Regulation of Mesenchymal Cell Fate by Transfer of Active Gasdermin-D via Monocyte-derived Extracellular Vesicles

Abstract

Fibrosis is characterized by inappropriately persistent myofibroblast accumulation and excessive extracellular matrix deposition with the disruption of tissue architecture and organ dysfunction. Regulated death of reparative mesenchymal cells is critical for normal wound repair, but pro-fibrotic signaling promotes myofibroblast resistance to apoptotic stimuli. A complex interplay between immune cells and structural cells underlies lung fibrogenesis. However, there is a paucity of knowledge on how these cell populations interact to orchestrate physiologic and pathologic repair of the injured lung. In this context, gasdermin-D (GsdmD) is a cytoplasmic protein that is activated following cleavage by inflammatory caspases and induces regulated cell death by forming pores in cell membranes. This study was undertaken to evaluate the impact of human (Thp-1) monocyte-derived extracellular vesicles (EVs) and GsdmD on human lung fibroblast death. Our data show that active GsdmD delivered by monocyte-derived EVs induces caspase-independent fibroblast and myofibroblast death. This cell death was partly mediated by GsdmD-independent induction of cIAP-2 in the recipient fibroblast population. Our findings define a novel paradigm by which inflammatory monocytes may orchestrate the death of mesenchymal cells in physiologic wound healing, illustrating the potential to leverage this mechanism to eliminate mesenchymal cells and facilitate the resolution of fibrotic repair.

Introduction

Fibrosis in the lungs and other organs is the consequence of an aberrant repair response to tissue injury characterized by the inappropriate persistence of myofibroblasts and excessive accumulation of extracellular matrix which disrupts tissue architecture and ultimately leads to organ dysfunction. Worldwide, fibrotic diseases account for an estimated 45% of deaths in developed countries (1). Idiopathic pulmonary fibrosis (IPF) is a diffuse and typically progressive fibrotic lung disease that is associated with radiologic and pathologic features of usual interstitial pneumonia (UIP) without a recognized single precipitating cause (2–4). Despite the approval of pharmacologic treatments that decrease the rate at which fibrosis progresses, no therapies reverse established fibrosis or improve quality of life in patients, highlighting the ongoing need to improve our fundamental understanding of disease pathogenesis and the urgent need for novel therapeutic interventions (5–9). Accumulating evidence indicates the there is a complex interplay between immune cells and structural (epithelial and mesenchymal) cells that underlies lung fibrogenesis. While each cell type has been studied in isolation, there is a paucity of research focused on how these cell populations interact to orchestrate physiologic and pathologic outcomes in response to lung injury (10–12).

Secreted extracellular vesicles (EVs) function to regulate intercellular communication by delivering cargo such as proteins, mRNA and microRNA to recipient cells in close approximation or at a distance to the secretory cell. Growing evidence indicates that EVs play an important role in IPF and other fibrotic diseases (13,14). A deeper understanding of the specific roles of immune-cell derived EVs in fibrosis may facilitate the identification of novel disease mechanisms and therapeutic targets. We previously identified inflammasome proteins including active caspase-1 and active gasdermin-D (GsdmD) encapsulated in EVs released by LPS-stimulated monocytes, and showed that these inflammasome proteins have a role in monocyte/macrophage mediated immune responses and cell dysfunction (15–19). We have also shown elevated levels of active GsdmD encapsulated in circulating EVs in the plasma of patients with sepsis-induced acute respiratory distress syndrome (ARDS) (20).

The resolution phase of normal (non-fibrotic) wound repair is associated with the robust regulated death of reparative mesenchymal cells (fibroblasts and differentiated myofibroblasts), although the physiologic trigger(s) and cellular mechanisms that drive mesenchymal cell death have not been defined (2). In contrast to this normal homeostatic resolution, we and others have established that the persistence of activated mesenchymal cells in fibrotic repair is due, at least in part, to reduced apoptosis susceptibility attributed to activation of pro-fibrotic signaling pathways and upregulation of anti-apoptotic proteins (2, 21). In murine models of fibrosis involving the lung and other organs, inhibition of these signaling pathways or inhibition/knockdown of anti-apoptotic proteins can promote mesenchymal cell death and facilitate the resolution of lung fibrosis (21–27).

Recent studies describe a role of EVs produced by mesenchymal stem cells (MSCs) as key regulators of bleomycin-induced pulmonary fibrosis in mice and demonstrate that EVs in bronchoalveolar lavage fluid from patients with IPF can function as intercellular signaling mediators (14,28). However, the potential role of monocyte-derived EVs in the regulation of mesenchymal cell phenotypes in the context of lung injury, repair and fibrosis remains largely unexplored. The goal of this study was to test the hypothesis that monocyte-derived EVs with active inflammasome mediators would impact fibroblast survival. Our data show that active GsdmD delivered by monocyte-derived EVs induces fibroblast and myofibroblast death. Surprisingly, this cell death is mediated, in part, by the GsdmD-independent induction of cIAP-2, a member of the Inhibitor of Apoptosis family protein family, in the recipient fibroblast population. This novel paradigm of monocyte-mesenchymal cell interactions mediated by EVs enhances our understanding of homeostatic wound repair and identifies a mechanism that could be leveraged as a therapeutic approach to eliminate mesenchymal cells and facilitate the resolution of fibrotic repair.

Materials and Methods

Cells and reagents

Lipopolysaccharide (LPS), E.coli 0111:B4 was purchased from InvivoGen (San Diego, CA). RPMI 1640 was purchased from Mediatech Inc. (Manassas,VA), phosphate buffered saline (PBS) from Life Technologies (Grand Island, NY) and fetal bovine serum (FBS) from R&D Systems (Minneapolis, MN). The pan-caspase inhibitor, z-Val-Ala-Asp (O-Methyl) fluoromethyl ketone (zVADfmk) and acetyl–tyrosyl-valyl-alanyl-aspartyl–chloromethylketone (YVAD-cmk) were purchased from EMD Biosciences (San Diego,CA) and Millipore Sigma (Burlington, MA) respectively. Mouse anti-GsdmD was obtained from Abnova (Taipei, Taiwan), cIAP-2 antibody from Sigma (St. Louis, MO) and Hsp90 from Cell Signaling (Danvers, MA). AT406 was from Selleck Chemicals, Houston, TX. All other reagents were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise specified. In all experiments we used Thp1 cells (American Type Culture Collection, Manassas, VA). Cas9 and CRISPR/Cas9 GsdmD knockout (KO) cells were generated in Thp1 monocytes by Dr. Seth Masters, WEHI, Australia. (19).

The IMR90 line of primary normal human fetal lung fibroblasts was obtained from ATCC (Coriell Institute for Medical Research, Catalog # I90–10). Control and cIAP-2ko IMR90 fibroblasts were generated using standard CRISPR technology (45,46). Briefly, oligonucleotides encoding either guide RNA (CACCGGTTCCGCGTTACATAACTTA; AAACTAAGTTATGTAACGCGGAACC) or cIAP-2 (CACCGAATGGGCTTTAGTAGAAGCC; AAAcGGCTTCTACTAAAGCCCATTC) were cloned into the plasmid lenti-CRISPRv2-mNEON. The vector was generated by replacing the puromycin ORF in lenti-CRISPRv2-puro with Neon coding region. Lentiviruses were generated using standard methods and IMR90 cells were transduced (29,30). Cells were selected for GFP fluorescence after 4 days. Fibroblasts deficient in endogenous GsdmD were generated using siRNA transfected with Lipofectamine 2000 according to manufacturer’s conditions using control non-targeting (D-001810-10-05) or pooled GsdmD specific siRNA (Dharmacon SMARTpool: L-016207-00-0005; final concentration 20 nM). Briefly, IMR90 fibroblasts were seeded in DMEM containing 5% serum overnight and transfected with 20 nM siRNA (control or GSDMD). Medium was changed 6 hours and 48 hours post-transfection and cells were then transferred to 48-well plates to complete experiments.

Primary normal and IPF lung fibroblasts.

Primary normal and IPF lung fibroblasts were isolated and cultured under The Ohio State University Institutional Review Board-approved protocols 2017H0309 and 2021H0180 as previously described (31). There were no identifiers, other than clinical diagnosis of “IPF” or “normal” associated with the specimens. Normal fibroblasts were isolated from the normal lungs of decedents with no evidence of pulmonary disease who were HIV, hepatitis C, and SARS-CoV-2 negative and had no evidence of significant trauma. Written consent was obtained from the families. IPF lung fibroblasts were collected from explanted lungs with minimum ischemia time. Once lung samples were received, the visceral pleura was dissected and removed. Parenchyma was then washed with PBS to eliminate RBCs. Minced tissue was transferred to GentleMACs C-Tubes containing the enzymatic cocktail (Elastase 10 U/ml, Collagenase IV 450 U/ml, Dispase 2 U/ml, and DNAse I 100 ug/ml) and incubated at 37°C. After 1-hour, cold serum was added to reduce the enzymatic activity. The filtered lung cell suspension was resuspended in RBC lysis solution and incubated for 7 min. The lung cell suspension was plated in a 75cm flask at a density of 5,0×10⁶ to 1,0×10⁷ cells and incubated for 24 hours to allow them to adhere to the plate. Non-adherent cells were removed, and fresh media was added every two days.

Cell culture.

Human Cas9 Thp1 monocytes were cultured in RPMI 1640 media supplemented with 10% heat inactivated fetal bovine serum (FBS) and antibiotics at 37°C in a humidified CO₂ incubator. All fibroblasts were maintained in DMEM containing 2mM glutamine and 10% fetal bovine serum. For all experiments, 30,000 fibroblasts/well were plated on 48 well plates in DMEM containing 2 mM glutamine and 10% fetal bovine serum. Medium was removed after 4 hours and replaced with serum-free DMEM containing 2 mM glutamine which was replaced with fresh serum-free DMEM with 2 mM glutamine after 16 hours.

Extracellular vesicle (EV) isolation and quantification:

EV isolation was performed following our published procedure (17). Briefly, monocytes were stimulated with either LPS (1μg/ml) or not for 2h. Supernatants were collected and centrifuged at 2000 g for 5 min and 16,000 g for 5 min to remove floating cells, cell debris and apoptotic bodies. The supernatants were further centrifuged at 100,000 g for 1h to isolate EVs. EVs isolated from Thp1 cells treated with LPS (LPS EV, abbreviated as LMV) or left untreated (control EV, abbreviated as CMV) were subjected to quantification analysis for normalization purposes. First, total proteins were measured from the EVs. Equal proteins from EVs were then also subjected to quantification using the NanoSight technology following the company manual. The Malvern NanoSight range of instruments utilizes Nanoparticle Tracking Analysis (NTA) to characterize nanoparticles from 10nm-2000nm in solution. Each EV particle is individually but simultaneously analyzed by direct observation and measurement of diffusion events. Based on our analysis, control and LPS EVs, when normalized using protein quantification, showed similar number of particles. Based on this observation, all EVs were normalized for equal loading and experimental designs based on protein normalization.

Cell death assays:

Cell death was quantified by lactate dehydrogenase (LDH) release in cell culture medium using NAD⁺ reduction assay as per manufacturer’s protocol (Roche Applied Science) and Annexin/PI assay following manufacturer’s protocol (BD Pharmingen). IMR90 fibroblasts plated in a 48-well plate at the density 30,000 cells/well were stimulated with EVs isolated from Thp1, Cas9 or Cas9 GsdmD knockout (KO) monocytes that were untreated (CMVs) or LPS treated (LMVs) for 18 hrs. Cell culture medium was collected, clarified by centrifugation at 300 x g for 10 min, and used for detection of LDH. Cell culture medium alone was used as a blank. LDH concentration in the medium was detected at a wavelength 490 nm and data represented as the measured absorbance. Recipient fibroblasts were separated from the culture medium and stained with annexin/propidium iodide. For a positive control, fibroblasts were treated with 1 μM Staurosporine for 24 hours. Samples were analyzed by BD FACSymphony^™ A1 Cell Analyzer (BD Biosciences, San Jose, CA) and FlowJo 10 software (FlowJo, LLC, Ashland, OR).

Immunoblotting:

Culture medium from Thp1 monocytes or fibroblasts was removed from cells and subjected to differential centrifugation to collect cell pellets. Thp1 supernatants were also subjected to sequential centrifugation for EV isolation. Cell and EV pellets were lysed in lysis buffer (50mM Tris-HCl pH8.0, 125mM NaCl, 10mM EDTA, 10mM sodium fluoride, 10mM sodium pyrophosphate and 1% Triton X-100 containing protease inhibitor cocktail from Sigma and 50μM N-methoxysuccinyl–Ala-Ala-Pro-Val chloromethylketone). Fibroblasts were washed with PBS and lysed in SDS buffer (62.5mM Tris pH 6.8 and 2%SDS) supplemented with 1 mM PMSF, Protease Inhibitor Cocktail from Sigma and 100 μM neutrophil elastase inhibitor N-methoxysuccinyl-ala-ala-pro-val-chloromethyl ketone (Sigma-Aldrich, St. Louis, MO). The protein concentration in the cell lysates and EVs was determined using Dc Lowry protein assay reagent from BioRad. Equal amounts of total protein were resolved by SDS-PAGE. Transferred membrane was then blocked with 10% nonfat dry milk in TBST (25mM Tris-HCl pH 7.5, 150mM NaCl, .1% Tween 20) for 2h at room temperature. The membranes were then probed with primary antibodies as indicated followed by peroxidase conjugated secondary antibodies. Protein bands were visualized by enhanced chemiluminescence (ECL, GE Healthcare). Presence of active GsdmD and cIAP-2 in EVs and fibroblasts was confirmed by using cIAP-2 antibody from Sigma (St. Louis, MO) and Gasdermin from Abnova (Taipei, Taiwan). Secondary antibodies were either donkey anti-rabbit polyclonal antibody or sheep anti-mouse monoclonal antibody (GE Healthcare), both conjugated to HRP (for ECL) respectively. Figures show representative Western blots from at least 3 experiments, and densitometry reflecting the pooled analysis of experimental replicates are included with the Supplement Figure 3 and Supplement Figure 4.

RNA isolation and gene expression analysis:

For RNA isolation, cells were lysed in presence of Trizol and RNA using the Zymoresearch RNA isolation kit according to manufacturer’s recommendations. RNA was dissolved in 8 μl per well per sample and reverse transcribed using Superscript III cDNA synthesis kit. qPCR was carried out using Power SYBR Green from Applied Biosystems in a BioRad CFX384 thermocycler. All samples were normalized to GAPDH and gene expression was quantified using the delta delta Ct method. Primers for the genes tested were generated using following sequences:

• XIAP: for GCTTGCAAGAGCTGGATTTT; rev TTGTTCCCAAGGGTCTTCAC

• cIAP-1: for AGCTTGCAAGTGCTGGTTTT; rev CTCCAGATTCCCAACACCTC

• cIAP-2: for CCAAGTGGTTTCCAAGGTGT; rev TGGGCTGTCTGATGTGGATA

• GAPDH: for GACAAGCTTCCCGTTCTCAG; rev GATCATCAGCAATGCCTCCT

Statistical analysis.

Data are represented as the mean ± standard error of the mean (SEM) from at least three independent experimental replicates. Statistical analysis was completed using ordinary one-way ANOVA with Tukey’s multiple comparison test (Graphpad Prism v9.4). p<0.05 was considered to represent statistical significance. * indicates p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.

Results

Extracellular Vesicles Isolated from LPS-stimulated monocytes induce caspase-independent fibroblast and myofibroblast death

To examine how inflammatory monocyte interactions with mesenchymal cells impact mesenchymal cell behavior, Thp1 monocytes were stimulated with LPS (1 μg/ml) for 2 hours to induce inflammasome activation and generation of active GsdmD. Extracellular vesicles (EVs) from both control (CMV) and LPS stimulated (LMV) Thp1 cells were isolated by differential centrifugation. Recipient mesenchymal cells were either quiescent or differentiated to myofibroblasts with TGF-β1 (2 ng/ml for 24 hours) prior to exposure to EVs at different doses, and fibroblast/myofibroblast cell death was assessed at 24 hours using an LDH-release assay. Untreated fibroblasts and TGF-β differentiated myofibroblasts had no measurable LDH release at baseline (Figure 1A) or in response to direct treatment with LPS (Figure 1B). Exposure of fibroblasts or myofibroblasts to control extracellular vesicles (CMV) induced low levels of LDH release, and there was no statistically significant difference between exposure to 12 μg, 24 μg (not shown) or 60 μg doses. In contrast to CMV, exposure to EVs isolated from LPS-stimulated monocytes (LMV) led to significant increases in cell death at 24 μg and further increased death at 60 μg (Figure 1A). Notably, susceptibility to cell death induced by LMV was no different in TGF-β differentiated myofibroblasts than the quiescent fibroblasts. Based on these data, subsequent experiments were conducted using a dose of extracellular vesicles of 24 μg.

Figure 1. LPS-stimulated monocyte derived extracellular vesicles (EVs) induce caspase-independent fibroblast and myofibroblast death.

A) Thp1 monocytes were stimulated with LPS (1 μg/ml) for 2 hours and EVs were isolated from supernatants by sequential centrifugation. EVs released from monocytes stimulated with or without LPS were incubated in a dose dependent manner (12, 24 and 60 μg) with IMR90 human primary pulmonary fibroblasts or fibroblasts differentiated with TGF-β1 (2 ng/ml) for 24 hours (myofibroblasts). Fibroblast cell death was analyzed by LDH assay and represented as % positive control (triton X). B) EVs were isolated from LPS stimulated monocytes in the presence or absence of a pan-caspase inhibitor, ZVAD (50 μM). Concurrently, fibroblasts were pre-treated with caspase inhibitors, ZVAD (50 μM) or VX765 (40 μM) prior treatment with LPS stimulated monocyte derived EVs. N=4 experiments.

To examine whether the cell death induced by LMVs was dependent on caspase activation in the recipient cells, fibroblasts were pre-treated with a pan-caspase inhibitor (Z-VAD-FMK), a caspase-1/11 inhibitor (VX 765) or a caspase-1 inhibitor (YVAD-cmk) and then exposed to CMV or LMV (24 μg) for 24 hours (Figure 1B and Supplement Figure 1A). Neither the pan-caspase inhibitor nor either of the caspase-1 inhibitors prevented LMV-induced fibroblast cell death, suggesting that death mediated by the LMVs bypassed the caspase-dependent apoptotic machinery in fibroblasts (Figure 1B, Supplement Figure 1A). However, if monocytes were treated with the caspase inhibitor, ZVAD or YVAD prior to LPS exposure and isolation of the EVs, the subsequent LDH release from fibroblasts was substantially diminished to levels similar to fibroblasts treated with CMV (Figure 1B, Supplement Figure 1A). As GsdmD activation in monocytes is caspase-1 dependent and prevented by treatment with the pan-caspase inhibitor (19), this finding suggested that active GsdmD in the EVs generated by LPS-treated monocytes may be transferred to the recipient fibroblasts and directly contribute to fibroblast cell death that is independent of caspase activation within the fibroblasts themselves.

Active Gasdermin-D (GsdmD) in monocyte-derived extracellular vesicles is necessary for induction of fibroblast death

Gasdermin-D is activated through caspase-1 dependent cleavage, and active GsdmD promotes cell death by insertion of “pores” within cell membranes. Given the caspase-independent mechanism of fibroblast cell death induced by LMV, we speculated that active GsdmD might be transferred from the LPS-stimulated monocytes to recipient fibroblasts via extracellular vesicles. To explore this possibility, we assessed the presence of active GsdmD in cell extracts and microvesicles from control monocytes (Cas9) and monocytes with stable knockout of GsdmD (Cas9-GsdmD^−/−) that were stimulated with/without LPS. Figure 2A shows that there is minimal detectable active GsdmD in the unstimulated monocyte cell extracts or EVs and that active GsdmD is enriched in EVs following LPS stimulation. As expected, there is no active GsdmD identified in the cell extracts or EVs from GsdmD knockout monocytes, regardless of LPS stimulation. EVs from unstimulated control or unstimulated GsdmD-deficient monocytes failed to induce fibroblast cell death, but robust cell death (LDH release and Annexin/PI positive cells) was observed in fibroblasts treated with LMV derived from control monocytes (Figure 2B, C). Consistent with a critical role for active GsdmD, fibroblasts exposed to LMV isolated from the GsdmD-deficient monocytes had significantly reduced death compared to fibroblasts exposed to EVs containing active GsdmD (Figure 2B, 2C and Supplement Figure 1). Pre-treatment of fibroblasts with TGF-β1 had no impact on fibroblast death (Supplement Figure 2A).

Figure 2. Active Gasdermin-D (GsdmD) in monocyte-derived extracellular vesicles is necessary for induction of fibroblast death.

A) EVs released from Cas9 (controls) or Cas9/GsdmD^−/− Thp1 monocytes stimulated with or without LPS were confirmed for presence or absence of active GsdmD (p30). Active GsdmD was only detected in Cas9 LPS EVs. Hsp90 was used as loading control for whole cell extracts (CE) and for EVs. CD63 and Tsg101 were used for markers of monocyte derived EVs. B) IMR90 fibroblasts were incubated with LPS EVs (24 μg) from Cas9 and Cas9/GsdmD^−/− monocytes for 24 hours and analyzed for cell death by LDH assays. C) IMR90 fibroblasts were incubated with LPS EVs (24 μg) from Cas9 and Cas9/GsdmD^−/− monocytes for 24 hours and cell death was analyzed using flow cytometry for Annexin V/ Propidium iodine (PI). Total fibroblast cell death was designated as combination of Annexin V(–)/PI(+), Annexin V(+)/PI(+), Annexin V(+)/PI(–) and viable cells as Annexin V(–)/PI(–). N=3 experiments, flow cytometry data is provided as Supplement Figure 1B. D) and E) Active GsdmD transfer from normal (D) or GsdmD-deficient (E) monocytes treated with/without LPS to recipient fibroblasts via EVs was analyzed by immunoblot. Hsp90 was used as loading control. N=4 experiments. F) Fibroblasts were transfected with either control siRNA (si control) or siRNA targeting GsdmD (si GSDMD) and treated with or without LPS (1μg/ml), with EVs isolated from normal monocytes treated with or without LPS (24 μg; CMV or LMV) or with EVs isolated from Cas9/GsdmD^−/− monocytes (LMV GsdmD ko) for 24h and analyzed for cell death by LDH assays. N=3 experiments. G) Active GsdmD transfer in EVs from monocytes to recipient control (WT; control siRNA transfected) and GsdmD deficient (GsdmD kd; transfected with siRNA targeting GsdmD) fibroblasts was analyzed by immunoblot. Hsp90 was used as loading control. N=3 experiments. H) Cas9 (control monocytes) were treated with disulfiram (DS; 10 μM) or DMSO vehicle for 1 hour prior to LPS exposure for 2 hours to functionally inhibit the pore-forming ability of active GdsmD. DS or DMSO exposed CMV (24 μg) and LMVs (24 μg) were then used to treat fibroblasts for 24 hours. Fibroblast death was analyzed by LDH assay. N=3 experiments.

To further establish whether this LPS EV-induced fibroblast death was due to the transfer of active GsdmD from monocytes to fibroblasts via the extracellular vesicles, we examined GsdmD expression in fibroblasts (Figure 2D). At baseline, normal fibroblasts did express intact (uncleaved; p52) GsdmD, but no active GsdmD (p30) was detectable. Neither direct LPS nor TGF-β1 treatment led to GsdmD activation in the fibroblasts. Following treatment with LMV, but not CMV, we observed a significant increase in active GsdmD within the fibroblasts that was not impacted by pre-treatment of the fibroblasts with TGF-β1 (Figure 2D). Finally, we similarly assessed GsdmD expression in fibroblasts treated with extracellular vesicles derived from control or GsdmD-deficient monocytes (Figure 2E). Consistently, active GsdmD was only present in fibroblasts treated with control LMV and no active GsdmD was identified in fibroblasts treated with LPS, CMV or GsdmD-deficient LMV.

To confirm that monocyte-derived active GsdmD transfer via EVs mediated fibroblast death, we generated normal lung fibroblasts in which endogenous GsdmD expression was knocked-down using siRNA (compared to scrambled control siRNA; Supplement Figure 2B. Fibroblasts deficient in GsdmD and controls were then analyzed for cell death following exposure to either CMV or LMV from Cas9 or Cas9-GsdmD^−/− Thp1 monocytes (Figure 2F). As expected, CMV treatment had no effect on control fibroblasts (siControl) or fibroblasts deficient in endogenous GsdmD (siGsdmD). In contrast, exposure to LMV from control monocytes (LMV) significantly induced fibroblast death in both control fibroblasts and GsdmD-deficient fibroblasts. Fibroblast cell death was significantly abrogated only when the cells (control or GsdmD-deficient) were treated with LMV from monocytes with stable knockout of GsdmD (LMV-GsdmD-ko) (Figure 2F). Uptake of exogenous active GsdmD from monocyte derived EVs was also analyzed in the recipient fibroblasts. Active GsdmD was detected in both normal control fibroblasts and GsdmD deficient fibroblasts when subjected to LMV from Cas9 monocytes. No active GsdmD was detected in fibroblasts when exposed to LMV from Cas9-GsdmD^−/− monocytes (GsdmD^−/− LMV), CMV or direct stimulation with LPS (Figure 2G).

Using an alternative approach, control monocytes (Cas9) were treated with disulfiram, which inhibits the pore-forming function of GsdmD without impacting its cleavage to the active form. Fibroblasts treated with EVs from the Disulfiram-exposed (Cas9 LMV DS), but not vehicle-exposed (Cas9 LMV DMSO) monocytes had significantly reduced cell death. In a finding consistent with a key role of monocyte-derived GsdmD transferred via EVs and the dispensable role of endogenous fibroblast GsdmD, treatment of recipient fibroblasts with Disulfiram prior to LMV exposure had no impact on cell death induced by the LMV (Supplement Figure 2C). Taken together, these data demonstrate that functionally active GsdmD is transferred from LPS-stimulated monocytes to recipient fibroblasts and myofibroblasts, and that the transfer of functionally active GsdmD is necessary for induction of mesenchymal cell death by these inflammatory extracellular vesicles.

cIAP-2 induction in fibroblasts is stimulated by extracellular vesicles from LPS-treated monocytes through a GsdmD-independent mechanism

Having previously shown that IAP family proteins regulate fibroblast susceptibility to apoptosis, we next sought to determine if IAPs have a role in the regulation of EV mediated fibroblast cell death by measuring the effect of monocyte-derived EVs on a subgroup of structurally homologous cellular IAPs (XIAP; BIRC4, cIAP-1; BIRC2 and cIAP-2; BIRC3). As with our published studies, treatment of normal fibroblasts with TGF-β1 increased expression of XIAP (Supplement Figure 2D). However, exposure to LMV had no significant impact on XIAP expression compared to untreated control fibroblasts, although LMV treatment did antagonize XIAP induction by TGF-β. Neither TGF-β nor LMV impacted cIAP-1 expression (Supplement Figure 2E). While TGF-β had no statistically significant effect on cIAP-2 compared to baseline controls, we were surprised to find that LMV exposure led to a substantial (> 20-fold) increase in cIAP-2 mRNA (Figure 3A, left). At the protein level, cIAP-2 expression in normal IMR90 fibroblasts was not detected under baseline conditions, but induction was seen following exposure to LMV (Figure 3A, right).

Figure 3. cIAP-2 induction in fibroblasts by monocyte derived EVs is GsdmD independent.

IMR90 normal lung fibroblasts were treated with/without TGF-β1 (2 ng/ml) for 24 hours and then exposed to extracellular vesicles derived from untreated Thp monocytes (Control EV, 24 μg) or EVs isolated from Thp1 monocytes treated for 2 hours with LPS (LPS EV, 24 μg) for 24 hours. cIAP-2 transcripts were assessed by quantitative real-time RT-PCR and expressed as “fold-change” from baseline controls (represented as 1.0). A) cIAP-2 gene and protein expression. N=3 experiments. B and C) cIAP-2 induction in fibroblasts treated with EVs from Cas9 and Cas9/GsdmD^−/− monocytes (treated +/− LPS) was analyzed by gene expression (B) and protein (C). D) cIAP-2 induction in fibroblasts transfected with control (si control) or GsdmD-targeting siRNA (si GsdmD) by LPS EVs from Cas9 and Cas9/GsdmD^−/− monocytes was analyzed by immunoblot. Data is representative of N=3 experiments.

Beyond their ability to function as apoptosis inhibitors, IAP family proteins have been shown to impact a variety of cellular activities that are not directly related to apoptosis, including the regulation of inflammasome activation (32–35). Given the robust induction of cIAP-2 in fibroblasts treated with LMV and our finding that the transport of active GsdmD as cargo in the LMV was critical for induction of fibroblast death, we asked whether GsdmD transfer was necessary for the observed increase in cIAP-2. Using LMV from control or GsdmD-deficient monocytes, we found that the absence of GsdmD in the monocytes did not alter the induction of cIAP-2 in recipient fibroblasts at gene or protein levels (Figure 3B, C). We further analyzed expression of cIAP-2 induction in normal fibroblasts and fibroblasts deficient in endogenous GsdmD. As seen in Figure 3D, cIAP-2 expression was detected in both normal (WT) and GsdmD deficient (GsdmD kd) fibroblasts in the presence of LMV from wild type monocytes (Ca9 LMV) or LMV from monocytes lacking GsdmD (GasdmD^−/−). This experiment confirmed that cIAP-2 protein induction required exposure to LMV and was independent of GsdmD in either the monocytes generating the LMV or in the recipient fibroblasts exposed to the LMV (Figure 3D). Collectively, these experiments show that while active GsdmD transferred from inflammatory monocytes to fibroblasts through extracellular vesicles is essential for fibroblast death, it coincides with the GsdmD-independent induction of cIAP-2.

cIAP-2 induction in fibroblasts contributes to LMV-induced cell death

The induction of cIAP-2 in normal lung fibroblasts that were undergoing cell death suggested that either cIAP-2 induction was a response meant to mitigate death or that, despite its name, cIAP-2 had a contributory role in fibroblast death following inflammatory EV exposure. To distinguish between these possibilities (and the possibility that cIAP-2 induction had no positive or negative effect on fibroblast survival/death), we generated normal lung fibroblasts deficient in cIAP-2 (Cas9-cIAP-2 KO). cIAP-2 knockout Cas9/IMR90 cells were validated for cIAP-2 expression by assessment of RNA by quantitative real-time RT-PCR (Figure 4A) and then analyzed for cell death following exposure to either CMV or LMV. As expected, CMV treatment had no effect on normal fibroblasts or Cas9 control fibroblasts, while exposure to LMV significantly induced fibroblast death. However, cIAP-2 deficiency was associated with a partial, but significant, reduction in LMV-induced cell death (Figure 4B). To confirm the contributory role of cIAP-2 in LMV-induced death, we treated normal IMR90 fibroblasts with AT406, a SMAC mimetic that inhibits cellular IAP protein function and promotes IAP protein degradation, prior to LMV exposure and noted a similar reduction in fibroblast death (Supplement Figure 2F). Assessments of cell lysates consistently failed to show cIAP-2 expression in Cas9 control fibroblasts at baseline and induction of cIAP-2 protein expression following LMV treatment. Consistent with effective knockdown, LMV failed to induce cIAP-2 protein expression in the Cas9-cIAP-2^−/− fibroblasts (Figure 4C). Included in Figure 4C as positive controls for cIAP-2 expression are normal human (h) IMR90 fibroblasts treated with LMV, and murine(m) fibroblasts which we found have increased levels of detectable cIAP-2 protein at baseline. Having previously shown that active endogenous GsdmD is not present in fibroblasts (Figure 2), we additionally found that transfer of active GsdmD from LMV to fibroblasts was independent of cIAP-2 expression in the fibroblasts, as there was a robust increase in active GsdmD observed following treatment of cIAP-2 deficient fibroblasts with LMV (Figure 4D). Next, we again assessed the role of caspases and found that neither pan-caspase inhibition nor caspase-1 inhibition in fibroblasts altered the induction of cell death in LMV-treated fibroblasts (Figure 4E). Moreover, treatment of fibroblasts with these caspase inhibitors did not alter the partial protection from death that was observed in fibroblasts lacking cIAP-2 (Figure 4E). Finally, inhibition of these caspases in the recipient fibroblasts lacking cIAP-2 had no effect on cIAP-2 induction in the control (Cas9) fibroblasts. As expected, there was no cIAP-2 induction in the cIAP-2 knockout fibroblasts. Moreover, active GsdmD transfer by LMVs was observed in both control Cas9 and fibroblasts lacking cIAP-2 (Figure 4F). Taken together, these studies demonstrate that LMV-induction of fibroblast cell death is GsdmD-dependent, although the GsdmD-independent induction of cIAP-2 has a contributory role in fibroblast death induced by LMV.

Figure 4. cIAP-2 induction in fibroblasts contributes to LMV-induced cell death.

A) IMR90 cells with CAS9 or CAS9-cIAP-2^−/− were treated with/without LMV and analyzed for cIAP-2 expression (fold change) using RT-PCR to confirm generation of cIAP-2 knockout (ko) fibroblasts. B) IMR90 normal lung fibroblasts, Cas9 IMR90 and Cas9/cIAP-2^−/− IMR90 myofibroblasts were subjected to EVs released from monocytes stimulated with or without LPS, incubated for 24 hours and analyzed for LDH release as measure of cell death. C) and D) Induction of cIAP-2 in fibroblasts and transfer of extracellular active GsdmD from monocyte derived LPS EVs were measured using immunoblot. E) and F) IMR90 normal lung fibroblasts, Cas9 IMR90 and Cas9/cIAP-2^−/− IMR90 myofibroblasts pretreated with pan-caspase inhibitors, ZVAD and VX765 were subjected to EVs released from monocytes stimulated with or without LPS and incubated for 24 hours. Cell death was analyzed by LDH assays. cIAP-2 expression and presence/transfer of active p30 GsdmD in the fibroblasts by EVs were analyzed by immunoblot. N=3 experiments

Fibrotic lung fibroblasts are not resistant to cell death induced by monocyte-derived active GsdmD

Acquisition of an apoptosis-resistant fibroblast phenotype is an established feature of fibrotic disease in the lung (36). To determine whether resistance to regulated cell death is circumvented by the transfer of active GsdmD from inflammatory monocytes, we compared cell death in primary lung fibroblasts isolated from patients with normal or IPF lungs (Figure 5A). As with the IMR90 normal lung fibroblast cell line, exposure to CMV failed to induce cell death in patient-derived normal lung fibroblasts while LMV induced significant cell death. Importantly, the IPF lung fibroblasts showed no resistance to death induced by LMV exposure. Consistent with the findings in the normal lung fibroblasts, cIAP-2 induction was seen in the normal patient-derived and IPF fibroblasts exposed to LMV, and active GsdmD was seen only in the fibroblasts treated with LMV (Figure 5B). We then assessed whether this LMV mediated death in the patient-derived normal and IPF lung fibroblasts was similarly dependent on active GsdmD within encapsulated EVs. Figure 6A confirms significantly reduced death in normal and IFP fibroblasts exposed to EVs generated from GsdmD deficient monocytes. cIAP-2 protein levels in control and IPF fibroblasts were observed following LMV exposures, and this was not dependent on GsdmD in the EVs (Figure 6B). To further evaluate the role of cIAP-2 in the LMV-mediated cell death, normal and IPF fibroblasts were treated with AT406 prior to LMV from control and GsdmD deficient monocytes. Consistent with prior reports, AT406 treatment led to a significant reduction of cIAP-2 expression (Figure 6C, upper blot). Notably, however, AT406 did not impact transfer of extracellular active GsdmD although quantitatively there was a decrease in active GsdmD seen in the AT406-treated cells (Figure 6C, lower blot). As we observed in the IMR90 fibroblasts, treatment of normal and IPF fibroblasts with AT406 significantly reduced cell death by LMV (Figure 6D). These findings confirm that LMV-mediated fibrotic lung fibroblast cell death is GsdmD-dependent and that GsdmD-independent cIAP-2 induction has a contributory role in this pathway in primary human lung fibroblasts and that IPF fibroblasts are equally susceptible to cell death induced by transfer of active GsdmD from monocytes via extracellular vesicles.

Figure 5. Fibrotic lung fibroblasts are not resistant to cell death induced by EVs from LPS-stimulated monocytes.

Primary normal and IPF lung fibroblasts (n = 3 of each) were collected from explanted lungs and the lung cell suspension was plated at a density of 5.0 × 10⁶ to 1.0 × 10⁷ cells and incubated for 24 hours to allow adherence. Non-adherent cells were removed, and fresh media was added every two days. A) Primary normal and IPF lung fibroblasts were then subjected to EVs from LPS treated monocytes and analyzed for cell death by LDH assays. B) cIAP-2 induction and transfer of active GsdmD by monocyte EVs was analyzed by immunoblot.

Figure 6. Monocyte derived EV GsdmD is essential for fibrotic lung fibroblast cell death.

Primary normal and IPF lung fibroblasts (N = 3 each) were subjected to LPS EVs from Cas9 and Cas9/GsdmD^−/− Thp1 monocytes for 24 hours. A) Cell death was analyzed by LDH; B) cIAP-2 induction in fibroblasts by EVs and transfer of GsdmD was analyzed by immunoblot. C) and D) Primary normal and IPF lung fibroblasts were subjected to LPS EVs from Cas9 and Cas9/GsdmD^−/− Thp1 monocytes for 24 hours in the presence or absence of Smac mimetic AT406 (1.0 μM). C) cIAP-2 expression (upper panel) and transfer of extracellular EV active GsdmD (bottom panel) in normal and IPF fibroblasts were analyzed by immunoblot. Hsp90 was used as loading control. D) Effect of AT406 on EV-GsdmD mediated fibroblast cell death was analyzed by LDH assay in normal and patient fibroblasts. N=3 experiments.

DISCUSSION:

Fibroblasts (and activated myofibroblasts) are essential contributors to the normal wound repair response and the resolution of homeostatic repair coincides with a reduction in myofibroblast numbers due to regulated cell death and, possibly, dedifferentiation to a quiescent state (36–38). In contrast to this homeostatic resolution of repair with restoration of normal tissue structure and function, fibrotic repair is characterized by the persistence of activated myofibroblasts which resist and/or evade signals of regulated cell death and perpetuate aberrant wound repair (39). While we, and others, have defined multiple mechanisms that allow these mesenchymal cells to resist death, the physiologic triggers for mesenchymal cell death in the context of normal repair have not been previously identified. We have presented a novel paradigm in which inflammatory cell-mesenchymal cell communications mediated via extracellular vesicles can orchestrate mesenchymal cell death. Mechanistically, we show that monocytes stimulated by LPS secrete EVs containing active GsdmD. We show that transfer of this active GsdmD into recipient mesenchymal cells is essential for the subsequent death of the mesenchymal cells. Moreover, caspase-activation within the recipient cells is not required for the mesenchymal cell death, indicating that this mechanism circumvents many of the anti-apoptotic mechanisms induced by TGF-β1 that have been implicated in the resistance of fibrotic lung fibroblasts to intrinsic and extrinsic apoptotic stimuli (40,41). Additionally, we show secreted EVs induce a spike in cIAP-2 within the recipient mesenchymal cells in a GsdmD independent manner, and that this induction of cIAP-2 paradoxically contributes to the extent of mesenchymal cell death. Notably, differentiated myofibroblasts that have been shown to resist cell death in a variety of contexts are equally susceptible to death on exposure to EVs containing active GsdmD. Taken together, these findings define a novel mechanism by which inflammatory monocytes may orchestrate the death of mesenchymal cells in physiologic wound healing and illustrate the potential to leverage this mechanism to overcome mesenchymal cell resistance to cell death as a therapeutic approach in fibrotic disease.

Extracellular vesicles (EVs) released by activated or apoptotic cells are known to carry various factors including proteins, microRNAs and lipid mediators which can be transferred to other cells at a distance (42,43). Our group and others have reported that EVs regulate vascular function, cellular responses and survival in different diseases (14–16,44–49). Specifically, we have shown that monocyte derived circulating EVs regulate vascular cell dysfunction and death in sepsis and sepsis-induced acute respiratory distress syndrome (ARDS) (17,18,44). We have also shown that monocytes activated by LPS release exosomes encapsulating active inflammasome proteins, including active GsdmD, resulting in vascular endothelial cell injury (17–20,44). GsdmD is a 487 amino acid cytoplasmic protein that contains a poorly characterized gasdermin domain and lacks any obvious transmembrane segment or signal peptide. Activated by caspase-1 dependent cleavage, the processed p30 amino terminal fragment of GsdmD induces pyroptotic cell death due to its pore forming capacity (19,20,50–58). In the present study, we provide clear evidence that EVs released from monocytes (Thp1) containing active GsdmD potently induce fibroblast and myofibroblast cell death. The necessity of active GsdmD is confirmed by our studies showing that EVs from monocytes deficient in GsdmD, or from monocytes treated with a caspase-1 inhibitor (which prevents cleavage to GsdmD to its active form) prior to LPS, failed to induce fibroblast death. Notably, fibroblast death required the transfer of active GsdmD from the monocytes, as fibroblasts deficient in endogenous GsdmD remained susceptible to death. Treatment of monocytes with Disulfiram to functionally block the pore-forming activity of GsdmD did not prevent transfer of cleaved GsdmD but did attenuate fibroblast death induced by the monocyte-derived EVs, directly implicating the pore-forming function of active GsdmD in fibroblast death. Supporting the direct transfer of active GsdmD from LPS-treated monocytes to fibroblasts, we show active GsdmD is only present in recipient fibroblasts that are exposed to EVs with active GsdmD. Indeed, active GsdmD was clearly identified in fibroblasts in which endogenous GsdmD had been silenced with siRNA, and treatment of fibroblasts with broad-spectrum caspase inhibitors to prevent cleavage of endogenous GsdmD did not impact cell death. Thus, our experiments conclusively demonstrate that active GsdmD from LPS-stimulated monocytes is transferred to recipient fibroblasts and that they are essential for the induction of caspase-independent fibroblast death.

Interestingly, treatment of the mesenchymal cells with TGF-β prior to exposure to the EVs containing active GsdmD did not afford protection from cell death. Moreover, fibroblasts isolated from patients without lung disease and fibroblasts isolated from patients with the fibrotic lung disease, IPF, which have been shown to have increased resistance to apoptotic stimuli, were equally susceptible to death induced by these EVs. Prior studies have established that TGF-β induced mesenchymal cell resistance to apoptosis is mediated by activation of pro-survival kinases, downregulation of Fas expression, and induction of the endogenous inhibitor of apoptosis proteins which function, in part, by preventing caspase cleavage (40,59–61). Thus, EV-mediated transfer of active GsdmD circumvents the extensive anti-apoptotic mechanisms that may contribute to the persistence of myofibroblasts in fibrotic disease states. While the mechanism of cell death in this case remains to be determined, we speculate that it is a direct function of active GsdmD insertion into the recipient cell membrane. Notably, the cargo packaged within EVs from LPS-treated monocytes is not fully characterized, and our studies do not exclude a role for other transported mediators. Indeed, our studies suggest that additional contents do play a role, as induction of cIAP-2 was GsdmD-independent.

Among the mechanisms and mediators that promote mesenchymal cell survival are members of the Inhibitor of Apoptosis Protein (IAP) family. The IAP family represents a group of proteins that are characterized by the presence of at least one baculoviral IAP repeat (BIR) domain (33). In each of these proteins, the BIR domain allows the protein to bind to caspases and prevent caspase cleavage, thereby conferring the in vitro anti-apoptotic capacity from which the protein family derives its name. We have shown that two IAP family proteins, the single BIR-domain containing Survivin and X-linked IAP (XIAP, a member of the subset of “cellular IAP” proteins characterized by three BIR domains and a C-terminal RING domain with E3 ligase function), are induced by the pro-fibrotic cytokine TGF-β1, expressed at increased levels in IPF lung fibroblasts, and that their inhibition or knockdown enhances fibroblast susceptibility to apoptosis (21,42,61,62). Moreover, we have shown that a small molecule SMAC mimetic (AT-406) that functionally inhibits XIAP and the structurally homologous cIAP-1 and cIAP-2 diminishes lung fibrosis in a murine model (22,42). Notably, however, XIAP, cIAP-1 and cIAP-2 (each of which also has a CARD domain between BIR3 and RING) have all been shown to have a broad array of cellular activities mediated via their RING domains (63). Among these, the cellular IAPs have been implicated in necroptosis and NF-kB activation (64,65).

We initially interrogated the IAP family to determine whether the monocyte-derived inflammatory EVs would promote cell death by suppressing IAP expression. To our surprise, cIAP-2 transcription was strongly induced in the recipient mesenchymal cells. In fact, while cIAP-2 protein levels were not readily identified in untreated fibroblasts, treatment with the inflammatory EVs led to detectable levels of cIAP-2. In contrast to active GsdmD, cIAP-2 protein in the mesenchymal cells was endogenously generated and not transferred from the monocytes. The induction of cIAP-2 was independent of the presence of active GsdmD in the vesicles. Functionally, this increase in cIAP-2 contributed to mesenchymal cell death, as knockdown of cIAP-2 in fibroblasts or treatment with the SMAC mimetic AT-406 (which promotes cIAP-2 degradation) actually reduced fibroblast death induced by active GsdmD. These paradoxical findings in which cIAP-2 appears to contribute to cell death highlight emerging complexity of the non-redundant IAP family proteins. The potential mechanisms by which cIAP-2 is induced by EVs from LPS-treated monocytes and contributes to GsdmD-mediated cell death remain to be elucidated, and we speculate that this involves ubiquitin-mediated downregulation of alternative anti-apoptosis proteins. This is not without precedent, as a prior study showed that cIAP-1 mediated ubiquitylation led to the degradation of XIAP and promoted apoptosis in cancer cells (66).

As an additional point, we note that induction of cIAP-2 in IPF fibroblasts treated with EVs from LPS-treated monocytes may be blunted in comparison to the induction observed in the normal fibroblasts. We interpret this observation with caution, as these experiments were not designed to provide a direct comparison of the role of cIAP-2 in the susceptibility of normal and IPF fibroblasts to cell death. Nevertheless, if cIAP-2 induction is important in the regulation of fibroblast death in vivo and cIAP-2 induction in response to EVs from inflammatory monocytes is blunted in IPF fibroblasts, then our findings allow speculation that reduced cIAP-2 could represent an additional mechanism by which reduced fibroblast cell death contributes to the pathogenesis of fibrotic disease.

What emerges from our data is a novel paradigm related to the cross-regulation of mesenchymal cell viability by inflammatory monocytes through the elaboration of active GsdmD delivered via the transfer of extracellular vesicles. While GsdmD activation by the inflammasome has been implicated in inflammatory cell death via pyroptosis, no role of GsdmD has been described in the context of normal wound repair or fibrosis and this study motivates additional exploration of how inflammatory cells communicate with structural cells to orchestrate the evolution and resolution of normal and pathologic repair. In our experiments, monocytes were separated from mesenchymal cells and EVs collected, purified and transferred. Additional studies will need to define the impact of EVs in vivo during normal and fibrotic repair. Moreover, delineation of the mechanisms by which GsdmD triggers death in the mesenchymal cells has not been determined, and the mechanisms by which monocyte-derived EVs regulate cIAP-2 will need to be elucidated.

Key Points:

Active Gsdm-D is transferred from monocytes to fibroblasts via extracellular vesicles

Transfer of active Gsdm-D promotes caspase-independent fibroblast death

Gsdm-D independent cIAP2 induction in recipient fibroblasts contributes to cell death

Abbreviations

EV

extracellular vesicles

CE

cell extract

CMV

control EV

LMV

LPS EV

LPS

lipopolysaccharide

GsdmD

gasdermin-D

cIAP-2

cellular inhibitor of apoptosis 2