Extracellular vesicles promote autophagy in human microglia through lipid raft-dependent mechanisms

Autophagy dysfunction has been closely related with pathogenesis of many neurodegenerative diseases and therefore represents a potential therapeutic target. Extracellular vesicles (EVs) may act as a potent anti-inflammatory agents and also modulators of autophagy in target cells. However, the molecular mechanisms by which EVs modulate autophagy flux in human microglia remain largely unexplored. In the present study we investigated the effects of EVs derived from human oral mucosa stem cells on the autophagy in human microglia. We demonstrate that EVs promoted autophagy and autophagic flux in human microglia and that this process was dependent on the integrity of lipid rafts. LPS also activated autophagy, but combined treatment with EVs and LPS suppressed autophagy response indicating interference between these signalling pathways. Blockage of Toll-like receptor 4 (TLR4) with anti-TLR4 antibody suppressed EV- induced autophagy. Furthermore, blockage of EV- asscoiated HSP70 chaperone which is one of the endogenous ligands of the TLR4 also suppressed EV- induced lipid raft formation and autophagy. Pre-treatment of microglia with selective inhibitor of αvβ3/αvβ5 integrins cilengitide inhibited EV-induced autophagy. Finally, blockage of purinergic P2X4 receptor (P2X4R) with selective inhibitor 5-BDBD also suppressed of EV-induced autophagy. In conclusion, we demonstrate that EVs activate autophagy in human microglia through interaction with HSP70/TLR4, αVβ3/αVβ5, and P2X4R signalling pathways and that these effects depend on the integrity of lipid rafts. Our findings could be used for development of new therapeutic strategies targeting disease-associated microglia.


Introduction
Microglial cells regulate immune homeostasis in the central nervous system (CNS) by constant monitoring of the brain tissue for signs of damage and clearing cellular debris [1,2].
Abnormal activation of microglia has been associated with many acute and chronic inflammatory CNS disorders, therefore targeting of disease-associated microglia represents a promising therapeutic approach [1,3]. Extracellular vesicles (EVs) are nanosized lipid bilayer -enclosed vesicles, that carry different proteins, RNAs, lipids, and other bioactive molecules [4]. Increasing evidence demonstrate that EVs may act as a potential immunomodulatory and anti-inflammatory agents [5,6]. Moreover, some unique characteristics of EVs such as biocompatibility, specific targeting and accumulation in pathologically affected areas, low immunogenicity and the ability to cross biological barriers make them an attractive therapeutic tool for a variety of neurological disorders [7][8][9][10]. Intravenously administered EVs derived from mesenchymal stem cells (MSC) reduced neuroinflammation and switched microglia towards restorative functions after cortical injury in the aged brain of monkeys [11]. EVs administered either by intracerebroventricular, or by intranasal routes, protected mice from the neonatal stroke by direct interaction with microglial cells [12]. EVs also suppressed neuroinflammation and rescued cognitive impairments after traumatic brain injury [13]. However, the mechanisms by which EVs regulate neuroinflammatory response of microglia remain largely unexplored.
Autophagy dysfunction has been closely related with pathogenesis of many neurodegenerative diseases [14]. In this regard, it is important to note that autophagy can also serve as controlling mechanism of the metabolic and immune status of microglia and thus balance neuroinflammatory response [14][15][16]. For instance, ATG7 gene-deficient microglial cells, show transcriptional and functional similarity to the functionally incompetent microglia from the aged wild-type mice [17]. Induction of autophagy in aged mice using disaccharide trehalose resulted in the restoration of microglial phagocytic activity and remission of the neuroinflammation [17]. Autophagy also negatively regulates inflammasomes in microglia during Aβ-induced neuroinflammation [18] and experimental autoimmune encephalomyelitis [19,20]. Several experimental evidence indicate bidirectional relationship between autophagy and EVs. Autophagy is crucial for the synthesis, secretion and degradation of the EVs. Increased levels of autophagy have been associated with inhibition of EV release due to the increased fusion of multivesicular bodies with autophagic vacuoles [21][22][23]. On the other hand, exogenous EVs can modulate autophagic flux in target cells.
Microglia-derived EVs promoted autophagy and mediated multi target target signalling during microglia microglia crosstalk in vitro [24]. Neural stem cell-derived EVs attenuated apoptosis and neuroinflammation after traumatic spinal cord injury by activating autophagy [25]. By contrast, in ischemic stroke model, EVs from human iPSC-derived MSC suppressed autophagy and promoted angiogenesis in vivo and in vitro [26]. These findings suggest that EVs can be used for targeting autophagy in the disease-associated microglia. However molecular mechanisms by which exogenous EVs modulate autophagic flux in target cells remain unknown.
Our previous findings demonstrate that EVs may act as a potent immunomodulators of human microglia by suppressing TLR4/NFκB signalling pathway, promoting phagocytosis and inducing metabolic reprogramming [5]. More recently we demonstrated that EVs increased migration of human microglia through milk fat globule-epidermal growth factor-VIII (MFG-E8)/ purinergic P2X4R receptor (P2X4R) pathways [27]. EVs also induced lipid raft formation in human microglia through Toll-like receptor 4 (TLR4), P2X4R and integrin α Vβ3/αVβ5 signalling pathways [28]. In the present study we investigated the effects of EVs on the autophagy in human microglia. We demonstrate that EVs activate autophagy and autophagic flux in human microglia through TLR4, α Vβ3/αVβ5, and P2X4R signalling pathways and that these effects depend on the integrity of lipid rafts.

Materials and methods
Human oral mucosal stem cell line After adhesion of the explants (approximately 3 to 4 hours), each well was filled with medium. Explants were maintained at 37 ºC in a humidified atmosphere with 5% CO 2 , and the medium was routinely changed twice per week. After the appearance of migrating cells, the explants were removed from the wells and the medium changed every three days until the cell cultures reached subconfluence.

Human microglial cell line
The immortalized (SV40) human microglial cell line was acquired from ABM.

Isolation of extracellular vesicles
EVs were isolated by differential centrifugation according to Thery et al [29]. with some modifications. All centrifugation steps were performed at 4 ºC. using the ChemiDoc MP system (Bio-Rad).

Transmission electron microscopy
Transmission electron microscopy (TEM) of EVs was conducted following a previously established protocol [29] with some adjustments. In summary, extracellular vesicles (EVs) suspended in PBS were treated with 2% paraformaldehyde (PFA) for 40 minutes on ice to ensure fixation. Copper grids coated with Formvar and carbon were placed on a 10-µl droplet of the fixed EV suspension at room temperature for 20 minutes.
Subsequently, the grids were rinsed with PBS and transferred to a 30-µl droplet of 1% glutaraldehyde solution for 5 minutes at room temperature. The grids underwent eight consecutive washes by transferring them from one droplet of distilled water to another. To enhance contrast, the samples were treated with 2% neutral uranyl acetate on 30 µL droplets for 5 minutes at room temperature in the absence of light. Afterward, the grids were air-dried for 5 minutes. Analysis of the samples was carried out using a FEI Morgagni 268 transmission electron microscope.

Immunocytochemistry
The autophagy process in microglial cells was evaluated immunocytochemically by staining against autophagy marker LC3B protein [30]. Immediately after treatments, cells  After 48 hours of incubation with BacMam reagent, cells were exposed to different agents depending on the experimental design. Then samples were prepared and analyzed as described previously, the only difference is that in this case both GFP and RFP fluorescence were analyzed and quantified separately. After specific loss of GFP fluorescence following fusion with lysosomes ( Fig.1), calculation of the RFP:GFP ratio was used as an indicator of autophagic flux.

Experimental design
To assess the effect of EVs on autophagy and autophagic flux, microglial cells were

Statistical analysis
Statistical analysis was performed using data from at least three independent biological experiments. Plots show the mean and standard error of the mean (SEM). Since all our data did not meet the assumptions of parametric methods (failed the Shapiro-Wilk normality test or variances were unequal), nonparametric tests were used for statistical analysis. Difference between two groups was determined using the Mann-Whitney U test, and differences between 3 or more groups were determined by Kruskal-Wallis one-way analysis of variance followed by Dunn's post-hoc test. All results were considered significant when p < 0.05. Data was analyzed using Graph Pad Prism® 8.0.1 version software (Graph Pad Software, Inc., San Diego, CA, USA).

EVs promote autophagy and autophagic flux in human microglia
We first tested how OMSC-derived EVs affect LC3B expression in human microglial cells. For this purpose, the cells were incubated with 1 AU of EVs for 2, 6 and 16 hours  Kruskal-Wallis test followed by Dunn's post-hoc test (** p<0.01). E -Estimation of autophagic flux by RFP:GFP ratio. Data represent the mean ± SEM from three independent experiments (n = 3). Statistical significance was determined using the Mann-Whitney U test (** p<0.01). F -Confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit after exposure to EVs for different durations. G -Quantification data of GFP-LC3B and RFP-LC3B protein expression. Data shown represent the mean ± SEM from 15 fields of view (n = 3). Statistical analysis was performed using Kruskal-Wallis test followed by Dunn's post-hoc test for GFP-LC3B (H) and RFP-LC3B (I) protein expression (* p<0.05; *** p<0.001) These results were further confirmed with RFP-GFP-LC3B reporters (Premo™ Autophagy Tandem Sensor Kit). Exposure to EVs for 16 hours significantly increased autophagosome formation (p = 0.0135, n = 3, Fig. 1 D) and autophagic flux (p = 0.0049, n = 3, Fig. 1 E) in microglia cells. We also found that 48 hours after exposure to EVs autophagy response declined to the basal levels (p Fig.1 F, G, H, I).

Disruption of lipid rafts prevents EV-induced autophagy in microglia
We have previously demonstrated that EVs promoted lipid raft formation in human microglia [27,28]. We therefore asked whether EV-induced autophagy depends on the integrity of lipid rafts. For this purpose cells were pre-treated for 1 hour with 5 mM cholesterol removing agent methyl-β-cyclodextrin (MβCD) and then exposed to the 1 AU of EVs for 16 hours (Fig. 2). Disruption of lipid raft suppressed EV-induced autophagy to the basal levels (by 28.6%, p <0.0001, n = 3; Fig. 2 B). Our results demonstrate that lipid raft integrity is essential for the EV-induced autophagy in human microglia.

Combined treatment with EVs and LPS attenuate autophagy response in human microglia
We have recently demonstrated that combined treatment with EVs and LPS decreased lipid raft formation in microglia [28]. We therefore tested how combined treatment with EVs and LPS affect autophagy response. Microglial cells were exposed to EVs (1 AU), or LPS (5 µg/ml) and combination of both EVs and LPS for 16 hours (Fig. 3). Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test in GraphPad Prism 8.0.1 software (* p<0.05; **** p<0.0001). C -Confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit. D -Quantification data of GFP-LC3B and RFP-LC3B protein expression. Data shown represents the mean ± SEM from 15 fields of view (n = 3). Statistical analysis was performed using Kruskal-Wallis test followed by Dunn's post-hoc test for GFP-LC3B (E) and RFP-LC3B (F) protein expression (* p<0.05; ** p<0.01; *** p<0.001). G -Estimation of autophagic flux by RFP:GFP ratio. Data represents the mean ± SEM from three independent experiments (n = 3). Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test (* p<0.05; *** p<0.001).
We next tested how combined treatment with EVs and LPS affect autophagy and autophagy flux in microglia expressing tagRFP-eGFP-LC3B reporters (Fig. 4 C). We found that both EVs and LPS significantly increased signal intensity of GFP-LC3B protein (EVs by 94.5 %, p = 0.0202 and LPS by 93.1%, p = 0.0059, n = 3, Fig. 3 E). By contrast, combined treatment suppressed GFP-LC3B signal to the control levels ( Fig.3 E). These results indicate that when Our findings show mutual interference between EVs and LPS signalling during induction of autophagy in human microglia.

Blockage of TLR4 prevents EV-induced autophagy
We further investigated possible crosstalk between EVs and LPS signalling in microglia and tested the effects of specific blockage of TLR4 on the EV-induced autophagy (Fig. 4). As expected, pre-treatment with anti-TLR4 antibody prevented LPS-induced autophagy (p = 0.0295, n = 3, Fig. 4 A, B). Importantly, blockage of TLR4 also suppressed EVs-induced autophagy in microglia (decreased by 39.9 %, p = 0.0119, n = 3; Fig. 4 B). In contrast, pre-treatment with control IgG antibody did not affect EV-induced autophagy. Our results demonstrate that EVs trigger autophagy response in microglia through interaction with TLR4.

EV-associated HSP70 chaperones promote formation of lipid rafts and autophagy in human microglia
HSP70 protein is one of the endogenous ligands of the TLR4 [34]. HSP70/TLR4 interaction is also important for the protective effect of exosomes against neomycin-induced hair cell death [32]. We detected high expression levels of HSP70 in our EV preparations (Supplemental figure 1 C). We therefore used neutralizing anti-HSP70 antibody for blockage of the EV-associated HSP70 and then investigated lipid raft formation and autophagy in human microglia cells. m, magnification -63x). B -Mean fluorescence intensity of labeled ganglioside GM1 per cell measured using LAS X software. Data represents the mean ± SEM from 15 fields of view (n = 3), results normalized to control. Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test in GraphPad Prism 8.0.1 software (*** p<0.001; **** p<0.0001). C -Confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit. D -Quantification data of GFP-LC3B and RFP-LC3B protein expression. Data shown represents the mean ± SEM from 15 fields of view (n = 3). Statistical analysis was performed using Kruskal-Wallis test followed by Dunn's post-hoc test for GFP-LC3B (E) and RFP-LC3B (F) protein expression (* p<0.05; ** p<0.01; *** p<0.001).
Pre-treatment of EVs with a neutralizing anti-HSP70 antibody (1 μ g/ml) for 2 hours resulted in a significant decrease in their ability to induce lipid raft formation in microglial cells (p = 0.0009, n = 3, Fig. 5 B). Pre-treatment with IgG control antibody did not affect EVinduced lipid raft formation. We also found that blockage of vesicular HSP70 significantly integrins and phosphatidylserine (PS) exposed on the EV membranes. (Fig.7 A) [35,36]. Our recent data demonstrate that blockage of α vβ3 and α vβ5 integrins with cilengitide prevented EV-induced lipid raft formation in microglia cells [28]. We therefore tested the effects of cilengitide on the EV-induced autophagy in microglia (Fig. 6 B, C).

Inhibition of P2X4 receptor suppresses EVs-induced autophagy
Our previous study showed a close association between MFG-E8 protein and P2X4 receptor, which increased after exposure to EVs [27]. In this study we tested how blockage of P2X4R with selective antagonist 5-BDBD affect EV-induced increase of LC3B protein in human microglia cells (Fig. 7 A).  Fig. 7 B) in microglia.

Discussion
The primary goal of this study was to examine the effects of OMSC-derived EVs on the autophagy process in human microglia cells. We demonstrate that EVs significantly increased autophagy and autophagic flux and therefore could be potentially used for targeting autophagy in the disease-associated microglia.
It has been shown that activation of autophagy can promote neuroprotective properties of microglia during neuroinflammation whereas inhibition of this process can lead to the increased neurodegeneration [37][38][39]. On the other hand, over-enhancement of autophagy may exert adverse effects on cells, leading to the cell death [40,41]. It is therefore important to understand molecular mechanisms responsible for the autophagy-promoting effects of EVs in human microglia.
We have previously demonstrated that EVs promoted lipid raft formation in human microglia [27,28]. In this study we show that EV-induced autophagy depends on the integrity of the lipid rafts. We also demonstrate that when used separately, both EVs and LPS increased autophagy, but combined treatment suppressed autophagy response. Blockage of LPS receptor TLR4 with anti-TLR4 antibody prevented EV-induced autophagy in human microglia. Furthermore, blockage of HSP70 chaperone which is highly expressed in our EV preparations and is one of the endogenous ligands of the TLR4 also suppressed EV-induced lipid raft formation and autophagy. We therefore suggest that EVs can at least partially trigger lipid raft formation and autophagy response in microglia through interaction with TLR4s. Autophagy promotes elimination of intracellular pathogens and is closely associated with different innate immunity signalling pathways [42,43]. Different types of TLR ligands enhanced autophagy in immune cells [44]. In particular, stimulation of TLR4s activated autophagy by affecting Bcl-2 -Beclin 1 interactions in macrophages [45]. Interestingly, several studies emphasize a close interplay between lipid rafts and autophagy machinery. For instance, it has been shown that lipid rafts can regulate autophagy by interacting with autophagosomes and autophagy-related proteins, such as ATG5 and ATG12 [46]. However, the exact molecular mechanisms linking EV/TLR4 -induced lipid raft formation with activation of autophagy in human microglia remain unclear.
Microglia are highly motile cells and we previously showed, that EV-induced migration of microglial cells depends on the α vβ3 and α vβ5 integrins [27]. We also demonstrated that α vβ3/ α vβ5 integrin signalling pathway was indispensable for EV-and LPS-induced lipid raft formation in microglia [28]. In the present study we demonstrate that pre-treatment of microglia with specific inhibitor of α vβ3 and α vβ5 integrins cilengitide suppressed EV-induced autophagy. EV preparations used in our study were highly enriched with MFG-E8 protein (Supplemental figure 1) which can act as a molecular bridge connecting phosphatidylserine (PS) exposed on the outer membranes of the EVs with microglial integrin α Vβ3/αVβ5 receptors [47]. We therefore hypothesize that EV-associated MFG-E8 protein may interact with microglial α Vβ3/αVβ5 receptors and promote downstream signalling events leading to the increased migration and autophagy. It has been shown that autophagy regulates integrin-mediated cell adhesion and may promote cell migration affecting the turnover of focal adhesions [48,49]. Interestingly, integrins may also play an important role in the autophagosome formation by direct interactions with LC3 proteins [50,51]. We speculate that EVs can affect bidirectional interplay between integrin recycling and autophagy thereby promoting remodeling of focal adhesions and microglial motility.
We have recently demonstrated that pharmacological inhibition of purinergic receptor P2X4R signalling suppressed EV-induced lipid raft formation, migration and phagocytic activity of microglia [27,28]. P2X4Rs undergo rapid and constitutive endocytosis from the plasma membrane to the lysosomes and are activated by both intraluminal ATP and by a decrease in acidity [52]. Activated P2X4Rs initiate release of Ca 2+ from the lysosomes and promote fusion of the lysosome membranes with the autophagosomes stimulating autophagic flux [53]. In the present study we show that blockage of P2X4R with selective antagonist 5-BDBD inhibited EV -induced autophagy in microglia. Our results are in agreement with other studies showing that P2X4R signalling promotes autophagy activation in different experimental models [54,55]. It is presently unclear, however, whether EVs can directly interact with P2X4Rs on the plasma membrane or induce P2X4R-dependent effects indirectly. Purinergic receptors, TLRs and integrins are enriched in the lipid rafts and therefore are in close proximity [56][57][58], therefore possible EVs may also directly interact with P2X4Rs. Further studies are needed to elucidate interactions between EVs and P2XR4 signalling pathway.
Another interesting observation of this study is functional interdependence between TLR4, α Vβ3/αVβ5, and P2X4R signalling pathways as autophagy promoters in response to the treatment with EVs (Fig. 8 A). We have recently demonstrated that these signalling pathways are functionally interdependent promoters of EV-and LPS-induced lipid raft formation in microglia [28]. We therefore propose that TLR4, α Vβ3/αVβ5 integrin, and P2X4R comprise a network of functionally interdependent signalling units regulating the induction and enlargement of lipid rafts leading to the autophagy initiation (Fig. 8). In conclusion, we demonstrate that EVs activate autophagy in human microglia through interaction with TLR4/HSP70, α Vβ3/αVβ5, and P2X4R signalling pathways and that these effects depend on the integrity of lipid rafts. Our findings could be used for development of new therapeutic strategies involving extracellular vesicles for targeting disease-associated microglia.

Conflict of interest
There is no conflict of interests to disclose.  μ m, magnification -63x). B -Mean fluorescence intensity of LC3B protein per cell measured using LAS X software. Data represent the mean ± SEM from 20 fields of view (n = 4), results normalized to control. Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test in GraphPad Prism 8.0.1 software (* p<0.05; *** p<0.001; **** p<0.0001). C -Representative confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit. D -Statistical analysis of GFP-LC3B and RFP-LC3B protein expression. Data shown represent the mean ± SEM from 15 fields of view (n = 3). Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test (** p<0.01). E -Estimation of autophagic flux by RFP:GFP ratio. Data represent the mean ± SEM from three independent experiments (n = 3). Statistical significance was determined using the Mann-Whitney U test (** p<0.01). F -Confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit after exposure to EVs for different durations. G -Quantification data of GFP-LC3B and RFP-LC3B protein expression. Data shown represent the mean ± SEM from 15 fields of view (n = 3). Statistical analysis was performed using Kruskal-Wallis test followed by Dunn's post-hoc test for GFP-LC3B (H) and RFP-LC3B (I) protein expression (* p<0.05; *** p<0.001) μ m, magnification -63x). B -Mean fluorescence intensity of LC3B protein per cell measured using LAS X software. Data represents the mean ± SEM from 15 fields of view (n = 3). Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test in GraphPad Prism 8.0.1 software (* p<0.05; **** p<0.0001). C -Confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit. D -Quantification data of GFP-LC3B and RFP-LC3B protein expression. Data shown represents the mean ± SEM from 15 fields of view (n = 3). Statistical analysis was performed using Kruskal-Wallis test followed by Dunn's post-hoc test for GFP-LC3B (E) and RFP-LC3B (F) protein expression (* p<0.05; ** p<0.01; *** p<0.001). G -Estimation of autophagic flux by RFP:GFP ratio. Data represents the mean ± SEM from three independent experiments (n = 3). Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test (* p<0.05; *** p<0.001). μ m, magnification -63x). B -Mean fluorescence intensity of labeled ganglioside GM1 per cell measured using LAS X software. Data represents the mean ± SEM from 15 fields of view (n = 3), results normalized to control. Statistical significance was determined using Kruskal-Wallis test followed by Dunn's post-hoc test in GraphPad Prism 8.0.1 software (*** p<0.001; **** p<0.0001). C -Confocal images of microglial cells infected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3 Kit. D -Quantification data of GFP-LC3B and RFP-LC3B protein expression. Data shown represents the mean ± SEM from 15 fields of view (n = 3). Statistical analysis was performed using Kruskal-Wallis test followed by Dunn's post-hoc test for GFP-LC3B (E) and RFP-LC3B (F) protein expression (* p<0.05; ** p<0.01; *** p<0.001).