N2a neuronal cells were purchased from the Chinese Academy of Sciences Cell Bank, and cultured in DMEM (Gibco) containing 10% FBS (Gibco) and 100 U/ml penicillin and streptomycin (ThermoFisher).

APP/PS1 mice and C57 mice (purchased from Shanghai Model Organisms Center) are housed and bred in the Comparative Medicine animal facilities of Tongji University School of Medicine (TUSM). All procedures were conducted according to protocols approved by the Institutional Animal Care and Use Committee of TUSM. Mouse genotype was validated by PCR.

The design of the present study was approved by the ethics committee of Tongji Hospital affiliated to Tongji University School of Medicine (Shanghai, China), and the written informed consents were obtained from all participants. A total of three AD patients (males, mean age 67.3 ± 4.7) and three healthy donors (males, mean age 70.0 ± 7.1) were selected for this study. Peripheral blood was collected from each patient after fasting for 12 h. The serum was separated by centrifugation at 3,000 g for 10 min in room temperature, followed by centrifugation at 12,000 g for 5 min at 4°C.

Exosomes were isolated from the culture medium of N2a cells as previously described (Ma et al., 2018). Briefly, 3.5 × 10⁶ N2a cells were culture in 10 cm dish. 24 h before collecting supernatants, the culture medium was replaced with FBS free medium containing DMEM (Gibco) and 100 U/ml penicillin and streptomycin (ThermoFisher) only. After conditioned media collection, exosomes were isolated by sequential centrifugation: conditioned media was first centrifuged at 300 g for 10 min to eliminate flowing cells, at 3,000 g for 20 min to eliminate cellular debris, at 10,000 g for 30 min to eliminate intracellular organelles and then at 100,000 g for 2 h to precipitate exosomes. All steps of centrifugation were handled at 4°C. For brain exosome isolation, fresh mouse brains were dissected and treated with 20 units/ml papain (Worthington) in Hibernate E solution (5 ml/brain) for 30 min at 37°C. The brain tissue was gently homogenized in 10 ml cold Hibernate E solution. The brain homogenate was sequentially filtered through a 40 μm mesh filter (BD Biosciences) and a 0.2 μm syringe filter (Thermo Scientific). Exosomes in the filtrate were collected through gradient centrifugation as described above. Exosomes in the serum were isolated using ExoQuick Exosome Precipitation kit (System Biosciences). Exosomes were resuspended in PBS and stored in −80°C until required.

Nanoparticle Tracking Analysis (NTA) was carried out as previously described (Ma et al., 2019). Briefly, 1 ml of exosome suspension was used for NanoSight analysis. NTA was assessed on NanoSight NS300 system (Malvern Instruments, UK) with a sCMOS camera. The conditions of the measurements were set at 25°C, 1 cP viscosity, 25 s per capture frame, and 60 s measurement time. Three individual measurements were applied for measuring the sizes of exosomes.

Transmission Electron Microscop (TEM) was carried out as previously described (Ma et al., 2019). Briefly, purified exosomes were negatively stained and then spread on the copper grids. The droplets of exosomes were removed with filter paper and air-dried at room temperature. Images were obtained using transmission electron microscopy (JEM-1230, JEOL Ltd.).

The plasmids for the ectopic expression of human APP full-length, APP CDS, and APP 3′UTR sequences were constructed by Generay (Generay Biotech Co., Ltd., Shanghai). Plasmid backbone pcDNA3.1 was purchased from Generay. APP transcript sequences were obtained from UCSC Genome Browser (http://genome.ucsc.edu/) and inserted into pcDNA3.1 using restriction enzymes NheI and NotI for genomic digestion. The mimics control, miR-185-5p mimics, inhibitor control, and miR-185-5p inhibitor were purchased from GenePharma (GenePharma Co., Ltd., Shanghai). Transfection was performed using the Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s instruction.

The messenger RNA (mRNA) and miRNA were isolated from cell samples using RNeasy mini kit (Qiagen) according to the manufacturer’s instructions. Genomic DNA was removed and cDNA was synthesized using DNase I digestion kit (Qiagen) and miScript II RT kit (Qiagen), respectively. Transcripts were amplified using specific primer sets (Supplementary Table S1) and SYBR green PCR kit (Qiagen) with the ABI7500 (Applied Biosystems). Reactions were run in triplicates for each sample and no-template blanks were used as negative controls. Values were normalized to the Gapdh (for mRNA) and U6 snRNA (for miRNA).

Western blotting was performed as previously described (Gao et al., 2019). Exosomes or cells were lysed in RIPA lysis and extraction buffer (ThermoFisher) containing a protease inhibitor cocktail (Sigma). Protein concentrations were determined with a BCA Protein Assay Kit (Pierce). Proteins (20–30 mg) from exosome and cell lysates were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and electrophoretic transferred to polyvinylidene fluoride membranes (Millipore and Bio-Rad). Membranes were incubated with primary antibodies for APP (rabbit, Abcam, 1:1,000), CD9 (rabbit, Abcam, 1:2000), Flottlin2 (mouse, BD Biosciences; 1:5,000), overnight at 4°C followed by a secondary anti-rabbit or anti-mouse antibody (Cell Signaling Technologies, 1:10,000) incubation. Antigen-antibody complexes were visualized by Pierce ECL Western Blotting Substrate (ThermoFisher). For data quantification, films were scanned with a CanonScan 9950F scanner; the acquired images were analyzed using ImageJ program.

Total RNA was extracted from N2a cell-derived exosomes and 3 μg total RNA per sample was used as input material for the small RNA library. Sequencing libraries were generated using NEBNext® Multiplex Small RNA Library Prep Set for Illumina® (NEB). The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq SR Cluster Kit v3-cBot-HS (Illumia). After cluster generation, the library preparations were sequenced on an Illumina Hiseq 2,500/2000 platform and 50 bp single-end reads were generated. Raw data (raw reads) of fastq format were firstly processed through custom perl and python scripts for quality control. The small RNA tags were mapped to reference sequence by Bowtie without mismatch to analyze their expression and distribution on the reference. Mapped small RNA tags were used to looking for known miRNA. miRBase 20.0 was used as reference for known miRNA, miRDeep2 and sRNA-tools-cli were used to obtain novel miRNAs and draw the secondary structures, respectively. miRNA expression levels were estimated by TPM (transcript per million) and differentially expressed miRNAs between two groups were determined (p ≤ 0.05). The targets of differentially expressed miRNAs were predicted using TargetScan and miRanda databases according to the scoring criteria of each software. The target genes whose context score percentage is less than 50 or max energy is greater than −10 were removed in targetscan algorithm and Miranda algorithm, respectively. The intersection of predicted targets of both databases was taken as the final target genes of differential expressed miRNAs. Predicted targets of differentially expressed miRNAs were mapped to Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways analysis. GO and KEGG enrichment analysis was performed using The Database for Annotation, Visualization and Integrated Discovery (DAVID) (http://david.ncifcrf.gov/).

Dual-luciferase reporter assay was performed as previously described (Xia et al., 2019a). The APP wild type 3′UTR (APP WT 3′UTR) and miR-185-5p binding site-mutated 3′UTR (APP MU 3′UTR) were synthesized by Genewiz (Genewiz, Suzhou, China) and cloned into the PmeI and SacI site of the pmirGLO vector (Promega, Beijing, China), downstream of the firefly luciferase gene. For the luciferase assay, 3 × 10⁴ 293T cells were cultured in 96-well plates with DMEM (Gibco) containing 10% FBS (Gibco) and 100 U/ml penicillin and streptomycin (ThermoFisher). Cells were co-transfected with miR-185-5p mimics and either APP WT 3′UTR or APP MU 3′UTR dual-luciferase reporter vector after reaching ∼70% confluency. Serum-free Opti-MEM (ThermoFisher) was used to prepare the transfection solution, and Lipofectamine 2000 reagent (Invitrogen) was used for transfection. 24 h post transfection, Dual-Luciferase® Reporter Assay System (Promega) was used to determine the luciferase activities on SpectraMax M5 microplate readers (Molecular Devices). The activity of Renilla luciferase was used to normalize that of firefly luciferase.

All results are the means of at least three independent experiments ± s. d. Data from two groups were evaluated statistically by two-tailed, paired or unpaired student t test. Data from multiple groups were evaluated statistically by one-way ANOVA followed by Tukey’s post hoc test. Significance was considered when p < 0.05.

To investigate the effects of exosomes on APP expression, we first isolated exosomes from cortical and hippocampal tissues of 7-month-old APP/PS1 mice (AD-EXO) and control C57 mice (Ctl-EXO). The purity of exosomes was examined by NTA and western blotting. NTA results suggested that the sizes of AD-EXO and Ctl-EXO were between 100 and 200 nm, matching with the typical size distribution of exosomes (Figure 1A). The morphology of exosomes was visualized via TEM, which showed distinct cup-shaped structure of exosomes (Figure 1B). Western blotting detected strong expression of exosome positive markers, Flotillin2, Flotillin1, CD9, TSG101, HSP70, and CD9, in both AD-EXO and Ctl-EXO (Figure 1C). Moreover, the negative markers of exosomes, Apoa1, Apob, and Calnexin, were absent in both AD-EXO and Ctl-EXO, further confirming the purity of exosomes (Figure 1D). To determine the effects of exosomes on APP expression, we co-cultured N2a cells with either 20 μg/ml AD-EXO or Ctl-EXO for 2 days. RT-qPCR analysis showed higher levels of APP transcripts in AD-EXO-treated group, compared with Ctl-EXO group (Figure 1E). Western blotting further revealed a significant increase of the expression levels of APP proteins in AD-EXO-treated cells versus Ctl-EXO group (Figure 1F). Thus, our results suggest an important role of exosomes from AD mouse brains in APP dysregulation in neuronal cells.

To further confirm the exosome-mediated APP dysregulation and investigate the underlying mechanisms, we established an in vitro AD model by ectopically expressing full-length APP mRNAs in N2a cells to mimic the pathological conditions of AD in vivo. The overexpression of APP transcripts and APP proteins in N2a was validated by RT-qPCR (Figure 2A) and western blotting (Figure 2B), respectively. APP-overexpressed (APP OE) cell-derived exosomes (APP-EXO) or empty vector-transfected (control) cell-derived exosomes (EXO) were then collected, visualized, and validated via NTA, TEM, and western blotting assays (Figure 2C–F). To determine the effects of APP-EXO on APP expression, we co-cultured N2a cells with either 20 μg/ml APP-EXO or EXO for 2 days. Both RT-qPCR and western blotting results revealed a significant increase of the expression levels of APP transcripts (Figure 2G) and APP proteins (Figure 2H) in APP-EXO-treated cells versus EXO group, indicating the positive effects of APP-EXO on APP expression in neuronal cells.

We next investigated the possible mechanisms of exosome-induced APP dysregulation. The simplest and most obvious one is that exosomes may directly transfer APP transcripts and APP proteins to target cells (Perez-Gonzalez et al., 2012). To test this premise, we first compared the expression levels of APP transcripts and APP proteins in APP-EXO and EXO. Both RT-qPCR and western blotting results indicated that APP-EXO contains higher levels of APP transcripts (Figure 3A) and APP proteins (Figure 3B), compared with EXO. To further confirm our observations, we then transfected N2a cells with plasmid that overexpresses the coding DNA sequence (CDS) region of APP transcript. The empty (control) plasmid and plasmid containing the 3′ UTR of APP transcript were used as controls. Both RT-qPCR and western blotting results suggested a significant up-regulation of APP CDS and APP proteins in APP CDS-overexpressed (CDS OE) N2a cells, compared with APP 3′UTR-overexpressed (3′UTR OE) and control cells (Figures 3C,E). Moreover, APP 3′UTR levels were elevated only in 3′UTR OE group versus CDS OE and control groups (Figure 3D). Our observations suggested that CDS OE has similar promotional effects to APP OE on the expression of APP transcripts and proteins in N2a cells. Based on the results, we proposed that exosomes derived from CDS OE cells (CDS-EXO) should have also similar promotional effects to APP-EXO on the expression of APP transcripts and proteins in recipient cells if exosomes regulated APP expression through the direct transferring of APP gene products among cells.

We then collected CDS-EXO and exosomes derived from 3′UTR OE cells (UTR-EXO). The sorting of APP gene products into exosomes was confirmed by RT-qPCR and western blotting. The results showed that APP CDS expression levels were significantly higher in CDS-EXO, but not in UTR-EXO, compared with EXO (Figure 3F). APP 3′UTR levels were significantly higher in UTR-EXO, compared with both CDS-EXO and EXO (Figure 3G). Moreover, APP protein levels were significantly elevated in CDS-EXO, but not in UTR-EXO, compared with EXO, confirming the enrichment of APP gene products in CDS-EXO (Figure 3H). Afterwards, we co-cultured N2a cells with either CDS-EXO, UTR-EXO, or EXO for 2 days. Surprisingly, both RT-qPCR and western blotting results showed that CDS-EXO failed to elevate the expression levels of APP transcripts and APP proteins, respectively (Figures 3I,J). In contrast, UTR-EXO significantly increased the expression of APP transcripts and APP proteins in co-cultured cells (Figures 3I,J). Therefore, our results suggested that APP-EXO regulated APP expression in recipient cells independent to the direct transferring of APP gene products.

To understand the mechanisms of UTR-EXO-mediated APP regulation, we first carried out small RNA sequencing to determine the miRNA signatures of APP-EXO and EXO, given the importance of mRNA 3′UTR-miRNA interaction in mediating the content sorting and functions of exosomes (Squadrito et al., 2014). 796 miRNAs were detected in either APP-EXO or EXO (Supplementary Table S2). 435 and 56 miRNAs were identified that were significantly up-regulated and down-regulated, respectively, in APP-EXO versus EXO (Figure 4A, Supplementary Table S3). The heatmap that represented the top 60 differentially expressed miRNAs also demonstrated the distinct profiles of exosomal miRNAs between two exosome populations (Figure 4B). GO analysis on biological process revealed that the predicted target genes of differentially expressed miRNAs were strongly associated with response to stimulus, cell differentiation, protein phosphorylation, and apoptotic process that might play roles in the pathogenesis of AD (Figure 4C). More importantly, KEGG analysis of the predicted target genes identified synaptic vesicle cycle, p53 pathway, Pyruvate pathway, glycolysis, and glycosphingolipid biosynthesis that were linked to AD (Figure 4D). Taken together, our study demonstrated distinct miRNA signatures in exosomes derived from N2a cells with and without APP dysregulation, implying an involvement of exosomal miRNAs in the regulation of AD-related pathogenic networks.

To further identify the key miRNAs that mediate the effects of APP-EXO on APP expression in recipient cells, we identified the overlapping miRNAs (miR-185-5p, miR-31-5p, miR-144-3p, miR-369-3p, and miR-15b-5p) between the down-regulated miRNAs in APP-EXO and miRNAs that conservatively bind to APP 3′UTR (Figure 5A). Among these miRNAs in EXOs, miR-185-5p had the largest readcounts (Figure 5B). Furthermore, miR-31a-5p and miR-185-5p were the two with the largest fold change (Figure 5C). Hence, miR-185-5p was selected for following studies. The small RNA sequencing results were corroborated via RT-qPCR which demonstrated a significant decrease in the levels of miR-185-5p in APP-EXO versus EXO (Figure 5D). To make sure that our observations in vitro match with the situation in vivo, we examined the exosomal miR-185-5p levels in AD mouse brains via RT-qPCR and found a significant decline of miR-185-5p expression levels in AD-EXO compared with Ctl-EXO (Figure 5E).

Next, we carried out dual-luciferase assay to validate the direct binding of miR-185-5p with APP 3′UTR (Figure 5F). Co-transfection of miR-185-5p mimics and dual-luciferase reporter constructs containing the wild-type APP 3′UTR, but not miR-185-5p target site-mutated APP 3′UTR, significantly decreased the firefly activity in N2a cells, normalized by the Rellina activity, suggesting that miR-185-5p directly targets APP transcripts. We further tested that whether or not the sorting miR-185-5p in exosomes is regulated by APP 3′UTR. RT-qPCR results showed that miR-185-5p levels was significantly reduced in UTR-EXO, but not in CDS-EXO, compared with EXO, indicating APP 3′UTR but not CDS as the key element in regulating the sorting of miR-185-5p into exosomes (Figure 5G). Taken together, our study identified miR-185-5p as an EXO-enriched miRNA whose loading into exosomes is greatly influenced by endogenous APP 3′UTR.

To determine whether miR-185-5p mediates the effects of APP-EXO, we first examined the roles of miR-185-5p in regulating APP expression in neuronal cells via miR-185-5p loss-of-function (LOF) and gain-of-function (GOF) approaches. For miR-185-5p LOF, we transfected N2a cells with either miR-185-5p inhibitor (=LOF group) or inhibitor control for 2 days. RT-qPCR analysis showed a significant decrease in the levels of miR-185-5p in miR-185-5p LOF group, validating the efficiency of transfection (Figure 6A). A significant increase of APP transcript levels was observed in LOF group, compared with controls (Figure 6B). RT-qPCR results were confirmed by western blotting, which showed a significant increase of APP protein levels in the cell lysate of miR-185-5p LOF group versus that of control group (Figure 6C). For miR-185-5p GOF, N2a cells were transfected with either miR-185-5p mimics (=GOF group) or mimics control for 2 days. The efficiency of transfection is validated by RT-qPCR (Figure 6D). Moreover, both RT-qPCR and western blotting analyses demonstrated a significant decline of the expression levels of APP transcripts (Figure 6E) and APP protein (Figure 6F), respectively, in miR-185-5p GOF group, compared with control group. Thus, our results suggested that miR-185-5p negatively regulated APP expression in neuronal cells in vitro.

Afterwards, we transfected N2a cells with miR-185-5p inhibitor to knockdown (KD) intracellular miR-185-5p and collected exosomes from the conditioned medium of miR-185-5p KD cells (185 KD-EXO). The KD of exosomal miR-185-5p was validated by RT-qPCR (Figure 6G). N2a cells were co-cultured with either EXO control or 185 KD-EXO for 2 days. RT-qPCR results showed a significant increase of APP transcripts in 185 KD-EXO treated cells versus controls (Figure 6H). The RT-qPCR data were further confirmed by western blotting (Figure 6I), indicating miR-185-5p as an essential content of exosomes for inhibiting APP expression in recipient cells.

Since the small RNA sequencing analysis revealed multiple miRNAs whose abundance in exosomes were regulated by APP 3′UTR, we next addressed that whether exosomal miR-185-5p mediates the promoting effects of UTR-EXO on APP expression. Except for EXO and UTR-EXO, we also collected exosomes from the conditioned medium of N2a cells co-transfected with APP 3′UTR and miR-185-5p mimics (UTR&185-EXO). RT-qPCR results revealed that the APP 3′UTR overexpression-induced decline of the levels of exosomal miR-185-5p was abrogated by miR-185-5p GOF in exosome-secreting cells (Figure 6J). N2a cells were then co-cultured with either EXO, UTR-EXO, or UTR&185-EXO for 2 days. RT-qPCR analysis found no significant difference of APP transcripts in UTR&185-EXO group, compared to control group, which indicated that the UTR-EXO-induced excessive expression of APP transcripts in recipient cells was rescued by the overexpression of exosomal miR-185-5p (Figure 6K). Western blotting results also displayed a similar APP protein expression pattern to that of RT-qPCR (Figure 6L). In conclusion, our findings suggest miR-185-5p as an important exosomal cargo in modulating APP expression in recipient cells, which may contribute to the dysregulation of APP metabolism in AD.

As our in vitro observations indicated exosomal miR-185-5p as a key regulator of APP expression that may be involved in AD pathogenesis, we examined the potential translational value of exosomal miR-185-5p in AD diagnosis. We first compared the exosomal miR-185-5p levels in the sera collected from three AD patients and three age-/gender-matched healthy donors. RT-qPCR results suggested that serum exosomal miR-185-5p was significantly down-regulated in AD group versus controls (Figure 7A). We also collected sera from three AD mice and three age-/sex-matched C57 mice. Similarly, RT-qPCR results showed that the expression levels of serum exosomal miR-185-5p were significantly lower in AD mice, compared with controls (Figure 7B). Hence, our results implied a possibility of utilizing serum exosomal miR-185-5p in AD diagnosis, which requires extensive investigations including the introduction of large AD cohort and complicated statistical analyses.