Alk5/Runx1 signaling mediated by extracellular vesicles promotes vascular repair in acute respiratory distress syndrome

Data and sample collection was done following the Rush University Medical Center (RUMC) Institutional Review Board (IRB), using an approved protocol (IRB# 14030705-IRB01) for investigational use of un-used diseased blood samples, drawn for routine medical care. Blood was collected from 33 patients admitted to RUMC ICU; all patients included, 18 years and older, were identified within 24 h of diagnosis and met moderate-to-severe ARDS criteria per “The Berlin Definition of ARDS” [13]. Age < 18 years, patients with isolated left heart failure and active malignancy and patients who received immunosuppressant or chemotherapy during ARDS hospitalization were excluded. Age, sex, race, Acute Physiology and Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology Score II (SAPS II), Sequential Organ Failure Assessment (SOFA) score, Lung Injury Score (LIS), P/F (PaO₂/FiO₂) ratio on the day of ARDS diagnosis, cause of ARDS, ventilator-free days, extracorporeal membrane oxygenation (ECMO), and length of stay were collected. Mortality from all causes was recorded at day 100. A detailed clinical data set is included in Additional file 1: Table S1.

Pathological slides (paraffin-embedded lung tissue) of five ARDS subjects and three non-disease controls (ND-Ctrl) identified from autopsy files were provided by the Department of Pathology, RUMC. Frozen lung tissue, normal and ARDS, was obtained from the National Disease Research Interchange. Clinical diagnosis, underlying conditions, and other pertinent clinical and laboratory data were reviewed.

CD1 male mice, 6–8 weeks old, 20–25 g weight, from Jackson Laboratory (Bar Harbor, ME), kept under standardized housing and feeding conditions were used. All mouse studies were approved and performed under the guidelines of RUMC Institutional Animal Care and Use Committee. The experiments were done under anesthesia [ketamine (60 mg/kg), acepromazine (2.5 mg/kg) and xylazine (2.5 mg/kg)] in 0.1–0.2 ml phosphate buffered saline (PBS)]. Three to five mice per experimental condition (wt-mice, LPS- and LPS ± EVs-treated mice) were used; all experiments were repeated at least three times. No mouse mortality occurred during the study.

Mouse lungs were inflated with 1% low-melting-point agarose in 10% formalin at a constant pressure of 25 cm H₂O, allowing for homogenous expansion of lung parenchyma, and then fixed in 4% paraformaldehyde for 48 h and paraffin-embedded [12]. Thin sections (4–5 μm), cut longitudinally, were stained with hematoxylin/eosin (H&E). Images were acquired with a 20× lens using a Zeiss AxioImager M1 motorized upright microscope equipped with AxioCam ICc1 R3 RGB color digital camera (Carl Zeiss MicroImaging, Inc., Thornwood, NY). Quantification of perivascular cuffing area was performed on small and medium-sized (20 μm ≥ diameter ≤ 100 μm) blood vessels using the NIH ImageJ software version 1.8.0_112 as described previously [14]. A minimum of 25 vessels per section was used (three sections/mouse, 3–5 mice/experimental condition). All experiments were performed at least three times with reproducible results.

IHC on paraffin-embedded human lung tissue sections was performed using the Prestige ITSN-1 Ab (C-terminal epitope; the only commercially available Ab efficient in IHC), CD31 Ab and Ki67 Abs. All were followed by the appropriate Alexa Fluor 488- or Alexa Fluor 594-conjugated reporters as previously described [12]. CD31 and Ki67 Abs were used at 1:200 dilution in 0.1% BSA in PBS, whereas ITSN-1 and was used at 1:100 dilution.

EVs were isolated from the blood of ARDS patients (EV_ARDS) and control healthy subjects (EV_Ctrl). Whole blood was subjected to a first centrifugation [1.5×g; (3000 rpm) for 15 min, at 4 °C; Eppendorf microfuge, 5702R], to obtain the platelet-free plasma, that was then subjected to ultracentrifugation (Beckman Coulter, Optima™ MAX-XP, TLA-55 fixed angle rotor, 45-degree angle) at 79,700×g; (36,000 rpm), for 2 h, at 4 °C [15]. The EVs pellet was washed (3 × 10 min) in sterile PBS, subjected to ultracentrifugation as above and resuspended in 200 μl sterile PBS. To minimize potential changes, if any, in EVs’ stability, only freshly prepared EVs were used for functional and morphological studies. EVs stored in liquid N₂ were used only for biochemical studies. To achieve high scientific rigor regarding reproducibility, the EVs were standardized according to the guidelines established by the International Society of Extracellular Vesicles for EVs isolation and analyses [16, 17]. We performed the following: (i) use of plasma for EVs retrieval; (ii) venipuncture and 0.109 M Na citrate as coagulant; (iii) room temperature (RT) for blood storage before first centrifugation; (iv) EVs isolation within 2 h of blood collection; (v) ultracentrifugation and immunobead magnetic separation for EVs isolation/enrichment, (vi) flow cytometry and calibrated/counted beads to establish the EVs’ cellular origin and counts, (vii) collection of blood at the same time of the circadian day. Other key standardization factors specific to this study are: (i) constant blood volume to isolate EVs, (ii) constant 200 μl PBS to resuspend the EVs pellet, (iii) establish the Runx1p66/p52 ratio in the EVs isolated from the blood samples collected in week 2 of ICU stay (days 7–14), when the patients are in the sub-acute, proliferative phase of the disease, characterized by repair of the damaged alveoli and restoration of the barrier function [18].

ECs, passages 3–5, were grown in Endothelial Basal Medium-2 and medium 199-supplemented with 20% fetal bovine serum (FBS) as previously described [19]. To mimic the inflammatory EC dysfunction, the ECs were treated with 1 μg/ml LPS for 6 h (EC_LPS). Following LPS treatment, EC_LPS were exposed to three doses of EVs for 30 h, with 1 μg/ml LPS still present in the growth media.

Cells were used for the experimental protocols between passages 2 and 4. The MSCs were cultured in α-Minimum Essential Medium without ribonucleosides and deoxyribonucleosides containing 2 mM l-glutamine, 10% FBS (Atlanta Biochemicals, Inc., Flowery Branch, GA), 100 units/ml penicillin and 100 mg/ml streptomycin (Thermo Fisher Scientific, Hanover Park IL), in a humidified incubator at 5% CO₂ and 37 °C under sterile conditions [20, 21].

EV_MSC were obtained from the conditioned medium of MSCs at different time points of culture. MSCs were incubated for 24 h in medium depleted of FBS-derived EVs [22]. The conditioned was centrifuged at 1.5×g; (3000 rpm), for 15 min to remove cellular debris, then at 79,700 × g; (28,000 rpm), Beckman Coulter XL-90 ultracentrifuge, 70 Ti rotor, for 2 h at 4 °C. EV_MSC pellet was washed in PBS and subjected to a second ultracentrifugation. The EV_MSC were resuspended in sterile PBS according to the count of MSCs, usually 500 μl sterile PBS for 2 × 10⁶ cells. EV_MSC were lysed for 1 h, at 4 °C in 50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 1% NP-40 and protease inhibitor cocktail, used as per manufacturer’s instructions. The protein content of the EV_MSC was quantified by the BCA assay with bovine serum albumin as standard.

EV_ARDS and EV_Ctrl were lysed, and protein concentration was determined as described for EV_MSC. EVs lysates (individual samples) were analyzed for Alk5, TβRII and Runx1 protein content by WB via Alk5 (1:500), TβRII (1:500), Runx-1, CD9, CD81, CD63, syntenin-1, mitofilin (1:1000) Abs followed by the appropriate horseradish peroxidase-conjugated reporters [15]. WB analyses of EV_ARDS and EV_Ctrl lysates were normalized to equivalent μl blood or the total protein, as specified in the text. Actin cannot be used as loading control, as there is no evidence that EVs contain actin.

Total number/percentages of EVs were determined by standardized flow cytometry (Beckman Coulter Gallios flow cytometer with Kaluza G software for acquisition and Kaluza1.3 for analysis; [15, 25]), where the number of EVs was correlated to a fixed number of counting beads. EVs gating was accomplished by preliminary standardization experiments using Spherotech nano fluorescent size standard beads. Data are presented as dot plot and histograms and the results of data analysis as average % of total gated events (at least 10,000 events/sample) ± SEM. Percentage labeled EVs was determined by comparing with unstained control EVs, with gates set to exclude beads, background as determined by a buffer alone sample, as well as aggregates. Isotype-matched Abs served as control.

Immunobead magnetic separation of Alk5-positive EVs was achieved in three steps as follows: EV_ARDS were labeled with Alk5 polyclonal Ab [15]; then a biotin-conjugated IgG secondary Ab was used as a bridge to allow the binding of the MagSi-STA 1.0 magnetic silica beads to the EV_ARDS. All three steps were followed by successive washings in PBS followed by ultracentrifugation as above. Isolation of CD105-positive EVs was achieved via biotinylated Abs using a similar approach. Aliquots of EVs bound to the magnetic beads were analyzed by SDS PAGE and WB for Alk5, Runx1, CD105, CD34 and CD45 immunoreactivity. EV_ARDS preparations were normalized to the blood volume used for isolation. Incubations of the EVs with the biotin-conjugated IgG secondary Ab and the MagSi-STA 1.0 magnetic silica beads by omitting the first Ab were used for controls.

EVs were labeled with biotin/neutrAvidin–Alexa Fluor 594 or double labeled with Alk5 rabbit Ab/anti-rabbit IgG Alexa Fluor 488 and biotin/neutrAvidin–Alexa Fluor 594 reporters [15]. Final pellets were resuspended in PBS and fixed in 1% paraformaldehyde; aliquots were mounted on glass slides with Prolong Antifade Reagent. Isotype-matched IgG was used as a control. EVs were examined and photographed using a Zeiss AxioImager M1 microscope.

The human analysis used a Kaplan–Meier test (SPSS software, Version 22), to determine if the survival is longer among the Runx1p66 immunoreactive subjects. An effect size of d = 1.20 was obtained [28], which may be a little optimistic. Thus, assuming a more conservative effect of d = 0.80 (Cohen’s estimation of a ‘large’ effect), for our sample size of 29 ARDS patients (four patients out of the total of 33, were used only for flow cytometry; no Runx1 expression pattern was analyzed), a one-tailed alpha of 0.05, a power of 0.97 are obtained.

To look for differences in the two groups as shown in Table 3, we used χ² test for differences in mortality and sex and utilized student t test to look for differences in age, P/F ratios, APACHE II, SAPS II, SOFA score, and LIS.

As chronic deficiency of ITSN in a murine model of acute lung injury (ALI)/ARDS triggers a repair process characterized by EC proliferation and vascular remodeling [12], we investigated whether ITSN deficiency is a feature of ARDS human lung tissue. ITSN/CD31 IHC was used to study the cellular distribution of ITSN in the lung tissue of ARDS and ND-Ctrl specimens. ITSN/Alexa Fluor 594 staining is limited in the ARDS specimen (55 years old, male; pneumonia; Fig. 1a, inset, arrows). The typical ARDS morphology with tissue congestion, inflammatory infiltrates, and edema (Fig. 1a, arrowheads) is noticeable. The ND-Ctrl shows significant ITSN immunoreactivity and co-localization with the EC marker CD31 (Fig. 1b, inset). For biochemical evaluation of ITSN expression, total protein was extracted from frozen human ND-Ctrl lung (62 years old male, myocardial infarction) and ARDS tissue (36 years old male, pneumonia), from three different locations. WB followed by densitometry indicate that expression of ITSN protein is decreased 40–70% in ARDS samples compared to ND-Ctrl (Fig. 1c). This finding is highly relevant to ECs of the lung, as ECs represent about 50% of the lung tissue [29].

As ECs proliferation and increased microvascular density were critical for the repair process of murine lung injury [12], we next used lung autopsy slides of 5 ARDS patients who survived 6–24 days in ICU and subjected them to dual Ki67 (proliferation marker)/CD31 IHC to investigate the extent of EC proliferation. ARDS specimens show limited ECs proliferation (Fig. 1e, f), most likely the cause of the poor lung architecture and increased fibrosis in these patients. Ki67/CD31 co-localization on a lung section of a ND-Ctrl specimen is shown for comparison (Fig. 1d, arrows).

EV_ARDS and EV_Ctrl were isolated from the blood of ARDS and control subjects within less than 2 h of blood withdrawal. Accurate data for EVs concentration were obtained by standardized flow cytometry [25]. EVs’ count indicated mean values of 13,680 EV_ARDS/1 μl blood and 8615 EV_Ctrl/1 μl blood, Fig. 2a. The EVs were also quantified by comparing the total protein amounts in the EV_ARDS and EV_Ctrl pellets. The protein content of EV_ARDS is ~ 2-fold higher compared to the EV_Ctrl, Fig. 2b, an expected higher ratio given the aggregates content, excluded from the gated flow cytometry count. This ratio was recorded regardless of the timing of blood collection from all ARDS patients studied, consistent with reports of elevated levels of circulating EVs in disease settings [9, 11, 30]. As ARDS lung tissue is deficient of ITSN, we investigated whether EV_ARDS contain the TβRI/Alk5, similar to the EVs released in the circulation of ITSN-deficient mice. EV_ARDS and EV_Ctrl lysates were analyzed for their Alk5 content by WB, Fig. 2c. Alk5 was found in both EV_Ctrl and EV_ARDS. However, as indicated by densitometric analyses of representative blots, Fig. 2f, Alk5 expression was sevenfold greater in EV_ARDS compared to EV_Ctrl, suggestive of a greater expression of Alk5 in EV_ARDS compared to the EV_Ctrl. Since TβRI signals through a heteromeric TβRI/TβRII complex and since the EVs of ITSN-deficient mice were immunoreactive for the TβRII [15], we also analyzed the EV_ARDS for their TβRII content by WB. EV_ARDS were immunoreactive to the TβRII Ab, Fig. 2d, suggesting that the human EVs, bear the Alk5/TβRII heteromers. Furthermore, time course analyses of Alk5 protein expression in the EV_ARDS from different ARDS subjects at several time points of ICU stay indicate no significant changes in Alk5 from day 5 to day 28 of ICU stay, Fig. 2e, the time interval used to collect blood and isolate EV_ARDS.

As the size ranges of EVs, from exosomes (diameter 40–120 nm) to microparticles, diameter 200–1000 nm [31], cannot be considered absolute and given the unfeasibility to isolate a pure microparticle fraction, exclusively [16], we also evaluated the exosomal content by NTA approach (Fig. 2g). No discernible peak was detected for particle sizes of 40–120 nm. We also evaluated the purity of the isolated EVs fractions by WB analyses using specific Abs against CD9, CD81, syntenin-1 and CD63, reported to be enriched in exosomes, and actinin-1 and mitofilin enriched in non-exosomal EVs populations [22]. As shown in Fig. 2h, the EVs fractions of three different ARDS patients showed scarcely detectable immunoreactivity for exosomal markers and strong immunoreactivity for the non-exosomal EVs, consistent with an enriched non-exosomal fraction of EV_ARDS. Thus, the isolation procedure applied indicates that the exosome and microparticle fractions are distinct in both size and biochemical makeup.

It is widely accepted that it is difficult to distinguish between two subpopulations of EVs based only on the size and protein composition [22]. Since a precise definition for exosomes and microparticles remains to be resolved, we refer to the population of membrane vesicles isolated according to our experimental protocol (79,700 × g, 2 h), as EVs.

Labeling of EV_ARDS using allophycocyanin-Alk5, phycoerythrin- and pacific blue-conjugated cell surface markers, followed by flow cytometry analyses of the mean values/percentages recorded, indicate: (i) a high content (> 50%) of EV_ARDS immunoreactive to Alk5 Ab, (ii) immunoreactivity of EV_ARDSAlk5 to CD105 and CD73 Abs, suggestive of MSC-origin, (iii) 4.7-fold increase in the count of EV_ARDS of MSC-origin compared to EV_Ctrl, as determined by tri-color labeling of EVs (Table 1). For data analysis, scatter and fluorescence noise was determined by running double filtered buffer alone, and a gate was drawn over the background noise to exclude those events (Fig. 3a, Gate A). Representative forward (FS) and side scatter (SS) plot for unstained EV_ARDS and background fluorescence in PE and APC channels for the unstained samples are shown as dot plot (b) and individual histograms (c, d). Samples of isolated EVs (Gate D) and 2 μm beads (Gate B) were run, to determine approximate sizing of the EVs relative to the beads as well as to determine background fluorescence (Fig. 3e, i). The majority of EVs were found to be around 1 μm in size (Gate D). Next, EV_ARDS stained with Alk5-APC/CD105-PE Abs were analyzed by flow cytometry to evaluate the co-expression of Alk5 and CD105 on the EV_ARDS (Fig. 3e, Gate D). A representative plot of double stained EV_ARDS (Fig. 3f) and the individual histograms for PE-CD105 (Fig. 3g) and APC-Alk5 (Fig. 3h) are shown. Similarly, EV_ARDS co-stained with Alk5-APC/CD73-PE Abs were analyzed by flow cytometry to evaluate the co-expression of Alk5 and CD73 on the EV_ARDS (Fig. 3i, Gate D). A representative flow cytometry plot of Alk5-APC/CD73-PE-labelled EV_ARDS is shown in Fig. 3j. The individual histograms for CD73-PE (Fig. 3k) and Alk5-APC (Fig. 3l) are shown.

MSCs origin of EV_ARDS was also confirmed by no detectable immunoreactivity to the hematopoietic markers CD45/CD34 [32], as shown by magnetic bead immunoisolation of CD105-positive EV_ARDS followed by WB (Fig. 3m). As the cellular origin of EVs is a defining factor for the magnitude of the physiological effect elicited [33], and considering the 4.7-fold increase in the count of the EV_ARDS of MSC origin compared to controls, the most significant increase compared to the EV_ARDS of other cellular origins identified, we focused on this disease specific subset of EV_ARDS.

Given the involvement of Runx1 transcription factor in angiogenesis and MSCs proliferation [34] and our findings that lung tissue of non-surviving ARDS patients shows deficient EC proliferation, we investigated whether the EV_ARDS contain Runx1 by WB analysis. Two Runx1 isoforms were detected—a 52-kDa protein present in all EV_ARDS as well as in EV_Ctrl, and a less characterized 66-kDa isoform [35], present only in some EV_ARDS samples, Fig. 4a. This observation is of particular interest, as out of 29 samples analyzed, 20 samples show Runx1p66 immunoreactivity; 14 out of these 20 EV_ARDS samples also display a higher than 0.5 Runx1p66/p52 ratio and belong to ARDS patients still alive, Table 2. Three samples (S5, S14, S35) belong to ARDS patients that died despite a high Runx1p66/p52 ratio (> 0.5) due to special circumstances, as described in Table 2 (legend), while the remaining three samples (S11, S12, S15) have a low Runx1p66/p52 ratio (< 0.5) and also belong to non-surviving ARDS patients. Other 4 EV_ARDS samples (S7, S8, S20, and S22) with no Runx1p66 immunoreactivity belong to non-surviving ARDS patients, as well. Notably, the last 5 EV_ARDS samples belong to long-term ARDS survivors that spent on average only 12.4 days in ICU; these EV_ARDS show no Runx1p66, Table 2, raising the possibility that the circulatory EV_ARDSRunx1p66 might be needed only when the endogenous lung MSCs [36, 37] and thus, the ability of lung tissue to repair, a hallmark of the normal course of recovery from ALI in some patients [38], are not sufficient. The ratio Runx1p66/p52 was established in all cases in EV_ARDS isolated from blood samples collected in week 2 of ICU stay, a key standardization factor for patients’ disease stage. Histologically, days 7–14 are recognized as the sub-acute, proliferative ARDS phase, characterized by lung cell proliferation and attempts of repairing the damaged alveoli and restoration of the barrier function [18]. These observations strongly suggest that expression of Runx1p66 and a high Runx1p66/p52 ratio provide a survival benefit to ARDS patients. It seems that despite the heterogeneous etiology of ARDS and various treatment/medication received, concurrent illness, mechanical ventilation, active inflammation, etc., the transient expression of Runx1p66 and a high Runx1p66/p52 ratio are part of a common pathway of lung response leading to ARDS resolution.

We also performed a time course analyses of Runx1 expression in EV_ARDS of long-term ARDS survivors, Fig. 4b, and non-survivors, Fig. 4c; Runx1p52 is detected at each time point analyzed, while Runx1p66 is detected only at day 14 in ICU for the ARDS survivor, in the case shown. This transient expression of Runx1p66 was also detected in cultured human bone marrow-derived MSCs, Fig. 4d. Briefly, we analyzed cultures of human bone marrow-derived MSCs, passage 2–3, for their Runx1 expression, each day starting at about 5% confluence (day 2) and after the cultures had expanded to about 80% confluence (day 8). Runx1p52 was expressed starting at day 2 at each time point analyzed, while Runx1p66 was transiently expressed (days 6 and 7), and released in the EVs recovered from the growth medium.

We then analyzed the ARDS patients and their EV_ARDS′ Runx1p66 and p52 isoform content regarding to the severity of the disease, demographics and mortality, Table 3. Only 3/17 patients with a high ratio (> 0.5) died as compared to 7/12 patients who had a low ratio (< 0.5) or Runx1p66 negative. This difference in mortality was statistically significant by Chi square test (χ² = 5.15, p = 0.023). The likelihood ratio is 5.218. Furthermore, we compared these two groups for differences in age, sex, SAPS II score, APACHE II score, SOFA score, LIS and P/F ratio at the onset of ARDS and ICU length of stay. None of these were statistically significant, but patients with high Runx1p66/p52 ratio tended to be younger (mean age 41.76 ± 6.9 vs. 50.25 ± 9.4). Despite the relatively small sample size of this study, there is no question that the survival/censoring times for participants with this Runx1p66 isoform was much larger than it was for those without it; the patients are censored as the total survival time cannot be accurately determined for all participants [39]. Combining mortality and censored times, the mean time for the Runx1p66 group is 32.10 days (sd = 24.35) while the mean time for the Runx1 p66-negative group was 12.89 days (sd = 6.37). A simple, heterogeneous t-test between the two groups is statistically significant (t = 3.29, df = 23.90, p = 0.003).

The differences between the two conditions are also clearly seen in a life table, Fig. 4e. While the Kaplan–Meier statistic for this table is not quite significant (χ² = 3.65, df = 1, p = 0.056), it is also clear that the improved mortality for the Runx1p66 group begins at about 15 days, and then persists from then on. Another remarkable finding was that 100% of the high p66/p52 ratio sample was still alive at day 100 (not plotted), Additional file 1: Table S1. Thus, correlation of EV_ARDS immunoreactivity for Runx1 with survival data indicates that the transient expression of Runx1p66 and a high Runx1p66/p52 ratio provide a survival benefit to ARDS patients.

We used magnetic separation of Alk5-positive EV_ARDS via MagSi-streptavidin beads to address whether Alk5 and Runx1 reside on the same EV_ARDS population. Briefly, Alk5-positive EV_ARDS were labeled with an anti-Alk5 goat Ab, followed by a biotinylated rabbit anti-goat IgG and MagSi-streptavidin beads. WB analyses of bound and unbound EV_ARDS indicated that all EV_ARDS immunoreactive to Alk5 possess Runx1 (Fig. 4f), suggesting that a significant sub-set of EV_ARDS comprises both Alk5/TβRI receptor and Runx1 transcription factor. However, a small fraction of Runx1-positive EV_ARDS does not contain Alk5, Fig. 4f (unbound fractions).

First, EV_ARDS were visualized by biotin/neutrAvidin–Alexa Fluor 594 labeling, as recently described by us for the EVs isolated from the blood of ITSN-deficient mice [15]. Similar to the mouse EVs, fluorescently labeled human EVs show a continuous, donut-shape (Fig. 5a, arrows; b, inset). As human ARDS lung tissue is deficient in EC proliferation, we next evaluated the biological activity of EV_ARDS on cultured human lung ECs proliferation. To mimic the inflammatory EC dysfunction, the ECs were treated with 1 μg/ml LPS for 6 h (EC_LPS), and then exposed to three doses of EV_ARDSAlk5Runx1p66 or EV_ARDSAlk5Runx1p52 [0.8 × 10³, 1.6 × 10³ (the dose shown in Fig. 5e, f) and 3.2 × 10³ per 1 ml growth medium] for 30 h, followed by BrdU cell proliferation assay [24]. ECs with no LPS and no EV_ARDS exposure (EC_Ctrl, Fig. 5c), EC_LPS (Fig. 5d) and EC_LPS exposed to EV_Ctrl at 2:1 (EV_ARDS:EV_Ctrl) ratio, according to the in vivo situation (not shown), were used for comparison. Morphometric analyses of the BrdU-positive ECs nuclei (Fig. 5g), indicated that 1 μg/ml LPS for 6 h minimally inhibits ECs proliferation (less than 10%). Exposure of EC_LPS to 1.6 × 10³ EV_ARDSAlk5Runx1p66 caused about 38% increase in the BrdU-positive cells compared to EC_LPS. The EV_ARDSAlk5Runx1p52 did not induce EC_LPS proliferation, regardless of the dose used. Exposure of EC_LPS to EV_Ctrl (2:1 ratio EV_ARDS;EV_Ctrl) increased the number of BrdU-positive cells less than 10%. The proliferative effects of EV_ARDSAlk5Runx1p66 at a lower or, the higher doses used were less evident. Furthermore, the MTT assay [24] to evaluate the effects of the EV_ARDS on LPS-injured ECs (Fig. 5h) biochemically, indicated that 1.6 × 10³ EV_ARDSAlk5Runx1p66/1 ml growth medium at day 3 post-exposure, increase the number of EC_LPS by 44%, compared to EC_Ctrl (Fig. 5h). EC_LPS exposed to EV_Ctrl show similar proliferation as EC_LPS, both slightly lower than EC_Ctrl. EC_LPS exposed to 1.6 × 10³ EV_ARDSAlk5Runx1p52/1 ml growth medium showed some variability in their proliferative effects depending on the EV_ARDS sample used; on average however, the EV_ARDSAlk5Runx1p52 slightly decreased, by less than 8%, the number of the BrdU-positive cells compared to EC_LPS (Fig. 5h). Based on these data, we concluded that the under inflammatory conditions, EV_ARDSAlk5 Runx1p66 stimulate ECs proliferation.

The proof-of-concept that Runx1p66 isoform accounts for EC proliferation was obtained when EC_LPS were exposed to EV_MSC, isolated from the conditioned media of cultured MSCs (passage 3–4), as described under Methods. EV_MSC were isolated at day 7, when Runx1p66 immunoreactivity is detected and at days 5 and 8, with no Runx1p66 is detected; two EV_MSC doses were used: 5 μg/ml (low dose) and 10 μg/ml (high dose). The quantitative assessment indicated that 10 μg/ml EV_MSC increased proliferation of EC_LPS at day 7 by 40%, compared to EC_Ctrl, Fig. 5I (bar d vs. bar a).

Minimal EC proliferation, with no statistical significance was noted at days 5 and 8, for the high EV_MSC dose. No EC proliferation was noticed for the low EV_MSC dose, regardless of Runx1p66 content; the finding is consistent with an EV_MSC basal threshold for inducing proliferation, as already reported by us [15] and with the beneficial effects of the high Runx1p52/p66 ratio. The data support the causal impact of Runx1p66 expression on EC proliferation.

As the integrity of the interendothelial junctions (IEJs) is vital for barrier function and edema resolution, we investigated the biological activity of freshly isolated EV_ARDS in mitigating the LPS effects on IEJs. Cultured ECs treated with 1 μg/ml LPS for 6 h, as above followed by 30 h exposure to three doses (0.8 × 10³, 1.6 × 10³ and 3.2 × 10³ per 1 ml growth medium) of EV_ARDSAlk5Runx1p66, were subjected to VE-cadherin Ab/Alexa Fluor488 immunocytochemistry. The integrity of IEJs in EC_LPS treated with 1.6 × 10³ EV_ARDSAlk5Runx1p66 (Fig. 6d) was improved compared to EC_LPS with no EV_ARDSAlk5Runx1p66 exposure (Fig. 6b, c) or exposed to an equal dose of EV_ARDSAlk5Runx1p52 of a non-surviving ARDS patient (Fig. 6e). Figure 6a shows for comparison the IEJs in EC_Ctrl. Interendothelial gaps are significantly larger at 30 h post-LPS exposure without any EV_ARDS treatment (Fig. 6c) compared to 6 h LPS time point (Fig. 6b). Quantitative assessment of intercellular gaps (NIH ImageJ), shows that EV_ARDS bearing Runx1p66 reduced significantly, more than 60% the interendothelial gap area in LPS-treated ECs, while the EV_ARDS with no Runx1p66 by only 27%, Fig. 6f. The surface of the interendothelial gaps at 30 h post-LPS is 40% greater compared to 6 h LPS time point with no EV_ARDS exposure. The effects of the lower and the higher EV_ARDS doses were less noticeable.

Lung injury was induced in mice by intraperitoneal delivery of 3.5 mg/kg LPS (sub-lethal dose) as described in [6]; 8 h post-LPS treatment, mice were injected retro-orbitally with two EV_ARDS doses: 2.9 × 10⁵ and 5.8 × 10⁵ (dose shown in Fig. 7) in 100 μl sterile PBS. Assessment of lung pathology on H&E stained mouse lung sections, 48 h after EV_ARDSAlk5Runx1p66 administration shows reduction of inflammatory cells, decreased perivascular cuffing and less thickening of the interstitium (Fig. 7e) compared to LPS-only treated mice (Fig. 7b–d). Lung histological severity is not improved by EV_ARDSAlk5Runx1p52 (EV_ARDS of the same patient in week 1 in ICU, when no Runx1p66 immunoreactivity is detected), Fig. 7f, or EV_ARDS a non-surviving patient (week 2 ICU), Fig. 7g. The 48 h time point post-EV_ARDS has been selected as mice need at least 72 h for recovery after sub-lethal LPS, with no other treatment [40]. The benefit of the low EV_ARDSAlk5Runx1p66 dose was less evident (not shown). The lung histology of mice treated with LPS only and mice treated with LPS followed by EV_Ctrl display no detectable differences at this time point (not shown). Mice injected with PBS served as controls (Fig. 7a). Quantitative assessment of perivascular cuffs, sign of a compromised endothelial barrier indicates ~ 60% reduction in EV_ARDSAlk5Runx1p66-treated mice, when the high dose has been used, compared to LPS-treated mice (Fig. 7h). The beneficial effects of the low EV_ARDSAlk5Runx1p66 dose were less visible (Fig. 7h). Mice treated with EV_ARDSAlk5Runx1p52, regardless of the source of the EV_ARDS preparation—non-surviving ARDS subjects or long-term survivors—show no decline in the perivascular cuffing (Fig. 7h). These data strongly suggest that despite the complex biochemical makeup of EV_ARDS, Runx1p66 contributes significantly to improve the junctional integrity and reduce perivascular cuffing of LPS-injured ECs.