MicroRNA-29c-3p in Dual-Labeled Exosome is Potential Marker of Alzheimer's Disease

Objective We aimed to establish a method to determine whether peripheral blood NCAM/amphiphysin 1 dual-labeled exosomal proteins and microRNA might be serve as a marker for the diagnosis of Alzheimer's disease. Methods It was a multicenter study using a two-stage design. The subjects included 45 SCD, 50 aMCI, 40 AD patients, and 30 controls in the discovery stage; the results were conrmed in the verication stage (47 SCD, 45 aMCI, 45 AD, and 30 controls). Peripheral blood NCAM single-labeled and NCAM/amphiphysin 1 double-labeled exosomes were captured and detected by immunoassay, respectively. Results Levels of Aβ42, Aβ 42/40 , Tau, P-T181-tau, and miR-29c-3p in plasma NCAM single-labeled and NCAM/amphiphysin 1 double-labeled exosomes of the aMCI and AD groups were signicantly higher than those of the SCD group, control group and VaD group (all P<0.05). The levels of Aβ42 and miR-29c-3p in peripheral blood NCAM/amphiphysin 1 dual-labeled exosome was higher than that in the control and VaD groups (all P<0.05). The levels of exosomal Aβ42, Aβ 42/40 , Tau, P-T181-tau, and miR-29c-3p in peripheral blood were correlated with that in CSF (all P<0.05). This study rst provided a method for sorting specic surface marker exosomes using a two-step immune capture technology. The plasma NCAM/amphiphysin 1 dual-labeled exosomal Aβ 42/40 and miR-29c-3p had potential advantages in the diagnosis of SCD. and the plasma exosomal miR-29c-3p increased signicantly, compared with the wild-type mice [12]. The decrease of MiR-29c-3p level may lead to the progression of AD [13]. These results suggested that miR-29c-3p in CSF was transported to peripheral blood through NCAM/amphiphysin 1 dual-labeled exosomal NCAM single-labeled NCAM/amphiphysin 1 dual-labeled exosomal Aβ42, 42/40 , T-tau, P-T181-tau, and miR-29c-3p. results from discovery validated P-T181-tau in plasma exosomes of the aMCI AD groups signicantly higher than of the SCD group, control group and VaD group, and the AD group higher than the aMCI group. The levels of plasma exosomal Aβ42, Tau and P-T181-tau in the SCD group were slightly higher than those in the control group and the VaD group, but there was no signicant difference. The level of NfL in plasma exosomes in the aMCI and AD groups was signicantly higher than that in the SCD group, control group and VaD group, and the AD group was higher than that in the aMCI group. The NfL level of plasma exosomes in the SCD group was slightly higher than that in the control group and the VaD group, but there was no signicant difference. The levels of miR-29c-3p in plasma exosomes in the aMCI and AD groups were signicantly higher than that in the SCD, control, and VaD group; while the AD group was higher than the aMCI group. The level of plasma exosomes miR-29c-3p in the SCD group was slightly higher than that in the control group and the VaD group, but there was no signicant difference. The change trend of the test results of the specimens in the verication stage is consistent with the discovery stage (Fig. We rst used the specimens from the discovery stage to detect the NCAM/amphiphysin 1 dual-labeled exosomal proteins extracted in the second stage of the exosome extraction experiment. The results showed that the levels of Aβ 42/40 , Tau, and P-T181-tau in plasma exosomes of the aMCI and AD groups were signicantly higher than those of the SCD group, control group and VaD group, and the AD group was higher than the aMCI group. The levels of plasma exosomes Aβ 42/40 , Tau, and P-T181-tau in the SCD group were slightly higher than those in the control group and the VaD group, but there was no signicant difference. There was no signicant difference in the level of exosomal Aβ40 among the groups, and only the AD group decreased slightly. The level of NfL in plasma exosomes in the aMCI and AD groups was signicantly higher than that in the SCD group, control group and VaD group, and the AD group was higher than that in the aMCI group. The NfL level of plasma exosomes in the SCD group was slightly higher than that in the control group and the VaD group, but there was no signicant difference. The change trend of the test results of the specimens in the verication stage was consistent with the discovery stage. In all disease groups and control groups, the levels of Aβ42 and Aβ 42/40 in NCAM/amphiphysin 1 dual-labeled exosomes were higher than those of NCAM single-labeled exosomes, considering that the concentration of exosomes in each group has been standardized by CD81. The levels of Aβ42 and miR-29c-3p in plasma NCAM/amphiphysin 1 dual-labeled exosomes of the SCD, aMCI and AD groups were signicantly higher than those of the control and VaD groups, the AD group was higher than the aMCI group, and the aMCI group and 0.906 (P = 0.000), 0.915 (P = 0.000), respectively; the AUCs for the SCD/aMCI and aMCI/AD were 0.778 (P = 0.000) and 0.670 (P = 0.000), respectively. The results of DeLong test showed that the AUC of NCAM/amphiphysin 1 dual-labeled exosomal Aβ 42/40 for diagnosis of SCD was higher than that of Aβ42 (P = 0.032), T-tau (P = 0.001) and P-T181-tau (P = 0.025); the AUC of NCAM/amphiphysin 1 dual-labeled exosomal miR-29c-3p for diagnosis of SCD was higher than that of Aβ42 (P = 0.027), Aβ 42/40 (P = 0.001), T-tau (P = 0.001), P-T181-tau (P = 0.008), and NfL (P = 0.000). These data indicate that plasma NCAM/amphiphysin 1 dual-labeled exosomal Aβ 42/40 and miR-29c-3p had potential advantages in the diagnosis of SCD. considered to be the prime modulator of cholesterol homeostasis in the brain. Current study showed that the diagnostic ecacy of miR-29c-3p of peripheral blood NCAM single-labled exosomes for AD and aMCI is similar to that of Aβ42, but their diagnostic value for SCD was limited. However, miR-29c-3p in NCAM/amphiphysin 1 dual-labeled exosomes of peripheral blood not only has better diagnostic performance for AD and aMCI, but also has acceptable diagnostic performance for SCD, which is not available in peripheral blood exosomes Aβ42. In addition, for T-tau, P-T181-tau and NfL, there was no signicant difference in the levels of NCAM single-labeled exosomes and NCAM/amphiphysin 1 double-labeled exosomes, as well as the diagnostic eciency of AD diagnosis and staging, which suggested that there is no need to detect NCAM/amphiphysin 1 dual-labeled exosomal T-tau, P-T181-tau and NfL for diagnosis and staging of AD. These data indicate that blood neuron-derived exosomes are ideal biomarker carriers for AD screening, especially for large-scale population screening. In summary, this study rst provided a method for sorting specic surface marker exosomes using a two-step method; and veried that peripherald neuronal-derived exosomal Aβ42, Aβ 42/40 , T-tau, P-T181-tau and miR-29c-3p may reect AD pathological changes in the brain and therefore have the capacity to diagnose AD. The detection of NCAM/amphiphysin 1 dual-labeled exosomal molecules show a better potential diagnostic value for SCD than NCAM single-labeled exosomes. However, these ndings need further conrmation in longitudinal studies.


Introduction
Diagnosis of Alzheimer's disease (AD) requires physical signs, psychological tests, syndrome judgment, and biomarker testing. AD biomarkers can generally be divided into two categories: "imaging markers" and "clinical laboratory diagnostic markers". In vivo diagnostic imaging markers such as amyloid PET (positron emission computed tomography) imaging have good diagnostic speci city on AD, and it also has potential value in the diagnosis of SCD, however the expensive costs of PET limit its application as a routine inspection item [1,2]. At present, there are no drugs that can effectively treat patients with DAT, but patients in the middle and early stages such as amnestic mild cognitive impairment (aMCI) and subjective cognitive decline (SCD) could avoid or delay the development of the disease through a series of treatments [3,4]. Therefore, it is more important to make a clear diagnosis in the early stages when Aβ has not yet signi cantly accumulated [5]. Recent studies have shown that the increase and aggregation of amyloid β-protein (Aβ) outside the cell may not be the earliest molecular changes in AD: a series of molecular biological changes have begun to occur in neuronal cells before the occurrence of Aβ aggregation that can be detected by PET, which are considered to be the earliest (ultra-early) changes that lead to the occurrence, progression and clinical onset of AD [5][6][7]. In recent years, with the conclusion of a number of large-scale long-term longitudinal studies, the clinical application value of multiple new familial AD ultra-early diagnostic markers has been initially clari ed. For example, molecules such as peripheral blood neuro lament light chain (NfL) are expected to provide early warning of the onset of familial AD with presenilin 1 mutant 10-15 years in advance, so as to achieve ultra-early early warning [8]. However, for sporadic AD, which accounts for 90% of the total number of AD, there is still a lack of effective peripheral blood ultra-early warning biomarkers.
CSF is in direct contact with the brain, and the material exchange with the brain is more direct than that of peripheral blood and other body uids. Therefore, CSF is considered to be a "clinical laboratory diagnostic biomarker" specimen that can diagnose AD and other neurological diseases more speci cally and earlier [2]. However, because CSF needs to be collected invasively by lumbar puncture, there is a certain clinical risk, and patient acceptance is low, and it is di cult to carry out routine and large-scale development. Due to the existence of the blood-brain barrier (BBB) transport mechanism, CSF and peripheral blood can also exchange certain substances. In recent years, a series of studies have suggested the possibility of using peripheral blood as a specimen source for AD diagnostic markers. In addition, peripheral blood collection is less traumatic and patient acceptance is high. If markers with high disease diagnosis e ciency can be found, it will greatly promote the AD marker routine detection and effectively increase the AD detection rate. However, facts have proved that some potential AD markers (such as proteins, miRs, etc.) that have greater clinical value in CSF, when directly detected in peripheral blood, have unsatisfactory diagnostic performance such as positive rates. The reason may be the BBB transport process leads to changes in its "quality" or "quantity" or interference caused by the secretion of the same or similar substances in other tissues and organs of the body [2,9].
At present, more researched are centered on the non-invasive approaches of the AD early diagnosis or warning. The series of protein biomarker panels, metabolomics biomarker panels, and nucleic acid biomarker panels were initially clari ed [10]. MicroRNAs (miRs) are a class of small (18-25 nucleotides), single-stranded non-coding RNAs involved in the post-transcriptional regulation of gene expression. In recent years, miRs have been shown to play important roles in several diseases, such as cancer, cardiovascular disease, and diabetes, as well as central nervous system diseases. Because of their stable character, altered miRs in tissues and organs may lead to the deregulation of miRs in body uids such as CSF, serum and urine, either by cell destruction or secretion. Therefore, miRs are attractive targets in the search of novel biomarkers [10,11]. In view of the complexity of miR sources, how to detect miRs that representing AD pathological changes has become one of the solutions to reduce false negatives in diagnosis.
In a previous study, we reported that miR-29c-3p was down-regulated in hippocampal neurons of APP/PS1 double transgenic mice, at the same time, the total miR-29c-3p in CSF increased signi cantly, suggesting that the reduced miR-29c-3p in neuronal cells may be released into CSF. However, the total miR-29c-3p in mouse plasma did not change signi cantly, and the plasma exosomal miR-29c-3p increased signi cantly, compared with the wild-type mice [12]. The decrease of MiR-29c-3p level may lead to the progression of AD [13]. These results suggested that miR-29c-3p in CSF was transported to peripheral blood through exosomes across the BBB. Further exosome protein mass spectrometry results showed that expression of the amphiphysin 1 in the CSF of AD patients was signi cantly higher than that of controls, suggested that amphiphysin 1 may be related to the increase of peripheral blood exosomal miR-29c-3p.
Previous multi-center studies have shown that the concentration of neural-derived exosomal Aβ42, T-tau, and P-T181-tau in peripheral blood can re ect their changes in CSF [14]. In the present study, the neural cell adhesion molecule (NCAM) single-labeled exosomes were rst captured, and then speci c NCAM/amphiphysin 1 dual-labeled exosomal components are screened for corresponding detection. All subjects were tested for NCAM single-labeled and NCAM/amphiphysin 1 dual-labeled exosomal Aβ42, Aβ 42/40 , T-tau, P-T181-tau, NfL and miR-29c-3p. The results from discovery stage were validated a in a validation stage with more samples in different centers.

Participants
A total of 392 subjects were randomly selected from the Beijing Anding Hospital (center 1), Xuanwu Hospital of Capital Medical University (center 2), Minhang Hospital, Fudan University (center 3), Air Force General Hospital, Chinese People's Liberation (center 4), from May 2018 to March 2021; of which 92 subjects were in the SCD group, 95 subjects were in the MCI group, and 85 subjects were in the DAT group. Another 60 randomly selected healthy subjects were recruited as the control group. For the evaluation of the differential diagnosis, 60 patients with vascular dementia were also included. The diagnosis of AD subject was made according to the criteria of the National Institute on Aging and Alzheimer's Association (NIA-AA) [15]. The diagnosis of aMCI subject was based on published criteria [16]. The diagnosis of SCD subject made according to published criteria and researches [17,18]. The diagnosis of SCD subject made according to published criteria [19,20]. The experimental protocol was approved by the Ethics Committees of participating centers. Written informed consents were obtained from all participants or their legal guardians.
According to the patients enrolled in three central hospitals in Beijing as the discovery stage of the study, the patients enrolled in the two central hospitals in Shanghai and Heilongjiang province as the veri cation stage. Subjects with endocrine system, liver, kidney or cardiovascular diseases would be excluded. All samples were stored and detected in a blinded manner.

Sample collection and pretreatment
The EDTA anticoagulated venous blood was drawn in the morning after 8 ~ 12 hours fasting (Vacutainer K 2 -EDTA tube, BD, Franklin Lake, USA). To reduce the error caused by the exchange of exosomes and blood, centrifugation is required within 30 minutes after specimen collection. Subjects undergoing blood cell exosomes release experiments were collected two tubes of specimens, and other subjects were collected one tube of specimens. To obtain plasma, the specimens were centrifuged at 4,200 g for 10 minutes at 4°C; to obtain blood cells, blood separation medium was used according to the manufacturer's protocol (Solarbio, Beijing, China) [14]. The plasma was stored in liquid nitrogen before testing and was only allowed to freeze and thaw once. The blood cells were used in the next experiment within 1 hour.
The CSF samples were collected within 1 hour after blood samples collection in accordance to the CSF collection and biobanking guideline [21]. In brief, the patient was positioned in a left lateral position, and the puncture point is between the waist 3 and 4. Each subject collected 5 ml to 15 ml of CSF. The collected CSF was centrifuged at 2,000 g at 4°C for 10 min [13,14]. The supernatant was immediately collected into polypropylene tube, stored in liquid nitrogen before testing, and allowed to freeze and thaw only once.

Blood cell separation and culture
The separated RBCs and WBCs were washed in 37°C pre-warmed phosphate buffered saline (PBS) 3 times and then cultured in Roswell Park Memorial Institute 1640 (RPMI 1640) medium (Invitrogen, Carlsbad, USA) containing 10 % exosome-depleted serum (Gibco, Thermo Fisher Scienti c, MA, USA) at a concentration of 1 × 10 7 /L for 0 h and 2 h. The NCAM in the cells and medium were detected by enzyme linked immunosorbent assay (ELISA) method (Abcam, Cambridge, UK) according to the following procedure. The randomly selected and gender/age-matched SCD, aMCI, DAT, VaD and control samples from XuanWu Hospital were all run in triplicate.

Neuronal-derived exosomes collection from plasma
First of all, the total exosome was extracted from plasma. In brief, 0.5 ml plasma was incubated with 0.15 ml thromboplastin-D (Amresco, Boise, USA) for 60 min at room temperature, then 0.33 ml magnesium and calcium-free Dulbecco's phosphate-buffered saline (D-PBS, Amresco), 0.02 ml EDTA-free protease and phosphatase inhibitor cocktails (Abcam) were added. Next, 0.5 ml of the mixture was added with 0.5 ml of D-PBS, and the mixed solution was centrifuged at 1,500 g for 20 minutes at 4°C. The supernatant was mixed with equal volume of D-PBS, and the mixed solution was centrifuged at 1,500 g for 20 minutes.
Then, speci c neuronal-derived exosomes de ned by multiple studies were separated by a novel immunomagnetic bead method [14,22]. In short, 0.2 ml of supernatant of previous step was mixed with 0.05 ml of ExoQuick ™ exosome precipitation solution (System Biosciences, CA, USA) and incubated on ice for 1 hour. After centrifugation at 1,500 g for 30 minutes, the pellet was resuspended in 200 µl D-PBS. 200 µl of each sample was incubated with the NCAM monoclonal antibody (mouse anti-human, Abcam) coated Dynabeads (1 µm, Thermo Fisher Scienti c) for 30 min in microplate on a shaker at 37°C. Then use a magnet with a shape that ts the microplate to draw on the magnetic beads and discard the supernatant. Remove the magnet, each sample was resuspended in 210 µl 0.05 M glycine-HCl (pH 3.0, Amresco) mixed with 0.10 % Tween 20 and 10 µl inhibitor cocktails by shaking for 30 seconds and draw the beads with the magnet again. Then the supernatant was harvested and the pH of the supernatant was adjusted to 7.0 with 1 M Tris-HCl (pH 8.6, Amresco).
Afterwards, speci c subcomponent of neurogenic exosomes was captured. In brief, 200 µl of the supernatant from the previous step was added into the plate of a human amphiphysin 1 ELISA kit (Abcam). After 30 minutes incubation at 37°C, the plate was washed three times with the wash solution provided in the kit and then drained. Speci c exosomes were adsorbed on the well wall for the next step. The extraction reagent is directly added to the well to extract protein or nucleic acid for subsequent detection, according to the following procedure ( Fig. 1).

Transmission electron microscopy
Exosomes adsorbed on the well wall were eluted with glycine-HCl (pH 3.0) and adjusted pH to 7.0 with Tris-HCl (pH 8.6), then the transmission electron microscopy (TEM) was performed according to published protocol [23]. Sections (65 nm) were stained with uranyl acetate and Reynold's lead citrate. The experiment was performed on JEM-2100F TEM (JEOL, Akishima-shi, Japan).

MiR puri cation and miR analysis
Total miR was puri ed using a magnetic bead method Kit (Thermo Fisher Scienti c) according to the manufacturer's protocol. Total RNA yields were ~ 20 ng/mL and ~ 50 ng/mL in CSF and plasma respectively, as assessed by Quant-iT ™ RiboGreen ™ RNA reagent (Invitrogen). Total miR in blood cells was extracted by a column-based method Kit (Thermo Fisher Scienti c). MiRs were reverse transcribed into cDNA with 10 µl reaction volume, followed by TaqMan qPCR according to the manufacturer's protocol (Thermo Fisher Scienti c). Brie y, a 20 µl PCR reaction contained 2 µl cDNA, 300 nM TaqMan probe, 300 nM sense primer, and 300 nM anti-sense primer. Cycling parameters were 95°C for 8 min, followed by 35 cycles of 95°C for 15 s and 60°C for 1 min (Roche Light Cycler 480, Basel, Switzerland). For body uid samples, miRNeasy Serum/Plasma Spike-In Control (C. elegans miR-39 miR mimic, Qiagen, Dusseldorf, Germany) served as control; while for cell samples, U6 snRNA (RNU6B, Qiagen) served as endogenous control. The relative levels of miRs were calculated by the 2 -ΔΔCt method [24].

Western blot
Western blot was performed to detect exosomal biomarker, Alix, as well as NCAM and amphiphysin 1 using monoclonal anti-human antibodies (Santa Cruz, Dallas, USA). The dilution ratios of the primary and secondary antibodies of Alix, NCAM and amphiphysin 1 were 1:1000/1:400, 1:500/1:300, 1:1000/1:300 respectively. The captured exosomes and the corresponding supernatant in all steps were tested for the above three proteins.

ELISA measurements of protein
The Aβ42, Aβ40, T-tau, P-T181-tau, and NfL (Abcam) ELISA tests were performed in strict accordance with the instructions of the kit. The determination of CD81 content is used for the normalization of exosomes content. The mean value of CD81 levels in each group was set to 1.00, and the relative values for each sample were used to normalize their recovery [14].

Statistical analyses
Statistical analyses were performed using SPSS 18.0 for Windows (SPSS, Inc., Chicago, IL, USA). For normally distributed data, results are expressed as the mean ± standard deviation. The differences between groups were assessed using the t-test or one-way ANOVA. Rates were compared using the chi-square (χ 2 ) test. Correlations were determined by computing the Spearman Rank Correlation. The Delong test was used to compare the performance of two ROC curves. A binary logistic regression model was used to calculate the predicted values, where age, gender, and APOE status were used as covariates. P < 0.05 was considered to indicate a statistically signi cant difference. Table 1 depicts the characteristics of participants. There were no differences in the ages or ratios of males/females among the SCD, aMCI, AD, VaD and control groups in both the discovery stage and the validation stage. The percentages of APOE ε4, Mini-Mental State Examination (MMSE), and Clinical Dementia Rating (CDR) scores were signi cantly different between aMCI patients and controls, AD patients and controls, VaD patients and controls, and AD and aMCI patients (all P < 0.05). * P < 0.05 compared to controls; † P < 0.05 compared to SCD; § P < 0.05 compared to aMCI.

Identi cation of exosomes
The results of TEM and particle size analysis show that the shape and size of the extract were consistent with the characteristics of exosomes ( Fig. 2A, B).
Secondly, western blot results show that Alix and L1CAM in the stage 1 immune capture product were highly expressed but not detected in supernatants (Fig. 2C, D). Third, the results of ELISA showed that in the stage 1 and stage 2 immune capture, the levels of NCAM and amphiphysin 1 in the captured product were higher than that of the corresponding supernatant respectively (all P < 0.05) (Fig. 2E, F).
The concentration of exosomes in all specimens were standardized using CD81 test results before testing for other proteins. Overall, there was no signi cant difference in CD81 among the groups (all P < 0.05) (Fig. 2G).
More important, after 2 hours of cultivation, NCAM levels in WBCs and RBCs did not decrease signi cantly (all P > 0.05), the NCAM levels in WBC and RBC medium did not increase signi cantly (all P > 0.05) (Fig. 3). The result suggested that the content of NCAM in RBCs and WBCs were low (~ 30 µg/L), RBC and WBC almost did not release NCAM into the medium, suggested that specimen storage and pre-processing will not affect subsequent test results.
3.3 Levels of Aβ42, Aβ 42/40 , T-tau, P-T181-tau, NfL, and miR-29c-3p in NCAM singlelabeled exosomes of plasma We rst use the specimens from the discovery stage to detect the NCAM-labeled exosomal proteins extracted in the rst stage of the exosome extraction experiment. Exosomes of this component are generally considered neuron-derived exosomes. The results showed that the levels of Aβ42, Aβ 42/40 , Tau and P-T181-tau in plasma exosomes of the aMCI and AD groups were signi cantly higher than those of the SCD group, control group and VaD group, and the AD group was higher than the aMCI group. The levels of plasma exosomal Aβ42, Tau and P-T181-tau in the SCD group were slightly higher than those in the control group and the VaD group, but there was no signi cant difference. The level of NfL in plasma exosomes in the aMCI and AD groups was signi cantly higher than that in the SCD group, control group and VaD group, and the AD group was higher than that in the aMCI group. The NfL level of plasma exosomes in the SCD group was slightly higher than that in the control group and the VaD group, but there was no signi cant difference. The levels of miR-29c-3p in plasma exosomes in the aMCI and AD groups were signi cantly higher than that in the SCD, control, and VaD group; while the AD group was higher than the aMCI group. The level of plasma exosomes miR-29c-3p in the SCD group was slightly higher than that in the control group and the VaD group, but there was no signi cant difference. The change trend of the test results of the specimens in the veri cation stage is consistent with the discovery stage (Fig. 4).
3.4 Levels of Aβ42, Aβ 42/40 , T-tau, P-T181-tau, NfL, and miR-29c-3p in NCAM/amphiphysin 1 dual-labeled exosomes in blood and CSF We rst used the specimens from the discovery stage to detect the NCAM/amphiphysin 1 dual-labeled exosomal proteins extracted in the second stage of the exosome extraction experiment. The results showed that the levels of Aβ 42/40 , Tau, and P-T181-tau in plasma exosomes of the aMCI and AD groups were signi cantly higher than those of the SCD group, control group and VaD group, and the AD group was higher than the aMCI group. The levels of plasma exosomes Aβ 42/40 , Tau, and P-T181-tau in the SCD group were slightly higher than those in the control group and the VaD group, but there was no signi cant difference. There was no signi cant difference in the level of exosomal Aβ40 among the groups, and only the AD group decreased slightly. The level of NfL in plasma exosomes in the aMCI and AD groups was signi cantly higher than that in the SCD group, control group and VaD group, and the AD group was higher than that in the aMCI group. The NfL level of plasma exosomes in the SCD group was slightly higher than that in the control group and the VaD group, but there was no signi cant difference. The change trend of the test results of the specimens in the veri cation stage was consistent with the discovery stage. In all disease groups and control groups, the levels of Aβ42 and Aβ 42/40 in NCAM/amphiphysin 1 dual-labeled exosomes were higher than those of NCAM singlelabeled exosomes, considering that the concentration of exosomes in each group has been standardized by CD81.
The levels of Aβ42 and miR-29c-3p in plasma NCAM/amphiphysin 1 dual-labeled exosomes of the SCD, aMCI and AD groups were signi cantly higher than those of the control and VaD groups, the AD group was higher than the aMCI group, and the aMCI group was higher than the SCD group (Fig. 5).
Of interest, the increase ratio of NCAM/amphiphysin 1 double-labeled exosomes Aβ42 and miR-29c-3p in all disease groups was higher than that of T-tau, P-T181-tau, and NfL (Fig. 6).

Correlation analysis between CSF and plasma exosomal biomarkers
We conducted a correlation analysis of exosomes and CSF biomarkers and found that the levels of Aβ42, P-T181-tau, as well as miR-29c-3p in neuron-derived exosomes were highly correlated with their levels in CSF in all groups (all P < 0.05), while the level of NfL in neuron-derived exosomes has no correlation with CSF (all P > 0.05). This correlation was also observed in NCAM/amphiphysin 1 dual-labeled exosomes, and the correlation coe cients were even higher (Fig. 7).

Diagnostic power analysis of plasma exosomal biomarkers
We used the ROC curve to analyze the diagnostic performance of Aβ42, Aβ 42/40 , T-tau, P-T181-tau, NfL, and miR-29c-3p in plasma NCAM single-labeled (Fig. 8, Table 2) and NCAM/amphiphysin 1 dual-labeled (Fig. 9, Table 3 and miR-29c-3p had potential advantages in the diagnosis of SCD.   experiments and clinical studies have shown that with the progress of AD, the level of miR-29c-3p in cerebrospinal uid and peripheral blood exosomes keep rising, which suggests that the decrease of miR-29c-3p in neuronal cells may be due to excessive miR-29c-3p being transported from neurons to cerebrospinal uid and peripheral blood. The pathogenesis of AD is diverse. At present, the mainstream ones include amyloid theory, genetic theory, tau protein theory and lipid metabolism theory. In the theory of lipid metabolism, apolipoprotein E4 (ApoE4) occupies an important position, which plays a role in both amyloid production and tau protein phosphorylation [15]. Similar to ApoE4, amphiphysin 1 is widely distributed in various tissues and organs including brain tissue [27]. It can transport cholesterol and other lipids from the cell to the extracellular environment on the basis of consuming adenosine triphosphate, which is important in lipid metabolism. Studies have shown that amphiphysin 1 can interact with ApoE4 and act as a component of the amyloid clearance channel, so the change of amphiphysin 1 expression may be involved in the occurrence and development of AD [28]. The main modeling principle of APP/PS1 model mice is to promote the excessive deposition of amyloid. Therefore, our research group found that the signi cant changes in the expression of amphiphysin 1 in exosomes may be caused by abnormal amyloid metabolism. As a membrane protein, amphiphysin 1 also may be actively loaded on the exosomal membrane by the cell during the exosome production stage and secreted to the outside of the cell. There are multiple ways of Aβ elimination, such as endocytosis, proteases degradation, lymphatic drainage, CSF ow and active transportation across the BBB. Soluble Aβ also can be actively transported into BBB endothelial cells via amphiphysin 1, ApoE and LRP1 pathway, and be degraded in endothelial cells or pass through BBB to the peripheral blood [28,29]. The neuronal parenchymal cells are redistributed by the amphiphysin 1 transporter, which is considered to be the prime modulator of cholesterol homeostasis in the brain.
Current study showed that the diagnostic e cacy of miR-29c-3p of peripheral blood NCAM single-labled exosomes for AD and aMCI is similar to that of Aβ42, but their diagnostic value for SCD was limited. However, miR-29c-3p in NCAM/amphiphysin 1 dual-labeled exosomes of peripheral blood not only has better diagnostic performance for AD and aMCI, but also has acceptable diagnostic performance for SCD, which is not available in peripheral blood exosomes Aβ42.
In addition, for T-tau, P-T181-tau and NfL, there was no signi cant difference in the levels of NCAM single-labeled exosomes and NCAM/amphiphysin 1 double-labeled exosomes, as well as the diagnostic e ciency of AD diagnosis and staging, which suggested that there is no need to detect NCAM/amphiphysin 1 dual-labeled exosomal T-tau, P-T181-tau and NfL for diagnosis and staging of AD. These data indicate that blood neuron-derived exosomes are ideal biomarker carriers for AD screening, especially for large-scale population screening.
In summary, this study rst provided a method for sorting speci c surface marker exosomes using a two-step method; and veri ed that peripherald neuronalderived exosomal Aβ42, Aβ 42/40 , T-tau, P-T181-tau and miR-29c-3p may re ect AD pathological changes in the brain and therefore have the capacity to diagnose AD. The detection of NCAM/amphiphysin 1 dual-labeled exosomal molecules show a better potential diagnostic value for SCD than NCAM singlelabeled exosomes. However, these ndings need further con rmation in longitudinal studies.

Declarations
Ethics approval The experimental protocol was approved by the Ethics Committees of participating centers.

Consent to participate
Written informed consents were obtained from all participants or their legal guardians.

Consent for publication
Not applicable Authors' contributions Ying Li, Ming Xia, Shuang Meng, and Di Wu tested patient specimens and analyzed the data. Ying Li performed the cell culture and was a major contributor in writing the manuscript. Sihai Ling and Xiali Chen participated in data analysis. Chengeng Liu mainly participated in the experimental design. All authors read and approved the nal manuscript.

Con icts of Interest
The authors declare that they have no con icts of interest.

Data Availability
The data used to support the ndings of this study are available from the rst author or corresponding author upon request.
Acknowledgments    The fold changes of biomarkers. A shows the ratio of each biomarker relative to the control group in NCAM single-labeled exosomes (NE) and NCAM/amphiphysin 1 dual-labeled exosomes (NAM) from VaD, SCD, aMCI, and AD. B shows the ratio of NAE to NAM.