GDF15 is an epithelial-derived biomarker of idiopathic pulmonary fibrosis

Abstract

Idiopathic pulmonary fibrosis (IPF) is the most common and devastating of the interstitial lung diseases. Epithelial dysfunction is thought to play a prominent role in disease pathology, and we sought to characterize secreted signals that may contribute to disease pathology. Transcriptional profiling of senescent type II alveolar epithelial cells from mice with epithelial-specific telomere dysfunction identified the transforming growth factor-β family member, growth and differentiation factor 15 (Gdf15), as the most significantly upregulated secreted protein. Gdf15 expression is induced in response to telomere dysfunction and bleomycin challenge in mice. Gdf15 mRNA is expressed by lung epithelial cells, and protein can be detected in peripheral blood and bronchoalveolar lavage following bleomycin challenge in mice. In patients with IPF, GDF15 mRNA expression in lung tissue is significantly increased and correlates with pulmonary function. Single-cell RNA sequencing of human lungs identifies epithelial cells as the primary source of GDF15, and circulating concentrations of GDF15 are markedly elevated and correlate with disease severity and survival in multiple independent cohorts. Our findings suggest that GDF15 is an epithelial-derived secreted protein that may be a useful biomarker of epithelial stress and identifies IPF patients with poor outcomes.

Keywords: MIC-1; NAG-1; SASP; aging.

Conflict of interest statement

D. J. Kass reports collaborative research funding from Regeneron Pharmaceuticals in pulmonary hypertension, which is unrelated to this article. R. Lafyatis has received consulting fees from PRISM Biolab, Merck, Bristol Myers Squibb, Biocon, Formation, Genentech/Roche, UCB, and Sanofi and grant support from Elpidera and Regeneron, not related to the submitted work. K. F. Gibson reports membership on the advisory board of Bayer Pharmaceuticals, outside the scope of the submitted work. E. L. Herzog has received grant funding from Sanofi, Bristol Myers, and Biogen, and consulting fees from Boehringer Ingelheim, Genentech, and Merck, all unrelated to the submitted work. M. Rojas reports funding from Regeneron and MedImmune, unrelated to this work. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

Figures

Fig. 1.

Growth and differentiation factor 15 (Gdf15) is upregulated in response to telomere dysfunction. A: schematic of our analysis strategy for identifying secreted proteins from transcriptional profiling data. Transcriptional data were obtained as described previously (3). Differentially upregulated genes that were also annotated as secreted proteins are identified. B: quantitative real-time PCR for Gdf15 from sorted type II alveolar epithelial cells (AEC2s) from Trf2Fl/+Sftpc-CreER (control) and Trf2Fl/FlSftpc-CreER (senescent) AEC2s. Cells were sorted 10 days after treatment with tamoxifen based on green fluorescent protein expression from mTmG reporter allele (3). Gene expression was normalized to Hprt and B2m. C: representative images of RNA in situ hybridization staining for Gdf15 from mouse lungs 6 wk after treatment with tamoxifen showing AEC2-specific expression of Gdf15. AEC2s were identified by expression of the Sftpc transcript. Scale bar is 50 µM. D: quantitation of the colocalization of Gdf15 and Sftpc transcripts (n = 4 mice per group). E: bronchoalveolar lavage cell counts from mice treated with 2 µg of GDF15. GDF15 or sterile saline was instilled directly into the lungs and the bronchoalveolar lavage was collected thereafter at the indicated times. Total viable cells were quantitated by trypan blue staining, and a differential count was performed on >100 cells. Values are means and standard deviation (SD). Student’s t test (two-tailed) was used to compare groups.

Fig. 2.

Bleomycin induces expression of growth and differentiation factor 15 (GDF15). A and B: quantitative ELISA of GDF15 levels in bronchoalveolar lavage (BAL; A) and plasma (B) from mice treated with intratracheal bleomycin or saline. Mice were treated on day 0, and groups of 6–7 mice (at least 3 male and 3 female at each time point) were harvested at the indicated time points. C: representative RNA in situ hybridization of day 3 lungs showing alveolar expression of Gdf15. D: quantitation of colocalization of Sftpc and Gdf15 transcripts in RNA in situ hybridization staining. Values are means and SD. Student’s t test (two-tailed) was used to compare groups. ****P < 0.0001, ***P < 0.001, and **P < 0.01.

Fig. 3.

Growth and differentiation factor 15 (GDF15) is upregulated in idiopathic pulmonary fibrosis (IPF) and expressed by epithelial cells. A: box-and-whisker plot of GDF15 expression data from the Lung Genomics Research Consortium (LGRC). Horizontal line marks the median value, box boundaries show the upper and lower quartiles, and whiskers show high and low values. Relative expression was calculated from normalized hybridization signal from microarray data. Welch’s t test, two tailed, was used to compare groups. B: correlation between natural log of GDF15 expression from IPF patients and carbon monoxide diffusion capacity (DlCO) in LGRC samples (Pearson correlation = −0.24). C: violin plots of GDF15 and EPCAM (epithelial cell adhesion molecule) expression in scRNA-seq data demonstrating epithelial specific expression of GDF15 (43). Data were processed and clustered exactly as described (43). The identity of each cluster is listed in the legend on the right. D: violin plot comparing GDF15 expression in donor and IPF lungs. AEC1 and AEC2, type I and II alveolar epithelial cell, respectively; NK, natural killer.

Fig. 4.

Expression of putative growth and differentiation factor 15 (GDF15) receptors. A: Gfral expression was measured in the mouse tissues shown (n = 3 for all tissues, except medulla for which only a single sample was measured). No signal was detected in the majority of samples, except skeletal muscle and medulla. Values are means and SD. B: violin plots of TGFBR1 and TGFBR2 (transforming growth factor-β receptors I and II, respectively) in scRNA-seq data from human lungs showing macrophage predominant expression of TGFBR1 and broad expression of TGFBR2 (43). AEC1 and AEC2, type I and II alveolar epithelial cell, respectively; NK, natural killer.

Fig. 5.

Growth and differentiation factor 15 (GDF15) is expressed by honeycomb cyst epithelial cells. A: representative photomicrographs from three independent donor and idiopathic pulmonary fibrosis (IPF) lungs. Slides were stained for GDF15 (brown) and counterstained with hematoxylin. GDF15 expression is present in macrophages from healthy lungs (arrowheads) but rarely in epithelial cells. In contrast, GDF15 expression was abundant in epithelial cells (arrows) and macrophages in fibrotic lungs. Scale bar in micrographs is 2 mm and 100 μm in insets. B: GDF15 RNA in situ hybridization (RNA-ISH) in donor and IPF lungs showing epithelial-specific expression of GDF15. Scale bar is 100 μm. C: Western blot of whole lung lysate from donor and IPF lungs for proGDF15 and GAPDH as a load control. D: quantitation of proGDF15 in Western blot in C. Values are means and SD. Student’s t test, two tailed was used for comparison in D. IHC, immunohistochemistry.

Fig. 6.

Growth and differentiation factor 15 (GDF15) is a biomarker of idiopathic pulmonary fibrosis (IPF) and identifies high-risk patients. A–C: plasma levels of GDF15 in three cohorts of controls and IPF patients. GDF15 was measured using Luminex assay in cohort 1 (A) and ELISA in cohorts 2 and 3 (B and C, respectively). Values are means and SD. D and E: Kaplan-Meier graph showing the proportion of patients who were alive or transplant free as a function of time. IPF patients were split into two groups based on GDF15 level. Patients from the upper quartile were compared with the lower three quartiles. Age-adjusted P values are from Cox regression analysis. Comparisons in A, B, and C used Welch’s t test, two tailed.