MicroRNA-153 impairs presynaptic plasticity by blocking vesicle release following chronic brain hypoperfusion

Background Chronic brain hypoperfusion (CBH) is closely related to Alzheimer’s disease (AD) and vascular dementia (VaD). Meanwhile, synaptic pathology plays a prominent role in the initial stage of AD and VaD. However, whether and how CBH impairs presynaptic plasticity is currently unclear. Methods In the present study, we performed a battery of techniques, including primary neuronal culture, patch clamp, stereotaxic injection of the lentiviral vectors, morris water maze (MWM), dual luciferase reporter assay, FM1–43 fluorescence dye evaluation, qRT-PCR and western blot, to investigate the regulatory effect of miR-153 on hippocampal synaptic vesicle release both in vivo and in vitro. The CBH rat model was generated by bilateral common carotid artery ligation (2VO). Results Compared to sham rats, 2VO rats presented decreased field excitatory postsynaptic potential (fEPSP) amplitude and increased paired-pulse ratios (PPRs) in the CA3-CA1 pathway, as well as significantly decreased expression of multiple vesicle fusion-related proteins, including SNAP-25, VAMP-2, syntaxin-1A and synaptotagmin-1, in the hippocampi. The levels of microRNA-153 (miR-153) were upregulated in the hippocampi of rats following 2VO surgery, and in the plasma of dementia patients. The expression of the vesicle fusion-related proteins affected by 2VO was inhibited by miR-153, elevated by miR-153 inhibition, and unchanged by binding-site mutation or miR masks. FM1–43 fluorescence images showed that miR-153 blunted vesicle exocytosis, but this effect was prevented by either 2′-O-methyl antisense oligoribonucleotides to miR-153 (AMO-153) and miR-masking of the miR-153 binding site in the 3′ untranslated region (3’UTR) of the Snap25, Vamp2, Stx1a and Syt1 genes. Overexpression of miR-153 by lentiviral vector-mediated miR-153 mimics (lenti-pre-miR-153) decreased the fEPSP amplitude and elevated the PPR in the rat hippocampus, whereas overexpression of the antisense molecule (lenti-AMO-153) reversed these changes triggered by 2VO. Furthermore, lenti-AMO-153 attenuated the cognitive decline of 2VO rats. Conclusions Overexpression of miR-153 controls CBH-induced presynaptic vesicle release impairment by posttranscriptionally regulating the expression of four vesicle release-related proteins by targeting the 3’UTRs of the Stx1a, Snap25, Vamp2 and Syt1 genes. These findings identify a novel mechanism of presynaptic plasticity impairment during CBH, which may be a new drug target for prevention or treatment of AD and VaD. Video Abstract Graphical abstract


(Continued from previous page)
Conclusions: Overexpression of miR-153 controls CBH-induced presynaptic vesicle release impairment by posttranscriptionally regulating the expression of four vesicle release-related proteins by targeting the 3'UTRs of the Stx1a, Snap25, Vamp2 and Syt1 genes. These findings identify a novel mechanism of presynaptic plasticity impairment during CBH, which may be a new drug target for prevention or treatment of AD and VaD.

Background
Recently, chronic brain hypoperfusion (CBH) has been considered serving as a key predictor of the conversion of mild cognitive impairment (MCI) to Alzheimer's disease (AD) and vascular dementia (VaD) [1]. Previous studies have documented that CBH can produce cognitive decline by affecting many pathological processes including white matter attenuation, neuronal death, neuroinflammation, oxidative stress and even the formation of Aβ and the hyperphosphorylation of Tau [1][2][3][4][5][6]. Synaptic pathology plays a prominent role in the initial stage of AD and VaD [7][8][9]. CBH also attenuates synaptic plasticity by changing the expression of several proteins and remodeling synaptic dendrites and spines [10,11], however, whether and how CBH impairs presynaptic plasticity is currently unclear.
Presynaptic plasticity in chemical synapses is based on neurotransmitter release through synaptic vesicle exocytosis, which relies on highly conserved protein machinery. Neurotransmitter release is well understood to be dependent on the fusion of presynaptic vesicles with presynaptic membranes. This process is controlled by the interaction between the Ca 2+ sensor protein synaptotagmin-1 (Syt1) and the soluble N-ethylmaleimide-sensitive factor attachment receptor proteins complex (SNAREs), which comprises vesicle associated membrane protein 2 (VAMP-2), synaptosomal-associated protein 25 (SNAP-25) and syntaxin-1 [12,13]. Other important proteins that are required for exocytosis, including Munc-18, Munc-13 and complexins, have been found to collaborate with SNAREs to enhance the fusion probability between vesicles and presynaptic membranes [14,15]. Importantly, the expression of SNAP-25, VAMP-2, syntaxin-1 and Syt1 has been reported to be lower in postmortem tissues from AD and VaD patients than in those from healthy controls [16][17][18][19]. Whether CBH directly affects presynaptic plasticity by interrupting this highly conserved and elaborate protein machinery is unclear.
Previous studies have reported that microRNAs (miR-NAs) could regulate presynaptic plasticity by targeting various vesicle-related proteins. For example, miR-137 impaired vesicle release in a mouse model of schizophrenia [20]; miR-34c was reported mediate amyloid-beta (Aβ)-induced synaptic failure by targeting VAMP2 [21]; knockdown of miR-135a triggered an early stress response by disturbing presynaptic vesicle release [22]; miR-153 was found to be a contextual fear-induced miRNA in the dentate gyrus [23] and was reported to regulate vesicle release in zebrafish [24]. However, these studies did not analyze CBH-induced changes in miRNA levels or their function in presynaptic plasticity.
Here we show that CBH obstructs presynaptic vesicle fusion with presynaptic membrane via miR-153-mediated downregulation of multiple synaptic vesicle-related proteins. Therefore, we consider that miR-153 may be a new drug target for the treatment of dementia.

Animals
Male Sprague-Dawley (SD) rats (5~6 months old) were obtained from the Animal Center of the Second Affiliated Hospital of Harbin Medical University (Harbin, Heilongjiang Province, China). All animals for experiments were housed at 23 ± 1°C with 55 ± 5% humidity and maintained on a 12-h dark-light artificial cycle (lights on at 7:00 AM) with food and water available ad libitum. Samples for quantitative real-time PCR (qRT-PCR) and western blot assay were obtained from the hippocampi and/or cortices of rats after they had been anaesthetized with 10% chloral hydrate (500 mg/kg, intraperitoneal, Aladdin, Cat. No. C104202, Shanghai, China) followed by confirmation of death by exsanguination. Tissues for primary neuron culture were obtained from neonatal SD rats after administration of 20% isoflurane and confirmation of death by cervical dislocation. All animal procedures were approved by the Institutional Animal Care and Use Committee at Harbin Medical University (No. HMUIRB-2008-06) and the Institute of Laboratory Animal Science of China (A5655-01). All procedures were conformed to the Directive 2010/63/EU of the European Parliament.

Permanent, bilateral common carotid artery occlusion (2VO) in the rat
The method used for preparation of 2VO rats has been describedin previous studies [3,25] (www.bio-protocol. org/e2668). Briefly, after being anaesthetized by 10% chloral hydrate (300 mg/kg) and maintained under anaesthesia using 0.5-1.0% isoflurane, rats were placed on an electric heating pad (Dongxiyi, Cat. No. wi95919, Beijing, China) to maintain body temperature at 37°C. Then the fur around the neck was removed using an electric shaver and the area was sterilized by a 75% alcohol cotton ball. A 2-cm incision was made above the manubrium along the anterior midline of the neck. The bilateral common carotid arteries of rats were exposed by vertical separation of the omohyoideus muscle and then carefully separated from the vagus and aortic depressor nerve followed by permanent ligation with 3-0 silk suture (Jinhuan, model: 3-0, China). After the surgical procedures, all the anterior cervical muscles were returned to their original location. To avoid potential postoperative infection, the wounds were washed before closing with 20 mg/mL gentamycin sulfate solution (Sangon Biotech, Cat. No. B540724). The wounds were then sutured and rats were allowed to recover from anaesthesia before being returned back to their cages.

Stereotaxic injection of the lentiviral vectors
After anaesthesia, rats were placed on a stereotaxic frame (RWB Life Science Co. Ltd., China), and the skull was exposed. Then, 2 μL (10 8 TU/mL) lenti-pre-miR-153-3p, lenti-2′-O-methyl antisense oligoribonucleotides to miR-153 (AMO-153), lenti-mis-pre-miR-153-3p and/ or leni-mis-AMO-153-3p was injected into the bilateral CA1 region of the hippocampus using a 5-μL Hamilton syringe with a 33-gauge tip needle (Hamilton, Bonaduz, Switzerland) at a rate of 0.5 μL/min. The injection coordinate relative to the bregma was as follows: anteroposterior (AP), − 3.60 mm; mediolateral (ML), ±2.30 mm; dorsoventral (DV), − 3.00 mm below the surface of the dura using coordinates derived from the atlas of Paxinos and Watson. The needle was maintained in the place for another 2 min after injection and then withdrawn very slowly to prevent backflow of the solution. Finally, the skin incision was sutured, and the animal was returned to its housing after recovering from anaesthesia. Subsequent experiments were performed 8 weeks after virus injection [3].

Morris water maze (MWM)
The maze consisted of a black circular pool of 2.0 m diameter, filled with opaque water (25 ± 1°C) via the addition of black food pigment. A submerged escape platform (20 cm in diameter, top surface 2.0 cm below water level) was located in the centre of the first quadrant. Before training, the pupillary light reflex of all rats was tested, and those with impaired pupillary light reflex were excluded from the experiment to avoid the influence of the animal's vision on the test. For cued training (three trials per day for 5 d), the rats were released into the water facing the sidewalls, and each rat was allowed 120 s to find the platform, if the rats did not find it in the time allowed, they were guided to the platform and permitted to rest for at least 20 s. After the last cued trial of day 5, the platform was removed from the pool, and each rat was tested on one 120-s swim probe trial on day 6. Escape latency (s) and the number of platform crossings were monitored using an online DigBehav-Morris Water Maze Video Analysis System (Mobile Datum Software Technology Co. Ltd., Shanghai, China) [3,26].

Functional magnetic resonance imaging (fMRI) measurements
After rats were anaesthetized with chloral hydrate (300 mg/kg,i.p.), a dose of 0.06 mmoL/kg of Dimeglumine gadopentetate was injected into the rat's tail vein, and an animal brain coil was used for T2-weighted imaging to acquire fMRI measurements using a 3.0-T animal MIR scanner (PHILIPS: ACHIEVA 3.0-TX) with fast spinecho plus sequences. The parameters were follows: data matrix = 100 × 92, repetition time (TR) = 2534 ms, effective echo time (TE) = 40 ms, echo train length = 27, field of view = 0.8 × 0.25 cm, twenty-five 1-mm slices, and four signal averages [27].

2,3,5-Triphenyltetrazolium chloride (TTC) staining
After the rats were anaesthetized, the brains of rats were immediately removed and cut into 2-mm sections, which were then immersed sequentially into a phosphatebuffered 2% TTC solution at 37°C for 30 min, and then fixed in phosphate-buffered 4% paraformaldehyde (PFA) solution at 4°C [3].

Extracellular recordings
Concentric bipolar microelectrodes (CBARC75, FHC, USA) were placed in the Schaffer collateral (SC) domain of CA3 300 μm away from recording pipettes, which were placed in the stratum radiatum of CA1. Recording pipettes were pulled from borosilicate glass (BF100-58-10, Sutter Instrument) with resistances of 2-3 MΩ when filled with a NaCl (3 mol/L) solution. The field excitatory postsynaptic potential (fEPSP) of CA3-CA1 was evoked by current stimulation of 0.033 Hz with stimulation steps of 20-to 160-μA (10-steps, 100 ms) using a stimulatory isolator (ISO-Flex (AMPI, Jerusalem, Israel), controlled by a Master-8 pulse generator (AMPI, Jerusalem, Israel). The stimulus intensities, which corresponded to that necessary to evoke~50% of the maximum amplitude of fEPSP, were used for the next synaptic plasticity evaluation [29]. Analogue signals were bypass filtered and digitized at 6 kHz using Digidata 1550A and pClamp10 software (Molecular Devices, US). Paired-pulse facilitation (PPF) in hippocampal CA1 was recorded in separate slices and measured by evoking fEPSP with varying interpulse intervals from 20 to 500 ms.
Signal analysis Off-line analysis was performed using Clampfit software (Molecular Devices, US). The basic synaptic transmission was assessed by the normalized amplitude of fEPSP calculated by the amplitude of each fEPSP relative to the amplitude evoked by 20 μA stimulation. Presynaptic vesicle release was evaluated by PPF which is expressed paired-pulse ratio (PPR): fEPSP2 amplitude/fEPSP1 amplitude [30].

Extraction of synaptosomes
After anaesthetization and perfusion with chilled normal saline, rats were sacrificed with 10% chloral hydrate (500 mg/kg), and the hippocampi were quickly removed and immediately placed on an ice-cold plate, followed by suspension in 10% (w/v) 320 mM sucrose HEPES buffer and homogenization with a grinding rod. The HEPES buffer contained the following (in mM): 145 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 5 glucose and 5 HEPEs buffer with pH 7.3-7.4. Furthermore, the homogenate was centrifuged at 4-8°C for 10 min at 600×g. The supernatant was then diluted at a 1:1 ratio with 1300 mM sucrose HEPES buffer to yield a suspension at a final concentration of 800 mM sucrose, and centrifuged at 4-8°C for 15 min at 12000×g. The supernatant was discarded. The pellet consisting of synaptosomes was washed twice with HEPES buffer and centrifuged at 12,000×g for 15 min at 4-8°C to wash out the impurities [31]. To obtain the synapse proteins for western blot experiments, the resulting synaptosomal preparation was disassociated in RIPA buffer with 0.2% TritonX-100 and 1% protease inhibitor. After standing for 30 min on ice, the suspension was further centrifuged at 20,000×g for 30 min at 4-8°C.

FM1-43 fluorescent dye evaluation
After the neonatal rat hippocampal and cortical neurons (NRNs) were cultured for 10-14 d in cover glasses, they were incubated with FM 1-43 fluorescent dye (T35356, Invitrogen, Oregon, USA) for 3 min at room temperature to allow FM1-43 to bind to the outer membrane of NRNs. Next, 70 mM KCl was added to the FM1-43 loaded NRNs for 3 min at room temperature to allow for the internalization of FM1-43 dye by endocytosis. After this, the NRNs were washed gently with 37°C 0.9% NaCl to remove the extracellular FM1-43 dye. Finally, 70 mM KCl was added again to elicit synaptic vesicle exocytosis. The FM1-43 fluorescent dye signal was monitored by confocal microscopy (Olympus FV1000, Japan), and fluorescence images were acquired every 1 s [32,33].

Primary hippocampal and cortical neuron culture
Rat pups from postnatal days 1-3 (P 1-3 ) were first anaesthetized with isoflurane and decapitated, and then the hippocampi and cortices were removed and placed in cold phosphate-buffer solution (PBS, Solarbio, Cat. No. P1010, Beijing). After dissection and trituration, tissues were digested in 0. C8768, USA) to inhibit astrocyte proliferation. For the western blot and qRT-PCR experiments, the neurons were used 5-7 d after plating. For the FM1-43 dye experiments, the neurons were plated on cover glass precoated with 10 μg/mL PDL and cultured for 12 d. Half the volume of culture media was changed every three days [3].

Dual luciferase reporter assay
Plasmid construction was performed according to the method previously described [3,34]. Genome DNA templates were first extracted from rat tails with Wizard® genome DNA purification kit (Promega, Cat. #A1125), and DNA electrophoresis was used to verify the quality of extracted genomic DNA using DL2000DNA marker (Fig. S1a). The sequences of the full-length 3′ untranslated region (3'UTR) and mutant 3'UTR of the Snap25, Vamp2, Stx1a and, Syt1 genes 3 are shown in Additional file 1. Amplification of the acquired genomic DNA was performed by PCR and the PCR products were recollected with a DNA gel extraction kit (Dongsheng Biotech) (Fig. S1a-e). The primers for the amplification of the full-length 3'UTR of the Snap25, Vamp2, Stx1a, and Syt1 genes are shown in Table S1. Next, the purified PCR products were incubated with psiCHECK-2 vectors using the XhoI and NotI enzyme for enzymatic digestion. The enzyme digestive products were purified on an agarose gel with a DNA Gel extraction kit and ligated into psiCHECK-2 vectors using T4 DNA ligase (TaKaRa, D2011A). The ligated products were transformed into E.coli DH5α competent cells. The positive recombinant clones were screened by agarose gel electrophoresis and BLAST analysis of cloned DNA sequences relative to the 3'UTRs of the genes published in NCBI was employed to verify vector construction (Fig. S2).
Mutagenesis of nucleotides was carried out using direct oligomer synthesis for the 3'UTR region of Syt1, Stx1a, Snap25 and Vamp2. Point mutations were introduced into a possible miR-153 binding site located in the 3'UTR region. Plasmid construction for carrying the mutated 3'UTRs of these four genes were constructed using the same method as that used for wild-type plasmid construction. The PCR primers for the mutagenesis of the 3'UTRs of these four genes are shown in Table S1.

Transfection procedures
After six days of cell culture, 75 pmoL/mL miR-153 mimics and/or AMO-153, oligodeoxynucleotides (ODNs), scrambled miR-153 or scrambled AMO-153 (Table S2) were transfected into cultured neonatal hippocampal and cortical neurons with X-treme GENE siRNA transfection reagent (catalog#04476093001, Roche, USA) according to the manufacturer's instructions. Forty-eight hours after transfection, cells were collected for subsequent total RNA isolation, protein purification or FM1-43 staining. Transfection was performed at 12 d for the FM1-43 dye experiments [3].

Immunofluorescence detection
After 48 h, the cultured cells were fixed in 4% paraformaldehyde for 30 min. After the cells were blocked, they were incubated with the Tuj1 antibody (Cat.#T8578; 1: 200; Sigma, USA) to label the neurons and then the cultured cells on the glass were washed and incubated with the secondary antibodies conjugated to Alexa Fluor 488 (Invitrogen) for 1 h at 37°C. After incubation with secondary antibody and DAPI as usual, neurons were mounted on coverslips to obtain confocal images by FluoView™ FV300 (Olympus) using × 60 objective with the same condition at a resolution of 1024 × 1024 pixels (12 bit) [27].

Western blot
The concentrations of the protein extracts were measured using the Bio-Rad Protein Assay kit (Bio-Rad, Hercules, CA, USA) with bovine serum albumin standards. Protein samples fractionated on a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis gel were transferred to a nitrocellulose membrane. Table S4 displays all of the primary antibodies used in this study. Western blot bands were captured using the Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA) and quantified using Odyssey version 3.0 software by measuring the band intensity (areaoptical density (OD)). Protein expression in each group was normalized to the internal control β-actin (1:1000, G8795, Sigma).

Human blood sample preparation
Vascular dementia patients (between 70~90 years old), from the Second Affiliated Hospital of Harbin Medical University, were recruited for this experiment based on the criteria of the National Institute of Neurological and Communicative Disorders (NINCDS) (Sarazin, etal., 2012). Written consent was obtained from all subjects, and the study protocol was approved by the Ethics Committee of Harbin Medical University (HMUIRB20140029). To prepare plasma samples for miRNA evaluation, whole blood samples (2.5 mL per patient) were collected from subjects via a direct venous puncture into tubes containing sodium citrate and centrifuged at 1000×g for 5 min, and then the supernatant (plasma) was carefully transferred into an RNase-free tube for RNA extraction [35].

Statistical analysis
All data are represented by the mean and standard error of the mean (s.e.m). Each data set was analysed for its ability to meet the statistical assumptions for equality of the variance. The independent sample test was calculated using the Levene variance equality test. If P > 0.05, independent Student's t-test was used for the comparison between two groups; if P < 0.05, the Kruskal-Wallis rank sum test was performed. One-way ANOVA was performed for the comparison among multivariate groups, and post hoc analyses of significant main effects were further examined using Fisher's procedures for learning systems design (PLSD) tests. For repeated measures data, Mauchly's test of sphericity was first performed to evaluate the relationship among the repeatedly measured data. If P > 0.05, a general linear model was selected for further analysis; if P < 0.05, Greenhouse-Geisser corrected results or multivariate ANOVA was used for further analysis of multiple comparisons. All statistical analyses were performed in SPSS software (version 11).

CBH impairs presynaptic function in rats
To evaluate presynaptic plasticity in the hippocampi of CBH rats, we performed 2VO surgery on rats [3,36]. As shown in Fig. 1a and b, similar to our previous study [3], 2VO surgery elicited markedly low cerebral blood flow (CBF) at 8 weeks (8 W), as indicated by both TTC staining and fMRI evaluation. We then prepared brain slices and monitored the fEPSP in the stratum radiatum of CA1 by electrically stimulating the SC of the CA3 pathway (Fig. 1c). Based on the input-output (I-O) curve, compared with the sham group, the 2VO group exhibited reduced fEPSP responses (Fig. 1d, P = 0.002). For example, the fEPSP amplitude in the 2VO rat slices was~75% of that observed in the sham controls when the stimulus intensity was 60 μA, and the percentage was further decreased to~60% at 120 μA stimulation (Fig. 1e, P = 0.003). To evaluate the presynaptic vesicle release probability, PPF (60 μA) measurements were performed and evaluated by PPR [30]. The PPR in the brain slices of the 2VO group, compared with that of the sham controls, was significantly increased when the interstimulus interval (ISI) ranged from 20~100 ms (Fig. 1f, P < 0.0001). The PPR was 1.48 ± 0.01 in the sham group and 1.85 ± 0.02 in the 2VO rats when the ISI was 40 ms (Fig. 1g, P < 0.0001). These findings demonstrate an impaired presynaptic function in the 2VO rats.
Previous elegant studies have demonstrated that synaptic vesicle fusion is a crucial step for presynaptic neurotransmitter release triggered by the influx of Ca 2+ , which is controlled by the core fusion machinery composed of the SNARE-complex and Sec1/Munc18-like (SM) proteins [12,37] (Fig. 1h). We speculated that synaptic vesicle fusion-related proteins might be changed under CBH conditions. Western blotting experiments showed that the protein levels of all three SNAREcomplex proteins (SNAP-25, P = 0.0017; syntaxin-1A, P = 0.005 and VAMP-2, P = 0.02) and the Ca 2+ -sensor protein Syt1 (P = 0.015) were significantly lower in the hippocampi of 2VO rats than in those of sham rats ( Fig.  1i and j). However, the protein levels of Munc-13, Munc-18-1 and Complexin1/2, coregulators of vesicle exocytosis, were unchanged ( Fig. 1k and l, P > 0.05). These findings suggest that the decreased protein expression of SNARE complex proteins and Syt1 may participate in the impaired presynaptic vesicle release observed in CBH rats.

MiR-153 targets SNARE complex-related genes and the Syt1 gene
Compared to the levels in sham rats, the SNARE complex protein and Syt1 protein levels were decreased in 2VO rats, while their mRNA levels were not reduced, indeed, and the level of SNAP25 mRNA was even increased in 2VO rats (Fig. 1m, P = 0.025). Therefore, we considered that a posttranscriptional regulatory mechanism might be involved in the observed changes. Multiple miRNAs have been previously reported to regulate synaptic fusion-related proteins at the mRNA level. For example, miR-153 regulates Snap25, Vamp2, Snca, Trak2, Bsn and Pclo genes at the mRNA level [23]; miR-137 inhibits complexin-1, Nsf and Syt1 in mouse model of schizophrenia [20]; miR-34c targets VAMP-2 [21] and miR-135a binds the complexin-1 and complexin-2 genes in the amygdala [22]. To identify which one of the above miRNAs is involved in the CBH induced downregulation of fusion-related proteins in the hippocampi of rats, we performed qRT-PCR test. As displayed in Fig. 2a, compared to the sham rats, 2VO rats exhibited~2-fold increase in the expression of miR-153 (P = 0.011), however, the miR-137 and miR-135a levels were unchanged (P > 0.05). In contrast, miR-34c expression was reduced in the hippocampi of 2VO rats. These data suggest that the decreased expression of SNAP-25 or VAMP2 in the hippocampi of 2VO rats may result from upregulated expression of miR-153, but do not correlate with the observed downregulation of miR-34c. Additionally, the decrease in Syt1 protein expression was not associated with miR-137 expression. Furthermore, to elucidate the potential clinical significance of miR-153, 12 male subjects (7 controls and 5 VaD patients) with higher education backgrounds were recruited to our study and assessed by the Mini-Mental State Examination (MMSE), Montreal Cognitive Assessment (MoCA) and Hamilton Depression Scale (HAMD) ( Table S5). As Fig. 1 Impairment of presynaptic vesicle release in CBH rats 8 W following 2VO surgery. a TTC staining was used to identify brain ischaemia in rats at 8 W after 2VO. Red represents normal tissue and white represents infarct tissue. b fMRI brain images of rats at 8 W after 2VO. fMRI, functional magnetic resonance imaging. c Schematic diagram showing the placement of the recording (CA1) and stimulating electrodes (CA3) in hippocampal slices. d Comparison of I-O curves between sham and 2VO rats. F (14,196)   shown in Fig. 2b, the levels of miR-153 in the plasma of VaD patients were significantly higher than those in the plasma from the control group (P = 0.019), indicating that miR-153 is potentially involved in the pathophysiological process of human presynaptic vesicle release disorders. Bioinformatics methods (TargetScan Human 5.1 and RNAhybrid database) were then employed to identify the potential miRNAs regulating Syt1 and syntaxin-1A expression. Surprisingly, we found that miR-153 alone had conserved binding sites in the 3'UTRs of the Snap25 and Vamp2 genes as well as targets in the 3'UTRs of the Stx1a and Syt1 genes (Fig. 2c).
To better understand how miR-153 regulates all four genes, we first performed a dual luciferase reporter gene assay to evaluate the binding ability of miR-153 with these genes. The full-length 3'UTRs of these four genes containing the miR-153 binding sites were separately cloned, purified, and ligated into the psiCHECK™-2 vector at the site between the Renilla luciferase gene and the synthetic poly (A) tail (Fig. S1a-e and S2a-c). Thereafter, the effects of miR-153 on reporter activities were assessed in HEK293T cells. As predicted, cotransfection of chemically synthesized miR-153 mimics and the psiCHECK™-2 vector plasmid consistently produced lower Renilla luciferase activities than transfection of the empty psiCHECK™-2 vector plasmid alone ( Fig. 2d-g, P < 0.0001). The silencing effect of miR-153 reached~60% on the Snap25 transcript (Fig. 2d),~50% on the Vamp2 transcript (Fig. 2e),~35% on the Stx1a transcript (Fig. 3f) and~70% on the Syt1 transcript (Fig. 2g), and these effects could be prevented by cotransfection of the psi-CHECK™-2 vector plasmids with AMO-153. However, the transfection of psiCHECK™-2 vector plasmids with scrambled oligoribonucleotides of either mis-miR-153 or antisense mis-AMO-153 in all four genes failed to affect luciferase activity (Fig. 2d-g). Furthermore, when we cloned the full-length 3'UTR of the Snap25 (1057-1063 bp), Vamp2 (320-327 bp), Stx1a (841-848 bp) and Syt1 (1622-1628 bp) with mutant binding sites (Fig. 2c), the repressive effects of miR-153 on the luciferase activities of these genes were abolished ( Fig. 2d-g, P > 0.05). These results suggest that miR-153 directly target the 3'UTRs of the Snap25, Vamp2, Stx1a and Syt1 genes.
To confirm that miR-153 gain-of-function affects the expression of these four proteins in neurons, chemically synthesized miR-153 mimics and AMO-153 were Fisher's PLSD test was used for the post hoc analyses of the two-group comparisons. *P < 0.05 vs NC; # P < 0.05 vs miR-153 successfully transfected into cultured NRNs by the Xtreme GENE siRNA transfection reagent (Fig. 3a, P < 0.0001 and Fig. S3). The miR-153 mimics effectively inhibited the expression of all four proteins with an inhibitory efficiency of~48% for SNAP-25 (P = 0.002),4 7% for Syt1 (P = 0.001),~43% for syntaxin-1A (P = 0.002) and~28% for VAMP-2 (P = 0.005), whereas the scrambled NC failed to affect the levels of these proteins (Fig. 3be). In contrast, AMO-153 rescued the downregulation of all proteins elicited by the miR-153 mimics. Subsequently, to confirm that miR-153 mediates the downregulation effect on these four proteins by directly targeting the binding sites of the four genes, the miRNA-masking antisense ODNs (miRmasking) technique was employed as previously reported [3]. An ODN is an antisense oligodeoxynucleotide fragment designed to fully base pair to a protein-coding mRNA at the sequence motif spanning the binding site for an endogenous miRNA of interest. Since a miR-mask only acts on the target gene with minimal effects on other target genes that may also be targeted by a same miRNA, the anti-miRNA action of a miR-mask is gene-specific. Based on this mechanism, we designed four miR-153 masks that could base pair the miR-153 binding sites in the 3'UTRs of the Snap25, Vamp2, Stx1a and Syt1 genes and labelled these masks Snap25-ODN, Vamp2-ODN, Stx1a-ODN and Syt1-ODN, respectively, and evaluated whether these ODNs could shield the action of miR-153 on the four genes. As predicted, after successful transfection of the four ODNs (Fig. 3f), these ODNs effectively blocked the repressive effects of miR-153 on the protein expression of SNAP-25 (P = 0.002), syntaxin-1A (P = 0.004), VAMP-2 (P = 0.008) and Syt1 (P < 0.0001) (Fig. 3g-j), highlighting the sequence specificity of the actions of miR-153. To determine whether miR-153 gain-of-function impairs synaptic vesicle release, we loaded NRNs with FM1-43 fluorescent dye, which can be used to monitor vesicle release by labelling synaptic vesicles [32]. As illustrated in Fig. 4a and b, after treatment with phosphate buffer saline (PBS), the red fluorescent signal of the synaptic boutons was not changed in the NRNs transfected with NC, which was sharply depleted by adding KCl. The F/F0 ratio was reduced to~0.4 at 10 s and further decreased to~0.3 at 40 s. However, the miR-153 mimics markedly blocked the fast-declining fluorescent signal of the boutons with the F/F0 ratio as high as 0.8 at 10 s and 0.7 at 40 s after KCl stimulation; this effect was fully reversed by the cotransfection with AMO-153 ( Fig. 4a-c, P < 0.0001). Overall, these data provide evidence that miR-153 gain-offunction impairs presynaptic vesicle release.
Knockdown of miR-153 attenuates cognitive decline of 2VO rats Subsequently, we investigated whether miR-153 knockdown could prevent the dementia phenotype induced by 2VO. As expected, compared with age-matched 2VO rats that were transfected with lenti-NC, 2VO rats transfected with lenti-AMO-153 exhibited a significantly shortened latency to arrive at the platform (Fig. 7a, P = 0.001). In the probe trial, lenti-AMO-153 treatment increased the number of platform crossings in 2VO rats ( Fig. 7b and c, P = 0.08). These results imply that knockdown of miR-153 improved spatial memory in 2VO rats.

Discussion
We are the first to report that CBH impairs presynaptic plasticity by disturbing synaptic vesicle release. The disturbed process was due to a blockade of presynaptic vesicle fusion with the presynaptic membrane controlled by miR-153, which posttranscriptionally repressed the expression of fusion-related proteins, including the SNARE-complex proteins SNAP-25, VAMP-2 and syntaxin-1A and the Ca 2+ sensor protein Syt1 by targeting the 3'UTRs of the Snap25, VAMP2, Stx1a and Syt1 genes (Fig. S4). This mechanism not only deepens our understanding of CBH-induced brain dysfunction, but also leads to a new preventive or treatment strategy for AD or VaD.
The number of formed SNARE complexes is widely known to be the rate-limiting index controlling the quantity and speed of presynaptic vesicle release. Although 2~3 SNARE-complexes are sufficient to mediate the fusion process between vesicles and presynaptic membranes, more SNARE complexes trigger greater vesicle release [38]. Here, CBH downregulated the expression of SNAP-25, VAMP-2 and syntaxin-1A, suggesting that CBH induced vesicle release impairment may reduce the number of SNARE complexes. Synaptic vesicle fusion is not only mediated by SNARE proteins but also co-regulated by SM proteins that control fast vesicle release by opening the fusion-pore by engaging in the formation of the trans-SNARE/SM-complex [15,39]. However, in our study, Munc-18-1 expression was unchanged in the hippocampi of 2VO rats, suggesting that the CBH-induced impairment in synaptic fusion is associated with reduced SNARE protein expression but not SM protein expression. Furthermore, prior to fusionpore opening, synaptic vesicles need to be docked and Fig. 7 Knockdown of miR-153 attenuates learning and memory deficits in the 2VO model. a Knockdown of miR-153 by injecting lenti-AMO-153 reversed the increase in the daily average escape latency to locate the platform from three nontarget quadrants in 2VO rats. χ 2 Mauchly = 16.20, P = 0.063;; F total (4, 104) = 7.404, P = 0.001. b Average number of platform crossings during the probe trial. P Levene = 0.005, one-way ANOVA: F = 4.736, P = 0.008. n = 8, *P < 0.05 vs lenti-NC; # P < 0.05 vs 2VO + lenti-AMO-153. One-way ANOVA, Fisher's PLSD test were used for the post hoc analyses of the two-group comparisons. c Represent path tracing of the probe trial test on day 6 in the MWM test for each group