Key Points
MSC EV may stimulate increased antimicrobial activity during bacterial pneumonia.
Increased antimicrobial activity is associated with increased LTB4 production.
MSC EV may increase LTB4 production via transfer of miR145 to target cells.
Abstract
Human mesenchymal stem cell (MSC) extracellular vesicles (EV) can reduce the severity of bacterial pneumonia, but little is known about the mechanisms underlying their antimicrobial activity. In the current study, we found that bacterial clearance induced by MSC EV in Escherichia coli pneumonia in C57BL/6 mice was associated with high levels of leukotriene (LT) B4 in the injured alveolus. More importantly, the antimicrobial effect of MSC EV was abrogated by cotreatment with a LTB4 BLT1 antagonist. To determine the role of MSC EV on LT metabolism, we measured the effect of MSC EV on a known ATP-binding cassette transporter, multidrug resistance–associated protein 1 (MRP1), and found that MSC EV suppressed MRP1 mRNA, protein, and pump function in LPS-stimulated Raw264.7 cells in vitro. The synthesis of LTB4 and LTC4 from LTA4 are competitive, and MRP1 is the efflux pump for LTC4. Inhibition of MRP1 will increase LTB4 production. In addition, administration of a nonspecific MRP1 inhibitor (MK-571) reduced LTC4 and subsequently increased LTB4 levels in C57BL/6 mice with acute lung injury, increasing overall antimicrobial activity. We previously found that the biological effects of MSC EV were through the transfer of its content, such as mRNA, microRNA, and proteins, to target cells. In the current study, miR-145 knockdown abolished the effect of MSC EV on the inhibition of MRP1 in vitro and the antimicrobial effect in vivo. In summary, MSC EV suppressed MRP1 activity through transfer of miR-145, thereby resulting in enhanced LTB4 production and antimicrobial activity through LTB4/BLT1 signaling.
Introduction
Leukotrienes (LTs) are a family of lipid mediators associated with acute and chronic inflammatory diseases. They act in host defense, intercellular communication, and signal transduction. LTs include LTB4 and cysteinyl LTs (CysLTs) (LTC4, LTD4, and LTE4) (1). CysLTs play a significant role in the pathogenesis of inflammatory-related diseases. In asthma, they mediate bronchoconstriction, bronchial hyperactivity, edema, and eosinophilia (2, 3). LTB4 was initially identified as an activator of granulocytes (4) and is known to exert broad proinflammatory effects (5, 6). Multiple studies have demonstrated the antimicrobial effects of LTB4, which may be mediated through augmented phagocytosis and the release of antimicrobial agents (7, 8). LTB4 can enhance host defense against pneumonia and sepsis (9–12). The biosynthesis of LTs occurs predominantly in leukocytes. In response to stimuli (immune and inflammatory signals), arachidonic acid (AA) is liberated from membrane phospholipids, catalyzed by phospholipase A2 (PLA2). AA is then converted to LTA4 via the action of 5-lipoxygenase. From LTA4 there are two competitive metabolic routes leading to the synthesis of LTB4 or LTC4 via the enzymes LTA4 hydrolase (LTA4H) or LTC4 synthase (LTC4S), respectively. LTC4 is transported across the plasma membrane by MRP1, also named ABCC1, and subsequently converted into LTD4 and LTE4 in the extracellular space.
MRP1 is a member of the ATP-binding cassette transporter superfamily along with multidrug resistance protein 1 (MDR1), also named ABCB1 or P-glycoprotein. MRP1 is a ∼190-kDa membrane protein and functions as a transporter. LTC4 is a particularly high-affinity substrate of MRP1 (13, 14), and MRP1 is the major efflux pump for LTC4 (15). In LT-mediated inflammation, MRP1 plays a crucial role in LTC4 release (9, 16). Surprisingly, mice lacking Mrp1 are more resistant to bacterial pneumonia compared with wild-type mice (9). Because of a lack of Mrp1 expression, efflux of LTC4 is impaired and intracellular LTC4 is accumulated (9). Accumulated LTC4 downregulates LTC4S thus increasing the availability of LTA4 for LTA4H. LTA4 thus becomes more available for conversion to LTB4, leading to enhanced LTB4 release (17). In addition, LTA4H, which can also be released into the extracellular space, possesses significant aminopeptidase activity against the matrikine proline–glycine–proline (PGP) (18). PGP is generated from the extracellular matrix (ECM) collagen, via enzymes such as matrix metalloproteinase (MMP)-9, followed by cleavage by prolyl endopeptidase (PE) (19). The degradation of PGP by LTA4H help facilitates the resolution of inflammation (18) (Fig. 1). The extracellular role of LTA4H in suppressing inflammation through PGP degradation has been reported in various pulmonary disorders (18, 20–22).
Human MSC extracellular vesicle (EV) have been studied in various acute lung injury models as a therapeutic because of their ability to reduce inflammation, decrease lung permeability, and reduce bacterial pneumonia (23–26). However, the mechanism underlying the antimicrobial activity of MSC EV remains largely unknown. MSC-derived EV are small membrane-bound vesicles originating from intracellular multivesicles or from budding off the plasma membrane. MSC and MSC EV contain numerous proteins and RNA. The RNA components carried by MSC EV are mainly small RNAs (<100 bp), including miRs (27–29). miRs regulate gene expression and function not only within the cells where they are transcribed but can be transferred to target cells to mediate gene expression and regulate cell function (30). miR-145 has been found in MSC cells (31, 32), MSC conditioned medium (CM) (29), and MSC-derived exosomes (27), and it is one of the top 10 most abundant miRs detected (27). Recently miR-145 was found to directly regulate MRP1 expression in breast cancer (33) and gallbladder cancer (34).
In the current study, we hypothesized that miR-145 present in MSC EV may reduce MRP1 expression in leukocytes and subsequently enhance the production of LTB4 and increase overall antimicrobial activity.
Materials and Methods
Mesenchymal stem cell extracellular vesicle isolation and characterization
Human MSC were obtained from a National Institutes of Health repository from Texas A&M Health Science Center (Temple, TX). Human bone marrow–derived MSC from three different donors with passages 3–8 were used in this study. These human MSC were well characterized (23, 35). Their capability for osteogenic, adipogenic, and chondrogenic lineage differentiation was previously demonstrated. Flow cytometry results showed >99% cells display the standard MSC surface markers such as CD105, CD90, CD73, CD44, and CD166. Their phenotype and function met all the criteria for MSC as defined by the International Society of Cellular Therapy (36). Normal adult human lung fibroblasts (NHLF) were used as cellular controls (Lonza Group, Basel, Switzerland).
EV were obtained from the supernatants of MSC and NHLF using ultracentrifugation as previously described (23, 37). Briefly, MSC or NHLF were cultured until confluent and then serum starved for 48 h in fresh CM without FBS but containing 0.5% BSA (MP Biomedicals, LLC, Solon, OH). To isolate EV, the CM of MSC or NHLF was centrifuged at 3000 rpm for 20 min to remove cellular debris and then at 100,000 × g (Beckman Coulter, Brea, CA) to sediment the EV for 1 h at 4°C. The EV were washed in PBS and then submitted to a second ultracentrifugation before being resuspended according to the final cell count after 48 h of serum starvation (10 μl per 1 × 106 cells).
MSC EV were characterized by morphology, size, protein, RNA content, and surface receptor. MSC EV were examined by electron microscopy for morphology and approximate size. Both size distribution (both mean and modal) and the concentration of MSC EV as reconstituted in PBS were analyzed by nanoparticle tracking analysis with NanoSight NS300 (Malvern Instruments, Malvern, U.K.). This technique calculates size based on tracking of Brownian motion and concentration by particle observation on a frame-by-frame basis by the high-sensitivity sCMOS camera. Data were analyzed with NTA2.3 software. The content of total protein and RNA of MSC EV were quantified by BCA protein assay, Western blot, and RT-PCR. To characterize the surface molecules, MSC EV were fluorescently labeled with PKH26 dye (Sigma-Aldrich, St. Louis, MO) or FITC-conjugated CD44 or CD9 (BD Biosciences, San Jose, CA) following the manufacturer’s instructions. FITC nonimmune isotypic IgG were used as a control. MSC EV were labeled with PKH26 to separate out EV from protein debris. To detect CD44 or CD9 on MSC EV, a BD FACSAria Fusion Special Order (SORP) cell sorter (BD Biosciences) with 100-nm nozzle and ND filter 1 was used. The threshold was set on the SSC 200. Collected data were analyzed by FACSDiva software (BD Biosciences). For fluorescence detection, we used a 586/15 band-pass filter for PKH26 and 525/50 band-pass filter for CD9, CD44, and their IgG controls. An unstained sample was used to detect autofluorescence and set the photomultiplier for all the considered channels. The instrument was rinsed with particle-free rinse solution before running samples and between each sample to eliminate the background. Standard silica beads (Apogee Mix for Flow Cytometer; Apogee Flow Systems, Hemel Hempstead, U.K.), with a similar refractive index of vesicles, were used as internal size standards and to determine the gate for analysis.
Models and assessment of acute lung injury
C57BL/6 male mice (10–12 wk old, ∼25 g; The Jackson Laboratory, Bar Harbor, ME) were used in all experiments. The Institutional Animal Care and Use Committee at the University of California, San Francisco, approved all experimental protocols. The mice were first anesthetized with ketamine (90 mg/kg) and xylazine (10 mg/kg) (i.p.).
E. coli endotoxin-induced acute lung injury.
Acute lung injury was induced by the intratracheal (IT) instillation of a nonlethal dose of endotoxin from E. coli O111: B4 (Sigma-Aldrich) at 4 mg/kg; MSC (750,000 cells per mice) were given simultaneously as treatment. PBS was used as the carrier control and NHLF as a cellular control (750,000 cells per mouse). In separate experiments, MK-571 (Cayman Chemical, Ann Arbor, MI), a nonspecific inhibitor of MRP1, was instilled (i.v.) after endotoxin instillation at doses of 15, 35, or 50 mg/kg. Mice were sacrificed at 12, 24, or 48 h. MK-571 is both a LTD4 antagonist and MRP1 inhibitor (38, 39).
E. coli pneumonia.
Acute lung injury was induced by the IT instillation of E. coli K1 strain (1.5 × 106 CFU). Four hours later, MSC EV were injected i.v. with 90 μl of MSC EV per mouse (1 × 1010 particles). NHLF EV was used as a negative control. In separate experiments, in addition to MSC EV treatment, LY293111 (Cayman Chemical), an inhibitor of LTB4 receptor 1, was instilled (i.p.) at a dose of 18 mg/kg. In additional experiments, Reversan (Sigma-Aldrich), a specific inhibitor of both MRP1 and MDR1, was instilled (i.p.) after E. coli instillation at a dose of 40 mg/kg. Mice were sacrificed at 24 or 28 h.
At the end of the experiment, bronchoalveolar lavage fluid (BALF) from the lungs were collected for assessment of leukocyte counts, cytokine levels, bacterial load, protein levels, and histology. Total WBC count and differential were obtained using the Hemavet HV950FS (Drew Scientific, Miami Lakes, FL). Mouse MIP-2, neutrophil chemokine (KC), TNF-α, LTB4, and PGE24/D4/E4 Biotrak Enzyme immunoassay System (GE Healthcare Life Science, Buckinghamshire, U.K.). For histologic analysis, the mouse lungs from each group were fixed in 4% paraformaldehyde and embedded in paraffin, cut into 5-μm sections, and stained with H&E.
miR-145 mimic and antagomir experiments
27). Briefly, MSC EV were incubated with Lipofectamine RNAiMAX-prepared miR-145 antagomir or antagomir negative control according to the protocol. The mixture was then incubated at 37°C for 24 h followed by ultracentrifugation to pellet the transfected MSC EV. Residual antagomir or antagomir control was removed by washing the MSC EV with PBS. Antagomir-transfected MSC EV were then cocultured with Raw267.4 cells. Raw267.4 cells were seeded in six-well plates and treated with LPS (100 ng/ml) and 90 μl of MSC EV (1 × 1010 particles), pretransfected with miR-145 antagomir or a negative control.
Multidrug resistance assay
A fluorometric multidrug resistance assay kit (Abcam, Cambridge, U.K.) was used to determine MRP1 activity. Mouse Raw264.7 cells (Sigma-Aldrich) were seeded into 96-well flat clear-bottom black-wall microplates and incubated for 24 h with or without stimulation with LPS (100 ng/ml). Cells were then treated with MSC CM, MSC CM with EV removed (CM supernatant), MSC EV (at doses of 2.5, 5, and 10 μl, equivalent to 3, 6, and 12 × 108 particles of MSC EV), MK-571: 1 μl (0.001 mM) and 2 μl (0.002 mM), or MSC EV transfected with miR-145 antagomir or its negative control. For miR-145 agonist experiments, Raw267.4 cells were pretreated with miR-145 mimic or its negative control. A total of 100 μl of MDR dye solution was added to each well and incubated at 37°C for 2 h in the dark. Intracellular fluorescence intensity, as an indicator of MDR pump activity after coincubation, was measured with a FLUOstar OPTIMA fluorescent plate reader (BMG Labtech, Cary, NC) at an excitation wavelength of 490 nm and an emission wavelength of 525 nm.
E. coli bacteria quantification
Bacterial growth in the BALF following E. coli pneumonia or in the supernatants of primary cultures of human monocytes exposed to E. coli bacteria with or without MSC EV was quantitated by counting CFU. The samples were cultured on LB agar plate (TEKnova, Hollister, CA) overnight at 37°C. Individual colonies (CFU) were then counted.
Phagocytosis of GFP-labeled E. coli bacteria by Raw267.4 cells
Raw267.4 cells were seeded in six-well plate and treated with miR-145 mimic or its negative control and LPS for 24 h. After washing, cells were incubated with GFP-labeled E. coli (1 × 107 CFU/well [25922GFP; American Type Culture Collection]) for 90 min. After washing, intracellular fluorescence intensity was measured with a FLUOstar OPTIMA fluorescent plate reader following lysis of the cells.
MRP1 Western blot analyses
RNA isolation and RT-PCR
Total RNA was isolated from MSC EV using the RNeasy Mini Kit (Qiagen Sciences, Germantown, MD). After isolation, RNA samples were treated with DNase I for 60 min at room temperature. The quality of the RNA was assessed with the NanoDrop ND-1000 UV-Vis Spectrophotometer (NanoDrop Technologies, Wilmington, DE) according to the manufacturer’s instructions. The primers used for RT-PCR were human Ang-1 and human KGF (SABiosciences, Qiagen, Valencia, CA). The RT-PCR assays were conducted using the SuperScript III One-Step RT-PCR protocol as described by the manufacturer (SABiosciences, Qiagen). For cDNA amplification, an initial reverse transcription step (52°C for 30 min) was followed by a denaturing step (94°C for 2 min) and then by 40 cycles of denaturing (94°C for 20 s), annealing (60°C for 30 s), and extending (68°C for 30 s), followed by 5 min at 72°C for elongation. GAPDH gene amplification was used as an internal control. The amplified DNA products were run on a 1.4% agarose gel, and bands were visualized with the use of ethidium bromide.
Quantitative real-time PCR
The average cycle threshold (Ct) value of two technical replicates was used in all calculations with specific gene expression assays. The average Ct value of the internal controls GAPDH or U6 was used to calculate ΔCt values for the array samples, as this combination of reference genes displayed the lowest SD among groups. GAPDH or U6 alone was used to calculate ΔCt values for specific gene expression assays. The initial data analysis was performed using the 2−ΔΔCt method.
LTA4H quantification
MSC EV were transfected with miR-145 antagomir or negative control for 24 h. A total of 90 μl (1 × 1010 particles) MSC EV was then cocultured with Raw267.4 cells in six-well plates stimulated with LPS (100 ng/ml). LTA4
Statistics
Data were presented as the mean ± SD or median with interquartile range (IQR). Shapiro–Wilk normality test and Kolmogorov–Smirnov test with Dallal-Wilkinson-Lillie for p value were used to determine if the values were from a Gaussian distribution. Comparisons between two groups were made using Student t test (for parametric data) or the Mann–Whitney U- test (for nonparametric data). Comparisons between more than two groups were made using one-way ANOVA with the Bonferroni correction (for parametric data) or Kruskal–Wallis test with Dunn correction (for nonparametric data) for multiple comparison testing. A p value <0.05 was considered statistically significant. All statistical analysis was performed using GraphPad Prism 6.0 software (GraphPad, San Diego, CA).
Results
Isolation and characterization of MSC EV
EV were isolated from the conditioned medium of human bone marrow–derived MSC. Human MSC were serum starved for 48 h. The viability of serum-starved MSC before EV isolation was >95% (trypan blue exclusion). Conditioned medium was collected, and MSC EV were isolated using ultracentrifugation as previously described (23, 24, 37). Scanning electron microscopy showed that the isolation technique yielded homogeneous population of spheroid particles the size of ∼200 nm (Supplemental Fig. 1A). NanoSight analysis from multiple samples showed MSC EV mean size was 227 ± 25 nm, and mode size was 146 ± 22 nm with a concentration of 1.2 ± 0.2 ×1011 particles per ml or 1.2 ± 0.2 ×108 particles per μl (Supplemental Fig. 1B).
Total protein and RNA content of MSC EV were 1.1 ± 0.3 μg (n = 5) and 0.9 ± 0.4 ng/μl EV (n = 5), respectively (Supplemental Fig. 1C), which falls into a range of concentrations found in previous studies (23, 24, 37). By RT-PCR, MSC EV expressed mRNA for angiopoietin 1 and keratinocyte growth factor, which are soluble factors secreted by MSC and known to participate in the therapeutic effects of MSC in acute lung injury (Supplemental Fig. 1D) (23, 24). Based on previous dose-response experiments (23), 90 μl of MSC EV per mice (1 × 1010 particles) was the optimal dose used in the in vivo experiments.
CD9 is a known marker on exosomes, and CD44 is a key plasma membrane receptor involved in MSC trafficking (40); we previously demonstrated that MSC EV expressed CD44 with Western blot analyses (23). For flow cytometry, we labeled MSC EV with PKH26 to quantify only membrane-bound vesicles and exclude protein debris. Flow cytometry analyses showed >73% (74 ± 8) of MSC EV expressed CD44, and <0.3% (0.3 ± 0.4) expressed CD9 (Supplemental Fig. 1E), suggesting that the majority of EV were microvesicles. Furthermore, similar to their parent MSC, MSC EV expressed CD44 on their surface, which was instrumental in incorporating MSC EV into target cells in previous studies (Fig. 1).
Role of LTs in the antimicrobial activity of MSC or MSC EV. A schematic illustration of the possible mechanisms by which MSC or MSC EV increase bacteria killing. Numbers 2–7 refer to subsequent figures as a mechanism of MRP1 inhibition by MSC or MSC EV.
MSC EV decreased MRP1 protein levels in LPS stimulated human monocytes
To determine whether MSC EV suppressed MRP1 expression and activity, we cultured normal adult human blood monocytes stimulated with LPS with or without MSC EV. Western blot analyses demonstrated a 190-kDa band corresponding to MRP1 protein using a mAb specific for MRP1 (no cross-reaction with P-glycoprotein) (Fig. 2A). Compared to controls, MSC EV suppressed MRP1 protein levels in human monocytes, which was associated with increased phagocytosis of E. coli bacteria (Fig. 2B).
MSC EV decreased E. coli bacterial CFU levels and MRP1 protein levels and MRP1 activity. (A) Administration of MSC EV decreased MRP1 protein level (Western blot) compared with controls, human monocytes exposed to LPS alone. n = 3. (B) Administration of MSC EV increased human monocyte phagocytosis of E. coli bacteria in vitro compared with PBS. The decrease in bacterial CFU counts was associated with decrease in MRP1 protein levels. n = 5. (C) A fluorometric MDR assay kit was used to determine MRP1 pump function; in the assay, cellular fluorescence increases if MRP1 activity is decreased. MSC CM decreased MRP1pump activity in Raw267.4 cells with or without LPS stimulation, suggesting that either a soluble factor or EV may be involved. n = 4–7. (D) MSC CM had a stronger effect on decreasing MRP1 pump activity in Raw267.4 cells compared with MSC CM supernatant (with EV removed). n = 6. (E) MSC EV decreased MRP1 pump activity in Raw267.4 cells stimulated with LPS in a dose-dependent manner. n = 5. (F) MK-571 decreased MRP1 pump activity in Raw267.4 cells stimulated with LPS in a dose-dependent manner. n = 3. (G) EV transfected with miR negative control decreased MRP1 pump activity. This effect was abolished when miR-145 antagomir was transfected into EV. n = 3. (H) Raw264.7 cells transfected with miR-145 mimic decreased MRP1 pump activity compared with that of negative control. n = 3–7. Data were expressed as mean ± SD. Student t test was performed on (A)–(D) and (H). One-way ANOVA with Bonferroni correction was performed on (E)–(G). *p < 0.05, **p < 0.01, ***p < 0.001. Antagomir, MSC EV transfected with miR-145 antagomir; Ctrl, Raw cells transfected with miR mimic negative control; Mimic, Raw cells transfected with miR-145 mimic; MK, MK-571; Neg Ctrl, MSC EV transfected with miR antagomir negative control; Raw, Raw267.4 cells.
MRP1 functional assay
To determine if MSC EV suppressed MRP1 pump function, a fluorometric MDR assay kit was used. PGP (MDR1) and MRP1 are members of the ABC transporter family; they are cell membrane proteins and function as ATP-dependent drug efflux pumps. This MDR assay kit uses a fluorescent MDR indicator dye to assay these two pump activities. The hydrophobic fluorescent dye rapidly penetrates cell membranes and becomes trapped in cells. In the MDR1- and/or MRP1-expressing cells, this dye is extruded by MDR transporters, thus decreasing cellular fluorescence intensity. However, when MDR1 and/or MRP1 pump activity are suppressed, their intracellular dye cannot be pumped out, thus increasing intracellular fluorescence intensity. A mouse macrophage cell line Raw267.4, which highly expresses MRP1 upon stimulation with LPS, was used in this assay. We cultured Raw267.4 stimulated with LPS with or without MSC CM or MSC EV. Similar to MRP1 protein expression, administration of MSC CM downregulated MRP1 pump activity on Raw267.4 cells as evidenced by increased intracellular fluorescence intensity compared with PBS (Fig. 2C). MSC CM that contains EV had a stronger effect on suppressing MRP1 pump activity compared with CM with EV removed (named CM supernatant) by ultracentrifugation but containing soluble factors (Fig. 2D). MSC EV also inhibited MRP1 activity in a dose-dependent manner (Fig. 2E) at dosages of 2.5, 5, and 10 μl, (equivalent to 3, 6, and 12 ×108 particles of MSC EV). We then compared MSC EV to MK-571, a nonspecific MRP1 inhibitor, in the MDR pump assay. MSC EV (5 μl, equals 6 × 108 particles) had a similar effect on inhibiting MDR pump activity as MK-571 (2 μl, equals 0.002 mM) (Fig. 2F). We also studied the role of miR-145 by transfection of EV with miR-145 antagomir (synthetic miR inhibitor that degrades miR-145). Similar to EV in Fig. 2E and 2F, EV transfected with miR negative control showed increased fluorescent intensity (Fig. 2G). However, EV transfected with miR-145 antagomir abolished this effect. In contrast, transfection of Raw264.7 cells with a miR-145 mimic (synthetic miR-145) showed increased fluorescent intensity compared with that of negative control (Fig. 2H).
MSC or MSC EV treatment increased LTB4 levels in BALF of mice with acute lung injury
We measured LTB4 levels in mice with acute lung injury from LPS or E. coli bacteria. Baseline level of LTB4 in BALF was low but increased with stimulation with LPS or E. coli bacteria (data not shown). MSC or MSC EV treatment following acute lung injury resulted in significantly higher LTB4 level in BALF compared with PBS treatment (Fig. 3). The level of LTB4 was significantly higher in MSC or MSC EV treatment groups compared with NHLF or NHLF EV groups, which was used as cellular controls. LTB4 level was increased in the plasma after MSC treatment (Fig. 3A). We also measured CysLTs (LTC4D4E4) level in BALF in mice with acute lung injury from LPS or E. coli bacteria. There were no significant differences in CysLTs level in all groups.
MSC or MSC EV administration increased LTB4 levels in the BALF of mice injured with LPS or E. coli bacterial-induced lung injury. (A) Treatment with MSC significantly increased LTB4 levels in BALF of mice injured with IT LPS at 48 h (Kruskal–Wallis test with Dunn correction). Plasma levels of LTB4 were significantly increased with administration of MSC as well (one-way ANOVA with Bonferroni correction). There were no significant differences in CysLT level in all groups (Kruskal–Wallis test with Dunn correction). n = 5–8. (B) Administration of MSC EV as therapy significantly increased LTB4 levels in BALF of mice injured with IT E. coli pneumonia at 24 h. Administration of NHLF or its released EV had no significant effect on LTB4 levels. Data were mean ± SD or median with IQR. There were no significant differences in CysLT level in all groups. n = 5–13. Kruskal–Wallis test with Dunn correction. *p < 0.05, ***p < 0.001.
PGE2/LTB4 ratio in acute lung injury mice treated with MSC or MSC EV
Similar to LTB4, baseline level of PGE2 in BALF was low but increased with stimulation with LPS or E. coli bacteria. Unlike LTB4, the level of PGE2, a lipid mediator with anti-inflammatory properties known to be secreted by MSC, was unchanged in the BALF after MSC or MSC EV treatment compared with PBS-treated mice injured with LPS or E. coli (data not shown), which is consistent with our previous study (41). However, the ratio of PGE2 (nanogram per milliliter)/LTB4 (picogram per milliliter) decreased with treatment with MSC or MSC EV (Supplemental Fig. 2A, 2B).
MRP1 inhibitor reduced lung injury
MK-571, a commonly used nonspecific MRP1 inhibitor, was administered in LPS-induced acute lung injury in mice to corroborate the importance of MRP1 activity during lung injury. Compared with PBS-treated, LPS-injured mice, MK-571 administration significantly reduced WBC and neutrophil counts (Fig. 4A) and proinflammatory cytokines (TNF-α, MIP-2, and KC) levels (Fig. 4B) and demonstrated a trend toward decreased protein levels in BALF (Fig. 4C). These results suggested that MRP1 inhibition by MK-571 had significant anti-inflammatory properties.
Administration of MK-571 decreased LPS-induced acute lung injury. MK-571 administration significantly (A) reduced the influx of inflammatory cells, (B) downregulated proinflammatory cytokines (TNF-α, MIP-2, and KC), and was associated with a trend in reducing (C) total protein levels in the injured alveolus in a dose-dependent manner. Compared to LPS-injured mice, MK-571 treatment increased LTB4 levels at 24 h (D), which was associated with reduced CysLT levels (E). Data were expressed as mean ± SD or median with IQR, n = 5–11 (12-h time point) and 7–14 (24-h time point). One-way ANOVA with the Bonferroni correction analyses was performed for WBC, NE, MIP-2, 12 h TNF-α, 24 h KC, 12-h protein concentration, 24-h LTB4, 12-h CysLTs; Kruskal–Wallis test with Dunn correction analyses was performed for 24-h TNF-α, 12-h KC, 24-h protein concentration, 12-h LTB4, 24-h CysLTs. *p < 0.05, **p < 0.01, ***p < 0.001.
LTB4 level in BALF was barely detected in all groups at 12 h, whereas higher levels of LTB4 were observed only in MK-571–treated mice compared with PBS-treated, LPS-injured mice at 24 h (Fig. 4D). In contrast, CysLTs levels in all groups were higher at 12 h than that at 24 h (Fig. 4E). Compared to PBS-treated, LPS-injured mice, MK-571 administration significantly reduced CysLTs production at 12 h (Fig. 4E). These results suggested that MK-571 may inhibit MRP1-mediated transport of LTC4 and thus reduce CysLTs and increase LTB4 production. CysLTs are proinflammatory lipids that can increase vascular permeability.
A more selective MRP1 inhibitor, Reversan, was used in mice with E. coli pneumonia. Administration of Reversan significantly reduced the bacterial CFU counts (Fig. 5A) and the total WBC and neutrophil counts (Fig. 5B) in the BALF following E. coli pneumonia.
Role of MRP1 Inhibition and LTB4 in the antimicrobial activity of MSC EV. Administration of Reversan, a specific inhibitor of MRP1, in mice with E. coli pneumonia decreased (A) total bacterial CFU counts and (B) the influx of inflammatory cells, corroborating the effects seen with MK-571. Administration of LY293111, a LTB4 receptor 1 antagonist, reduced the therapeutic effects of MSC EV in mice injured with E. coli pneumonia in terms of reducing (C) total bacterial CFU counts, (D) the influx of inflammatory cells, and (E) inflammatory cytokines/chemokines in the BALF. Data were expressed as mean ± SD or median with IQR, n = 6–9. Student t test was performed for (A) and (B); one-way ANOVA with the Bonferroni correction was performed for (C) and (D), WBC count, and (E), Kruskal–Wallis test with Dunn correction (D, NE count). *p < 0.05, **p < 0.01, ***p < 0.001. D, DMSO; LY, LY293111.
Administration of MRP1 inhibitors (MK-571 and Reversan) behaved in a similar manner as MSC or MSC EV in mice with acute lung injury, suggesting that MSC or MSC EV may share a common mechanism with MK-571 or Reversan (i.e., inhibition of MRP1 activity). Similar to MSC or MSC EV treatment, PGE2/LTB4 ratio was lower in MK-571–treated mice although not statistically significant (Supplemental Fig. 2C).
LY293111, a LTB4 BLT1 antagonist, inhibited the effect of MSC EV on E. coli bacterial levels
To determine if the antimicrobial effect of MSC EV was mediated by LTB4, LY293111, a specific antagonist of LTB4 BLT1, was administered in mice with E. coli pneumonia treated with MSC EV.
Mice were infected with E. coli bacteria, and 4 h later, MSC EV alone or together with LY293111 were administrated to the mice. Treatment with MSC EV significantly reduced the bacterial CFU counts in the BALF compared with PBS treatment. However, administration of MSC EV together with LY293111 abolished this effect (Fig. 5C). MSC EV treatment also reduced the total inflammatory cells and levels of TNF-α and MIP-2 in the BALF, which was abolished by LY293111 cotreatment (Fig. 5D, 5E).
MiR-145 regulates the expression of MRP1
We next determined whether MSC or MSC EV downregulated MRP1 activity through the transfer of miR-145. We first cocultured LPS-stimulated Raw267.4 cells with MSC pretreated with or without miR-145 antagomir. Real-time PCR showed that MRP1 mRNA expression decreased with MSC treatment. However, the effect of MSC on MRP1 suppression was abolished when MSC was pretreated with miR-145 antagomir (Fig. 6A).
miR-145 regulates the expression of MRP1 and the phagocytic activity of Raw267.4 cells. (A) Administration of MSC suppressed MRP1 expression in part by transfer of miR-145. Similar to MSC, administration of MSC EV significantly increased (B) miR-145 levels, which was associated with a decrease in (C) MRP1 expression. MSC EV transfected with miR-145 antagomir eliminated the effect of MSC EV on MRP1 expression (D). Raw264.7 cells transfected with miR-145 mimic increased miR-145 levels, which was associated with a decrease in (E) MRP1 expression. (F) Raw267.4 cells transfected with miR-145 mimic had increased phagocytosis activity against GFP-labeled E. coli bacteria. (G) Representative images of phagocytosis of GFP-labeled E. coli bacteria by Raw267.4 cells, pretransfected by miR-145 mimic or negative control. Scale bars, 100 μm. n = 3–5 for (A)–(E). n = 9 for (F). Data were expressed as mean ± SD. One-way ANOVA with the Bonferroni correction was performed for (A)–(C), and Student t test was performed for (D)–(F). **p < 0.01, ***p < 0.001.
To confirm the effect of MSC EV on MRP1 suppression was due to transfer of miR-145 by MSC EV, we transfected MSC EV with miR-145 antagomir or negative control prior to coculture. MiR-145 level in the EV was reduced by miR-145 antagomir (Fig. 6B). More importantly, compared with PBS treatment, MSC EV transfected with negative control significantly decreased MRP1 expression, whereas MSC EV with miR-145 knockdown (miR-145 antagomir) failed to do so (Fig. 6C). These results suggested that the transfer of miR-145 from MSC EV to target Raw267.4 cells contributed to the suppression of MRP1 expression.
We then determined if increased miR-145 expression in Raw264.7 cells regulated the expression of MRP1. Raw267.4 cells were transfected with miR-145 mimic or negative control. Real-time PCR showed that miR-145 level was significantly increased by miR-145 mimic (Fig. 6D), and MRP1 expression decreased accordingly (Fig. 6E). Our results suggested that Mrp1 was the target of miR-145, and miR-145 negatively regulated MRP1 expression.
MiR-145 mimic increased the phagocytosis of E. coli bacteria by Raw267.4 cells
We examined whether increased miR-145 expression increased the phagocytic capacity of Raw267.4 cells against GFP-labeled E. coli bacteria. Raw267.4 cells transfected with miR-145 mimic showed increased fluorescent intensity compared with that of mimic negative control (Fig. 6F, 6G).
Knockdown of miR-145 in MSC EV inhibited the effect of MSC EV on E. coli bacterial levels
We next determined whether miR-145 antagomir treatment would have an impact on in vivo antimicrobial activity of MSC EV. MSC EV pretreated with either miR-145 antagomir or negative control was administrated to E. coli pneumonia–injured mice. As shown in Fig. 7A, IT administration of E. coli bacteria resulted in a significant bacterial load at 28 h in the BALF, which was reduced with MSC EV therapy. However, antagomir pretreatment of MSC EV increased CFU counts more than 3-fold over MSC EV negative control treatment. In addition, the beneficial effects of MSC EV treatment on reducing the total WBC and neutrophil counts and levels of TNF-α and protein concentration in the BALF was abolished by miR-145 antagomir treatment (Fig. 7B–D). Thus, knockdown of miR-145 in MSC EV significantly decreased the in vivo antimicrobial and anti-inflammatory effect of MSC EV. Our results suggested that miR-145 level was inversely correlated with MRP1 at the levels of mRNA (Fig. 6A–E), protein function (Fig. 2G, 2H), and phagocytosis (Fig. 6F, 6G). The therapeutic activity of MSC EV in acute lung injury was mediated via miR-145 (Fig. 7) by downregulating MRP1 function.
Knockdown of miR-145 in MSC EV inhibited the antimicrobial effect of MSC EV on E. coli bacterial growth. Compared with negative control–transfected MSC EV (Neg Ctrl), miR-145 antagomir–transfected MSC EV (Antagomir) reduced the therapeutic effects of MSC EV in terms of reducing (A) total bacterial CFU counts, (B) the influx of total WBC and neutrophils, (C) total protein concentration, and (D) TNF-α in the BALF. (E) Representative H&E staining of lung sections. Scale bars, 50 and 10 μm (insert). Data were expressed as mean ± SD, n = 6–7. One-way ANOVA with the Bonferroni correction. *p < 0.05, **p < 0.01, ***p < 0.001.
Effect of MSC EV on extracellular LTA4H and MMP-9 expression
Extracellular LTA4H can reduce inflammation by degrading PGP, which is generated from ECM collagen via enzymes such as MMP-9. We measured LTA4H and MMP-9 protein levels in the supernatant of injured Raw267.4 cells coincubated with MSC EV (pretreated with miR-145 antagomir or negative control). LTA4H was elevated at 14 h in MSC EV negative control treatment group compared with PBS (Supplemental Fig. 3A). MMP-9 protein level was found to be elevated in all three groups over 24 h (data not shown). However, at 24 h, treatment with MSC EV negative control significantly reduced MMP-9 mRNA expression and demonstrated a trend toward reduced MMP-9 protein expression (Supplemental Fig. 3B, 3C). The effect of increasing LTA4H and decreasing MMP-9 expression was abolished when MSC EV was pretreated with miR-145 antagomir (Supplemental Fig. 3A–C).
Discussion
Our study demonstrated that (1) MSC and MSC EV administration led to significantly higher LTB4 levels in the BALF in both LPS-induced acute lung injury and E. coli pneumonia in mice (Fig. 3) (2); MSC EV had significant antimicrobial activity in mice with E. coli pneumonia, which was mediated in part by LTB4 activity through BLT1 receptor (Fig. 5C) and which was abolished by the knockdown of miR-145 (Fig. 7A) (3); MSC EV treatment decreased MRP1 mRNA expression (Fig. 6C), protein levels (Fig. 2A), and pump activity (Fig. 2E) (these effects were abolished by knockdown of miR-145) (Figs. 2G, 6C) (4); similar to MSC EV, administration of MRP1 inhibitors (MK-571 or Reversan) decreased the influx of inflammatory cells (Figs. 4A, 5B), bacteria (Fig. 5A), and LTC4 (Fig. 4E), and subsequently increased LTB4 levels (Fig. 4D) in mice with acute lung injury (5); and similar to MSC EV, Raw267.4 cells transfected with the miR145 mimic had decreased MRP1 mRNA expression (Fig. 6E), MRP1 pump activity (Fig. 2H), and enhanced phagocytosis activity against E. coli bacteria (Fig. 6F). Taken together, our study revealed for the first time, to our knowledge, that MSC EV suppressed MRP1 expression through transfer of miR-145, which resulted in reduced extracellular level of LTC4 and enhanced LTB4 production Increased LTB4 levels then mediated increased antimicrobial activity through LTB4/BLT1 signaling (Fig. 1).
We and other investigators have demonstrated the therapeutic effects of MSC or MSC EV in acute lung injury models of inflammation and infection (23–26). We showed that MSC EV reduced the total bacterial load in both mice (23) and ex vivo perfused human lung (26) injured with E. coli pneumonia. However, the mechanisms underlying the antimicrobial activity of MSC EV remain largely unknown. Recent studies have demonstrated significant antimicrobial effects of LTB4 in pneumonia and sepsis (9–12). These studies provided the biological rationale for our hypothesis that the antimicrobial effect of MSC or MSC EV in acute lung injury were in part through increased LTB4 activity. High level of LTB4 in BALF was detected in both MSC EV–treated E. coli pneumonia mice and MSC-treated LPS-induced acute lung injury mice (Fig. 3). MSC EV treatment also showed improved bacterial clearance in E. coli pneumonia mice. However, this effect was abrogated by cotreatment with LY293111 (Fig. 5C), suggesting that the LTB4/BLT1 axis was a key determinant in the antimicrobial effect of MSC EV.
To determine if MSC or MSC EV reduced MRP1 expression and thus increased LTB4 levels, we first studied the effect of MSC EV on MRP1 protein expression and function in vitro. Western blot analyses showed that MRP1 expression on LPS-stimulated monocytes was reduced after MSC EV treatment (Fig. 2A), which was associated with increased antimicrobial activity (Fig. 2B). MDR assay showed that MSC CM or isolated MSC EV increased intracellular fluorescence intensity in LPS-stimulated Raw267.4 cells, which resulted from decreased MRP1 pump activity (Fig. 2C–E).
We then administered MK-571 to mice with LPS-induced acute lung injury to explore whether the inhibition of MRP1 affected the production of LTs. Similar to MSC EV, administration of MK-571 inhibited the influx of inflammatory cells and reduced proinflammatory cytokines in BALF in a dose-dependent manner (Fig. 4A, 4B). Efflux of LTC4 was impaired after MK-571 treatment as indicated by decreased CysLT levels at 12 h (Fig. 4E), which was associated with an increase in LTB4 levels at 24 h (Fig. 4D). A more selective MRP1 inhibitor, Reversan, was used in mice with E. coli pneumonia. Similar to MSC EV, Reversan significantly reduced the bacterial CFU counts and the infiltration of inflammatory cells in the BALF (Fig. 5A, 5B). Administration of MK-571 or Reversan behaved in a similar fashion as MSC or MSC EV, suggesting that they may share a common mechanism (i.e., the inhibition of MRP1). However, compared with MK-571, MSC EV treatment resulted in significantly higher LTB4 levels in either LPS-injured or E. coli pneumonia in mice (Figs. 3, 4D). This suggested that MSC or MSC EV may have a stronger effect on MRP1 suppression than MK-571 or there were other mechanisms in which MSC or MSC EV may enhance the production of LTB4. The MDR pump assay showed that MSC EV had a similar effect as MK-571 on inhibiting MDR pump activity (Fig. 2F). Thus, there might be some other signals involved in LTB4 production. We previously demonstrated that MSC itself secreted LL-37 (42), a peptide with broad-spectrum bactericidal activity. LTB4 is known to interact with LL-37. LTB4 can trigger the release of LL-37 in a BLT1-dependent way. Meanwhile, LL-37 can cause the translocation of the enzyme 5-lipoxygenase to the perinuclear membrane, which can promote the production of LTB4. LL-37 can also enhance LTB4-induced phagocytosis (43, 44). In the current study, MSC EV–suppressed MRP1 activity was associated with enhanced production of LTB4. LTB4 may work with LL-37 in a synergistic fashion to suppress inflammation and decrease bacteria growth (Fig. 1).
Although LTs are secreted predominantly by leukocytes, treatment with MSC, MSC EV, and MK-571 reduced the influx of inflammatory cells (23–26, 45), which was associated with increased LTB4 levels in the BALF (Figs. 3, 4). There are several possibilities why LTB4 level can be high with low leukocyte numbers. Studies have showed that MSC can inhibit neutrophil apoptosis, prolong leukocyte survival, and enhance their function (46, 47), possibly leading to enhanced antimicrobial activity, decreasing the trafficking of immune cells to the injured alveolus. In addition, the actual conversion to LTB4 may result from trans-cellular biosynthesis. The synthesis of LTB4 requires two steps from AA via individual enzymes. Once LTA4 is produced by inflammatory cells, it can be transferred to LTA4H-containing cells to generate LTB4 or LTC4S-containing cells to generate LTC4 (48–52), possibly reducing the overall production of LTB4 by neutrophils. In addition, AA can also be transferred between cells (52–54).
MSC and MSC EV showed similar protective effects on acute lung injury both in vivo and in vitro by decreasing inflammation and lung permeability and enhancing antimicrobial activity (23–26, 45). To understand the mechanism, we studied the activity of intracellular or vesicular miR-145, as it was found in both MSC and MSC EV (29, 31), and it was previously shown to regulate MRP1 expression (33, 34). By qPCR and pump activity assay, MRP1 mRNA expression and MRP1 pump function on LPS-stimulated Raw267.4 cells were downregulated with MSC EV treatment, and this effect was eliminated by knockdown of miR-145 (Figs. 2G, 6C). MSC EV treatment improved bacterial clearance in E. coli pneumonia mice, and this effect was abrogated by miR-145 knockdown (Fig. 7A). In corroboration, Raw267.4 cells, transfected with miR145 mimic, had decreased MRP1 mRNA expression (Fig. 6E), decreased pump activity (Fig. 2H), and improved bacterial clearance (Fig. 6F). Our data suggested that miR-145 is a key regulator of MRP1 expression and function, in agreement with recent studies (33, 34) that miR-145 directly regulated MRP1 expression by increasing MRP1 mRNA degradation by targeting its 3′-untranslated region of the Mrp1 gene.
Extracellular LTA4H was reported to help facilitate the resolution of inflammation in various pulmonary disorders through the degradation of PGP, which is generated from ECM collagen via MMP-9 (18, 20–22). We previously showed that MMP-9 level in the BALF was significantly reduced in MSC-treated acute lung injury mice compared with control (25). In the current study, we showed that MSC EV treatment led to increased LTA4H levels and decreased MMP-9 mRNA expression. These effects were miR-145 dependent (Supplemental Fig. 3A–C). Increased extracellular LTA4H and reduced MMP-9 may be involved in the anti-inflammatory effect of MSC EV as well (Fig. 1).
AA can also be metabolized to PGH2 through the enzyme cyclooxygenase, which is then converted to PGE2 and transported outside via multidrug resistance protein 4. Inhibition of PGE2 may decrease bacterial growth. A recent study showed that the protective effect of adipose tissue–derived MSC against Pseudomonas aeruginosa pneumonia was in part by inhibiting PGE2, which subsequently increased macrophage phagocytosis (55). In our study, PGE2 level did not change after MSC EV treatment in E. coli pneumonia in mice, suggesting the therapeutic effect may not be attributed to reduced PGE2 level alone. It remains controversial whether PGE2 is beneficial or plays a negative role in infection (55–58). Previous studies showed that LTB4 can enhance, whereas PGE2 can inhibit phagocytosis of bacteria by alveolar macrophage (11, 59, 60) via cAMP (11, 59). In addition, other studies have shown that increased PGE2/LTB4 ratio was associated with increased Leishmania infantum infection (61). MSC and PGs can both affect ABC transport protein expression and function as well (62–65). For example, PGE2 can induce MRP1 mRNA expression in human bronchial epithelium (63). Thus, the ratio between PGE2 and LTB4 may reflect MRP1 levels and overall antimicrobial activity. Our data showed that decreased PGE2/LTB4 ratio was associated with the therapeutic effects of MSC or MSC EV (Supplemental Fig. 2). The importance of PGE2 and LTB4 on inflammation and infection with MSC therapy needs further study.
There are some limitations to the current study that require further study: 1) We cannot exclude the possibility that other target gene(s) besides Mrp1 might be involved in the therapeutic effect of MSC EV via miR-145. Recently, Shinohara and his colleagues (66) showed that miR-145 within cancer EV can target HDAC11 and promote IL-10 expression and M2 macrophage polarization. Our previous studies showed that MSC or MSC EV administration can upregulate IL-10 mRNA or protein expression (23, 41). Whether there is some connection between MSC EV miR-145 and IL-10 production will need to be investigated; 2) LY293111, the antagonist of BLT1, only partially reduced the antimicrobial and anti-inflammatory activity of MSC EV (Fig. 5C–E), suggesting other LT-derived factors may be involved in the anti-inflammatory effect. We and others have recently found that human MSCs promoted the resolution of LPS-induced acute lung injury or following polymicrobial sepsis in mice in part through the secretion of specialized proresolving lipid mediators such as lipoxin A4 or resolvins (67, 68). Lipoxin A4 is derived directly from LTA4 and has anti-inflammatory properties in both sterile and infectious animal models of injury (67, 69). Studies are ongoing whether EV derived from MSCs can generate specialized proresolving lipid mediators, similar to the parent cells, and what role they have if any in the overall therapeutic effect; 3) and the MDR pump assay assessed both MDR1 and MRP1 function, not MRP1 alone. The effect of MSC EV on MDR1 activity in the overall therapeutic effect will need to be explored.
In summary, MSC EV suppressed MRP1 expression in part through the transfer of miR-145, which resulted in enhanced LTB4 production and subsequently increased bacterial phagocytosis through LTB4/BLT1 signaling. This represents a previously undescribed mechanism underlying the antimicrobial activity of MSC or MSC EV.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We thank Dr. Airan Liu for help with MSC EV analyses, Dr. Hyungsun Lim for assistance with the animal experiments, and Dr. Stephane Gennai for assistance with the preparation of MSC EV.
Footnotes
This work was supported by Grant HL-113022 from the National Heart, Lung and Blood Institute at the National Institutes of Health (to J.W.L.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- AA
- arachidonic acid
- BALF
- bronchoalveolar lavage fluid
- CM
- conditioned medium
- Ct
- cycle threshold
- CysLT
- cysteinyl LT
- ECM
- extracellular matrix
- EV
- extracellular vesicle
- IQR
- interquartile range
- IT
- intratracheal
- LTA4H
- LTA4 hydrolase
- LTC4S
- LTC4 synthase
- MDR1
- multidrug resistance protein 1
- MMP
- matrix metalloproteinase
- MRP1
- multidrug resistance–associated protein 1
- MSC
- mesenchymal stem cell
- MSC CM
- MSC conditioned medium
- MSC EV
- MSC extracellular vesicle
- NHLF
- normal adult human lung fibroblast
- PGE2
- prostaglandin E2
- PGP
- proline–glycine–proline.
- Received November 26, 2018.
- Accepted July 30, 2019.
- Copyright © 2019 by The American Association of Immunologists, Inc.
References
In this issue
