Gene expression profiling reveals novel TGFβ targets in adult lung fibroblasts

Primary adult lung fibroblasts were cultured from three control samples (unaffected lung from patients undergoing cancer-resection surgery) and from open-lung biopsy samples of lung fibrosis patients, three with idiopathic pulmonary fibrosis (IPF) [8] and three with pulmonary fibrosis associated with the fibrotic disease systemic sclerosis [9]. Independent reviews of the clinical (SV, ER) and histopathologic diagnosis (AGN) were performed. All the idiopathic pulmonary fibrosis biopsies were characterized by a usual interstitial pneumonia pattern (UIP), whereas all of the scleroderma-associated pulmonary fibrosis were classified as non-specific interstitial pneumonia (NSIP) [8]. Verbal and written consent was given by all subjects; authorization was given by the Royal Brompton Hospital Ethics Committee. Fibroblast culture conditions were as previously described [20]. At confluence, lung fibroblasts (all between passages 4–5) were serum-deprived for 16 hours, and exposed to either 4 ng/ml of activated TGF-β1 (R&D Systems) or serum-free culture medium for four hours. The concentration and time point of TGFβ used in our experiments was determined from ongoing studies within our laboratory, in which a 4 hour treatment with TGFβ 4 ng/ml was found to show significant induction of selected known direct TGFβ target genes, including CTGF.

At the end of the treatment period with or without TGFβ, total RNA was harvested (Trizol, Life Technologies), quantified, and integrity was verified by denaturing gel electrophoresis.

Preparation of RNA samples for chip hybridization followed Affymetrix (Affymetrix, Santa Clara, California) protocols. Each RNA sample derived from an individual fibroblast line was hybridized on a separate microarray chip. Hybridization of cRNA to Affymetrix human U95Av2 chips, containing approximately 12,000 well characterized human genes, signal amplification and data collection were performed using an Affymetrix fluidics station and chip reader, following Affymetrix protocol. Scanned files were analyzed using Affymetrix Version 5.0 software (MAS5). Chip files were analyzed by scaling to an average intensity of 150 per gene, as recommended by Affymetrix. Reproducibility was assessed using two pairs of RNA samples from the same control line, TGFβ-treated/untreated; the concordance correlation coefficients were of 0.979 and 0.983, respectively.

TGFβ response was analyzed by using a combination of the MAS5 Affymetrix change algorithm and of ANOVA models. According to Affymetrix criteria, in each TGFβ-treated/medium only pair, genes were defined as differentially regulated (either up or down) by TGFβ only when identified as significantly increased (I) or decreased (D) as determined by the Affymetrix change algorithm, with a change p value<0.001, and were detected as Present (according to the "absolute call"obtained by an Affymetrix algorithm) at least in the samples with the highest count (i.e. medium only in the case of D and TGFβ in the case of I). Genes were defined as TGFβ-responsive in normal human lung fibroblasts when they fulfilled all of the following three conditions: a) they were detected as TGFβ-regulated by Affymetrix criteria (see above) in at least two of the three control pairs; b) they showed a mean fold change after TGFβ of at least 2 (or lower than 0.5) in control fibroblasts; c) either a two-way ANOVA including only control fibroblasts detected a significant (p < 0.05) increase or decrease in control fibroblasts after TGFβ or they were also found to be responsive in at least four of the six fibrotic fibroblast lines and a significant effect (p < 0.05) of treatment (with TGFβ) was detected by a repeated measure ANOVA model including all the samples and adjusting for individual samples, disease, and interaction between treatment and disease. All statistical analyses were performed on log transformed data to reduce inequalities of variance. Thus, the latter ANOVA model could detect genes which were equally up- or down-regulated in normal and fibrotic fibroblasts, taking advantage of the larger number of samples, while the first model (equivalent to a paired t test) could detect changes possibly occurring in controls but not in fibrotic cell lines.

Except for unknown genes, all gene symbols and names are given according to the nomenclature proposed by the Human Genome Organization (HUGO) Gene Nomenclature Committee.

Real time PCR (RT-PCR) was performed to confirm selected novel TGFβ targets in lung fibroblasts. Adult lung fibroblast lines [three control and three fibrotic (IPF)] were treated with or without TGFβ (4 ng/ml) for four hours. Total RNA was isolated from treated and untreated samples using Trizol (Life Technologies) and the integrity of the RNA was verified by gel electrophoresis. Total RNA (1 microgram) was reverse transcribed in a 20 μl reaction volume containing oligonucleotide dTs (dT₁₈) and random decamers (dN₁₀) using M-MLV reverse transcriptase (Promega) for 1 hour at 37°C. The cDNA was diluted to 100 μl with DEPC-treated water and 1 μl was used per real-time PCR reaction. A set of eight standards containing a known concentration of target amplicon was made by PCR amplification, isolation by gel electrophoresis through a 2% agarose gel followed by gel purification using QIAquick PCR purification spin columns (Qiagen). The concentration of the amplicon was measured by spectrophotometry and diluted in DEPC-treated water containing transfer RNA (10 μg/ml) to make standards of 10 fold dilutions from 100 pg/ μl to 0.01 fg/ μl. The target was measured in each sample and standard by real-time PCR using FastStart DNA Master SYBR Green (Roche Applied Science) as described by the manufacturer, in half the reaction volume (10 μl). Samples and standards were amplified for 30 to 40 cycles with the appropriate primers (Molecular Biology Unit, KCL School of Biological Sciences) at least in duplicate. The amount of target in the sample in picograms was read from the standard curve and values were normalised to 28S ribosomal RNA (pg of target/pg of 28S ribosomal-RNA). The oligonucleotide primer sequences are listed (5'-3'): angiotensin II receptor type1 (AGTR1) primers: forward TGC TTC AGC CAG CGT CAG TT and reverse GGG ACT CAT AAT GGA AAG CAC; SMAD specific E3 ubiquitin protein ligase 2 (SMURF2): forward AAC AAG AAC TAC GCA ATG GGG and reverse GTC CTC TGT TCA TAG CCT TCT G; nuclear receptor co-repressor 2 (NCOR2): forward CAG CAG CGC ATC AAG TTC AT and reverse GTA ATA GAG GAC GCA CTC AGC; bone morphogenetic protein 4 (BMP4) primers: forward CTA CTG GAC ACG AGA CTG GT and reverse GAG TCT GAT GGA GGT GAG TC.

The results were analyzed using Student's paired t-test after logarithmic transformation, and statistical significance was taken as a p value of <0.05.

Lung fibroblasts were grown to confluence in DMEM with 10% FCS. At confluence, lung fibroblasts (all between passages 2–5) were serum-deprived overnight, and exposed to either 4 ng/ml of activated TGF-β1 (R&D Systems) or serum-free culture-medium with the addition of 0.1% BSA for 24 hours. To determine the signalling pathways through which TGFβ induces AGTR1, lung fibroblasts were treated with specific inhibitors 30 minutes before treatment with TGFβ. These included the dual MKK1/MKK2 inhibitor U0126 (10 μM) and predominant MKK1 inhibitor PD98059 (50 μM), known to inhibit MKK2 only weakly [21], as well as the p38 MAPK inhibitor SB 202190 (30 μM). Cell layer lysates were examined. Cell protein (10 μg/sample) was heated to 99°C for 5 min, loaded into sample wells, resolved on a 12% tricine SDS-polyacrylamide gel (Novex, San Diego, CA), and run at 120 V for 2 h. The separated proteins were transferred onto nitrocellulose membranes at 30V for 90 minutes. Membranes were blocked by incubation for one hour with 5% non-fat milk in phosphate buffered saline (PBS) containing 0.1% Tween 20. They were then washed and incubated overnight at 4°C in a 1:500 dilution of rabbit anti-angiotensin II receptor 1 polyclonal antibody (Santa Cruz Biotechnology), followed by a three-time wash in PBS and incubation in 1:1000 goat anti-rabbit biotinylated IgG (Vector Laboratories, Peterborough, UK) for 60 min at room temperature. Membranes were washed three times in PBS, and the signal was amplified/detected by using the ECL protocol as described by the manufacturer (Amersham plc, Little Chalfont, UK). Films were analysed by laser scanning densitometry on an Ultrascan XL (LKB-Wallac, UK). Data were analyzed by using Student's paired t test after log transformation and a p value<0.05 was considered significant.

The distribution of staining for AGTR1 was assessed by immunohistochemistry in surgical lung biopsies from four patients with idiopathic pulmonary fibrosis (IPF), meeting the diagnostic criteria of the American Thoracic Society/European Respiratory Society Consensus Classification [8], and in control biopsies (normal periphery of resected cancer) from three patients undergoing cancer resection surgery. Paraffin-embedded sections were dewaxed with xylene, hydrated and heated in the microwave at 120 degrees for 30 minutes in citrate buffer (10 mM pH 6.0).

Slides were then briefly rinsed in PBS, blocked with 10% normal goat serum for 20', incubated with rabbit polyclonal anti-human AGTR1 antibody (N-10, 1:50, Santa Cruz Biotechnology, Santa Cruz, Calif) for one hour at room temperature. After washing with PBS, sections were incubated with biotinylated goat anti-rabbit IgG diluted in PBS (1:200) for 30 minutes, rinsed, and finally incubated with Vectastain Elite STR-ABC reagent (Vector Laboratories) for 30 minutes. After washing, sections were visualized using 3-amino-9-ethylcarbazole chromogen and H₂O₂ as substrate (SK-4200; Vector Laboratories). Sections were then washed in tap water, counterstained with Carrazzis hematoxylin, and mounted with Gelmount (Biomeda, Foster City, CA) for examination using an Olympus BH-2 photomicroscope. Controls included an exchange of primary antibodies with goat matched antibodies. To confirm staining specificity, sections were also incubated with either nonimmune rabbit IgG control or secondary antibody only.

According to the criteria outlined in the methods, a four hour treatment with TGFβ was found to regulate 129 transcripts in human lung fibroblasts. TGFβ-responsive transcripts included genes with roles in gene expression, matrix formation, cytoskeletal remodelling, signalling, cell proliferation, protein expression and degradation, cell adhesion and metabolism. A complete list of TGFβ-regulated genes is provided (see Additional File 1). The complete set of gene array data has been deposited in the Gene Expression Omnibus database with GEO serial accession number GSE1724 http://​www.​ncbi.​nlm.​nih.​gov/​geo.

We did not observe a substantial degree of difference in the response to TGFβ between the two fibrotic groups (idiopathic pulmonary fibrosis and scleroderma-associated pulmonary fibrosis) and control lung fibroblasts. Once the criteria outlined in the methods section and the p-value for interaction with treatment had been taken into account, there were no significant differences in the response to TGFβ among the three groups except for two genes, KIAA0261 (probe N: 40086_at), an unknown gene more upregulated in IPF (median fold change 2.2) than in scleroderma-associated pulmonary fibrosis (1.5) and in controls (1.3), and BTG1 (probe N: 37294_at), which was only slightly more downregulated in scleroderma-associated pulmonary fibrosis (fold change:0.4) than in IPF (0.6) and in controls (0.7). As both the number of genes and the magnitude of the differences were minimal, they were not considered meaningful and were not investigated further. Among genes responding significantly to TGFβ in control lung fibroblasts, as assessed by ANOVA analysis, none changed in opposite directions in either of the fibrotic groups. All the genes that responded significantly in the control group alone, were also TGFβ-responsive when analysis was extended to include the fibrotic cell lines. Furthermore, none of these genes responded differently to TGFβ between the two fibrotic groups, which are thus presented together in Tables 1 and 2.

For the purpose of this study, we will concentrate on genes involved in transcriptional regulation, cytoskeletal/extracellular matrix organization, and signal transduction (Tables 1 and 2).

Several of the genes regulated by TGFβ confirmed previously published findings, thus validating our methods, including JUN-B, SMAD7, connective tissue growth factor, elastin, and SERPINE1 [17, 18, 16, 30]. To further consolidate our analysis, we selected a small group of novel TGFβ targets to be confirmed by RT-PCR in both control and fibrotic lung fibroblasts. These novel fibroblast TGFβ-responsive genes included potential key candidates in the regulation by TGFβ of lung tissue fibrosis and included angiotensin II receptor type 1 (AGTR1), SMURF2, a gene involved in terminating TGFβ signalling, NCOR2, a transcriptional co-repressor and BMP4, a member of the TGFβ family. Compared to untreated samples, we confirmed that TGFβ upregulated AGTR1 (ratio = 2.4; p = 0.002), SMURF2, (ratio = 1.8, p = 0.003), NCOR2 (ratio 1.4; p = 0.004), and downregulated BMP4 (ratio = 0.4; p = 0.009), with no difference in the response between control and fibrotic fibroblasts (Figure 1).

We focused on AGTR1 protein because, as shown by microarray analysis, it was the most highly TGFβ-induced gene among signaling molecules in both control and fibrotic fibroblasts (Table 2). To verify whether AGTR1 mRNA upregulation corresponded to an increase in protein levels, we performed Western analysis on primary human adult lung fibroblasts exposed to TGFβ or medium alone in serum-free conditions for 24 hours. The intensity of the angiotensin II receptor 1 immunoreactive band was significantly increased in TGFβ-treated fibroblasts compared to those treated with medium alone (2.4 fold; p < 0.001) (Figure 2). To identify the signalling pathways through which TGFβ induces AGTR1, we evaluated whether the ability of TGFβ to induce AGTR1 expression in lung fibroblasts was blocked by specific signaling pathway inhibitors. A 30 minute preincubation with the dual MKK1/MKK2 inhibitor U0126 significantly inhibited TGFβ induction of AGTR1 protein (p < 0.01), whereas predominant MKK1 inhibitor PD98059 and p38 MAPK inhibitor SB202190 had no significant effect (Figure 2).

We assessed staining for AGTR1 in lung biopsies from four patients with idiopathic pulmonary fibrosis and compared it to that of three control lungs. In particular we aimed to evaluate AGTR1 staining in fibroblastic foci, aggregates of fibroblasts/myofibroblasts in close contact with alveolar epithelial cells. Both in control and in idiopathic pulmonary fibrosis lung biopsies, AGTR1 immunoreactivity was observed in alveolar epithelial cells and alveolar macrophages. In addition, the fibroblasts within the fibroblastic foci present in idiopathic pulmonary fibrosis biopsies stained positive for the receptor (Figure 3).