Extracellular vesicles-transmitted miR-21a-5p altered microglia polarization after hypoxia- ischemic injury in neonatal mice via STAT3 pathways


 Background We previously reported that mesenchymal stromal cells (MSCs)-derived extracellular vesicles (EVs) exhibit protective effects in hypoxia-ischemia (HI) brain damage. The neuroprotective action was connected with its anti-inflammatory effect. However, the mechanisms involved with this effect have not been determined. Methods A modified version of the Rice-Vannucci method was performed on postnatal day 7 mouse pups to induce neonatal HI brain injury. The model of oxygen-glucose deprivation (OGD) was established in BV-2 cells to mimic HI injury in vitro. Mice or BV-2 cells received EVs and EVs-miR-21ainhibitor at indicative time post-injury. In vivo, brain water content and TTC staining were used to evaluate the effects of EVs on HI brain injury. Immunofluorescence staining was used to observe the effect of EVs on the polarization of microglia. The effect of EVs on p-STAT3 was assessed by Western blot. In vitro, the effect of EVs on cell survival was evaluated by CCK8. Expression of miR-21a-5p and inflammatory factors was measured using qRT-PCR. Dual-Luciferase Reporter Assay was performed to illustrate the link between miR-21a-5p and STAT3. The role of miR-21a-5p in EVs on HI injury and ODG injury was further investigated by using EVs-miR-21ainhibitor.Results By using OGD mimicking HI injury in vitro, we found that MSCs-EVs treatment elevated cell viability following OGD exposure in BV-2 cells. MSCs-EVs treatment impeded microglia-mediated neuroinflammation, shifted microglia toward M2 polarization, and suppressed the phosphorylation of selective signal transducer and activator of transcription 3 (STAT3) in microglia after HI exposure in vitro and in vivo. In light of miR-21a-5p being the most highly expressed miRNA in MSCs-EVs interacting with the STAT3 pathway, further work focused on this pathway. Notably, MSCs-EVs treatment increased HI-reduced miR-21a-5p levels in BV-2 cells. Diminishing miR-21a-5p in MSCs-EVs partially attenuated its effect on microglia polarization and STAT3 phosphorylation following HI exposure in vitro and in vivo. Conclusions Our study suggested that MSCs-EVs attenuated HI brain damage in neonatal mice via shuttling miR-21a-5p, which induced microglia M2 polarization by targeting STAT3.


Introduction
Neonatal hypoxia-ischemia (HI) brain damage is one of the major causes of death and/or long-term neurodevelopmental disabilities [1]. Evidence suggests that neuroin ammation via releasing proin ammatory mediators lead to the alteration of neuronal functions and contribute to persistent HI brain damage [2,3]. As the main effectors of neuroin ammation following HI insult, microglia can be polarized into proin ammatory M1 or anti-in ammatory M2 [4]. The balance between the M1 phenotype and the M2 phenotype plays an important role in regulating neuroin ammation and brain homeostasis.
Mesenchymal stromal cells (MSCs) possess broad immunoregulatory properties and could modulate in ammation-related diseases [5]. For example, human umbilical cord-derived MSCs exerted anti-diabetic effects and alleviated islet dysfunction in type 2 diabetic mice by switching macrophages from M1 phenotype to M2 phenotype [6]. Evidence suggests the immunomodulatory properties of MSCs related to its paracrine effect [7]. For example, stimulation of MSCs with lipopolysaccharide (LPS) can secrete certain factors to in uence macrophage function [5,8]. Recently, extracellular vesicles (EVs), recognized as an important MSCs paracrine factor, contribute to the positive effects of MSCs [9]. EVs were classi ed into small EVs (50-100 nm), medium EVs (100 nm-1 μm) and large EVs (1 μm-5 μm) according to their size or density [10]. MSCs-EVs may promote an immunosuppressive response through the induction of immature dendritic cell, switch macrophages from M1 toward M2 phenotype, secreting anti-in ammatory cytokines [9]. The latest research shows that MSCs-EVs play important roles in ischemic brain damage.
Within our own laboratory, we have found that MSCs-EVs administration exerts the neuroprotective and anti-in ammatory effect on HI injury in neonatal mice [16,17]. However, its underlying mechanism is still unknown.
In the present study, we used HI injury in neonatal mice in vivo and oxygen-glucose deprivation (OGD) mimicking HI injury in vitro. We found that miR-21a-5p in MSCs-EVs has dramatic effects on the regulation of microglia polarization to resolve neuroin ammation following HI insult by targeting selective signal transducer and activator of transcription 3 (STAT3).

Materials
Reagents and primer sequences are listed in Table S1 and Table S2 of supplementary material respectively.

HI model and treatments
The HI model was induced on postnatal day 7 (P7) according to the methods of Rice et al, as described in our previous publication [17]. Brie y, C57BL/6J male mouse pups (P7) were anesthetized with 2% iso urane and the right common carotid artery was separated and ligated. After skin suturing and disinfection, the pups were put back to their dam for 60 min, and then exposed in a hypoxia chamber (humidi ed at 8% O 2 +92% N 2 ) at 37°C for 90 min. Finally, the mice were put back to the dam to continue feeding for the follow-up experiment. In the Sham operation group mice, only the common carotid artery was exposed without ligation and hypoxia treatment.
The mice were randomly divided into the following groups: Sham + vehicle (PBS) group, HI + vehicle (PBS) group, HI + EVs group, HI + EVs-miR-21a INC group and HI + EVs-miR-21a inhibitor group. A total of 100 μg EVs dissolved in 50 µL PBS was administered via an intracardial injection at 24 h after HI insult for one time [17].

Culture and identi cation of MSCs
MSCs from bone marrow were harvested from C57BL/6J mice (4 weeks old) as previously described and modi ed [18]. Brie y, the cells were washed out from the femur and tibia of the donor mice with a syringe with 20 needles, ltered through 70 μm lter, centrifuged at 350 × g for 5 min, and then suspended in the complete medium (DME/F-12 containing 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin), and cultured in 5% CO 2 air environment at 37 °C. After 3 h, the fresh complete medium was replaced to remove the non-adherent cells. The medium was replaced every 8 h in the next 24 h, and then every 3 d. Adherent cells (passage 0) reached approximately 90% con uence in 14 d and were lifted by incubating in 0.25% trypsin/1 mM ethylenediaminetetraacetic acid.
Lifted cells were cultured in 5% CO 2 air environment at 37 °C, and the complete medium was replaced Harvest and identi cation of MSCs-EVs and contents analysis As previously described, the total EVs isolation kit (qEV, iZonScience, #1001622) was used for isolation and identi cation of EVs [17]. In short, when MSCs were fused to about 60%, the medium was replaced by EVs-free serum medium for 24 h, then collect the medium and centrifuge at 5752 × g for 30 min to obtain the supernatant and lter (0.22 μm). The supernatant (about 15 mL) was transferred to a 100000 MWCO ultra ltration tube and centrifuged at 4000 × g for 30 min. The sample in each ultra ltration tube was concentrated to about 200-300 μL and transferred to the qEV column. The remaining steps are carried out according to the manufacturer's instructions. Finally, 3 mL of PBS suspension containing EVs was used immediately or stored at -80 °C.
Western blot was used to measure the common markers of EVs, such as CD9 and TSG-101 [10]. The morphology of EVs was observed with use of TEM (HITACHI Limited, Japan). The qNano platform (Izon Sciences Ltd, NZ) was used to determine the size and concentration of EVs.

MSCs-EVs labeled with PKH67
In order to examine the distribution of EVs, the green uorescent dye PKH67 (Sigma-Aldrich Co., St Louis, MO, USA) was used to mark the EVs with according to the manufacturers' direction. Brie y, PKH67 dye (4 μL) was mixed into Diluent C (1 mL) to obtain the PKH67 solution. Then, PKH67 solution (1 mL) and diluent EVs (1 mL) were combined within centrifugation tube for 5 min, 2 mL 1% bovine serum albumin (BSA) was added to centrifugation tube to stop dyeing. Then, the mixture was ultracentrifuged (100000 × g) for 70 min to obtain EVs precipitate, followed by washing again with PBS (100000 × g) for 70 min. Finally, PKH67-labeled EVs were resuspended in PBS.

Tracking the distribution and location of MSCs-EVs
In order to determine the distribution and localization of EVs in BV-2 cells, cells seeded in 24-well plates were treated with OGD and then incubated with PKH67-labeled EVs (10 μg/mL) for 24 h. The slides of cells were collected for xation and block, and then stained with primary antibody Iba-1 (1:100) at 4 °C for 16 h. The next day, slides were incubated with uorescent secondary antibody for 30 min at 37 °C followed by a 5 min application of DAPI. Fluorescence images were captured with uorescent microscopy (OLYMPUS-BX51).

Oxygen-glucose-deprivation (OGD) model and treatments
For the induction of OGD, BV-2 cells were incubated at 37°C in a hypoxic chamber (Sanyo Electric, Osaka, Japan) (1 % O 2 , 5 % CO 2 ) in glucose-free medium for different durations of time. Afterwards, glucose was replenished to normal levels and treated with EVs, EVs-miR-21a INC or EVs-miR-21a inhibitor for 24 h under 5% CO 2 air environment at 37 °C (reoxygenation).

Cell Transfection
MiR-21a-5p inhibitor and its inhibitor negative control (INC) were designed by GenePharma Corporation (Suzhou, China). MSCs were transfected miR-21a-5p inhibitor and INC by using a Lipofectamine 2,000 reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. At 6 h after transfection, the culture was continued with EVs-free serum medium for 24 h and then EVs-miR-21a INC and EVs-miR-21a inhibitor were collected.

RNase A and TritonX-100-treatment
To degrade RNA, EVs were incubated in the presence of 20 μg/mL RNase A (Thermo Fisher Scienti c, PA, USA) and/or with 0.5% TritonX-100 (Solarbio, Beijing, China) for 30 min at 37°C. RNase digestion was stopped by addition 2 μL/mL RNase inhibitor (Thermo Scienti c, PA, USA) and RNA-isolation was performed as previous described.

Statistical analysis
The SPSS software program was used to analyze the data. All values presented are expressed as the mean ± standard deviation (SD). Data were analyzed with use of a one-way ANOVA and Bonferroni corrections for multiple post-hoc comparisons of means. A p value<0.05 was required for results to be considered statistically signi cant. All "Ns" in each group for histological ndings refer to the number of animals.

Characterization of MSCs and MSCs-EVs
TEM analysis revealed that MSCs-EVs were almost around 100 nm and a rounded morphology ( Figure  1A). MSCs-EVs were identi ed as small vesicles ranging from 60 to 120 nm with qNano ( Figure 1B).
Consistently, micrglia polarization was as assessed by immuno uorescence staining. The results revealed that the number of M1 phenotypes (CD16 + /Iba1 + cells) the right cortex was signi cantly reduced ( Figure 6A) and the number of M2 phenotypes (CD206 + /Iba1 + cells) in the ipsilateral cortex signi cantly increased ( Figure 6B) in the HI + EVs group as compared with that in response to HI alone when determined at 3 d post-HI.

Discussion
Recently, MSCs-EVs have been an attractive therapeutic approach for the treatment of in ammatory diseases and tissue injury. In this study, we found that MSCs-EVs alleviated neuroin ammation, associating with switch of microglia a pro-to an anti-in ammatory state following HI insult in vivo and in vitro. MiR-21a-5p shuttled by MSCs-EVs might participate in microglia polarization by targeting STAT3 to regulate neuroin ammation and maintain brain homeostasis.

MSCs-EVs suppressed neuroin ammation by switch of microglia from a pro-to an anti-in ammatory state
Therapeutic effects of MSCs-EVs in ischemic brain are related to its immunosuppressive properties. MSCs-derived microvesicles dampened LPS-induced in ammatory responses in BV-2 cells [23]. MSCs-EVs suppressed the microglia activation, switched microglia toward an anti-in ammatory phenotype, increased dendritic spine density in experiment Alzheimer's disease animal [24]. MSCs-derived EVs suppressed early in ammatory responses by modulating microglia/macrophage polarization after traumatic brain injury in rat [25]. Umbilical cord MSCs-derived EVs are internalized by microglia cells and dampened neuroin ammation following HI brain damage in newborn rats [15]. MSCs-EVs treatment signi cantly reduced microgliosis and prevented reactive astrogliosis following LPS-stimulated brain injury [26]. We previous found that MSCs-EVs by intracardial injection were found to be localized in the microglia after HI insult in vivo and in vitro [16,17]. MSCs-EVs treatment reduced neuroin ammation by suppressing osteopontin expression in microglia/macrophage after HI insult in neonatal mice [27]. In the current study, MSCs-EVs, internalized in microglia, decreased BV-2 cells apoptosis, reduced the expression of the M1 microglial markers (including IL-1β, iNOS and TNF-α), while increased the expression of the M2 microglial markers (including CD206, TGFβ and CD206) following OGD BV-2 cells. Con rmed with in vitro study, we found that MSCs-EVs treatment reduced the number of M1 phenotypes (CD16 + /Iba1 + cells), while increased the number of M2 phenotypes (CD206 + /Iba1 + cells) in the ipsilateral cortex, indicating that MSCs-EVs might be a promising therapeutic treatment for HI brain damage by modulating microglial polarization.
Anti-in ammatory mechanisms of miR-21a-5p in MSCs-EVs One of the main mechanisms responsible for the therapeutic properties of EVs injected in the brain ischemia is the transfer of miRNAs between cells. We previously reported that miR-21a-5p was found to be highly abundant in MSCs-EVs [17]. Several studies indicated that miR-21 play important role in the anti-in ammatory response in in ammatory-related diseases, tumor, infection, and so on [28,29] . MiR-21 overexpression markedly inhibited the in ammatory cytokine production and improved cardiac dysfunction after myocardial infarction [30]. EVs from hypoxia-preconditioned MSCs rescued synaptic dysfunction and regulated in ammatory responses to improve learning and memory function via replenishment of miR-21 in Alzheimer disease animal modle [31]. Moreover, overexpression of miR-21 in EVs remarkedly reduced cell apoptosis and improved cardiac function after myocardial infarction [32]. We found that MSCs-EVs treatment markedly increased OGD-reduced miR-21a-5p expression in BV-2 cells. MiR-21a-5p inhibitor abrogated the bene cial effects of MSCs-EVs on microglial polarization, in ammatory cytokines and cell survival, showing that miR-21a-5p transmitted by MSCs-EVs play considerably role in neuroprotective effect on HI exposure. STAT3 signaling pathway contributed to MSCs-EVs' effect on microglial M2 polarization The STAT3 signaling pathway is important for cellular growth, differentiation, and survival [33], involved in a variety of in ammatory and anti-in ammatory responses [34]. Activated STAT3 was predominantly localized in the macrophages/microglia in the post-ischemic brain [22]. Activation of microglial STAT3 induced expression of pro-in ammatory factors following cerebral ischemia [35,36]. In the present study, STAT3 was activated after HI insult in vivo and in vitro, in consistent with increased pro-in ammatory cytokines, decreased anti-in ammatory cytokines. MSCs-EVs treatment suppressed STAT3 activation, in consistent with switch of microglia a pro-to an anti-in ammatory state following HI insult in vivo and in vitro. After con rming a direct binding by Dual-Luciferase Reporter Assay, we found a negative correlation between STAT3 and miR-21a-5p after OGD in microglia. In addition, miR-21a-5p inhibitor abrogated STAT3 activation in microglia in vivo and in vitro. Taken together, the data suggested that MSCs-EVs regulated microglial activation by transferring miR-21a-5p to microglia and targeting STAT3 pathway.
This study has several limitations. First, the level of miR-21a-5p was decreased after OGD exposure in microglia. We previously reported that the level of miR-21a-5p decreased in the right hemisphere at 24 h and 48 h post-HI, whereas increased at 72 h post-HI insult. We did not observe the miR-21a-5p expression in microglia in vivo. Additionally, limited intracardial injection of MSCs-EVs administrations (24 h after HI insult) was tested in this study. Whether multiple treatments of MSCs-EVs using different time courses or modes of administration would produce better effects against HI will require future investigations.