The ARDS microenvironment enhances MSC-induced repair via VEGF in experimental acute lung inflammation

Abstract

Clinical trials investigating the potential of mesenchymal stromal cells (MSCs) for the treatment of inflammatory diseases, such as acute respiratory distress syndrome (ARDS), have been disappointing, with less than 50% of patients responding to treatment. Licensed MSCs show enhanced therapeutic efficacy in response to cytokine-mediated activation signals. There are two distinct sub-phenotypes of ARDS: hypo- and hyper-inflammatory. We hypothesized that pre-licensing MSCs in a hyper-inflammatory ARDS environment would enhance their therapeutic efficacy in acute lung inflammation (ALI). Serum samples from patients with ARDS were segregated into hypo- and hyper-inflammatory categories based on interleukin (IL)-6 levels. MSCs were licensed with pooled serum from patients with hypo- or hyper-inflammatory ARDS or healthy serum controls. Our findings show that hyper-inflammatory ARDS pre-licensed MSC conditioned medium (MSC-CM_Hyper) led to a significant enrichment in tight junction expression and enhanced barrier integrity in lung epithelial cells in vitro and in vivo in a vascular endothelial growth factor (VEGF)-dependent manner. Importantly, while both MSC-CM_Hypo and MSC-CM_Hyper significantly reduced IL-6 and tumor necrosis factor alpha (TNF-α) levels in the bronchoalveolar lavage fluid (BALF) of lipopolysaccharide (LPS)-induced ALI mice, only MSC-CM_Hyper significantly reduced lung permeability and overall clinical outcomes including weight loss and clinical score. Thus, the hypo- and hyper-inflammatory ARDS environments may differentially influence MSC cytoprotective and immunomodulatory functions.

Keywords: ARDS; MSC; SARS-CoV-2; VEGF; permeability.

Graphical abstract

Figure 1

ARDS patient sub-phenotype stratification (A) Schematic created using Biorender.com. ARDS patient serum samples were obtained from patients of (B–D) differing backgrounds (age, World Health Organization [WHO] score, O₂ level) and (E and F) stratified into a hypo- or hyper-inflammatory sub-phenotype based on their IL-6 levels at the time of admission (n = 7). Patients with <50 pg/mL IL-6 were considered hypo-inflammatory, and patients with >50 pg/mL IL-6 were considered hyper-inflammatory. The samples were then pooled, and the (G) IL-6 (BioLegend), (H) IL-8 (R&D), (I) MIF (R&D), and (J) TNF-α (BioLegend) levels were observed in each pool. Serum pooled from 6 healthy age-matched patients was used as a control. Data are presented as mean ± SEM; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 2

The MSC secretome in response to healthy or hypo- or hyper-inflammatory ARDS serum (A) Schematic created using Biorender.com. hBM-MSCs (n = 3 donors) were seeded at a density of 1 × 10⁵ in a 6-well plate, left to attach, and exposed to 20% ARDS patient serum or healthy serum control. The serum was removed after 24 h, the cells were washed with PBS, and serum-free DMEM was added for a further 24 h to generate MSC-CM. The (B) IL-6 (BioLegend), (C) IL-8, (D) MIF, and (E) VEGF (R&D) levels were then analyzed via ELISA. RT-PCR was also carried out on (F) vegf-a and (G) kdr. Data are presented as mean ± SEM; n = 3 per group; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 3

The MSC-CM enhancement of tight junction expression (A) Schematic created using Biorender.com. CALU-3 cells (passages 10–12) were seeded at a density of 5 × 10⁵ in a 6-well plate, left to attach, and exposed to 10 μM SU-5416, a VEGFR2 inhibitor, for 4 h before being stimulating with 2 μg/mL endotoxin for 48 h. Some groups were subsequently exposed to 2 mL of MSC-CM containing ∼1,000 pg/mL VEGF for 24 h. The cells were then harvested for gene expression studies of tight junction genes: (B) tjp1, (C) ocln, and (D) cldn4. Data are presented as mean ± SEM; n = 3 per group; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 4

Barrier integrity of CALU-3 cells in response to MSC-CM (A) Schematic created using Biorender.com. CALU-3s were seeded at a density of 7 × 10⁵ in a Transwell (0.4 μm) and grown under air-liquid interface conditions from day 8 of culture. On day 15, the cells were stimulated with 2 μg/mL of endotoxin, and on day 17, they were exposed to 500 μL of MSC-CM from the groups depicted in Figure 2E (± bevacizumab biosimilar or the appropriate IgG isotype control). (B) TEER measurements were taken 3 times a week for 3 weeks, and (C) final measurements were taken on day 19. (D–F) Gene expression studies highlighted differences in tight junction expression. Data are presented as mean ± SEM; n = 3 per group; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 5

Cytokine levels in BALF and clinical scoring from ALI mouse model BALF was harvested from C57BL6/J mice (n = 5) exposed to our ALI model and (A) IL-6 and (B) TNF-α were quantified by ELISA. (C) Clinical score and (D) percentage of weight loss were also assessed. Data are presented as mean ± SEM; n = 5 per group; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 6

The impact of MSC-CM on lung permeability in vivo (A) Schematic created using Biorender.com. C57BL6/J mice (n = 5) were exposed to 2 mg/kg of endotoxin intratracheally at T0. (B and C) At T4, 500 μL of MSC-CM that had been concentrated down to 30 μL (± bevacizumab biosimilar or an appropriate IgG isotype control) was administered intranasally. 1 h before harvesting (T47), a 10% solution of Evans Blue dye was administered intravenously, and mice were sacrificed at T48. (D and E) Evans Blue dye and (F) wet:dry weight ratio were used as indicators of permeability. Data are presented as mean ± SEM; n = 5 per group; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.