Extracellular Vesicle ASC: A Novel Mediator for Lung–Brain Axis in Preterm Brain Injury

Abstract

Bronchopulmonary dysplasia (BPD) and neurodevelopmental impairment are among the most common morbidities affecting preterm infants. Although BPD is a predictor of poor neurodevelopmental outcomes, it is currently uncertain how BPD contributes to brain injury in preterm infants. Extracellular vesicles (EVs) are involved in interorgan communication in diverse pathological processes. ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain) is pivotal in inflammasome assembly and activation of inflammatory response. We assessed expression profiles of the alveolar macrophage (AM) markers CD11b, CD11c, and CD206 as well as ASC in EVs isolated from the plasma of preterm infants at risk for BPD at 1 week of age. We found that infants on higher fraction of inspired oxygen therapy (HO₂⩾30%) had increased concentrations of AM-derived EV-ASC compared with infants on lower fraction of inspired oxygen (LO₂<30%). To assess the function of these EVs, we performed adoptive transfer experiments by injecting them into the circulation of newborn mice. We discovered that mice that received EVs from infants on HO₂ had increased lung inflammation, decreased alveolarization, and disrupted vascular development, the hallmarks of BPD. Importantly, these EVs crossed the blood–brain barrier, and the EVs from infants on HO₂ caused inflammation, reduced cell survival, and increased cell death, with features of pyroptosis and necroptosis in the hippocampus. These results highlight a novel role for AM-derived EV-ASC in mediating the lung-to-brain cross-talk that is critical in the pathogenesis of BPD and brain injury and identify potential novel targets for preventing and treating BPD and brain injury in preterm infants.

Clinical Relevance

Our study suggests a critical role of extracellular vesicle ASC in mediating the lung–brain axis, which is pivotal in preterm brain injury. Targeting extracellular vesicle ASC may present a novel approach for preventing and treating bronchopulmonary dysplasia–associated brain injury in preterm infants.

More than 15 million infants are born preterm worldwide each year (1, 2). Extremely premature infants born at less than 28 weeks of gestational age (GA) are at great risk of having multiorgan injury predominantly involving the lungs and brain (3–5). Born with immature lungs, these premature infants experience respiratory failure soon after birth, and they often require supplemental oxygen and mechanical ventilation to survive. However, this life-sustaining high fraction of inspired oxygen (FiO₂) can cause lung inflammation that ultimately leads to bronchopulmonary dysplasia (BPD), characterized by disrupted alveolar and vascular development and reduced lung function (3, 4). The immature brain in these premature infants is also affected by oxygen therapy, resulting in inflammation and leading to short- and long-term neurodevelopmental sequelae such as intraventricular hemorrhage, encephalopathy of prematurity, cerebral palsy, intellectual disability, and cognitive deficits (5, 6). Preterm infants with BPD develop not only long-term lung disease but also long-term sequelae involving the brain, leading to neurodevelopmental impairment (NDI). Moreover, there is mounting clinical evidence that severe BPD is an independent risk factor for adverse neurodevelopmental outcomes, even without catastrophic brain injury (6–8). However, the mechanisms by which BPD contributes to brain injury and NDI remain to be elucidated.

Extracellular vesicles (EVs) are lipid membrane–encircled vesicles secreted by cells into the extracellular environment (9–12). EVs carry complex cargoes of proteins, lipids, and nucleic acids, and their cargo composition is highly dependent on the biological function of the parental cells. Being membranous, EVs protect their cargo from the extracellular environment, allowing safe transport and delivery of their intact cargo to target cells. This results in the modification of the target cells’ signaling pathways and overall function. In the lung, both alveolar epithelial cells (AECs) and alveolar macrophages (AMs) can release bioactive EVs upon inflammatory injury. AEC- and AM-derived EVs isolated from BAL fluid have been shown to regulate inflammatory responses in adult lung diseases (13–18). Furthermore, increased numbers of AEC-derived EVs were detected in tracheal aspirate fluids from preterm infants with severe BPD compared with infants without BPD (19). Similarly, brain cell–derived EVs have the potential to serve as biomarkers and therapeutic agents for perinatal brain injury (20). We recently found that neonatal rats exposed to 85% O₂ have increased AEC-derived EVs in their circulation. Importantly, we showed that when adoptively transferred into the circulation of normal newborn rats, these EVs could cross the blood–brain barrier (BBB), be taken up by brain tissues, and induce neural cell death and brain injury, suggesting the existence of lung-to-brain cross-talk (21).

Inflammasomes are multiprotein complexes that mediate proteolytic cleavage and activation of GSDMD (gasdermin D), pro–IL-1β, and pro–IL-18 by caspase-1 (22–25). ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain) is pivotal in inflammasome assembly and activation of caspase-1 (26, 27). Cleavage of GSDMD by caspase-1 releases a 30-kD N-terminal domain, GSDMD-p30, which oligomerizes in the cell membrane to form pores that cause pyroptosis. In addition, the GSDMD pores allow rapid release of active IL-1β and IL-18, resulting in secondary inflammation. Growing evidence shows activation of the inflammasome cascade in the lung and brain secondary to oxygen exposure (21, 27, 28). Previous studies have demonstrated early activation of the NLRP3 inflammasome with an increased IL-1β:IL-1ra ratio is a key mechanism in developing BPD (29, 30). Recently published studies from our laboratory have highlighted the crucial role of GSDMD and circulating EVs in hyperoxia-induced neonatal lung and brain injury (21). However, whether circulating EVs have increased cargo of ASC in preterm infants at risk for BPD and the functions of these EVs in mediating lung-to-brain cross-talk leading to brain injury in preterm infants are unknown.

The aim of this study was to determine the expression profile of AM-derived EV-ASC in the circulation of preterm infants at 1 week of age who are at risk for developing BPD and to investigate the effects of these EVs on lung and brain injury by adoptively transferring them into newborn mice. We found that plasma EVs from preterm infants who were on higher FiO₂ (HO₂) had higher concentrations of AM-derived EV-ASC than infants who were on lower FiO₂ (LO₂). Furthermore, the adoptive transfer of EVs from infants on HO₂ into the circulation of newborn mice induced BPD-like pathology and brain injury. Our results help fill a gap not only in understanding the novel role of EV-ASC in mediating the lung-to-brain cross-talk that is critical in the pathogenesis of BPD and brain injury but also in identifying potential novel targets for preventing and treating BPD and brain injury in preterm infants.

Methods

See the data supplement for material lists.

Human Study Design

The Institutional Review Boards of Jackson Health System and the University of Miami Miller School of Medicine approved the study (protocol #20200736). This prospective, longitudinal, and observational study was conducted in the neonatal ICU (NICU) at the Holtz Children’s Hospital in the Jackson Health System and University of Miami Miller School of Medicine. The study population consisted of neonates admitted to our NICU between August 2021 and August 2022 with GA 23^0/7–29^6/7 weeks. Written informed parental consent was obtained at <7 days of age upon admission to the NICU. The exclusion criteria included major congenital anomalies. The patients’ demographic characteristics were prospectively collected and entered in the database, including GA, birth weight, antenatal steroid use, gender, Apgar scores, FiO₂ requirements at day of life (DOL) 1, DOL 7, and DOL 28, and BPD. BPD was defined by the bedside clinicians using the 2001 Eunice Kennedy Shriver National Institute of Child Health and Human Development workshop criteria (3). Mild BPD is defined as treatment with >21% O₂ for at least 28 days but breathing room air at 36 weeks’ postmenstrual age. Moderate BPD is defined as requiring <30% O₂, and severe BPD is defined as needing ⩾30% O₂ and/or positive pressure support at 36 weeks’ postmenstrual age.

Blood Sample Collection and EV Isolation

Blood specimens were obtained at 1 week of age, and the plasma was collected in EDTA tubes and stored at −80°C for future analyses. We randomly selected seven plasma samples from infants who were on <30% FiO₂ (LO₂), seven plasma samples from infants who were on ⩾30% FiO₂ (HO₂), and one adult human plasma sample (Sigma-Aldrich), and we isolated EVs from these samples using ExoQuick (System Biosciences) according to the manufacturer’s instruction. Briefly, 50 μl of plasma sample was centrifuged at 3,000 × g for 10 minutes to eliminate cells and cell debris. The supernatant was collected and mixed with an appropriate volume of precipitation solution in an Eppendorf tube for 30 minutes at 4°C. The mixes were centrifuged at 1,500 × g for 30 minutes, and the EV pellets were resuspended in 50 μl PBS. Protein concentration in each EV preparation was measured by BCA protein assay using a commercial kit from Pierce Biotechnology, Inc.

EV Characterization

Four microliters from each EV sample were analyzed for particle numbers and size distributions by nanoparticle tracking analysis using the NanoSight NS300 system (Malvern Instruments), as previously described (21). Nanoparticle concentrations were expressed per milligram of EV protein concentrations. For transmission EM (TEM) examination, EV suspensions were diluted 1:10 and then fixed in 4% paraformaldehyde for 1 hour. The samples were loaded onto a carbon copper grid for 30 minutes, rinsed in phosphate buffer and then double-distilled water, fixed with 2% glutaraldehyde, and stained with a 2% aqueous uranyl acetate solution. The grid was kept overnight protected from light and was viewed at 80 kV in a JEOL JEM-1400 TEM (JEOL USA, Inc.), and images were captured using a Gatan ES1000W digital camera. The average EV sizes were calculated from a minimum of 25 EVs from each EV preparation. For western blot analysis, total proteins (12 micro gram/sample) were separated by 4–20% SDS polyacrylamide gel electrophoresis (Bio-Rad) and probed with a CD9 antibody and then CD63 and CD81 antibodies, as we previously described (21).

Assessing AM-derived EVs

Flow cytometry was performed with double staining of the AM markers CD11b, CD11c, CD206 (31, 32) and ASC. Briefly, 5 μg/sample of plasma EVs were captured on antitetraspanin-conjugated (CD9, CD63, and CD81) magnetic beads (Exo-Flow Exosome Purification Kit, System Biosciences); stained sequentially with an anti-ASC antibody and a FITC-labeled secondary antibody, followed by a phycoerythrin-labeled anti-CD11b, anti-CD11c, or anti-CD206 antibody; and then analyzed using a flow cytometer (CytoFLEX; Beckman Coulter) (12, 21).

EV Depletion

To confirm that AMs produced plasma EVs in preterm infants, we performed an EV depletion assay by incubating the plasma EVs with Exo-Flow magnetic beads (System Biosciences) coupled to a cocktail of antibodies to macrophage surface markers (CD11b, CD11c, and CD206). The depleted EVs were then bound to Exo-Flow beads coupled to a cocktail of antibodies to universal EV membrane markers, the tetraspanins CD9, CD63, and CD81. These Exo-Flow bound EVs were then incubated with fluorescently labeled antibodies to CD11b, CD11c, and CD206 as well as ASC and analyzed using flow cytometry.

Animals

C57BL/6 mice were purchased from the Jackson Laboratory. The Animal Care and Use Committee of the University of Miami Miller School of Medicine approved the experimental protocol. All animals were cared for according to the National Institutes of Health guidelines for the use and care of animals. The study is reported in accordance with the Animal Research: Reporting of In Vivo Experiments guidelines.

In Vivo Tracking of EVs

To assess whether circulating EVs from preterm infants can cross the BBB, we labeled EVs (10 μg/sample) with ExoGlow-Vivo, a near-infrared dye (SBI) (21). Labeled EVs and sham-labeled PBS (negative control) were then adoptively transferred into normal newborn mice by retroorbital sinus injection at postnatal day 10 (P10). The lungs and brains were dissected at 2 hours for ex vivo imaging. Lung and brain tissues were fixed in 4% paraformaldehyde overnight, and immunostaining of these tissue sections was performed with a cocktail of antibodies to the pan-EV makers CD9, CD63, and CD81.

Adoptive Transfer EVs

We pooled seven EV samples from infants who were on LO₂ and seven samples from infants who were on HO₂ and diluted them with PBS to a concentration of 1 μg/μl for adoptive transfer experiments. Adult human plasma EVs were used as the control for monitoring nonspecific immune responses. Newborn mice from six litters were randomized on P3. They received EVs (10 μg/animal, 5 mg/kg mouse body weight, 0.1 mg/ml mouse blood volume) from infants in the LO₂ group, infants in the HO₂ group, and normal adult human via retroorbital sinus injection. These calculations were on the basis of the newborn mouse body weight of ∼2 g and blood volume of ∼0.1 ml (33). This dose was chosen on the basis of published studies using 0.2–15 mg/kg EV intravenous injection in adult mice (34–36). This dose was lower than the EV concentrations (0.6–4.5 mg/ml) we detected in our preterm infants. The mice grew in room air, had routine care, and were killed on P14 under anesthesia by 0.1% isoflurane.

Lung Tissue Section

Lung tissues were first pressure fixed by infusing 4% paraformaldehyde through a tracheal catheter at 20 cm H₂O for 5 minutes and then fixed overnight in 4% paraformaldehyde. The fixed lung tissues were embedded in paraffin wax, and 5-μm sections were prepared.

RNA Isolation and qRT-PCR

Total RNA was extracted from frozen lung tissues using the RNeasy Universal Mini Kit (Qiagen) according to the manufacturer’s instructions. RNA quality and integrity were verified using the Agilent 2100 Bioanalyzer (Agilent Technologies). All samples had RNA integrity numbers >7. Gene expressions of apoptosis-, pyroptosis-, and necroptosis-related molecules were determined using qRT-PCR, as previously described (21). The relative expression amounts of target genes were normalized to 18S rRNA.

Assessing ASC and GSDMD Expression in the Lung

To determine ASC and GSDMD expression, lung tissue sections were immunostained with an anti-ASC or an anti-GSDMD antibody (21).

Assessment of Lung Inflammation

Macrophage infiltration was determined by immunostaining using an anti-Mac3 antibody. The number of MAC-3–stained cells in the alveolar airspaces of lung tissue sections was counted from five random high-power views (HPVs) taken from the 20× objective on each slide (28, 37). Immunostaining for inducible nitric oxide synthase (iNOS; an M1 macrophage marker) and Ym1 (an M2 macrophage marker) were also performed to identify M1 and M2 macrophage phenotypes (38).

Lung Histology, Morphometry, and Vascularization Assessment

Hematoxylin and eosin–stained lung tissue sections were used for histology and morphometry (21). The mean linear intercept assessment was performed by a staff member unaware of the experimental condition, as previously described (37). Immunofluorescent staining for von Willebrand factor, an endothelial marker, was performed to determine pulmonary vascular density, which was quantified by the number of von Willebrand factor–positive vessels (<50 μm in diameter) per HPV in five randomly selected, nonoverlapping, parenchymal fields on lung sections from each animal.

Hippocampal Tissue Collection and Sectioning

Hippocampal tissues were collected from the left side of the brain and frozen at −80°C for RNA isolation. The right-side brain tissue was fixed in 10% formalin and paraffin embedded. Ten 10-μm coronal sections were processed, as previously described (39).

Assessment of ASC and GSDMD Expression in Hippocampal Tissues

Expression of ASC and GSDMD was determined using immunostaining with an anti-ASC or anti-GSDMD antibody on brain tissue sections. To determine specific brain cells that express ASC and GSDMD, double immunofluorescent staining was performed with antibodies for AIF1 (allograft inflammatory factor 1) (a microglial marker) (40), GFAP (glial fibrillary acidic protein) (an astrocyte marker) (41), SOX10 (an oligodendrocyte marker) (42), and an anti-ASC or anti-GSDMD antibody.

Assessment of Hippocampal Inflammation

Immunostaining with an antibody for AIF1 was performed to detect microglial infiltration. The number of AIF1-stained microglial cells was counted from three random HPVs taken from the 20× objective in the hippocampal sections on each slide (28). Gene expressions of inflammatory mediators were determined using qRT-PCR and normalized to 18S rRNA, as described above (21).

Assessment of Hippocampal Cell Proliferation and Death

To determine cell proliferation, immunofluorescent staining for Ki67, a nuclear proliferation marker, was performed. The proliferative index was calculated as the average percentage of Ki67-positive nuclei in total nuclei in five random HPVs on hippocampal sections from each animal (28, 39). A TUNEL assay was used to detect cell death. The cell death index was calculated as the average percentage of TUNEL-positive nuclei in total nuclei in five random HPVs on hippocampal sections from each animal (21, 28, 39). Gene expressions of apoptosis-, pyroptosis-, and necroptosis-related molecules were assessed using qRT-PCR, as described above. Double immunofluorescent staining was performed with an anti-ASC or anti-GSDMD antibody and an antibody for MLKL (mixed-lineage kinase domain-like) (a necroptosis mediator).

Data Management and Statistical Analysis

Data are expressed as mean ± SD. Comparisons between two human groups were performed using the Mann-Whitney test. ANOVA and Tukey post hoc analysis were used for three-group comparisons in animal studies. A P value ⩽0.05 was considered to indicate statistical significance.

Results

Clinical Characteristics

The clinical characteristics of the infants in this study (n = 14) are summarized in Table 1. The infants in the HO₂ group were born earlier and had lower birth weights compared with those in the LO₂ group. Each of these groups had six female and one male infant. There was no difference in antenatal corticosteroid therapy. The two groups of patients were on similar concentrations of FiO₂ on DOL 1, but the infants in the HO₂ group were on HO₂ therapy on DOL 7 and DOL 28 and eventually developed BPD.

Table 1.

Patient Characteristics and Respiratory Outcomes

Patient Demographics and Outcomes	LO2 (n = 7)	HO2 (n = 7)	P Value
GA, wk, mean ± SD	27.8 ± 1.3	24.3 ± 0.5	<0.0001
BW, g, mean ± SD	997.5 ± 200.9	576.7 ± 59.4	<0.001
Female:male	6:1	6:1	No significance
ANSC, %	93	100	No significance
FiO2 at DOL 1, %, mean ± SD	30.1 ± 6.2	38.8 ± 28.7	No significance
FiO2 at DOL 7, %, mean ± SD	22.1 ± 1.9	36.4 ± 9.4	<0.01
FiO2 at DOL 28, %, mean ± SD	27.1 ± 1.9	69.3 ± 24.2	<0.01
Moderate and severe BPD	0	7	<0.001

Definition of abbreviations: ANSC = antenatal corticosteroid; BPD = bronchopulmonary dysplasia; BW = body weight; DOL = day of life; FiO₂ = fraction of inspired oxygen; GA = gestational age; HO₂ = higher fraction of inspired oxygen; LO₂ = lower fraction of inspired oxygen.

Infants on HO₂ Have Increased AM-derived ASC in Their Plasma EVs

We isolated EVs from these patients and analyzed them using NanoSight tracking, TEM, western blot, and flow cytometry (12). NanoSight tracking showed that the majority of EV particles were between 100 and 150 nm in size in both groups (Figures 1A and 1B), and there was no difference in EV particle concentrations between the two groups (LO₂ group, 2.39 ± 5.32 × 10¹¹; HO₂ group, 5.52 ± 3.25 × 10¹¹; P = 0.06). TEM showed that the EVs from infants on HO₂ were smaller than those from infants on LO₂ (Figures 1C–1E; 49.75 ± 6.18 vs. 70.84 ± 10.00; P < 0.05). Western blot detected EV markers CD9, CD63, and CD81 in these EVs (Figure 1F), and the concentrations of CD63 (Figure 1G) and CD81 (Figure 1H) were higher in the LO₂ group than in the HO₂ group. Flow cytometry analysis (Figures 2A and 2B) demonstrated that the EVs from HO₂ plasma had higher ASC⁺ (Figure 2C; P < 0.04), CD11b⁺/ASC⁺ (Figure 2D; P < 0.05), CD11c⁺/ASC⁺ (Figure 2E; P < 0.01), and CD206⁺/ASC⁺ (Figure 2F; P < 0.05) EVs than the LO₂ plasma EVs. These results suggest that a major portion of circulating ASC⁺ EVs are of AM origin. We further performed an EV depletion assay by incubating these plasma EVs with magnetic beads coupled to a cocktail of antibodies to CD11b, CD11c, and CD206. The depleted EVs were analyzed again using flow cytometry for ASC, CD11b, CD11C, and CD206. As shown in Figures 2G–2I, the CD11b antibody effectively depleted 66% of CD11b⁺/ASC⁺ EVs. These results highlight that the majority of circulating EV-ASC in preterm infants is produced by AMs.

Figure 1.

Characterization of plasma extracellular vesicles (EVs). (A and B) NanoSight tracking analysis of EVs from infants on lower fraction of inspired oxygen (LO₂) (A) and EVs from infants on higher fraction of inspired oxygen (HO₂) (B). (C–E) Transmission EM of LO₂-EVs (C; red arrow) and HO₂-EVs (D; red arrow) showed that the HO₂-EVs were smaller in size (E). (F–H) Western blots demonstrated expression of EV markers CD9, CD63, and CD81 in the LO₂-EVs and the HO₂-EVs (F), with LO₂-EVs having higher expressions of CD63 (G) and CD 81 (H). Scale bars, 200 nm; n = 4 or 5 per group. *P < 0.05 and **P < 0.01.

Figure 2.

Plasma EVs from infants on HO₂ contain higher percentages of alveolar macrophage–derived ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain). (A–F) Flow cytometry analysis of LO₂-EVs (A) and HO₂-EVs (B) demonstrated that HO₂-EVs had higher percentages of ASC⁺ (C), CD11b⁺/ASC⁺ (D), CD11c⁺/ASC⁺ (E), and CD206⁺/ASC⁺ (F) EVs compared with LO₂-EVs. (G–I) Flow cytometry of nondepleted EVs (G) and depleted EVs (H) showed that the depletion protocol effectively removed the majority of CD11b⁺/ASC⁺ EVs (I). n = 5–7 per group. *P < 0.05, **P < 0.01, and ***P < 0.001.

Plasma EVs Can Cross the BBB

We assessed if the adoptive transfer of plasma EVs from the HO₂ group and LO₂ group can cross the BBB. As demonstrated in Figure 3, both LO₂-EVs and HO₂-EVs were visualized in the brain tissues of newborn mice by ex vivo imaging at 2 hours (Figure 3A). Immunostaining of sections of these brain tissues with a cocktail of antibodies to the EV markers CD9, CD63, and CD81 detected these EVs in brain cells, particularly in the subgranular zone (SGZ) (Figure 3B) and subventricular zone (SVZ) (data not shown), brain areas known to be rich in neural stem cells (NSCs). They were also detected in the lung tissues by ex vivo imaging (Figure 3C) and in small vessels and alveolar septum by immunostaining (Figure 3D). These data suggest that circulating EVs from preterm infants can cross the BBB and be taken up by brain cells.

Figure 3.

Tracking adoptively transferred EVs. ExoGlow-Vivo–labeled LO₂-EVs, HO₂-EVs, and PBS were injected into the retroorbital sinus of normal newborn mice at postnatal day 10. (A and C) The brains (A) and lungs (C) were dissected at 2 hours for ex vivo imaging. (B and D) Immunostaining for pan-EV markers CD9, CD63, and CD81 showed EVs (brown, red arrows) in the subgranular zone of the brain (B) and small vessels and alveolar septum of the lungs (D) in both LO₂-EV–injected and HO₂-EV–injected mice. Scale bars, 50 μm.

Adoptive Transfer of Circulating EVs from Infants on HO₂ Increases Lung Expression of Inflammasome Components, Apoptosis Regulators, and a Necroptosis Regulator

We examined the protein expression of ASC and GSDMD in the lungs of mice that received adoptive transfer of EVs. As shown in Figure 4, ASC was detected with increased intensities on sections of HO₂-EV–injected lungs compared with LO₂ and control EVs in both small pulmonary vessels and AM-like cells, whereas GSDMD was detected more in AM-like cells. We assessed gene expression of NLRP3 inflammasome components and found that HO₂-EVs increased Gsdmd compared with the other two groups. In addition, the expressions of casp3, casp8, and Ripk3 (receptor-interacting protein kinase 3) were also induced to higher degrees in the lungs of HO₂-EV–injected mice than the other two groups. GSDMD regulates pyroptosis, caspase-3 and caspase-8 regulate apoptosis, and RIPK3 regulates necroptosis, and these results suggest that HO₂-EVs could modulate these three cell death pathways in the lungs.

Figure 4.

HO₂-EV–injected mice have increased ASC and GSDMD (gasdermin D) expression in the lungs. (A) Immunostaining with an anti-ASC antibody detected increased ASC expression in small vasculatures and infiltrated AMs in the HO₂-EV–injected lung sections. (B) Zoomed magnification in the red box. (C) Immunostaining with an anti-GSDMD antibody detected increased GSDMD expression in the alveolar macrophages in the HO₂-EV–injected lung sections. (D) Zoomed magnification in the red box. Objective magnification, 20×. Scale bars, 50 μm. (E–H) qRT-PCR detected increased Gsdmd (E), casp3 (F), casp8 (G), and Ripk3 (H) gene expression in HO₂-EV–injected lungs compared with the other two groups. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. n = 6 per group. AMs = alveolar macrophages; rRNA = ribosomal RNA.

Adoptive Transfer of Circulating EVs from Infants on HO₂ Induces Lung Inflammation in Neonatal Mice

We next investigated if the adoptive transfer of circulating EVs with increased cargo of CD11b⁺/ASC⁺, CD11c⁺/ASC⁺, and CD206⁺/ASC⁺ EVs could cause inflammatory lung injury in normal neonatal mice. On hematoxylin and eosin–stained lung tissue sections, we found that HO₂-EV injection resulted in increased infiltrated cells in the alveolar airspaces compared with the other two groups (Figure 5A). We performed immunostaining with an antibody for Mac3, an AM marker, and found that the lungs of HO₂-EV–transferred mice had a significant increase in Mac3-positive cell counts compared with LO₂-EV–injected and control EV–injected groups. The LO₂-EV–injected group also had a slight increase in Mac3-positive cell counts compared with the control EV–injected group (Figures 5B and 5C). We further investigated whether the adoptive transfer of HO₂-EVs affected macrophage phenotypes by staining lung tissue sections with iNOS (an M1 marker) and Ym1 (an M2 marker) and found that HO₂-EVs altered both M1 and M2 phenotypes (Figures 5E–5H). Overall, these results suggest that HO₂-EVs cause lung inflammation.

Figure 5.

Adoptive transfer of HO₂-EVs increases lung inflammation. (A) H&E staining showed increased cellular infiltration into the alveolar airspaces in HO₂-EV–injected lung tissue sections. (B) Immunostaining for Mac3 demonstrated increased alveolar macrophages (AMs) in HO₂-EV–injected lungs compared with the other two groups (C). The LO₂-EV–injected lungs also had slightly higher macrophage counts compared with the control EV–injected lungs (C). (D) Zoomed magnification of the red box. (E) Immunostaining for iNOS detected positive stained AMs in HO₂-EV–injected lungs. (F) Zoomed magnification of the red box. (G) Immunostaining for Ym1 detected positive stained AMs in HO₂-EV–injected lungs. (H) Zoomed magnification of the red box. Objective magnification, 20×. Scale bars, 50 μm. n = 5 per group. *P < 0.05 and ****P < 0.0001. HPV = high-power view; iNOS = inducible nitric oxide synthase.

Circulating EVs from Infants on HO₂ Inhibit Alveolarization and Vascular Development in Normal Neonatal Mice

Impaired alveolarization and vascular development are hallmarks of BPD. Thus, we next evaluated alveolar and vascular development in neonatal mice that received EV adoptive transfer. Microscopy of lung tissue sections showed that mice injected with HO₂-EVs had larger and more simplified alveolar structures (Figure 6A). Morphometric analysis demonstrated increased mean linear intercept in HO₂-EV–injected mice compared with mice that received LO₂-EVs or control EVs (Figure 6B). Moreover, vascular density was significantly reduced in mice injected with HO₂-EVs (Figures 6C and 6D). These results confirm that circulating EVs from infants on HO₂ at 1 week of age can induce BPD-like pathology.

Figure 6.

Adoptive transfer of HO₂-EVs induces bronchopulmonary dysplasia–like pathology. (A and B) Assessment of lung morphometry (A) showed that HO₂-EV–injected lungs had increased MLI compared with LO₂-EV–injected and control EV–injected lungs (B). (C) Immunofluorescence staining for vWF demonstrated that HO₂-EV–injected lungs had reduced vWF⁺ vessels compared with LO₂-EV–injected and control EV–injected lungs (D). n = 5 per group. Objective magnification, 20×. Scale bars, 50 μm. *P < 0.05, **P < 0.01, and ****P < 0.0001. MLI = mean linear intercept; ns = not significant; vWF = von Willebrand factor.

Detection of ASC and GSDMD in Mouse Hippocampus

We examined ASC and GSDMD expression in the hippocampus of mice that received EV adoptive transfer. Some ASC was detected in the SGZ and SVZ (data not shown) areas, but the highest intensity of ASC was found in the dentate gyrus hilus and areas peripheral to the dentate gyrus in the HO₂-EV–injected brain (Figure 7A). Double immunofluorescent staining was performed to colocalize ASC with AIF1 (a microglial marker), GFAP (an astrocyte marker), and SOX10 (an oligodendrocyte marker). ASC was colocalized with AIF1 and GFAP (Figures 7B and 7C), suggesting that ASC is expressed in microglial cells and astrocytes. Similar to ASC, GSDMD was detected in the same areas as ASC, with increased intensity in HO₂-EV–injected mice compared with the other two groups of mice (Figure 7D). Double immunofluorescent staining colocalized GSDMD with AIF1, GFAP, and SOX10 (Figures 7E–7G), indicating that GSDMD is expressed in microglial cells, astrocytes, and oligodendrocytes.

Figure 7.

Detection of ASC and GSDMD in brain cells. (A) ASC immunostaining showed that ASC expression was increased in the hippocampal area of HO₂-EV–injected and LO₂-EV–injected brains (red arrows) compared with the hippocampal areas of control EV–injected brains. Representative red arrows point to focal enlarged areas of ASC⁺ cells in the red boxes. Objective magnification, 20×. Scale bars, 50 μm. (B) Double immunofluorescent staining for ASC (green), AIF1 (allograft inflammatory factor 1; red), and DAPI (blue). ASC was colocalized with AIF1 in the HO₂-EV–injected hippocampus (white arrow and white box). Scale bars, 50 μm. (C) Double immunofluorescent staining for ASC (green), GFAP (glial fibrillary acidic protein; red), and DAPI (blue). ASC was colocalized with GFAP in control EV–injected, LO₂-EV–injected, and HO₂-EV–injected hippocampi (white arrows and white arrows). Objective magnification, 20×. Scale bars, 50 μm. (D) GSDMD immunostaining showed that its expression was increased in the hippocampal area of HO₂-EV–injected brains (red arrows) compared with the hippocampal areas of control EV–injected and LO₂-EV–injected brains. Representative focal enlarged areas of GSDMD⁺ cells are in the red boxes. (E) Double immunofluorescent staining for GSDMD (green), AIF1 (red), and DAPI (blue). GSDMD was colocalized with AIF1 in the hippocampus of all three groups, but only the LO₂ and HO₂ groups had microglial cells exhibiting active forms (white arrows and white boxes). (F) Double immunofluorescent staining for GSDMD (green), GFAP (red), and DAPI (blue). GSDMD was colocalized with GFAP in the hippocampal areas of three groups (white arrows and white boxes). (G) Double immunofluorescent staining for GSDMD (green), SOX10 (red), and DAPI (blue; not shown). GSDMD was colocalized with SOX10 in the hippocampal areas of all three groups (white arrows and white boxes). Objective magnification, 20×. Scale bars, 50 μm.

Circulating EVs from Infants on HO₂ Induce Hippocampal Inflammation in Neonatal Mice

We next examined hippocampal sections for microglial infiltration by immunostaining to assess whether adoptive transfer of EVs induces hippocampal inflammation. Histologically, many more microglial cells were positive for AIF1 in the HO₂-EV– injected hippocampus than the other two groups (Figure 8A). These cells had activated microglia features such as enlargement of the cell body and irregular cell shape (red boxes). Quantitative analysis showed that the AIF1⁺ microglial count in the HO₂-EV–injected mice was more than 2-fold higher than in the control EV–injected mice (P < 0.0001; Figure 8B) and 1.4-fold more elevated than the LO₂-EV–injected mice. We also assessed the gene expression of NLRP3 inflammasome cascade in the total RNA isolated from the hippocampal area. As demonstrated in Figures 8C–8E, expression of Nlrp3, Asc, and Il18 were significantly increased in HO₂-EV–injected mice compared with control EV–injected mice. The expressions of Nlrp3 and Il18 were also higher in HO₂-EV–injected brains than in LO₂-EV–injected brains (Figures 8C and 8E). Thus, HO₂-EVs induce brain inflammatory responses when adoptively transferred into normal newborn mice.

Figure 8.

HO₂-EVs induce brain inflammation. (A) Immunostaining for AIF1 (brown, red arrows) showed that microglia cells in the HO₂-EV–injected hippocampus were disorganized and had enlarged bodies compared with hippocampi from LO₂-EV–injected and control EV–injected mice. Representative focal enlarged areas of microglial cells are in the red boxes. Objective magnification, 20×. Scale bars, 50 μm. (B) There was a significant increase of AIF1⁺ microglial cells in the HO₂-EV–injected group compared with the other two groups. (C–E) qRT-PCR demonstrated increased gene expression of the inflammasome components Nlrp3, Asc, and Il18 in the hippocampus of HO₂-EV–injected mice compared with control EV–injected and LO₂-EV–injected mice. n = 5 or 6 per group. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Circulating EVs from Infants on HO₂ Decrease Cell Survival and Increase Cell Death in Mouse Hippocampus

The inflammasome–ASC–GSDMD cascade plays a crucial role in pyroptosis. We found that HO₂-EV–injected mice had a 1.7-fold decrease in cell proliferative index compared with control EV–injected mice (Figures 9A and 9B; P < 0.0001) and a 1.5-fold reduction in cell proliferative index compared with LO₂-EV–injected mice (Figures 9A and 9B; P < 0.001) in the SGZ as determined by immunofluorescence staining for a cell proliferation marker Ki67. When we assessed cell death, our data showed that the HO₂-EV–injected mice had a 2.0-fold increase in cell death index compared with the mice that received control EV injection (Figures 9C and 9D; P < 0.0001), as well as a 1.5-fold increase in cell death index compared with LO₂-EV injection (Figures 9C and 9D; P < 0.01) in the SGZ area. Our previous study showed that SGZ is rich in NSCs (39). To reveal potential cell death mechanisms, we assessed GSDMD and MLKL expression and colocalized them in clusters of cells in the SGZ area of HO₂-EV–injected mice (Figures 9E and 9F). We also evaluated gene expression of molecules important in inducing apoptosis and necroptosis and found that HO₂-EVs increased expression of casp3 and casp8 (apoptosis regulators), and Tnf, Irpk1, Irpk3, and Mlkl (necroptosis regulators) compared with the other two groups (Figures 9G–9L). Thus, HO₂-EVs reduce cell survival and increase cell death, which is possibly caused by apoptosis, pyroptosis, and necroptosis in the SGZ, where the loss of NSCs may contribute to long-term NDI.

Figure 9.

HO₂-EVs decrease cell survival and increase cell death in the hippocampus. (A) Merged representative immunofluorescence staining for Ki67 (pink, white arrow) and DAPI nuclear staining (blue) in the hippocampus of control EV–injected, LO₂-EV–injected, and HO₂-EV–injected mice. (B) Quantification of the cell proliferative index (Ki67⁺ nuclei/total nuclei × 100) showed a highly significant decrease in the HO₂-EV–injected group compared with the control group and LO₂ group, while the LO₂-EV–injected group also had a small but significant reduction compared with the control EV–injected group. (C) TUNEL assay (teal, white arrow) and DAPI nuclear stain (blue) were used to identify dead cell nuclei. (D) Quantification of cell death index (TUNEL⁺ nuclei/total nuclei × 100) revealed that it was increased in the hippocampus of HO₂-EV–injected mice compared with LO₂-EV–injected mice and control EV–injected mice. n = 5 per group. (E) Double immunofluorescent staining with GSDMD (green) and MLKL (red) showed they were colocalized in some cluster cells near the subgranular zone in HO₂-EV–injected mice. (F) Zoomed magnification in the white box. (G–L) qRT-PCR detected increased expression of casp3 (G) and casp8 (H) (apoptosis related) and Tnf (I), Ripk1 (J), Ripk3 (K), and Mlkl (L) (necroptosis related) in the hippocampus of HO₂-EV–injected mice compared with control EV–injected and LO₂-EV–injected mice. n = 6 per group. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. MLKL = mixed-lineage kinase domain-like. Objective magnification, 20×. Scale bars, 50 μm.

Discussion

BPD and brain injury are among the most common complications affecting extremely preterm infants, and currently, there are no effective and safe therapies for either condition. Many clinical studies indicate that BPD is associated with advanced brain injury and long-term NDI (43, 44). However, the mechanistic link between BPD and NDI remains to be explored. In this study, we focused our investigations on the expression of inflammasome cascade molecules in the plasma EVs of preterm infants at risk for developing BPD. We also explored the mechanisms by which plasma EVs from preterm infants induce lung and brain injury when adoptively transferred into normal newborn mice. We provide evidence, for the first time to the best of our knowledge, the plasma EVs from infants on HO₂ at 1 week of age had increased ASC in AM-released EVs, as their plasma had higher concentrations of CD11b⁺/ASC⁺, CD11c⁺/ASC⁺, and CD206⁺/ASC⁺ EVs. Importantly, the adoptive transfer of these EVs induced lung inflammation while reducing alveolarization and pulmonary vascular development. Furthermore, these EVs caused inflammation and cell death, possibly through apoptosis, pyroptosis, and necroptosis mechanisms in the hippocampal area of the brain, as summarized in Figure 10. Overall, these findings highlight a novel lung–brain axis mediated by AM-derived EVs with an increased cargo of ASC that is critical in inducing lung and brain injury in neonates, which suggests that targeting EV-ASC may have therapeutic value in preventing BPD-associated brain injury and NDI.

Figure 10.

EV-mediated lung–brain axis. The AMs in early injured lungs release EVs that contain an increased cargo of ASC. These EVs contribute to bronchopulmonary dysplasia pathogenesis by inducing lung inflammation and inhibiting alveolarization and vascularization. These EVs can be released to the circulation, cross the BBB and be taken up by neural cells, and the ASC cargo can activate GSDMD in specific neural cells and result in brain injury by activating microglia and inducing cell death, possibly through apoptosis, pyroptosis, and necroptosis mechanisms. The molecular and cellular changes can lead to long-term NDI. BBB = blood–brain barrier; NDI = neurodevelopmental impairment.

Since it was first described by Dr. Northway in 1967, the pathogenesis of BPD has evolved from an airway disease to a disease of both alveoli and pulmonary vasculature (3). Although various prenatal and perinatal factors can lead to lung injury and BPD development, the lung injury, regardless of the cause, is believed to be due mainly to inflammatory responses mediated by macrophages and neutrophils that invade the endothelium and alveolar spaces of premature lungs (45, 46). Although inflammation is known to play a vital role in BPD pathogenesis, there are currently no antiinflammatory drugs that can prevent or treat BPD, except postnatal corticosteroids. However, corticosteroid use is linked to various NDI in preterm infants (47, 48).

Inflammasomes are multiprotein complexes that mediate proteolytic cleavage of GSDMD, pro–IL-1β, and pro–IL-18 by caspase-1 (22–25). Previous studies from our laboratory have defined a crucial role for the inflammasome in hyperoxia-induced neonatal lung, eye, and brain injury and the efficacies of a caspase-1 inhibitor and GSDMD gene knockout in preventing hyperoxia-induced organ injury (28, 37, 39). In this study, we investigated whether plasma EVs from infants on HO₂ in early life have increased ASC and the functions of these EVs on lung and brain injury upon adoptive transfer into normal newborn mice.

One of the key findings of this study is the discovery that plasma EVs from infants on HO₂ have higher expression of AM-derived ASC⁺ EVs than EVs from infants on LO₂ at 1 week of age. We used flow cytometry analysis to identify that infant on HO₂ had higher proportions of ASC⁺, CD11b⁺/ASC⁺, CD11c⁺/ASC⁺, and CD206⁺/ASC⁺ EVs in their plasma. These three markers are considered AM markers (32, 45). The percentages of CD11b⁺/ASC⁺, CD11c⁺/ASC⁺, and CD206⁺/ASC⁺ EVs were more than threefold higher in infants on HO₂ than in those on LO₂. These data, combined with the results from EV depletion with an anti-CD11b antibody, suggest that early lung injury stimulates AMs to release EVs containing an increased cargo of ASC into circulation. To our knowledge, this is the first study reporting increased AM-derived ASC in circulating EVs in patients at risk for BPD or in any neonatal patients. There are reports that other inflammasome components are higher in plasma microparticles and airway fluids. Increased plasma microparticle GSDMD was detected in adult septic patients and increased EV caspase-1 in airway fluids from patients with cystic fibrosis (49, 50). These results suggest that the activation of cellular inflammasome components under inflammatory conditions leading to EV release into the circulation and airway fluid may be common to many inflammatory lung diseases.

We explored the biological functions of circulating EVs from infants on HO₂ and LO₂ to determine if they are novel mediators for a lung–brain axis that induces lung and brain injury. We performed adoptive transfer experiments, as we have previously done with circulating EVs from hyperoxia-induced newborn rat models of lung injury (21). We first confirmed that the patient’s EVs can cross the BBB and be taken up by brain cells. We then systematically evaluated lung and brain injury in EV adoptively transferred mice.

Mice that received HO₂-EV injections had increased lung ASC expression in small vasculature and AMs and increased GSDMD expression in AMs, and these were associated with increased lung inflammation. The majority of infiltrated cells were AMs, as we have described, although possible neutrophil and lymphocyte infiltration was also seen (data not shown). We further evaluated macrophage phenotype by immunostaining for iNOS, an M1 marker, and Ym1, an M2 marker. Our results showed that HO₂-EV injections increased the expression of both proteins in AMs. Changing macrophage phenotypes is a process that hinges on an intricate regulatory network influenced by myriad signaling molecules and transcription factors (51). In general, M1 macrophages are considered inflammatory macrophages, and M2 macrophages are considered antiinflammatory macrophages (52). Increased expression of NLRP3, ASC, and caspase-1 has been associated with altered phenotype in AMs from BAL fluid in LPS-induced adult lung injury (53). As both ASC and GSDMD were detected in AMs in our study, it is plausible to suggest that ASC and GSDMD might affect AM phenotypes. We also assessed gene expression of inflammasome cascade molecules and found that Nlrp3, Asc, and Il18 were increased in the lungs of HO₂-EV–injected mice. Taken together, these results suggest that the upregulation of inflammasome components contributes to the induction of lung inflammation by HO₂-EVs in the recipient mice.

We observed decreased alveolarization and vascular density, two hallmarks of BPD in HO₂-EV–injected mice, suggesting that the adoptive transfer of HO₂-EVs was responsible for creating a BPD phenotype in these otherwise normal newborn mice. Similar results were reported in our previous study with hyperoxia-induced EV-injected rats (21). Interestingly, ASC expression was detected in some of the small vessels and AMs, but GSDMD expression was detected only in the AMs. We hypothesize that the BPD-like pathology is secondary to lung inflammation. Given that we examined the lungs after 14 days of exposure to HO₂-EVs, our observations likely reflect chronic responses. Future studies will be needed to evaluate how HO₂-EV affects lung development at earlier times, which may uncover how lung inflammation precedes deranged alveolar and vascular development.

Our investigation yielded pivotal data that support a novel hormone-like mechanism in which circulating EVs from preterm infants can cross the BBB and be taken up by neural cells. More important, we discovered that HO₂-EVs are still biologically active after crossing the BBB. They induced brain inflammatory injury similar to hyperoxia-induced neonatal brain injury in rodent models, as we have previously observed (21, 28, 39). We investigated if the adoptive transfer of circulating EVs from infants on HO₂ containing increased ASC cargo could induce GSDMD expression in specific neural cells and found that both ASC and GSDMD expressions were increased in the hippocampus of HO₂-EV–injected mice. ASC was colocalized with microglial cells and astrocytes and GSDMD with microglia, astrocytes, and oligodendrocytes. Microglia are highly dynamic cells, apart from their well-known immune functions (54). Microglial activation has been demonstrated almost ubiquitously in perinatal brain injury, and these responses are key cell orchestrators of brain pathology but also may be involved in repair (55). During brain development, astrocytes are implicated in neurogenesis, synaptogenesis, controlling BBB permeability, and maintaining extracellular homeostasis (56). Oligodendrocytes are responsible for the myelination of axons, and they also play a role in neural plasticity by inducing activity-dependent myelination, which is required for motor learning (57). Impaired oligodendrocyte functions are linked to white matter injury in preterm infants (57). Increased expression of ASC and GSDMD in these cells, as we observed, could lead them to pyroptosis and inflammation that greatly reduces their function in the developing brain. Collectively, our data reveal a novel in vivo paradigm that links early lung injury, circulating AM-derived EV-ASC, and increased ASC and GSDMD expression in specific neural cells that play crucial roles in brain development, neurogenesis, motor learning, and inflammation in hippocampal areas. Injuries to these cells could lead to long-term NDI, which is under investigation.

We demonstrated that EVs from infants on HO₂ activated microglial cells in the hippocampus, and this activation was accompanied, as we had seen in the lung, by increased expression of the inflammasome components NLRP3, ASC, and IL-18. Previous studies have shown that IL-1α, IL-18, and TNF-α are important mediators of neonatal inflammatory brain injury (58, 59). We further found that HO₂-EVs induced cell death in the SVZ (data not shown) and SGZ areas of recipient mice. We have previously found that SGZ and SVZ are the sites of cell death in hyperoxia-exposed mice and hyperoxia-induced EV-injected rats (21, 39). It is well known that both SVZ and SGZ are the main sites of postnatal neurogenesis in mammalian brains (60–63). The developing SVZ and SGZ are enriched in NSCs that can proliferate and generate transient neural progenitor cell clusters, which then further differentiate into astrocytes, oligodendrocytes, and neurons (60–63). Since the neural progenitor cell clusters lie in close proximity to blood vessels, and circulating mediators are known to significantly affect their biological functions (64–66), they would be logical HO₂-EV targets. Our previous studies have identified NSCs as the primary cells undergoing death upon exposure to hyperoxia or hyperoxia-induced circulating EVs (21, 39). One of our interesting findings was that we colocalized GSDMD with MLKL in some cells around the SGZ. MLKL is a crucial factor for necroptosis, which involves membrane pore formation and lytic cell death (67). TNF-α is a major activator of necroptosis, and the process is mediated by several kinases, including RIPK1, RIPK3, and MLKL (67). Inhibition of necroptosis has been shown to markedly ameliorate brain injury and long-term neurobehavioral abnormalities of periventricular leukomalacia mice (68). We did find that HO₂-EVs increased TNF-α gene expression, and this combined with increased MLKL expression in GSDMD-expressing cells strongly suggest that there could be both pyroptosis and necroptosis occurring in these cells.

Our study has limitations. Our clinical study has small sample sizes for both HO₂ and LO₂ groups, and the infants on HO₂ had lower GAs and birth weights. Future studies will enroll additional patients in each group to control the confounding effect of GA and determine if there is a correlation between AM-derived EV-ASC and the severity of BPD.

We did not investigate whether other cell types, such as AECs and vascular endothelial cells, contribute to increased circulating EV-ASC concentrations in the infants on HO₂. Thus, future studies are needed to determine if circulating ASC⁺ EVs from preterm infants are of epithelial or endothelial origin.

Another limitation of this study is that although we performed AM EV depletion, we did not explore whether the depleted EVs still have effects on lung and brain injury. Future studies will be performed to definitively prove that AM-ASC is required for circulating EV-induced lung and brain injury in neonatal mice. Alternatively, studies directed at inhibiting EV formation or cellular uptake will also be important in understanding the mechanisms by which EV-ASC mediates the lung–brain axis that appears to play a pivotal role in BPD-associated brain injury.

We recognize that our study was limited as we examined only lung and brain structural development, inflammation, and gene expression at P14. We plan to perform longer studies to evaluate how HO₂-EVs alter lung function and neurobehavior at later time points, such as 1–6 months of age in mouse models.

Conclusions

In this study, we demonstrate that plasma EVs from infants on HO₂ contain higher concentrations of AM-derived EV-ASC than infants on LO₂ at an early age. These EVs induce lung inflammatory injury and BPD-like pathology in adoptively transferred mice. More important, these EVs can cross the BBB and induce brain inflammation and cell death. Notably, the effects of HO₂-EVs on lung and brain development, inflammation, and gene expression in newborn mice observed here are strikingly similar to what has been shown by us and other investigators in hyperoxia-induced lung and brain injury in newborn rodent models (28, 39, 69, 70). Although there is no doubt that hyperoxia alone can cause both lung and brain injury in newborn rodents and possibly preterm infants by directly inducing oxidative stress and inflammatory responses in lung and brain cells, our study suggests that HO₂-EVs also play a significant role in inducing lung and brain cell damage. Moreover, this work indicates that early lung injury causes AMs to release EV-ASC, which serves as a novel mediator for the lung–brain axis that may be critical for BPD-associated brain injury, paving the way for further exploration of potential EV-ASC–targeted therapies for preventing and treating BPD-associated brain injury and long-term NDI in preterm infants.