Background
Idiopathic pulmonary fibrosis (IPF) is characterized by alveolar epithelial cell hyperplasia and increased myofibroblast with the interstitial deposition of extracellular matrix (ECM) [
1]. The disease course is highly variable due to interactions between chronic inflammatory and fibrosis-related processes [
2,
3]. Exploration of global gene expression in lung tissues may facilitate the identification of novel candidate genes to further explain the complex mechanism and to predict the clinical courses of IPF. In a human study, 164 differentially expressed genes were demonstrated in IPF lung tissues [
4]. In this study, fibrotic lungs showed changes in the expression of genes involved in ECM formation and degradation. A comparison of rapid-and slow-progressor patients revealed 437 differentially expressed genes involved in morphogenesis [
5]. In another study, integration of the expression levels of 134 genes enabled the discrimination of progressive and stable subjects [
6]. Almost identical patterns of gene expression to those of stable IPF were reported in cases of acute exacerbation [
7]. These studies have demonstrated novel candidate genes related with IPF.
The use of whole-lung tissues, however, may be a limitation of transcriptomic studies because transcriptomic changes are cell-type-specific [
8]. The pathologic characteristics of IPF include mixed features with normal lungs, alveolar inflammation, interstitial fibrosis, and honeycomb changes [
2]. Furthermore, the extent of fibrosis and inflammation varies markedly during the disease course. Accordingly, selective separation of homogenous cells from diseased lungs would be optimal, but is problematic. Among the various cell types present in lung tissue, fibroblasts are easily obtained and maintained, and the biologic properties of IPF fibroblasts differ from those of normal lung fibroblasts [
9,
10]. To further investigate the molecular mechanisms of IPF lungs, a global transcriptome analysis was conducted using fibroblasts obtained from the lung tissues of 8 patients with IPF and normal lungs of 4 subjects with localized lung lesions. The differential expression of CCL8 was validated using an additional number of fibroblasts and bronchoalveolar lavage (BAL) fluid samples from normal controls (NC), patients with IPF and those with other interstitial lung diseases including non-specific interstitial pneumonia (NSIP), hypersensitivity pneumonitis (HP), and sarcoidosis.
Discussion
In our study, 178 genes were found to be differentially expressed by the fibroblasts derived from IPF lungs. Among the top 15 genes of them, only
FGF7 and
POSTN have been previously reported as related with IPF. The new genes identified in our study may provide insight into the pathogenesis of IPF. Among the top genes,
CCL8 expression was >20-fold higher in the IPF-fibroblasts.
CCL8 gene was reported to overexpress in 26 IPF lungs compared with 11 normal lungs (
p = 0.099) in the GDS1252 dataset, as well as in 6 sarcoidosis lung compared with 6 normal lungs (
p = 0.016) in the GDS3580 dataset of NCBI GEO DataSet Browser (
https://www.ncbi.nlm.nih.gov/geo/).
CCL8 was included in the ontology categories of extracellular region, receptor binding, heparin binding, G-protein-coupled receptor binding, chemokine activity, carbohydrate derivative binding, and glycosaminoglycan binding, which are essential pathways in the pathogenesis of IPF. In validation,
CCL8 levels in BAL fluids appeared useful as a candidate marker for the differential diagnosis from NC and other chronic interstitial lung diseases. Thus, we identified several novel genes, and demonstrated the clinical relevance of
CCL8 as a candidate marker for the diagnosis and prognosis of IPF patients for the first time to the best of our knowledge.
CCL8 is secreted into the peripheral circulation. However, no correlation between
CCL8 protein levels in BAL fluids and peripheral bloods was observed. This may be due to sources of
CCL8 other than fibroblasts or individual variance in the diffusion of
CCL8 from IPF-fibroblasts into the bloodstream. Among the CC chemokine ligand family,
CCL26 expression was increased in IPF patients compared with controls (2.39-fold increase,
p = 0.048) in our study. In other studies of IPF-patients,
CCL2 are present in metaplastic epithelial cells and vascular endothelial cells [
28] and
CCL3,
CCL4, and
CCL7 expressions are elevated in BAL fluids [
29,
30]. The discrepancy between our data and these reports may be due to the presence of other sources of
CCL2,
CCL3,
CCL4, and
CCL7 in the lung.
CCL8 activates various immune cells, including mast cells, eosinophils, basophils, monocytes, T cells, and NK cells [
31]. Recently, diverse functions of
CCL8 have been discovered in infection, immunity, and allergic inflammation.
CCL8 recruits gamma/delta T cells, which preferentially express
IL-17F and synergistically enhance neutrophil chemotaxis in the presence of IL-8 [
32].
CCL8 is induced by fibroblasts and endothelial cells when co-stimulated with
IL1-β and interferon
(IFN)-γ. IFN-γ has also a synergistic effect with activation of
TLR2, TLR3 or
TLR4. The application of both
IFN-γ and dsRNA via TLRs resulted in the synergistic induction of
CCL8 expression [
33]. All of them are known mediators involved in the development of IPF. Thus,
CCL8 appears to be related with innate immune response in the development of IPF, but the exact role of
CCL8 remains to be solved in near future.
In the ontology analysis of the 178 genes, regulation of fibroblast growth factor production, extracellular region, positive regulation of the EMT, and positive regulation of cell morphogenesis were the most relevant groups. The fibroblasts used in our study expresses markers of smooth muscle differentiation, such as α-SMA [
34]. Lindahl GE and collaborators performed Gene ontology analysis using 843 differentially expressed genes between IPF and scleroderma fibroblasts and normal control fibroblasts. Enriched functional groups represent 12 broad categories as follows: anatomical structural development, regulation of cell cycle, response to stress and wounding, regulation of apoptosis, cell migration and smooth muscle contraction in upregulation and inflammatory and immune response, response to biotic stimulus, regulation of apoptosis, regulation of cell migration, regulation of cell proliferation, and regulation of I-kB/NF-kB cascade in down regulation [
35]. The enriched functional groups are in part compatible with our findings. Studies using human whole lungs also showed elevated expression of genes related to tissue remodeling/reorganization and ECM formation/degradation [
4,
5]. Interestingly, the expression of
CCL8 was positively correlated with those of
IL-8, IL-13RA2, and
CCL2, ADAMTS1, ADAMTS8, MMP10, MMP2, MMP3, TIMP2, ECM1, TGFBI, and
CLEC3B, and inversely correlated with
FN1 (
p < 0.05, respectively) (Additional file
1: Tables S7 and S8). However, the contents of the gene ontology are different between the studies. In our study, the following 18 genes were included in the extracellular region:
PF4V1,
NID1,
PTHLH,
CREG1,
TFPI,
FBN2,
RSPO3,
TSKU,
EPDR1,
BMP2,
CCL26,
POSTN,
MYOC,
CCL8,
C1QTNF9B,
CLEC18C,
FGF7, and
TUBA4A. Among them,
FGF7 [
36],
POSTN [
37],
TSKU [
38], and
TFPI [
39] have been suggested to be involved in the pathogenesis of IPF. Two genes (
PTGS2 and
RGCC) in our study were included in the fibroblast growth factor production and EMT categories. Among them,
PTGS2 is known to be involved in the development of IPF [
40]. These results indicate that more than half of the genes in our study are newly identified.
Our study had the following limitation: the small number of lung tissue samples available for microarray analysis. Therefor we used the
t-test and nonparametric TNoM scoring method to compare the differences in gene expression between NC and IPF groups because of no gene passed of less than 5% correction for multiple testing [
21]. The use of unadjusted
P-values may be less problematic than omitting informative genes in studies aimed at identifying target genes responsible for biological mechanisms [
41]. In addition, when the differentially expressed genes in our study was compared with those with Lindahl GE’s study using cultured fibroblasts (
n = 744, 10 controls and 3 IPF subjects) [
35], those with Sridhr S’s study using cultured fibroblasts (
n = 1813, 4 controls and 10 IPF subjects; GSE44723), those with Ronzani C’s study using (
n = 3, 5 controls and 5 IPF subjects; GES45686) and those with Emblom-Callahan’s study using uncultured fibroblast (
n = 1, 6 controls and 12 IPF subjects) [
8]. Only 3–10% of the differentially expressed genes were overlapped between the studies as seen in a Venn diagram in the Supplement (Additional file
6: Figure S5). This discrepancy may be due to the small numbers of fibroblasts in each study in addition to the phenotypic changes of fibroblasts during the passage in culture.
Fibroblasts derived from IPF lungs have distinct biological characteristics: a high percentage of apoptotic cells, and increased collagen, fibronectin, gelatinase B,
TIMPs,
β-FGF, and
PDGF expression (i.e., a pro-fibrotic secretory phenotype) [
42], and a reduced capacity to secrete anti-fibrotic molecules, such as prostaglandin E2 and hepatocyte growth factor [
9]. Interestingly, the above-mentioned genes were not identified in our study. Recently, a genomic expression in non-cultured fibroblasts obtained from IPF-lungs demonstrated 1,813 significantly differentially expressed transcripts from those of normal fibroblasts [
8]. When they were compared with the 178 genes of our study, only 9 genes, including
ALDH3A2, CDC42EP3, IGFBP2, MOXD1, NBEAL2, PITX1, POSTN, TMEM51, and
UBE2K, were overlapped. This may be due to biological differences between the cultured fibroblasts used in our study and the uncultured ones from IPF-lungs. However, we validated the
CCL8 concentration in BAL fluid in 86 patients with IPF and those in 41 controls and
CCL8 protein expression using immunohistochemical stain with antibodies to
CCL8 and α-SMA-positive cells in the lung tissues of IPF. Because the BAL fluids and lung tissues were obtained in un-cultured condition, so
CCL8 protein may be generated per se, not solely due to a phenotypic shift of the fibroblasts. However, replication of the result is mandatory for useful biomarkers. Another limitation was the use of control fibroblasts from the resected cancer specimens. It cannot be excluded that the gene expression in fibroblasts derived from the lungs in which cancer has developed may be different form fibroblasts derived from the normal lungs.
Epigenetic changes, such as CpG methylation and miRNA, are widespread throughout the genome in IPF-lung tissues and may regulate the expression of important genes [
43‐
46]. Although fibroblasts have been studied mainly at the steady-state RNA level, there is evidence for the abnormal regulation of mRNA translation in the fibroblasts [
47]. Furthermore, the surrounding environment influences the gene expression of fibroblasts. Recently, diseased ECM was reported to be the predominant driver of pathological gene expression, and the expression of ECM-sensitive genes is regulated primarily at the translational level [
48]. Genes encoding IPF-associated ECM proteins are targets for miR-29, which is downregulated in fibroblasts grown on IPF-derived ECM. Other candidate miRNAs are localized to the chromosome 14q32 microRNA cluster, many of which belong to the miR-154 family [
49]. Thus, the differentially expressed genes in our study should be analyzed together with global changes in CpG methylation and miRNA expression in the future.