Extracellular vesicles trigger ATP release and promote migration of human microglia through the P2X4 receptor / MFG-E8 – dependent mechanisms

Extracellular vesicles (EVs) effectively suppress neuroinflammation and induce neuroprotective effects in different disease models. However, the mechanisms by which EVs regulate neuroinflammatory response of microglia remain largely unexplored. Here, we addressed this issue by testing the action of EVs derived from human exfoliated deciduous teeth stem cells (SHEDs) on immortalized human microglial cells. We found that EVs induced a rapid increase in intracellular Ca 2+ and promoted a significant ATP release in microglial after 20 min of treatment. Boyden chamber assays revealed that EVs promoted microglial migration by 20 %. Pharmacological inhibition of different subtypes of purinergic receptors demonstrated that EVs activated microglial migration preferentially through the P2X4R pathway. Proximity ligation and co-immunoprecipitation assays revealed that EVs promote association between milk fat globule-epidermal growth factor-factor VIII (MFG-E8) and P2X4 receptor proteins. Furthermore, pharmacological inhibition of αVβ3/αVβ5 integrin suppressed EV induced cell migration and formation of lipid rafts in microglia. These results demonstrate that EVs promote microglial motility through P2X4 R/ MFG-E8 – dependent mechanisms. Our findings provide novel insights into the molecular mechanisms through which EVs target human microglia that may be exploited for the development of new therapeutic strategies targeting disease associated neuroinflammation.


Introduction
Microglia regulate immune homeostasis in the central nervous system (CNS) by initiating, maintaining and terminating neuroinflammatory response to all types of injuries [1]. Dysregulated microglial response is important for the development and propagation of various neurological disorders and therefore targeting of neuroinflammatory microglia has been considered as a promising therapeutic approach [2,3]. Extracellular vesicles (EVs) containing multiple proteins, RNAs, lipids and metabolites [4] are capable of crossing blood-brain barrier (BBB) and thus can be delivered into the brain by minimally invasive intranasal route [5][6][7]. Intranasal delivery of EVs has been shown to suppress neuroinflammation and promote neuroprotection in a different experimental models [8][9][10][11][12]. EVs specifically target and accumulate in pathologically affected areas of the brain [6,13]. Intranasally administered curcumin-encapsulated exosomes are selectively taken up by microglia and suppress secretion of pro-inflammatory cytokines in LPS-treated mice [9]. EVs also reduce microglial activation and prevent the induction of plethora of pro-inflammatory cytokines after status epilepticus in mice [8]. However, the mechanisms by which EVs regulate microglial responses remain largely unexplored.
During CNS injury damaged cells release several signaling molecules, including extracellular adenosine triphosphate (eATP) nucleotides that induce recruitment of microglia to the site of the injury and subsequent phagocytosis of apoptotic cells and neuronal debris. ATP potentiates bacterial killing by macrophages through purinergic P2X4 receptors by increasing the production of mitochondrial reactive oxygen species (ROS) [14]. Blockade of P2X4R signaling inhibits myelin phagocytosis in the experimental autoimmune encephalomyelitis (EAE) model [15]. Purinergic signaling is also crucial for ATP-induced chemotaxis [16,17]. Disruption of local ATP microgradients associated with neuronal hyperactivity during epilepsy impairs microglial motility and phagocytosis [18]. Microglia not only sense eATP by multiple purinergic receptors but can also release ATP by exocytosis [19]. Stimulation of Toll -like receptors (TLRs) triggers Ca 2+dependent release of ATP from microglia [19] and macrophages [20]. It has therefore been suggested that controlled ATP release by inflammatory cells may be used to fine-tune autocrine /paracrine responses during acute and chronic inflammation [21]. We have previously reported that EVs derived from stem cells from the dental pulp of human exfoliated deciduous teeth (SHEDs) upregulates human microglia phagocytosis [22]. In the present study we show that EVs increase intracellular Ca2+, trigger ATP release and significantly enhance microglial motility through milk fat globule-epidermal growth factor-factor VIII (MFG-E8)/ P2X4 receptor -dependent mechanisms. Our findings provide novel insights into the molecular mechanisms through which EVs target human microglia.

Characterization of EVs
Transmission electron microscopy (TEM) of EV samples isolated from SHEDs identified vesicles with a typical cup -shaped morphology (Fig. 1A), nanoparticle tracking analysis showed that size distribution of the EVs was around 150 nm (Fig.1B). Western blot analyses revealed that EV fractions were positive for the Milk fat globule-epidermal growth factor-factor VIII (MFG-E8) (Fig1C). Figure 1. Characterization of extracellular vesicles (EVs) isolated from stem cells from the dental pulp of human exfoliated deciduous teeth (SHEDs). (A) Transmission electron microscopy of EVs isolated from SHEDs (×120,000 magnification). (B) Concentration and particle size of EVs derived from SHEDs as analysed by nanoparticle tracking analysis using NanoSight LM10 instrument (Malvern Panalytical). Size distribution of the EVs was around 150 nm. (C) Samples from the cell lysates (L) and extracellular vesicles (EVs) were subjected to electrophoresis, blotted and the membrane was probed with antibodies against MFG-E8. Bands were visualized by incubation with appropriate horseradish peroxidase-conjugated secondary antibodies and chemiluminescence substrate. Full blot is available in the Supplementary Figure 1.

EVs increase intracellular Ca 2+ levels in human microglia
Increase in mobilization of intracellular Ca 2+ triggers many functions of microglia including activation, motility and release of ATP [19,20]. We used live calcium imaging to test how shortterm treatment of human microglia with EVs isolated from SHEDs alter microglial intracellular Ca 2+ levels (Fig 2). Our data show that acute 1 min application with EVs (4 AU/ml) significantly increased Ca 2+ levels in microglia (by 3.6 folds when compared to baseline level (EV-free BS); p < 0.0001, (Fig. 2C). (B, C) Response of human microglial cells to acute 1 min treatment of EVs. Intracellular Ca 2+ concentrations in single cells were visualized with the sensitive imaging system (Till Photonics, Germany). The peak of fluorescence in each individual cell during EV treatment was normalized to peak of fluorescence after response to ionomycin and presented as a percent of intracellular calcium signal. The graph represent mean ± SEM, statistically significant difference was determined by Mann-Whitney test, **** p < 0.0001 (Control N = 85 cells, EVs; N = 127 cells; cells from 3 experiments).

EVs trigger ATP release in human microglia
Different biologically relevant stimuli can induce microglial ATP release [19]. We therefore investigated levels of the extracellular ATP (eATP) in human microglia cultures stimulated with EVs (1 AU) for 20 and 60 min. Treatment with EVs for 20 min significantly (by 2 folds; p = 0.0158) increased the levels of eATP in the microglia cultures (Fig. 3). This increase was no longer evident after 60 min of incubation (p = 0.6417) (Fig. 3) suggesting that that EVs induce a rapid and transient ATP release in human microglia. . EVs trigger ATP release from human microglial cells. Microglial cells were exposed to EVs (1 AU/well) diluted in BS for 20 and 60 minutes. Supernatants from each well were subjected to ATP assay. ATP was measured using ATPlite Luminescence Assay System with Perkin Elmer Wallac 1420 Victor2 instrument using Wallac 1420 software. Each bar represents mean ± SEM, statistically significant difference was determined by two-way ANOVA, Sidak's multiple comparisons test, * p < 0.05 (n = 15 -20 wells from 3 experiments).

EVs promote P2X4R-dependent migration of human microglia
It is well known that eATP serves as a powerful trigger for motility of microglia [23,24]. We therefore hypothesized that EV-triggered ATP release may promote microglial motility via autocrine and (or) paracrine mechanisms. Indeed, EVs significantly increased microglial migration by 20 % (n = 22, p = 0.0001) in the Boyden chamber assay. (Fig.4B). Since microglia are highly sensitive to eATP through several subtypes of purinergic receptors such as P2X4 and P2Y12 [25].
we tested different inhibitors to determine which purinergic signaling pathway is responsible for the observed effects. Whilst the nonselective antagonist of ATP-gated P2 receptors suramin failed to significantly alter cellular migration (n =3, p = 0.1321; Fig.4C), a highly potent P2Y12 antagonist AR-C 69931 significantly decreased the migration by 30 % (n = 3, p = 0.0009; Fig. 4D). AR-C69931 failed to prevent the EV induced migration (n =3, p = 0.0002; Fig. 4D). These results indicate that suramin and AR-C69931 did not suppress EV-induced migration of microglial cells.
P2X4R is relatively insensitive to the blockade by the suramin, we therefore used a potent selective P2X4R antagonist 5-BDBD [26]. We found 5-BDBD significantly decreased microglial migration in the presence and absence of EVs (n = 5, p = 0.0263) (Fig.4E). Interestingly, EVs also neutralized inhibitory effect of ATP degrading enzyme apyrase on the migration of microglia (n = 3, p = 0.0001) (Fig. 4F). Taken together our data indicate that EVs activate microglial migration via P2X4R pathway and that this effect does not depend on the EV-triggered ATP release. For these experiments apyrase and 1 AU of EVs were added simultaneously to the upper wells and cells incubated overnight. Data represent measurements from three independent experiments (only treatment with 5-BDBD from five). The graph represent mean ± SD values, groups were compared with one-way ANOVA, Tukey's multiple comparisons test,*p <0.05, **p<0,01, *** p < 0.001.
We used proximity ligation assay (PLA) to detect protein -protein interactions between MFG-E8 and P2X4 receptor in the human microglial cells. Our results show an association between MFG-E8 and P2X4 receptor proteins in control (EV-untreated) cells. Exposure to EVs for 2 hours significantly (by 4.3 folds compared to control; p<0.0001) promoted association between MFG-E8 and P2X4 receptor proteins (Fig.5) suggesting a close association between MFG-E8 and P2X4 receptor proteins in human microglia which is significantly promoted in cells exposed to EVs. Co-immunoprecipitation assays also confirmed association between MFG-E8 and P2X4 receptor proteins in the microglial cells (Fig.6). Inhibition of MFG-E8 receptor with cilengitide suppressed EV -induced migration and formation of lipid rafts in microglia MFG-E8 protein interacts with target cells through αVβ3 and αVβ5 integrins [27]. Cilengitide selectively blocks activation of the αVβ3 and αVβ5 integrins and is effectively used as an inhibitor of MFG-E8 signaling [30,31]. We therefore tested the effects of cilengitide on EV -induced migration of microglia. Pretreament with 10 µM of cilengitide significantly suppressed EV -induced migration of microglia (by 17 % compared to control; n= 8; p= 0.0001) (Fig.7). For migration experiments (see above) microglial cells were pre-treated for 30 min with inhibitor MFG-E8 receptor cilengitide (10 µM), then treated with 1 AU of EVs. Data represents mean ± SD, **p < 0.01 ,***p < 0.001; N = 3; One way ANOVA, Tukey's multiple comparisons test.
Lipid rafts serve as organizing platforms enriched in different signaling proteins and receptors initiating inflammatory signaling and different cellular responses [32]. We tested how EVs affect formation of lipid rafts in the microglia. Treatment of human microglia with EVs for 30 min significantly increased lipid raft formation (by 113 % compared to control; n = 3; p= 0.0001) (Fig.8).

Discussion
In this study we demonstrate for the first time that EVs trigger ATP release and promote migration of human microglia through the P2X4 receptor/MFG-E8 -dependent mechanisms. Our findings provide novel insights into the molecular mechanisms through which EVs target human microglia.
Our initial observation that EVs increased intracellular Ca 2+ levels and induced a rapid ATP release in microglia prompted us to test whether these effects were responsible for the increased microglial motility. Blockage of P2X4 receptor with selective inhibitor 5-BDBD prevented EV -induced increase in microglial migration. However, our findings do not support the model according to which EV-triggered ATP release promote microglial motility via autocrine and (or) paracrine mechanisms. First of all, use of ATP degrading enzyme apyrase did not suppress the effects of EVs on the microglial migration. Furthermore, it is unlikely, that EV -induced transient (20 min) increase of eATP could significantly affect overnight migration of cells through the membrane in the Boyden chamber unless it triggers early remodeling supporting formation of the cell migration mechanisms. Several reports have demonstrated that MFG-E8 secretory glycoprotein is highly expressed in the different types of EVs. By its discoidin domain MFG-E8 protein associates with PS exposed on the membranes and this property has already been used for the in vivo identification of apoptotic and EV-bound cells [29]. EVs used in our study expressed high levels of MFG-E8 protein (Fig.1C). We therefore tested whether MFG-E8 and P2X4 receptor proteins directly interact in human microglia. Indeed, sensitive in situ proximity ligation assays (PLA) revealed a close association between MFG-E8 and P2X4 receptor proteins in EV -treated and untreated cells (Fig.5). Microglial cells constitutively express and secrete MFG-E8 protein [27,33], it has therefore been impossible to distinguish between microglial and vesicular MFG-E8 fractions. Nevertheless, treatment with EVs remarkably promoted association between MFG-E8 and P2X4 receptor proteins ( Fig.5). To our knowledge, this is the first demonstration of direct interaction between MFG-E8 and P2X4 receptor proteins. The MFG-E8 protein has epidermal growth factor domain which recognizes integrins αVβ3 and αVβ5 on the target cells [27]. MFG-E8 has been shown to inhibit necrotic cell-induced and ATP-dependent IL-1β production by macrophages through mediation of integrin β3 and P2X7 receptor interactions in primed cells [34]. Several studies have also demonstrated a direct physical association between P2X4 and P2X7 receptors in different types of cells [35,36]. We therefore propose that EV -associated MFG-E8 may interact with P2X4 receptors in human microglia. This statement is further supported by our finding that cilengitide, which is a cyclized RGD-containing pentapeptide that selectively blocks activation of the αvβ3 and αvβ5 integrins and is effectively used as inhibitor of MFG-E8 signaling [30,31], suppressed EVinduced migration of microglia (Fig.7). P2X4R signaling is important for phagocytosis during EAE [15]. On the other hand, MFG -E8 -mediated phagocytic clearence of apoptotic cells by microglia is crucial for the proper regulation of the neuroinflammatory response in the CNS [27] [37]. We have also previously demonstrated that EVs increased phagocytic activity of human microglial cells [22]. Although in this study we did not test the effects of cilengitide on the EV -induced phagocytic activity of microglia, we propose that EVs may also promote phagocytic response through the MFG-E8 -dependent pathway. We also suggest that after therapeutic administration EVs can be selectively recognized and internalized by microglial cells through MFG-E8 -αVβ3/αVβ5 -dependent mechanisms. This model may at least partially explain selective EV targeting to the pathologically affected areas [6,13] and specific accumulation in microglial cells [8][9][10][11][12]. Further studies, especially using MFG-E8 -deficient animals are needed to clarify these issues.
Different stimuli promote formation and enlargement of lipid rafts serving as organizing platforms initiating inflammatory signaling and different cellular responses [32]. Lipid rafts are enriched in the innate immune receptors TLRs 2 and 4, TREM2, IFNγR, purinergic receptors P2X and P2Y, integrins and other signaling proteins [32,38,39]. We therefore reasoned that EVs may also affect lipid raft formation in target cells. Indeed, exposure to EVs for 30 min greatly increased lipid raft formation in microglia (Fig.8). Furthermore, pretreatment with cilengitide prevented EV -induced lipid raft formation showing that EVs trigger lipid raft formation through MFG-E8 -αVβ3/αVβ5depending mechanisms.
Based on these findings, we propose a novel mechanism for the EV action on microglial cells ( Fig.9). . Proposed mechanism for the EV action on microglial cells. EVs carrying MFG-E8 proteins associated with phosphatidylserine exposed on the outer membrane are recognized by the αVβ3/αVβ5 integrin receptors of microglial cells and trigger lipid raft formation, interaction with P2X4 receptors and possibly other molecules enriched in the lipid rafts such as components of TLR4 multireceptor complex. These events lead to the upregulation of intracellular Ca 2+ , release of ATP and increased motility of microglia.
EVs carrying MFG-E8 proteins are recognized by the αVβ3/αVβ5 integrin receptors of microglial cells and trigger lipid raft formation, interaction with P2X4 receptors and possibly other molecules enriched in the lipid rafts such as components of TLR4 multireceptor complex. These events lead to the upregulation of intracellular Ca 2+ , release of ATP and increased motility of microglia.
In conclusion our study demonstrates the importance of MFG-E8 -P2X4 signaling pathway for the immunomodulatory action of EVs in human microglia. Our findings could be potentially exploited for the development of new therapeutic strategies targeting neuroinflammatory microglia.

Materials and Methods
Culture of stem cells from the dental pulp of human exfoliated deciduous teeth (SHEDs) and human microglial cells SHEDs were obtained from human exfoliated deciduous teeth of a child, whose parents had signed an informed consent. Material was collected under the approval of the Lithuanian Bioethics committee (Nr. 6B-08-173; 2008-04-22). SHEDs were isolated according previously described protocol [22] For the isolation of extracellular vesicles (EVs) SHEDs from the 3-5rd passages were grown until the cultures reached subconfluence, then standard medium was changed to the serum-free medium MSC NutriStem XF (Biological Industries, Kibbutz Beit Haemek, Israel). Immortalized (SV40) human microglial cell line was purchased from ABM. Human microglial cells were cultivated on cell culture tissue flasks coated with 50 µg/ml of rat tail collagen I (Gibco) in high glucose (4.5 mg/ml) DMEM supplemented with glutamax (Gibco) and 10 % of EV-depleted FBS (Biochrom).

Isolation and characterization of extracellular vesicles
Isolation of extracellular vesicles (EVs) was performed using differential centrifugation according to the described protocol [40] with some modifications. All centrifugation steps were performed at 4 °C. Supernatants collected from SHEDs cultivated in serum-free medium MSC NutriStem XF images were taken with a Olympus Quemesa camera, using the iTEM 5.2 software (Fig.1A).
The yield of EVs prepared from the supernatants of SHEDs grown until subconfluence on the cell culture flasks (representing surface area 37.5 cm 2 ) and conditioned for 72 hours in serum-free medium MSC NutriStem XF was defined as 1 activity unit (AU). According to the NTA measurements ( Fig.1B) 1 AU corresponded to the 3.65 x 10 8 of EVs for migration assays and 5.2 x 10 8 of EVs for Ca 2+ -imaging and ATP assays.

Measurements of intracellular Ca 2+ concentration
Human microglial cells were plated on rat tail collagen type I pre-coated coverslips (4000 cells/cm 2 ) for 24 hours in complete DMEM medium (with 10 % FBS and 1 % P/S) depleted of EVs by ultracentrifugation at 100 000 x g for 6 hours at 4 ⁰C. Medium was removed and coverslips were  and P2X4 receptor for 1 hour at room temperature. After incubation with primary antibodies membranes were washed three times in PBS-Tw. After washing, membranes were incubated further with horseradish peroxidase (HRP)-conjugated secondary antibody (Thermo Scientific) for 1 hour at room temperature. Washing procedure was repeated and immunoreactive bands were detected with Clarity ECL Western blotting substrate (Bio-Rad) using ChemiDoc MP system (Bio-Rad).

Proximity ligation assay
To detect protein -protein interactions between MFG-E8 and P2X4 receptor in the human microglial cells proximity ligation assay (PLA) was conducted using the Duolink® in situ detection

Inhibition of MFG-E8 receptor
Human microglial cells were plated on rat tail collagen type I pre-coated 24-well plates (for ELISA assays) and on the glass coverslips (20 000 cells/well) for 72 hours in complete DMEM medium depleted of EVs (for lipid raft labeling). Microglial cells were pre-treated with 10 µM of cilengitide (Sigma) for 2 hours and exposed to EVs (1 AU) for 30 min then lipid raft labeling was performed.

Statistical analysis
Statistical analysis was performed from data of at least three independent biological experiments.
Graphs represent mean and standard deviation (SD) or standard error of the mean (SEM) values.
Differences between 2 groups were compared by Student's t-test, while differences between three or more groups were compared by one-way ANOVA. Data, which did not passed the Shapiro-Wilk test of normality, were analyzed with non-parametric one-way ANOVA (Kruskal-Wallis one-way analysis of variance) using Dunn's multiple comparison post-hoc test. All results were considered as significant, when p < 0.05. Data analyzed using Graph Pad Prism® software version 8.0.1 (Graph Pad Software, Inc., USA).

Acknowledgements
This work was supported by funding from European Regional Development Fund