Study design and patients

This retrospective observational cohort study was conducted with IPF patients hospitalized in the Second Xiangya Hospital of Central South University) in Hunan, China, from January 2014 to October 2018. Institutional approval was provided by the Second Xiangya Hospital of Central South University Biomedical Research Ethics Committee (Hunan, China). Written informed consent was waived because of the retrospective observational design. All patient data were anonymously recorded to ensure confidentiality. Patients were admitted to the hospital with a diagnosis of IPF based on the 2011 IPF definition¹⁴. The diagnosis of IPF requires the following: 1) Exclusion of other known causes of interstitial lung disease (ILD); 2) The presence of a UIP pattern on high-resolution computed tomography (HRCT) in patients not subjected to surgical lung biopsy; and 3) Specific combinations of HRCT and surgical lung biopsy pattern in patients subjected to surgical lung biopsy.

Smoking-related information was collected from all patients and a telephone follow-up to inquire about their current living conditions was conducted. Never smokers were defined as those who smoked <100 cigarettes in their lifetime. The primary outcome was mortality.

Statistical analysis of the Bioscience database

RNA sequencing of idiopathic pulmonary fibrosis lung and healthy control normal tissues was downloaded from GEO (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE134692). The batch effect of data was removed by the R package limma and expression was log2CPM normalized. We visualized the distribution of data between the healthy control group and IPF patient group using the t-SNE method to perform quality control, which included 46 IPF and 26 normal lung tissues. The statistics of the two-group comparisons were compared with the Wilcoxon test, and the correlation method was Spearman’s test.

Experimental pulmonary fibrosis model

Animal preparation

C57BL/6 mice (male, 6–8 weeks old) were acquired from the Experimental Animal Center of the Second Xiangya Hospital of Central South University. The mice were used for experiments and controls were fed for 1 week under the experimental conditions. Establishment of a bleomycin (BLM)-induced pulmonary fibrosis model in mice involved a lung fibrosis model established after intratracheal injection of BLM (5 mg/kg, Sigma) under anesthesia with intraperitoneal injection of 2% pentobarbital by volume weight. Mice were injected with normal saline as a control. According to the different treatment groups after 72 h, the transplantation group mice were intratracheally injected with 50 μL shLINC00665 (GenePharma, Shanghai), and the transplantation control group mice were injected with 50 μL shNC (GenePharma, Shanghai). The samples were collected 7, 14 and 28 days later.

Extraction, isolation, identification, culture, and treatment of lung fibroblasts (LFBs)

C57BL/6 mice (male, 6–8 weeks old) were sacrificed under anesthesia and painless conditions, and their lungs were rapidly collected in a sterile state. After rinsing with PBS 3 times, the samples were cut into 1 mm³ tissue blocks, which were evenly spread at the bottom of a 10 cm petri dish and digested with 0.25% trypsin. The isolated cells were cultured in 10% bovine serum DMEM with 100 IU/mL penicillin and 100 IU/mL streptomycin in a humidified incubator at 37^oC with 5% CO₂ and a constant temperature for 2 to 3 days. When the cells reached 90% confluence, they were digested and subcultured. After 4 generations of culture, the cells showed typical spindle shapes. Vimentin was strongly positive in the cytoplasm, as demonstrated by immunocytochemistry, and the cells were confirmed to be fibroblasts for further experiments. LFBs were cultured at 37°C with 5% CO₂. Recombinant TGF-β1 (10 ng/mL, Sigma) was used to establish a pulmonary fibrosis model in vitro. Cells were treated with an inducer of ER stress, thapsigargin (TG, 0.2 μM, T9033, Sigma) to regulate the expression of XBP-1 in lung fibroblasts¹⁵.

Preparation of CSE

CSE was prepared as previously described^16-18. One cigarette (Baisha, China Tobacco Hunan Industrial Corporation, Changsha, Hunan, China) burned yielded 25 mg of tar, 10 mg of CO and 0.1 mg of nicotine under a standard smoking regimen, and the smoke was passed through 20 mL of phosphate-buffered saline via a vacuum pump. This 100% CSE solution was adjusted to 7.2–7.4 and filtered through a 0.22 m membrane filter to remove large particles and bacteria before use. Then, CSE was diluted with PBS to obtain concentrations of 1% (0.1 M) and 10% (1.0 M) and these samples were used for the cell experiments within a period no longer than 30 min after preparation.

Cell transfection

According to the lncRNA database, the gene sequence of LINC00665 was obtained, and double-stranded DNA and iRNA vectors were synthesized. Then, the double-digested lentivirus vector was linked to the cloned double-stranded DNA. The constructed vector and lentivirus packaging plasmid vector pcDNA3.1 were coinfected in 293T packing cells by Lipofectamine 2000, and the virus genome was generated. It was determined by the pore-release method. The prepared lentivirus vector was stored at -80°C. Plasmid construction and lentivirus packaging were completed by Hunan Pratze Biotechnology Co., Ltd.

Immunofluorescence

The cells in each group were inoculated in 24-well plates at 1×10⁴ cells/well and cultured to 50% confluence. The original medium was discarded and replaced with serum-free basic medium for 24 h. We extracted the original culture medium, washed the cells with PBS for 3 times, fixed with 4% paraformaldehyde for 15 minutes, and washed with PBS for 3 times, 5 minutes each time. We added Triton X-100 (0.3%) to the membrane for 15 min, and washed with PBS for 3 times, 5 min each time. Goat serum was added for 1 h. The sealing solution was aspirated without rinsing, and mouse anti-rat α-SMA was added and incubated at 4°C overnight. The next day, the primary antibody was recovered, and the samples were rinsed with PBS 3 times for 5 min each. The immunofluorescent FITC-labelled goat anti-mouse IgG secondary antibody was added dropwise and incubated at room temperature for 1–2 h. The fluorescent secondary antibody was recovered, and then rinsed 3 times with PBS for 5 min each time. After rinsing with PBS, DAPI staining solution was added for 10 min at room temperature and then washed 3 times with PBS for 5 min each time.

Relative luciferase activity

MiR-214-3p was predicted to be a potential biological target of LINC00665 by StarBase. The double luciferase reporter assay further confirmed its targeting relationship. The 3’ untranslated regions (UTR) of LINC00665 and miR-214-3p were identified, and then, plasmids containing potential miR-214-3p binding sites and LINC00665 wild-type (WT) or mutant (MUT) sites were cotransfected with miR-NC or miR-214-3p mimic. The LINC0065 WT group and LINC00665 MUT group were compared to verify whether miR-214-3p was the direct target of LINC00665 through relative luciferase activity. Based on the binding sequence of the target gene 3’UTR to miR-214-3p, the design was generated using the miRNABase database. The XBP-1 target gene fragment was synthesized by PCR, and the synthesized target gene fragment was cloned into the XhoI and NotI sites, while one end of the synthesized gene fragment was cloned into the pluC-UTR luciferase reporter gene expression vector. The plasmid was named pluC-XBP-1-3’UTR, and the construct was confirmed by sequencing. However, the seed region sequence was mutated by point mutation. The two groups of luciferase expression vectors were named. The luciferase reporter gene expression vectors were wild-type (XBP-1-WT) and mutant (XBP-1-MUT).

A dual luciferase assay kit (Promega) was used. Enzyme activity was determined according to the instructions. XBP-1 was cloned into a 3’UTR plasmid or a 3’UTR vector. Then, 293T cells were cotransfected with miR-214-3p mimics and controls. Twenty-four hours after transfection, the cells were collected in cold PBS solution, and 100 μL of cell lysate buffer was added to the cells. After centrifugation at 4°C, 20 μL of supernatant was added to 96-well plates, and 30 μL of LAR II was added to mix. Determination of firefly luciferase activity was performed on a microplate reader (F value). The luciferase activity (R value) of Renilla was determined by adding 30 μL of Stop & Glo reaction solution. The ratio of the F value to the R value reflected the comparative expression level of the reporter gene.

RNA binding protein immunoprecipitation (RIP) assay

The binding relationship between endogenous LINC00665 and miR-214-3p was studied by the RIP method. The lysate transfected with LFB was treated with RIP buffer containing magnetic beads coupled with anti Ago2 antibody (Millipore, Billerica, MA, USA), and the negative IgG was used as a control. We washed the pellets with washing buffer, and incubate with 0.1% SDS/protease K to remove protein. Finally, real time quantitative PCR (qRT-PCR) was used to detect their expression levels.

qRT‑PCR

Total RNA was extracted from lung tissue or cultured LFBs with TRIzol reagent (Invitrogen, USA). A reverse transcriptase kit was used to synthesize cDNA. An ABI 7500 Fast Real-Time PCR instrument (California, USA) was used to determine gene expression with SYBR Green I.

Western blot analysis

After each group LFBs were treated at a certain concentration for 48 h, followed by a mixture of the BRIPA lysis buffer and a protease inhibitor cocktail (GenDEPOT; P3100, P3200) to prepare the whole cell proteins.

We measured the cell protein concentrations with a bicinchoninic acid assay kit. Protein samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. We incubated the membrane with the following antigen recognition antibody at 4^oC overnight: α-SMA, fibronectin, collagen I, E-cadherin and XBP-1,blocking with 5% nonfat milk later. Incubation of membrane with polyclonal anti rabbit/mouse or goat HRP binding secondary antibody (1:5000 dilution; LI-COR Biosciences, Lincoln, NE, USA) for 1 h at room temperature was done, after washing with Tris-buffered saline Tween. Next, we used Tris-buffered saline-Tween to wash the membranes 3 times, and using an ECL detection solution to detect. We used the Odyssey infrared imaging system (Gene Company, Beijing, China) to scan and quantify the band intensities.

Statistical analysis

We used SPSS 22.0 (SPSS, Inc., Chicago,IL, USA) and GraphPad Prism 8.0 (GraphPad Software Inc. USA) to perform the data analysis. We used Kaplan-Meier plots and log-rank tests to compare the survival rate of each group. We calculated an independent predictor of mortality with stepwise binary logistic regression variables for the regression values of p<0.05 (one variable was entered when p<0.05, and one was deleted when p>0.10). The odds ratio (OR), p-value and 95% CI were used to represent the results. The data are expressed as the mean ± SEM, except where otherwise indicated. The Kruskal-Wallis test was used to determine the difference detection for different groups, and the Mann-Whitney U test was used to analyze the differences between individual variables from the 2 groups. A p<0.05 was considered significant.

The relationship between smoking status and survival rate in IPF patients, and statistical analysis of the Bioscience database

We retrospectively analyzed IPF patients hospitalized in the Department of Respiratory and Critical Care Medicine of Xiangya Second Hospital of Central South University from January 2014 to October 2018. There were 87 patients in total, 41 (47.13%) of whom had a smoking history. Furthermore, these patients were followed up and 25 patients were lost. Among 52 patients who were followed up successfully, the influence of smoking as a single factor on survival time was compared. Thirty smoker patients and 22 never smoker patients were followed up for survival time in weeks. It was found that the survival time of patients with a smoking history was shorter than that of non-smokers (p=0.0318, Figure 1).

Figure 1

Kaplan-Meier survival curve for patients with IPF using the cut-off values for the smokers and never-smokers obtained by ROC analysis (log-rank test)

The COX multivariable survival analysis beyond smoking with more complete data is shown in Table 1).

The database includes 46 IPF and 26 normal lung tissues (Supplementary file Figure 1A). The Wilcoxon test was used for comparisons between the two groups, and Spearman’s test was used for correlations. The results of the data analysis showed that compared with that in normal lung tissue, the expression of LINC00665 in IPF lung tissues was obviously higher than that in normal healthy control lung tissues (p=0.000006, p<0.001) (Supplementary file Figure 1B), and the expression of XBP-1 in IPF lung tissue was significantly increased (Supplementary file Figure 1C, p=0.014). The expression levels of LINC00665 and XBP-1 in the former smoker lung tissues were significantly higher than those in the never smoker lung tissues (Supplementary file Figure 1D, p=0.018; and 1E, p=0.029). The expression level of LINC00665 in the former smokers’ IPF lung tissues was higher than that of the former smokers’ healthy controls (Supplementary file Figure 1F, p=0.037).

CSE regulates the LINC00665/XBP‑1 in the transformation of pulmonary fibroblasts

The fibroblasts were cultured in the presence or absence of an increasing concentration of CSE (0.1 M, 1.0 M) for 48 h; immunocytochemical analysis of α-SMA protein expression showed that the fibroblasts treated with 1.0 M CSE for 48 h resulted in more obvious expression of α-SMA than that in the 0.1 M group. (Supplementary file Figure 2A). Then, we determined LINC00665 expression in the control group, 0.1 M CSE and 1.0 M CSE group at 24 h and 48 h time points respectively, and the expression of LINC00665 increased significantly in the 1.0 M CSE group compared to the 0.1 M CSE group at different time points (Supplementary file Figure 2B, p<0.05). We used shRNA targeting LINC00665 to knockdown the expression of LINC00665, and the effect of shRNA1 was more obvious (Supplementary file Figure 2C, p<0.05) thus, this shRNA was selected for use in follow-up experiments. We used western blotting to monitor α-SMA and XBP-1 protein expression levels in different groups and examined the effect of different combinations of CSE, shLINC00665 and shNC. The results revealed that the expression of α-SMA and XBP-1 was obviously upregulated after treatment with CSE alone and CSE combined with shNC group, but was obviously downregulated after transfection with shLINC00665 after CSE compared to treatment with CSE alone. (Supplementary file Figure 2D, p<0.05). The expression level of XBP-1 in LFBs was significantly increased with TG treatment alone. However, the expression level of α-SMA in fibroblasts was substantially decreased after treatment with shLINC00665, but could be reverted upwards after restoring XBP-1 levels by TG with shLINC00665 in vitro (Supplementary file Figure 2E, p<0.01; and 2F, p<0.05).

LINC00665 regulated the expression of XBP‑1 by targeting miR‑214‑3p

We predicted that miR-214-3p was a potential biological target of LINC00665 by StarBase (http://starbase.sysu.edu.cn) and further confirmed its targeting relationship by a double luciferase reporter assay (Supplementary file Figure 3A). The 3’UTR loci of LINC00665 and miR-214-3p were identified, and then plasmids containing potential miR-214-3p binding sites and LINC00665 WT or MUT sequences were cotransfected with miR-NC or miR-214-3p mimic. The relative luciferase activity of the LINC0065 WT group and the LINC00665 MUT group was compared to verify whether miR-214-3p was the direct target of LINC00665 (Supplementary file Figures 3B and 3C). After TGF-β1 treatment of LFBs, the expression of miR-214-3p decreased compared with that in the control group. However, the expression of miR-214-3p in LFBs increased significantly after treatment with shLINC00665 (Supplementary file Figure 3D, p<0.01). Based on the binding sequence of the target gene 3’-UTR to miR-214-3p, the design was made using the miRNA Base database XBP-1 target sequence. The XBP-1 target gene fragment was synthesized by PCR. The results of double luciferase reporter gene analysis showed that miR-214-3p may bind to XBP-1 through the predicted binding site (Supplementary file Figures 3E and 3F). The expression of miR-214-3p in LFBs infected with miR-214-3p mimics was evidently higher than that in the control group (Supplementary file Figure 3G, p<0.01). These results indicated that miR-214-3p mimics can clearly improve the expression level of miR-214-3p, and the expression of miR-214-3p in LFBs transfected with miR-214-3p inhibitor was obviously lower than that in the inhibitor-transfected control group (Supplementary file Figure 3G, p<0.01). Moreover, the mRNA expression of XBP-1 in LFBs infected with miR-214-3p mimics was significantly lower than that in LFBs infected with mimics-NC, and the miR-214-3p inhibitor significantly increased the expression level of XBP-1 (Supplementary file Figure 3H, p<0.01). We used western blotting to monitor XBP-1 protein expression levels in LFBs with miR-214-3p mimics and inhibitors. The results revealed that miR-214-3p mimics obviously reduced the protein expression level of XBP-1 in cells compared with the control group, but miR-214-3p inhibitors obviously increased the protein expression level of XBP-1 (Supplementary file Figure 3I, p<0.01). We used western blotting to monitor XBP-1 protein expression levels in different groups and examined the effect of different combinations of TGF-β1, miR-214-3p inhibitor and shLINC00665. The results revealed that the expression of XBP-1 was obviously downregulated after transfection with shLINC00665 after TGF-β1 compared to treatment with TGF-β1 alone; however, the effect of shLINC00665 was reversed by inhibiting miR-214-3p (Supplementary file Figure 3J, p<0.01). To summarize, these results demonstrated that LINC00665 could regulate the expression of XBP-1 by targeting miR-214-3p.

Role of LINC00665 in BLM‑induced pulmonary fibrosis

HE staining was performed at 3 different times (7, 14, and 28 day) after BLM treatment to observe the pulmonary fibrosis in mouse lung tissue. Supplementary file Figure 4A provides the results of HE staining; the alveolar cavity structure of the control mice was intact, only a few inflammatory cells infiltrated in the interstitium, and the alveolar epithelium and the capillaries were normal. However, in the BLM-treated group, pulmonary fibrosis, structural disorder of bronchi and alveoli, chronic inflammatory interstitial infiltration, alveolar wall fracture and alveolar cavity fusion increased significantly. Moreover, we determined the LINC00665 mRNA expression in the NS or BLM group, and the expression of LINC00665 mRNA increased significantly in the BLM group compared to the control group at the same series of time points (Supplementary file Figure 4B, p<0.05 and p<0.01, respectively). The BLM-induced mouse pulmonary fibrosis was modelled by endotracheal administration of BLM under anesthesia, and normal saline solution was used as a control. According to the different treatment groups after modelling 72 h after injection, the transplantation group was intratracheally injected with shLINC00665, and the control group was injected with shNC. Next, we determined the expression of marker molecules (α-SMA, E-cadherin, collagen 1 and fibronectin) in each group. In the BLM-induced groups, the expression levels of α-SMA, collagen 1 and fibronectin were significantly increased in the merged processing groups with shNC compared with the groups combined with shLINC00665 (Supplementary file Figure 4C, p<0.05). The expression of E-cadherin was significantly decreased in the combined treatment of BLM with shNC group in contrast, its level was upregulated with shLINC00665 (Supplementary file Figure 4C, p<0.01). As shown in the HE and Masson staining images, treatment with shLINC00665 significantly reduced BLM-induced pulmonary fibrosis, and there was no difference in lung tissue between the saline-treated groups with or without shLINC00665 (Supplementary file Figures 4D and 4E). In summary, LINC00665 was highly expressed in BLM-induced lung tissues, and knockdown of LINC00665 alleviated BLM-induced pulmonary fibrosis.