Human alveolar epithelial cells type II are capable of TGFβ-dependent epithelial-mesenchymal-transition and collagen-synthesis

Abstract

Background: The origin of collagen-producing cells in lung fibrosis is unclear. The involvement of embryonic signaling pathways has been acknowledged and trans-differentiation of epithelial cells is discussed critically. The work presented here investigates the role of TGFB in cytoskeleton remodeling and the expression of Epithelial-Mesenchymal-Transition markers by Alveolar Epithelial Cells Type II and tests the hypothesis if human alveolar epithelial cells are capable of trans-differentiation and production of pro-fibrotic collagen.

Methods: Primary human alveolar epithelial cells type II were extracted from donor tissues and stimulated with TGFβ and a TGFβ-inhibitor. Transcriptome and pathway analyses as well as validation of results on protein level were conducted.

Results: A TGFβ-responsive fingerprint was found and investigated for mutual interactions. Interaction modules exhibited enrichment of genes that favor actin cytoskeleton remodeling, differentiation processes and collagen metabolism. Cross-validation of the TGFβ-responsive fingerprint in an independent IPF dataset revealed overlap of genes and supported the direction of regulated genes and TGFβ-specificity.

Conclusions: Primary human alveolar epithelial cells type II seem undergo a TGFβ-dependent phenotypic change, exhibit differential expression of EMT markers in vitro and acquire the potential to produce collagen.

Keywords: Alveolar epithelial cells; Collagen; EMT; Fibrosis.

Fig. 1

Transcriptional analysis of hAECII upon TGFβ stimulation. hAECII different donors were stimulated each with 5 ng/ml TGβ1, 10 μM SB431542, 5 ng/ml TGFβ1 and 10 μM SB431542 or left untreated for 48 h (N = 4). Significantly regulated genes were computed by Repeated-Measures ANOVA (RM-ANOVA) with a Benjamini-Hochberg multiple testing correction cut-off of ≤0.05. Averaged gene expression data of results from RM-ANOVA is shown as a Heatmap with Hierarchical Clustering of samples and entities according to Pearson Centered algorithm with Ward’s linkage rule. A Fold-Change Filter was applied to further analyze only probes that were at least ≥2 fold up-regulated compared to medium control (a). Out of those genes, a list of targets was extracted that was exclusively regulated by TGFβ and designated TGFβ fingerprint (b). The TGFβ fingerprint genes were further investigated for mutual interactions by querying the String protein-protein interaction database with those genes that exhibited an annotated GeneSymbol. A global map of these genes was constructed based on interaction scores within Cytoscape (c). The presence of intrinsic modules within the interactions of the TGFβ fingerprint was computed by a spectral partition based cluster algorithm from the Reactome FIViz app and respective modules encoded by different colours. Further details of interactions within each module are shown in (d). Global enrichment of GSEA Hallmark gene sets (e) and Reactome pathways (f) from whole list of TGFβ fingerprint genes are displayed by their FDR q-value

Fig. 2

Word clouds of significantly enriched GO terms highlight activity of different biological processes governed by each module. Each module of the TGFβ fingerprint was investigated for significant enrichment of GO terms with a FDR cut-off of ≤0.05. The Top15 leading edge GO terms of Biological Process, by ranking of their corrected FDR p-value, were subjected to a word cloud generator for summary. Here, the relative abundance of each word within the submitted list of terms is directly related to the size within the word cloud. Words with a higher abundance exhibit an increased font size

Fig. 3

TGFβ induces actin cytoskeleton remodeling in hAECII. hAECII (N = 5) were seeded on coverslips and cultivated for 48 h in medium a, 5 ng/ml TGFβ b, 10 μM SB431542 (c) or both (d). The actin cytoskeleton was visualized by Rhodamine-Phalloidin and nuclei by DAPI. A representative image from one experiment shows the actin cytoskeleton at a magnification of 200× with a scale bar of 100 μm. Quantification of the actin area per cell was done with ImageJ and depicted as scatter dot plot (E) with the mean value and SD of 5 biological replicates. For statistical analysis, p ≤ 0.05 (*), 0.01 (**) and 0.001 (***) were regarded as significant

Fig. 4

TGFβ regulates E-Cadherin, Vimentin and Collagen I on protein level in hAECII. Immunocytochemistry (a) was used to assess protein expression of E-Cadherin, Vimentin and Collagen I in paraffin-embedded hAECII from cell culture experiments (N = 2). Representative image shown with a scale bar = 100 μm. Red colour indicates positive signals. b-e Double immunofluorescence staining was used to show an up-regulated collagen metabolism in AECII lineage by TGFβ-stimulation. The thyroid transcription factor I (TTF-1) of the surfactant molecules was targeted as a proof of AECII lineage and the collagen chaperon HSP47 was used to display an elevated collagen synthesis. Nuclei are stained by DAPI (blue). TTF1 was visualized with a TRITC-conjugated secondary antibody (red) and HSP47 with an Alexa488-conjugated (green) secondary antibody. Exemplary image of 3 biological replicates with scale bar =50 μm show the expression of HSP47 in medium control b, with 5 ng/ml TGFB1 C, with 10 μm SB431542 (d) or both, TGFB1 and SB431542 (e). For means of cross-validation of results from ICC and IF, In-Cell Western analysis was conducted (f)s to investigate regulation of Collagen I, HSP47, N-Cadherin and ZO1 on protein level in hAECII . Cells were seeded in 96 well plates and stimulated for 48 h. Protein expression of Collagen I, HSP47, ZO1 and N-Cadherin/CDH2 was detected by means of ICW assay with primary antibodies against the targets and near-infrared conjugated secondary antibodies. To-Pro3 was used to stain all cells as loading control. Data is shown as semi-quantitative fluorescent signal normalized to total amount of cells, corrected for background signal and normalized to medium controls. N = 5 (HSP47, ZO1, N-Cadherin) or N = 9 (Collagen I). Semi-quantitative measurement of collagen produced by hAECII as stained by SiriusRed (N = 7)

Fig. 5

Consensus between the TGFβ fingerprint and up-regulated genes in tissues of IPF patients. An external dataset (GSE10667) of microarray results from the tissues of 23 IPF patients and 15 healthy controls was used to investigate the potential overlap with genes from the TGFβ fingerprint in hAECII. 53 genes were shared between both datasets (a). The Log Fold Changes of these genes (IPF compared to normal lung or to the unstimulated hAECII control, respectively) are depicted as bar charts with the respective direction (b)