Introduction

Idiopathic Pulmonary Fibrosis (IPF) is a progressive and chronic disorder that has been characterized by excessive wound repair and fibrosing interstitial pneumonia of unknown etiology1,2. IPF is likely driven by abnormal epithelium and propagated by dysregulated overabundant, heterogeneous fibroblast population in various states of activation3,4,5. It has been reported that transforming growth factor-β1 (TGF-β1) plays a key role in the development of IPF and TGF-β1 gene polymorphisms may affect disease progression in patients with IPF6. Survival rate of the advanced-stage IPF patients at 5 years is much lower than that of early-stage IPF7. Although emerging evidence indicates that genetic studies may hold promise in the connections between early-stage and advanced-stage disease8, no special study focuses on analyzing the difference of gene expression during different stages of IPF. Thus, a comprehensive understanding of the pathogenesis and progression mechanisms involved in IPF remains elusive.

Long non-coding RNAs (lncRNAs) are one type of non-coding RNAs with lengths greater than 200 nucleotides. The aberrant expression of lncRNAs has been linked to multiple biological processes involved in IPF9,10,11. Some studies developed IPF signatures based on lncRNAs expression analysis. For example, lncRNA H19 functions as a molecular sponge of miR-196a, which provides a novel therapeutic target for IPF12,13,14. LncRNA MALAT1, E2F1, and YBX1 may be key regulators for the pathogenesis of IPF, in the peripheral blood of IPF patients15. LncRNA sirt1 antisense was reported to inhibit TGF-β1-mediated epithelial-mesenchymal transition in vitro and alleviate the progression of IPF in vivo16. Crosstalk among non-coding RNAs plays a crucial regulatory role in the progression of IPF17. Savary et al. demonstrated that lncRNA DNM3OS regulates myofibroblast activation by giving rise to profibrotic mature miRNAs, such as miR-199a-5p/3p and miR-214-3p18. Interfering with DNM3OS function not only prevents lung fibrosis but also improves established pulmonary fibrosis. LncRNA ZFAS1 promotes the progression of pulmonary fibrosis and facilitates the proliferation and phenotypic differentiation of fibroblasts into myofibroblasts via functioning as a ceRNA19, and lncRNA Hoxaas3 promotes lung fibroblast activation and fibrosis by targeting miR-450b-5p to regulate Runx120. Huang et al. demonstrated that lncRNA FENDRR exhibits anti-fibrotic activity in pulmonary fibrosis21. Recently, we have reported several functional lncRNAs as targets for the treatment of fibrosis, and designated three pulmonary fibrosis regulatory associated lncRNAs (PFRL, PFAL, PFAR) that promote lung fibrosis by competitively binding miRNA22,23,24.

Previous studies provided insights into the crosstalk between mRNAs and lncRNAs to explore the regulatory mechanism in pulmonary fibrosis25,26. The changes in lncRNAs expression may affect the stability and translation of genes involved in lung fibrosis26. Hao et al. reported that lncRNA AP003419.16 regulates its adjacent gene RPS6KB2, which regulates the process of IPF27. Increasing studies suggest new theories for the pathogenesis and treatments of IPF25. Thus, systematically exploring the deregulation mechanism between lncRNAs and mRNAs in different stages of IPF will enhance our understanding of the progression of IPF.

In this study, we identified IPF pathogenic and progression related mRNAs and lncRNAs by constructing a pathogenic network and two dynamic progression networks for IPF. Our study revealed a core IPF regulatory sub-network centered on the lncRNA CTD-2528L19.6, which was up-regulated in the early-stage of IPF patients but down-regulated in advanced-stage IPF patients. In vitro assays, consistent with bioinformatics analysis of IPF progression in context: CTD-2528L19.6 was proven to alleviate fibroblast activation by regulating LRRC8C in MRC-5 cells and the silencing CTD-2528L19.6 promoted pulmonary fibrosis. In summary, our study highlighted that CTD-2528L19.6 may prevent the progression of IPF by alleviating fibroblast activation.

Results

Identification of IPF pathogenic signature and IPF progression signature

To evaluate gene expression patterns in the occurrence and development of IPF, we identified differentially expressed (DE) genes (DEGs) for the following three comparisons: (1) early-stage IPF vs. normal; (2) advanced-stage IPF vs. normal; (3) advanced-stage IPF vs. early-stage IPF. The methodological workflow for this study is summarized in Supplementary Fig. S1. Two lncRNAs and 48 mRNAs were significantly differentially expressed in all three comparisons (P < 0.01, Student’s t-test, Fig. 1A; Supplementary Fig. S2A). LncRNA CTD-2528L19.6, NR2F1-AS1 were significantly up-regulated in early-stage of IPF compared with normal lung and then down-regulated in advanced-stage of IPF (P < 0.01, Student’s t-test, Fig. 1B, C). Notably, the expression level of lncRNAs CTD-2528L19.6 and NR2F1-AS1 was still higher in advanced-stage IPF than that in normal lung (P < 0.01, Student’s t-test, Fig. 1B, C). Also, some genes, such as DCLRE1C, S100A8, THBS1, showed reverse tendency among the normal, early-stage and advanced-stage of IPF (P < 0.01, Student’s t-test, Supplementary Fig. S2B-D).

Fig. 1: Identification of IPF lncRNAs signatures.
figure 1

A Venn diagram of overlapping DE lncRNAs among three groups. B, C Distribution of expression levels for key lncRNAs CTD-2528L19.6 and NR2F1-AS1. *P < 0.05, **P < 0.01 in Student’s t-test. D, E Heatmaps show the global expression levels of IPF pathogenic lncRNAs (D) and IPF progression lncRNAs (E) in GSE24206.

Full size image

Then, 43 DE lncRNAs and 835 DE mRNAs that were differentially expressed in both early-stage and advanced-stage IPF compared to normal samples, were defined as IPF pathogenic signatures. IPF pathogenic lncRNAs and mRNAs showed consistently up-regulated or down-regulated expression in both early-stage and advanced-stage IPF compared with normal samples (Fig. 1D for lncRNAs, Supplementary Fig. S2E for mRNAs). 14 DE lncRNAs and 264 DE mRNAs that were differentially expressed in advanced-stage IPF compared to both normal and early-stage samples, were defined as IPF progression signature. Interestingly, all IPF progression lncRNAs were sharply up-regulated in the transition from normal lung to early-stage IPF, then there was a “rebound” tendency. The expression level of IPF progression lncRNAs was obviously down-regulated during the transition to advanced-stage IPF, but was still higher compared to normal lung (Fig. 1E). A similar “rebound” tendency was observed in IPF progression mRNAs (Supplementary Fig. S2F).

Additionally, we performed pathway enrichment analyses to investigate the potential functional mechanisms involved in pathogenesis and progression of IPF. IPF pathogenic mRNAs participated in pathways related to immune system and inflammation, including “Phagosome”, “Antigen processing and presentation”, “Cell adhesion molecules (CAMs)”, “Focal adhesion” and “ECM-receptor interaction” (FDR < 0.05, Hypergeometric test, Supplementary Fig. S2G). IPF progression mRNAs are involved in some IPF-related pathways, including “PI3K/Akt signaling” and immune system such as “B cell receptor signaling pathway”, “Chemokine signaling pathway” and “Jak-STAT signaling pathway” (FDR < 0.05, Hypergeometric test, Supplementary Fig. S2H). The above results indicated that dysregulation of lung fibrosis and immune systems plays important roles in IPF28,29.

Analysis of IPF pathogenic co-expression network

To capture the core lncRNA-mRNA regulatory module over pathogenesis in IPF, we constructed IPF pathogenic co-expression network (Fig. 2A). IPF pathogenic co-expression network exhibited the power-law behavior in “scale-free” network models (R2 = 0.99, P < 0.01, Goodness of fit test) (Fig. 2B). The lncRNAs and mRNAs with co-expression relationships were equally distributed across the genome (Fig. 2C). A sub-network centered on CTD-2528L19.6 emerged in the IPF pathogenic co-expression network. In the dataset of GSE24206, expression of CTD-2528L19.6 was increased in early-stage of IPF and decreased in advanced-stage IPF. CTD-2528L19.6 showed a similar differential expression tendency in three independent datasets (GSE10667, SRP10849 and GSE73854, Supplementary Fig. S3A-C). Some hub lncRNAs, such as LINC00342, RP11-1008C21.1, RP11-363E7.4 and TP53TG1, may regulate fibrogenesis by regulating mRNAs CD4, CCNB1, XAF1 and PAK1 in fibroblast related gene sets, respectively (Supplementary Fig. S4). Besides, mRNAs up-regulated in the IPF pathogenic network were involved in “Primary immunodeficiency”, “Axon guidance”, and “T cell receptor signaling pathway” (FDR < 0.05, Hypergeometric test, Fig. 2D). mRNAs down-regulated in the network were involved in pathways in cancer and cancer-related signaling pathways (FDR < 0.05, Hypergeometric test, Fig. 2E).

Fig. 2: LncRNA-mRNA co-expression networks during the pathogenesis of IPF.
figure 2

A IPF pathogenic lncRNAs (diamond)-mRNAs (oval) co-expression network in IPF. The nodes marked with red color represent the up-regulated IPF pathogenic genes. The nodes marked with blue color represent the down-regulated IPF pathogenic genes. The solid lines (dotted lines) represent the positive (negative) correlation between mRNAs and lncRNAs (|r| >0.8, P < 0.01, Pearson Correlation Test). B Distribution of the degree of genes in the IPF pathogenic network. C Circos plots display chromosomal interactions between co-expressed lncRNAs and mRNAs. The location of lncRNA CTD-2528L19.6 was marked with red arrows. D, E KEGG pathways enrichment with IPF pathogenic up-regulated mRNAs (D) or down-regulated mRNAs (E). FDR < 0.05, Hypergeometric test. Size of the bubble indicates the number of genes in the corresponding annotation. Color shade corresponds to the -log10(p) value. The closer the color is to red, the more significant the enrichment is.

Full size image

Dynamic IPF progression lncRNA-mRNA co-expression networks

To explore the dynamic regulatory mechanism in the IPF progression, IPF early-stage specific and advanced-stage specific lncRNA-mRNA co-expression networks were constructed, respectively. IPF early-stage specific network exhibited the power-law behavior in “scale-free” network models (Fig. 3A; R2 = 0.99, P < 0.01, Goodness of fit test, Fig. 3B), and CTD-2528L19.6 was co-expressed with 15 mRNAs in this network (Fig. 3C). IPF advanced-stage specific network also exhibited the power-law behavior in “scale-free” network models (Fig. 3D; R2 = 0.99, P < 0.01, Goodness of fit test, Fig. 3E), and CTD-2528L19.6 was co-expressed with 6 mRNAs in this network (Fig. 3F). LRRC8C, which was correlated with CTD-2528L19.6 in both networks, has been approved to be an IPF biomarker30 and is involved in multiple fibrosis related gene sets (Fig. 4A, B). These results emphasized the regulatory relationship between CTD-2528L19.6 and LRRC8C in the dynamic progression of IPF.

Fig. 3: IPF co-expression networks in early-stage and advanced-stage of IPF.
figure 3

A IPF progression lncRNAs (diamond)-mRNAs (oval) co-expression network in early-stage IPF. The solid lines (dotted lines) represent the positive (negative) correlation between mRNAs and lncRNAs (|r| >0.8, P < 0.01, Pearson Correlation Test). The nodes marked with green color represent the early-stage IPF specific genes. The nodes marked with yellow color represent the common genes in early-stage and advanced-stage IPF progression network. Several important IPF biomarkers and fibroblast related mRNAs that co-expressed with CTD-2528L19.6 were marked with red arrows. B, C Distribution of the degree of genes in early-stage IPF progression network. D IPF progression lncRNAs (diamond)-mRNAs (oval) co-expression network in advanced-stage IPF. The nodes marked with blue color represent specific genes in advanced-stage IPF. The nodes marked with yellow color represent the common genes in early-stage and advanced-stage IPF progression network. Several important IPF biomarkers and fibroblast related mRNAs that co-expressed with CTD-2528L19.6 were marked with red arrows. E, F Distribution of the degree of genes in the advanced-stage IPF progression network.

Full size image
Fig. 4: Illustration of fibrosis related gene sets that contain the mRNAs co-expressed with lncRNAs in the progression of IPF.
figure 4

A, B Sankey diagrams display the connection between CTD-2528L19.6 correlated mRNAs and fibrosis related gene sets in early-stage IPF (A) and advanced-stage IPF (B). The expression of mRNAs in diagrams are correlated with the expression of CTD-2528L19.6 (|r| >0.8, P < 0.01, Pearson Correlation Test). Path connecting an mRNA to a fibrosis related gene set represents the mRNA participates in the gene set.

Full size image

mRNAs whose expression correlated with CTD-2528L19.6 are members of many fibrosis related biological processes. In early-stage IPF co-expression network, mRNAs S100A8 (r = 0.85, P = 6.99E-3, Pearson correlation test) and TLR7 (r = 0.89, P = 3.01E-3, Pearson correlation test) that co-expressed with CTD-2528L19.6 have been reported to be novel biomarkers of IPF fibrosis31,32. DDIT4, whose expression was significantly positively co-expressed with CTD-2528L19.6 (r = 0.87, P = 4.24E-3, Pearson correlation test), was contained in 42% of fibrosis related gene sets, which were annotated in the MSigDB database (Fig. 4A). In advanced-stage co-expression network, lncRNAs RP11-196G18.23, LINC00894 and CTD-2528L19.6 were frequently co-expressed with other genes. IPF biomarker LRRC8C was positively co-expressed with RP11-196G18.23 (r = 0.85, P = 3.67E-3, Pearson correlation test) and CTD-2528L19.6 (r = 0.81, P = 8.71E-3, Pearson correlation test). THBS1, which showed significant positive co-expression with CTD-2528L19.6 (r = 0.81, P = 8.7E-3, Pearson correlation test), was contained in 84% fibrosis related gene sets (Fig. 4B). These results suggest that CTD-2528L19.6 affects lung fibrosis in early-stage and advanced-stage of IPF by regulating different fibrosis related genes.

LncRNA CTD-2528L19.6 expression is negatively correlated with fibroblast genes

To investigate the association between CTD-2528L19.6 and fibrosis, we performed Pearson correlation coefficient test between CTD-2528L19.6 and six IPF cell markers at single-cell level33. The results from Fig. 5 showed negative correlations between CTD-2528L19.6 and IPF cell markers KRT5 (r = −0.51, P = 0.036, Fig. 5A), NGFR (r = −0.57, P = 0.018, Fig. 5B), and strong negative correlations between CTD-2528L19.6 and IPF cell markers MUC5AC (r = −0.75, P = 4.94E-4, Fig. 5C), MUC5B (r = −0.76, P = 4.41E-4, Fig. 5D), SCGB1A1 (r = −0.65, P = 4.69E-3, Fig. 5E) and SPDEF (r = −0.64, P = 6.12E-3, Fig. 5F) in IPF patients. Notably, three of the six IPF cell makers (KRT5, NGFR, MUC5B) showed significantly up-regulation in advanced-stage of IPF and the other two makers (SCGB1A1, MUC5AC) were marginally significantly up-expressed in advanced-stage of IPF compared with early-stage of IPF (Supplementary Fig. S5).

Fig. 5: Correlation between CTD-2528L19.6 and six IPF cell markers in IPF patients.
figure 5

A-F The correlation relationship between CTD-2528L19.6 and six IPF cell markers (KRT5, NGFR, MUC5AC, MUC5B, SCGB1A1, SPDEF) was tested by Pearson correlation test. The fitting model was constructed using the “lm” method function in R package ggplot2.

Full size image

Silencing lncRNA CTD-2528L19.6 promotes fibroblast activation in MRC-5 cells

To determine the effect of lncRNA CTD-2528L19.6 in lung fibrosis, we first examined the localization of CTD-2528L19.6 in human lung fibroblast MRC-5 cells by fluorescence in situ hybridization (FISH). FISH results showed that CTD-2528L19.6 was expressed in both the nucleus and the cytoplasm (Fig. 6A). Then, we used a smart silencer against CTD-2528L19.6 (SSi-CTD) to explore the effects of CTD-2528L19.6 knockdown on the proliferation, migration and trans-differentiation of MRC-5 cells (Fig. 6B). As illustrated in Fig. 6C, SSi-CTD resulted in the up-regulation of Fn1 and Collagen 1α1 at mRNA levels. In addition, silencing CTD-2528L19.6 promoted the expression of Fn1 and Collagen I at protein levels (Fig. 6D). Meanwhile, as illustrated in Fig. 6E–G, SSi-CTD apparently increased the ability of cell migration, proliferation and facilitated the trans-differentiation of fibroblasts into myofibroblasts. The above results suggest that silencing lncRNA CTD-2528L19.6 can promote the activation of MRC-5 cells.

Fig. 6: Silencing CTD-2528L19.6 promotes fibroblast activation of MRC-5 cells.
figure 6

A Fluorescence in situ hybridization detects the localization of CTD-2528L19.6 in MRC-5 cells. For clarity, CTD-2528L19.6 was abbreviated as CTD in the figures. 18 S and U6 are the markers of cytoplasm and nuclei, respectively. Blue nuclei are stained with DAPI. Scale bar: 50 μm. B, C The expression of CTD-2528L19.6, Fn1, Collagen 1α1 after silencing CTD-2528L19.6 in MRC-5 cells were detected by qRT-PCR. n = 4. D Western blot was used to detect the expression of Fn1 and Collagen I after SSi-CTD transfection. n = 6. Wound-healing migration assay (E) and EdU proliferation assay (F) were used to assay the effect of SSi-CTD on migration and proliferation in MRC-5 cells. n = 4. Scale bars, 200 μm in E and 50 μm in F. G Immunofluorescence was applied to examine the fibroblasts into myofibroblasts transition after transfecting SSi-CTD. Scale bar, 20 μm. *P < 0.05; **P < 0.01.

Full size image

Overexpression of CTD-2528L19.6 alleviates fibroblast activation induced by TGF-β1

Next, we transfected CTD-2528L19.6 overexpression plasmid into MRC-5 cells to determine the effect of CTD-2528L19.6 on fibroblast activation (Fig. 7A). As shown in Fig. 7B, TGF-β1 promoted the expression of Fn1 and Collagen 1α1, which was inhibited by overexpression of CTD-2528L19.6. Furthermore, western blot assay showed that forced expression of CTD-2528L19.6 inhibited the up-regulation of Fn1 and Collagen I induced by TGF-β1 at protein levels (Fig. 7C). Fibrosis occurs with abnormal activation and excessive proliferation of fibroblasts. Wound-healing migration assay showed that overexpression of CTD-2528L19.6 attenuated TGF-β1-induced cell migration and inhibited wound healing (Fig. 7D). Through the proliferation experiment of the EdU proliferation assay, we found that overexpression of CTD-2528L19.6 suppressed the MRC-5 proliferation induced by TGF-β1 (Fig. 7E). More importantly, we detected the expression of α-SMA, a marker for fibroblast-myofibroblast transition, in MRC-5 cells by immunofluorescence assay. As shown in Fig. 7F, TGF-β1 could significantly increase α-SMA expression, whereas this effect was inhibited by the overexpression of CTD-2528L19.6. These results suggest that overexpression of lncRNA CTD-2528L19.6 can suppress activation of MRC-5 cells induced by TGF-β1.

Fig. 7: LncRNA CTD-2528L19.6 inhibits the TGF-β1-induced fibroblast activation in MRC-5 cells.
figure 7

qRT-PCR analysis of the expression of CTD-2528L19.6 (A) and Fn1, Collagen 1α1 (B) in MRC-5 cells after transfection CTD-2528L19.6 under TGF-β1 induced fibroblast activation. n = 4. C Western blot results demonstrate that the overexpression of CTD-2528L19.6 inhibits the pro-fibrotic effect of TGF-β1 in MRC-5. n = 9. Wound-healing assay (D) and EdU proliferation assay (E) reveals the inhibitory effects of CTD-2528L19.6 on TGF-β1-induced migration and proliferation in MRC-5 cells. n = 4. Scale bars, 200 μm in D and 50 μm in E. F Fibroblast to myofibroblast transition was observed by immunofluorescence. Scale bar, 20 μm. *P < 0.05; **P < 0.01.

Full size image

Silencing LRRC8C alleviated the inhibitory effect of CTD-2528L19.6 on fibroblast activation

In the co-expression network, expression of CTD-2528L19.6 was positively correlated with LRRC8C in both early-stage (r = 0.95, P = 2.82E-4, Pearson correlation test, Fig. 3A) and advanced-stage IPF (r = 0.81, P = 8.71E-3, Pearson correlation test, Fig. 3D). Therefore, we performed qRT-PCR assay to examine the effect of CTD-2528L19.6 on LRRC8C. Results showed that knockdown of CTD-2528L19.6 inhibited the expression of LRRC8C in MRC-5 cells (Fig. 8A), whereas overexpression of CTD-2528L19.6 promoted LRRC8C expression at mRNA level (Fig. 8B).

Fig. 8: LncRNA CTD-2528L19.6 regulates LRRC8C in MRC-5 cells.
figure 8

A, B qRT-PCR shows that silencing or forced overexpression of CTD-2528L19.6 inhibited (A) or promoted (B) the expression of LRRC8C. C-E Silencing of LRRC8C restored the expression levels of Fn1 and Collagen 1α1 in TGF-β1-treated MRC-5 cells. n = 4; *P < 0.05, **P < 0.01.

Full size image

Then, we constructed three small interference RNAs (siRNAs) against LRRC8C to further explore the effects of LRRC8C knockdown on collagen deposition. qRT-PCR results showed that all three siRNAs could inhibit the expression of LRRC8C in MRC-5 cells (Fig. 8C). As shown in Fig. 8D, E, silencing of LRRC8C alleviated the inhibitory effect of CTD-2528L19.6 on Fn1 and Collagen 1α1 in TGF-β1-treated MRC-5 cells.

Discussion

In this study, by systematical analysis of transcriptome profile, we revealed a set of differentially expressed lncRNAs in IPF compared with normal lung tissues. Interestingly, some lncRNAs were up-regulated in the transition from normal lung to early-stage IPF, subsequently down-regulated during the transition to advanced-stage IPF. The lncRNA-mRNA co-expression network-based transcriptome analysis revealed a key fibrosis regulator lncRNA CTD-2528L19.6 in IPF. We demonstrated that lncRNA CTD-2528L19.6 participates in the pathogenesis and progression of IPF in different stages with different mechanisms. In MRC-5 cells, up-regulation of CTD-2528L19.6 prevented the fibrosis progression of IPF by alleviating fibroblast activation. LncRNA CTD-2528L19.6 inhibited fibroblast activation through regulating the expression of LRRC8C in vitro assays. Silencing of CTD-2528L19.6 promoted fibroblast activation, which may be a novel biomarker for IPF.

Expression of CTD-2528L19.6 was increased in early-stage of IPF and decreased in advanced-stage IPF. The dysregulation and effect of CTD-2528L19.6 in IPF is similar to the MMP13, whose expression is up-regulated during two stages after fibrosis whereas knockdown of MMP13 could aggravate the progression of lung fibrosis34,35. Thus, we inferred that, at the early-stage of IPF, the adaptive up-regulating CTD-2528L19.6 is sufficiently to partially offset the activation of fibroblasts. However, with long-term stress, a sustained increase of the fibroblasts beyond a certain threshold could counteract the function of CTD-2528L19.6 up-regulation, leading to the development of fibroblast activation. Also, the biological experiments in our work revealed that enhanced expression of lncRNA CTD-2528L19.6 prevents the activation of fibroblast, which confirmed our hypothesis.

CTD-2528L19.6 may prevent the progression of IPF and exhibit anti-fibrotic activity by regulating the expression of fibrosis associated mRNAs. In IPF pathogenesis network, mRNA DCLRE1C is a member of the non-homologous end joining pathways, and deleterious mutations which can cause severe combined immunodeficiency36. In early-stage IPF progression networks, mRNA S100A8 has been reported in bronchoalveolar lavage fluid as a biomarker of IPF fibrosis31. In advanced-stage IPF progression network, mRNA THBS1 is co-expressed with CTD-2528L19.6, which activates the reproduction of fibroblast cells in mice37. TLR7 has been revealed to be a novel IPF associated gene and biomarker32, and is dysregulated in IPF pathogenesis38. In IPF progression network, genes S100A8 and TLR7, which were positively correlated with the expression of CTD-2528L19.6, act as immune cell markers32. The crosstalk analysis between mRNAs and lncRNAs highlights the important role of immune regulation in IPF, which warrants further detail in vitro or in vivo biological experiments. The functional relationship between LRRC8C and CTD-2528L19.6 in fibrosis was proved in vitro assays. LncRNA CTD-2528L19.6 inhibited fibroblast activation through regulating the expression of LRRC8C. Limited by the sample amount of IPF data set, we did not control the FDR when performing differential expression analysis. However, the roles of CTD-2528L19.6 in IPF predicted by bioinformatics analysis could be validated by cellular experiments and some correlated genes were proved to be involved in IPF by functional annotation. Another key lncRNA NR2F1-AS1 with many transcripts was not tested in this study, which is worthy of our follow-up in-depth research. Notably, the lncRNA TP53TG1, which is most frequently co-expressed with mRNAs in IPF pathogenesis network, also deserves our following study.

Overall, we predicted IPF pathogenic and progression related lncRNAs and mRNAs through analyzing transcriptional profile, then constructed an IPF pathogenic lncRNA-mRNA co-expression network and dynamic IPF progression networks. A key lncRNA CTD-2528L19.6 was captured to regulate the pathogenesis and progression of IPF as a potential biomarker. CTD-2528L19.6 may regulate fibroblast activation in IPF progression by mediating the expression of some mRNAs, such as LRRC8C, DDIT4, THBS1, S100A8 and TLR7. Our results suggest that during early-stage, CTD-2528L19.6 was up-regulated to prevent IPF, while CTD-2528L19.6 was down-regulated to promote fibroblast activation from early-stage to advanced-stage transition. This finding provides a framework for designing interventions that could prevent the development or progression of fibrosis at various stages of IPF development. The present research also reveals some lncRNAs as potential IPF biomarkers.

Materials and methods

Microarray data and processing

Gene expression dataset (GSE24206)39, with 17 IPF (8 early-stage IPF and 9 advanced-stage IPF) and 6 normal lung tissues, was downloaded from the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) (Supplementary Fig. S1A). Expression for the Affymetrix U133 Plus 2.0 GeneChips was estimated by robust multi-array average then log2 transformed. Data were filtered prior to analysis: If a gene was mapped to multiple probe sets, the expression value for the gene was generated by averaging. Probes that did not map to any Gene ID or map to multiple Gene IDs were deleted40.

GSE10667 and GSE73854 were collected from the GEO and SRP10849 was download from The Sequence Read Archive (SRA, https://www.ncbi.nlm.nih.gov/sra/).

Identification of IPF signature

We performed three comparisons using Student’s t-test: (1) early-stage IPF vs. normal; (2) advanced-stage IPF vs. normal; (3) advanced-stage IPF vs. early-stage IPF. We selected genes with P < 0.01 as DEGs, including DE lncRNAs and DE mRNAs. The DEGs derived from both (1) and (2) were defined as IPF pathogenic signatures. And, the DEGs detected from both (1) and (3) were defined as IPF progression signatures, as showed in Supplementary Fig. S1B.

Co-expression network construction

Pearson correlation test was used to test the expression correlation between differential expressed lncRNAs and mRNAs. The lncRNA-mRNA pairs with the absolute value of correlation coefficient r > 0.8 and P-value < 0.01 were included in the co-expression network.

Then, a pathogenic co-expression network between IPF pathogenic lncRNAs and mRNAs was constructed in IPF samples. Two IPF progression co-expression networks between IPF progression lncRNAs and mRNAs were constructed in early-stage IPF and advanced-stage IPF samples, respectively (Supplementary Fig. S1C).

Fibrosis related gene sets and IPF cell markers

Fibrosis related gene sets were collected from the Molecular Signatures Database (MSigDB, https://www.gsea-msigdb.org/gsea/msigdb), including fibroblast related gene sets. These annotated sets of genes involved in fibrosis related biochemical pathways, signaling cascades, expression profiles from research publications, and other biological concepts41. IPF cell markers (KRT5, MUC5AC, MUC5B, NGFR, SCGB1A1, SPDEF) were collected from the CellMarker database (http://biocc.hrbmu.edu.cn/CellMarker/)42.

Pathway enrichment analysis

Pathway information was downloaded from Kyoto Encyclopedia of Genes and Genomes (KEGG) on August 17, 2017 (http://www.genome.jp/kegg/), consists of six top categories (09100 to 09160) for KEGG pathway maps. The hypergeometric distribution model was used to test whether KEGG pathways were enriched with genes in the co-expression network. The P-value was adjusted by the Benjamini-Hochberg (BH) procedure. The pathways with a false discovery rate (FDR) less than 0.05 were considered to be significant.

MRC-5 cell culture and transfection

MRC-5 cell line was purchased from Cell Bank of Chinese Academy of Sciences, and cultured in 89% DMEM (Biological Industries, Israel) containing 10% FBS (Biological Industries, Cromwell, CT, USA) and 1% Penicillin-Streptomycin-Amphotericin B (Solarbio, China) then plated in a standard humidity incubator at 37 °C with 5% CO2. PCR was used to synthesize the full-length of lncRNA CTD-2528L19.6 and inserted into the pcDNA3.1 vector. pcDNA3.1 empty vector was used as control. LncRNA smart silencer SSi-CTD/SSi-NC (negative control) was constructed by RiboBio Tech (Guangzhou, China). CTD-2528L19.6 plasmid or SSi-CTD/SSi-NC and Lipofectamine 2000 (Invitrogen, Carlsbad, CA, U.S.A.) transfection reagents were mixed with Opti-MEM (GIBCO, Grand Island, NY, U.S.A.) in serum-free medium, respectively. Then incubated for 5 min and protected from the light. Mixed two liquids and placed for 15 min at room temperature, then added to the cell plate and cultured in incubator. After transfection 6 h replaced with a normal culture medium. After cells were cultured with 10 ng/ml TGF-β1 (Sigma-Aldrich, U.S.A.) for 48 h and prepared for further analysis (Supplementary Fig. S1D).

Fluorescence in situ hybridization

Fluorescence in situ hybridization (FISH) was performed using the lncRNA FISH Probe Mix kit (Ribobio, Guangzhou, China). MRC-5 cells were fixed in 4% paraformaldehyde (Solarbio, China) for 10 min and then added 1 ml precooling permeable solution (5 μl Triton-X + 1 ml PBS) for 5 min at 4 °C. Added 200 μl pre-hybridization buffer into each well then incubated at 37 °C for 30 min. Then discarded the pre-hybridization buffer and added 150 μl hybridization buffer with lncRNA FISH Probe Mix, hybridized overnight at 37 °C. Later washed with buffer I (4x SSC), buffer II (2x SSC) and buffer III (1x SSC) at 42 °C. The nuclei were stained with DAPI (Roche Molecular Biochemicals, Basel, Switzerland). The images were taken under the inverted fluorescence microscope (Olympus, IX73, Japan).

RNA extraction and quantitative RT-PCR

The total RNA of cells was extracted with TRIzol reagent. The concentration and purity of extracted RNA were determined by NanoDrop 8000 (Thermo, U.S.A). RNA reverse transcription for cDNA by using 5× All-in-One RT Master Mix. cDNA was used to detect the relative expression of mRNA by real-time quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) in the presence of SYBR Green fluorescent dye (Applied Biosystems, Foster City, CA). The relative expression level was calculated based on the Ct values and GAPDH was used as a normalized control.

Protein extraction and Western blot

Total proteins of MRC-5 cells were extracted and lysed with RIPA buffer (Beyotime, Jiangsu, China) containing protease inhibitor. Protein samples were separated on 8% sodium dodecyl sulfate-polyacrylamide gel and transferred to pure nitrocellulose (Pall Life Sciences, Ann Arbor, MI, USA) after electrophoresis. The membranes were probed with primary antibodies against β-actin (1:500, 66009-1-Ig, Proteintech, Wuhan, China), Fn1 (1:500, 15613-1-AP, Proteintech, Wuhan, China) and Collagen I (1:500, WL0088, Wanlei, Liaoning, China). The protein expression levels were detected by the Odyssey Infrared Imaging System (Odyssey CLX, Biosciences, USA).

Immunofluorescence staining

After transfected and treated with TGF-β1, MRC-5 cells were fixed by 4% paraformaldehyde at room temperature for 30 min, and then permeabilized with 0.1% Triton X-100. Then, the cells were blocked with 50% goat serum and incubated with anti-Rabbit α-SMA antibody (1:100, ab7817, Abcam). Finally, cells were incubated with anti-rabbit IgG (H + L) (Alexa Fluor 488 Conjugate) (1:500, 4412, CST). Nuclei were stained with DAPI. Immunofluorescence images were photographed under fluorescence microscope.

EdU cell proliferation assay

According to the Cell-Light EdU DNA cell proliferation kit (RiboBio, Guangzhou, China) instructions, the cells were incubated with 200 μl, 50 μM EdU solution. The cells were fixed with 4% paraformaldehyde and incubated in a 2 mg/ml glycine shaker. Afterwards, cells were stained with Apollo solution staining for proliferating cells. Nuclei were stained by DAPI. Images were finally photographed under the inverted fluorescence microscope.

Wound-healing migration assay

Wound-healing migration assay was performed as previously described24. MRC-5 cells were seeded in 6-well plates and grown until formation of confluent monolayer. Then cells were gently scratched with a 10 μl pipette tip. The scratch healing areas were observed and photographed under microscope (×20 objective). Afterwards, the cells were transfected and added with TGF-β1. The images were taken by the Nikon Ts100 microscope (Nikon, Tokyo, Japan) and analyzed by using Image-J. First, the images resolution was changed to 8-bit and adjust the contrast of the enhanced image. After that, smoothed the scratch edge then found edge again. Finally, set the appropriate image threshold and measured the scratch area. Wound healing rate (%)= 1-(24 h or 48 h scratch area /0 h scratch area)*100.

Statistics and analysis

Student’s t-test was used for statistical analysis between two groups. The fitting regression was constructed to modeling the relationship between CTD-2528L19.6 and IPF cell markers with linear fitting using formula = y~x with method = “lm” in R package ggplot2. All bioinformatics analyses were carried out using R software version 3.6.0 (http://www.r-project.org/). Cytoscape software (version 3.6.0) was used to visualize the lncRNA-mRNA co-expression networks. All experimental data were presented as mean ± SEM. One-way analysis of variance (ANOVA) was used to determine the statistically significant differences among multiple groups. Statistical analyses were carried out using GraphPad Prism 8.0.