Gene Expression Profiles of Acute Exacerbations of Idiopathic Pulmonary Fibrosis

Study Population

Lung tissue samples for microarray analysis were obtained through the University of Pittsburgh Health Sciences Tissue Bank (Pittsburgh, PA) as previously described (15) (and see the online supplement). Those included 23 lungs from patients with IPF, 8 lungs from patients with acute exacerbation of IPF (IPF-AEx; obtained from explanted lungs or via the warm autopsy protocol) (23), and normal lung histology samples from control subjects. Plasma samples of patients with stable IPF (n = 10), patients with IPF-AEx (n = 16), and healthy control subjects (n = 12) were obtained from Asan Medical Center (Seoul, South Korea). The diagnosis of IPF was based on the American Thoracic Society and European Respiratory Society definition (24). The definition of IPF-AEx was based on criteria provided by Collard and colleagues (3) or Akira and colleagues (25). All cases were reviewed by expert pulmonologists and pathologists. Detailed clinical information about the subjects with IPF-AEx is provided (see Tables E1 and E2 in the online supplement).

The mean forced vital capacity (expressed as a percentage of the normal expected value) (FVC%) and diffusing capacity for carbon monoxide (expressed as a percentage of the normal expected value) (Dl_CO%) of patients with stable IPF and patients with IPF-AEx are provided in Tables 1A and 1B. All studies were approved by the Institutional Review Boards at the University of Pittsburgh and Asan Medical Center.

Oligonucleotide Microarray Experiments

Total RNA extracted from snap-frozen lung tissue was used as template for the generation of labeled cRNA that was hybridized to Agilent 4 × 4 4k whole human genome microarrays and scanned with an Agilent scanner (Agilent Technologies, Santa Clara, CA) as recommended and previously described by us (16). The complete data set is available in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/; accession number GSE10667).

Real-time Quantitative Reverse Transcription-Polymerase Chain Reaction

The same RNA samples used for microarray experiments were used to run real-time quantitative reverse transcription-polymerase chain reaction (qRT-PCR) on TaqMan system (Applied Biosystems, Foster City, CA). PCR was performed with TaqMan universal PCR master mix (Applied Biosystems) for the following genes: CCNA2, DEFA1–3, AGER, COL1A2, MMP1, MMP7, and GUSB. The results were analyzed by the ΔΔC_t method and (GUSB, encoding β-glucuronidase) was used as a housekeeping gene. Fold change was calculated by taking the average over all the control samples as the baseline.

Immunohistochemistry

OCT-embedded sections of normal and IPF-AEx samples were used for fluorescence immunohistochemistry. Rabbit polyclonal antibody against cyclin A2 (CCNA2; Abcam, Cambridge, MA), prosurfactant protein C (Abcam), and mouse monoclonal antibodies for cytokeratin (Vector Laboratories, Burlingame, CA), vimentin (Vector Laboratories), Ki-67 (Abcam), and α-defensins (Hycult Biotechnology, Uden, The Netherlands) were used as described in the online supplement. Each antigen–antibody complex was labeled with biotinylated antibody against mouse or rabbit IgG, and visualized with fluorescein green or Texas red (Vector Laboratories). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (Sigma-Aldrich, St. Louis, MO).

ELISA

Plasma concentrations of α-defensins were determined with an ELISA kit for α-defensins (DEFA1–3) (Hycult Biotechnology).

Western Blot

Total protein was denatured by adding Laemmli sample buffer (Bio-Rad, Hercules, CA) 2-mercaptoethanol and boiling. Fifteen micrograms of total protein was used in the immunoblotting process.

Terminal Deoxynucleotidyltransferase dUTP Nick End Labeling

Formalin-fixed, paraffin-embedded tissue samples were used for the terminal deoxynucleotidyltransferase dUTP nick end labeling (TUNEL) assay, done with an in situ cell death detection kit (Roche Applied Sciences, Indianapolis, IN). After proteinase digestion the sections were incubated in a mixture containing terminal deoxynucleotidyltransferase and fluorescein isothiocyanate–labeled dUTP. The TUNEL conjugates were labeled with alkaline phosphatase, visualized with Vector red, and counterstained with hematoxylin. The samples were observed under a light microscope.

Statistical Analysis

Array images were processed according to the Agilent Feature Extraction protocol (26). All arrays were cyclic-LOESS normalized, using the Bioconductor package as previously described (27). For statistical analysis we applied significance analysis of microarrays (SAM) (28). A q value of 5, which corresponds to a 5% false discovery rate, was used as a cutoff of statistical significance in microarray data. Data visualization and clustering were performed with Genomica (28), Scoregene (29), and Spotfire DecisionSite 9 (TIBCO, Palo Alto, CA). For qRT-PCR, the Student t test was used and significance was defined as P < 0.05.

Compared with Control Samples, Global Gene Expression Patterns of IPF-AEx and IPF Are Similar

When compared with control samples, the global gene expression patterns of IPF-AEx are almost identical to those of stable IPF (Figure 1A). To better characterize this similarity we looked at genes that characterized IPF lungs in previous studies (Figure 1B) and compared their expression in IPF-AEx with that in stable IPF (Figure 1C). Impressively, all the highlighted genes were not significantly different between IPF and IPF-AEx (Figure 1C; see also Table 2). These results indicate that compared with control samples, IPF-AEx exhibits a fibrosis signature that is identical to that of stable IPF.

To confirm this observation we performed qRT-PCR for MMP1, MMP7, COL1A2, and AGER, which are among the genes that consistently distinguish patients with IPF from control subjects (16, 20) (Figure 2). As expected, COL1A2, MMP1, and MMP7 were significantly higher and AGER was significantly lower in IPF and IPF-AEx compared with control samples, but there was no significant difference in their expression between IPF and IPF-AEx.

Direct Comparison of IPF-AEx and Stable IPF

To identify the subtler gene expression changes that distinguish IPF-AEx from stable IPF, and to focus on clusters of genes that seemed differentially expressed in IPF and IPF-AEx (orange quadrangles; Figure 1A), we performed a direct comparison of IPF-AEx and stable IPF. We identified 579 significantly differentially expressed genes (q < 5) (Figure 3A). Among them were genes related to stress response such as heat shock proteins and α-defensins and mitosis-related genes including histones and CCNA2 (Figure 3B and Table 3).

Impressively, the gene expression signature of IPF-AEx did not exhibit an increase in inflammatory response compared with stable IPF. Genes known to be associated with the general inflammatory response, adaptive or innate immunity, were not significantly enriched in genes that characterize IPF-AEx. Similarly, only 2 (the α-defensins DEFA3 and DEFA4) of the 100 genes on the array that belong to gene ontology (GO) annotations associated with response to viral infection were significantly changed (see Figure E1 in the online supplement).

CCNA2 Is Overexpressed in IPF-AEx

When compared with stable IPF, CCNA2 was one of the top 20 up-regulated genes, with a q value of 0 and a 2.27-fold increase (Table 3). Considering that this gene is a regulator of the cell cycle we chose to validate and localize its expression. qRT-PCR confirmed the microarray data (Figures 3B and 3C), and Western blots indicated an increase in CCNA2 protein (Figure 3D). To localize CCNA2 overexpression we performed double-fluorescence labeling for CCNA2 with either cytokeratin or vimentin. The double labeling demonstrated coexpression of CCNA2 with cytokeratin (Figures 4A–4C), but not with vimentin (Figures 4D–4F), indicating that the increase in CCNA2 was localized to epithelial cells and not fibroblasts. Confocal microscopy revealed localization of CCNA2 in the alveolar epithelium as well as the basal portion of the bronchial epithelium (Figures 4G and 4H). To determine whether increases in CCNA2 were associated with cellular proliferation, we double-labeled IPF-AEx tissues with CCNA2 and Ki-67. CCNA2 and Ki-67 colocalized to the pulmonary epithelium (Figure 4I), suggesting accelerated epithelial cell proliferation, potentially as a compensatory response of the injured epithelium.

IPF-AEx Lungs Exhibit Widespread Epithelial Apoptosis

To determine whether increased epithelial proliferation was associated with epithelial cell death in IPF-AEx, we studied apoptosis in IPF-AEx tissues by in situ TUNEL assay. We observed significant and widespread positive epithelial TUNEL staining in IPF-AEx tissues (Figure 4K). This pattern was consistent with previous observations in IPF (30). In addition, positive TUNEL stains were also observed in the hyaline membranes typical of diffuse alveolar damage, a pathological hallmark of IPF-AEx (Figure 4J). In control lungs, rare TUNEL-positive structures were predominantly observed in apoptotic bodies engulfed by alveolar macrophages (Figure 4L).

α-Defensin Expression Is Increased in Lungs and Peripheral Blood of Patients with IPF-AEx

Gene expression of α-defensins (DEFA3 and DEFA4) was significantly increased in IPF-AEx lungs compared with stable IPF in the microarray data (Figure 3B), and the increase was confirmed by qRT-PCR (Figure 5A). To determine whether α-defensins may serve as peripheral blood markers for IPF-AEx, we analyzed their levels in the plasma of patients with IPF-AEx from Asan Medical Center. Plasma defensin concentrations were significantly higher in patients with IPF-AEx compared with control subjects (P = 0.0007) or patients with stable IPF (P = 0.025) (Figure 5B). To determine the cellular origins of α-defensins in IPF-AEx, we performed double labeling on IPF-AEx lung sections with antibodies against α-defensins and against the alveolar type II cell marker surfactant protein C (SFTPC). These experiments identified alveolar type II cells as the source of α-defensins in IPF-AEx (Figures 5C–5E).