in several cell lines [13].By querying a gene expression signature of disease patho-genesis against the CMap dataset, one can find com-pounds that elicit a pattern of gene expression that is theopposite to the disease-related gene expression profile.This can lead to the hypothesis that such compounds,since they reverse the disease-related gene expression pat-tern, are potential therapeutics for that disease. Thisapproach has been recently successful in the therapeuticrepositioning of the antiulcer drug cimetidine to lung ade-nocarcinoma and the anticonvulsant drug topiramate toinflammatory bowel disease [49,50]. In these studies, sig-natures for each disease were derived using several pub-licly available gene-expression datasets and queried in theCMap. Candidate compounds or drugs that could signifi-cantly reverse the disease-related signatures of geneexpression were further validated in vitro, showing thatthis computational method is a viable approach for identi-fying novel therapeutics.Using the CMap dataset, we identified a relationshipbetween the gene expression changes induced by the tri-peptide GHK and those that are repressed with increasingemphysema severity. Intriguingly, we further found thatCampbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 12 of 16GHK-treatment induced a pattern of gene expressionsimilar to that resulting from TGFb pathway activation.We replicated both of these findings in human lung fibro-blasts, which are the major interstitial cells that maintaintissue structural integrity by sculpting the connective tis-sue. GHK-Cu is a natural tripeptide that, in humanplasma, can be found at a concentration of 200 ng/ml atthe age of 20 years but drops to around 80 ng/ml by theage of 60 years [34]. Characterization GHK-Cu in skinwound repair models suggests that it induces wound con-traction, cell proliferation, angiogenesis, and increasedexpression of antioxidant enzymes and integrins [34,51].Direct evidence for the ability of GHK-Cu to promotewound healing comes from experimental rat modelswhere GHK treatment causes an acceleration of healingand a concentration-dependent increase of connective tis-sue and other ECM components [32,33]. These effects areconsistent with the gene expression alterations induced byGHK and TGFb treatment. Moreover, we confirmed thesesimilarities by demonstrating that GHK and TGFbinduced significantly higher expression and re-organiza-tion of actin and integrin-b1 in distal lung fibroblasts.We further assessed the ability of GHK and TGFb toinduce tissue contraction. As in previous studies [35],we demonstrated that distal lung fibroblasts derivedfrom COPD patients have intrinsic defects in collagen Icontraction compared to fibroblasts derived from formersmokers without COPD. When fibroblasts from COPDlungs were treated with GHK or TGFbcontraction andremodeling of collagen gels was induced to levels com-parable to fibroblasts from former smokers withoutCOPD. We further demonstrated that the collagen con-traction induced in COPD fibroblasts by GHK involvesthe organization of collagen I gels into collagen fibrilsusing multi-photon microscopy. Taken together, thesedata further support the hypothesis in which a wound-healing-like process is diminished as a function ofemphysema progression and further suggest that thisprocess is related to the TGFb pathway.While the number of subjects in this study for genomicanalysis was small, the analysis of eight specimens perlung representing different degrees of emphysema fromeach individual allowed us to detect gene expressionchanges specifically associated with regional emphysemaseverity. We further demonstrated that these genes areconcordantly differentially expressed in previous cross-sectional studies involving larger numbers of individualswith varying degrees of airflow limitation. These resultsvalidate the gene expression differences associated withregional emphysema severity in independent cohortsfrom different clinical settings and support the hypoth-esis that the genes whose expression is associated withregional emphysema severity reflect the activity of truedisease-associated processes. As demonstrated by ourmicro-CT data, COPD is a heterogeneous disease withinthe lung [12]. Further studies will be required to assesswhether COPD-associated differences in ECM remodel-ing by distal fibroblasts in vitro is associated with theregional disease severity in the tissue from which thefibroblasts are derived.ConclusionsThis study has provided insights into molecular pro-cesses associated with emphysematous destruction ofthe lung and revealed mechanisms that contribute tothe pathogenesis of COPD. Whole genome gene-expres-sion analysis supports the role of the immune responsein regional emphysema and elucidates additional path-ways involved in the process of emphysematous destruc-tion. The suggestion that progressive emphysematousdestruction is associated with down-regulation of genesinvolved in or downstream of tissue remodeling andwound repair pathways supports a role for defects inECM homeostasis and angiogenesis in the emphysema-tous destruction that occurs with chronic inflammationin COPD. We propose that these processes could belinked through decreased TGFb pathway activation.These data are supported by our identification of GHKas a compound with the potential to mimic TGFb path-way activity and induce collagen contraction, an impor-tant functional component of wound repair.Additional materialAdditional file 1: Supplementary methods.Additional file 2: Statistical results for gene expression analysis.Additional file 3: Functional categories enriched among genesassociated with regional emphysema severity.Additional file 4: Gene expression relevance network. Dark bluecircles are genes that have expression significantly correlated with Lm;light blue circles are all other genes. Edges are indicated by green(positive correlation) or red (negative correlation) lines.Additional file 5: RT-PCR validation of 14 genes associated withregional emphysema severity.Additional file 6: Confirmation of gene expression changesassociated with regional emphysema severity (Lm) within individualswith emphysema using GSEA. Genes associated with unscaled Lmmeasurements identified using all eight patients in the analysis areconcordantly enriched among genes associated with scaled Lmmeasurements (Z-scored within each patient) using only the fiveemphysema patients (FDR <0.001). These results demonstrate that the 127gene signature is related to regional emphysema severity within individualsand not to differences between donors and COPD patients or to differencesin levels of emphysema between COPD patients. Orange and blue colorbars represent the t-statistics from correlations of gene expression with Lm.The vertical black lines represent the position of genes in the gene setamong the ranked gene list. The length of the black lines corresponds tothe magnitude of the running enrichment score from GSEA.Additional file 7: Relation between gene expression changesassociated with regional emphysema severity (Lm) and studies ofTGFb-related gene expression using GSEA. Genes associated with Lmare enriched among the genes that are differentially expressed in responseto TGFb treatment in datasets from (a) Classen et al.[20], (b) Koinuma et al.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 13 of 16[22], and (c) Malizia et al.[23]. (d) Genes most induced by TGFb in sevenstudies [19-23,30,31] are enriched among the genes that are associated withLm. Orange and blue color bars represent the t-statistics from correlationsof gene expression with a continuous variable. Red and green color barsrepresent the fold change between samples treated with and withoutTGFb. The vertical black lines represent the position of genes in the geneset among the ranked gene list. The length of the black lines correspondsto the magnitude of the running enrichment score from GSEA. Enrichmentswith an FDR q-value <0.05 were considered significant.Additional file 8: Relation between gene expression changesassociated with regional emphysema severity (Lm) and geneexpression changes that occur with treatment of GHK or TGFb infibroblast cell lines using GSEA. (a) Genes increasing in expression inresponse to treatment with GHK or TGFb are enriched among genes thatdecrease with increasing emphysema severity. (b) Genes differentiallyexpressed with TGFb treatment or in response to GHK in theConnectivity Map are enriched among genes that change in expressionwith GHK (0.1 nM) in fibroblast cell lines. (c) Genes that are differentiallyexpressed with TGFb treatment or that are down-regulated withincreasing emphysema severity are enriched among genes that changein expression with GHK (10 nM) in fibroblast cell lines. (d) Genes that aredifferentially expressed in response to GHK are concordantly enrichedamong genes that change in expression with TGFb treatment infibroblast cell lines. Orange and blue color bars represent the t-statisticsfrom correlations of gene expression with a continuous variable. Red andgreen color bars represent the t-statistic between treated and untreatedsamples. The vertical black lines represent the position of genes amongthe ranked gene list. The length of the black lines corresponds to themagnitude of the running enrichment score from GSEA. Enrichmentswith an FDR q-value <0.05 were considered significant.Additional file 9: Localization of members of the TGFb superfamilyusing immunohistochemistry. Representative images of positiveSMAD2 staining (arrows) in the (a) alveolar and (b) small airway walltissue. (c) Representative image of positive SMAD6 staining in vascularendothelial cells (arrows) and macrophages (arrowheads). (d)Representative image of weak SMAD1 staining in vascular endothelialcells (arrows). Representative images are shown for control IgG stainingin the (e) alveolar wall tissue, (f) airway wall tissue, and (g) blood vessels.Scale bar = 200 µm.Additional file 10: Relation between gene expression changesassociated with regional emphysema severity (Lm) and cross-sectionalstudies of COPD-related gene expression using GSEA. Genes associatedwith Lm are enriched among the genes found to associated with thepresence of COPD or degree of airflow obstruction in datasets from(a) Golpon et al.[6], (b) Spira et al.[9], (c) Wang et al.[10], and (d) Bhattacharyaet al.[7]. (e) Genes previously found to be associated with COPD-relatedclinical variables [6,8-10] are enriched among the genes associated with Lm.Orange and blue color bars represent the t-statistics from correlations of geneexpression with a continuous variable. Red and green color bars represent thet-statistic from a t-test between cases and controls. The vertical black linesrepresent the position of genes in the gene set among the ranked gene list.The length of the black lines corresponds to the magnitude of the runningenrichment score from GSEA. Enrichments with an FDR q-value <0.05 wereconsidered significant.Additional file 11: Examples of genes associated with Lm andvalidated by quantitative RT-PCR that were also differentiallyexpressed in other COPD-related gene-expression datasets. Genessuch as ACVRL1, SMAD6, CCR7, and CXCL13 were associated with increasingregional emphysema severity and concordantly differentially expressed inother datasets such as Golpon et al.[6] and/or Wang et al.[10].Additional file 12: Subject demographics for lung fibroblast cultures.AbbreviationsBMP: bone morphogenetic protein; CMap: Connectivity Map; COPD: chronicobstructive pulmonary disease; CT: computed tomography; DLCO: diffusingcapacity of carbon monoxide; DMEM: Dulbecco’s modified Eagle’s medium;ECM: extracellular matrix; FBS: fetal bovine serum; FDR: false discovery rate;FEV1: forced expiratory volume in 1 second; FVC: forced vital capacity; GOLD:Global Initiative for Chronic Obstructive Lung Disease; GSEA: Gene SetEnrichment Analysis; Lm: mean linear intercept; PBS: phosphate-bufferedsaline; TGF: transforming growth factor; Vv: volume fraction.Authors’ contributionsAS, JCH, MEL, DAK, DSP, and WT conceived and designed the experiments.JEM, TLH, DVP, CAB, MS, JVG, GL, YOA, JX, XZ, and SH performed theexperiments. JDC, JEZ, TLH, DVP, and CAB analyzed the data. JDC and JCHcontributed materials. JDC, MEL, JCH, and AS wrote the manuscript. Allauthors read and approved the final version for publication.Competing interestsBoston University has intellectual property related to the work described inthis manuscript.AcknowledgementsWe thank A Wright, D Horng, and P Sanchez for supporting studies in thismanuscript; W Elliott for help with the immunohistochemistry; F Shaheen forhelp with cell culture; K Steiling for reviewing this manuscript; T Abraham atthe UBC James Hogg Research Centre Imaging Cellular Imaging andBiophysics Core facility for help with imaging the collagen gels. This workwas funded by the National Heart, Lung and Blood Institute (R01HL095388to AS and MEL), National Center for Advancing Translational Science(UL1 TR000157 to AS and MEL), National Science Foundation (IntegrativeGraduate Education and Research Traineeship to JDC and JEZ), BritishColumbia Lung Association, Canadian Institute for Health Research, Parker BFrancis Foundation (Senior Fellowship to TLH) and the Dutch AsthmaFoundation and European Respiratory Society (International ResearchFellowship to CAB).Author details1Division of Computational Biomedicine, Department of Medicine, BostonUniversity School of Medicine, 72 East Concord Street, Boston, MA 02118,USA. 2Bioinformatics Program, Boston University, 44 Cummington Street,Boston, MA 02215, USA. 3UBC James Hogg Research Centre, ProvidenceHeart + Lung Institute, St. Paul’s Hospital and Department of Pathology andLaboratory Medicine, University of British Columbia, 1081 Burrard St,Vancouver, BC V6Z 1Y6, Canada. 4Department of Pathology and MedicalBiology, University Medical Center Groningen, University of Groningen,Hanzeplein 1, 9713 Groningen, Netherlands. 5Department of Pathology andLaboratory Medicine, Boston University School of Medicine, 72 East ConcordStreet, Boston, MA 02118, USA. 6Hospital of the University of Pennsylvania,Division of Thoracic Surgery, 3400 Spruce Street 6 White Building,Philadelphia, PA 19104, USA. 7Department of Pulmonary Diseases, UniversityMedical Center Groningen, University of Groningen, Hanzeplein 1, 9713Groningen, Netherlands.Received: 5 March 2012 Revised: 14 August 2012Accepted: 16 August 2012 Published: 31 August 2012References1. Miniño AM, Xu J, Kochanek KD, Statistics V: Deaths: Preliminary Data for2008. National Vital Statistics Reports Volume 59 NVSS; 2010 [http://www.cdc.gov/nchs/data/nvsr/nvsr59/nvsr59_02.pdf].2. Lange P, Scharling H, Fabricius P, Vestbo J: Developing COPD: a25 year follow up study of the general population. Thorax 2006,61:935-939.3. Park JW, Ryter SW, Choi AMK: Functional significance of apoptosis inchronic obstructive pulmonary disease. COPD 2007, 4:347-353.4. Postma DS, Timens W: Remodeling in asthma and chronic obstructivepulmonary disease. Proc Am Thorac Soc 2006, 3:434-439.5. Rennard SI, Bailey KL: Chronic obstructive pulmonary diseaseexacerbations: accurate and easy measurement promises much. Am JRespir Crit Care Med 2012, 185:1139-1141.6. Golpon HA, Coldren CD, Zamora MR, Cosgrove GP, Moore MD, Tuder RM,Geraci MW, Voelkel NF: Emphysema lung tissue gene expressionprofiling. Am J Respir Cell Mol Biol 2004, 31:595-600.7. Bhattacharya S, Srisuma S, Demeo DL, Shapiro SD, Bueno R, Silverman EK,Reilly JJ, Mariani TJ: Molecular biomarkers for quantitative and discreteCOPD phenotypes. Am J Respir Cell Mol Biol 2009, 40:359-367.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 14 of 168. Ning W, Li C-J, Kaminski N, Feghali-Bostwick C, Alber SM, Di YP,Otterbein SL, Song R, Hayashi S, Zhou Z, Pinsky DJ, Watkins SC, Pilewski JM,Sciurba FC, Peters DG, Hogg JC, Choi AMK: Comprehensive geneexpression profiles reveal pathways related to the pathogenesis ofchronic obstructive pulmonary disease. Proc Natl Acad Sci USA 2004,101:14895-14900.9. Spira A, Beane J, Pinto-Plata V, Kadar A, Liu G, Shah V, Celli B, Brody JS:Gene expression profiling of human lung tissue from smokers withsevere emphysema. Am J Respir Cell Mol Biol 2004, 31:601-610.10. Wang I-M, Stepaniants S, Boie Y, Mortimer JR, Kennedy B, Elliott M,Hayashi S, Loy L, Coulter S, Cervino S, Harris J, Thornton M, Raubertas R,Roberts C, Hogg JC, Crackower M, O’Neill G, Paré PD: Gene expressionprofiling in patients with chronic obstructive pulmonary disease andlung cancer. Am J Respir Crit Care Med 2008, 177:402-411.11. Francis SMS, Larsen JE, Pavey SJ, Bowman RV, Hayward NK, Fong KM,Yang I: Expression profiling identifies genes involved in emphysemaseverity. Respir Res 2009, 10:81.12. McDonough JE, Yuan R, Suzuki M, Seyednejad N, Elliott WM, Sanchez PG,Wright AC, Gefter WB, Litzky L, Coxson HO, Paré PD, Sin DD, Pierce RA,Woods JC, McWilliams AM, Mayo JR, Lam SC, Cooper JD, Hogg JC: Small-airway obstruction and emphysema in chronic obstructive pulmonarydisease. N Engl J Med 2011, 365:1567-1575.13. Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, Lerner J,Brunet J-P, Subramanian A, Ross KN, Reich M, Hieronymus H, Wei G,Armstrong S, Haggarty SJ, Clemons P, Wei R, Carr S, Lander ES, Golub TR:The Connectivity Map: using gene-expression signatures to connectsmall molecules, genes, and disease. Science 2006, 313:1929-1935.14. Choong CK, Haddad FJ, Martinez C, Hu DZ, Pierce J, Meyers BF,Patterson GA, Cooper JD: A simple, reproducible, and inexpensivetechnique in the preparation of explanted emphysematous lungs for exvivo studies. J Thorac Cardiovasc Surg 2005, 130:922-923.15. Zhang X, Liu G, Lenburg ME, Spira A: Comparison of smoking-inducedgene expression on Affymetrix Exon and 3’-based expression arrays.Genome Inform 2007, 18:247-257.16. Dennis G, Sherman BT, Hosack D, Yang J, Gao W, Lane HC, Lempicki R:DAVID: Database for Annotation, Visualization, and Integrated Discovery.Genome Biol 2003, 4:P3.17. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B,Gautier L, Ge Y, Gentry J, Hornik K, Hothorn T, Huber W, Iacus S, Irizarry R,Leisch F, Li C, Maechler M, Rossini AJ, Sawitzki G, Smith C, Smyth G,Tierney L, Yang JYH, Zhang J: Bioconductor: open software developmentfor computational biology and bioinformatics. Genome Biol 2004, 5:R80.18. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette M,Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP: Gene setenrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 2005, 102:15545-15550.19. Qin H, Chan MWY, Liyanarachchi S, Balch C, Potter D, Souriraj IJ, Cheng ASL,Agosto-Perez FJ, Nikonova EV, Yan PS, Lin H-J, Nephew KP, Saltz JH,Showe LC, Huang THM, Davuluri RV: An integrative ChIP-chip and geneexpression profiling to model SMAD regulatory modules. BMC Syst Biol2009, 3:73.20. Classen S, Zander T, Eggle D, Chemnitz JM, Brors B, Büchmann I, Popov A,Beyer M, Eils R, Debey S, Schultze JL: Human resting CD4+ T cells areconstitutively inhibited by TGF beta under steady-state conditions.J Immunol 2007, 178:6931-6940.21. Renzoni E, Abraham DJ, Howat S, Shi-Wen X, Sestini P, Bou-Gharios G,Wells AU, Veeraraghavan S, Nicholson AG, Denton CP, Leask A, Pearson JD,Black CM, Welsh KI, du Bois RM: Gene expression profiling reveals novelTGFbeta targets in adult lung fibroblasts. Respir Res 2004, 5:24.22. Koinuma D, Tsutsumi S, Kamimura N, Taniguchi H, Miyazawa K,Sunamura M, Imamura T, Miyazono K, Aburatani H: Chromatinimmunoprecipitation on microarray analysis of Smad2/3 binding sitesreveals roles of ETS1 and TFAP2A in transforming growth factor betasignaling. Mol Cell Biol 2009, 29:172-186.23. Malizia AP, Keating DT, Smith SM, Walls D, Doran PP, Egan JJ: Alveolarepithelial cell injury with Epstein-Barr virus upregulates TGFbeta1expression. Am J Physiol Lung Cell Mol Physiol 2008, 295:L451-460.24. Noordhoek JA, Postma DS, Chong LL, Menkema L, Kauffman HF, Timens W,van Straaten JFM, van der Geld YM: Different modulation of decorinproduction by lung fibroblasts from patients with mild and severeemphysema. COPD 2005, 2:17-25.25. Pechkovsky DV, Hackett TL, An SS, Shaheen F, Murray L, Knight D: Humanlung parenchyma but not proximal bronchi produces fibroblasts withenhanced TGF-beta signaling and alpha-SMA expression. Am J Respir CellMol Biol 2010, 43:641-651.26. Hackett T-L, Warner SM, Stefanowicz D, Shaheen F, Pechkovsky DV,Murray L, Argentieri R, Kicic A, Stick SM, Bai TR, Knight D: Induction ofepithelial-mesenchymal transition in primary airway epithelial cells frompatients with asthma by transforming growth factor-beta1. Am J RespirCrit Care Med 2009, 180:122-133.27. Abraham T, Carthy J, McManus B: Collagen matrix remodeling in3-dimensional cellular space resolved using second harmonicgeneration and multiphoton excitation fluorescence. J Struct Biol 2010,169:36-44.28. Bignon J, Khoury F, Even P, Andre J, Brouet G: Morphometric study inchronic obstructive bronchopulmonary disease. Pathologic, clinical, andphysiologic correlations. Am Rev Respir Dis 1969, 99:669-695.29. Faith JJ, Hayete B, Thaden JT, Mogno I, Wierzbowski J, Cottarel G, Kasif S,Collins JJ, Gardner TS: Large-scale mapping and validation of Escherichiacoli transcriptional regulation from a compendium of expressionprofiles. PLoS Biol 2007, 5:e8.30. Chambers RC, Leoni P, Kaminski N, Laurent GJ, Heller RA: Global expressionprofiling of fibroblast responses to transforming growth factor-beta1reveals the induction of inhibitor of differentiation-1 and providesevidence of smooth muscle cell phenotypic switching. Am J Pathol 2003,162:533-546.31. Verrecchia F, Chu ML, Mauviel A: Identification of novel TGF-beta/Smadgene targets in dermal fibroblasts using a combined cDNAmicroarray/promoter transactivation approach. J Biol Chem 2001,276:17058-17062.32. Siméon A, Wegrowski Y, Bontemps Y, Maquart FX: Expression ofglycosaminoglycans and small proteoglycans in wounds: modulation bythe tripeptide-copper complex glycyl-L-histidyl-L-lysine-Cu(2+). J InvestDermatol 2000, 115:962-968.33. Maquart FX, Bellon G, Chaqour B, Wegrowski J, Patt LM, Trachy RE,Monboisse JC, Chastang F, Birembaut P, Gillery P: In vivo stimulation ofconnective tissue accumulation by the tripeptide-copper complexglycyl-L-histidyl-L-lysine-Cu2+ in rat experimental wounds. J Clin Invest1993, 92:2368-2376.34. Pickart L: The human tri-peptide GHK and tissue remodeling. J BiomaterSci Polym Ed 2008, 19:969-988.35. Togo S, Holz O, Liu X, Sugiura H, Kamio K, Wang X, Kawasaki S, Ahn Y,Fredriksson K, Skold CM, Mueller KC, Branscheid D, Welker L, Watz H,Magnussen H, Rennard SI: Lung fibroblast repair functions in patientswith chronic obstructive pulmonary disease are altered by multiplemechanisms. Am J Respir Crit Care Med 2008, 178:248-260.36. Hogg JC, Chu F, Utokaparch S, Woods R, Elliott WM, Buzatu L,Cherniack RM, Rogers RM, Sciurba FC, Coxson HO, Paré PD: The nature ofsmall-airway obstruction in chronic obstructive pulmonary disease. NEngl J Med 2004, 350:2645-2653.37. Brusselle GG, Demoor T, Bracke KR, Brandsma C-a, Timens W: Lymphoidfollicles in (very) severe COPD: beneficial or harmful?. Eur Respir J 2009,34:219-230.38. van der Strate BW, Postma DS, Brandsma C-A, Melgert BN, Luinge M,Geerlings M, Hylkema MN, van den Berg A, Timens W, Kerstjens HaM:Cigarette smoke-induced emphysema: A role for the B cell?. Am J RespirCrit Care Med 2006, 173:751-758.39. Blobe GC, Schiemann WP, Lodish HF: Role of transforming growth factorbeta in human disease. N Engl J Med 2000, 342:1350-1358.40. Morty RE, Königshoff M, Eickelberg O: Transforming growth factor-betasignaling across ages: from distorted lung development to chronicobstructive pulmonary disease. Proc Am Thorac Soc 2009, 6:607-613.41. Zandvoort A, Postma DS, Jonker MR, Noordhoek J, Vos JTWM, van derGeld YM, Timens W: Altered expression of the Smad signalling pathway:implications for COPD pathogenesis. Eur Respir J 2006, 28:533-541.42. Zandvoort A, Postma DS, Jonker MR, Noordhoek J, Vos JTWM, Timens W:Smad gene expression in pulmonary fibroblasts: indications fordefective ECM repair in COPD. Respir Res 2008, 9:83.43. Springer J, Scholz FR, Peiser C, Groneberg D, Fischer A: SMAD-signaling inchronic obstructive pulmonary disease: transcriptional down-regulationof inhibitory SMAD 6 and 7 by cigarette smoke. Biol Chem 2004,385:649-653.Campbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 15 of 1644. Leppäranta O, Myllärniemi M, Salmenkivi K, Kinnula VL, Keski-Oja J, Koli K:Reduced phosphorylation of the TGF-Beta signal transducer Smad2 inemphysematous human lung. COPD 2009, 6:234-241.45. Gosselink JV, Hayashi S, Elliott WM, Xing L, Chan B, Yang L, Wright C, Sin D,Paré PD, Pierce J, Pierce R, Patterson A, Cooper J, Hogg JC: Differentialexpression of tissue repair genes in the pathogenesis of chronicobstructive pulmonary disease. Am J Respir Crit Care Med 2010,181:1329-1335.46. Smolonska J, Wijmenga C, Postma DS, Boezen HM: Meta-analyses onsuspected chronic obstructive pulmonary disease genes: a summary of20 years’ research. Am J Respir Crit Care Med 2009, 180:618-631.47. van Diemen CC, Postma DS, Vonk JM, Bruinenberg M, Nolte IM,Boezen HM: Decorin and TGF-beta1 polymorphisms and development ofCOPD in a general population. Respir Res 2006, 7:89.48. van Diemen CC, Postma DS, Aulchenko YS, Snijders PJLM, Oostra B, vanDuijn CM, Boezen HM: Novel strategy to identify genetic risk factors forCOPD severity: a genetic isolate. Eur Respir J 2010, 35:768-775.49. Dudley JT, Sirota M, Shenoy M, Pai RK, Roedder S, Chiang AP, Morgan A,Sarwal MM, Pasricha PJ, Butte AJ: Computational repositioning of theanticonvulsant topiramate for inflammatory bowel disease. Sci Transl Med2011, 3:96ra76.50. Sirota M, Dudley JT, Kim J, Chiang AP, Morgan A, Sweet-Cordero A, Sage J,Butte AJ: Discovery and preclinical validation of drug indications usingcompendia of public gene expression data. Sci Transl Med 2011, 3:96ra77.51. Kang Y-A, Choi H-R, Na J-I, Huh C-H, Kim M-J, Youn S-W, Kim K-H, Park K-C:Copper-GHK increases integrin expression and p63 positivity bykeratinocytes. Arch Dermatol Res 2009, 301:301-306.doi:Cite this article as: Campbell et al.: A gene expression signature ofemphysema-related lung destruction and its reversal by the tripeptideGHK. Genome Medicine 2012 4:67.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitCampbell et al. Genome Medicine 2012, 4:67http://genomemedicine.com/content/4/8/67Page 16 of 16


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items