To mimic the general DM population, middle aged male Wistar rats (13 month of age) were subjected to intraperitoneal injection of NTM (210 mg/kg) and STZ (60 mg/kg). Non-fasting plasma glucose levels were measured 7 days after NTM/STZ injection and monthly thereafter using a portable blood glucose meter (AgaMatrix). Animals with glucose concentrations >200 mg/dL were enrolled. This model has been demonstrated to produce noninsulin-dependent DM syndromes that resemble human T2DM (Masiello et al., 1998). Aged-matched rats without NTM-STZ injection were used as controls.

Primary CECs were isolated from cerebral microvessels of healthy male adult rats (2–3 months of age) and aged-DM rats (15 months of age, 2 months of DM) according to published method (Teng et al., 2008). We have demonstrated that the purity of primary endothelial cells is above 90% (Teng et al., 2008). The sEVs were isolated from culture medium of passages 1 to 2 CECs using a ultracentrifugation method according to published protocol (Li et al., 2021). The size and particle number of sEVs were characterized by a nanoparticle tracking analyzer (NTA, NanoSight N300 System, Merkel Technologies, Israel). sEV morphology was examined by means of transmission electron microscopy (TEM, JEM-1500Flash, JEOL). Total proteins extracted from isolated sEVs were used for Western blot analysis for the identification of EV marker proteins. The following primary antibodies were used: CD63 (ab68418, 1:1,000, Abcam), Alix (ab88388, 1:1,000, Abcam), CD31 (ab124432, 1:1,000, Abcam), and Calnexin (ab22595, 1:1,000, Abcam).

To examine the therapeutic effects of N-CEC-sEVs, aged-DM rats at 2 months after NTM-STZ injection (15 months of age, n = 14/group) were randomly assigned to receive twice weekly intravenous administration of the N-CEC-sEVs at concentrations of 1 × 10¹¹ particles/injection, or the same volume of Vehicle (saline) for a total of four consecutive weeks. For detecting cell proliferation, bromodeoxyuridine (BrdU, 100 mg/kg), a thymidine analog, was injected (i.p) daily for 7 days starting 2M after NTM-STZ injection. Rats were euthanized 4 months after NTM-STZ injection (17 months) and brain tissues were collected for immunohistochemistry and Western blot analysis.

For evaluation of cognitive function, odor recognition test (Levy et al., 2004) and social interaction test (Sato et al., 2012) were performed at 2 and 4 months after DM onset. Morris water maze (MWM) test (Choi et al., 2006) was performed 4 months after DM onset (n = 10/group). These tests are well established in our laboratory and provide sensitive evaluation of cognitive function (Zhang et al., 2016).

The odor recognition test evaluates the non-spatial social olfactory memory that critically depends on adult neurogenesis (Levy et al., 2004). Briefly, rats were gradually introduced to four wooden beads consistent with two beads with its home cage odor (F1 and F2); one bead has a recently familiarized odor from a donor rat (N1), and one bead has a never encountered novel-odor from another donor rat (N2). The test was performed in three phases including a familiarization phase, an odor habituation phase, and an odor recognition memory test phase (Spinetta et al., 2008). During the initial familiarization phase, four unscented wooden beads (designated as F1–F4) were introduced into the home cage, where the testing rat would be familiarized with the presence of the beads for 24 h. The odor habituation phase of the task was done the next day after the now familiar beads with home cage odor (F1–F4) were removed for 1 h. The rat was exposed to four beads including three familiar beads (F1–F3) and a bead with odor from a donor rat (N1) for three 1-min trials with 1-min intertrial intervals. This procedure produces habituation to N1. The odor recognition memory test was conducted 24 h after the odor habituation phase. The testing rat was exposed to four beads including two with home cage odor (F1 and F2), one with recently familiarized odor (N1), and one with never encountered novel-odor from another donor rat (N2) following the same procedure outlined for the habituation phase. Exploration times for each bead were recorded. Data are presented as the percentage of exploration time on N2 relative to the total amount of exploration time spent on all beads. The focus of the test is to assess whether the test subject can identify the novelty of N2 by retaining the overnight memory for the N1 bead. Thus, a higher percentage of exploration time on N2 reflects better memory.

The social interaction test evaluates cognitive function in the form of sociability and social memory and novelty (Sato et al., 2012). Briefly, the social interaction test was performed in a three-chambered apparatus (Sato et al., 2012). After the testing rat habituated to the apparatus, a social target rat (R1) was randomly placed in the left- or right-compartment (the sociability test phase). The time spent by the rat exploring the R1 was recorded. The testing rat was then introduced to R1 rat and to a novel social target-never met before rat (R2) by randomly placing R1 and R2 in the left- or right-compartments (the social memory/novelty test phase). The times spent with R1 and R2 were recorded. The social memory/novelty data are presented as the percentage of time spent with R2 relative to the total amount of time spent with R1 and R2.

The MWM test is a widely used behavioral test for hippocampal-dependent spatial learning and memory (Choi et al., 2006). Rats were subjected to daily trial for five consecutive days starting at 4M after NTM-STZ injection. Briefly, on each testing day, the rat was placed in to a circular water tank (140 cm in diameter) with a transparent platform (15 cm in diameter) submerged 1.5 cm below the water at a random location within the Northeast (correct) quadrant of the tank. The rat was placed from four starting points (north, south, east, and west) in random order, and allowed to swim for a maximum of 90 s. The latency to find the hidden platform and the time spent within the correct quadrant were recorded. Data are presented as the latency to find the hidden platform (seconds) and the percentage of time spent within the correct quadrant relative to the total amount of time spent in the tank.

Immunofluorescent staining for brain tissue (n = 8/group) and culture cells were performed according to our published protocols (Zhang et al., 2016). For each staining, four coronal sections (8 μm in thickness) spaced as 50–100 μm intervals, at the levels of lateral ventricle (bregma –0.4 to –1.4 mm) and dorsal hippocampus (bregma –2.8 to –3.8) per rat were used. The following primary antibodies were used: mouse anti endothelial barrier antigen (EBA, 836804, 1:1000, Biolegend), goat anti fibrin/fibrinogen (YBGMFBG, 1:1000, Accurate Chem.), mouse anti BrdU (M0744, 1:100, Dako), goat anti doublecortin (DCX, 1:100, sc-8066, Santa Cruz), mouse anti-Nestin (556309, 1:100, BD Biosciences), and mouse anti-β-III tubulin (TuJ-1, 801202, 1:1000, Biolegend). For quantification of vascular damage, the numbers of EBA positive vessels with fibrin/fibrinogen immunoreactivity within hippocampus were counted and are presented as the density of immunoreactive vessels relative to the scan area (mm²) determined with an MCID image analysis system (Imaging Research). For quantitative analysis of cell proliferation, the numbers of BrdU positive cells were counted throughout the subventricular zone (SVZ) of the lateral ventricle and granule cell layer of dentate gyrus (DG). Data are presented as the density of immunoreactive cells relative to the area of the SVZ and DG. For quantification of neuroblasts, the numbers of DCX positive cells in the DG were counted and presented as the density of immunoreactive cells relative to the DG area. Moreover, the number and the total length of branches emerging from the soma of DCX positive cells in the DG were measured. The DCX immunoreactive areas within the SVZ were measured and presented as a percentage of positive immunoreactive area relative to the area of SVZ. The percentage of BrdU cells double-labeled for DCX and Nestin within the SVZ were measured.

Subventricular zone and DG tissues were manually collected from frozen coronal brain sections (n = 6/group), as previously described (Guo et al., 2012). Total RNAs were isolated using the miRNeasy mini Kit (Qiagen), followed by reverse transcription. Individual RT and TaqMan real-time PCR reactions of miR-1 and –146a were performed on an ABI 7000 instrument (Applied Biosystems). Each TaqMan assay was done in triplicate for each sample tested. Relative quantities were calculated using the 2-ΔΔCT method with U6 small nuclear RNA TaqMan miRNA control assay (Applied Biosystems) as the endogenous control. Ingenuity Pathway Analysis (IPA, Qiagen) was used to generate the microRNA and target gene interaction network. IPA network analysis revealed that miR-1 and –146a can directly and/or indirectly target genes of thrombospondin 1 (TSP1) and myeloid differentiation primary response gene 88 (MYD88). Protein levels of MyD88 and TSP1 in isolated SVZ and DG tissues were measured by Western blot. The following primary antibodies were used: MyD88 (PA5-19918, 1:1,000, Invitrogen), TSP1 (37879s, 1:1,000, Cell Signaling Technology), and β-actin (1:4,000, Abcam). Signal bands were visualized by the ECL system (Amersham). The relative densities among the blot lanes were analyzed using the MCID system (Imaging Research).

To examine the brain distribution of intravenously administered CEC-sEVs, N-CEC-sEVs carrying CD63-GFP (N-CEC-sEVs-GFP) were isolated from CECs transfected with a plasmid carrying pEGFP-CD63 vector (Addgene) according to our previously published protocols (Zhang et al., 2021). Normal adult rats (n = 4) were subjected to intravenous injection of N-CEC-sEVs-GFP (1 × 10¹¹ particle/injection) via a tail vein. Brain tissues were collected at 4 h after administration of N-CEC-sEVs-GFP. To examine the internalization of N-CEC-sEVs-GFP at the ultra-structural level, the lateral ventricle tissues were subjected to immunogold staining in which we employed streptavidin-gold nanoparticle (10 nm) conjugate (25269, EMS) and an anti-rabbit monoclonal antibody against GFP (G10362, ThermoFisher). TEM analysis was performed according to our published protocols (Zhang et al., 2021).

A sEV uptake assay was used to evaluate the internalization of exogenous sEVs by CECs. Briefly, N-CEC-sEVs or DM-CEC-sEVs were fluorescent-labeled with the ExoGlow Protein EV labeling kit (BSI# EXOGP300A-1) according to the manufacturer’s instruction. Primary CECs isolated from healthy young adult rats were cultured with fluorescent-labeled N-CEC-sEVs and DM-CEC-sEVs (3 × 10⁷ particles/ml) for 1 h, respectively. To further examine whether endocytosis pathways mediate the cellular uptake of sEVs, CECs pretreated with Pitstop-2 (clathrin inhibitor, 20 μM) and Nystatin (caveolin inhibitor, 20 μM) for 24 h were cultured with fluorescent-labeled sEVs (3 × 10⁷ particles/ml). The fluorescent signal in CECs were determined with an MCID image analysis system (Imaging Research).

To investigate whether sEVs modulate the barrier function of CECs, we harvested CECs from aged-DM rats (DM-CECs) 2M after NTM-STZ injection and age-matched non-DM rats (N-CECs). Endothelial monolayer barrier assay were performed according to our published methods³³. Briefly, CECs harvested non-DM rats (N-CECs) and DM rats (DM-CECs, passage 3 to 4) were cultured for 7 days at a density of 5 × 10⁴ per well and were then seeded as monolayer onto the inner chamber of trans-well insert (Corning) in the presence or absence of DM-CEC-sEVs or N-CEC-sEVs (3 × 10⁷ particles/ml) for 48 h, respectively. Fluorescent-conjugated dextran (0.5 mg/mL, 70 kDa, D1830; Thermo Fischer Scientific) was then added to the top of the trans-wells for 30 min. Fluorescent signals in the bottom medium were measured using a plate reader at excitation and emission wavelengths of 595 and 615 nm, respectively. Trans-endothelial permeability was calculated as % signals = (optical density experimental-optical density vehicle)/optical density vehicle × 100. To examine whether blocking sEV uptake abolishes its effects on endothelial cell barrier function, CECs pretreated with endocytosis inhibitors were cultured in the presence and absence of DM-CEC-sEVs or N-CEC-sEVs. All data were obtained from 3 individual experiments.

In the present study, the term of NSCs was used which includes neural progenitor and stem cells. In rodent brain, NSCs are present in the SVZ of the lateral ventricle and in the subgranular zone of the DG, and these cells generate new neurons throughout life (Zhang et al., 2004; Deng et al., 2009). To examine whether CEC derived sEVs modulate NSC function, the SVZ were dissociated from aged-DM rats (15 months of age, after 2 months of DM) and age-matched non-DM rats, as previously reported (Zhang et al., 2004). The cells were cultured in the medium containing 20 ng/ml basic fibroblast growth factor and epidermal growth factor at a cell density of 1 × 10⁴ cells/ml. To test the effects of DM-CEC-sEVs on NSCs harvested from healthy non-DM rats, SVZ cells from age-matched non-DM rats were cultured with and without DM-CEC-sEVs (3 × 10⁷ particles/ml). To test whether N-CEC-sEVs ameliorate cell proliferation and neuronal differentiation of NSCs harvested from aged-DM rats, SVZ cells from aged-DM rats were cultured with and without CEC-sEVs (3 × 10⁷ particles/ml). The number and size of neurospheres were measured after 7 day culture. To examine the effects of CECs derived sEVs on cell proliferation and neuronal differentiation, neurosphere cells were plated onto laminin-coated glass coverslips for differentiation assay according to our previously published protocols (Zhang et al., 2004). For cultured cells, BrdU and TuJ1 positive cells as well as total cells identified by 4’,6-diamidino-2-phenylindole nuclei counterstain were counted and presented as the percentage of each cell type in the total cells numbers.

Repeated measure of analysis (ANCOVA) was performed to study the effects of CEC-sEV treatment on odor test, social chamber test, and MWM over time, respectively, and analysis considered treatment-by-time interaction. A significant time by treatment interaction (p-value < 0.05) indicated the treatment effects are dependent on time. One-way ANOVA with post hoc Bonferroni tests was used for multiple groups comparisons. Two-sample t-test was used for two group comparisons. Data are presented as mean ± standard error. A p-value < 0.05 was considered significant.

To examine the effect of DM-CEC-sEVs on neurogenesis, we performed ex vivo experiments. CEC-sEVs were isolated from the medium of cultured primary CECs harvested from normal healthy young adult (2 to 3 months, N-CEC-sEVs) and aged-DM (15 month of age, 2M after NTM-STZ injection, DM-CEC-sEVs) male rats. Using multiple methods including NTA, TEM, and Western blot analysis, we found that N-CEC-sEVs and DM-CEC-sEVs had an average particle size of 70.5 ± 1.1 (N-CEC-sEVs) and 70.8 ± 1.4 (DM-CEC-sEVs), cup-shaped morphology, and exhibited sEV membrane marker proteins, but not intracellular protein calnexin, respectively, (Figure 1), which are consistent with the Minimal Information for Studies of Extracellular Vesicles 2018 (MISEV2018) report (Théry et al., 2018). There were no significant differences of the particle size, morphology, and protein component between N-CEC-sEVs and DM-CEC-sEVs.

We then harvested NSCs from DM rats and age matched non-DM rats. Compared to NSCs harvested from non-DM rats, NSCs from DM rats showed significant reduction of neurosphere formation, cell proliferation, and neuronal differentiation (Figures 2A,C,D), which is consistent with published studies (Bonds et al., 2020). Treatment of NSCs harvested from DM rats with N-CEC-sEVs significantly increased neurogenesis (Figures 2C,D), whereas treatment of NSCs harvested from non-DM rats with DM-CEC-sEVs robustly suppressed neurogenesis (Figures 2A,B). These ex vivo data indicate that sEVs derived from dysfunctional CECs inhibit neurogenesis, whereas sEVs derived from healthy CECs alleviate DM-impaired neurogenesis.

To extend aforementioned ex vivo finding, DM rats at age of 15 months (2M after NTM-STZ injection) were treated with N-CEC-sEVs (1 × 10¹¹ particles/injection) two times a week for four consecutive weeks via a tail vein. To examine the effect of N-CEC-sEVs on neurogenesis, we employed a BrdU chasing approach that is a standard method to study neurogenesis (Wojtowicz and Kee, 2006), in which BrdU was given for seven consecutive days to the rat at age of 15 months while BrdU immunoreactive cells were examined 2 months after termination of BrdU treatment. Compared to age matched non-DM rats, DM rats had significant reduction of DCX+ neuroblasts and BrdU+ cells in the DG and the SVZ (Figures 3A–F), which is consistent with our previous findings (Zhang et al., 2016). Treatment of DM rats with N-CEC-sEVs significantly increased the number of BrdU+ cells and the DCX+ neuroblasts in the DG and the SVZ compared to saline treated DM rats (Figures 3A–F). Moreover, N-CEC-sEV treatment significantly increased BrdU+/DCX+ double labeled cells but not the BrdU+/Nestin+ cells in the SVZ compared to saline treatment (Figures 3G,H). Confocal microscopy analysis of DCX immunoreactive cells in the DG revealed that the N-CEC-sEV treatment significantly increased branch number and length of DCX⁺ neuroblasts compared to DCX+ neuroblasts in saline treated DM rats (Figures 3I,J), suggesting that DCX⁺ neuroblasts acquire a more mature phenotype after the N-CEC-sEV treatment. Collectively, these data indicate that N-CEC-sEV treatment enhances neurogenesis in DM rats.

Neurogenesis contributes to cognitive function (van Praag et al., 2002; Deng et al., 2009). An array of behavioral tests was employed to examine cognitive function, which included an odor recognition test to evaluate the social recognition memory, a social interaction test to assess social memory and novelty, and a MWM test to specifically detect hippocampal-dependent spatial learning and memory (Levy et al., 2004; Choi et al., 2006; Sato et al., 2012). Compared to age matched non-DM rats, DM rats treated with saline spent significantly less time to explore a novel odor object and to interact with a new rat (Figures 4A,B). In contrast, DM rats treated with N-CEC-sEVs spent significantly more time (p < 0.05) on a novel object and to interact with a new rat compared to DM rats treated with saline. In addition, the MWM test revealed that DM rats treated with N-CEC-sEVs significantly increased their percent time in the correct quadrant and the escape latency compared to DM rats treated with saline (Figure 4C). These data indicate that the N-CEC-sEV treatment improves cognitive function in DM rats. The N-CEC-sEV treatment did not significantly alter blood glucose levels and body weight in DM rats (Figures 4D,E).

We previously demonstrated that DM elicits cerebrovascular disruption characterized by microvascular thrombosis and blood brain barrier (BBB) leakage mainly localized to the hippocampus (Zhang et al., 2016). Thus, in addition to neurogenesis, we examined the effect of N-CEC-sEVs on DM-induced vascular damage. Treatment of DM rats with N-CEC-sEVs significantly reduced vascular thrombosis and BBB leakage as indicated by the decreased number of vessels with intra and extravascular fibrin deposition, respectively, in the hippocampi compared to DM rats treated with saline (Figure 5).

Using a cerebral endothelial permeability assay, we then examined a direct effect of N-CEC-sEVs on CECs. Compared to primary CECs harvested from age matched non-DM rats, CECs from DM rats showed significant leakage. However, addition of N-CEC-sEVs to CECs harvested from DM rats robustly reduced endothelial cell leakage (Figure 6A). Together, these in vivo and ex vivo data indicate that N-CEC-sEVs can reduce DM-induced cerebral vascular damage.

The clathrin and caveolin signaling pathways mediate endocytosis of extracellular vesicles (Mulcahy et al., 2014). We thus, examined whether these two pathways are involved in the up-take of sEVs by CECs. After incubation of CECs harvested from non-DM rats with fluorescently labeled CEC-sEVs derived from non-DM rats (N-DM-CEC-sEVs) or from DM rats (DM-CEC-sEVs), fluorescent signals were detected within the cytosol of CECs, indicating that endothelial cells internalize sEVs (Figure 6B). There were no significant differences of EV uptake between N-DM-CEC-sEVs and DM-CEC-sEVs (Figure 6B). Pretreatment of endothelial cells with pitstop2, a cell-permeable clathrin inhibitor, or nystatin that inhibits caveolin-dependent endocytosis, significantly reduced uptake of N-DM-CEC-sEVs and DM-CEC-sEVs (Figure 6B), suggesting both signaling pathways are involved in cerebral endothelial cell endocytosis of CEC-sEVs. Moreover, inhibition of endocytosis by the two inhibitors significantly reduced DM-CEC-sEVs increased endothelial cell permeability (Figure 6A), indicating that impairment of endothelial cell permeability by DM-CEC-sEVs is specific.

We previously demonstrated that intravenously administered CEC-sEVs cross the BBB (Li et al., 2021). However, whether systemic administration of N-CEC-sEVs reach to NSCs has not been investigated. To track N-CEC-sEV in neurogenic regions, N-CEC-sEVs carrying CD63-GFP (N-CEC-sEVs-GFP) were intravenously administered to non-DM adult rats. The rats were sacrificed 4 h after the treatment and their brains were processed for analysis. Confocal microscopic analysis (Figures 7A–C) showed that GFP signals were detected within NSCs in the neurogenic regions of the SVZ and the subgranular zone of DG (Figures 7B,C). Ultrastructural analysis (Figures 7D–F) revealed that GFP-immunogold positive particles were localized to cytoplasm and nucleus of NSCs in the SVZ (Figures 7E,F), indicating that NSCs internalize CEC-sEVs. Consistent with our previous findings (Li et al., 2021), GFP signals (Figure 7A) and GFP-immunogold positive particles (Figure 7D) were also detected in cerebral endothelial cells.

We and others have demonstrated that miRNAs regulate neurogenesis through modulating their target genes (Chamorro-Jorganes et al., 2011). Small EVs impact recipient cell biological function by delivering their cargo materials including miRNAs (O’Brien et al., 2020; Zhang et al., 2021). Using qRT-PCR analysis of N-CEC-sEVs, we found that miR-1 and –146a were 2.2 and 1.9-fold higher than DM-CEC-sEVs (Figure 8A). Next, we measured these two miRNA levels in isolated DG and SVZ tissues. Compared to age matched non-DM rats, DM dramatically decreased miR-1 and –146a levels, whereas the N-CEC-sEV treatment significantly increased miR-1 and –146a levels compared to the tissues isolated from saline treated DM rats (Figure 8B). Bioinformatics analysis revealed that miR-1 and –146a can directly and/or indirectly target genes of MYD88 and TSP1. MYD88 related toll-like receptor (TLR) signaling is involved in vascular dysfunction, while TLR regulate adult neurogenesis (Rolls et al., 2007; Okun et al., 2011). Western blot analysis of DG and SVZ tissues showed that compared to age-matched non-DM rats, aged-DM rats exhibit significantly increased MYD88 and TSP1 levels. In contrast, the N-CEC-sEV treatment significantly reduced MYD88 and TSP1 levels compared to the saline treatment (Figures 8C,D). An inverse relationship of the miR-1 and –146a and the protein levels of MYD88 and TSP1 in the DG and SVZ suggests that alterations of this network by CEC-sEVs likely contribute to the therapeutic effect of CEC-sEVs on DM rats.