LTBP2 is secreted from lung myofibroblasts and is a potential biomarker for idiopathic pulmonary fibrosis

Idiopathic pulmonary fibrosis (IPF), pathologically known as usual interstitial pneumonia (UIP), is one of the most common interstitial lung diseases and has a poor prognosis [1]. Although the pathophysiology of lung fibrosis remains unclear, the currently accepted mechanism involves recurrent or repeated alveolar epithelial injuries; modulation by profibrotic cytokines, including transforming growth factor-β (TGF-β); activation and proliferation of fibroblasts; and the differentiation of fibroblasts into myofibroblasts, resulting in excessive production of extracellular matrix (ECM) proteins and aberrant fibrosis [2–4]. Amongst these, the differentiation of fibroblasts into myofibroblasts via the Smad and Akt signaling pathways is thought to be the main driving factor of lung fibrogenesis [5,6].

The clinical course of IPF is heterogeneous and often unpredictable [1]. Although pulmonary function parameters, including forced vital capacity (FVC), diffusion capacity for carbon monoxide (DLCO), and the scoring system based on these parameters are reliable biomarkers for IPF [1,7], measuring or re-evaluating these parameters is occasionally difficult in certain situations, such as for ventilated patients and patients with poor general health. Therefore, blood biomarkers that are easy to evaluate and are predictive of disease severity and/or prognosis are helpful in clinical settings.

To date, several blood biomarker candidates have been proposed for IPF, including Krebs von den Lungen-6 (KL-6), surfactant protein A and D, matrix metalloproteinase 1 and 7, CC chemokine ligand 18, C-X-C motif chemokine 13, and their combinations [8–11], but few have been sufficiently validated. Additionally, to our knowledge, no biomarker has been identified that reflects the essential fibrotic process ‘differentiation of lung fibroblasts into myofibroblasts’. Although myofibroblasts are generally defined as fibroblasts up-regulating α-smooth muscle actin (αSMA) expression, the αSMA, an intracellular protein, is not measurable as a blood biomarker. Therefore, there is a need to find another protein that is excessively secreted into blood during fibroblast differentiation. Based on this context and our preclinical findings, in the present study, we focussed on a secreted protein, latent TGF-β binding protein-2 (LTBP2), and aimed to evaluate the potential of LTBP2 as a novel blood biomarker for IPF.

C57BL/6 male mice, aged 11–14 weeks and weighing 25–28 g, were purchased from Japan SLC. The mice were maintained in a pathogen-free mouse facility at our institution. All in vivo experiments complied with the Animal Research: reporting in vivo Experiments guidelines and were performed in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. In addition, the protocol of these in vivo experiments was approved by the Animal Care and Use Committee of Hamamatsu University School of Medicine (approval number: 2015008).

Mice were anesthetized and then intratracheally administered with a single microspray (MicroSprayer, Penn-Century, U.S.A.) of 3 mg/kg bleomycin sulphate (Wako, Japan) in 50 μl saline to induce lung fibrosis. Mice were killed on day 10–14, an early fibrotic stage, and the harvested lungs were dissociated for FACS or used for immunostaining.

After perfusion of the lungs with 5 ml DPBS (Gibco, U.S.A.) via the right ventricle and removal of the bronchus and pulmonary vessels, lungs were harvested and incubated with 200 U/ml collagenase type 2 (Worthington, U.S.A.) and 100 U/ml DNase 1 (Worthington, U.S.A.) for 30 min at 37°C in DPBS. Digested lung tissues were then cut using a gentleMACS Dissociator (Miltenyi Biotechnology, Germany) according to the manufacturer’s instructions, and cells were filtered through a nylon screen (BD Biosciences, U.S.A.) to remove cell/tissue aggregates. The resulting single cell suspension was centrifuged and rinsed twice with FACS buffer (1% HEPES buffer, 2% heat-inactivated FBS, 120 μg/ml penicillin, 100 μg/ml streptomycin in HBSS).

Mouse lung cells were incubated with antibodies against the following lineage-specific cell surface markers: CD45, TER-119, CD324 (E-cadherin), CD31, LYVE-1, and CD146. When isolating steady-state fibroblasts from untreated mice, anti-PDGFRα (CD140a) antibody was added. In contrast, for isolation of activated fibroblasts and myofibroblasts from bleomycin-treated mice, anti-Sca-1 and biotin-conjugated anti-CD49e (integrin α5) antibodies were added. Incubation with these antibodies was performed on ice for 60 min. After centrifugation and rinsing twice in FACS buffer, only for the bleomycin-treated mouse lung cells, incubation with streptavidin for 60 min on ice was additionally done. Based on the results of our previous study but with a few modifications [12], we defined steady-state fibroblasts as lineage-negative and PDGFRα-positive cells from untreated mouse lungs; activated fibroblasts as lineage-negative and Sca-1-positive cells from bleomycin-treated mouse lungs; and myofibroblasts as lineage-negative, Sca-1-negative, and CD49e-positive cells from bleomycin-treated mouse lungs (Figure 1A). On the contrary, each lineage-specific cell surface marker (CD45, CD324, CD31, LYVE-1, and CD146) was also used for the isolation of blood cells, epithelial cells, vascular endothelial cells, lymphatic endothelial cells, and smooth muscle cells, respectively. Before sorting, Sytox-Red or -Green (1:1000; Molecular Probes, U.S.A.) was used to remove the dead cells.

The protein level of intracellular αSMA, a myofibroblast marker, was quantitated by permeabilizing 3000 isolated fibroblasts/myofibroblasts using IntraPrep (Beckman Coulter, U.S.A.), according to the manufacturer’s instructions. After centrifugation and rinsing in DPBS, the cells were incubated with anti-αSMA antibody for 60 min at room temperature. After additional centrifugation and rinsing in DPBS, fluorescence intensities of FITC-conjugated anti-αSMA antibody in steady-state fibroblasts, activated fibroblasts, and myofibroblasts were measured for confirmation (Figure 1B).

The antibodies and their concentrations used in FACS are listed in Supplementary Table S1. Isotype antibodies were used for setting cut-off levels. To obtain a high purity, the samples were subjected to sorting twice. All sorting and analyses were performed using FACSAria (BD Biosciences, U.S.A.). FACS data were analyzed using FlowJo software version 7.6.5 (Tree Star, U.S.A.).

Total RNA was extracted from two independently prepared batches of 50000 cells of steady-state fibroblasts, myofibroblasts, or lung homogenates (as a control). Additionally, RNA was amplified using the Ovation Pico WTA System V2 (NuGEN Technologies, U.S.A.). Gene expression profiling was commissioned to an outside agency (SurePrint G3 Mouse Gene Expression 8x60K v2 Microarray, Takara Bio, Japan). The experiments and description followed the MIAME guidelines; the raw data are available at the GEO database (accession number: GSE111043). The heatmaps of gene expression profiles are shown in Figure 1C and in Supplementary Figure S1; genes that were up-regulated or down-regulated in myofibroblasts and the ontology profiles are summarized in Supplementary Tables S2 and S3, respectively. RNA levels of several genes were confirmed using reverse transcription-PCR (RT-PCR).

Total RNA was extracted from freshly isolated mouse cells or cultured human cells using TRIzol (Invitrogen, U.S.A.) with glycogen as a carrier. RNA was extracted as per the manufacturer’s instructions. Extracted RNA was treated with RNase free DNase 1 (Ambion, U.S.A.) in the presence of RNase Inhibitor (Invitrogen, U.S.A.) for 10 min at 37°C. RNA was purified using the RNeasy Mini Kit (Qiagen, Germany) according to the manufacturer’s instructions. After first-strand cDNA was synthesized using SuperScript Reverse Transcriptase (Invitrogen, U.S.A.) with random primers (Invitrogen, U.S.A.), cDNA equivalent to 100 cells was used for each PCR. Gene-specific primers were designed using Primer 3 software (v.0.4.0) to generate short amplicons (100–150 bp). PCR was performed using the SYBR Green quantitative RT-PCR (qRT-PCR) Kit (Applied Biosystems, U.S.A.). The PCR cycling program comprised 1 cycle at 95°C for 10 min, 40 cycles at 95°C for 15 s and 60°C for 1 min, and 1 cycle at 95°C for 15 s and 60°C for 1 min (ABI StepOnePlus, Applied Biosystems, U.S.A.). The cycle threshold (C_T) for each gene was measured, and the values were expressed as 2^−ΔCT (ΔC_T: C_T of target gene minus C_T of glyceraldehyde-3-phosphate dehydrogenase). Primer sequences for RT-PCR are listed in Supplementary Table S4.

Primary human adult lung fibroblasts (HPF-a #3310, passage number: less than three) were purchased from the ScienCell Research Laboratories (U.S.A.). In addition, primary human fetal lung fibroblasts (TIG-1-20, passage number: unknown) were also purchased from the Japanese Collection of Research Bioresources Cell Bank (Japan). These fibroblasts were grown on a 24-well plate to confluence in growth medium containing DMEM (Gibco, U.S.A.), 10% FBS, 2 mM glutamine, and a 1:1000 dilution of antibiotics solution. Cells were maintained in an incubator at 37°C in a 5% CO₂ atmosphere. Confluent monolayers were washed twice with DPBS. The medium was then replaced with chemically defined serum-free medium (Lonza, U.S.A.) containing recombinant human TGF-β1 (#0117209-1, PeproTech, U.S.A.) at concentrations of 0, 2.5, or 5 ng/ml. After incubation for 48 h, the culture supernatant was collected and used for ELISA according to the manufacturer’s protocol (SEB630Hu, Cloud-Clone, U.S.A.). Cells were collected from the plates for RT-PCR.

A retrospective database review at Hamamatsu University Hospital between 2000 and 2014 identified 152 IPF patients based on the criteria used in a recent clinical trial [13]. Amongst these patients, 36 patients who had or developed malignant diseases such as lung cancer and prostate cancer within 6 months from IPF diagnosis were excluded because their clinical course and serum concentrations of LTBP2 would likely be affected by the malignancies [14–22]. Subsequently, 116 IPF patients were included in the present study. The patients’ medical records were reviewed to obtain the clinical data and the clinical course, including prognosis (deaths by all causes or respiratory events) and complications of acute exacerbation of IPF [23]. The present study was approved by the Institutional Review Board of Hamamatsu University School of Medicine (approval number: 15-197).

Serum samples were obtained at the time of IPF diagnosis and stored at −80°C for future analysis. Written informed consent regarding serum conservation was obtained from all patients (approval number: 15-165). As a control, serum samples were also collected from 31 healthy volunteers who worked or had worked at Hamamatsu University School of Medicine. Serum LTBP2 concentrations were quantitated using ELISA (SEB630Hu, Cloud-Clone, U.S.A.).

Lung specimens were obtained from three untreated mice, three bleomycin-treated mice (killed on day 10–14), ten patients with IPF (extracted by surgical lung biopsy), a patient who died of acute exacerbation of IPF (extracted by autopsy), and two patients with spontaneous pneumothorax and early-stage lung cancer, respectively, as normal controls (extracted by lung resection).

These lung specimens had been fixed in 10% formalin and embedded in paraffin. Deparaffinized sections (4-μm-thick) were preheated to 120°C for 15 min (pH = 6). After inactivating endogenous peroxidase with 0.3% H₂O₂ for 10 min, the sections were incubated with blocking solution (10% goat serum in PBS and 0.1% Triton X-100 in PBS) for 10 min and then incubated with anti-LTBP2 antibody (1:1000; Proteintech, U.S.A.) or anti-αSMA antibody (1:300; Dako, U.S.A.) as a primary antibody for 1 h.

Lung sections were incubated with peroxidase-labeling goat anti-rabbit IgG antibody (Nichirei Biosciences, Japan) for 15 min. The immunoreaction was visualized using 3,3′-diaminobenzidine (Dako, U.S.A.) and then counterstained with Hematoxylin. Hematoxylin–Eosin staining and Elastica van Gieson staining were also performed on other serial sections.

In addition to the anti-LTBP2 antibody and anti-αSMA antibody, for human lung sections, anti-CD45 antibody (1:100; Dako, U.S.A.), anti-CD324 antibody (1:20; Dako, U.S.A.), anti-CD31 antibody (1:100; Dako, U.S.A.), anti-podoplanin antibody (1:200; Dako, U.S.A.), and anti-h-caldesmon antibody (1:5; MyBioSource, Canada), anti-elastin antibody (1:500; Abcam, U.S.A.), and anti-fibrillin-1 antibody (1:200; EMD Millipore, U.S.A.) were also used as primary antibodies. The sections were incubated with Hoechst 33342 (1:1000; Sigma, U.S.A.) with Alexa fluor conjugated secondary antibodies for 30 min. The sections were visualized using a confocal microscope (Leica, U.S.A.). In the present study, human lung myofibroblasts in immunofluorescence images were defined as single cells with positivity for αSMA and negativity for lineage markers: CD45 (blood cells), CD324 (epithelial cells), CD31 (vascular endothelial cells), podoplanin (lymphatic endothelial cells), and h-caldesmon (smooth muscle cells).

For in vitro experiments, data were expressed as mean ± S.E.M.; group comparisons with each control were performed using the Student’s t test. For analyses using human materials, data were described as number (%) or median (interquartile range) unless otherwise noted; group comparisons were performed using Fisher’s exact test or the Mann–Whitney U test. Using the curve of receiver operating characteristic (ROC) and the Youden index method, a cut-off value with favorable sensitivity and specificity was calculated to discriminate patients with IPF from healthy controls. Spearman’s rank correlation coefficients were evaluated between serum LTBP2 concentrations and the other clinical factors. Overall survival was defined as the time from the date of IPF diagnosis to the date of censoring or death (by all causes or respiratory events). Patients were censored if they remained alive until 31 December 2016 or if they dropped out of the follow-ups. The Cox proportional hazard model was used to evaluate the prognostic impact of serum LTBP2 concentrations. Additionally, the bootstrap method with 1000-fold replication was utilized for estimating variability of hazard ratios of serum LTBP2 concentrations [24]. To estimate the ideal cut-off value of serum LTBP2 concentrations in patients with IPF, a time-dependent ROC analysis was performed [25]. The survival curves were generated using the Kaplan–Meier method and compared using log-rank test. Statistical analyses were performed using R software version 2.15.1 (the R Foundation for Statistical Computing, Austria) and SPSS software version 13.0 (SPSS, U.S.A.). P<0.05 was considered significant.

Microarray comparisons between FACS-sorted steady-state fibroblasts and myofibroblasts from mouse lungs revealed 44 genes that were highly up-regulated in myofibroblasts at the transcriptional level based on our criteria (Supplementary Table S2). These genes were commonly annotated with ‘ECM’ or ‘proteinaceous ECM’ (Supplementary Table S3). Amongst them, we selected one gene, Ltbp2, because it encodes an extracellular protein and its signal was most similar to Acta2, a gene encoding αSMA, in the clustering analysis (Figure 1C). Consistent with the microarray results, qRT-PCR analyses confirmed that the Ltbp2 mRNA levels were significantly higher in myofibroblasts than in steady-state fibroblasts (Figure 2A). Intermediate Ltbp2 mRNA levels were observed in bleomycin-induced activated fibroblasts. These differences in Ltbp2 mRNA levels seemed parallel to Acta2 mRNA levels (Figure 2B). On the other hand, the Ltbp2 mRNA levels in CD45-positive blood cells, CD324-positive epithelial cells, CD31-positive vascular endothelial cells, LYVE-1-positive lymphatic endothelial cells, and CD146-positive smooth muscle cells (pericytes and endothelial cells) were commonly low and not significantly different between untreated and bleomycin-treated mouse lungs (Figure 2C), suggesting that these lineage cells were not the main source of LTBP2 in mouse lungs.

To determine whether the results in mouse cells were reproducible in human cells, we analyzed the LTBP2 mRNA levels in human lung fibroblasts with or without the stimulation by TGF-β1. As shown in Figure 3A,B, the TGF-β1 stimulation significantly up-regulated LTBP2 mRNA levels, which almost corresponded to the change in Acta2 mRNA levels. In addition, protein concentrations of LTBP2 in culture supernatants, evaluated by ELISA, were significantly higher in TGF-β1-induced myofibroblasts than in controls, suggesting that the extracellular secretion of LTBP2 proteins was higher from TGF-β1-induced myofibroblasts than from untreated fibroblasts (Figure 3C).

Figure 4 shows the results of immunohistochemistry. In normal lung sections, scarce positivity was observed except at the peribronchovascular lesions (Figure 4A,B). In contrast, LTBP2 staining was broadly positive in the fibrotic interstitium in biopsied IPF lungs. Unexpectedly, the bodies of fibroblastic foci, which are thought to be aggregated myofibroblasts, were not clearly positive for LTBP2 staining (Figure 4C–F). Elastica van Gieson staining showed that the distribution of LTBP2 partially overlapped with the elastic fibers (Figure 4G,H).

Consistent with the immunohistochemistry results, immunofluorescence staining for LTBP2 was extensively positive in the fibrotic interstitium but negative in the bodies of fibroblastic foci in biopsied IPF lungs (Figure 5A and Supplementary Figure S2A). However, some αSMA-positive myofibroblasts, which did not configure fibroblastic foci, stained positive for LTBP2 in the cytoplasm, suggesting that LTBP2 was secreted from the myofibroblasts (Figure 5B). These findings are consistent with the results seen in bleomycin-treated mouse lungs (Supplementary Figure S3). On the contrary, LTBP2 staining was not observed in blood cells (CD45), epithelial cells (CD324), vascular endothelial cells (CD31), lymphatic endothelial cells (podoplanin), and vascular smooth muscle cells (h-caldesmon) (Figure 5C–F). In contrast, as an ECM protein, LTBP2 partially colocalized with elastin and fibrillin-1, which are major elastinogenesis-related proteins (Supplementary Figure S2B,C).

To evaluate the clinical significance of LTBP2, serum concentrations of LTBP2 in patients with IPF and healthy controls were analyzed using ELISA. Baseline characteristics are summarized in Table 1. Serum LTBP2 concentrations in patients with IPF were significantly higher than those in healthy controls (median: 18.6 ng/ml (mean: 21.4) compared with 12.2 ng/ml (mean: 12.4), P<0.001; Figure 6A). Figure 6B shows the ROC curve with an area under curve (AUC) of 0.862. The ideal cut-off value was calculated as 15.1 ng/ml; the sensitivity and specificity for discriminating patients with IPF from healthy controls were 78.4 and 87.1%, respectively. Additionally, as shown in Table 2, serum LTBP2 concentrations in patients with IPF showed a significant negative correlation with % predicted FVC (r = −0.369).

We next sought to study the association between LTBP2 and acute exacerbation of IPF. Interestingly, patients who had acute exacerbation of IPF at the time of IPF diagnosis or who developed the event within 6 months of their diagnosis, had significantly higher serum LTBP2 levels than the other patients (median: 21.4 ng/ml (mean: 29.0) compared with 17.9 ng/ml (mean: 19.4), P<0.01; Figure 7A). Immunostaining of autopsied lung sections from a patient who died of acute exacerbation of IPF revealed diffuse alveolar damage with a lot of αSMA-positive and non-fibroblastic foci myofibroblasts within the alveolar walls. Most of these myofibroblasts were stained or surrounded by LTBP2. In addition, interestingly, hyaline membranes in the alveolar spaces seemed to contain LTBP2 proteins (Figure 7B−F). In contrast with the biopsied lungs without acute exacerbation, LTBP2 positivity was also detected in or around proliferated smooth muscle cells at the fibrotic alveolar walls (Figure 7G). However, even with acute exacerbation, myofibroblasts configuring fibroblastic foci were not clearly stained by LTBP2.

The median follow-up period for patients with IPF was 36 months (interquartile range: 16–67 months). During the follow-up period, 60 of the 116 patients died (of which 55 died of respiratory events); no patients received lung transplantation; 16 patients became lost during follow up at least 6 months after IPF diagnoses. The Cox proportional hazard model analyses identified baseline PaO₂, % predicted FVC, % predicted DLCO, and serum LTBP2 concentrations as potential prognostic factors in patients with IPF (Table 3). The hazard ratios for an increase in serum LTBP2 concentration of 1 ng/ml were 1.038 for all-cause death (95% confidence interval: 1.024–1.053) and 1.040 for death by respiratory events (95% confidence interval: 1.026–1.054). The bootstrap method with 1000-fold replication estimated the mean hazard ratio commonly as 1.046 (S.D.: 0.014, minimum: 1.024, maximum: 1.138 for all-cause death; S.D.: 0.015, minimum: 1.024, maximum: 1.138 for death by respiratory events), which was comparable with the original value. Using time-dependent ROC curves, the ideal cut-off level for predicting a better/worse prognosis was calculated as 18 ng/ml (Figure 8). The ROC curves at the 60-month follow-up are shown in Supplementary Figure S4. Comparisons of clinical factors and survival curves based on this cut-off value are shown in Supplementary Table S5 and Figure 8, respectively.

We identified Ltbp2 as one of the several highly up-regulated genes in myofibroblasts by analyzing transcriptome data from freshly isolated steady-state fibroblasts and myofibroblasts from mouse lungs. The up-regulation of Ltbp2 was reproduced in human lung myofibroblasts induced by TGF-β1. Importantly, the LTBP2 protein was excessively secreted from these differentiated cells in vitro. Serum concentrations of LTBP2 were associated with baseline % predicted FVC values, as well as the prognosis of patients with IPF, suggesting that serum LTBP2 is potentially a novel prognostic biomarker reflecting the level of differentiation of lung fibroblasts into myofibroblasts in patients with IPF.

LTBP2, a fibrillin/LTBP superfamily protein, is an ECM protein with a molecular weight of 195–240 kDa. It is commonly expressed in the lungs, skin, and large vessels [26–28]. Although LTBP1, 3, and 4 can combine with latent TGF-β, only LTBP2 is present without combining; thus, the function of LTBP2 has not been fully clarified [28]. Although Ltbp2-knockout mice are reported to exhibit normal-structured lungs [29], a recent study demonstrated that a double knockout mouse lacking the Ltbp2 and Ltbp4S genes exhibited a worsened emphysema compared with the single knockout mice lacking only Ltbp4S [30]. These data suggest a positive role of LTBP2 in lung elastinogenesis.

In addition to elastinogenesis, LTBP2 may be associated with fibrosis or tissue remodeling in several organs, including the skin, heart, and kidneys [31–33]; however, its biological role remains unknown. An in vitro study has revealed that LTBP2 competes with LTBP1 by binding to fibrillin-1, suggesting an indirect role of LTBP2 in the storage and activation of TGF-β, a major profibrotic growth factor [34]. In addition, Sideek et al. [35] recently reported that a bioactive region of LTBP2 can directly up-regulate TGF-β1 expression. On the other hand, previous studies suggested the antifibrotic role of LTBP2 via inhibition of fibroblast growth factor-2-induced fibroblast proliferation and fibroblast adhesion to fibronectin [36,37]. Therefore, future experiments using Ltbp2-knockout mice to assess lung fibrosis, as well as gene and/or protein analyses of isolated lung (myo)fibroblasts, may be necessary to determine the exact role of LTBP2 in fibrosis.

In the present study, up-regulation of LTBP2 at the mRNA and protein levels in human lung myofibroblasts induced by TGF-β1 was found. This is consistent with the results from a previous study that speculated the involvement of Ras or another related small GTP-binding protein in the signaling pathway [38]. In biopsied IPF lung sections, LTBP2 staining was almost negative in blood cells, epithelial cells, vascular endothelial cells, lymphatic endothelial cells, and (peribronchovascular and interstitial) smooth muscle cells, whereas the fibrotic interstitial space was broadly stained. Based on our immunofluorescence results, the majority of LTBP2 presented as an ECM protein; however, several non-aggregated (non-fibroblastic foci) myofibroblasts also stained positive for LTBP2, suggesting an origin of LTBP2 in fibrotic lungs. In contrast, as shown in Figure 7G, when a patient develops acute exacerbation of IPF, proliferated interstitial smooth muscle cells as well as non-fibroblastic foci myofibroblasts seem to be one of the sources of LTBP2. Collectively, during lung fibrosis, LTBP2 is produced by these abnormal mesenchymal cells and is secreted into the ECM where it likely colocalizes with several extracellular proteins, including elastin, fibrillin-1, and possibly other collagenous proteins, such as collagen І and fibronectin [39,40]. In addition, some LTBP2 enters into blood vessels and can be detected as a serum protein.

To the best of our knowledge, no studies evaluating LTBP2 as a blood biomarker for IPF have been reported. Based on a previous proteome analysis using human plasma, higher LTBP2 concentrations were significantly associated with a worse prognosis, commonly due to pulmonary death, in patients who developed acute dyspnea [41]. The differentiation of lung fibroblasts into myofibroblasts is frequently observed, regardless of the underlying disease, when a patient develops acute respiratory distress syndrome or diffuse alveolar damage [42,43]. These findings may support our data showing that patients who developed acute exacerbation within 6 months of their IPF diagnosis, often with diffuse alveolar damage superimposed on the underlying UIP fibrosis [23], had significantly higher serum LTBP2 concentrations than those who did not. Moreover, although the mechanism is unclear, the LTBP2 positivity for hyaline membranes, in which various kinds of serum proteins are included [44], might also be a supportive evidence for the serum data of acute exacerbation. Taken together, LTBP2 as a biomarker can be used to evaluate not obsolete fibrosis but acute-phase and ongoing fibrogenesis. This leads to speculation that LTBP2-positive or -producing myofibroblasts are differentiating or newly differentiated cells compared with LTBP2-negative myofibroblasts configuring fibroblastic foci.

As previous researchers have indicated, serum LTBP2 concentrations can increase in several non-pulmonary diseases, such as cancer, heart failure, and renal dysfunction [14–22,32,45]. Additionally, in the present study, there was a considerable overlap of data in the serum LTBP2 concentrations between IPF patients and healthy subjects. These results imply that serum LTBP2 is insufficient as a ‘diagnostic’ biomarker for IPF. In order to improve the diagnostic accuracy, it may be worth considering combining LTBP2 with other existing markers like KL-6.

Our study has several limitations, particularly in the clinical aspect. First of all, this is a small single-cohort retrospective study. To validate our preliminary results, a future large-scale, multi-cohort prospective study is warranted. In addition, we could not evaluate the chronological changes of serum LTBP2 concentrations in the same IPF patients. Perhaps, such changes might be more important than the one-point value itself, which may explain why several patients showed relatively low serum LTBP2 concentrations even during the acute exacerbation of IPF. Moreover, to clarify the utility of LTBP2, it is necessary to study the serum concentrations in patients with other pulmonary diseases including infectious pneumonias, bronchial asthma, and particularly interstitial lung diseases except IPF. These can be future research topics.

Within these limitations, in conclusion, we demonstrated that serum LTBP2, which may correspond to TGF-β1-induced differentiation of lung fibroblasts into myofibroblasts, is a potential prognostic biomarker for IPF.

• IPF is a chronic, progressive, fibrotic lung disease with high mortality. Although differentiation of lung fibroblasts into myofibroblasts is known to be important in disease progression, few biomarkers reflecting the fibrotic process have been discovered.

• Our transcriptome analysis identified Ltbp2, the gene encoding LTBP2, as a significantly up-regulated gene in myofibroblasts isolated from bleomycin-treated mouse lungs using FACS. This up-regulation was reproduced at the mRNA and protein levels in human lung myofibroblasts induced by TGF-β1. Intriguingly, serum LTBP2 concentrations in IPF patients were higher than those in healthy volunteers and the increase was significantly associated with a worse prognosis.

• Serum LTBP2, which may correspond to TGF-β1-induced differentiation of lung fibroblasts into myofibroblasts, is a potential prognostic biomarker in patients with IPF.

We thank all staff of Department of Regenerative and Infectious Pathology, Hamamatsu University School of Medicine (Mr Kaneta, Ms Suzuki, Ms Kawashima, and Ms Muranaka) for their technical assistance. We also thank all staff of Second Division, Department of Internal Medicine, Hamamatsu University School of Medicine, particularly Dr Furuhashi, Dr Suzuki, Dr Karayama, Dr Hozumi, Dr Kono, Dr Yasui, Dr Oyama, Dr Yoshimura, Dr Nishimoto, Dr Akiyama, Ms Oishi, Ms Takeuchi, and Ms Onaka for their data acquisition. We appreciate the statistical advice from Dr Yamakage (SATISTA Inc., Japan).

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by grant-in-aid for Scientific Research (C) from the Japan Society for the Promotion of Science [grant number 16K08736].

Y.E. and T.I. designed the study. Y.E., S.Matsushima, K.S., Y.A., H.Y., S.Meguro, H.K., I.K., T.F., N.E., N.I., Y.N., and T.S. collected data. All authors contributed to data interpretation. Y.E. and T.I. wrote the original manuscript. All authors contributed to the revisions of the manuscript, and agreed to be accountable for all aspects of the work.

αSMA

α-smooth muscle actin

AUC

area under curve

C_T

cycle threshold

DLCO

diffusion capacity for carbon monoxide

ECM

extracellular matrix

FVC

forced vital capacity

IPF

idiopathic pulmonary fibrosis

KL-6

Krebs von den Lungen-6

LTBP2

latent transforming growth factor-β binding protein-2

qRT-PCR

quantitative reverse transcription-PCR

ROC

receiver operating characteristic

RT-PCR

reverse transcription-PCR

TGF-β

transforming growth factor-β

UIP

usual interstitial pneumonia