Extracellular Vesicles Contribute to the Pathophysiology and Progression of Pleural Fibrosis by Promoting Mesothelial-to-Mesenchymal Transition and Neoangiogenesis

Abstract

Severe pleural space inflammation associated with exudative pleural effusions leads to the development of pleural fibrosis (PF). Pathological tissue remodeling in PF is associated with profibrotic changes in the pleural mesothelium and neoangiogenesis within the fibrotic region. However, the factors that promote these processes remain poorly understood. This study investigates the role of extracellular vesicles (EVs) in the development and progression of PF, focusing on mesothelial-to-mesenchymal transition and neoangiogenesis. Human pleural mesothelial cells (HPMCs) were treated with coagulation proteases FXa (factor Xa) and thrombin, and EV production was quantified using nanoparticle tracking analysis. The functional relevance of these EVs was assessed by evaluating their ability to promote a profibrotic phenotype in HPMCs and induce tube formation in endothelial cells. FXa and thrombin treatments significantly increased EV generation from HPMCs via PAR (protease-activated receptor)-mediated cell signaling. Our studies showed that these EVs primed HPMCs toward a profibrotic phenotype and enhanced tube formation in endothelial cells. Further investigations in preclinical mouse models of PF revealed elevated EV levels in pleural fluids from injury-induced mice, compared with saline control mice. In clinical specimens, exudative pleural effusions from patients with empyema and parapneumonic effusions exhibited significantly elevated EV numbers compared with transudative effusions from patients with congestive heart failure. More importantly, EVs isolated from exudative effusions promoted a profibrotic phenotype in naive HPMCs and enhanced tube formation similar to the effects observed with FXa- and thrombin-generated EVs. These findings offer new insights into PF pathogenesis by identifying EVs as previously unknown contributors that modulate mesothelial-to-mesenchymal transition and neoangiogenesis.

Clinical Relevance

Our work elucidates the novel role of extracellular vesicles (EVs) in the pathogenesis of pleural fibrosis, revealing previously unrecognized mechanisms by which EVs contribute to disease progression. By demonstrating how EVs mediate key signaling events and angiogenesis, our findings introduce a new aspect of pleural fibrosis pathogenesis that has not been explored before. This research expands the current understanding of fibrotic remodeling and highlights EVs as potential therapeutic targets for intervention.

Pleuritis with pleural effusions occurs in a wide range of diseases, with resolution occurring spontaneously or through dedicated treatments (1). However, exudative pleural effusions associated with persistent pleural organization may lead to the development of loculation and pleural fibrosis (PF) (2). The precise mechanisms and/or factors that drive PF remain unclear, which complicates treatment strategies (2, 3). One of the earliest events noted in PF is the activation of the coagulation cascade, which plays an important role in the development of pleural inflammation, as pleural injury is characterized by florid fibrin deposition (4). Significant differences were observed in coagulation activation between transudative and exudative effusions, which may offer insights into the mechanism of PF development and may influence diagnostic and therapeutic approaches to manage organizing pleural effusions (5). Inflammation and tissue injury are the primary triggers to the activation of coagulation cascade (3, 4, 6).

The visceral and parietal pleural mesothelium is the primary site of injury in PF. Furthermore, pleural mesothelial cells (PMCs) play a pivotal role in the pathogenesis of PF, as they contribute to local fibrotic repair (7, 8). Activated PMCs can promote extravascular fibrin deposition, leading to pleural organization with loculation (9–11). PMCs can also undergo profibrotic transition to α-SMA–expressing myofibroblasts via a process we termed mesothelial-to-mesenchymal transition (MesoMT). This phenotypic change accelerates neomatrix deposition and pleural rind formation, leading to scarring of the pleural surface (12–14). MesoMT is induced by various factors that activate distinct signaling pathways mediated by diverse surface receptors (12, 13, 15, 16). We recently reported that mTORC2/Rictor-mediated activation of the SGK1/NDRG1 signaling axis is critical for the progression of MesoMT and PF (17). However, it remains unclear if additional factors contribute to MesoMT in PF.

Increased cellular infiltration and proliferation increase the demand for essential nutrients and oxygen. Combined with excessive matrix deposition, these pathological processes may lead to hypoxic conditions, thereby triggering neoangiogenesis. In our previous publication, we showed increased angiogenesis in the thickened pleura due to the progression of PF in a mouse model induced by Streptococcus pneumoniae (18). Prior studies have shown increased angiogenesis in rabbits in an empyema model of pleural injury (19). However, the factors driving neoangiogenesis in PF remain unidentified, representing a key gap in our understanding of processes that support the development and progression of PF (9, 20, 21).

EVs are increasingly acknowledged to play an important and novel role in intercellular communication, distinct from direct cell-to-cell contact or the transfer of secreted molecules (22). EVs are bilayer-membrane structures released by nearly all cells, which cannot replicate on their own (23). They can, however, intricately regulate the activity of diverse cell types through the transfer of complex cargo from donor cells to recipient cells, substantially contributing to various pathophysiological functions (23, 24) that support fibrogenesis (25–28). However, the specific contribution of EVs to the development and progression of PF has not been studied and remains unknown.

The present study aimed to investigate biogenesis and the role of EVs, focusing specifically on MesoMT and angiogenesis in PF. In this study, we demonstrate that coagulation proteases FXa and thrombin promote EV generation from human pleural mesothelial cells (HPMCs) in a time- and dose-dependent manner through the activation of distinct PARs (protease-activated receptors). We observed significantly increased EV numbers in the conditioned media of HPMCs treated with these proteases compared with untreated cells. Importantly, these in vitro findings were consistent with our in vivo studies, in which we found elevated EV levels in pleural lavages collected from our mouse models of pleural injury (S. pneumoniae or carbon black+bleomycin [CBB]-induced pleural injury) compared with saline-challenged control mice. More importantly, elevated EV numbers were also detected in pleural exudates from patients with empyema (EMP) and parapneumonic pleural effusions (ParaPN) compared with transudative effusions from patients with congestive heart failure (CHF), highlighting the clinical relevance of our findings. Furthermore, EVs isolated from the conditioned media of HPMCs treated with FXa or thrombin, as well as those from pleural effusions of patients with EMP and ParaPN, promoted a profibrotic phenotype in naive HPMCs and induced endothelial tube formation in human umbilical vein endothelial cells (HUVECs).

Methods

See the data supplement for additional details.

Pleural Fluid Collection

The Institutional Review Board at The University of Texas Health Science Center at Tyler approved an Exempt Protocol to collect pleural fluids from patients. Pleural fluids were obtained from deidentified patients clinically diagnosed with EMP, ParaPN, or CHF using standard thoracocentesis, as previously described (11). Briefly, the fluids were aspirated using polypropylene tubes containing 3.8% sodium citrate at a ratio of 9:1 (pleural fluid:sodium citrate). The fluids were centrifuged at 400 × g for 10 minutes at room temperature to remove debris and intact cells. The resulting supernatants were aliquoted and immediately stored at −80°C until further use. LDH (lactate dehydrogenase) concentration was determined by the clinical laboratory, and total protein concentration was assessed by the bicinchoninic acid method. The effusions were classified as transudative or exudative based on both clinical evaluation using Light’s criteria and the underlying clinical condition of the patient (21, 29).

HPMC Isolation and Primary Cell Culture

HPMCs were isolated from transudative pleural fluids collected from patients with CHF as described previously (1, 2), with the collection approved under an exempt protocol as mentioned above. HPMCs were maintained in a humidified incubator at 37°C in 5% CO₂–95% air and cultured in bronchial epithelial growth medium medium (containing the Bullet Kit minus epinephrine and retinoic acid; Lonza) containing 3% FBS (Gibco), 2% antibiotic-antimycotic (Lonza), and 1% GlutaMAX (Invitrogen), as previously described (3). The primary cells were initially screened for the mesothelial-specific marker calretinin by flow cytometry, and only those cells with >85% positivity were used for the indicated studies. Although other cell types may also express calretinin, their presence in pleural effusions is unlikely. Our earlier isolation protocols visualized a four-marker panel provided by our resident pathologist to ensure cellular identity, including HBME, cytokeratin 5/6, thrombomodulin, and calretinin. Because of the heterogeneity of expression associated with some of these markers (30), we began using calretinin expression exclusively as detected via FACS (16). When necessary, mesothelin was used as a secondary marker for further validation. HPMCs were subjected to a maximum of five passages (4).

Results

Coagulation Factors Xa and Thrombin Promote EV Generation in HPMCs

We previously showed that diverse mediators promote MesoMT, thereby contributing to the pathogenesis of PF (15, 17). Here, we evaluated the ability of these mediators to enhance EV biogenesis from unperturbed HPMCs. FXa and thrombin significantly increased EV production in HPMCs compared with control and other treatments (Figure 1A). TGFβ, plasmin, and uPA failed to enhance EV generation (Figure 1A). Increasing TGFβ dose or treatment duration did not potentiate the release of EVs from HPMCs (see Figures E1A and E1B in the data supplement).

Figure 1.

Coagulation factors FXa (factor Xa) and thrombin promote extracellular vesicle (EV) generation in human pleural mesothelial cells (HPMCs). (A) HPMCs were serum starved for 8 hours and treated with TGFβ (5 ng/ml), FXa (10 nM), thrombin (10 nM), uPA (20 nM), or plasmin (0.5 μg/ml) in fresh serum-free medium for 16 hours. EVs released into cell supernatants were isolated and quantified by nanoparticle tracking analysis (NTA). Serum-starved HPMCs were treated with FXa to assess (B) time response, where 10 nM FXa was used for varying time intervals, or (C) dose response, where different doses of FXa (1, 2, 7, and 10 nM) were used for 16 hours. At the end of the treatment, EVs were harvested and analyzed by NTA. (D) Size variation of EVs generated after FXa treatment at different concentrations. Similarly, serum-starved HPMCs were treated with thrombin to assess (E) time response, where 10 nM thrombin was used for varying time intervals, or (F) dose response, where different doses of thrombin (0.1, 0.5, 1, 2, 5, and 10 nM) were used for 16 hours. At the end of the treatment, EVs were harvested and analyzed by NTA. (G) Size variation of EVs generated after thrombin treatment at different concentrations. (H) HPMCs were incubated with PKH67 dye (20 mM) for 30 minutes at 37°C in a serum-free medium. Cells were then fixed and stained with DAPI (1:1,000) for 30 minutes and subjected to imaging by a confocal microscope. (I) PKH67-labeled cells were serum starved for 1 hour, followed by treatment with FXa or thrombin (10 nM each) for 16 hours together with untreated control. EVs were isolated from the culture supernatant and were quantified by measuring PKH67 fluorescence intensity. All data are representative of a minimum of three independent experiments and are presented as means ± SEM. Data were evaluated by one-way ANOVA with Šidák post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. ns = no statistically significant difference.

Further investigation showed that FXa enhanced EV generation in a time- (Figure 1B) and dose-dependent (Figure 1C) manner. Thrombin, likewise, induced EV generation in a time- (Figure 1E) and dose-dependent manner (Figure 1F) from HPMCs. A significant change in FXa and thrombin-mediated EV production was observed by 8 hours after treatment (Figures 1B and 1E). Cryo-phase–transmission electron microscopy of HPMC EVs confirmed the characteristic membrane lipid bilayer (Figure E2). Further characterization of the EVs showed that FXa and thrombin-generated EVs were significantly smaller than the EVs generated by control HPMCs (Figures 1D and 1G). EV generation by FXa and thrombin was also confirmed by fluorescent cell surface labeling of HPMCs with the cell-impermeable fluorescent dye PKH67 (Figure 1H). PKH67-labeled cells were treated with FXa (10 nM) or thrombin (10 nM). EVs isolated from the supernatants of labeled cells showed a significant increase in fluorescence intensity in FXa- and thrombin-treated samples, indicating an increase in EV number in FXa- and thrombin-treated HPMCs (Figure 1I). Analysis of exosomes, isolated by centrifugation of the conditioned media at 100,000 × g for 1 hour, revealed that FXa and thrombin treatment had no impact on their release (data not shown).

Although differential centrifugation is an accepted and widely used method for EV isolation, this technique has some limitations. One such limitation is the potential presence of protein aggregates and lipids that are sedimented with EVs during centrifugation. This may pose a problem when using nanoparticle tracking analysis (NTA). To address this issue, we used size-exclusion chromatography to validate our results, per guidelines established by the Minimal Information for Studies of Extracellular Vesicles consortium 2023 (22). Izon columns were used to purify control, FXa-, or thrombin-induced EVs. Analysis of EVs obtained after size-exclusion chromatography showed significant increases in EV numbers from HPMCs stimulated with FXa or thrombin compared with the control EVs. These results confirmed our findings, as observed by other methods (Figure E3).

Protease-activated Receptors Play a Crucial Role in Thrombin- and FXa-mediated Generation of Extracellular Vesicles

Because PARs mediate FXa and thrombin signaling (31), we evaluated the role of PAR1 and PAR2 in the generation of EVs by these proteases. Specific siRNA was used to silence PAR1 (Figure 2A) or PAR2 expression (Figure 2B) before exposing the cells to FXa or thrombin. Treatment with specific siRNAs resulted in >95% reduction in mRNA levels of both PAR1 and PAR2, ensuring effective knockdown before subsequent experimental treatments (Figures 2A and 2B). In addition, the knockdown of PAR1 and PAR2 was verified by immunoblot analysis (Figures E4A and E4B). As shown in Figure 1, treating HPMCs with FXa or thrombin markedly increased EV generation. PAR2 silencing significantly reduced EV generation induced by FXa (Figure 2C). PAR1 silencing had no effect on FXa-mediated EV release. Conversely, PAR1 knockdown significantly impacted EV generation by thrombin. However, PAR2 knockdown did not affect thrombin-mediated EV release. Control siRNA had minimal impact on EV generation induced by FXa and thrombin. These findings underscore the selective PAR-signaling mechanisms used by FXa and thrombin for EV generation in HPMCs.

Figure 2.

Protease-activated receptors are essential for the generation of extracellular vesicles induced by thrombin and FXa. HPMCS were transfected with control, PAR1 (protease-activated receptor 1)-, or PAR2-specific siRNAs (200 nM each). Transfected cells were serum starved for 8 hours and then treated with FXa or thrombin (10 nM each) in fresh serum-free medium for 16 hours (overnight). At the end of the incubation, cells were lysed, and RNA was extracted to confirm the knockdown of (A) PAR1, and (B) PAR2 by measuring their mRNA expression levels using quantitative PCR analyses. The effect of PAR1 or PAR2 knockdown on EV generation was assessed by quantifying the EVs released into cell supernatants after treatment with (C) FXa or (D) thrombin, using the NTA method as previously described. Data are representative of a minimum of three independent experiments and are presented as mean ± SEM and evaluated by one-way ANOVA with Šidák post hoc test. *P < 0.05, **P < 0.01, and ****P < 0.0001. At least two reference genes were used to calculate the relative expression levels of target genes in quantitative PCR studies.

FXa- and Thrombin-generated EVs Activate Profibrotic Signaling Pathways in HPMCs

The transitioning of HPMCs to myofibroblasts via induction of MesoMT plays a critical role in the progression of PF (15). We previously demonstrated that coagulation proteases FXa and thrombin induce MesoMT in HPMCs (15, 17). However, it remains unclear whether EVs generated by these proteases contribute to MesoMT. To explore this possibility, we investigated whether these EVs could promote MesoMT in naive HPMCs and evaluated their physiological relevance by examining their impact on HPMC phenotype. An initial pilot study conducted using three concentrations of EVs (10⁸, 2 × 10⁸, and 4 × 10⁸/ml), showed that 4 × 10⁸ EVs/ml is the optimal dose to induce MesoMT; therefore, 4 × 10⁸ EVs/ml was used for all subsequent experiments. Serum-starved HPMCs were treated with control, FXa-, or thrombin-generated EVs for 48 hours. Distinct morphological changes were observed in naive HPMCs 48 hours after treatment with both FXa- and thrombin-generated EVs (Figure E5). Immunoblot analysis of treated HPMCs showed that FXa- and thrombin-generated EVs induced significant AKT activation via phosphorylation at both Thr308 and Ser473 compared with PBS controls. Conversely, control EVs failed to phosphorylate AKT both at Thr308 and Ser473 (Figure 3A). Thrombin EVs exhibited robust induction of the myofibroblast marker α-SMA, whereas FXa EVs resulted in a modest increase. However, the control EV-treated sample also showed a mild increase in α-SMA levels (Figure 3A). It should be noted here that exposure of HPMCs to FXa or thrombin releases four- to fivefold more EVs than control vehicle (Figure 1). However, in the above studies, we used equal numbers of control, FXa, or thrombin EVs.

Figure 3.

FXa- and thrombin-generated EVs induce mesothelial-to-mesenchymal transition in HPMCs and induce tube formation in human umbilical vein endothelial cells (HUVECs). (A) HPMCs were starved for 8 hours and treated with FXa or thrombin (10 nM each) in fresh serum-free medium for 16 hours (overnight). EVs were isolated from the conditioned media, and FXa- or thrombin-generated EVs (4 × 10⁸) were added to serum-starved HPMCs cultured in a 6-well plate. After 48 hours, cell lysates were harvested, and an equal amount of proteins of cell lysates were subjected to immunoblot analysis to assess the phosphorylation status of AKT (Thr 308), AKT (Ser 473), and expression levels of α-SMA. GAPDH was used as the loading control. Quantitative analysis of Western blot data revealed that both FXa- and thrombin-generated EVs significantly increased AKT phosphorylation at Thr308 and Ser473 compared with PBS control cells. (B) HPMCs were starved overnight with or without AKT inhibitor, AKT VIII (200 nM), then treated with control, FXa, or thrombin EVs (4 × 10⁸). After 48 hours, cell lysates were prepared for immunoblot analysis to assess phosphorylation of AKT (Thr308, Ser473), S6 (Ser235, Ser244), and NDRG1 (Thr346), together with α-SMA and total AKT expression. GAPDH was used as the loading control. Equal volumes of supernatants were immunoblotted to evaluate fibronectin (Fn). Next, HPMCs were starved overnight with or without mTORC1/C2 inhibitor INK128 (200 nM) and then treated with (C) control or FXa EVs (4 × 10⁸) or (D) control or thrombin EVs (4 × 10⁸). After 48 hours, cell lysates were analyzed by immunoblotting for AKT (Thr308, Ser473), S6 (Ser235, Ser244), NDRG1 (Thr346), and α-SMA, with GAPDH as the loading control. Supernatants were also analyzed for Fn. (E) For endothelial tube formation assay, HUVECs were grown in a 6-well plate for 48 hours to 70–75% confluency. The cells were incubated with FXa- or thrombin-generated EVs (4 × 10⁸) per well for 16 hours (overnight). Next day, HUVECs incubated with EVs were trypsinized and counted and plated at 4 × 10⁴ cells per well of a 24-well plate coated with growth factor reduced Matrigel and incubated at 37°C and 5% CO₂ for 6–8 hours. Endothelial tube formation was observed using a Nikon eclipse Ti microscope in phase contrast and photographed at 4× magnification using Nikon digital sight DS-Fi1 camera and NIS elements BR software. For each sample, four random images were captured, and endothelial tubular network formation was quantified by analyzing the number of (F) nodes, (G) junctions, (H) meshes, and (I) total tube length using the Angiogenesis Analyzer plugin available in ImageJ. The data obtained for quadruplets were averaged and presented. Each data point represents a single patient sample. All data are representative of a minimum of four independent experiments and are presented as means ± SEM and evaluated by one-way ANOVA with Šidák post hoc test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Our previous studies demonstrated that the activation of the PI3K/AKT pathway is crucial for inducing MesoMT (16). Phosphorylation of AKT at Thr308 activates the mTORC1/C2 complexes, which in turn promote MesoMT (17). To evaluate the role of AKT in FXa and thrombin EV-induced MesoMT, HPMCs were serum starved for 16 hours (overnight) with or without the AKT inhibitor (AKT VIII, 200 nM), followed by treatment with control, FXa, or thrombin EVs. Both FXa and thrombin EVs significantly activated AKT, as evidenced by increased phosphorylation at Thr308 and Ser473. In addition, these EVs also activated mTORC1, as shown by increased phosphorylation of S6 at Ser235 and Ser244, and mTORC2, indicated by phosphorylation of NDRG1 at Thr346. Both FXa and thrombin EVs also induced increased expression of α-SMA and fibronectin, markers of MesoMT. Inhibition of AKT with AKT VIII effectively prevented the phosphorylation of AKT at both Thr308 and Ser473, without affecting total AKT levels (Figure 3B). Moreover, the inhibition of AKT blocked the phosphorylation of both C1 (S6) and C2 (NDRG1) and also suppressed the increased expression of α-SMA and fibronectin (Figure 3B) induced by FXa and thrombin EVs, confirming the involvement of the AKT pathway in the process of MesoMT.

Because AKT inhibition with AKT VIII indicated a direct role of mTORC1/C2 in MesoMT induction, we further investigated the involvement of these complexes by directly inhibiting them with INK128, an mTORC1/C2 inhibitor. HPMCs were treated with control, FXa, or thrombin EVs in the presence or absence of INK128 (200 nM) to assess the specific role of mTORC1/C2. INK128 treatment significantly reduced the phosphorylation of mTORC1 substrates (S6 at Ser235/Ser244) and mTORC2 substrates (AKT at Ser473 and NDRG1 at Thr346) in both control and FXa EVs (Figure 3C) or control and thrombin EVs (Figure 3D). INK128 inhibition of mTORC1/C2 did not diminish the increased AKT phosphorylation at Thr308, indicating that this phosphorylation occurs upstream of mTORC1/C2 activation. INK128 suppressed the FXa and thrombin EV-induced expression of α-SMA and fibronectin, further supporting the essential role of mTORC1/C2 signaling in EV-driven MesoMT (Figures 3C and 3D). These findings provide strong evidence that mTORC1/C2 is a critical mediator of EV-induced MesoMT in pleural fibrosis.

FXa- and Thrombin-generated EVs Promote Tube Formation in HUVECs

In the healthy human lung, a single layer of mesothelial cells lines the pleural surface. Capillaries are confined to the submesothelial and alveolar regions. We previously reported increased angiogenesis within the pleural thickening associated with the fibrotic areas in mouse lungs with S. pneumoniae–induced PF (18). To assess angiogenesis in clinical disease, human nonspecific pleuritis tissue sections were immunostained for the angiogenic marker CD31 (32). Our analysis revealed an increase in blood vessel density (black arrows) within the fibrotic region (red line) (Figure E6). We next assessed the angiogenic potential of FXa- and thrombin-induced EVs by endothelial tube formation assay. HUVECs treated with FXa- or thrombin-induced EVs (4 × 10⁸) displayed significantly greater tube formation than those treated with a similar number of control EVs (Figure 3F). Compared with HUVECs incubated with control EVs, cells exposed to FXa- and thrombin-induced EVs (n = 4 each) exhibited a significantly increased number of nodes (302.4 ± 32.04 vs. 629.1 ± 44.01 and 574.2 ± 50.81, respectively) (Figure 3F), junctions (87.23 ± 8.55 vs. 177.9 ± 14.41 and 168.5 ± 13.88, respectively) (Figure 3G) and meshes (14.79 ± 2.38 vs. 56.5 ± 6.12 and 50.56 ± 6.23, respectively) (Figure 3H) (33). Total tube length was also significantly higher in FXa- and thrombin-incubated endothelial cells compared with the control EV–incubated cells (Figure 3I) (33). The effects mediated by FXa- and thrombin-generated EVs could have been attributed to the presence of FXa and thrombin in the EV preparations (34, 35). To exclude this possibility, we measured FXa and thrombin levels in EVs in sensitive chromogenic substrate-based assay (36). We found minimal to no detectable levels of FXa or thrombin in EVs (data not shown). This finding confirms that the effects of EVs derived from FXa and thrombin in HPMCs on MesoMT and angiogenesis are mediated by the EVs themselves rather than by FXa or thrombin that may be bound to or released by them.

Elevated EV Levels Occur in Human Exudative Pleural Effusions and Mouse Pleural Lavage Fluids after Injury

To investigate the clinical relevance of our findings, we examined potential differences in EVs from transudative (CHF) and exudative (EMP, ParaPN) pleural effusions by isolating EVs from equal volumes (1 ml) of pleural effusions collected from 65 patients (CHF, n = 27; EMP, n = 24; and ParaPN, n = 14) (36). EV numbers and size were determined using NTA (Figures 4A–4D). We observed fivefold higher EV numbers in exudative effusions (EMP, 12.59 ± 1.343 × 10⁹ and ParaPN, 11.74 ± 0.9217 × 10⁹) compared with those obtained from transudative effusions (CHF, 2.462 ± 0.356 × 10⁹) (Figure 4A). EMP and ParaPN EVs were also smaller in size, with a mean diameter of 169.6 ± 9.032 and 180.8 ± 12.29 nm, respectively, compared with CHF EVs, which had a mean diameter of 273.5 ± 13.42 (Figures 4B and 4D). The size distributions of EMP and ParaPN EVs were also significantly smaller than CHF EVs (Figure 4C).

Figure 4.

Elevated EV levels found in human exudative pleural effusions and mouse pleural lavages after injury. EVs were isolated from 1 ml of pleural effusions from patients and analyzed by NTA using NanoSight for (A) number of EVs and (B and C) size. For congestive heart failure (CHF), n = 27; empyema (EMP), n = 24; and parapneumonic pleural effusions (ParaPN), n = 14. (D) Representative spectrum showing the diameter range of EVs analyzed by NTA. EVs were isolated from the pleural lavages of mice injured with (E) Streptococcus pneumoniae (1.8 × 10⁸ cfu) for 48 hours (Saline, n = 6; Strep, n = 8) or 7 days (Saline, n = 6; Strep, n = 7), or (F) carbon black/bleomycin (CBB) for 14 days (Saline, n = 6; CBB, n = 6) and were analyzed by NTA. Data are expressed as means ± SEM and were evaluated by one-way ANOVA with Šidák post hoc test or nonparametric Mann-Whitney test. **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Empyema and complicated parapneumonic effusions are frequently associated with bacterial infections (37–39). Because all cells, including bacteria, can produce EVs (22), we next investigated whether the increase in EVs observed in exudative effusions could be attributed to bacterial EVs rather than human-derived EVs. To address this possibility, we assessed the percentage of 16S prokaryotic ribosomal RNA (16S rRNA) in the total EVs isolated from 1 ml of patient pleural effusions (40, 41). A known quantity of S. pneumoniae–derived EVs was used to amplify 16S rRNA using universal primers, and a standard curve was generated to calculate the number of bacterial EVs in EVs isolated from pleural effusions of above patient groups. The results indicated that both exudative and transudative effusions contained either no detectable or a negligible number of (<0.5%) bacterial EVs (data not shown). To further confirm the mammalian origin of these EVs, we followed MISEV guidelines (23). Western blot analysis of the EV lysates that showed the presence of cytosolic proteins (HSP70 and GAPDH) and transmembrane proteins (TF, CD63) confirmed their mammalian origin (Figure E7) (23). Because the mesothelium is the primary site of disease initiation, the mesothelial origin of these EVs was investigated using the mesothelium-specific marker mesothelin. Mesothelin expression was also detected in patient EV lysates, suggesting a mesothelial origin among these EVs (Figure E7).

Next, we determined if EV numbers increase during disease progression in mouse models of pleural injury. EVs were isolated from the pleural lavages of two distinct pleural injury mouse models and saline controls (15, 18). We found a significant (∼2.5-fold) increase in EV number in S. pneumoniae–induced PF at 48 hours (saline: 1.505 ± 0.108 × 10⁸; Streptococcus infected: 3.634 ± 0.45 × 10⁸). Similar results were observed at 7 days (saline: 1.607 ± 0.083 × 10⁸; Streptococcus infected: 4.101 ± 0.465 × 10⁸) (Figure 4E). We next assessed our sterile preclinical model of pleural injury using CBB (Figure 4F). In the CBB pleural injury model, pleural effusions were collected 14 days after injury. EV numbers were 2.7-fold higher in pleural lavages of CBB-injured mice compared with the control mice (saline: 1.719 ± 0.12 × 10⁸; CBB: 4.663 ± 0.276 × 10⁸) (Figure 4F).

EVs from Exudative Pleural Effusions Activate HPMCs

We next investigated whether EVs from pleural effusions could promote profibrotic signaling pathways in unperturbed HPMCs. We treated quiescent HPMCs with exudative or transudative EVs for 48 hours and evaluated their profibrotic potential by assessing markers of MesoMT and activation of profibrotic signaling pathways. Most exudative EVs (EMP and ParaPN) induced dual phosphorylation of AKT at Thr308 and Ser473 (Figure 5A) and significantly increased α-SMA expression (Figures 5A and 5B). In contrast, treatment with transudative EVs mostly resulted in phosphorylation at Ser473 but not at Thr308 (Figure 5A). Although some transudative EVs elicited dual phosphorylation, the expression of α-SMA in these samples was minimal (Figure 5B). Similarly, exudative EV-treated HPMCs showed significantly higher fibronectin levels than those treated with transudative EVs (Figure 5D). Although no significant differences in collagen production were observed between HPMCs treated with exudative versus transudative EVs (Figure 5C), Western blot data analysis showed a 37% increase in collagen production in HPMCs treated with exudative EVs compared with PBS-treated controls, whereas transudative EVs resulted in only an 8.5% increase in collagen levels. When analyzed separately, collagen expression in HPMCs treated with EVs isolated from EMP exudative effusions was significantly higher than in those treated with EVs from transudative CHF. However, there was no significant difference in collagen expression between cells treated with ParaPN EVs and those treated with transudative CHF EVs (Figure E8), suggesting intrinsic differences between the EVs isolated from exudative effusions of patients with EMP or ParaPN conditions.

Figure 5.

EVs from exudative pleural effusions activate mesothelial cells. Serum-starved HPMCs were treated with EVs isolated from 1 ml of transudative (CHF) or exudative pleural effusions (EMP and ParaPN) from patients for 48 hours. (A) After 48 hours of treatment, cell lysates were harvested, and equal concentrations were immunoblotted to assess the phosphorylation status of AKT (Thr308 and Ser473) and expression levels of α-SMA. GAPDH was used as the loading control. Equal volumes of cell supernatants were immunoblotted to probe for the presence of matrix proteins collagen1a (Col-1a) and Fn. For statistical analysis, data from four independent Western blots with different patient samples were combined. Quantified data are shown for (B) α-SMA, (C) collagen, and (D) fibronectin. Statistical significance was determined by the Mann-Whitney test. *P < 0.05.

EVs from Exudative Pleural Effusions Induce Endothelial Cell Signaling and Promote Tube Formation

To evaluate the angiogenic potential of EVs isolated from pleural effusions, HUVECs were incubated with exudative or transudative EVs, and their effect on endothelial tube formation was assessed. VEGF-treated HUVECs were used as positive control. Exudative EVs significantly increased tube formation in HUVECs compared with transudative EV–treated cells (Figure 6A). Specifically, exudative EVs (n = 10) significantly increased the number of nodes (536.7 ± 29.67) (P < 0.001; Figure 6B), junctions (157.4 ± 8.4) (Figure 6C) and meshes (51.87 ± 4.602) (P < 0.001; Figure 6D) compared with transudative EV–treated HUVECs (nodes: 289.3 ± 28.52; junctions: 85.77 ± 8.31; and meshes: 18.03 ± 3.15; n = 10). The number of nodes, junctions, and meshes exhibited by exudative EV–treated HUVECs were similar to VEGF-treated HUVECs (30 ng/ml; nodes: 576.9 ± 33.56; junctions: 171.9 ± 9.06; meshes: 59.89 ± 5.58; n = 3) (Figures 6B–6D) (33). A significant increase in total tube length was also observed with exudative EVs compared with transudative EVs (P < 0.001; Figure 6E). These data suggest that exudative EVs strongly promote endothelial tube formation, supporting their angiogenic potential to occur in vivo.

Figure 6.

EVs from exudative pleural effusions promote endothelial tube formation. (A) For endothelial tube formation assay, HUVEC cells were grown in a 6-well plate for 48 hours to 70–75% confluency. The cells were incubated with EVs isolated from 1 ml of exudative or transudative pleural effusions from patients for 16 hours (overnight). Next day, HUVECs incubated with EVs were trypsinized, counted, and plated at 4 × 10⁴ cells per well in a 24-well plate coated with growth factor reduced Matrigel. Plates were then incubated at 37°C and 5% CO₂ for 6 hours. Endothelial tube formation was observed using a Nikon eclipse Ti microscope in phase contrast and photographed at 4× magnification using Nikon digital sight DS-Fi1 camera and NIS elements BR software. For each sample, four random images were captured, and endothelial tubular network formation was quantified by analyzing the number of (B) nodes, (C) junctions, (D) meshes, and (E) total tube length using the Angiogenesis Analyzer plugin available in ImageJ. The data obtained for quadruplets were averaged and presented. Each data point represents a single patient sample. Data are presented as means ± SEM and evaluated by one-way ANOVA with Šidák post hoc test. ***P < 0.001 and ****P < 0.0001. SFM = serum free medium.

Discussion

PF is a debilitating condition marked by progressive scarring of the lung surface and impaired lung function. A comprehensive understanding of the mechanisms involved in the pathogenesis of PF warrants continued study (33). Exudative pleural effusions, commonly associated with PF, may hold the key to identifying factors contributing to its development (2). Once viewed as cellular debris, EVs are now recognized as vital communicators found in various bodily fluids, capable of exerting significant influence over both normal and pathophysiologic processes (23–27). Understanding their role in PF remains rudimentary. In this study, we demonstrate, for the first time, that coagulation proteases FXa and thrombin induce EV generation from pleural mesothelial cells via distinct PAR-mediated signaling. We also provide compelling evidence that these EVs are biologically active and likely play a significant role in disease progression by promoting profibrotic responses and angiogenesis—two key processes in the development of PF. Moreover, we observed a significant increase in the number of functionally active EVs in exudative pleural effusions, which are commonly associated with fibrosing pleural injury. This elevated EV presence suggests an active role in the local microenvironment, where they may contribute to the progression of PF. The increase in EV numbers and their enhanced biological activity point to a dynamic mesothelial response to injury, likely driven by inflammatory and profibrotic stimuli. These insights address critical gaps in our understanding of PF pathogenesis and highlight the potential role of EV-mediated intercellular communication in processes such as mesothelial cell activation and neoangiogenesis. The findings presented herein offer a predicate for future therapeutic interventions.

The relationship between inflammation of the pleura and the activation of the coagulation cascade presents a conundrum. Inflammation triggers the expression and activation of proteases involved in coagulation (42, 43). Conversely, tissue injury activates these coagulation proteases, which in turn initiate an inflammatory response (4, 44). PF develops because of various inflammatory conditions in which pleural injury activates mesothelial cells and recruits inflammatory cells into the pleural space. Both resident mesothelial cells and infiltrating immune cells produce cytokines and chemokines, which activate signaling pathways that upregulate tissue factor, a critical initiator of coagulation (7, 8). The expression and activation of tissue factor initiates the coagulation cascade, converting FX to Xa and generating thrombin, contributing to the inflammatory environment and fibrotic processes associated with pleural fibrosis (3, 4). Our earlier studies reported increased thrombin–antithrombin complex levels in S. pneumoniae–induced PF, indicating enhanced activation of coagulation (18). We also demonstrated that the activated coagulation proteases FXa and thrombin, resulting from this enhanced coagulation, contribute to the pathology of PF by promoting MesoMT (15, 17).

Recent studies, including ours, suggest that coagulation proteases trigger the release of EVs from endothelial cells, which are involved in regulating hemostasis, inflammation, and barrier disruption (45–48). Endothelial EVs, together with those from other cell types, have been implicated in tissue remodeling and several fibrotic diseases, including idiopathic pulmonary fibrosis (IPF) (25, 49–53). Although EVs have been implicated in lung injury and inflammation, their specific role in the development and progression of PF remains unclear (54, 55). In the present study, we report that both FXa and thrombin generate EVs from unperturbed HPMCs in a time- and dose-dependent manner. In parallel studies, FXa failed to elicit EV generation in endothelial cells (data not shown), suggesting cell-specific variations in the underlying signaling mechanisms. In HPMCs, EV generation occurs through distinct PAR activation pathways. FXa activates PAR2, whereas thrombin operates through PAR1 (31). Although PARs are known to promote PF progression through direct cell signaling (6, 16), our study uncovers their novel role in indirectly facilitating this process via the generation of EVs.

In healthy individuals, pleural fluid levels are maintained in a state of homeostasis, balancing production from the parietal pleura and resorption via lymphatic channels, typically averaging 0.26 ± 0.1 ml/kg (56). Disruption of this homeostasis due to underlying pathophysiological conditions, such as CHF and renal and/or kidney diseases, can lead to transudative effusions (57). Exudative effusions typically arise from increased permeability of the pleural membranes and microvasculature in response to injury or insult to the pleural compartment (58). Our study reveals, for the first time, significantly increased EVs in exudative effusions compared with transudative effusions. Exudative EVs were also significantly smaller than transudative EVs. These differences suggest variation in their biogenesis, in part originating from the pleural mesothelium. Moreover, our studies in mouse pleural injury models also show increased EVs in different forms of fibrosing pleuritis. Earlier studies have reported elevated thrombin–antithrombin complexes in exudative effusions, indicating activation of the coagulation cascade that result in factor Xa and, subsequently, thrombin generation (5). These data suggest that factor Xa and thrombin in exudative effusions may be responsible for the elevated EVs observed in exudative effusions.

PF is characterized by the increased abundance of α-SMA–expressing myofibroblasts, derived from PMCs undergoing MesoMT. These proliferating myofibroblasts resist apoptosis, express pathological amounts of matrix proteins, and acquire a contractile phenotype leading to fibrotic repair of the pleura (13, 15, 59, 60). Our recent study revealed the critical role of the SGK1/NDRG1 axis mediated by mTORC2 activation in driving these processes (15). EVs are known to carry profibrotic signals, contributing to the pathogenesis of several fibrotic diseases, including IPF (25, 52). In IPF, for example, EVs have been shown to promote fibroblast proliferation in an autocrine fashion, by transferring molecules such as WNT5A (Wingless/integrase 1-5A), thus contributing to disease pathogenesis (25). WNT5A is implicated in several disease conditions, including fibrosis (61). Huang and colleagues demonstrated that injured mesothelial cells release EVs enriched in integrin-like kinase, which activate fibroblasts via the p38 MAPK pathway, promoting peritoneal fibrosis (62). Our recent studies show that coagulation proteases promote endothelial EV release, influencing coagulation, inflammation, and barrier disruption. FVIIa- and APC-induced EVs carry antiinflammatory microRNAs (miRNAs) that suppress inflammation by transferring miRNAs to monocytes, macrophages, and endothelial cells (45, 46, 48, 63). In the current investigation, we found that both FXa- and thrombin-generated EVs promoted a profibrotic phenotype in unperturbed HPMCs evidenced by a modest induction of the MesoMT marker α-SMA. Furthermore, these findings are strongly supported by our results showing that exudative EVs also activated profibrotic signaling pathways in these cells. Although the induction of these profibrotic markers is modest, it can be inferred that these EVs may be activated by other factors, perhaps concurrently, thereby enhancing this pathological process. Overall, our data strongly suggest their participation in the pathogenesis of PF. Our studies suggest that these EVs carry profibrotic and proangiogenic cargo (miRNAs, surface and intracellular proteins, metabolites, etc.), which they may transfer to recipient cells to induce these effects.

Neoangiogenesis is commonly observed in diverse forms of tissue fibrosis, indicating its importance in progression of such conditions (64, 65). However, the role of angiogenesis in PF remains controversial. We and others have shown increased vessel formation in the lungs of patients with PF (18, 19). In this study, we similarly identified increased angiogenesis in tissue samples of patients with pleuritis. By contrast, vascular density is strikingly decreased in fibrotic foci of IPF tissue sections (66, 67). The angiogenic potential of FXa and thrombin is well documented, as both are reported to induce the expression of diverse angiogenic factors (34, 35). However, at the concentrations used to generate EVs, FXa and thrombin did not induce endothelial tube formation (data not shown). However, HPMC-derived EVs generated by FXa and thrombin robustly induced endothelial tube formation. This apparent discrepancy reinforces the suggestion that the mechanisms underlying EV-mediated effects may be cell specific and warrants further investigation.

Our investigation also showed distinct differences in the ability of transudative and exudative EVs to induce endothelial tube formation. These differences suggest variation in their composition. Specifically, exudative EVs significantly increased angiogenesis, as evidenced by the enhanced formation of nodes, junctions, and meshes. These distinctive effects suggest EVs exert both autocrine and paracrine cellular effects. Numerous molecules, including miRNA found in EVs, are reported to promote angiogenesis (68). However, the specific molecule(s) responsible for inducing endothelial tube formation in HPMC-derived EVs remain unidentified and are under active investigation.

In our experiments with HPMCs and endothelial cells, we used patient-derived EVs based on equal volumes of effusions rather than equal EV numbers, resulting in treatment with fewer transudative EVs. One could argue that the differences observed may stem from these variations in EV numbers. However, it is important to note that normal human pleural fluid volume is very low (0.26 ± 0.1 ml/kg) compared with patients with pleural effusions (56). Under disease conditions, EV concentrations in the pleural space are likely much higher locally and represent just one of many components in the exudative milieu.

Our recent studies suggest that TGFβ, FXa, and thrombin induce MesoMT in mesothelial cells via the activation of the mTORC2/SGK1/NDRG1 signaling axis (17). However, only FXa and thrombin generated EVs from HPMCs. The discordance in EV production between these proteases and TGFβ, despite their shared ability to induce MesoMT, suggests that EVs alone may not be sufficient to drive MesoMT. Rather, these EVs may prime the cells within the microenvironment to amplify their contribution to PF. Our current focus is to identify the specific cargo molecules responsible for the observed functional effects mediated by exudative EVs as well as FXa- and thrombin-generated EVs.

Our study addresses a critical knowledge gap in understanding the pathogenesis of PF, specifically the role of EVs in neoangiogenesis and profibrotic signaling. Traditionally, the analysis of pleural effusions has focused on biochemical, cytological, and microbiological aspects. Although these approaches provide valuable diagnostic information, many pleural effusions lack a clear etiology (20, 69, 70). Integrating EV analysis alongside conventional techniques could potentially enhance diagnostic precision and facilitate development of tailored management strategies for patients with fibrosing pleural injury. Our ongoing research seeks to identify the specific cargo and signaling pathways regulated by EVs in PF and to translate these findings into clinically tractable therapies for the treatment of PF.