Pleural inhibition of the caspase-1/IL-1β pathway diminishes profibrotic lung toxicity of bleomycin

Eight-week-old C57BL/6J mice (Charles River, Saint Germain-sur-l’Arbresle, France) were housed accordingly to the guidelines of the “Ministère de la Recherche et de la Technologie”. All experiments were approved by the “Comité d’Ethique de l’Expérimentation Animale (C2EA) du grand campus Dijon, n°105” (ref number : 4612). Mice were intravenously injected three times per week with bleomycin (BLM. Calbiochem) at a dose of 20 mg/kg for a total of 6 injections (Additional file 1: Figure S1A). Il-1β signaling was blocked with IL-1ra (anakinra) that was injected (0.1, 1 or 5 mg) intra-pleura every other day from day 0 to day 14 (Additional file 1: Figure S1B). Caspase-1 was activated by intrapleural injection of nigericin (Sigma Aldrich). Three, 14 or 21 days after the beginning of the injections and after a sample of blood was collected, mice were euthanized by abdominal aortic bleeding. Bronchoalveolar lavage fluid (BALF) and pleural lavage fluid (PLF) were then collected as previously described [5]. For histological analysis, lungs were harvested, inflated and fixed in formalin. Intrapleural injection of β-Galactosidase coding adenoviruses (AdLacZ) were administered as previously described [4, 5]. After being harvested, the lungs were placed in a solution containing β-Galactosidase substrate. Hydrolysis of this substrate released a blue-colored product highlighting pleural adenovirus-infected cells.

ibrosis in pleural and subpleural areas was assessed on lung sections after H&E staining using a scoring method based on the modified Ashcroft score previously described [20]. Lung sections were analyzed in a double-blinded test to assess the pleural area using ×100 magnification and were graded from 0 (normal lung) to 8 (completely fibrotic lung). When varying degrees of fibrosis were present, the highest score was retained. Collagen was quantified using a histomorphometric assay on picrosirius red stained sections [5]. Ten areas targeting the pleura were randomly acquired under polarized light (×100. Nikon Eclipse E600) with a high-resolution microscope camera (Nikon DS-Ri1). Signal intensity was quantified using a homemade ImageJ macro and plotted as signal intensity according to distance to the pleura (up to 500 μm into the lung parenchyma) (Additional file 2: Figure S2). Briefly, the user inputs a pleura limit, a parenchymal side (above, below, left, right), a background area and areas to exclude from the analysis (staining artifacts, vessels and large airways). The program adjusts the pleural limits, the beginning of the pleura being set as the first of two consecutive pixels (in an order from the background to the parenchyma) with an intensity higher than the mean background signal plus two standard deviations from the background. From there on, the program maps the distance from the parenchymal area to the pleura (Additional file 2: Figure S2A). It then measures the intensity of each parenchymal pixel in the picrosirius red channel. The program returns a text file with the intensity and distance to pleura of each measured pixel. The data are plotted to form a curve with the mean intensity of all pixels with a given distance to the pleura (Additional file 2: Figure S2B). Statistical analyses were performed by comparing areas under the curve over the whole range in the pleura (0–60 μm) or subpleural (60–500 μm) area. Picrosirius red was also monitored in the parenchymal area as previously described [21]. For collagen quantification, vessels and large airways were excluded. Total TGF-β1 levels in BALF or PLF were assessed with a specific ELISA for mouse TGF-β1 (Quantikine® ELISA. R&D Systems) according to the manufacturer’s guidelines.

Human pleural mesothelial cells Met5A were cultured in Medium-199 (Lonza) supplemented with 10% serum (Hyclone, Fisher Scientific). Proliferating cells were treated with BLM (Calbiochem). For Il-1β assessment in supernatant, cells were treated with BLM in serum-free medium. When indicated, caspase-1 was blocked using YVAD (Bachem AG) as previously described [19].

Formalin fixed, paraffin embedded lung sections were dewaxed with xylene and endogenous peroxidases were inhibited (H₂O₂ 3%). Frozen sections were fixed with ice-cold acetone and cells fixed with PFA and permeabilized with triton. Before saturation (BSA 5%), samples were incubated overnight at 4 °C with specific antibody. HSP47 (LF-PA41903) was from Ab Frontier, WT-1 (ab89901) from Abcam, Ki-67 (PA1-21520) from Thermo, HSP27 (SPA-803) and αB-crystallin (SPA-223) from Enzo Life Sciences, α-SMA (ab5694) from Abcam and NLRP3 (Cryo-2) from Adipogen. After washing, sections were incubated with HRP-coupled anti-rabbit antibody (Jackson Immunoresearch Laboratories) or Alexa568- or Alexa488-coupled antibody (Invitrogen). NovaRED (Vector Labs) was used to detect HRP followed by counterstaining with hematoxylin for immunohistochemistry. DAPI-containing ProLong® Gold (Invitrogen) was used for immunofluorescence experiments. For caspase-1 fluorescent staining, frozen lung sections were labeled according to the manufacturer’s instructions (FAM-FLICA YVAD. ImmunoChemistry Technologies). TGF-β1 expression was followed by a FISH analysis on dewaxed sections. Briefly, sections were incubated in hybridization solution (Formamide 20%, SSPE 2X) and then incubated with a mix of Quasar 570-coupled probes (“Stellaris”, Biosearch Technologies) specifically targeting murine TGF-β1 mRNA. After incubation, excess was eliminated from the probes and sections were mounted. Images were acquired with a Nikon Eclipse E600 with a high-resolution microscope camera (Nikon DS-Ri1). Fluorescent images were visualized using an Axio Imager.M2 microscope (Zeiss).

Proteins from cells or lung tissues were extracted using a Triton X-100 and anti-protease (Roche) containing buffer for 30 to 40 min on ice. Proteins were quantified using a colorimetric method (Bio-Rad Protein Assay. Bio-Rad) before thermic denaturation in loading buffer containing SDS and β-mercaptoethanol. For Il-1β assessment, supernatants were collected and centrifuged to pellet debris. Samples were concentrated using a methanol-chloroform method before denaturation. To evaluate active MMP expression, a non-denaturant loading buffer was used to deposit BALF or PLF for the SDS-PAGE. 30 μg of proteins for lysates or constant volume for supernatant were deposed for migration on 12–15% acrylamide gels according to the molecular weight of the targeted proteins. After electrophoresis, proteins were transferred on to PVDF membranes (GE Healthcare Europe GmbH) in a boric acid containing buffer. For IL-1β transfer, an ethanol-supplemented glycin buffer was used. Membranes were saturated with BSA before incubation with primary antibodies. MMP-9 (BML-SA620) from Enzolife Science, MMP-2 (sc-13594) and HSC70 (clone B-6) from Santa Cruz Biotechnology, HSP47 (LF-PA41903) from Abfrontier, Caspase-1 (AHZ0082) from Invitrogen, Il-1β (clone 166926) from R&D, α-SMA (ab5694) from Abcam, E-Cadherin (24E10) from Cell Signaling and β-actin (AC-74) from Sigma-Aldrich. After washing in TBS-Tween 0.1%, membranes were incubated with HRP-coupled secondary antibody (Jackson Immunoresearch Laboratories). The signal was revealed using a chemiluminescent reagent (Western Blotting Luminol Reagent. Santa Cruz Biotechnology) and was followed by a ChemiDoc XRS System (Bio-Rad).

Relative levels of several cytokines were assessed with the “Mouse Cytokine Array Panel A Array kit” (R&D Systems). Mouse plasma was tested as recommended by the manufacturer. Signal intensity was quantified by ImageJ software and data expressed as a percentage of intensity normalized to positive controls.

Histological analysis of lung sections, quantified by modified Ashcroft scoring and picorsirius red quantification, showed that repeated intravenous injections of bleomycin (BLM) induced progressive subpleural morphological changes in mouse lungs, compared with NaCl controls (Fig. 1a, Additional file 3: Figure S3A, B). By D14, morphological changes were mainly observed in the subpleural areas, and progressed towards the inner parenchyma by D21 (Fig. 1a, Additional file 3: Figure S3C–E). We set up a novel method to quantify picrosirius red staining by taking into consideration the distance to the pleura, and performed in-depth analyses (Additional file 2: Figure S2). We thus showed that, following intravenous administration of BLM, collagen started to accumulate in the pleura at D3 and progressed towards the lung parenchyma in a time-dependent manner (Fig. 1c, d). In line with this, in the lungs of BLM-treated mice but not in those of control mice, HSP47 was overexpressed in pleural cells as early as D3 and in fibrotic subpleural areas at D14 (Additional file 2: Figure S2F, G). Intravenous BLM promoted the upregulation of HSP27 and αB-crystallin in the fibrotic areas (Fig. 1e). These proteins have been reported to promote pleural mesothelial cell transformation [4, 9]. We also observed an increase in the expression of the marker of Ki-67 proliferation in pleural cells next to the fibrotic areas (Fig. 1e). Moreover, we detected an increase in proMMP-9 expression and MMP-2 activation in the pleural lavage fluid (PLF) from BLM-treated mouse lung, compared with the control group (Fig. 1f). We thus investigated the migration of pleural cells after intravenous BLM injection. Following intrapleural injection of LacZ-coding adenovirus (AdLacZ), which specifically labels pleural cells [5], we noticed a change in pleural cell morphology (arrows) at D5 in BLM-treated mice, compared with control mice (Fig. 1g). Furthermore, at D10 after the BLM injection, pleural cells were found within the subpleural lung parenchyma whereas in control animals the pleural cells remained in the pleural layer. These results demonstrate that systemic BLM injections trigger the transformation of pleural cells, which then migrate into the lung parenchyma.

Compared with controls, intravenous BLM induced an increase in the total number of cells in PLF starting at D14 (Fig. 2a). At D14 and D21 after the BLM systemic administration, we also found an increase in the percentage of neutrophils and leucocytes in the PLF (Fig. 2b). This pro-inflammatory profile at the pleural level was also found in the broncho-alveolar lavage fluid (BALF) following BLM (Additional file 4: Figure S4A, B). This suggests that the inflammatory response to the intravenous administration of BLM may involve the lung as well as the pleural area. Moreover, no differences were found in total cell count in the blood (monocytes, granulocytes, lymphocytes) or in serum levels of the major pro-inflammatory cytokines. (Additional file 4: Figure S4C, D). We then focused on the caspase-1/IL-1β pathway. Systemic administration of BLM in mice triggered the activation of caspase-1 in the lung in particular in the pleural and subpleural area (Fig. 2c, d). In the human pleural mesothelial Met5A cell line, BLM induced caspase-1 activation as well as active IL-1β secretion into the supernatant (Fig. 2e, f), and caspase-1/IL-1β activation correlated with the presence of NLPR3 aggregates, suggesting inflammasome formation (Additional file 5: Figure S5).

We observed at D14 that cells localizing in the pleura exhibited transformation features such as Ki-67 or HSP27/αB-crystallin overexpression. To investigate if pleural cells were also involved in the establishment of a profibrotic milieu, we investigated TGF-β1 production at this same time-point (e.g. D14). We observed an upregulation of TGF-β1 in the PLF from BLM-injected mice compared with controls (Fig. 3a). A Fluorescent In Situ Hybridization (FISH) analysis of lung sections showed that BLM treatment increased the expression of TGF-β1 mRNA in pleural cells whereas barely any expression was observed in control animals (Fig. 3b). A similar upregulation of TGF-β1 production was observed at D21 (Additional file 6: Figure S6). As inflammation is linked to TGF-β1 upregulation, we next investigated the relationship between BLM-induced IL-1β production by pleural cells and fibrosis markers such as TGF-β1 upregulation. Repeated intra-pleural administration of a specific inhibitor of IL-1β (IL-1ra) during the inflammation step (from day 0 to day 14) decreased TGF-β1 induction in the lungs (both mRNA and protein) induced by the intravenous administration of BLM at D21 (Fig. 3c, d). FISH analysis showed that the injection of IL-1ra was able to inhibit TGF-β1 expression in both pleural cells and cells localized in subpleural areas (Fig. 3e).

We studied whether the intra-pleural delivery of IL-1ra interfered with BLM pro-fibrotic toxicity in the lung. Pleural delivery of IL-1ra increased expression levels of E-Cadherin and decreased those of PAI-1 (Fig. 4a). In line with this finding, IL-1ra reduced collagen accumulation in mouse lung following BLM injection in a dose-dependent manner as observed by histomorphometric analysis (Fig. 4b). Further, as shown in Fig. 4c, in BLM-injected mice compared with the NaCl control, α-SMA was overexpressed in lung areas where caspase-1 was activated. Altogether, these data suggest that caspase-1/IL-1β signaling is involved in the transformation process occurring in the subpleural area, which represents a triggering event in fibrogenesis.

We then investigated the role of caspase-1 in pleural cell differentiation in vitro. In the presence of BLM, Met5A cells exhibited a mesenchymal phenotype with the overexpression of α-SMA (Fig. 5a). The specific caspase-1 inhibitor YVAD was able to counteract these changes. Furthermore, YVAD also abolished the E-Cadherin downregulation induced by TGF-β1 (Fig. 5b). To further confirm the involvement of caspase-1 in the process of pleural cell transformation, Met5A cells were cultured in the presence of the caspase-1 activator nigericin. As shown in Fig. 5c, E-cadherin expression gradually decreased as caspase-1 was activated. Compared with the control, nigericin induced other MMT-like changes such as α-SMA overexpression (Fig. 5d). In the same way, intra-pleural delivery of nigericin in mice induced a fibrotic response in the pleura (Fig. 5e). Moreover, when injected into the pleura, nigericin induced an increase in cell recruitment in the pleura of these animals (Fig. 5f). Collectively, these data indicate that caspase-1 is involved in the activation and subsequent transformation of pleural cells triggered by BLM or TGF-β1.