Full-length IL-33 augments pulmonary fibrosis in an ST2-and Th2-independent, non-transcriptomic fashion

Abstract

Mature IL-33 (MIL33) acting through its receptor, ST2, is known to regulate fibrosis. The precursor, full-length IL-33 (FLIL33), may function differently from MIL33 and independently of ST2. Here we report that genetic deletion of either IL-33 or ST2 attenuates pulmonary fibrosis in the bleomycin model, as does Cre-induced IL-33 deficiency in response to either acute or chronic bleomycin challenge. However, adenovirus-mediated gene delivery of FLIL33, but not MIL33, to the lungs of either wild-type or ST2-deficient mice potentiates the profibrotic effect of bleomycin without inducing a Th2 phenotype. In cultured mouse lung cells, FLIL33 overexpression induces moderate and distinct transcriptomic changes compared with a robust response induced by MIL33, whereas ST2 deletion abrogates the effects of both IL-33 forms. Thus, FLIL33 may contribute to fibrosis in an ST2-independent, Th2-independent, non-transcriptomic fashion, suggesting that pharmacological targeting of both FLIL33 and MIL33 may prove efficacious in patients with pulmonary fibrosis.

Introduction

Pulmonary fibrosis, the defining feature of interstitial lung disease (ILD), continues posing a major therapeutic challenge, particularly in patients with idiopathic pulmonary fibrosis (IPF), systemic sclerosis (SSc; scleroderma)-associated ILD, rheumatoid arthritis (RA)-ILD, and several other connective tissue diseases. There is no cure for pulmonary fibrosis, and the available therapies only alleviate some of the symptoms and slow disease progression, at best. A better understanding of disease mechanisms needs to be attained to allow for identification of new therapeutic targets and subsequent development of novel, more efficacious, remedies for lung fibrosis. Interleukin (IL)-33, an essential regulator of a remarkably wide array of homeostatic and pathological processes [1, 2], regulates pulmonary fibrosis [3, 4] and so does its cell-surface receptor, ST2 [4, 5]. Pulmonary IL-33 overexpression and bleomycin challenge synergized in inducing lung fibrosis in mice [3], whereas either treatment of bleomycin-challenged mice with neutralizing anti-IL-33 antibody or genetic deletion of ST2 attenuated collagen deposition [4, 5]. Considering the specific functional link between IL-33 and ST2, the notion of the IL-33 – ST2 axis controlling lung fibrosis and posing a viable therapeutic target thus appears plausible.

However, several inconsistencies with this notion should be noted and clarified to enable rational targeting of IL-33 and/or ST2 in pulmonary fibrosis. Less bleomycin-induced collagen seemed to accumulate in anti-IL-33-treated wild-type mice than in ST2 KO mice on the same C57BL/6 background (compare Fig. 2E and Fig. 1E in [4]). Others reported a statistically significant, yet moderate degree of protection against bleomycin-induced collagen accumulation in ST2 KO than in wild-type mice on C57BL/6 background (see Fig. 4B in [5]). Yet others noted that ST2 deletion did not protect against collagen accumulation in less bleomycin-sensitive BALB/c mice [3]. Taken together, these observations suggest a possibility of a partial functional disconnect between IL-33 and ST2 in the regulation of lung fibrosis. A possible explanation for such a disconnect rests on the dual nature of IL-33, which is active as both a full-length IL-33 (FLIL33) precursor and a mature IL-33 (MIL33) cytokine [6, 7]. In this report, we use the designation “IL-33” to generally refer to either or both forms of the molecule, whereas specific references to the FLIL33 precursor and the MIL33 cytokine are used to emphasize their distinguishing characteristics. Previous studies focused mostly on the MIL33 – ST2 axis in organ fibroses [8–22] and, specifically, in the bleomycin model of ILD [4, 5, 23], whereas FLIL33 has received less attention [3, 24]. MIL-33 acts exclusively through its specific cell-surface receptor, ST2, forming the IL-33 – ST2 signaling axis. FLIL33 can also bind to and act through ST2, although with an ~10-fold lesser potency [7, 25–29]. Unlike MIL33, FLIL33 can also function in an ST2-independent fashion [3, 24, 30–33]. Functionally, the effects of FLIL33 may be partially similar to [7, 25–28, 34] or notably different from [3, 24, 30–44] those of the MIL33 cytokine. Adding to this complexity, ST2 exerts opposite functions in its membrane-bound, signaling form, and in its soluble, IL-33-neutralizing “decoy receptor” form. Furthermore, ST2 may signal in an IL-33-independent fashion [45].

Figure 2.

Pulmonary effects of induced (A – D) and constitutive (E) IL-33 deficiency in the chronic bleomycin injury model. A. Time course schematic of the chronic bleomycin injury model in Cre^tg/tgIL33^fl/fl mice. Mice were treated with tamoxifen (TAM) to induce IL-33 deficiency or with oil control during the 4^th week of life. Starting at week 12, repeated intraperitoneal injections of bleomycin (BLM) or PBS control were performed twice a week for 33 days. B. Trichrome staining of lung sections at lower (main) and higher (inset, upper right) magnifications from Cre^tg/tgIL33^fl/fl mice induced with tamoxifen (TAM) or oil control and treated with bleomycin (BLM) or PBS control, as indicated. The black scale bar indicates 300 μm for to the lower magnification. The peripheral and subpleural collagen deposits appear in blue; selected areas of interstitial and pleural collagen deposition are indicated with arrowheads. C. Quantitative assessment of collagen staining-occupied (left) and nuclear staining-occupied (right) areas in a total of 10 randomly selected subpleural microscopic fields from 3 mice per treatment group. Each circle represents an individual microscopic field. D. Mean total lung collagen, μg per mg of wet lung tissue ± SD (left), and mean IL-33, pg/ml ± SD (middle), in lung homogenates from mice treated as indicated. The scatterplot on the right shows correlation between individual total lung collagen and pulmonary IL-33 levels, with each circle representing a separate mouse. E. Collagen levels, μg per mg wet tissue ± SD, in the lungs of wild-type and IL-33 KO (constitutive ubiquitous IL-33 deletion) mice. Animals were treated with PBS or BLM in the chronic model as indicated. In panels C, D, and E, single asterisks indicate significant (p < 0.05) elevations induced by bleomycin compared to PBS-treated controls, whereas double asterisks indicate the attenuated (p < 0.05) response to bleomycin challenge in tamoxifen-induced (C, D) or IL-33 KO (E) mice.

Figure 1.

Pulmonary effects of induced IL-33 deficiency in the acute bleomycin injury model. A. Time course schematic of the acute bleomycin injury model in Cre^tg/tgIL33^fl/fl mice. Mice were treated with tamoxifen (TAM) to induce IL-33 deficiency or with oil control during the 4^th week of life. A single intratracheal instillation of bleomycin (BLM) or PBS control was administered in the beginning of the 12^th week of life, with analyses performed 2 weeks later. B. Trichrome staining of lung sections at lower (main) and higher (inset, upper right) magnifications from mice induced with tamoxifen or oil and challenged with bleomycin or PBS, as indicated. The black scale bar indicates 300 μm for the lower magnification. Collagen deposits appear in blue; selected areas of interstitial collagen deposition are indicated with arrowheads. C. Quantitative assessment of collagen staining-occupied (left) and nuclear staining-occupied (right) areas in a total of 10 randomly selected microscopic fields from 3 mice per treatment group. Each circle represents an individual microscopic field. D. Mean total lung collagen, μg per mg of wet lung tissue ± SD (left), and mean IL-33, pg/ml ± SD (middle), in lung homogenates from mice treated as indicated. The scatterplot on the right shows correlation between individual total lung collagen and pulmonary IL-33 levels using the same data, with each circle representing a separate mouse. E. Total and differential cell counts in mice treated as indicated. In panels C, D, and E, single asterisks indicate higher (p < 0.05) levels in oil-treated, bleomycin-challenged mice compared to oil-treated, PBS-challenged controls. Double asterisks indicate lower (p < 0.05) levels in tamoxifen-induced, bleomycin-challenged mice compared to oil-treated, bleomycin-challenged mice.

Figure 4.

Pulmonary effects of ST2 deficiency on the response to IL-33 overexpression combined with acute bleomycin injury. A. Time course schematic of the combined IL-33 overexpression and bleomycin injury. Wild-type or ST2 KO mice received intratracheal instillation of IL-33-encoding adenoviruses (AdV-FLIL33, AdV-MIL33, or AdV-NULL control) and were challenged with intratracheal bleomycin 7 days later. Readouts were performed after an additional 14 days. B, C. Mean total and differential cell counts ± SD in bronchoalveolar lavage samples from AdV-NULL-, AdV-FLIL33-, or AdV-MIL33-infected, bleomycin-challenged wild-type (B) and ST2 KO (C) mice. Single asterisks indicate differences (p < 0.05) in AdV-FLIL33-infected compared with AdV-NULL-infected mice, whereas double asterisks indicate differences (p < 0.05) in AdV-MIL33-infected compared with AdV-FLIL33-infected mice. D. Periodic-acid Schiff staining of lung sections from wild-type and ST2 KO mice infected with AdV-MIL33. Arrows point to selected areas of airway mucus accumulation. E. Total lung collagen, μg per mg wet lung tissue ± SD, in AdV-NULL-, AdV-FLIL33-, or AdV-MIL33-infected, bleomycin-challenged wild-type and ST2 KO mice. Single asterisks indicate differences (p < 0.05) from control mice that were not infected with an AdV and were challenged with intratracheal PBS instead of bleomycin, whereas double asterisks indicate differences (p < 0.05) in AdV-FLIL33-infected or AdV-MIL33-infected compared with AdV-NULL-infected mice.

The potential complexity of the IL-33 – ST2 interplay in pulmonary fibrosis is further suggested by the discrepancy between, on the one hand, a potent ST2-mediated pro-Th2 function of MIL33 in the lungs [24] and, on the other hand, modest to absent Th2 features in many cases of pulmonary fibrosis. Indeed, MIL33 is well known to induce, in a strictly ST2-dependent fashion, a robust skewing of the cytokine milieu towards the characteristic type 2 pattern (IL-4, IL-5, IL-13), accumulation of eosinophils, and hyperplasia of mucus-producing epithelial goblet cells. However, despite confirmed IL-33 and ST2 elevations in the majority of ILDs, including IPF, neither humans with ILD nor mice in the bleomycin-induced models of lung fibrosis overtly manifest such a pulmonary phenotype. Furthermore, human patients with allergic asthma and experimental animals in asthma models manifest the characteristic IL-33 – ST2 axis-associated type 2 phenotype, yet the nature and severity of fibrosis in both humans and mice are underwhelming compared to ILD and its models. These combined considerations outlined above suggest a possibility that IL-33 may regulate fibrosis through mechanisms beyond the classical MIL33 – ST2 axis. Such a possibility was addressed in this study.

Materials and Methods

Experimental animals

The animal studies were performed in accordance with a research protocol reviewed and approved by the University of Maryland Institutional Animal Care and Use Committee. Animals were maintained in sterile microisolator cages with sterile rodent feed and water. Daily maintenance of mice was performed at the Baltimore VA Medical Center Research Animal Facility and University of Maryland Animal Facility, which are approved by the Association for Assessment and Accreditation of Laboratory Animal Care. All mouse strains were on the C57BL/6 background. Wild-type (WT) C57BL/6 mice aged 10–12 weeks (The Jackson Laboratory, Bar Harbor, ME), IL-33 germline-deleted mice, and ST2 germline-deleted mice were used. IL33^‒/‒ (IL-33 KO) mice [46] were kindly provided by Dr. Roland Kolbeck (MedImmune AstraZeneca) and ST2^‒/‒ (ST2 KO) mice [47] were kindly provided by Dr. Andrew N.J. McKenzie (Medical Research Council Laboratory of Molecular Biology, Cambridge, UK). Each of these strains was backcrossed to the C57BL/6 background for more than 10 generations.

We also developed an inducible model of IL-33 deficiency by homozygously floxing exons 5 – 7 and breeding these IL-33^fl/fl mice with R26-CreERT2 mice (strain #008463, The Jackson Laboratory) to homozygocity for Cre recombinase expression, which is designated here as Cre^tg/tg. The resulting Cre^tg/tgIL-33^fl/fl mice were further bred to homozygosity on the C57BL/6 background through sibling mating for more than 10 generations and used to achieve tamoxifen (TAM)-inducible IL-33 deficiency. For some experiments, additional mice were generated by crossing Cre^tg/tgIL-33^fl/fl with Ai14(RCL-tdT)-D (strain #007914, The Jackson Laboratory), which express tdTomato (tdT) fluorescence following Cre-mediated recombination, with subsequent backcrossing to Cre^tg/tgIL-33^fl/fl. The resulting strain IL-33^fl/flCre^tg/‒tdT^‒/tg on the C57BL/6 background was homozygously floxed for IL-33 and was heterozygously transgenic for each Cre and tdTomato. To induce Cre-based recombination, 2 mg/mouse/day of tamoxifen dissolved in corn oil was injected daily in 4-week-old mice for 5 days; control mice were injected with oil vehicle. After 8 weeks of recovery from possible Cre-independent confounding effects of tamoxifen, mice were tested in bleomycin injury models.

Bleomycin injury models

Acute intratracheal, as well as chronic systemic, bleomycin injury models were used. To induce acute bleomycin injury model manifesting in pulmonary inflammation and fibrosis, a single dose of 0.075 U of bleomycin (Sigma-Aldrich, St. Louis, MO) diluted in 50 μL of sterile phosphate-buffered saline (PBS) was delivered intratracheally to mice on day 0, as described [3, 48–51]. Briefly, a minor anterior midline neck incision was made to reveal the trachea, and a Micro Sprayer (Penn-Century, Wyndmoor, Philadelphia, PA) was inserted intratracheally and bleomycin or sterile PBS instilled. On day 14, mice were euthanized by CO₂ asphyxiation, followed by cervical dislocation.

Immediately postmortem, bronchoalveolar lavage (BAL) samples and lungs were harvested. BAL samples were collected and analyzed as described [3, 48–51]. Briefly, installation and withdrawal of 1 mL of PBS twice via an 18-gauge blunt-end needle secured in the trachea were performed in each animal. The two aliquots of BAL were pooled and centrifuged. Total and differential cell counts in BAL samples were performed after staining of cytospin preparations with a PROTOCOL Hema 3 manual staining system (Thermo Fisher Scientific, Kalamazoo, MI) by at least two technicians who were blinded to the identity of the samples.

The lungs were processed for histological analysis by Masson’s trichrome staining. Stained sections were scanned in their entirety with a 20× objective and digitally stored utilizing an Aperio Digital Pathology system (Leica BioSystems, Vista, CA). To quantitatively assess accumulation of collagen fibers based on trichrome staining, internally developed software, TriChromePick, was used to isolate blue-stained collagen fibers and dark-purple-stained cell nuclei. The software also calculates the total area of the microscopic field occupied by collagen and, separately, by nuclei. For each experiment in which trichrome-stained lung sections were analyzed, a total of 10 visual fields were randomly selected in sections from 3 mice for TriChromePick-based quantification.

Lung homogenates were analyzed by ELISA for IL-33 (R&D Systems, Minneapolis, MN), which measures both FLIL33 and MIL33 forms. Multiplex analyses for IL-4, IL-5, and IL-13 protein levels (Luminex, Austin, TX) were performed. Total collagen protein was measured based on the quantitative colorimetric determination of hydroxyproline obtained by acid hydrolysis of lung tissues, utilizing QuickZyme assays (QuickZyme BioSciences, Leiden, The Netherlands) according to the manufacturer’s recommendations, as previously described [3, 48–51].

To induce chronic bleomycin injury [50, 51], mice were injected intraperitoneally with 0.018 U of bleomycin or PBS control twice a week for 33 days. Mice were euthanized and the lungs extracted and processed in a similar fashion. Considering the limited inflammatory component in the chronic model [50, 51], changes in the cellular BAL composition were not analyzed.

Double-hit in vivo model based on IL-33 overexpression and acute bleomycin injury

To overexpress FLIL33 or MIL33 in mouse lung in the double-hit model [3], recombinant replication-deficient adenoviral (AdV) constructs were prepared and used as previously described [3, 24, 38, 49, 52–57]. The AdV-FLIL33 or AdV-MIL33 vectors encoding mouse FLIL33 or MIL33, respectively, or the AdV-NULL vector not encoding a cytokine [3, 24, 57] were instilled intratracheally, followed by intratracheal bleomycin challenge 7 days later. Analyses of pulmonary changes were performed on day 14 as described above for the acute bleomycin injury model.

RNA-Seq

Immediately following euthanasia, mouse lungs from wild-type and ST2 KO animals were perfused through the right ventricle with 10 ml of ice-cold PBS. Lungs were isolated, minced, and digested with collagenase and DNAse I (both from Sigma-Aldrich, St. Louis, MO). Cells were strained through a 70 μm filter and cultured at 37° C in a 5% CO₂ humidified air atmosphere, in T25 flasks at 5×10⁶ cells/flask, in RPMI medium supplemented with 0.5% dialyzed fetal calf serum. In one experiment, immediately after preparation, unseparated lung cells were infected in culture with mouse IL-33 isoform-encoding adenoviruses (AdV) described previously [24]: AdV-FLIL33, AdV-MIL33, or AdV-NULL at 2×10¹¹ plaque-forming units (pfu) per flask. In another experiment, primary mouse lung fibroblasts were enriched from such unseparated lung cells by two cycles of passaging plastic-adherent cells in RPMI medium supplemented with 10% serum. Before AdV infection, cells were rested for 24 h in the medium containing 0.5% dialyzed fetal calf serum, and then infected with AdV constructs similar to unseparated lung cells as described above. In each experiment, one culture from wild-type and one culture from ST2 KO mice was infected with AdV-FLIL33, one culture from wild-type and one culture from ST2 KO mice was infected with AdV-MIL33, and two cultures from wild-type and two cultures from ST2 KO mice were infected with AdV-NULL. In both experiments, infected cells were cultured for 72 h and total RNA was isolated using Trizol Reagent (Life Technologies, Grand Island, NY) and quality-controlled by capillary electrophoresis. Transcriptomic profiling of unseparated, AdV-infected lung cells was performed using RNA-Seq methodology at 1×75 single-end reads, >45 million reads per sample (Cofactor Genomics, Saint Louis, MO). AdV-infected primary mouse lung fibroblasts were also analyzed using RNA-Seq at 2×100 paired-end base reads, >50 million reads per sample (Beckman Coulter Genomics, Danvers, MA). The sequencing data and mouse reference transcriptome gencode.vM24 were used for quantification of transcripts using Salmon v1.6.0 software in quasi-mapping-based mode [58], with subsequent summarization to gene counts using tximport v1.14.2 [59]. Differential expression was analyzed using the DESeq2 package of Bioconductor [60]. The DESeq2-based pairwise comparisons were performed to identify genes that were differentially expressed depending on experimental conditions. The DESeq2 results were wrangled to separate up- and down-regulated genes in these pair-wise comparisons between experimental conditions, based on an arbitrarily chosen 2.0-fold change cutoff and p value ≤ 0.05. The RNA-Seq data have been deposited in the NCBI GEO database (accession number GSE197398). Gene functional enrichment analysis was performed using Metascape [61] in the express analysis mode.

Statistical analyses

Experimental data other than RNA-Seq were expressed as mean ± standard deviation values. Differences between sample groups were calculated using a two-tailed Student’s t-test.

Results

Inducible IL-33 deficiency protects against pulmonary fibrosis in the acute bleomycin injury model

The bleomycin model of lung fibrosis has been tested in ST2 KO mice in the past [3–5], where genetic IL-33 deficiency, to our knowledge, has not. We tested the effect of induced IL-33 deficiency by utilizing Cre^tg/tgIL33^fl/fl mice. The induction of Cre by tamoxifen was confirmed based on appearance and maintenance of red fluorescence in IL-33^fl/flCre^tg/‒tdT^‒/tg mice (Suppl. Fig. 1A), and the functionality of IL-33 depletion was inferred from attenuated pulmonary changes in tamoxifen-treated, bleomycin-challenged Cre^tg/tgIL-33^fl/fl mice compared to tamoxifen-treated, bleomycin-challenged wild-type controls (Suppl. Fig. 1B). We assessed the effect of inducible IL-33 depletion on the degree of pulmonary collagen accumulation in the traditional, acute, model of bleomycin challenge, which is based on a single intratracheal instillation of bleomycin (Fig. 1A). Based on trichrome staining, the lungs of tamoxifen-induced, PBS-treated mice appeared histologically indistinguishable from the lungs of unmanipulated mice (Fig. 1B, left). Indicative of the protective effect of IL-33 depletion, tamoxifen-induced, bleomycin-challenged mice manifested less pronounced pulmonary inflammatory infiltration and collagen accumulation in the acute bleomycin model (Fig. 1B, right) than did oil-induced, bleomycin-challenged controls (Fig. 1B, middle). To quantitatively assess histological changes, internally developed software, TriChromePick, was used, which isolates and quantitates areas occupied by collagen stain and nuclear stain in trichrome-stained tissue sections (Suppl. Fig. 2).

Consistent with the well-known patchy pulmonary response to acute bleomycin challenge, percent area occupied by collagen (Fig. 1C, left) as well as by nuclei (Fig. 1C, right) varied notably among the randomly selected microscopic fields within each of the tested groups of mice. The collagen and nuclear areas were elevated in response to bleomycin challenge in both tamoxifen- and oil-induced mice. Tamoxifen-induced depletion of IL-33 tended to attenuate (p = 0.10) the collagen-occupied (Fig. 1C, left) but not nuclei-occupied (Fig. 1C, right) areas, further suggesting a protective effect of IL-33 depletion on bleomycin-induced collagen accumulation.

To quantitatively account for the entirety of the effects IL-33 depletion on the whole lung, total lung collagen (Fig. 1D, left) and pulmonary levels of IL-33 protein (Fig. 1D, middle) were measured. Tamoxifen-induced mice accumulated significantly less collagen in response to the bleomycin challenge while demonstrating a correlative link between collagen and IL-33 levels (Fig. 1D, right). Despite seemingly attenuated parenchymal infiltration in IL-33-depleted mice (Fig. 1B), there were no differences in BAL cellularity between tamoxifen-induced and control oil-induced mice, including in BAL lymphocyte counts (Fig. 1E), nor did IL-33 depletion alleviate the decline in the body weight dynamics (Suppl. Fig. 3A). These findings indicate that IL-33 depletion attenuated collagen accumulation in the lungs, with limited effect on pulmonary inflammation and the overall disease severity.

Inducible IL-33 deficiency and genetic IL-33 deletion protects against pulmonary fibrosis in the chronic bleomycin injury model

Previous studies demonstrated the role of IL-33 in the acute bleomycin injury model [3–5], which is based on a single pulmonary delivery of bleomycin and characterized by a pronounced combination of inflammation and fibrosis. Despite its common use, the acute model does not fully represent the chronic nature of human ILD. The most debilitating and deadly human ILDs, including IPF, SSc-ILD, and RA-ILD, are characterized by their chronic nature, less pronounced inflammatory component, and more pronounced, geographically heterogeneous, often basolaterally distributed, fibrosis. Repeated systemic administration of bleomycin more precisely recapitulates human ILD, including less pronounced inflammation and the peripheral distribution of fibrosis in the lung [62–64], and in some instances, produces the results that are not only more relevant to human ILD but also different from those observed in the acute bleomycin model [51]. We therefore assessed the effect of tamoxifen-induced IL-33 depletion on the degree of pulmonary collagen accumulation in bleomycin-challenged mice in the more clinically relevant chronic bleomycin model (Fig. 2A).

Trichrome staining of the lungs revealed that repeated systemic administration of PBS did not induce histological changes in tamoxifen-induced mice (Fig. 2B, left), whereas repeated administration of bleomycin induced peripheral, subpleural histoarchitectural distortions in oil-treated controls (Fig. 2B, middle); bleomycin-induced changes were less pronounced in tamoxifen-induced controls (Fig. 2B, right). Quantification of collagen staining and inflammatory infiltrates in the peripheral areas of trichrome-stained lungs with TriChromePick (Suppl. Fig. 2) revealed an increase in collagen fibers (Fig. 2C, left), whereas the increase in nuclear staining tended to be somewhat elevated without reaching statistical significance (Fig. 2C, right). Tamoxifen-induced depletion of IL-33 attenuated nuclear staining (Fig. 2C, left). Considering the typical geographic heterogeneity of tissue engagement in the fibrotic response, which is reflected in the variable numerical assessment by TriChromePick, quantitative assessment of the whole lung was performed. Compared to oil-induced mice, tamoxifen-induced animals accumulated significantly less total lung collagen (Fig. 2D, left) and expressed significantly less IL-33 (Fig. 2D, middle) in response to the bleomycin challenge, while demonstrating a correlation between collagen and IL-33 levels (Fig. 2D, right). Tamoxifen-induced IL-33 deficiency did not alleviate the decline in the body weight dynamics after chronic bleomycin injury (Suppl. Fig. 3B). Due to the notably less prevalent pulmonary inflammation in such chronic model [50, 51], evaluation of changes in BAL cellularity was not performed. Constitutive IL-33 deletion similarly attenuated bleomycin-induced collagen accumulation to a pronounced degree (Fig. 2E). The findings presented in Figs. 1 and 2 indicate that IL-33 depletion was similarly protective against collagen accumulation in both acute and the more clinically relevant models of pulmonary fibrosis. Therefore, subsequent experiments were performed in the shorter, acute bleomycin injury, model.

Overexpression of FLIL33, but not MIL33, promotes pulmonary fibrosis independently of ST2

Consistent with the previous reports of others [4, 5], mice with constitutive ubiquitous ST2 deletion also accumulated less collagen than wild-type mice in response to bleomycin injury (Fig. 3A), but the decline in bleomycin-induced collagen accumulation in ST2 KO mice appeared to be less pronounced than the decline caused by IL-33 depletion (compare Fig. 3A to Fig. 1D, left). Whereas IL-33 depletion did not affect BAL cellularity, including lymphocyte counts (Fig. 1E), ST2 deletion significantly attenuated BAL cell counts, particularly of lymphocytes (Fig. 3B). Thus, in contrast to IL-33 depletion, ST2 deletion appears to have a significant but less pronounced attenuating effect on pulmonary collagen and a more pronounced effect on accumulation of lymphocytes in BAL. These combined findings suggest that the mechanisms by which IL-33 and ST2 control the severity of ILD may be in part dissociable.

Figure 3.

Pulmonary effects of constitutive ST2 deficiency in the acute bleomycin injury model. A. Collagen levels in mouse lungs, μg per mg wet tissue ± SD, in wild-type and ST2 KO mice challenged with bleomycin or PBS as indicated. B. Mean total and differential BAL cell counts ± SD in the same experiment. In both panels, single asterisks indicate differences (p < 0.05) in bleomycin-challenged wild-type mice compared to PBS-challenged controls, whereas double asterisks indicate differences (p < 0.05) in bleomycin-challenged ST2 KO mice compared to bleomycin-challenged wild-type mice.

To explore the possibility that IL-33 may promote fibrosis independently of ST2, we utilized a previously described alternative approach, in which IL-33 is not attenuated or deleted, but instead, overexpressed [3]. Neither FLIL33 or MIL33 overexpression alone in mouse lungs in vivo induces fibrosis [3, 24], but in a double-hit model, which combines IL-33 overexpression and bleomycin challenge, FLIL33 potently augments the profibrotic effect of bleomycin [3]. In contrast to the previous study [3], not only wild-type but also ST2 KO mice have now been studied, and not only FLIL33 but also MIL33 has been overexpressed. Mice were infected intratracheally with a replication-deficient adenovirus encoding either FLIL33, MIL33, or with the non-coding vehicle control, AdV-NULL. Seven days later, mice were challenged with a single intratracheal instillation of bleomycin, and analyzed after an additional 14 days (Fig. 4A). We reasoned that if IL-33 controls collagen accumulation in the bleomycin model exclusively through the MIL33 – ST2 axis, then the previously demonstrated potentiating effect of FLIL33 on lung fibrosis [3] would a) require maturation of FLIL33 into MIL33 in the mouse lungs, inducing a Th2 phenotype, and b) be abrogated by ST2 deletion.

Overexpression of MIL33, but not FLIL33, potently induced accumulation of eosinophils (Fig. 4B), and such MIL33-induced BAL eosinophilia was completely abrogated by ST2 deletion (Fig. 4C). Periodic acid-Schiff staining of lung tissues revealed goblet cell hyperplasia with mucus accumulation in the airways of MIL33-overexpressing wild-type, but not ST2 KO mice (Fig. 4D). Consistent with earlier observations [24], FLIL33 delivery did not induce goblet cell hyperplasia and mucus accumulation. These observations validate the conclusion that the MIL33 – ST2 axis controls such characteristic features of the Th2 phenotype as eosinophilia and goblet cell hyperplasia, yet FLIL33 does not.

If IL-33 in either form, FLIL33 and/or MIL33, controls fibrosis exclusively through ST2, then ST2 deletion should abrogate the augmenting effect of FLIL33 overexpression on the bleomycin-induced collagen accumulation [3]. Furthermore, since ST2 deletion attenuates the profibrotic effect of bleomycin alone (Fig. 3A), the effect of ST2 deletion on collagen accumulation in this double-hit model would be expected to be similar to the effect of ST2 deletion on the profibrotic effect of bleomycin alone (Fig. 3A). By contrast, both wild-type and ST2 KO FLIL33-overexpressing mice similarly accumulated more collagen in their lungs than did AdV-NULL-infected or MIL33-overexpressing mice when challenged with bleomycin (Fig. 4E). Histomorphologically, neither FLIL33-overexpressing nor AdV-NULL-infected ST2 KO mice were protected from bleomycin-induced inflammatory infiltration or collagen fiber accumulation compared to similarly treated wild-type mice (Suppl. Fig. 4A), nor were there differences in the dynamics of the total body weight (Suppl. Fig. 4B). Of note, the in vivo infections with these replication-deficient AdV constructs, including with AdV-NULL, eliminated the modest antifibrotic effect of ST2 deletion in bleomycin-challenged mice (compare Fig. 3A, in which the mice were not, and Fig. 4E, in which the mice were, infected with AdVs before the bleomycin challenge). Thus, while the effects of MIL33 are completely dependent on ST2 and are consistent with the Th2 phenotype, ST2 deletion does not protect from the potentiating effect of FLIL33 on the bleomycin-induced accumulation of pulmonary collagen. Therefore, in addition to the profibrotic regulation by the IL-33 – ST2 axis, FLIL33 contributes to collagen accumulation in an ST2-independent fashion.

A possibility cannot be excluded that the potentiating effect of FLIL33 on bleomycin-induced collagen accumulation is mediated by FLIL33 maturation into MIL33 with a subsequent induction of a mild Th2 phenotype. However, the observed lack of pulmonary eosinophilia or goblet cell hyperplasia with mucus accumulation in FLIL33-overexpressing mice does not support such a possibility. Additionally, Th2 cytokines were measured in lung homogenates (Fig. 5). A robust induction in the levels of IL-4, IL-5, and IL-13 was observed in MIL33-overexpressing wild-type but not ST2 KO mice (Fig. 5). By contrast, FLIL33 overexpression failed to induce Th2 cytokines in both wild-type and ST2 KO mice (Fig. 5). This finding is consistent with the notion that the Th2 phenotype is being driven by the MIL33 – ST2 axis but not by FLIL33 [3, 24].

Figure 5.

Th2 cytokine expression levels measured by multiplex assays in lung homogenates, mean pg/ml ± SD, of mice in the double-hit model, n = 5 per group. Wild-type or ST2 KO mice were infected intratracheally with indicated AdVs and subsequently challenged with bleomycin or PBS as indicated. Asterisks indicate significantly (p < 0.05) elevated levels in MIL33-overexpressing mice compared to AdV-NULL-infected mice.

The ST2-independent effect of FLIL33 is likely not transcriptionally regulated

To begin addressing the potential mechanisms of the ST2-independent profibrotic action of FLIL33, experiments in cell culture were performed. Primary mouse lung fibroblasts, as well as unseparated lung cells, were infected in culture with AdV-FLIL33, AdV-MIL33, or AdV-NULL, and transcriptomic changes analyzed by RNA-Seq (Suppl. Fig. 5). The identities of the mRNAs affected by IL-33 overexpression are included in the Supplementary Dataset. The transcriptome of FLIL33-overexpressing fibroblasts was only modestly affected (Fig. 6A) compared to the pronounced changes induced by MIL33 overexpression (Fig. 6B), based on the number of the affected genes as well as the magnitude of changes in the gene expression levels. Pathway and process enrichment analyses revealed that FLIL33 upregulated genes involved in response to IFN-β, regulation of viral processes, and other pathways listed in Suppl. Fig. 6A. Among the top enriched pathways were collagen metabolic processes, including elevations in expression of genes for MMP3, MMP13, MMP10, CXCL1, ARG1, and several other genes. FLIL33 overexpression downregulated several genes whose lowered expression has been linked to lung fibrosis (Suppl. Fig. 6B), including elastin. Among the pathways strongly upregulated by MIL33 were those related to the inflammatory response, cytokine production, cell activation, and other “acute” processes (Suppl. Fig. 6C). Multiple genes related to lung fibrosis were also upregulated, including MMP2, MMP9, CCL2, CCL3, TIMP1, and several other genes. At the same time, overexpression of MIL33 in lung fibroblasts strongly downregulated the extracellular matrix organization- and extracellular matrix-receptor interaction-related genes (Suppl. Fig. 6D), such as TGFB2, ENG, FN1, TNXB, FZD2, FZD4, FZD8, FBN1, ELN, POSTN, LOX, LOXL1, LOXL4, MMP11, MMP28, COL5A1, COL6A2, COL8A1, COL12A1, COL14A1, and a number of other extracellular matrix-related genes. Thus, MIL33 regulated a large number of fibrosis-related genes, but such regulation was inconsistent, with both augmentation and attenuation of expression of various profibrotic genes. By contrast, FLIL33 overexpression elicited a milder yet consistent, profibrotic transcriptomic effect. The transcriptomic effects of both FLIL33 and MIL33 overexpression in fibroblasts were nearly completely abrogated by ST2 deletion (Fig. 6C, D).

Figure 6.

Pairwise comparisons of gene expression levels based on RNASeq transcriptomic profiling of primary mouse lung fibroblasts (A – D) or unseparated mouse lung cells (E – H) infected in culture with AdV-NULL, AdV-FLIL33, or AdV-MIL33 as indicated. Each dot represents a pair of expression levels of a gene. For each gene, the horizontal coordinate represents the expression level in AdV-NULL-infected cells derived from either wild-type or ST2 KO mice, whereas the vertical coordinates represent the expression level in AdV-FLIL33- infected (A, C, E, G) or AdV-MIL33-infected (B, D, F, H) cells. The colorized dots represent genes with substantially different gene expression levels. The genes with 5-fold or higher (red), 2- to 5-fold higher (green), 2- to 5-fold lower (blue), and 5-fold or lower (purple) expression levels are shown, with the similarly colored numbers representing the counts of the corresponding genes.

Similar to fibroblasts, FLIL33 overexpression in unseparated lung cells induced moderate changes in the expression of numerous genes (Fig. 6E, Suppl. Fig. 6E), whereas changes induced by MIL33 overexpression were substantially more pronounced (Fig. 6F, Suppl. Fig. 6F). Again, ST2 deletion nearly abrogated these changes induced by either FLIL33 (Fig. 6G) or MIL33 (Fig. 6H) overexpression. At least two important observations are revealed by these combined data. First, some profibrotic effects of FLIL33 are ST2-dependent and transcriptionally mediated. Second, the mechanism of the ST2-independent, potentiating effect of FLIL33 on bleomycin-induced collagen accumulation (Fig. 4E) is unlikely to be transcriptional.

Discussion

Our findings suggest that in addition to acting through the MIL33-driven, canonical IL-33 – ST2 axis, IL-33 may also contribute to development of fibrosis in its FLIL33 form both transcriptionally, in an ST2-dependent fashion, and non-transcriptionally, without engaging ST2. Several lines of evidence presented above and revealed by earlier reports support such a possibility.

Previous reports revealed that both IL-33 and ST2 control pulmonary fibrosis in the bleomycin injury model [3–5], implying the role for the canonical IL-33 – ST2 axis. However, several apparent inconsistencies suggested a possibility that not only can MIL-33 drive fibrosis by acting through ST2, but other IL-33-driven processes may additionally contribute. Previous reports revealed that genetic ST2 deletion attenuates bleomycin-induced lung fibrosis in C67BL/6 mice [4, 5] but the magnitude of the decrease in collagen levels appeared to be relatively modest compared to the effect of antibody-mediated IL-33 blockade [4]. In a less bleomycin-sensitive mouse strain, BALB/c, ST2 deletion did not appear to notably affect the accumulation of pulmonary collagen following bleomycin challenge. Here, we compared the effects of IL-33 deficiency and ST2 deletion in a more consistent fashion. The induced depletion of IL-33 in the acute bleomycin injury model (Fig. 1), as well as in both induced and constitutive depletion of IL-33 in the chronic models of bleomycin injury (Fig. 2), led to a notable decline in the levels of bleomycin challenge-induced collagen. Genetic deletion of ST2 attenuated the levels of collagen significantly, yet less profoundly (Fig. 3A) than did IL-33, suggesting that some effects of IL-33 may be, in part, ST2-independent in this model.

To address this possibility, replication-deficient recombinant adenovirus-mediated overexpression of FLIL33 or MIL33 in bleomycin-challenged wild-type and ST2 KO mice was performed (Fig. 4). We previously reported that overexpression of FLIL33 potentiated pulmonary fibrosis by amplifying the bleomycin injury-induced production of proinflammatory and profibrotic, but not Th2, cytokines [3]. Of note, such potentiation was neither accompanied by accumulation of eosinophils nor by elevation in Th2 cytokines in the lungs, suggesting that the profibrotic amplification by IL-33 was not due to FLIL33 maturation into MIL33. A similar potentiation of bleomycin-induced collagen accumulation was now observed in wild-type C57BL/6 mice, and, importantly and perhaps surprisingly, in ST2 KO C57BL/6 mice (Fig. 4E). This effect is unlikely to have been driven by induction of the Th2 phenotype, because overexpression of MIL33, but not FLIL33, activated pulmonary eosinophilia and goblet cell hyperplasia in the ST2-dependent fashion (Fig. 4B – D), without potentiating the bleomycin-induced accumulation of collagen (Fig. 4E). Others demonstrated by western blotting that a direct pulmonary bleomycin challenge induced a relative elevation in MIL33 and a decrease in FLIL33, suggestive of IL-33 maturation into the active cytokine form [4, 65], and reports of mild elevations in type 2 cytokines in the bleomycin model are abundant in the literature.

Nevertheless, the published data are contradictory, both suggesting a contributing role for Th2 cytokines and chemokines in the bleomycin model [66, 67] and negating it [68]. In our study, the levels of Th2 cytokines IL-4, IL-5, and IL-13 were, as expected, strongly elevated in MIL33-overexpressing wild-type, but not ST2 KO mice, and were not elevated in FLIL33-overexpressing mice, further arguing against physiologically impactful maturation of FLIL33 into MIL33 in this model (Fig. 5). Thus, FLIL33 contributes to bleomycin-induced lung fibrosis at least in part in an ST2-independent fashion. Additionally, these findings further argue against involvement of a major Th2 mechanism in pulmonary fibrosis, supporting the previous similar claim of others based on their work in a different model that Th2-dominant inflammation in the lungs is not essential for the development of bleomycin-induced pulmonary fibrosis [68].

We then asked whether the contributions of FLIL33 to lung fibrosis are entirely or partially ST2-independent and whether such contributions are transcriptionally regulated. Transcriptomic changes caused by FLIL33 or MIL33 overexpression in primary pulmonary mouse fibroblasts, as well as in unseparated lung cells, were evaluated using an RNA-Seq approach (Suppl. Fig. 5). While MIL33 overexpression elicited extensive transcriptomic changes in an ST2-dependent fashion, overexpression of FLIL33 had a limited, yet discernable effect on the transcriptome of ST2^+/+ cells from wild-type mice in terms of the magnitude of the changes and the number of affected genes (Fig. 6A, B, E, F). By contrast, genetic deletion of ST2 strongly attenuated the transcriptional effects of either FLIL33 or MIL33 overexpression on ST2^−/− cells (Fig. 6C, D, G, H). It is likely that FLIL33 potentiates the profibrotic effect of bleomycin in wild-type mice in a both ST2-dependent and -independent, transcriptional and posttranscriptional, fashion, whereas in ST2 KO mice, such an effect (Fig. 4E) is likely to be mostly post-transcriptional.

These combined observations, combined with previous literature cited above, jointly suggest an integrated, three-pronged mechanism through which IL-33 controls fibrosis. The first, canonical MIL33 – ST2 axis, which requires FLIL33-to-MIL33 maturation, induces pronounced transcriptomic changes, and typically induces a pronounced Th2 phenotype [24]. The second, noncanonical FLIL33 – ST2 axis, also induces transcriptomic regulation, which is distinct from that induced by MIL33. This notion is supported by 1) the lack of overt Th2 manifestations in the bleomycin model suggestive of a limited contribution from MIL33, 2) the differential capacity of FLIL33 and MIL33 to engage distinct ST2-dependent transcriptomic changes (Fig. 6), and 3) the preserved partial ST2-dependence of collagen accumulation (Fig. 3A). In the third pathway, FLIL33, but not MIL33, contributes to the profibrotic regulation in an ST2-independent (Fig. 4E), non-transcriptomic (Fig. 6), fashion, possibly through some of the previously suggested mechanisms in other systems [3, 24, 30–33], which remain to be considered in the setting of pulmonary fibrosis.

In addition to this three-pronged mechanism by which IL-33 may contribute to pulmonary fibrosis, there are other innovative aspects to our study. First, the effects of genetic depletion of IL-33 in the bleomycin model of lung fibrosis have not been, to our knowledge, previously studied. Second, we studied the effects of such deficiency not only in the acute, but also in the more clinically relevant chronic model. The findings in both models were similar, expanding the notion of generalizable nature of the profibrotic regulation by IL-33, as well as providing a preclinical tool for future development of innovative targeting of IL-33-dependent downstream profibrotic pathways. The contributions of IL-33 to pulmonary fibrosis beyond the two versions of the bleomycin model, e.g., silica- or radiation-induced lung fibrosis, as well as to fibrosis of other organs, remain to be further explored. Third, similar to the majority of other tamoxifen-induced, Cre-mediated transgenic models, the depletion of IL-33 was incomplete and somewhat variable, allowing for correlating the levels of remaining IL-33 expression with the levels of collagen accumulating in response to bleomycin (Figs. 1D, 2D). The rather strong link between the magnitude of IL-33 depletion and decline in collagen levels is indicative of the possibility of precisely controlled, as opposed to binary (all-or-nothing), therapeutic regulation of connective tissue homeostasis though manipulation of IL-33 levels.

This study raises several important mechanistic and pathophysiological questions that will have to be addressed in subsequent studies. The multibranched process through which IL-33 likely regulates lung fibrosis needs to be understood in greater molecular details. These include a) the previously acknowledged [7, 25–29], but not explored in fibrosis, noncanonical FLIL33 – ST2 axis; b) the suggested [3, 24, 30–33], but minimally studied in fibrosis, ST2-independent non-transcriptional regulation by FLIL33; and c) the nearly completely unexplored possibility of an ST2-driven, IL-33-independent mechanism [45]. For example, among these questions, the relative contributions of IL1RAP and MyD88 to the differential transcriptomic regulation by MIL33 – ST2 versus FLIL33 – ST2 pathways need to be explored.

In summary, the presented findings indicate that the regulation of fibrosis by IL-33 is multifaceted in nature, with the precursor and mature forms of IL-33 differentially driving the fibrotic process in both ST2-dependent and -independent, transcriptional and non-transcriptional, fashions. This notion implies that therapeutic targeting of both FLIL33 and MIL33 is likely to be more efficacious than pharmacological targeting of the MIL33 – ST2 axis only in patients with pulmonary fibrosis.

• Genetic depletion or deletion of IL-33, or depletion of ST2, inhibited lung fibrosis

• Full-length IL-33 enhanced the profibrotic effect of bleomycin independently of ST2

• Full-length IL-33 enhanced fibrosis without induction of the Th2 phenotype

• ST2 deletion abrogated the modest transcriptomic effect of full-length IL-33

• Full-length IL-33 contributes to fibrosis distinctly from the mature IL-33 cytokine