Therapeutic effects of human mesenchymal stem cells in ex vivo human lungs injured with live bacteria

Abstract

Rationale: Mesenchymal stem cells secrete paracrine factors that can regulate lung permeability and decrease inflammation, making it a potentially attractive therapy for acute lung injury. However, concerns exist whether mesenchymal stem cells' immunomodulatory properties may have detrimental effects if targeted toward infectious causes of lung injury.

Objectives: Therefore, we tested the effect of mesenchymal stem cells on lung fluid balance, acute inflammation, and bacterial clearance.

Methods: We developed an Escherichia coli pneumonia model in our ex vivo perfused human lung to test the therapeutic effects of mesenchymal stem cells on bacterial-induced acute lung injury.

Measurements and main results: Clinical-grade human mesenchymal stem cells restored alveolar fluid clearance to a normal level, decreased inflammation, and were associated with increased bacterial killing and reduced bacteremia, in part through increased alveolar macrophage phagocytosis and secretion of antimicrobial factors. Keratinocyte growth factor, a soluble factor secreted by mesenchymal stem cells, duplicated most of the antimicrobial effects. In subsequent in vitro studies, we discovered that human monocytes expressed the keratinocyte growth factor receptor, and that keratinocyte growth factor decreased apoptosis of human monocytes through AKT phosphorylation, an effect that increased bacterial clearance. Inhibition of keratinocyte growth factor by a neutralizing antibody reduced the antimicrobial effects of mesenchymal stem cells in the ex vivo perfused human lung and monocytes grown in vitro injured with E. coli bacteria.

Conclusions: In E. coli-injured human lungs, mesenchymal stem cells restored alveolar fluid clearance, reduced inflammation, and exerted antimicrobial activity, in part through keratinocyte growth factor secretion.

Figures

Figure 1.

Schematic diagram of Escherichia coli pneumonia in the ex vivo perfused human lung. The right or left human lung declined for transplantation by the Northern California Transplant Donor Network is selected for perfusion if the total ischemic time is less than 48 hours and if the selection criteria as described in the Methods section are met. The lung is gently rewarmed and perfused with a crystalloid solution (Dulbecco's Modified Eagle's [DME] H-21 Medium with 5% albumin) over 1 hour and oxygenated with 10 cm H2O continuous positive airway pressure (CPAP) (FiO2 0.95). The perfusion rate or cardiac output is set at 0.2 L/min and the left atrial pressure at 0 mm Hg to prevent hydrostatic pulmonary edema. If the alveolar fluid clearance (AFC) rate is greater than or equal to 10% per hour in the control right upper lobe (RUL) or left upper lobe (LUL), then 109 CFU of E. coli bacteria (K1 strain) is instilled into the right middle lobe (RML) or left lower lobe (LLL) and 100 ml of fresh whole blood is added to the perfusate. For mesenchymal stem cells treatment groups, human mesenchymal stem cells are instilled intrabronchially into the RML or LLL or into the perfusate (intravenously) 1 hour after the initiation of the injury.

Figure 2.

Effect of clinical-grade, cryopreserved allogeneic human mesenchymal stem cells (MSC) on alveolar fluid clearance and on inflammatory cell infiltration into the injured lung lobe and histology after Escherichia coli pneumonia. (A) Instillation of clinical-grade, cryopreserved human allogeneic MSC intrabronchially (IB) or intravenously (IV) restored the decrease in alveolar fluid clearance (AFC) in the lung lobe injured by E. coli pneumonia at 6 hours. AFC was measured by the change in protein concentration of a 5% albumin instillate in the lung lobe over 1 hour and expressed as mean AFC (% per h, per 150 ml of bronchoalveolar lavage) ± SD for each condition. n = 20 for control lobe, n = 4 for E. coli–injured lung lobe, n = 3 for E. coli–injured lung lobe treated with MSC IB or IV or normal human lung fibroblasts (NHLF), *P < 0.008 compared with control lobe by analysis of variance (ANOVA; Bonferroni). (B) Instillation of clinical-grade MSC IB or IV decreased the influx of inflammatory cells, specifically neutrophils, into the lung lobe injured by E. coli pneumonia at 6 hours. Absolute neutrophil counts are expressed as mean total neutrophil counts (× 107 cells) ± SD for each condition. n = 20 for control lobe, n = 4 for E. coli–injured lung lobe, n = 3 for E. coli–injured lung lobe treated with MSC IB or IV or NHLF, *P < 0.0001 compared with control lobes by ANOVA (Bonferroni), √P < 0.0002 versus E. coli pneumonia by ANOVA (Bonferroni). (C) Human lungs exposed to E. coli bacteria with and without MSC were fixed in 10% formalin at 6 hours. Sections were stained with hematoxylin and eosin. The administration of human MSC 1 hour after E. coli pneumonia injury reduced the level of hemorrhage, edema, and cellularity in the injured lung lobe at 6 hours. Although not statistically significant by ANOVA, administration of human MSC IV after E. coli pneumonia also reduced the levels of inflammatory cytokines in a dose-dependent manner; doubling the dose of MSC to 10 × 106 cells reduced IL-1β level by 63% and tumor necrosis factor-α by 87% in the alveolar fluid compared with E. coli–injured lung lobes (IL-1β: 1,193 ± 549 for E. coli, 926 ± 156 treatment with MSC [×1], 437 ± 345 treatment with MSC [×2]; tumor necrosis factor-α: 3,019 ± 1,767 for E. coli, 1,604 ± 377 treatment with MSC [×1], 384 ± 189 treatment with MSC [×2], values are mean ± SD pg/ml, n = 3–4).

Figure 3.

Effect of allogeneic human mesenchymal stem cells (MSC) on alveolar Escherichia coli bacterial load after E. coli pneumonia. (A) Instillation of clinical-grade, cryopreserved human allogeneic MSC intrabronchially (IB) or intravenously (IV) decreased the total bacterial load in the lung lobe injured by E. coli pneumonia at 6 hours. Total bacterial counts were expressed as mean (× 106 CFU counts/ml) ± SD for each condition. n = 18 for control lobe, n = 3 for E. coli–injured lung lobe with or without administration of MSC IB or IV or normal human lung fibroblasts, *P < 0.0005 versus control lobe CFU counts per milliliter by analysis of variance (ANOVA; Bonferroni), √P < 0.0001 versus E. coli pneumonia CFU counts per milliliter by ANOVA (Bonferroni). (B) The alveolar fluid conditioned medium of lung lobes treated with human MSC IB or IV had increased antimicrobial activity against E. coli bacteria reexposed in vitro. Total bacterial counts were expressed as mean (× 104 CFU counts/ml) ± SD for each condition. n = 19 for control lobe, n = 4 for E. coli–injured lung lobe, n = 3 for E. coli–injured lung lobe treated with MSC IB or IV or normal human lung fibroblasts, *P < 0.0005 versus control CFU counts per milliliter by ANOVA (Bonferroni).

Figure 4.

Effect of allogeneic human mesenchymal stem cells (MSC) on alveolar macrophage phagocytosis of Escherichia coli bacteria. (A) A representative alveolar macrophage found from cytospin slides of bronchoalveolar lavage fluid from the injured alveolus under different treatment conditions: intrabronchial (IB) MSC, intravenous (IV) MSC, and IB keratinocyte growth factor. (B) The instillation of human MSC IB 1 hour after E. coli pneumonia significantly increased the percent phagocytosis and phagocytosis index of alveolar macrophages against E. coli bacteria at 6 hours. Although not statistically significant, the instillation of human MSC IV 1 hour after E. coli pneumonia increased the percent phagocytosis and phagocytosis index of alveolar macrophages against E. coli bacteria by 80% and 110%, respectively, at 6 hours. The percent phagocytosis and phagocytosis index were calculated by quantifying the number of alveolar macrophages containing bacteria out of 100 macrophages from the bronchoalveolar lavage fluid of three different lung preparations per condition and quantifying the average number of bacteria per macrophage in 100 macrophages containing the bacteria from three different lung preparations per condition, respectively. The values are expressed as mean ± SD for each condition. n = 3. *P < 0.05 for % phagocytosis and *P < 0.03 for phagocytosis index.

Figure 5.

Effect of antibiotics or human mesenchymal stem cells (MSC) on Escherichia coli pneumonia at 10 hours. We extended the ex vivo perfused human lung model injured with E. coli pneumonia to 10 hours to determine if MSC treatment would be therapeutic if given at a later time-point. We also increased the E. coli intrabronchial (IB) dose to 1010 CFU from 109 CFU to cause a transient bacteremia in the perfusate. Similar to the previous set of experiments at 6 hours, IB MSC restored all the parameters of lung injury when given 2 hours after the induction of E. coli pneumonia. (A) Loss of alveolar fluid clearance (AFC) is expressed as mean AFC (% per h, per 150 ml of bronchoalveolar lavage ± SD for each condition). n = 12 for control lobe, n = 3 for E. coli–injured lung lobe treated with or without administration of MSC or ampicillin IV ± MSC, *P < 0.0001 versus control lobe, √P < 0.0.0001 versus E. coli and #P < 0.0005 versus E. coli + IV Amp AFC by analysis of variance (ANOVA; Bonferroni). (B) Influx of neutrophils into the injured alveolus. Absolute neutrophil counts are expressed as mean total neutrophil counts (×107 cells) ± SD for each condition. (C) Increase in the total bacterial CFU counts in the injured alveolus. Total bacterial counts were expressed as mean (×106 CFU counts/ml) ± SD for each condition. *P < 0.002 versus control CFU counts per milliliter by ANOVA (Bonferroni). (D) Bacteremia present in the perfusate. The highest bacterial count found in the perfusate per hour was expressed as mean (×105 CFU counts/ml) ± SD for each condition. *P < 0.0001 versus control lobe, √P < 004 versus E. coli CFU counts per milliliter by ANOVA (Bonferroni). Treatment of E. coli pneumonia with ampicillin (0.2 g) IV at 1 hour restored most of the parameters of acute lung injury similar to MSC except for the restoration of AFC (A–D). The addition of IB MSC at 2 hours given after IV Amp at 1 hour after E. coli pneumonia had an additive effect in restoring AFC rate in the injured alveolus and decreasing the total bacterial load in the alveolus and perfusate (C and D).

Figure 6.

Effect of recombinant human keratinocyte growth factor (KGF) on Escherichia coli pneumonia. KGF instillation (100 ng) intrabronchially (IB) 1 hour after E. coli pneumonia restored many of the parameters of acute lung injury similar to mesenchymal stem cells (MSC). (A) Instillation of KGF IB restored the decrease in alveolar fluid clearance (AFC) in the lung lobe injured by E. coli pneumonia at 6 hours. AFC is expressed as mean AFC (% per h, per 150 ml of bronchoalveolar lavage) ± SD for each condition). n = 20 for control lobe, n = 3–4 for E. coli–injured lung lobe treated with or without KGF, *P < 0.009 versus control lobe AFC by analysis of variance (ANOVA; Bonferroni). (B) Instillation of KGF IB decreased the influx of neutrophils into the injured lung lobe at 6 hours. Absolute neutrophil counts are expressed as mean total neutrophil counts (× 107 cells) ± SD for each condition. n = 20 for control lobe, n = 4 for E. coli–injured lung lobe treated with or without KGF, *P < 0.0001 versus control lobe by ANOVA (Bonferroni), √P < 0.006 versus E. coli pneumonia by ANOVA (Bonferroni). (C) Instillation of KGF IB decreased the total bacterial load in the injured lung lobe at 6 hours. Total bacterial counts were expressed as mean (× 106 CFU counts/ml) ± SD for each condition. n = 18 for control lobe, n = 3 for E. coli–injured lung lobe with or without KGF, *P < 0.007 versus control CFU counts per milliliter by ANOVA (Bonferroni), √P < 0.0001 versus E. coli pneumonia CFU counts per milliliter by ANOVA (Bonferroni). (D) Instillation of KGF IB 1 hour after E. coli pneumonia significantly increased the percent phagocytosis and phagocytosis index of alveolar macrophages against E. coli bacteria at 6 hours. The values are expressed as mean ± SD for each condition. n = 3. *P < 0.03 for % phagocytosis and *P < 0.05 for phagocytosis index.

Figure 7.

Effect of recombinant keratinocyte growth factor (KGF) on peripheral blood monocytes. (A) The simultaneous addition of KGF (100 ng/ml) to isolated human blood monocytes increased Escherichia coli bacteria killing by 20%. Total bacterial counts were expressed as mean (% of control) ± SD for each condition. n = 14–16, *P < 0.0001 versus control CFU count. (B) By reverse-transcriptase polymerase chain reaction and Western blot analyses, human blood monocytes were found to express the mRNA and protein for FGFR2, the receptor for KGF. Primary cultures of human alveolar epithelial type II cells (TII) were used as a positive control. (C and D) The addition of KGF decreased lactate dehydrogenase (LDH) release after 24 hours, which was associated with an increase in intracellular AKT phosphorylation at 1 hour. LDH released was expressed as mean (% of total LDH) ± SD for each condition. n = 20–24, *P < 0.0008 versus control LDH release; the ratio of phosphorylated AKT/total AKT was expressed as mean (% of control) ± SD for each condition. n = 10, *P < 0.03 versus control P-AKT/Tot AKT ratio. The ratio of phosphorylated STAT5/total STAT5 was expressed as mean (% of control) ± SD for each condition. n = 10, *P > 0.05 versus control P-STAT5/Tot STAT5 ratio. GAPDH = glyceraldehyde phosphate dehydrogenase.

Figure 8.

Effect of human mesenchymal stem cells (MSC) on peripheral blood monocytes. (A) The simultaneous addition of human MSC, mixed together or separated by a Transwell plate, with isolated human blood monocyte increased Escherichia coli bacteria killing. Total bacterial counts were expressed as mean (% of control) ± SD for each condition. n = 8–15, *P < 0.0002 versus control CFU counts. (B) In addition, the level of tumor necrosis factor (TNF)-α was decreased and the level of IL-10 was increased, suggesting an antiinflammatory effect. The TNF-α level was expressed as mean (% of control) ± SD for each condition. n = 8–11, *P < 0.02 versus control and √P < 0.0001 versus MSC (Transwell) for TNF-α level. The IL-10 level was expressed as mean (% of control) ± SD for each condition. n = 8–11, *P < 0.03 versus control for IL-10 level. (C) Human MSC, whether mixed together or separated by a Transwell plate, with human blood monocytes increased granulocyte-macrophage colony–stimulating factor (GM-CSF) secretion by ×17 and ×6 from baseline. The GM-CSF level was expressed as mean (% of control) ± SD for each condition. n = 8–15, *P < 0.003 versus control and √P < 0.0001 versus MSC (Transwell) for GM-CSF level. (D) The addition of keratinocyte growth factor (KGF) or human MSC had the same antimicrobial effect on monocytes cultured for 7 days to take on the phenotype of alveolar macrophages. Total bacterial counts were expressed as mean (% of control) ± SD for each condition. n = 4, *P < 0.05 versus control CFU counts per milliliter and √P < 0.05 versus KGF CFU counts per milliliter.

Figure 9.

Effect of antihuman keratinocyte growth factor (KGF) or anti–granulocyte-macrophage colony–stimulating factor (GM-CSF) neutralizing antibody on the effect of human mesenchymal stem cells (MSC) on alveolar macrophages and peripheral blood monocytes. The inhibition of KGF by a neutralizing antibody abrogated the antimicrobial effect of human MSC on (A) alveolar macrophages in the ex vivo perfused human lung and (B) blood monocytes cultured in vitro. (C) Similar to anti-KGF Ab, the inhibition of GM-CSF by a neutralizing antibody abrogated the antimicrobial effect of human MSC in vitro. Total bacterial counts in the alveolar fluid for the perfused human lung were expressed as mean (× 106 CFU counts/ml) ± SD for each condition. n = 3–5, *P < 0.05 versus MSC CFU counts per milliliter treated with goat control IgG. Total bacterial counts for monocytes grown in vitro were expressed as mean (% of control) ± SD for each condition. n = 8–9. For anti-KGF Ab, *P < 0.05 versus MSC treated with goat control IgG. n = 14–16. For anti–GM-CSF Ab, *P < 0.04 versus MSC treated with goat control IgG.