Background
Idiopathic pulmonary fibrosis (IPF) is a debilitating interstitial lung disease (ILD) characterised by the excessive deposition of extracellular matrix by fibroblasts and an irreversible loss of lung function [
1]. With a median survival of approximately 3 years from diagnosis, the prognosis for patients with IPF is worse than many types of cancer [
2]. Despite its incidence continuing to increase, the aetiology and underlying pathophysiology of IPF remains unclear [
1]. Although two drugs are currently licensed to treat IPF, their failure to halt disease progression and use-limiting side effects [
3] demonstrate an urgent need to better understand the pathogenesis of IPF and identify new therapeutic targets.
The IL-1 family cytokine interleukin-33 (IL-33) is stored in the nucleus of multiple cell types and is released following cell damage or death [
4‐
7]. Upon its release, IL-33 can interact with its transmembrane receptor ST2L (also known as serum stimulated-2 (ST2) or Interleukin-1 Receptor Like 1 (
IL1RL1)) or be neutralised by its decoy receptor soluble ST2 (sST2) [
8]. ST2-dependent signalling can initiate inflammation and promote wound healing following tissue damage [
9]. Despite playing an important homeostatic role, dysregulation of the IL-33:ST2 axis has been implicated in the pathogenesis of several diseases including IPF [
9,
10]. Indeed, high levels of IL-33 have been observed in the bronchoalveolar lavage fluid (BALF) [
11], exhaled breath condensate [
12] and lung tissue [
13,
14] of IPF patients, with elevated concentrations of sST2 measured in IPF serum during exacerbations [
15]. Additionally, IL-33 has been implicated in the bleomycin (BLM) mouse model of pulmonary fibrosis with germline ST2 deletion, IL-33 neutralising antibodies and sST2 all leading to reduced lung fibrosis in vivo [
16‐
19]. Since IL-33 overexpression and treatment potentiates BLM-induced fibrosis [
14,
16], IL-33 has been proposed to act as a key pro-fibrotic mediator during the development of IPF. However, since IL-33 signalling has only ever been modulated during the inflammatory phase of the BLM mouse model [
14,
16‐
18], it is possible that these results reflect the importance of IL-33 during inflammation and repair rather than in fibrosis. Consequently, the therapeutic benefit of targeting the IL-33:ST2 axis in IPF is unclear.
Transforming growth factor-β (TGFβ) is a key pro-fibrotic cytokine that induces the production of extracellular matrix by primary human lung fibroblasts (HLFs) in vitro [
20,
21]. In vivo, TGFβ plays an important role in pulmonary fibrosis with therapies targeting TGFβ activation and signalling shown to reduce fibrosis in multiple model systems [
22‐
24]. Moreover, TGFβ can induce fibrotic changes in human precision-cut lung slices (PCLS) [
25] and explanted lung parenchymal tissue samples [
26].
To understand the role of IL-33 in established pulmonary fibrosis, we assessed the localisation of IL-33 in fibrotic lung tissue and determined whether IL-33 expression by HLFs could be regulated by TGFβ. Furthermore, the effects of blocking IL-33 signalling during the fibrotic phase of the BLM mouse model were established and the ability of IL-33 to induce fibrotic changes in human PCLS investigated.
Methods
IL-33 Immunohistochemistry (IHC)
Formalin fixed, paraffin embedded tissue samples from 43 ILD patients (patient characteristics previously reported in Saini et al. [
27]) and 4 non-ILD controls (normal adjacent tissue from patients undergoing lung cancer resections) were obtained from the University of Edinburgh and the Nottingham Respiratory Research Unit following informed consent and local ethics approval by South East Scotland SAHSC Bioresource and the MRC Nottingham Molecular Pathology Node respectively. IHC was performed as previously described [
27] using a mouse anti-IL-33 monoclonal antibody (Nessy-1, Abcam, Cambridge, UK).
Cell culture and treatment
Non-IPF and IPF patient-derived primary human lung fibroblasts (HLFs) were obtained from the Nottingham Respiratory Research Unit and cultured as previously described [
28]. Following 24 h serum-starvation, passage 6 HLFs were stimulated with 2 ng/ml TGFβ or 10 ng/ml IL-33 (R&D Systems, Abingdon, UK) with supernatants, RNA and protein collected as previously described [
29].
Bleomycin (BLM) mouse model
Animal studies using male C57BL/6 mice were approved by the University of Nottingham Animal Welfare and Ethical Review Board, carried out in accordance with Animals (Scientific Procedures) Act 1986 and planned and reported in compliance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines [
30]. Pulmonary fibrosis was induced with 60 IU BLM sulphate (Kyowa Kirrin, Slough, UK) as previously described [
24] and mice dosed every 72 h from day 14–28 with 10 mg/kg of ST2-Fc fusion protein (AstraZeneca, Cambridge, UK) via intraperitoneal (i.p.) injection. Upon completion of the study, mice were randomly allocated into groups for either biochemical or histological analysis.
For mice in the biochemical analysis group, bronchoalveolar lavage (BAL) was collected and lung tissue snap frozen as previously described [
24]. Total BAL cell counts were performed using a hemacytometer. For differential cell counts, 200 μl of BAL was cytospun onto glass slides and stained using a Diff-Quick staining kit (Siemens Healthineers, Camberley, UK) according to the manufacturer’s instructions. Differential cell counts were performed as previously described [
24] with the remaining BAL fluid (BALF) collected for analysis by ELISA. Lung tissue was ground into a powder using liquid nitrogen. Lung hydroxyproline content per lung set was measured as described previously [
29]. RNA and protein were extracted as previously described [
24,
29].
For mice assigned to the histological analysis group, the tracheas of these animals were cannulated and the pulmonary vasculature perfused with 40 U/ml of heparinised PBS (Sigma-Aldrich, St. Louis, MO, USA). Lungs were inflated with 10% formalin (VWR, Lutterworth, UK) under constant gravitational pressure (20 cm H
2O) prior to removal and fixation in 10% formalin. All tissue was wax embedded and 5 μm thick sections cut for histological analysis. Tissue was dewaxed in xylene and rehydrated in graded ethanol. Sections were subsequently stained with Mayer’s haematoxylin and eosin or Weigert’s haematoxylin and Sirius red. Masson’s trichome staining was performed as previously described [
29]. Stained sections were visualised using CaseViewer software (3DHISTECH, Budapest, Hungary) and the severity of lung fibrosis quantified by Ashcroft Scoring [
31].
ELISA analysis
Human (DY3625B) and mouse (DY3626) IL-33 and mouse ST2 (DY1004) ELISAs (R&D systems) were performed according to the manufacturer’s instructions.
Precision-cut lung slices (PCLS)
Normal adjacent lung tissue (NAT) and IPF lung tissue was obtained from the Royal Papworth Hospital Research Tissue Bank with written consent and study approval from the NRES Committee East of England. Patient characteristics for lung samples used in PCLS experiments are summarised in Table
1.
Table 1
Patient characteristics for PCLS lung samples
Age, years | 73 (67–76) (n = 3) | 61 (56–66) |
Gender, male/female | 1/2 (n = 3) | 2/0 |
Smoking status, ever/never-smokers | 3/0 (n = 3) | 1/1 |
FVC % predicted, % | 117.6 (109.8–121.0) (n = 3) | 46.5 (37.0–56.0) |
PCLS were generated as previously described [
32]. Once generated, PCLS were rested for 4 days in Small Airway Epithelial Cell Media supplemented with 2.5 mg/ml Bovine Serum Albumin, 0.004 ml/ml Bovine Pituitary Extract, 10 ng/ml EGF, 5 μg/ml Insulin, 0.5 μg/ml Hydrocortisone, 0.5 μg/ml Epinephrine, 6.7 ng/ml Triiodo-L-thyronine, 10 μg/ml Transferrin and 0.1 ng/ml Retinoic Acid (PromoCell, Heidelberg, Germany). Daily media changes were performed prior to stimulation every 24 h with 2 ng/ml TGFβ (R&D Systems) or 30 ng/ml IL-33 (Viva Biotech, Shanghai, China). PCLS (3–4/treatment) were pooled and processed for RNA or protein on day 7.
For RNA samples, PCLS were submerged in RNAlater (Sigma-Aldrich) for 24 h at 4 °C then stored at -80 °C prior to RNA isolation. To extract RNA from PCLS, 600 μl of TRI Reagent (Zymo Research, Irvine, CA, USA) was added to each thawed sample. Using a Retsch TissueLyser II and 5 mm stainless steel beads (Qiagen, Germantown, MD, USA), samples were homogenized at 20 Hz for 2 × 2 min. After centrifugation at 16,000 × g for 30 s to remove debris, RNA was isolated using the Direct-zol™ RNA Miniprep kit according to the manufacturer’s instructions.
For protein samples, 150 μl of Cell Lysis Buffer (Cell Signalling Technology, Danvers, MA, USA) supplemented with Halt™ Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific, Loughborough, UK) was added and samples stored at -80 °C. Thawed samples were homogenized using an OMNI Tissue Homogenizer with protein lysates clarified and quantified as previously described [
24].
Gene expression analysis
RNA from cells, murine tissue and human PCLS was converted into cDNA using Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI, USA), SuperScript® IV reverse transcriptase (Invitrogen, Carlsbad, CA, USA) and High-Capacity RNA-to-cDNA™ (Applied Biosystems, Vilnius, Lithuania) kits respectively. All reverse transcription reactions were performed according to the appropriate manufacturer’s instructions. cDNA from cells and murine lung tissue was analysed by qPCR using oligonucleotide primers (Eurofins, Luxemburg), KAPA SYBR® FAST qPCR master mix (Roche, Basel, Switzerland) and a MxPro3005 QPCR system (Agilent Technologies, Manchester, UK). All reactions were performed using the following program: initial denaturation at 95 °C for 30 s followed by 40 cycles of 95 °C for 5 s, 60 °C for 30 s and 72 °C for 15 s. Amplification of a single DNA product was confirmed by melting curve analysis. The following primer sequences were used for human (h) and mouse (m): hACTA2 forward 5’-TGTGCTGGACTCTGGAGATG-3’, reverse 5’-GACAATCTCACGCTCAGCAG-3’; hβ2M forward 5’-AATCCAAATGCGGCATCT-3’, reverse 5’-GAGTATGCCTGCCGTGTG-3’; hCOL1A1 forward 5’-CCAGCAAATGTTCCTTTTTG-3’, reverse 5’-AAAATTCACAAGTCCCCATC-3’; hCXCL8 forward 5’-ATGACTTCCAAGCTGGCCGTGGCT-3’, reverse 5’-TCTCAGCCCTCTTCAAAAACTTCTC-3’; hIL6 forward 5’-CAATAACCACCCCTGACCCA-3’, reverse 5’-GCGCAGAATGAGATGAGTTGTC-3’; hIL33 forward 5’-CCTGTCAACAGCAGTCTACT-3’, reverse 5’-TTGGCATGCAACCAGAAGTC-3’; mHPRT forward 5’-CCAGCAGGTCAGCAAAGAACT-3’, reverse 5’-TGAAAGACTTGCTCGAGATGTCA-3’ and mIL33 forward 5’-CACATTGAGCATCCAAGGAA-3’, reverse 5’-AACAGATTGGTCATTGTATGTACTCAG-3’. cDNA from PCLS was analysed by qPCR using TaqMan™ Assays, TaqMan™ Fast Advanced Master Mix and a 7900HT Fast Real-Time PCR System (Applied Biosystems). The following TaqMan™ Assays were purchased from Applied Biosystems: ACTA2 (Hs00426835_g1), β2M (Hs00187842_m1), COL1A1 (Hs00164004_m1) and FN1 (Hs01549976_m1) with all reactions performed as follows: initial denaturation at 95 °C for 20 s followed by 40 cycles of 95 °C for 1 s and 60 °C for 20 s. Using MxPro qPCR software (Agilent Technologies) or SDS 2.4 software (Applied Biosystems), cycle threshold (Ct) values for both the target and housekeeping genes (human β2-microglobulin (β2M) and murine HPRT) were determined. All data was analysed using the ΔΔCT method with the expression of each gene of interest, relative to a housekeeper, calculated and presented as a fold change versus the stated control condition/group.
Western blotting
Cell lysates were loaded into 12.5% bis/acrylamide (w/v) gels as previously described [
29]. PCLS protein lysates were loaded into Bolt™ 4–12% gels (Invitrogen). Western blotting was performed and analysed as previously described [
29,
33]. Primary antibodies used for western blotting included goat anti-IL-33 (AF3625; R&D systems), goat anti-ST2 (AF523; R&D systems), mouse anti-α-Tubulin (TU-02; Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-Fibronectin (ab2413; Abcam) and rabbit anti-GAPDH (ab181603; Abcam). Anti-goat HRP-conjugated secondary antibody (HAF109; R&D systems) and anti-rabbit (P0448) and anti-mouse (P0447) HRP-conjugated secondary antibodies (Dako, Glostrup, Denmark).
Statistical analysis
Statistical analysis was performed using GraphPad Prism 8 with P values ≤ 0.05 considered statistically significant. The distribution of all data was determined by Shapiro–Wilk normality test. Data are reported as either mean (parametric) or median (non-parametric) with all individual data points shown where appropriate. For parametric data, unpaired and paired t-tests were used to determine statistical significance between two unmatched and matched groups whilst One- and Two-way (ordinary and repeated measures) ANOVAs were used to compare multiple (unpaired and paired) groups and variables. For non-parametric unpaired data, statistical significance between two or more groups was assessed by Mann–Whitney U test and Kruskal–Wallis test respectively. For non-parametric paired data, statistical significance between two or more groups was assessed by Wilcoxon signed-rank test and Friedman test respectively.
Discussion
IPF is a debilitating interstitial lung disease with a poor prognosis and limited treatment options. Despite recent work suggesting that IL-33 may play an important fibrogenic role in IPF [
11‐
19], the use of prophylactic dosing regimens in all previous studies means that the therapeutic benefit of targeting IL-33:ST2 signalling in IPF is poorly understood. Here we show that despite being expressed in fibroblasts and upregulated by treatment with the potent pro-fibrotic cytokine TGFβ, extracellular IL-33 has no direct effect on the pro-fibrotic activity of these cells. In addition, using the BLM mouse model of lung fibrosis, we demonstrate that therapeutic dosing with an ST2-Fc fusion protein has no effect on the severity of established BLM-induced fibrosis. Furthermore, stimulation with IL-33 does not induce pro-fibrotic changes in human PCLS. Considered together, these findings suggest that the IL-33:ST2 axis is unlikely to play a central fibrogenic role in IPF.
Due to their role as one of the key effector cell types during the development of IPF [
1], the expression of IL-33 by fibroblasts in ILD lung sections was assessed by IHC. Although we cannot make definitive conclusions in the absence of cell-type specific markers, the spindle-like morphology of some IL-33 positive cells, our IL-33 gene and protein data in IPF HLFs and the reported expression of IL-33 by freshly isolated [
35] and cultured [
14] IPF fibroblasts, collectively suggest that a proportion of the IL-33 positive cells in our ILD lung sections were fibroblasts.
In agreement with our IHC data, IPF HLFs were found to express IL-33 although in contrast with a previous report we detected no difference in expression between non-IPF and IPF cells maintained in culture [
14]. These data are supported by single-cell RNA sequencing data from the IPF cell atlas [
35] which indicates that IL-33 is not overexpressed in interstitial lung fibroblasts from IPF patients. Although there is no clear explanation for the different patterns of IL-33 expression in non-IPF and IPF fibroblasts in our study and that of Luzina and colleagues, it may reflect differences in culture conditions and their inclusion of data from scleroderma patients [
14].
To determine if IL-33 expression by fibroblasts could be regulated by the pro-fibrotic microenvironment reported in the lungs of IPF patients, HLFs were stimulated with TGFβ. Interestingly, TGFβ increased IL-33 expression by HLFs in our study. As TGFβ has been previously shown to decrease IL-33 expression by smooth muscle cells [
36] and alveolar epithelial cells [
37], our results suggest that TGFβ-induced IL-33 expression is cell type specific. Moreover, given the increased numbers of fibroblasts in IPF tissue [
1], the elevated levels of TGFβ [
38], and the ability of this cytokine to drive epithelial to mesenchymal transition [
39], these findings may explain the higher expression of IL-33 reported in the lung tissue of IPF patients [
13,
14].
Although cell-associated IL-33 protein levels returned to baseline 24 h after stimulation, no secreted IL-33 could be measured in supernatants collected from TGFβ stimulated HLFs. As cell death by necrosis [
5], mechanical injury [
4], or viral infection [
6] is required for IL-33 release from other cell types, it is possible that excess TGFβ-induced IL-33 is degraded in the proteasome [
40] of HLFs rather than being released. However, as IL-33 release in the absence of cell death has also been reported [
41,
42], it is also possible that IL-33 is released from TGFβ treated HLFs but at a level below the lower limit of detection of the assay used [
43].
In the majority of our western blots assessing IL-33 expression by HLFs, a band at 22/23 kDa was detected in addition to the expected band at 31 kDa. As the full-length 31 kDa protein can be processed by cell-derived proteases [
44], it is possible that this band represents a cleaved form of IL-33. Based on reports from Scott et al. [
43] using a similar protein extraction method, we suspect the 22/23 kDa band is likely an artefact of mechanical cell lysis. As we are uncertain of the biological relevance of this band in our experiments, only bands at 31 kDa have been labelled and quantified as IL-33.
To assess whether IL-33 secreted either from fibroblasts themselves or from other cell types within the lung could mediate pro-fibrotic effects on fibroblasts, we stimulated non-IPF and IPF HLFs with exogenous IL-33 and found no evidence of increased expression of pro-inflammatory or pro-fibrotic genes suggesting that this cytokine does not promote fibrogenesis. Although studies assessing human [
45] and murine cells [
46] have demonstrated fibroblast responsiveness to IL-33, our results are consistent with those of Yagami and colleagues [
47] and suggest that extracellular IL-33 cannot directly increase the pro-fibrotic activity of fibroblasts during the development of IPF as they lack the IL-33 receptor ST2.
Given the predominantly nuclear pattern of IL-33 expression in IPF lung tissue, it is possible that TGFβ-induced IL-33 has intracellular rather than extracellular pro-fibrotic effects on HLFs. However, a recent report by Luzina et al. determined that although overexpression of nuclear IL-33 in fibroblasts resulted in phosphorylation of SMAD3, it did not induce collagen gene transcription and actually attenuated TGF-β-induced levels of collagen I and III mRNAs [
48]. Additionally, using human fibroblasts and an in vivo model of unilateral ureteral obstruction, Gatti et al. recently suggested a novel role for nuclear IL-33 as a repressor of interstitial cell extracellular matrix deposition rather than as a mediator of fibrosis [
49]. As comprehensive IL-33 knockdown failed to regulate the expression of any proteins other than IL-33 by endothelial cells [
50] and deletion of the nuclear localisation signal of IL-33 resulted in a fatal hyperinflammatory response in vivo [
51], it is possible that the nuclear pattern of IL-33 expression observed in fibrotic lung tissue may reflect the importance of nuclear storage in controlling the extracellular effects of IL-33 instead of an intracellular, fibrogenic role in transcription.
Consistent with previous studies using ST2 knockout mice [
16,
18], IL-33 neutralising antibodies [
16], and soluble ST2 [
17], treatment with the ST2-Fc fusion protein reduced the number of lymphocytes and neutrophils in the BAL of BLM treated mice. As adenoviral overexpression [
13,
14] and intranasal delivery of recombinant IL-33 [
16] has been reported to increase lymphocyte and neutrophil numbers, our results, in combination with previously published data, suggest that IL-33 signalling promotes inflammation in the early injury phase of the BLM model of lung fibrosis.
In contrast with previous studies targeting IL-33:ST2 signalling [
16‐
19], the ST2-Fc fusion protein failed to reduce BLM-induced fibrosis. Considering IL-33 is a potent pro-inflammatory cytokine [
9], the reduced fibrogenesis reported in all previous studies likely reflects a reduction in the inflammatory response required for the development of BLM-induced fibrosis, rather than a true anti-fibrotic effect. Indeed, as all previous studies in the BLM mouse model have blocked IL-33 signalling prior to or during the inflammatory phase [
16‐
19] we believe that our therapeutic dosing approach, intervening only during the fibrotic phase, explains why our findings differ from other reported studies. Furthermore, since IL-33 had no direct effect on HLFs and failed to induce pro-fibrotic changes in human PCLS, we propose that the IL-33:ST2 axis in isolation is unlikely to play a key pro-fibrotic role in IPF.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.