Extracellular vesicle-Serpine-1 affects neural progenitor cell mitochondrial functions and synaptic density: modulation by amyloid beta and HIV-1

Brain endothelial extracellular vesicles carrying amyloid beta (EV-Aβ) can be transferred to neural progenitor cells (NPCs) leading to NPC dysfunction. However, the events involved in this EV-mediated Aβ pathology are unclear. EV-proteomics studies identified Serpine-1 (plasminogen activator inhibitor 1, PAI-1) as a major connecting “hub” on several protein-protein interaction maps. Serpine-1 was described as a key player in Aβ pathology and was linked to HIV-1 infection as well. Therefore, the aim of this work was to address the hypothesis that Serpine-1 can be transferred via EVs from brain endothelial cells to NPCs and contribute to NPC dysfunction. HBMEC concentrated and released Serpine-1 via EVs, the effect that was potentiated by HIV-1 and Aβ. EVs loaded with Serpine-1 were readily taken up by NPCs, and HIV-1 enhanced this event. Interestingly, a highly specific Serpine-1 inhibitor PAI039 increased EV-Aβ transfer to NPCs in the presence of HIV-1. PAI039 also partially blocked mitochondrial network morphology and mitochondrial function alterations in the recipient NPCs, which developed mainly after HIV + Aβ-EV transfer. PAI039 partly attenuated HIV-EV-mediated decreased synaptic protein levels in NPCs, while increased synaptic protein levels in NPC projections. These findings contribute to a better understanding of the complex mechanisms underlying EV-Serpine-1 related Aβ pathology in the context of HIV infection. They are relevant to HIV-1 associated neurocognitive disorders (HAND) in an effort to elucidate the mechanisms of neuropathology in HIV infection.


Introduction
Extensive evidence indicates amyloid pathology in HIV-1 infection. Studies reported that HIV-infected brains have increased amyloid beta (Aβ) deposition mostly in the perivascular space [1][2][3][4] when compared to age-matched controls [1,[5][6][7][8][9]. These earlier ndings have recently been reevaluated and expanded, as new evidence from autopsy of HIV-1 brains did not con rm an overall enhanced Aβ deposition. Instead, duration of infection was correlated with Aβ accumulation independent of age. These ndings suggest that the length of HIV infection, and not the age of patients, predicts elevated brain Aβ levels, con rming an accelerated brain senescence in people living with HIV [10]. For example, methylome-wide analysis of chronic HIV-1 infection revealed ve-year increase in biological age of infected individuals [11]. HIV-related age acceleration has been shown to be associated with reductions in total gray matter using epigenetic age as a biomarker for age acceleration [12]. Evidence indicates that increased and persistent senescence during HIV-1 infection contributes to chronic in ammation, immune failure and mitochondrial dysfunction [13][14][15].
It was demonstrated that the blood-brain barrier (BBB) is critical for Aβ homeostasis and contributes to elevated Aβ deposition in the brain [16,17]. Aβ transport into the brain across the BBB involves BBB transfer mechanisms such as the receptor for advanced glycation end products (RAGE) [18]. Moreover, we found that HIV-1 contributes to Aβ accumulation in brain endothelial cells via upregulation of RAGE [19]. Extracellular vesicles (EVs) also appear to be important players in Aβ pathology by carrying biologically active proteins and genetic material. We demonstrated that exposure to HIV-1 increases EV release from human brain microvascular endothelial cells (HBMEC) and alter their Aβ cargo [20]. EVs France). HBMEC were cultured on collagen type I (BD Biosciences Pharmingen, San Jose, CA) coated dishes in EBM-2 medium (Clonetics, East Rutherford, NJ) supplemented with VEGF, IGF-1, EGF, basic FGF, hydrocortisone, ascorbate, gentamycin and 0.5% exosome depleted fetal bovine serum (Exo-FBS; System Biosciences, Mountain View, CA).

Human neural progenitor cells (NPCs)
An immortalized NPC line ReNcell VM, derived from 10-week human ventral mesencephalon, was obtained from Millipore and cultured according to the manufacturer's protocols. The cells were validated for high expression of Sox2 and nestin as well as for their self-renewal and differentiation capacity. Cells were grown on laminin coated tissue culture dishes in a maintenance medium (Millipore), containing 20 ng/ml FGF-2 and 20 ng/ml of rhEGF. Cells were used for experiments at < 80% con uence, three days after plating.

HIV-1 infection and treatment factors
HIV-1 stock was generated using human embryonic kidney (HEK) 293T cells (ATCC) transfected with pYK-JRCSF plasmid containing full-length proviral DNA. Throughout the study, HBMEC were exposed to HIV-1 particles at the p24 level of 30 ng/ml as previously reported [22]. Treatment was terminated by removing cell culture media containing HIV-1, followed by washing the cells with PBS.
Aβ  and Aβ  HiLyte 647 were purchased from Anaspec (San Jose, CA) and dissolved in PBS.
Freshly solubilized Aβ solutions without pre-aggregation were used for experiments as such a form of Aβ was demonstrated to induce proin ammatory reactions [33]. Aβ  HiLyte was dissolved rst in a basic buffer (0.1 M NH 4 OH) and then diluted further in PBS as suggested by the manufacturer. Cells were treated with Aβ  or Aβ  HiLyte at the concentration of 100 nM in complete medium. PAI039 (Tiplaxtinin, Catalog # PZ0295) was purchased from Millipore Sigma, Burlington, MA, USA).
PAI039 is a potent and selective Serpine-1 inhibitor [34] and demonstrated e cacy in vivo in multiple models of acute arterial thrombosis. A 20 mM stock solution was prepared in DMSO. In a typical experiment, NPCs were cotreated with isolated EVs and/or 2 µM PAI039 for 24 h. Literature indicates that 1 µM PAI-039 can effectively inhibit Serpine-1 activity in vitro [35]. PAI039 exerts its activity by binding close to the vitronectin binding site [36].

Treatment of brain endothelial cells and EV isolation
Con uent HBMEC were exposed to HIV-1, Aβ  and/or Aβ  HiLyte for 48 h. EVs were isolated from the media using ExoQuick-TC exosome precipitation solution (System Biosciences, Mountain View, CA) according to the manufacturer's speci cations. Brie y, 10 ml culture medium from 1.7 x 10 7 cells at con uency cultured in a P100 dish was centrifuged at 3000 g for 15 min to remove cells and debris.
Then, the samples were mixed thoroughly with 2 ml of Exo-Quick exosome precipitation solution and incubated overnight at 4°C. The next day, the samples were centrifuged at 1500 g for 30 min; the supernatants were removed and centrifuged again at 1500 g for 5 min. The EV pellets were resuspended in 400 µl PBS and used for further studies. The aliquots of 20 µl of EV suspension for every 100 µl of cell culture media was used for NPC treatment.
Transfection of brain endothelial cells HBMEC were transfected with the CD63 RFP and Serpine-1 GFP constructs (Vectorbuilder, Chicago, IL, USA) using Purefection Transfection Reagent (System Biosciences) following the manufacturer's protocol. Twenty-four hours post transfection, cells were exposed to HIV-1 or/and Aβ (1-40) HiLyte for 48 h followed by EV isolation from the media. We have observed that even minor changes of HIV-1 stock preparation can lead to differences in EV release from HBMEC exposed to HIV-1. If HEK293T cells were cultured in Exo-FBS, their morphology changed, and the resulting viral stock had vastly different effects on EV release and EV isolation. Therefore, HEK293T cells were cultured in media containing regular FBS consistent with our previous reports [20,22]. As a control for HIV-1 exposure, isolates from mocktransfected HEK293T cells were used.

Nanoparticle tracking analysis (NTA)
EVs were analyzed by NanoSight model NS300 (Malvern Instruments Company, NanoSight, Malvern, United Kingdom) as described earlier [20]. Brie y, EV pellet samples obtained in the process of EV isolation were resuspended in 4% paraformaldehyde-PBS and further diluted 50-fold in PBS for analysis.
During analysis, ve 15 s videos were recorded for each sample. The obtained data were analyzed using Nanosight NTA Analytical Software (Malvern Instruments Company) with the detection threshold optimized for each sample and screen gain at 10 to track the maximal number of particles with minimal background.
Brie y, a xed amount of tissue-type plasminogen activator (tPA) was added in excess to undiluted sample, which allowed Serpine-1 and tPA to form an inactive complex. The assay measures plasminogen activation by residual tPA in coupled assays that contain tPA, plasminogen, and a plasmin-speci c synthetic substrate. The amount of plasmin produced was quantitated using a highly speci c plasmin substrate releasing a yellow para-nitroaniline chromophore. The absorbance of the chromophore at 405 nm was inversely proportional to the Serpine-1 enzymatic activity. One arbitrary unit (AU) of inhibition was de ned as the amount of Serpine-1 that can inhibit one IU of tPA/ml under the testing conditions. EV-Serpine-1 activity was also measured in the presence of Tx-100 1%, which was used to lyse EVs.
Tissue plasminogen activator (tPA) assay tPA activity was measured with the Tissue type Plasminogen Activator Activity Assay Kit (Abcam, #ab108905) following the manufacturer's instructions. Brie y, the tPA assay protocol measures the ability of tPA to activate the plasminogen to plasmin in coupled or indirect assays that contain tPA, plasminogen, and a plasmin-speci c synthetic substrate. The amount of plasmin produced was quanti ed using a highly speci c plasmin substrate releasing a yellow para-nitroaniline chromophore.
The change in absorbance of the chromophore in the reaction solution at 405 nm was directly proportional to the tPA enzymatic activity.
Confocal microscopy HBMEC cultured on collagen I coated chambered glass slides (Becton Dickinson Biosciences, San Jose, CA) were transfected with the CD63 RFP and Serpine-1 GFP constructs as described above. The parent cells expressing CD63 RFP and Serpine-1 GFP were imaged live after 24 h using an inverted confocal laser-scanning microscope (Olympus Fluoview 3000). Twenty-four hours post transfection, cells were exposed to HIV-1 or/and Aβ (1-40) HiLyte 647 for 48 h. At the end of treatment, the culture media was removed for uorescent EV isolation; cells were washed with PBS and xed with ethanol for 30 min at 4°C. CD63 RFP, Serpine-1 GFP and Aβ (1-40) HiLyte 647 uorescence was imaged with an inverted confocal laser-scanning microscope (Olympus Fluoview 3000, objective lens UPLXAPO100XO 100X oil, numerical aperture 1.45) and analyzed using CellSens software.
After isolation from the media, uorescent EVs were pipetted onto cleaned glass slides, heat xed for 10 min at 95°C, and then xed with ethanol for 30 min at 4°C followed by PBS wash. Slides were mounted using ProLong Gold Antifade reagent with or without 4',6-diamidino-2-phenylindole (DAPI, Invitrogen, Carlsbad, CA, USA) to visualize the nucleic material in EVs. Specimens were covered with coverslips and the uorescent images were evaluated and captured under confocal microscopy. Red uorescence originating from EV-CD63 RFP, blue uorescence from DAPI, green uorescence from EV-Serpine-1 GFP and far-red uorescence from EV-Aβ HiLyte-Alexa Fluor 647 was acquired directly using confocal microscopy (Olympus, Fluoview 3000, 100x oil immersion lens, room temperature). Serpine-1 GFP positive, CD63 RFP positive EVs were counted with the CellSens software and expressed as percentage of the total DAPI positive EVs. Serpine-1 GFP and CD63 RFP double positive EVs were counted and expressed as percentage of the total Serpine-1 GFP positive EV number.
NPCs were seeded on laminin coated 8-well chambered glass slides (15,000 cells/well) and incubated overnight at 37°C in maintenance culture medium (Millipore), containing 20 ng/ml FGF-2 and 20 ng/ml of rhEGF. The following day, the medium was changed to maintenance medium without growth factors to induce differentiation. Cells were allowed to differentiate for a total of 3 days, with the last 24 h in the presence of the isolated EVs containing Serpine-1 GFP, CD63 RFP +/-Aβ (1-40) HiLyte 647. After 1 h of EV exposure, CD63 RFP, Serpine-1 GFP and Aβ  HiLyte 647 uorescence in the acceptor nonuorescent NPCs was imaged live with an inverted confocal laser-scanning microscope. After 24 h of EV treatment, NPCs media was removed, cells were washed with PBS and xed with ethanol for 30 min at 4°C. CD63 RFP, Serpine-1 GFP and Aβ (1-40) HiLyte 647 uorescence in the acceptor non-uorescent NPCs was imaged randomly as above. HBMEC were exposed to HIV-1 or/and Aβ (1-40) HiLyte 488 for 48 h followed by EV isolation from the conditioned media. NPCs cultured on chambered slides and differentiated as above, were exposed to EV-Aβ HiLyte 488 for 24 h. Then, the media was removed, cells were washed with PBS and xed with ethanol for 30 min at 4°C. Cells were washed again with PBS and transferred Aβ HiLyte 488 uorescence was assessed by confocal microscopy.
On some confocal images we have increased brightness or contrast for better visibility of the uorescence signal. These changes were consistent across all treatment groups to preserve the integrity of the data.

Mitochondrial stress Seahorse assay
The Seahorse XFe24 Analyzer was used to calculate oxygen consumption rate (OCR; a measure of mitochondrial respiration) and extracellular acidi cation rate (ECAR; a measure of glycolysis) in NPCs using the Agilent Seahorse XF Cell Mito Stress Test Kit (Agilent Technologies, Santa Clara, CA, USA). Cells were seeded in a 24XF cell culture microplate at 30,000 cells/well differentiated and treated with brain endothelial EVs as described above. After EV treatment for 24 h, the cell culture media were replaced with 500 µl of the prepared Seahorse medium (containing Seahorse XF DMEM media, 2 mM L-glutamine, 10 mM glucose, and 1 mM pyruvate) and incubated at 37°C for 1 h. Measurements of mitochondrial respiration and glycolysis were carried out as previously described [38]. In brief, cells were treated with 1.5 µM concentrations of oligomycin (ATP synthase inhibitor of complex V), 1 µM carbonyl cyanide-ptri uoromethoxyphenylhydrazone (FCCP, electron transport chain (ETC) uncoupler), and 0.5 µM rotenone with antimycin A (both ETC inhibitors of complex I and III, respectively) throughout the analysis. These treatments, which were added to the cells at speci ed time points, allowed for calculations of mitochondrial respiratory parameters, such as baseline OCR, ATP production, maximal respiration, proton leak, and non-mitochondrial oxygen consumption. After three basal measurements of OCR and ECAR were recorded, oligomycin was injected to inhibit ATP synthase and two more measurements were recorded to assess proton leak. Next, FCCP was injected to uncouple respiration and measure maximal respiration. Finally, antimycin A was injected to measure non-mitochondrial respiration. All OCR measurements were normalized to non-mitochondrial respiration and the nal values normalized to NPC protein concentration in each well. Reserve capacity is the difference between maximal respiration and basal respiration, while ATP-linked OCR is the difference between basal and proton leak. The data were analyzed using the Wave Software (Agilent Technologies). All conditions were measured in 4-6 samples/group, and three repeats were performed for this experiment.
Expression of synaptic proteins EV-treated NPCs cultured on laminin coated chambered glass slides (ibidi USA, Madison, WI, USA) were xed with ethanol for 30 min at 4°C. After washing with PBS and blocking with 3% bovine serum albumin in PBS for 30 min at room temperature, samples were incubated overnight at 4°C with the primary antibody: mouse anti-PSD95 monoclonal antibody (Abcam, Waltham, MA, USA, Catalog #192757, 1:1000) or rabbit anti-synaptophysin polyclonal antibody (Abcam, 1:1000). Then, the excess of primary antibody was removed, slides were washed with PBS, and incubated with Alexa Fluor 488/594conjugated secondary antibodies (1:1000, Invitrogen) for 2 h at room temperature. Nuclei were stained with Hoechst 33342 (Invitrogen, Catalog #H3570). The immuno uorescent images were evaluated and captured under confocal microscopy. Red uorescence originating from PSD95, blue uorescence from Hoechst 33342, and green uorescence from synaptophysin was acquired directly using confocal microscopy (Olympus, Fluoview 3000, 100x oil immersion lens, room temperature).
Quantitative analysis of synaptic protein expression was performed similarly to a previously published method, with modi cations [39]. After immuno uorescence staining for PSD95 and synaptophysin, elds were selected randomly using the guidance of DAPI nuclear staining (three images/treatment group, total 9 images from three independent experiments), then confocal images were taken. On each confocal image (bright eld), 20 identical rectangular areas were randomly superimposed on different segments of NPC projections. The selected areas were analyzed for green (synaptophysin) and red (PSD95) uorescence intensity using the NIH Image J software (Bethesda, MD, USA). Mean Fluorescence Intensity (MFI) for each area was normalized by subtracting the background uorescence intensity for that image. Synaptic protein densities were expressed relative to those in control NPCs. For total synaptophysin and PSD95 assessment, mean uorescence intensity on the acquired images was normalized to the number of NPC nuclei.

Statistical analysis
Data were analyzed using GraphPad Prism 9.0 (Graphpad Software, San Diego, CA). ANOVA was used to compare responses among treatments. Treatment means were compared using All Pairwise Multiple Comparison Procedures and p < 0.05 was considered signi cant.

Results
Serpine-1 is concentrated in EVs released from control and HIV-1 exposed brain endothelial cells Secretion of Serpine-1 in EVs was traced by cotransfection of HBMEC with the Serpine-1 GFP and CD63 RFP constructs. The tetraspanin CD63 is a membrane protein, which is predominantly localized to the vesicles, and; therefore, commonly used as a biomarker for EVs. As demonstrated by live uorescence microscopy 24 h post transfection, transfected cells appeared to concentrate and secrete Serpine-1 GFP in CD63 RFP-positive EVs. Indeed, green uorescence, corresponding to Serpine-1, noticeably overlapped with CD63 RFP-positive red uorescent EVs budding off from the parent cells (arrow heads for single markers, arrows for overlapping uorescence, Fig. 1A). We also measured the total number of EVs in the media originating from non-transfected HBMEC. EVs released from control HBMEC had a total EV concentration of 43.05 ± 6.65 × 10 8 particles/ml, while EVs released from HIV-1 treated HBMEC had a total EV concentration of 83.40 ± 34.60 × 10 8 particles/ml. This HIV-related increase in EVs release is consistent with our previous report [20].
Because both EVs and Serpine-1 were shown to play a role in Aβ pathology [20,30,31], we next evaluated the impact of HIV-1 on Serpine-1 and Aβ release via EVs. For these experiments, 24 h after cotransfection with Serpine-1 GFP and CD63 RFP, HBMEC were exposed to HIV-1 (30 ng p24/ml) and/or 100 nM Aβ (1-40) HiLyte 647 for 48 h. The treatment was terminated by removing the cell culture media for EV isolation, followed by washing with PBS and xing the parent cells. As illustrated in Figs. 1B and 1C, Serpine-1 GFP green uorescence and CD63 RFP red uorescence partially overlapped with the uorescent Aβ HiLyte 647 (yellow) taken up by the parent cells (arrow heads for single-, arrows for overlapping uorescence). The images were quanti ed for Serpine-1, CD63, or Aβ HiLyte colocalization (Fig. 1D), indicating that Serpine-1 GFP and Aβ HiLyte 647 colocalization signi cantly increased in the HIV + Aβ group when compared to control (Fig. 1D, lower right graph).
Next, Serpine-1 GFP, CD63 RFP and Aβ (1-40) HiLyte 647 uorescence was visualized in EVs isolated from cell culture media. Serpine-1 GFP was detected in EVs of different sizes in control, Aβ and/or HIV-1 treated samples ( Fig. 2A) indicating that the parent cells secrete Serpine-1 via EVs. Interestingly, overall fewer CD63-RFP positive EVs were present in all groups and Aβ HiLyte 647 occasionally colocalized with both Serpine-1 GFP and CD63-RFP in the secreted EVs. In the Aβ groups we frequently observed aggregates of Aβ HiLyte 647 with associated EVs ( Fig. 2A, arrow heads for single markers, arrows for overlapping uorescence). We also stained the EV genetic material with DAPI (blue uorescence). Most of the EVs showed DAPI uorescence indicating DNA/RNA cargo.
To further characterize the isolated EVs, we counted the number of Serpine-1 GFP-, CD63 RFP-and DAPI positive EVs from the confocal microscopy images. As quanti ed on the graphs from Fig. 2B, most of the DAPI containing EVs were also positive for Serpine-1 GFP. This percentage did not change signi cantly in any treatment group. In contrast, the percentage of CD63 RFP positive EVs was much lower and signi cantly decreased in the HIV-EV and the HIV + Aβ-EV groups as compared to the Aβ-EV group. Similarly, the number of Serpine-1 GFP and CD63 RFP double positive EVs changed in the same way (Fig. 2B).
HIV-1 impacts Serpine-1 levels and activity in the released EVs Serpine-1 levels in the EV lysates were next assessed by ELISA in non-transfected HBMEC and normalized either to cell culture media volume or to EV protein content (Figs. 3A). EVs isolated from control, Aβ, and/or HIV-treated HBMEC cultures contained Serpine-1. Importantly, Serpine-1 levels were signi cantly higher in the HIV-1 group as compared to the control or Aβ groups when normalizing to cell culture volume (Fig. 3A, left graph). Serpine-1 levels in the HIV + Aβ group were also signi cantly increased when compared only to control (Fig. 3A, middle graph). Nevertheless, EV Serpine-1 concentration was similar in all treatment groups after normalization to EV protein levels (Fig. 3A, right graph), presumably because exposure to HIV-1 increases the overall EV number produced as reported before [20]. Interestingly, Serpine-1 levels in the parent cells were in the picogram range and thus much lower as compared to the nanogram range of Serpine-1 in EVs. Moreover, Serpine-1 levels in the parent cells did not change after HIV-1 and/or Aβ treatment (Fig. 3B).
Serpine-1 is an enzyme, therefore, its activity was also measured from the isolated EVs lysed with 1% Tx100 (see Methods). HIV + Aβ-EVs had the highest Serpine-1 activity, which was statistically signi cant when compared to the control and Aβ groups and remained signi cant even after normalizing the results to EV protein content (Fig. 3C).
Because Serpine-1 is known to inhibit tPA activity [24]; therefore, we also measured tPA activity in the isolated EVs. As illustrated on Fig. 3D, tPA activity trends were the opposite of those of Serpine-1 activity. Treatment with Aβ, HIV-1 and HIV + Aβ of the parent cells had a signi cant impact on tPA activity in EVs, which signi cantly decreased in these groups as compared to control, when normalizing to cell culture volume (Fig. 3D, left graph). A decreasing trend was also observed when normalizing the results to EV protein content (Fig. 3D, right graph) but the changes were not statistically signi cant.
HBMEC-derived EVs transfer Serpine-1 cargo to recipient neural progenitor cells HBMEC are part of a functional unit at the BBB consisted of pericytes, perivascular astrocytes, microglia, and neurons, called the neurovascular unit [40]. Therefore, we hypothesized that HBMEC-derived EVs can transfer Serpine-1 to neighboring cells of the neurovascular unit, including neural progenitor cells (NPCs).
It was shown that ~ 47% of dividing progenitor and 46% of transit amplifying cells (precursors of neuroblasts) are located in close proximity to the brain endothelium [41,42].
In order to assess Serpine-1 transfer to NPCs, HBMEC transiently transfected with Serpine-1 GFP and CD63-RFP were exposed to 100 nM Aβ HiLyte 647 and/or HIV-1 for 48 h, resulting in secretion of Serpine-1 GFP positive and CD63 RFP positive EVs, with some of them containing uorescent Aβ cargo. EVs were isolated from the cell culture media and then employed to differentiating NPCs for 24 h. Green uorescence signals (corresponding to EV-derived Serpine-1 GFP), red uorescent signals (indicating EV-CD63 RFP) and yellow Aβ HiLyte 647 uorescence (indicating EV-Aβ cargo) in the acceptor nonuorescent NPCs was assessed by confocal microscopy.
Representative images of NPC cultures exposed for 1 h to EVs derived from control, HIV plus/or Aβtreated HBMEC are illustrated in Fig. 4A, with a variety of vesicular and non-vesicular structures. Some EVs show red uorescence due to the presence of the EV marker CD63 RFP. EVs also exhibit yellow uorescence indicating Aβ transfer via EVs derived from Aβ-treated HBMEC. Some of the EVs with uorescent Aβ cargo show an overlapping red or green uorescence indicating colocalization with Serpine-1 and/or CD63 (Fig. 4A, arrow heads for individual-, arrows for overlapping uorescence). After 24 h EV exposure, NPC media was removed, and the acceptor NPCs were xed and imaged again. Figure 4B visualizes Serpine-1 GFP, CD63 RFP and Aβ HiLyte transfer to NPCs by EVs derived from HIV and/or Aβ HiLyte-exposed HBMEC. Some of the transferred uorescent Aβ appeared to be concentrated in large yellow aggregates (arrows), particularly in the HIV + Aβ group (Fig. 4B).
We also quanti ed HBMEC-derived EV-Serpine-1 cargo transfer to NPCs. For these experiments, nontransfected HBMEC were exposed to 100 nM non-uorescent Aβ and/or HIV-1 for 48 h. EVs were isolated from the cell culture media and used for the subsequent NPC exposure for 24 h, followed by PBS wash.
First, Serpine-1 levels in the NPC culture media were assessed by ELISA 1 h and 24 h after EV exposure (Fig. 4C). As illustrated, 1 h after EV exposure, Serpine-1 levels were signi cantly higher in the HIV-EV and HIV + Aβ-EV treated groups as compared to the Control-EV group. In addition, Serpine-1 levels in all EVtreated groups were signi cantly higher when compared to the No EV group, indicating that Serpine-1 in the NPC culture media, indeed, originated from the HBMEC-derived EVs and was not secreted by the recipient NPCs. This trend was maintained at 24 h after EVs exposure as well, indicating that most of the Serpine-1 was still present in the culture media and originated from the employed EVs (Fig. 4C, right graph).
Next, NPC culture media samples were used to assess Serpine-1 activity 1 h and 24 h after EVs treatment (Fig. 4D). Although Serpine-1 activity measurements were low, they showed a statistically signi cant increase in the HIV + Aβ-EV group when compared to the control-EV group, corresponding to Serpine-1 levels measured by ELISA. An increasing trend in the HIV-EV group was also observed although values in the No EV group were scattered (Fig. 4D).
In the following experiments, we explored whether Serpine-1 in the NPC media can inhibit tPA activity. As illustrated on Fig. 4E (left graph), tPA activity after 1 h was the highest in the control-EV group and showed a decreasing trend in the other groups. tPA activity in the Control-EV and Aβ-EV groups was signi cantly higher as compared to the No EV group (Fig. 4E). Overall, tPA activity change trends were the opposite of the Serpine-1 activity trends, con rming that the transferred EV-Serpine-1 can inhibit the transferred EV-tPA activity. Interestingly, 24 h after EV exposure, tPA activity levels were similar in all EVtreated groups and most of them were signi cantly higher as compared to the No EV group (Fig. 4E, right graph), verifying that tPA activity in NPCs originated mainly from the employed EV exposure and that NPCs have no or minimal endogenous tPA activity.
Serpine-1 levels and activity were also measured in the recipient NPCs. As shown on Fig. 4F, Serpine-1 levels were much lower (picogram range) in the recipient NPCs as compared to the media (nanogram range). These levels were signi cantly higher in the HIV + Aβ-EV treated group as compared to the Control-EV group and to the No EV group, again indicating that Serpine-1 in the NPCs originated from the HBMECderived EVs. This trend was similar at 1 and 24 h after EVs exposure (Fig. 4F). Finally, we measured Serpine-1 activity and tPA activity in the recipient NPCs 1 h and 24 h after EVs exposure; however, the levels were mostly very low or undetectable (data not shown). These results are consistent with Serpine-1 and tPA being secreted enzymes; therefore, they may primarily exert their effects on the recipient NPCs from the extracellular side.

Serpine-1 is involved in EV-mediated transfer of Aβ cargo to recipient NPCs
After establishing that HBMEC-derived EVs can transfer both Serpine-1 and Aβ, we explored if Serpine-1 can be involved in Aβ transfer and/or uptake by NPCs. Non-transfected HBMEC were exposed to 100 nM Aβ HiLyte 488 and/or HIV-1 for 48 h, followed by isolation of EVs, which were then employed for NPC exposure for 24 h. Representative images of NPCs exposed to uorescent EVs for 24 h visualize the transferred Aβ HiLyte (green uorescence) and the NPC mitochondria traced with Mitotracker (red uorescence) (Fig. 5A). To assess the involvement of Serpine-1 in this process, NPCs were cotreated with the Serpine-1 inhibitor PAI039 for 24 h. PAI039 was used at low concentration of 2 µM, which did not affect NPC viability (Fig. 5B).
EV-derived Aβ HiLyte uorescence was next quanti ed in the recipient NPCs using a plate reader and normalized to nuclear DRAQ5 uorescence (Fig. 5C). A signi cant uorescence increase was observed upon treatment with EVs derived from Aβ HiLyte-treated HBMEC and HIV-1 plus Aβ HiLyte-treated HBMEC as compared to treatment with EVs from control HBMEC. PAI039 affected the EV-Aβ HiLyte transfer only in NPCs exposed to EVs derived from HIV-1 plus Aβ HiLyte HBMEC. Speci cally, cotreatment with PAI039 signi cantly increased Aβ transfer in this group as compared to the HIV + Aβ-EV group without inhibitor (Fig. 5C). These results indicated that transfer of EV-derived Aβ cargo could be modulated by Serpine-1 only in speci c conditions, such as HIV-1 exposure.
When we incubated only the isolated EVs with the inhibitor, we have observed that PAI039 paradoxically increased EV-Serpine-1 activity in the HIV groups and caused an increasing trend in the non-HIV groups when compared to the respective groups without PAI039 (Supplementary Fig. 1). Although observed in a different experimental set-up, these unexpected effects may have contributed to the impact of PAI039 on EV-Aβ transfer to NPCs in the context of HIV-1. On the other hand, PAI039 did not change tPA activity in the isolated EVs ( Supplementary Fig. 1).

Serpine-1 impacts mitochondrial networks and functions in NPCs exposed to HBMEC-derived EVs
In the next series of experiments, we investigated the implications of Serpine-1 transfer to NPCs via EVs by evaluating the mitochondrial networks and functions. In support of this line of investigation, there is evidence that proper mitochondrial functions of NPCs are essential for correct neurogenesis [43].
To assess mitochondrial morphology changes, we employed Mitotracker Deep Red to stain mitochondrial networks in the NPCs treated with EVs isolated from control, Aβ, and/or HIV-exposed HBMEC (Fig. 6A). The experiments also involved treatment with PAI039. Morphological analysis was performed using mitochondrial network analysis (MiNA), an Image J plug-in, as previously published [37] to quantify the networks pre-processed and skeletonized by the software (skeletonized images on Fig. 6A and graphs on Fig. 6B). Mitochondrial footprint (the total area of the image with the Mitotracker uorescence signal, normalized to the number of nuclei), was signi cantly increased in the HIV + Aβ-EV group when compared to the Aβ-EV and control-EV groups. In addition, mitochondrial footprint was signi cantly higher in the HIV-EV + PAI039 group as compared to the control-EV + PAI039 group, indicating HIV-1-mediated impact ( Fig. 6B, upper left graph). The mean branch length was also signi cantly increased in the HIV + Aβ-EV group when compared to the control-EV group, and this effect was signi cantly blocked by the PAI039 exposure (Fig. 6B, upper right graph). The presence of long branches may indicate hyperfusion, leading to an elongated branched network. The summed or total branch length showed a similar change with a signi cant increase in the HIV + Aβ-EV group when compared to the control-EV or Aβ-EV only groups and PAI039 abolished this effect (Fig. 6B, lower left graph). Consistent with these results, mean network branches were also signi cantly decreased in the HIV + Aβ-EV + PAI039 group when compared to the HIV + Aβ-EV group (Fig. 6B, lower right graph). Taken together, these results indicate that in the HIV + Aβ-EV group the mitochondrial branches are longer; however, PAI039 can reverse these effects.
Next, we assessed whether mitochondrial network changes were accompanied by functional changes as determined using the Seahorse Mito Stress assay. For these experiments, HBMEC were treated with Aβ and/or HIV, followed by isolation of EVs as in Fig. 3. Then, human NPCs seeded on 24-well Seahorse plates (30,000/well) were differentiated for 3 days with exposure to these isolated EVs for the last 24 h. Selected NPC cultures were also cotreated with PAI039 (2 µM) and EVs for 24 h. Mitochondrial oxidative phosphorylation was measured by the oxygen consumption rate (OCR) and glycolysis by analyzing the extracellular acidi cation rate (ECAR) in real-time in live NPCs (Fig. 7). Data from the Seahorse were normalized to NPC protein concentration per well. Figure 7A depicts the Seahorse XF Mito Stress Test Pro le, while Figs. 7B and C represent OCR and ECAR graphs, respectively. Non mitochondrial oxygen consumption did not change in NPCs after exposure to any EV groups (Fig. 7D). Interestingly, basal respiration was signi cantly higher in the NPC cultures exposed to EVs derived from HIV plus Aβ-treated HBMEC and cotreated with PAI039 as compared to cultures treated with PAI039 and Aβ-EVs or with PAI039 alone (Fig. 7E). Similar changes were observed for maximal respiration, with the highest values being recorded for the NPC cultures exposed to EVs derived from HIV plus Aβ-treated HBMEC and cotreated with PAI039 (Fig. 7F). Strikingly, maximal respiration in this group was signi cantly higher as compared to cultures treated with HIV + Aβ-EVs, HIV-EVs + PAI039, Aβ-EVs + PAI039, or control-EVs + PAI039 (Fig. 7F). Proton leak was signi cantly increased in the HIV + Aβ-EV + PAI039 group as compared to the Aβ-EV + PAI039 group (Fig. 7G). PAI039 did not change signi cantly the ATP production in the NPCs as compared to the corresponding treatment groups without PAI039. However, ATP production was signi cantly higher in the HIV + Aβ-EV + PAI039 group as compared to the Aβ-EV + PAI039 or control-EV + PAI039 groups (Fig. 7H). On the other hand, PAI039 signi cantly decreased spare respiratory capacity in the Aβ-EVs + PAI039 treated NPCs as compared to the Aβ-EVs only group (Fig. 7I). Although shown under different experimental conditions, the unexpected EV-Serpine-1 activity increase evoked by PAI039 (Supplementary Fig. 1) may have contributed, at least partly, to these effects.
Overall, the obtained results indicate a substantial impact of Serpine-1 inhibition on bioenergetics of mitochondria. While signi cant increases in basal and maximal respiration, elevated proton leak and ATP production were all observed in the HIV + Aβ + PAI039-EV group, a decrease in spare respiratory capacity in this group suggests that Serpine-1 inhibition may impair the ability of cells to respond to stress or metabolic demand.

Serpine-1 affects synaptic protein expression in the developing NPCs
Literature reports on potential neurotoxicity of Serpine-1 are con icting. There are indicators that Serpine-1 can block the damage of neuronal networks in vitro by increasing postsynaptic density protein 95 (PSD95) and synaptophysin [44,45]. In addition, Serpine-1 was demonstrated to be neuroprotective against NMDA-induced neuronal death [46]. In contrast, Serpine-1 was reported to inhibit tPA-mediated neurite outgrowth in NPCs [47].
In order to get a better understanding of the involvement of Serpine-1 in synaptic protein expression in differentiating NPCs, we evaluated PSD95 and synaptophysin expression pattern in NPCs exposed to HBMEC-derived EVs in the presence or absence of PAI039. As illustrated on Fig. 8A and in Supplementary   Fig. 2, synaptophysin exhibited a ne punctate immunoreactivity pattern (green uorescence) in all NPC groups with strikingly stronger uorescence in segments of the developing NPC projections (arrow heads). PSD95 showed a similar punctate immunoreactivity pattern (red uorescence). In addition to this ne punctate pattern, PSD95 uorescence was frequently observed in vesicular structures both inside and outside of NPCs in all groups (arrows) as revealed by z-stacking confocal microscopy (Fig. 8A). This was rarely the case for synaptophysin. We may speculate that PSD95 could possibly be also released via EVs, as was shown before [48], from the developing NPCs to be delivered to the developing synapses.
Quanti cation of total synaptophysin and PSD95 immunoreactivity from confocal z-stack images showed that both synaptic protein levels signi cantly decreased in NPCs exposed to HIV-EVs as compared to the Control-EVs treated group. This effect was reversed by PAI039 only for synaptophysin, indicating that PAI039 differentially affected synaptic protein levels overall (Fig. 8B). In addition, total PSD95 level was signi cantly decreased in NPCs after Aβ-EV exposure and PAI039 did not block this effect (Fig. 8B, right graph).
Next, we quanti ed synaptic protein uorescence intensity in the NPC projections (Figs. 8C-E). Because these were developing NPCs and not mature neurons, their projections were short (Fig. 8C) and their synaptic protein immuno uorescence varied greatly from no signal to very high intensity signal even within the same neurite (Figs. 8A and 8D). Synaptophysin (green uorescence) and PSD95 (red uorescence) showed a similar punctate immunoreactivity pattern in the NPC projections, visible at high magni cation (Fig. 8D). Interestingly, there appeared to be little overlap between synaptophysin and PSD95 signals in neurites on these high magni cation images. Using an unbiased approach, we highlighted identical rectangular areas randomly on NPC projections visible on bright eld images as depicted on Fig. 8C. Then, we measured synaptophysin (green) and PSD95 (red) uorescence intensity from the same areas. Quanti cation of these results are shown on Fig. 8E. Synaptophysin levels signi cantly increased in the Control-EV + PAI039 group as compared to the Control-EV group and in the HIV-EV + PAI039 group when compared to the HIV-EV only group. The Control-EV + PAI039 group was also signi cantly higher than the Aβ-EV + PAI039 group, suggesting that Aβ-EV treatment may have blocked the PAI039 effect (Fig. 8E, left graph). Changes in PSD95 uorescence were partly similar, with PSD95 levels being signi cantly increased in the Control-EV + PAI039 group as compared to the Control-EV group. Interestingly, the Control-EV + PAI039 group was also signi cantly higher than the Aβ-EV + PAI039, HIV-EV + PAI039 and HIV + Aβ-EV + PAI039 groups, suggesting that Aβ and HIV may have blocked the PAI039 effect (Fig. 8E, right graph).

Discussion
EVs are recognized as important contributors to Aβ pathology [49][50][51][52][53][54][55], including elevated Aβ deposition in HIV-1 infection [20,22]. We have shown previously that EV-Aβ can be transferred to cells of the neurovascular unit, including neural progenitor cells (NPCs) [20]; however, the mechanisms of EVmediated Aβ pathology remain elusive. In an effort to gain more insight into this process, we applied proteomics analysis to better characterize the EV protein cargo in the context of HIV-1, and Serpine-1 was identi ed as a main connecting "hub" on several EV protein-protein interaction maps [23]. This nding is important because Serpine-1 was previously described as a key player in Aβ pathology [30,31] and was linked to HIV-1 infection as well [29,56]. However, the role of EV-associated Serpine-1 in the Aβ pathology is not well understood. Since Serpine-1 was linked to Aβ pathology, HIV-1 comorbidities, and was found in EVs, we hypothesized that brain endothelial EV-Serpine-1 can be involved in HIV-1 and/or Aβ-induced NPC alterations. To the best of our knowledge, there are no reports in the literature on this subject.
The main physiological role of Serpine-1 is inhibition of tPA leading to inhibition of plasmin, which places Serpine-1 as the critical regulator of brinolysis pathways [24]. However, this process also affects Aβ levels because plasmin can degrade both APP and Aβ. Indeed, elevated levels of Serpine-1 favor a more procoagulant state by decreasing tPA activity, which, in turn, hinders plasminogen conversion to active plasmin leading to diminished Aβ degradation [28]. In support of this claim, Serpine-1 involvement in Aβ deposition in AD has been documented [30,31]. For example, cerebral blood vessels were shown to be Serpine-1 positive in AD transgenic mice overexpressing Tau [57]. Knock-out of Serpine-1 gene or inhibition of Serpine-1 signi cantly reduced brain Aβ load in the APP/PS1 AD mouse model. Oral administration of TM5275, a small molecule inhibitor of Serpine-1, increased the activities of tPA, uPA and plasmin, leading to decreased Aβ levels in the hippocampus and cortex with improved learning and memory functions [31]. Moreover, Serpine-1 inhibition with PAI039 restored Aβ-induced decreased tPA activity and altered neurovascular coupling. These effects were associated with reduced perivascular amyloid deposition and improved cognition [58]. In related studies, tPA was proposed to protect against elevated Aβ levels by accelerating Aβ degradation and inhibition of Aβ-mediated neurodegeneration [28].
Overall, the literature data provide evidence that inhibition of Serpine-1 and restoration of tPA activity could be of substantial therapeutic value in AD.
Regarding a link to HIV infection, antiretroviral therapy (ART) was shown to affect plasma Serpine-1 levels, which appeared to be a marker for HIV-1 related comorbidities. For instance, HIV-1 infected patients on protease inhibitors had higher plasma Serpine-1 levels [56]. Moreover, high plasma levels of Serpine-1 were associated with high risk of myocardial infarction in HIV-1 infected people [29]. Elevated expression of Serpine-1 was proposed to be one of the mechanisms of HIV-1 protein Tat-induced in ammation in vascular cells [59], and synthetic Tat In line with these reports, we have con rmed that primary human brain endothelial cell-derived EVs (HBMEC-EVs) contained Serpine-1 [23] implicating the brain endothelium and the BBB as important contributors to the Serpine-1 pool. In experiments in which HBMEC were double-transfected with Serpine-1 GFP and CD63 RFP, we found that these cells concentrated and released Serpine-1 via EVs (Figs. 1 and 2).
Interestingly, the isolated Serpine-1-positive EVs were rarely positive for CD63 RFP (Fig. 2B), suggesting that different sets of EVs with different cargoes were released by similar pathways from the parent cells, and Serpine-1 may be released in a speci c set of EVs. Nevertheless, there is also a possibility that Serpine-1 GFP transfection e ciency was higher than that of CD63 RFP, although the amount of CD63 RFP plasmid DNA used was double of that of Serpine-1 GFP plasmid. We also observed that Serpine-1 GFP was associated with EVs of different sizes ( Fig. 2A), underscoring the importance of evaluating total EVs in the HIV-related Aβ pathology as opposed to just a particular size-range EVs. Another interesting observation was that almost all Serpine-1 positive EVs were DAPI positive, which demonstrates their nucleic acid cargo ( Fig. 2A). Strikingly, the number of CD63 RFP positive EVs was signi cantly lower in the HIV and HIV + Aβ groups as compared to the Aβ only group (Fig. 2B). Similar pattern was observed for the Serpine-1 GFP/CD63 RFP double positive EVs (Fig. 2B), suggesting that HIV-1 exposure of the parent cells alters the protein cargo of the released EVs.
In general, only low levels of Serpine-1 protein were described in the brain and our results are consistent with these observations as only low levels of Serpine-1 were detected in HBMEC. However, our important results indicate that Serpine-1 was concentrated in EVs, achieving ~ 30x higher levels when compared to the parent cells (Figs. 3A-3B). When comparing the treatment groups, EV-Serpine-1 levels were signi cantly higher in the HIV-1 and the HIV + Aβ groups than in controls when normalizing the data to cell culture volume (Fig. 3A). These results were consistent with the activity data (Fig. 3C). On the other hand, there were no differences between the groups when EV-Serpine-1 levels were normalized to EV protein content (Fig. 3A) because exposure to HIV-1 increases the overall EV number as published before [20]. Overall, these observations suggest that one of the main Serpine-1 pools in the brain may be found in endothelial-EVs, pointing to a critical role of the BBB in Serpine-1 related brain pathologies such as AD or stroke or even systemic pathologies as BBB-EVs may also be released into the peripheral circulation.
Moreover, higher Serpine-1 levels in the HIV groups implicate EVs to creating a pro-coagulant environment in the vicinity of the BBB of HIV-1-infected brains. These observations are in line with the report that HIV-1 infection was associated with a more pro-coagulant state associated with high Serpine-1 levels in the plasma possibly increasing the risk of myocardial infarction [29]. These ndings are also consistent with tPA activity being signi cantly reduced in the EV groups originating from HIV-1 and/or Aβ-treated HBMEC as compared to control EVs (Fig. 3D). Indeed, a decrease in tPA activity was observed to be consistently associated with elevated amyloid deposition in the brain [66].
One of the main biological functions of EVs is intercellular communication, which is executed by transferring cargo between different cells and cell types. We have shown before that endothelial-derived EVs can transfer Aβ to NPCs affecting their neurogenesis [22]. Knowing that Serpine-1 levels are concentrated in EVs, we next evaluated if EVs can serve as carriers to deliver Serpine-1 from endothelial cells to NPCs and what are the outcomes of this process. The rationale for these experiments is the fact that a large pool of NPCs is in the neurogenic niches of the perivascular space in direct proximity to the brain endothelium [42]. We con rmed that Serpine-1 could be transferred to NPCs via EVs (Fig. 4).
Moreover, our results indicated that Serpine-1 is transferred together with Aβ cargo, potentially impacting the Aβ fate in the acceptor NPCs. Consistent with previously reported results, this process appeared to be enhanced in EVs derived from HIV-1-exposed HBMEC [20]. Nevertheless, Serpine-1 levels in acceptor NPCs were very low, suggesting that secreted EV-Serpine-1 acts on NPCs mostly extracellularly. This mode of action is consistent with the biological impact of Serpine-1 and tPA as both proteins are secreted and act in the extracellular environment.
To evaluate the interactions between Serpine-1 and Aβ, we examined whether Serpine-1 inhibition can affect EV-Aβ transfer to NPCs. Consistent with literature data [67], NPCs appeared to be sensitive to PAI039 toxicity, as 10 µM caused substantial toxicity (Fig. 5B) and 20 µM caused massive cell death; therefore, PAI039 was used at 2 µM. Inhibition of Serpine-1 had an unexpected effect on Aβ transfer to NPCs, enhancing this process in the HIV + Aβ-EV + PAI039 group as compared to the HIV + Aβ-EV group (Fig. 5C). The inhibitor did not affect Aβ transfer in other groups, suggesting a speci c impact of Serpine-1 on the transfer of EV-derived Aβ cargo in the context of HIV-1. Although this was a unique complex system, where EVs interacted with NPCs, separate observations that PAI039 exposure with EVs alone increased Serpine-1 activity in the HIV groups ( Supplementary Fig. 1) suggest that this phenomenon may have also affected EV-Aβ transfer.
The role of Serpine-1 in neuronal dysfunction is controversial with con icting reports in the literature. It was observed that Serpine-1 might be neuroprotective against Aβ-induced neurotoxicity, preserving neuronal networks, and promoting synaptogenesis by increasing PSD95 and synaptophysin [44,45]. Similarly, Serpine-1 was demonstrated to be neuroprotective against NMDA-induced neuronal death [46].
On the other hand, Serpine-1 can inhibit neuroprotective impact of tPA, which was shown to control neurite outgrowth in cortical neurons after stroke or in NPCs [47,68]. This is an important observation, because both Aβ and HIV pathologies are linked to increased incidents of strokes and administration of tPA is an approved intervention to restore blood ow to brain regions affected by a stroke. By blocking these bene cial effects, Serpine-1 can exert undesirable neurotoxic impact. To address these problems, we evaluated the effects of EVs carrying Serpine-1 and/or Aβ on the NPC mitochondrial networks, bioenergetics, and synaptic integrity. The results clearly demonstrated altered mitochondrial morphology in NPCs exposed to HBMEC-derived EVs in the HIV + Aβ-EV treated group, with PAI039 reversing the majority of these alterations (Fig. 6B). The characteristic of these changes related to long mitochondria branches and an increase in total branch length may point to mitochondrial hyperfusion. In support of a role of EVs and HIV-1 in these mitochondrial alterations, EVs from latent HIV-infected T cells were shown to enhance mitochondrial superoxide production, reduce mitochondrial membrane potential, and induce mitochondrial hyperfusion in primary human brain microvascular endothelial cells [69]. In addition, the obtained results were consistent with the observations that Aβ may cause NPC damage with mitochondrial alterations, which, in turn, may affect their functions [43]. Regarding NPC mitochondrial bioenergetics, inhibition of Serpine-1 appeared to impair the cells' ability to respond to stress or metabolic demand after Aβ-EV exposure as demonstrated by a decrease in spare respiratory capacity in this treatment group (Fig. 7F).
In order to assess the impact of Aβ and/or HIV-1 EVs on synaptic integrity, we evaluated the levels of synaptophysin and PSD95 in differentiating NPCs. Synaptophysin expression served as a marker of presynaptic plasticity and synaptogenesis [70]. Loss of synaptophysin was found in AD [71] and learning and memory de cits have been demonstrated in synaptophysin knockout mice [72]. PSD95 is a postsynaptic marker of synaptic integrity and its decrease can also lead to learning and memory impairment [73,74]. In contrast, up-regulation of PSD-95 was shown to improve memory [75], underscoring the importance of PSD95 in these key brain functions. We detected that even in early developmental stage (namely, three days of differentiation), NPCs expressed a ne punctate immunoreactivity for both synaptophysin and PSD95 in the developing neurite segment (Fig. 8D). In addition, PSD95-positive, but not synaptophysin-positive, uorescence was often demonstrated in vesicular structures both intra-and extracellularly on z-stacking confocal microscopy images (Fig. 8A), suggesting that it could be secreted via EVs from the developing NPCs. Exposure to EVs derived from HIVexposed HBMEC decreased both synaptophysin and PSD95 in NPCs, which is consistent with neurotoxicity and aberrant neurogenesis in HIV infected brains [76,77]. In addition, loss of synaptophysin was observed in HIV-1 infection in humanized mice [78] and exposure to HIV-1 proteins, like Tat and gp120, markedly decreased PSD95 in hippocampal neurons [79][80][81]. The levels of synaptophysin and PSD95 were also decreased in mice following exposure to HIV-1 Tat and methamphetamine [82].
Therefore, it was important that inhibition of Serpine-1 protected against HIV-EV-induced alterations in synaptophysin (Fig. 8B) and increased both synaptic proteins in the projections (Fig. 8E) of differentiating NPCs. These observations are consistent with the reports advocating for a bene cial effect of Serpine-1 inhibition on synaptic protein expression and neurite development, although EV-Serpine-1 activity increase evoked by PAI039 ( Supplementary Fig. 1) may have also contributed to these effects.
Overall, these data may be relevant in the context of HIV-1 associated neurocognitive impairments.
Conclusions. The results of the present study indicate that brain endothelial EVs contain active Serpine-1 cargo, which can be delivered to the recipient NPCs along with Aβ. These ndings represent a novel concept that endothelial-derived EVs constitute a major Serpine-1 pool in the brain, which can create a pro-coagulant environment at the BBB and lead to mitochondrial and synaptic alterations in NPCs. These processes may further contribute to HIV-1 associated neurocognitive disorders (HAND), especially in older brains, which are characterized by elevated Aβ depositions.

Availability of data and materials
All source data supporting the ndings of this manuscript are available from the corresponding authors upon request.

Competing interests
The authors declare that they have no competing interests.   Serpine-1 is released in EVs from control, Aβ, and/or HIV-1 exposed brain endothelial cells (A) Visualization by confocal microscopy of Serpine-1 GFP (green), CD63 RFP (red), and Aβ (1-40) HiLyte (yellow) (arrow heads) in EVs isolated from media of transfected and treated HBMEC as in Figure 1.

Figure 5
Involvement of Serpine-1 in the transfer of Aβ cargo from HBMEC-derived EVs to recipient NPCs. HBMEC were exposed to HIV (30 ng/ml) and/or 100 nM Aβ (1-40) HiLyte 488 for 48 h, followed by isolation of Figure 6 Mitochondrial network analysis (MiNA) of mitochondrial morphology in EV-exposed NPCs. Nontransfected HBMEC were treated as in Figure 3. EVs were isolated from the culture media and employed for NPC treatment for 24 h. Selected NPCs were cotreated with PAI039 (2 μM) and EVs for 24 h. (A) Confocal images of NPCs stained with Mitotracker Deep Red (red) for tracking the mitochondria. MiNA plugin on ImageJ was used to skeletonize the mitochondria. Scale bar: 10 µm. (B) Quanti cation of the mitochondrial footprint, mean branch length, total branch length and mean network branches. Values are mean ± SEM, n=8-10. One-, two-and three-way ANOVA with Tukey's multiple comparisons test. *Statistically signi cant p<0.05, **p<0.01.

Figure 7
Mitochondrial functions in recipient NPCs after exposure to HBMEC-derived EVs. Non-transfected HBMEC were treated as in Figure 3. EVs were isolated from the culture media and employed for NPC treatment for 24 h in the presence or absence of the Serpine-1 inhibitor PAI039 (2 μM).