Abstract

   Both Alzheimer's disease (AD) and osteoporosis (OP) are common
   age‐associated degenerative diseases and are strongly correlated with
   clinical epidemiology. However, there is a lack of clear pathological
   relationship between the brain and bone in the current understanding.
   Here, it is found that young osteocyte, the most abundant cells in
   bone, secretes extracellular vesicles (OCY^Young‐EVs) to ameliorate
   cognitive impairment and the pathogenesis of AD in APP/PS1 mice and
   model cells. These benefits of OCY^Young‐EVs are diminished in aged
   osteocyte‐derived EVs (OCY^Aged‐EVs). Based on the self‐constructed
   OCY‐EVs tracer transgenic mouse models and the in vivo fluorescent
   imaging system, OCY‐EVs have been observed to be transported to the
   brain under physiological and pathological conditions. In the
   hippocampal administration of Aβ40 induced young AD model mice, the
   intramedullary injection of Rab27a‐shRNA adenovirus inhibits
   OCY^Young‐EVs secretion from bone and aggravates cognitive impairment.
   Proteomic quantitative analysis reveals that OCY^Young‐EVs, compared to
   OCY^Aged‐EVs, enrich multiple protective factors of AD pathway. The
   study uncovers the role of OCY‐EV as a regulator of brain health,
   suggesting a novel mechanism in bone‐brain communication.

   Keywords: aging, Alzheimer's disease, extracellular vesicles,
   osteocyte, osteoporosis
     __________________________________________________________________

   It is found that the osteocyte‐derived extracellular vesicles (OCY‐EVs)
   isolated from young osteocytes can ameliorate cognitive impairment and
   pathogenies of AD, but not OCY‐EVs isolated from aged osteocytes.
   OCY‐EV can transfer to the brain under physiological and pathological
   conditions. The study uncovers the role of OCY‐EVs as a regulator of
   brain, suggesting a novel mechanism in bone‐brain communication.

   graphic file with name ADVS-9-2105316-g005.jpg

1. Introduction

   Alzheimer's disease (AD) is a progressive neurodegenerative disease
   that is mainly characterized by a progressive decline in memory and
   cognitive function.^[ [62]^1 ^] The prevalent features of AD
   pathogenesis are the appearance of β‐amyloid (Aβ) plaques and
   neurofibrillary tangles, which cause microglial activation, synaptic
   deficiency, and neuronal loss.^[ [63]^1 , [64]^2 ^] During aging,
   increased Aβ formation and weakened Aβ cleavage capacity lead to toxic
   product accumulation, which crucially contributes to the development of
   AD.^[ [65]^1 , [66]^3 ^] Osteoporosis (OP) is a systemic skeletal
   disease characterized by low bone mass and microarchitectural
   deterioration of bone tissue, with a consequent increase in bone
   fragility and susceptibility to fracture.^[ [67]^4 ^] Skeletal aging
   leads to changes in the cellular function of bone tissue and an
   imbalance in bone remodeling between bone resorption and formation.^[
   [68]^5 ^] Both AD and OP are common degenerative diseases related to
   aging and strongly correlate with clinical epidemiology.^[ [69]^6 ^]
   Several studies have demonstrated that elderly individuals with
   dementia present a lower bone mineral density (BMD) and a higher
   frequency of OP at the hip, and individuals with low BMD are also
   associated with an increased risk of AD.^[ [70]^7 ^] AD and OP also
   share various potential mechanisms, such as immunosuppression, lack of
   physical activity, low levels of vitamin D and K, and the presence of
   the apolipoprotein E (ApoE) 4 allele.^[ [71]^6 , [72]^8 ^] However,
   there is a lack of clear pathological relationship between AD and OP in
   the current understanding.

   Osteocytes (OCYs) are the most abundant cells (almost 90–95%) in the
   bone and play a vital role in bone homeostasis.^[ [73]^9 ^] OCYs,
   similar to neurons, present abundant cilia and cytoplasmic processes
   (≈40–100 per cell) radiating through the mineralized matrix to connect
   with adjacent cells.^[ [74]^10 ^] Considering the vast numbers and
   network‐like structures of OCY, it is possible that OCYs work like
   analogous endocrine cells and release secretory factors to regulate
   bone and distant extraskeletal (nonbone) organs. OCYs have been
   discovered to secrete the receptor activator for nuclear factor‐κB
   ligand (RANKL), fibroblast growth factor‐23 (FGF‐23), and leptin to
   regulate bone metabolism, kidney, and brain functions through the
   circulation, respectively.^[ [75]^11 ^]

   Recently, it has been recognized that extracellular vesicles (EVs) are
   important factors to mediate the exchange of intercellular biological
   signals.^[ [76]^12 ^] OCYs reside in the lacunae within the
   hard‐mineralized bone matrix; moreover, bone matrix vesicles are
   anchored within the matrix and serve as the initial mineralization
   site.^[ [77]^13 ^] These phenomena suggest that the anchored vesicles
   are probably osteocyte‐derived EVs (OCY‐EVs). The secreted OCY‐EVs are
   released from the bone matrix during the process of bone resorption or
   transferred to distant organs through the intraskeletal vasculature.
   For instance, we previously reported that OCY‐EVs accumulated in the
   prostate, affecting the development of benign prostatic hyperplasia.^[
   [78]^14 ^] These reports, together with the evidence‐based clinical
   relevance of AD and OP, suggest that the exchange of signals in the
   bone‐brain axis is likely to be mediated by OCY‐EVs.

   Herein, we confirmed the clinical relevance between bone and brain
   health. The transport of bone‐derived OCY‐EVs was observed in brain
   tissue using self‐constructed OCY‐EVs tracing mice. Then, we isolated
   and characterized OCY‐EVs from primary osteocyte derived from 2‐ or
   16‐month‐old mice (OCY^Young‐EVs or OCY^Aged‐EVs). In vitro, we
   compared the effects of OCY^Young‐EVs and OCY^Aged‐EVs on Aβ pathology
   and apoptosis of nerve cells transfected with AD pathogenic genes. In
   vivo, we investigated whether OCY^Young‐EVs could ameliorate cognitive
   impairment and the pathogenesis of AD in early or later stages of
   APP/PS1 AD model mice. Moreover, in Aβ40‐induced AD model mice, we
   tested whether the intramedullary injection of Rab27a‐shRNA adenovirus
   could inhibit EVs secretion from bone and aggravate cognitive
   impairment. Furthermore, potential protective factors for AD in OCY‐EVs
   were screened by proteomic quantitative analysis. This work showed that
   young bone‐derived OCY‐EVs played a protective role in AD through the
   bone‐brain axis.

2. Results

2.1. Bone Health Strongly Correlated with Brain Health in Clinical
Epidemiology

   To confirm the relationship between bone quality and cognitive
   impairment, bone mineral density (BMD) and Mini‐Mental State
   Examination (MMSE) tests were conducted in sex‐ and age‐matched
   subjects. The demographic and clinical characteristics of cognitively
   normal subjects (CN, n = 195) and AD patients (AD, n = 117) were
   summarized in Figure [79]1A. The age and gender distribution of the two
   cohorts were matched (P > 0.05; Figure [80]1A). The mean MMSE score was
   28.69 ± 1.44 in CN subjects and 15.50 ± 6.73 in AD patients (P = 0.000;
   Figure [81]1A), which was used to clinically estimate the cognitive
   impairment.^[ [82]^15 ^] According to World Health Organization (WHO)
   criteria, the bone density assessment was reported as T‐Scores.^[
   [83]^16 ^] The T‐Score value of AD patients (−2.58 ± 1.20) presented
   markedly lower than the CN subjects (−2.15 ± 1.32, P = 0.005). In the
   clinic, a T‐Score > −1.0, −2.5 < T‐Score ≤ −1.0, or T‐Score ≤ −2.5 was
   referred to as normal, osteopenia, or osteoporosis, respectively.^[
   [84]^16 ^] The frequency of bone loss (T‐Score ≤ −1.0) in AD patients
   was 91.5%, which was significantly higher than that in CN subjects
   (81.5%, P = 0.017; Figure [85]1A). The distributions of the three
   categories of BMD in CN subjects and AD patients were illustrated in
   Figure [86]1B. The percentage of AD patients with osteoporosis (57.3%)
   was higher than that of CN subjects (44.6%); nevertheless, the
   proportion of normal BMD in the AD (8.5%) and CN (18.5%) was reversed
   (P < 0.05; Figure [87]1A,B). Furthermore, there was a positive
   correlation between MMSE and T‐Score (Spearman correlation coefficient
   = 0.163, P = 0.007; Figure [88]1C).

Figure 1.

   Figure 1
   [89]Open in a new tab

   Bone Health Strongly Correlated with Brain Health in Clinical
   Epidemiology. A) Basic information and BMD of CN and AD subjects. a,b)
   each superscript letter denoted a subset of intervention categories
   whose column proportions did not differ significantly from each other
   at the 0.05 level (Bonferroni method). B) The distributions of three
   categories of BMD (Osteoporosis, Osteopenia, and Normal) in CN and AD
   subjects. C) Scatter plot of T‐Score and MMSE showing the relationship
   between BMD and cognition. D) Basic information and BMD of CN and AD
   subjects separated by gender. a,b) each superscript letter denoted a
   subset of intervention categories whose column proportions did not
   differ significantly from each other at the 0.05 level (Bonferroni
   method). The distributions of three categories of BMD (Osteoporosis,
   Osteopenia, and Normal) in E) male or F) female subjects.
   Representative western blot images of DMP1 and TSG101, and
   quantification of DMP1 intensity in plasma EVs from gender‐ and
   age‐matched CN subjects and AD patients with G,J) Osteoporosis, H,K)
   Osteopenia, or I,L) Normal. TSG101 was used as a marker of EVs. Each
   gel contained a calibration sample to adjust the comparison of two
   groups. n = 7 per group. M) Analysis of DMP1 intensity after correction
   for all samples in (G‐I). n = 21 per group. Data were presented as mean
   ± SD. For panel (A) and (D): Chi‐Square test for Gender, T‐Score
   classifications, and BMD classifications; unpaired, two‐tailed
   Student's t‐test for Age, MMSE, BMI, T‐Score value. For panel (C):
   Spearman Correlation Analysis. For panel (J‐L): unpaired, two‐tailed
   Student's t‐test. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P
   < 0.0001.

   Due to the incidence of AD and OP being different in male and female
   subjects, we then analyzed the data in a stratified manner by gender.
   The demographic and clinical characteristics of male and female
   subjects were summarized in Figure [90]1D. The BMI of AD patients
   (Male: 22.53 ± 2.80; Female: 21.54 ± 3.36) was significantly lower than
   that of CN subjects (Male: 24.47 ± 2.99, P = 0.002; Female: 23.67 ±
   3.04, P = 0.000). The T‐Score value of AD patients (Male: −1.72 ± 1.10;
   Female: –2.98 ± 1.02) presented markedly lower than that of CN subjects
   (Male: −1.23 ± 1.13, P = 0.036; Female: −2.64 ± 1.14, P = 0.034). In
   male subjects, the frequency of bone loss (T‐Score ≤ −1.0) in AD
   patients was 78.4%, which was significantly higher than that in CN
   subjects (58.8%, P = 0.044; Figure [91]1D). However, in female
   subjects, the frequency of bone loss (T‐Score ≤ −1.0) in AD patients
   was slightly higher than CN subjects in trend (97.5% and 93.7%
   respectively, P = 0.364; Figure [92]1D). The distributions of the three
   categories of BMD of male and female subjects were illustrated in
   Figure [93]1E,F. The percentage of AD patients with osteoporosis (Male:
   24.3%, Female: 72.5%) was higher than that of relevant CN subjects
   (Male: 13.2%, Female: 61.4%). These clinical epidemiologist studies
   suggested that individuals with lower T‐Score tended to exhibit poorer
   cognitive function.

   Changes in bone matrix properties and osteocyte dysregulation occurred
   with aging.^[ [94]^17 ^] Accompanied by bone loss and bone remodeling,
   the bone matrix and contents were absorbed and released into the
   surroundings and circulation.^[ [95]^18 ^] Among the extracellular
   secretions, EVs usually participated in distant interorgan
   communication.^[ [96]^12 , [97]^19 ^] Thus, plasma EVs from gender‐ and
   age‐matched AD and CN subjects were isolated to compare the different
   compositions. Western blot was used to identify the EVs classical
   marker of Tumor Susceptibility Gene 101 (TSG101), and measure the
   amount of osteocyte‐specific protein Dentin Matrix Protein 1 (DMP1) in
   plasma EVs. As shown in Figure [98]1D–F and quantified in
   Figure [99]1G–I, AD patients with osteoporosis (Figure [100]1D,G) or
   osteopenia (Figure [101]1E,H) had higher levels of OCY‐EVs in plasma
   than the related controls (n = 7 per group). However, there was no
   difference in patients and controls with normal BMD (Figure [102]1F,I,
   n = 7 per group). Representative plasma EVs from osteoporosis,
   osteopenia, and normal BMD were also displayed in one gel to avoid
   operation error (Figure [103]S1A,B, Supporting Information).
   Collectively, these results showed that patients with AD (n = 21) had
   higher levels of OCY‐EVs in plasma than the controls (n = 21;
   Figure [104]1J).

2.2. OCY‐EVs Were Transported to the Brain under Physiological and
Pathological Conditions

   To verify the participation of OCY‐EVs in the communication between
   bone and brain, we first observed whether OCY‐EVs could be transported
   to the brain under physiological conditions. OCY‐EVs tracer transgenic
   mice were constructed by our group. The tetraspanin protein CD63 was
   enriched in the EVs membrane and considered as a ubiquitous marker for
   almost all EVs.^[ [105]^20 ^] To generate EV reporter mice (Cd63^fl/+
   mice), the loxp‐cc‐mCherry‐polyA‐loxp‐cc‐eGFP expression frame was
   knocked in the termination codon in exon 8 of the Cd63 gene by
   homologous recombination CRISPR/Cas9 technology (Figure [106]2A).
   Before mating with Cre tool mice, the mCherry fusion protein was
   expressed. After mating with Cre tool mice, the eGFP fusion protein was
   conditionally expressed in specific cells. Sclerostin (SOST) was a
   well‐known specific marker of osteocytes.^[ [107]^9 ^] The Sost gene
   start codon site was knocked in CreERT2‐Wpre‐pA expression frame to
   construct Sost^CreERT2 mice, which expressing a tamoxifen‐activated Cre
   recombinase in osteocytes under the control of the sclerostin promoter
   (Figure [108]2A). The Sost^CreERT2 mice were mated with Cd63^fl/+ mice
   to obtain CD63 conditional mutagenic mice. The Sost^CreERT2;Cd63^fl/+
   mice (genotyping identified in Figure [109]S2A,D, Supporting
   Information) should transcribe eGFP in osteocyte and OCY‐EVs, and
   mCherry in other cells and secreted EVs (such as neurons and microglia;
   Figure [110]2B). As shown in the cortex area of brain sections in
   Figure [111]2C, mCherry red fluorescence signals were observed in both
   Sost^CreERT2;Cd63^fl/+ , and Cd63^fl/+ littermate control mice, while
   green fluorescence of eGFP was only observed in Sost^CreERT2;Cd63^fl/+
   mice. To further confirm these results, Dmp1^Cre mice expressing Cre
   recombinase under the control of the murine Dmp1 promoter were used as
   an alternative tool to generate osteocyte specific targeted mutants.^[
   [112]^21 ^] The obtained Dmp1^Cre;Cd63^fl/+ mice (genotyping identified
   in Figure [113]S2B,D, Supporting Information) showed similar eGFP and
   mCherry expression in the cortex area of brain sections (Figure
   [114]S3A,B, Supporting Information). Although Dmp1^Cre mice were widely
   used in osteocyte‐specific targeting, previous evidence had revealed
   that it also targeted cells in the mouse cerebellum and hindbrain.^[
   [115]^22 ^] To prove the osteocytic source of these green fluorescence
   signals, the brain sections were also stained with SOST antibody.
   SOST‐positive staining mostly overlapped with the eGFP signals in
   Dmp1^Cre;Cd63^fl/+ mice (Figure [116]S3B, Supporting Information).
   These results suggest that OCY‐EVs (SOST^+, DMP1^+, and CD63‐eGFP^+)
   could reach the brain under physiological conditions.

Figure 2.

   Figure 2
   [117]Open in a new tab

   OCY‐EVs were Transported to the Brain under Physiological and
   Pathological Conditions. A) Schematic diagram of the construction of
   Sost^CreERT2;Cd63^fl/+ mice model and B) its fluorescent appearances of
   EVs. C) Representative IF images of brain tissue sections of
   Sost^CreERT2;Cd63^fl/+ mice and Cd63^fl/+ littermate control mice. DAPI
   (blue), mCherry (red), and eGFP (green). The merged photos were
   overlaid on the right side, and the white square indicated the
   amplification regions. Scale bar: 10 µm. Representative D) cellular
   morphologies, E) western blot images of DMP1 and SOST expression, F)
   western blot images and G) quantification of P21 and P16 expression in
   OCY^Young and OCY^Aged. Scale bar: 100 µm. β‐actin was used as loading
   control. OB, Osteoblast; BMM, bone marrow macrophage. n = 3 per group.
   Representative H) western blot images of osteocyte markers (SOST and
   DMP1) and EVs classical markers (TSG101 and Flotillin‐1), I)
   morphological analysis, and J) diameter distribution of OCY^Young‐EVs
   and OCY^Aged‐EVs. Scale bar: 100 nm. K) Representative ex vivo
   fluorescent images and L) quantification of the fluorescent signals of
   brain and lower limb bone from WT and APP/PS1 AD mice treated with
   DiR‐labeled OCY‐EVs or solvent by intramedullary administration. n = 3
   per group. Representative ex vivo fluorescent images and quantification
   of the fluorescent signals of M) brain and N) lower limb bone from WT
   and APP/PS1 AD mice treated with DIR‐labeled OCY^Young‐EVs or
   DIR‐labeled OCY^Aged‐EVs by intravenous injection. n = 3 per group, *
   OCY^Young‐EVs WT (blue line) versus OCY^Aged‐EVs WT (green line), #
   OCY^Young‐EVs AD (red line) versus OCY^Aged‐EVs AD (purple line). Data
   were presented as mean ± SD. For panel (G), (M), and (N): unpaired,
   two‐tailed Student's t‐test. For panel (L): one‐way ANOVA or the
   Kruskal–Wallis test with Bonferroni post hoc correction. */#P < 0.05,
   **/##P < 0.01, ***/###P < 0.001, ****/####P < 0.0001.

   Next, we explored whether these OCY‐EVs could be transported from bone
   tissue to AD‐impaired brain tissue under pathological conditions by
   intramedullary injection of OCY‐EVs into APP/PS1 mice. Primary OCYs
   were derived from 2‐month‐old (OCY^Young) and 16‐month‐old (OCY^Aged)
   C57BL/6 mice. The biological features of osteocytes were characterized
   by cellular morphologies (Figure [118]2D), and DMP1 and SOST expression
   (Figure [119]2E).^[ [120]^9 ^] The expression of P21 and P16 proteins
   was measured to evaluate the senescence state of OCY
   (Figure [121]2F,G). Next, OCY‐EVs were isolated from the cell culture
   medium by ultracentrifugation and characterized with osteocyte markers
   (DMP1 and SOST) and EVs markers (TSG101 and Flotillin‐1) by western
   blot (Figure [122]2H). Moreover, transmission electron microscopy (TEM;
   Figure [123]2I) and nanoparticle tracking analysis (NTA;
   Figure [124]2J) results showed that OCY‐EVs displayed cup‐shaped
   morphologies with diameters of 127.1 ± 41.6 nm (OCY^Young‐EVs) or
   128.9 ± 45.8 nm (OCY^Aged‐EVs), consistent with previous reports.^[
   [125]^12 ^] To monitor the organ distributions, OCY^Young‐EVs were
   labeled with DiR near‐infrared fluorescent dye and then intramedullary
   (the femoral bone marrow) administered into wild type (WT) or APP/PS1
   AD mice (genotyping identified in Figure [126]S2C,D, Supporting
   Information). After 24 h, ex vivo fluorescence imaging revealed strong
   fluorescence signals in the brain tissues, indicating that OCY‐EVs
   could enter the brains of WT and AD mice (Figure [127]2K,L). DiR
   fluorescent signals were also detected in the heart, liver, spleen,
   lung, and kidney (Figure [128]S4A,B, Supporting Information). To
   further detect OCY‐EVs infiltration inside the tissues, red fluorescent
   dye DiL labeled OCY^Young‐EVs (intramedullary administered) were also
   observed in tissue sections of bone, brain, and other organs of WT and
   AD mice by fluorescence microscopy (Figure [129]S4C,D, Supporting
   Information).

   Furthermore, to compare the different distribution dynamics of
   OCY^Young‐EVs and OCY^Aged‐EVs in WT and AD mice, ex vivo fluorescent
   imaging of the brain and lower limbs was performed at 6, 24, and 48 h
   after intravenous administration of DiR‐labeled OCY‐EVs. Compared with
   OCY^Aged‐EVs, more OCY^Young‐EVs entered the brain and bone in both WT
   and AD mice (Figure [130]2M,N) but not always in the other tissues
   (Figure [131]S5A,B, Supporting Information). As shown in
   Figure [132]2M, the fluorescence signals of OCY^Young‐EVs and
   OCY^Aged‐EVs in the brain decreased over time in both WT and AD mice.
   However, the fluorescence signals of OCY^Young‐EVs in the brain seemed
   to decline at a slower rate in AD mice than in WT mice
   (Figure [133]2M). Combined with the DiL intensity in the CA3 region of
   brain sections (Figure [134]S6A,D, Supporting Information), it
   suggested that OCY^Young‐EVs might persist longer in the brains of AD
   mice.

2.3. OCY^Young‐EVs, but not OCY^Aged‐EVs, Rescued the Aβ Pathology and
Apoptosis of AD Model Cells

   We next asked whether OCY^Young‐EVs and OCY^Aged‐EVs were able to
   directly interact with neural cells. In vitro, the human neuroblastoma
   cell line SH‐SY5Y and the mouse hippocampal neuronal cell line HT22
   were adopted as normal neural model cells.^[ [135]^23 ^] They were also
   transfected with vectors carrying the APPswe fragment to generate
   SH‐SY5Y‐APPswe and HT22‐APPswe AD model cells. SH‐SY5Y‐APPswe and
   HT22‐APPswe cells highly expressed amyloid protein precursor (APP)
   mutant proteins and generated more pathogenic proteins (Figure
   [136]S7A–D, Supporting Information). The SH‐SY5Y and HT22 cells were
   transfected with control vectors. To evaluate the EVs internalized by
   AD model cells, OCY^Young‐EVs were labeled with PKH26 red fluorescent
   cell linker dye and the excess dye was disposed by ultrafiltration.
   PKH26‐labeled OCY^Young‐EVs were cocultured with SH‐SY5Y,
   SH‐SY5Y‐APPswe, HT22, or HT22‐APPswe cells for 12 h. After washing the
   free EVs, many red fluorescent signals were detected in the perinuclear
   region under confocal microscopy (Figure [137]3A,B), which showed that
   OCY‐EVs could be internalized by these normal and AD model cells.

Figure 3.

   Figure 3
   [138]Open in a new tab

   OCY^Young‐EVs, but not OCY^Aged‐EVs, Rescued the Aβ Pathology and
   Apoptosis of AD Model Cells. Internalization of the red fluorescent dye
   PKH26‐labeled OCY^Young‐EV or solvent by A) SH‐SY5Y and SH‐SY5Y‐APPswe
   cells, B) HT22 and HT22‐APPswe cells. Scale bar: 20 µm. ELISA analysis
   the concentration and ratio of Aβ42 and Aβ40 of C) SH‐SY5Y‐APPswe, D)
   SH‐SY5Y, E) HT22‐APPswe, and F) HT22 cells treated with OCY^Young‐EV,
   OCY^Aged‐EV, or solvent. n = 5 in (C), n = 6 in (D), or n = 3 in (E‐F).
   Representative fluorescent images and quantification of
   Hoechest‐positive cells (light blue) of G, H) SH‐SY5Y‐APPswe and
   SH‐SY5Y, I, J) HT22‐APPswe and HT22 cells treated with OCY^Young‐EV,
   OCY^Aged‐EV, or solvent. n = 6 per group. Scale bar: 50 µm. Data were
   presented as mean ± SD. All dot plots: one‐way ANOVA or the
   Kruskal–Wallis test with Bonferroni post hoc correction. * P < 0.05, **
   P < 0.01, *** P < 0.001, **** P < 0.0001.

   Excessive accumulation of β‐amyloid peptide (Aβ) as amyloid plaques and
   sequential neuronal cell apoptosis remained the primary neuropathologic
   criteria for AD.^[ [139]^24 ^] To assess the influence of OCY‐EVs on
   the process of AD pathology, extracellular Aβ accumulation and the
   apoptosis rate were detected in OCY^Young‐EV‐ or OCY^Aged‐EV‐treated
   SH‐SY5Y‐APPswe and HT22‐APPswe AD model cells. Aβ isoforms were
   composed of 39–43 amino acids, among which the 40‐residue Aβ isoform
   (Aβ40) was more abundant, and the 42‐residue Aβ isoform (Aβ42) was
   closely linked to the progression of AD.^[ [140]^25 ^] After treatment
   for 48 h, the concentrations of Aβ42 and Aβ40 in the medium supernatant
   of SH‐SY5Y‐APPswe, SH‐SY5Y, HT22‐APPswe, and HT22 cells were examined
   by enzyme‐linked immunosorbent assay (ELISA; Figure [141]3C–F).
   Compared with the solvent group, we observed a decreased concentration
   of Aβ42 and ratio of Aβ42/Aβ40 in the OCY^Young‐EV‐treated
   SH‐SY5Y‐APPswe (Figure [142]3C) and HT22‐APPswe cells (Figure [143]3E).
   However, Aβ42 and Aβ40 secretion were not affected in the OCY^Aged‐EVs
   groups. The SH‐SY5Y and HT22 cell groups presented a very low degree of
   Aβ42 accumulation, and no significant difference was observed after EVs
   intervention (Figure [144]3D,F). Finally, apoptotic neuronal cells were
   also detected by Hoechst staining. Compared with SH‐SY5Y and HT22
   cells, SH‐SY5Y‐APPswe and HT22‐APPswe cells exhibited high levels of
   apoptosis (Hoechst‐positive cells), which were markedly rescued by
   OCY^Young‐EVs (Figure [145]3G–J). Therefore, these results suggested
   that OCY^Young‐EVs might play a protective role in the pathogenesis of
   AD (Aβ pathology and apoptosis).

2.4. OCY^Young‐EVs, but not OCY^Aged‐EVs, Ameliorated Cognitive Impairment
and Aβ Pathology in the Early Stage of AD Model Mice

   To investigate the effects of OCY^Young‐EVs and OCY^Aged‐EVs on the
   process of AD in vivo, 4‐month‐old WT or APP/PS1 mice were
   intravenously administered equal amounts (2.4 × 10^6 particles per gram
   of body weight) of OCY^Young‐EVs, OCY^Aged‐EVs, or solvent once a week
   for eight weeks (Figure [146]4A). Four‐month‐old APP/PS1 mice began to
   exhibit Aβ deposition in the hippocampus and thus were considered as
   early stage AD model mice.^[ [147]^26 ^] After treatment, the Morris
   water maze (MWM), Y‐Maze, object location task (OLT), and novel object
   recognition task (NORT) were used to evaluate cognitive functions. The
   MWM and Y‐Maze tests are most commonly used to assess
   hippocampus‐dependent spatial learning and memory function.^[ [148]^27
   ^] As shown in Figure [149]4B–E, compared with the solvent‐ or
   OCY^Aged‐EV‐treated group, the behavior performance of
   OCY^Young‐EV‐treated APP/PS1 mice seemed to be improved in the escape
   latency time in the learning trials (Figure [150]4B, Day 2 and Day 3),
   the target quadrant occupancy (Figure [151]4C) and the platform
   crossing times (Figure [152]4D,E) in the following probe trials (Day
   6). In the Y‐Maze test, the time of the treated mice spent in the start
   arm, familiar arm, and novel arm (previously blocked) was recorded.
   OCY^Young‐EV‐treated APP/PS1 mice explored the novel arm significantly
   longer than the solvent‐ or OCY^Aged‐EV‐treated group (Figure [153]4F).
   Furthermore, OLT and NORT tests were commonly used as effective
   behavioral tasks to exploit the innate preference for the novelty to
   reveal memory for previously encountered objects.^[ [154]^28 ^] After
   OCY^Young‐EV intervention, the mice exhibited improved spatial learning
   (OLT test, which heavily relies on hippocampal activity;
   Figure [155]4G) and nonspatial learning of object identity (NORT test,
   which relies on multiple brain regions; Figure [156]4H). However, after
   OCY^Aged‐EV treatment, the mice maintained a comparable investigation
   time of moved or novel objects as the APP/PS1 control mice
   (Figure [157]4G,H).

Figure 4.

   Figure 4
   [158]Open in a new tab

   OCY^Young‐EVs, but not OCY^Aged‐EVs, Ameliorated Cognitive Impairment
   and Aβ Pathology in the Early Stage of AD Model Mice (4‐month‐old
   APP/PS1 mice). A) Experimental design. Treatment and detection time
   points were labeled. In the MWM test, B) the escape latency time, C)
   target quadrant occupancy, D) platform crossing times, and E)
   representative motion trajectories of different treated mice. The black
   arrow indicated the platform. F) Y‐Maze, G) OLT, and H) NORT
   performances of different treated mice. I) The levels and ratio of Aβ42
   and Aβ40 in the brain tissue homogenate of different treated mice. J)
   Representative IF staining of 6E10 antibody (Aβ) and K) quantification
   of amyloid plaques per mm^2 area, amyloid plaque size (µm^2), and 6E10
   staining intensity in the brain sections. Scale bar: 50 µm. L) Dual IF
   staining of Iba1 antibody (microglia) and 6E10 antibody (Aβ), and M)
   the ratio of Iba1/6E10 in the brain sections. Scale bar: 10 µm.
   Representative IF staining of N) synaptophysin (Syp) and P) mature
   neurons (Neu‐N), and quantification of O) relative Syp intensity and Q)
   Neu‐N positive cells per mm^2 in the cortex and hippocampus of the
   brain section. Scale bar: 10 µm (N, Cortex), 50 µm (N, Hippocampus; P,
   Cortex and Hippocampus). Solvent‐treated WT group, n = 4;
   solvent‐treated APP/PS1 group, n = 4; OCY^Young‐EV‐treated APP/PS1
   group, n = 6; OCY^Aged‐EV‐treated APP/PS1 group, n = 5. Data were
   presented as mean ± SD. All dot plots: one‐way ANOVA or the
   Kruskal–Wallis test with Bonferroni post hoc correction. * P < 0.05, **
   P < 0.01, *** P < 0.001, **** P < 0.0001.

   Since Aβ accumulation and induced nerve injury were the most recognized
   pathogenesis of AD and the dominant pathology in APP/PS1 mice, we
   subsequently investigated whether OCY^Young‐EVs could play a role in Aβ
   elimination and neuroprotection. First, the levels of soluble Aβ42 and
   Aβ40 in the supernatant from brain tissue homogenates were measured by
   ELISA. The concentrations of Aβ42 and the ratios of Aβ42/Aβ40 were
   significantly decreased in the OCY^Young‐EV‐treated group compared with
   the solvent‐ or OCY^Aged‐EV‐treated group (Figure [159]4I). The
   residual insoluble Aβ was also detected with the 6E10 antibody
   (recognizing the first 16 amino acids of the Aβ sequence) by western
   blot and presented the same trend (Figure [160]S8A,B, Supporting
   Information). Then, IF staining with 6E10 antibody was conducted to
   identify Aβ plaques in brain sections. Compared with solvent or
   OCY^Aged‐EVs, OCY^Young‐EVs treatment markedly reduced the number and
   size of Aβ plaques and 6E10 antibody staining intensity in the
   hippocampus (Figure [161]4J,K), suggesting that OCY^Young‐EVs
   ameliorated Aβ pathology in AD model mice. In addition, microglial
   cells commonly tended to phagocytose loosely organized Aβ to form
   dense‐core Aβ plaques, which served as a protective mechanism to
   prevent the dissemination of toxic preplaque Aβ oligomers throughout
   the brain.^[ [162]^29 ^] Dual IF staining of 6E10 and ionized
   calcium‐binding adaptor molecule 1 (Iba1; microglial marker) was
   conducted in different treated groups (Figure [163]4L). Although fewer
   microglia around plaques were observed in the OCY^Young‐EVs‐treated
   group (Figure [164]S8C, Supporting Information), the ratio of Iba1/6E10
   was higher than that in the solvent‐ or OCY^Aged‐EVs‐treated groups
   (Figure [165]4M). As a result, OCY^Young‐EV‐treated APP/PS1 mice
   revealed an increased fluorescence intensity of synaptophysin (Syp,
   Figure [166]4N,O) and a higher number of Neu‐N immunoreactive cells
   (Figure [167]4P,Q) in both the cortex and hippocampus, suggesting that
   synaptic deficits and neuronal survival were improved. Interestingly,
   consistent with the published phenomenon,^[ [168]^30 ^] APP/PS1 mice
   displayed a noteworthy decrease in trabecular bone mass, which was
   reversed by OCY^Young‐EV treatment (Figure [169]S8D,E, Supporting
   Information). Collectively, the results indicated that OCY^Young‐EVs
   ameliorated the behavioral performance and pathogenesis of early stage
   APP/PS1 mice.

2.5. OCY^Young‐EVs, but not OCY^Aged‐EVs, Rescued Cognitive Impairment and Aβ
Pathology in the Later Stage of AD Model Mice

   APP/PS1 mice had been reported to exhibit cognitive impairment at six
   months of age;^[ [170]^26 , [171]^31 ^] thus, we further explored
   whether OCY^Young‐EVs were able to reverse cognitive impairment and Aβ
   pathology. Six‐month‐old APP/PS1 mice were intravenously administered
   equal amounts (4.8 × 10^6 particles per gram of body weight) of
   OCY^Young‐EVs, OCY^Aged‐EVs, or solvent weekly for eight weeks (Figure
   [172]5A). In the OCY^Young‐EVs intervention group, the escape latency
   time in the learning trials (Figure [173]5B) and the target quadrant
   occupancy in the subsequent probe trials (Figure [174]5C) of the MWM
   test, and the time spent in the novel arm of the Y‐Maze test
   (Figure [175]5F) tended to be better than those in the other groups. In
   particular, the platform crossing times in the probe trials of the MWM
   test were significantly greater than those of solvent‐ or
   OCY^Aged‐EV‐treated mice (Figure [176]5D,E). Improved functional
   performance was also observed in OCY^Young‐EV‐treated APP/PS1 mice in
   both the OLT and NORT tests (Figure [177]5G,H). Although the levels of
   Aβ42 or Aβ40 and the ratio of Aβ42/Aβ40 were comparable among the
   different groups (Figure [178]5I), insoluble Aβ was dramatically
   decreased in the OCY^Young‐EV‐treated group compared with the
   solvent‐treated group (Figure [179]S9A,B, Supporting Information).
   Pathologically, OCY^Young‐EV treatment reduced the number of Aβ plaques
   and 6E10 antibody staining intensity in the brain sections of APP/PS1
   mice (Figure [180]5J,K). Dual IF staining of Iba1 and 6E10 revealed a
   comparable ratio of Iba1/6E10 staining intensities in the
   OCY^Young‐EV‐treated group (Figure [181]5L,M; Figure [182]S9C,
   Supporting Information). An increase in Syp intensity
   (Figure [183]5N,O) and a higher number of Neu‐N immunoreactive cells
   (Figure [184]5P,Q) in the cortex and hippocampus were observed in
   OCY^Young‐EV‐treated APP/PS1 mice. In contrast, the OCY^Aged‐EV‐treated
   group showed no significant difference in vitro and in vivo compared to
   the solvent group. As expected, higher percent bone volume of
   trabecular bone (Tb. BV/TV) and trabecular number (Tb. N) levels were
   also observed in OCY^Young‐EVs‐treated mice (Figure [185]S9D,E,
   Supporting Information). Together, these findings indicated that
   OCY^Young‐EVs rescued the behavioral performance and pathogenesis of
   later stage APP/PS1 mice.

Figure 5.

   Figure 5
   [186]Open in a new tab

   OCY^Young‐EVs, but not OCY^Aged‐EVs, Rescued Cognitive Impairment and
   Aβ Pathology in the Later Stage of AD Model Mice (6‐month‐old APP/PS1
   mice). A) Experimental design. Treatment and detection time points were
   indicated. In the MWM test, B) the escape latency time, C) target
   quadrant occupancy, D) platform crossing times, and E) representative
   motion trajectories of different treated mice. The black arrow
   indicated the platform. F) Y‐Maze, G) OLT, and H) NORT performances of
   different treated mice. I) The levels and ratio of Aβ42 and Aβ40 in the
   brain tissue homogenate of different treated mice. J) Representative IF
   staining of 6E10 antibody (Aβ), and K) quantification of amyloid
   plaques per mm^2 area, amyloid plaque size (µm^2), and 6E10 staining
   intensity in the brain sections. Scale bar: 50 µm. L) Dual IF staining
   of Iba1 antibody (microglia) and 6E10 antibody (Aβ), and M) the ratio
   of Iba1/6E10 in the brain sections. Scale bar: 10 µm. Representative IF
   staining of N) synaptophysin (Syp) and P) mature neurons (Neu‐N), and
   quantification of O) relative Syp intensity and Q) Neu‐N positive cells
   per mm^2 in the cortex and hippocampus of the brain section. Scale bar:
   10 µm (N, Cortex), 50 µm (N, Hippocampus; P, Cortex and Hippocampus).
   Solvent‐treated APP/PS1 group, n = 5; OCY^Young‐EV‐treated APP/PS1
   group, n = 5; OCY^Aged‐EV‐treated group, n = 4. Data were presented as
   mean ± SD. All dot plots: one‐way ANOVA or the Kruskal–Wallis test with
   Bonferroni post hoc correction. * P < 0.05, ** P < 0.01, *** P < 0.001,
   **** P < 0.0001.

2.6. Inhibition of EV Secretion from Bone Aggravated Cognitive Impairment in
Aβ40‐induced Young AD Model Mice

   The above results suggested that the isolated OCY^Young‐EVs improved
   the impaired brains of APP/PS1 mice. These outcomes prompted us to
   study the features of physiologically secreted OCY^Young‐EVs in vivo.
   In young mice, OCY^Young‐EVs were constantly secreted and transported.
   Rab27a was generally perceived as a regulator of EV biogenesis and
   secretion.^[ [187]^32 ^] Thus, the adenoviral entrapped Rab27a shRNA
   silencing vector system (Ad‐Rab27a‐shRNA) and the control vector
   (Ad‐Mock) were constructed in this study. Eight‐week‐old WT mice were
   subjected to bilateral intramedullary injection of Ad‐Rab27a‐shRNA to
   inhibit bone‐derived EV secretion (Figure [188]6A). As shown in
   Figure [189]6B, decreased expression of Rab27a was confirmed in the
   bone matrix and bone marrow following Ad‐Rab27a‐shRNA administration.
   IF staining of bone sections indicated that the intramedullary injected
   adenovirus (characterized with GFP, green) could infect the OCYs
   (characterized with SOST, purple; Figure [190]6C). As a result,
   markedly reduced OCY‐EVs were detected in the plasma EVs of
   Ad‐Rab27a‐shRNA‐treated mice (Figure [191]6D,E).

Figure 6.

   Figure 6
   [192]Open in a new tab

   Inhibition of EV Secretion from Bone Aggravated Cognitive Impairment in
   Aβ40‐induced Young AD Model Mice. A) Experimental design. Treatment and
   detection time points were labeled. B) qRT‐PCR analysis of Rab27a mRNA
   expression levels in bone matrix and bone marrow of Ad‐Rab27a‐shRNA or
   Ad‐Mock treated mice. n = 6 per group. C) Representative IF staining of
   GFP (Ad‐Rab27a‐shRNA; green), SOST (OCY; purple), and DAPI (blue) in
   cortical bone samples from different treated mice. Scale bar: 10 µm. D)
   Representative western blot images and E) quantification of DMP1
   intensity in plasma EVs from different treated mice. n = 6 per group.
   In the MWM test, F) the escape latency time, G) target quadrant
   occupancy, H) platform crossing times, and I) representative motion
   trajectories of different treated mice. The black arrow indicated the
   platform. J) Y‐Maze, K) OLT, and L) NORT performances of different
   treated mice. n = 10 per group. Data were presented as mean ± SD. For
   panel (B) and (E): unpaired, two‐tailed Student's t‐test. For panel
   (F), (G), (H), (J), (K), and (L): one‐way ANOVA or the Kruskal–Wallis
   test with Bonferroni post hoc correction. * P < 0.05, ** P < 0.01, ***
   P < 0.001, **** P < 0.0001.

   After one week, a certain amount of Aβ40 or solvent was administered
   into the bilateral hippocampus to mimic Aβ accumulation in this brain
   region using a high‐precision brain stereotaxic method (Figure [193]6A;
   Figure [194]S10, Supporting Information). Two weeks later, cognitive
   functions were assessed in Ad‐Mock/Solvent, Ad‐Mock/Aβ40,
   Ad‐Rab27a‐shRNA/Solvent, or Ad‐Rab27a‐shRNA/Aβ40 mice. In the MWM and
   Y‐Maze tests, spatial learning and memory were slightly impaired in the
   Ad‐Mock/Aβ40 and Ad‐Rab27a‐shRNA/Solvent groups (Figure [195]6F–J). The
   worst cognitive functional performance was observed in the
   Ad‐Rab27a‐shRNA/Aβ40 group, as determined by the longest escape latency
   time in the learning trials (Figure [196]6F, Day 2 and Day 3), the
   least target quadrant occupancy in the following probe trials
   (Figure [197]6G), the least platform crossing times (Figure [198]6H,I),
   and the shortest exploring time in the novel arm (Figure [199]6J).
   Furthermore, Ad‐Rab27a‐shRNA/Aβ40‐treated mice exhibited significantly
   decreased investigation times of moved or novel objects in the OLT and
   NORT tests (Figure [200]6K,L). Collectively, these results suggested
   that inhibition of OCY^Young‐EVs could accelerate severe cognitive
   impairment in young mice.

2.7. Multiple Functional Factors of Aβ Degradation and Mitochondrial Energy
Metabolism were Enriched in OCY^Young‐EVs

   Furthermore, to explore the mechanism of the protective role of
   OCY^Young‐EVs in AD pathway, global proteomics quantitative analysis of
   OCY^Young‐EVs and OCY^Aged‐EVs was performed using TMT labeling and
   LC–MS/MS. In total, 410 differently enriched proteins (higher than 1.5
   times, P < 0.05) were identified between OCY^Young‐EVs and OCY^Aged‐EVs
   (Table [201]S1, Supporting Information). Among these proteins, 310 and
   100 proteins were highly expressed in OCY^Young‐EVs and OCY^Aged‐EVs,
   respectively (Figure [202]7A). GO analysis was conducted to explore the
   biological interpretation and the involved biological processes.
   According to the functional enrichment results, 135 biological process
   (BP) terms, 104 cell component (CC) terms, and 83 molecular function
   (MF) terms were significantly enriched among the differentially
   expressed proteins (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5; Figure [203]7B).
   The most enriched terms were membrane and extracellular exosome terms,
   suggesting that OCY^Young‐EVs contained more membrane proteins and
   exosomal proteins (Figure [204]7B). Next, KEGG pathway enrichment
   analysis was performed to identify the most significantly enriched
   pathways (Figure [205]7C). Interestingly, most of them were related to
   neurodegenerative diseases. The top five enriched pathways were those
   involved in AD, Parkinson's disease (PD), Pathways of
   neurodegeneration‐multiple diseases, Prion disease, and Huntington's
   disease (Figure [206]7D). In the AD pathway, the highly enriched
   proteins could be roughly divided into two categories: Aβ degradation
   (4 proteins) and mitochondrial energy metabolism (38 proteins)
   (Figure [207]7D). The Mme gene encoded neprilysin (NEP: 3.30‐fold)
   could fully degrade Aβ peptides.^[ [208]^33 ^] Nicastrin (NCSTN;
   1.92‐fold) and Presenilin‐2 (PSEN2; 1.69‐fold), as components of
   β‐secretase, could continuously cleave the α‐secretase cleavage product
   and eventually form nontoxic products. Disintegrin and
   metalloproteinase domain‐containing protein 10 (ADAM10; 1.54‐fold), as
   the main component of α‐secretase, could regulate the alternative APP
   processing to prevent Aβ formation.^[ [209]^34 ^] They worked together
   to reduce the production and promote the degradation of Aβ. In
   addition, mitochondrial dysfunction was a fundamental pathological
   feature of AD because the accumulation of damaged neuronal mitochondria
   had been found in both clinical AD cases and in AD mouse models.^[
   [210]^35 ^] The higher ratio of mitochondrial proteins in OCY^Young‐EVs
   might help to provide extra energy for neurological function. In
   summary, the highly enriched Aβ degradation and functional factors of
   mitochondrial energy metabolism in OCY^Young‐EVs synergistically
   prevented Aβ deposition and neuronal damage.

Figure 7.

   Figure 7
   [211]Open in a new tab

   Multiple Functional Factors of Aβ Degradation and Mitochondrial Energy
   Metabolism were Enriched in OCY^Young‐EVs. A) Heatmap of the
   differentially enriched proteins between OCY^Young‐EVs and OCY^Aged‐EVs
   (more than 1.5‐fold, P < 0.05). The protein levels were normalized by
   Z‐Score. B) The top 30 enriched GO terms (BP, CC, and MF) of
   differentially expressed proteins between OCY^Young‐EVs and
   OCY^Aged‐EVs (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5). C) The network of top
   30 enriched KEGG pathway terms of differentially expressed proteins
   (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5). D) The network between enriched
   KEGG pathway (top 5) and genes. The fold change was used to show the
   ratio of protein expression levels between OCY^Young‐EVs and
   OCY^Aged‐EVs.

3. Discussion

   In our study, we present a novel mechanism underlying the correlation
   of OP and AD (Figure [212]8 ). Two significant findings involve the
   bone‐brain axis. First, signal exchange between bone and brain can be
   achieved by transferring functional EVs from osteocytes in the bone to
   cells in the brain. Second, OCY^Young‐EVs play a protective role in the
   AD pathway, while OCY^Aged‐EVs lose this function. The effects of
   OCY‐EVs during aging have been demonstrated to change due to the
   different sorting of EVs contents.

Figure 8.

   Figure 8
   [213]Open in a new tab

   Schematic diagram. OCY‐EVs as the significant bone source messengers
   rescued brain lesions by transferring protective factors in the
   bone‐brain axis.

   Among the clinical subjects in our study, the BMD of AD patients is
   lower than that of sex‐ and age‐matched CN subjects, which is
   consistent with previous studies.^[ [214]^7 ^] Cognitive function is
   positively correlated with bone density. However, a high prevalence of
   OP was observed in female subjects both in CN and AD groups, and there
   was no significant difference existed in comparison of either BMD or
   T‐Score classification between CN and AD. The average age of menopausal
   is around 51–52 years old,^[ [215]^36 ^] and all women aged 65 years or
   older should be screened for osteoporosis as estrogen deficiency in
   postmenopausal women results in an increased prevalence of OP.^[
   [216]^37 ^] Since the average age of female subjects in our cohort was
   67.68 years in the CN group and 68.23 years in AD patients, loss of
   estrogen protection became the major cause of OP. More younger female
   individuals, especially between 40 to 60 years old, should be collected
   to reanalyze the difference in future research work.

   Further exploration of plasma EVs has revealed that AD patients present
   more OCY‐EVs in plasma than CN subjects. This phenomenon may be due to
   the body compensation mechanism. The osteocytes try to release more
   OCY‐EVs to help the affected brain cells; unfortunately, OCY‐EVs
   derived from the elderly subjects lost their protective functions.
   Moreover, the OCY‐EVs tracer transgenic Sost^CreERT2;Cd63^fl/+ and
   Dmp1^Cre;Cd63^fl/+ mice have demonstrated that OCY‐EVs could reach the
   brain under physiological conditions. This connection under
   pathological condition is evaluated by the intramedullary injection of
   DiR‐labeled OCY‐EVs in APP/PS1 AD mice, which confirms that OCY‐EVs can
   be transferred from bone to the brain. The metabolic kinetics curve of
   OCY‐EVs in the brain indicates that more OCY^Young‐EVs enter the brain
   than OCY^Aged‐EVs, and OCY^Young‐EVs decline at a slower rate in the
   brains of APP/PS1 AD mice than in those of WT mice. Proteomic analysis
   data showed that OCY^Young‐EVs contain higher amounts of membrane and
   extracellular exosome proteins. These results probably explain why
   OCY^Young‐EVs more easily penetrate the blood‐brain barrier than
   OCY^Aged‐EVs.

   Subsequently, we find that OCY^Young‐EVs play a protective role in the
   pathogenesis of AD in vitro and in vivo. Decreased concentrations of
   Aβ42 and percentages of apoptotic cells are observed in
   OCY^Young‐EV‐treated SH‐SY5Y‐APPswe and HT22‐APPswe cells. After
   intervention with OCY^Young‐EVs, both early stage (4‐month‐old) and
   later stage (6‐month‐old) APP/PS1 AD model mice exhibit improved
   cognitive functions and fewer Aβ plaques, more synapses, and more
   neurons in the brain. However, the OCY^Aged‐EV‐treated group shows no
   significant difference in vitro and in vivo, suggesting that the
   protective effects of OCY‐EVs decline with age. In later stage AD mice,
   the intervention amount of OCY‐EVs is increased to a double dose, the
   protective effect of OCY^Young‐EVs is not as significant as the
   intervention in early stage AD mice. In OCY^Young‐EVs treated AD mice,
   although the benefits of OCY^Young‐EVs are observed in most of behavior
   and pathological tests, the small number of mice per group is a
   limitation in this study. Moreover, Ad‐Rab27a‐shRNA is designed to
   prevent the secretion of EVs. Aβ40 is used to induce young AD mice in
   8‐week‐old WT mice. After intramedullary treatment with
   Ad‐Rab27a‐shRNA, plasma OCY^Young‐EVs are reduced and the cognitive
   impairment of treated mice is aggravated. These results prove the
   effects of OCY^Young‐EVs in preserving cognitive functions. Previous
   studies have reported that mesenchymal stem cell‐derived extracellular
   vesicles (MSC‐EVs) exhibit neuroprotective potential in AD.^[ [217]^38
   ^] Intramedullary injection of Ad‐Rab27a shRNA could not only reach
   osteocytes but also block the secretion of EVs in other bone cells. The
   lack of a specific OCY‐EVs blocking tool is not addressed in this
   study.

   Furthermore, the potential molecular mechanism of OCY^Young‐EVs on
   neuroprotection is explored by proteomics analysis. KEGG pathway
   analysis of differentially enriched proteins
   (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5) show that neurodegenerative diseases
   pathways are primarily enriched. In highly enriched proteins of the AD
   pathway, the functions of upregulated proteins include reducing Aβ
   production (such as ADAM10, NCSTN, and PSEN2), increasing Aβ
   degradation (such as NEP), and preserving extra‐mitochondrial.

   In conclusion, our findings support that OCY^Young‐EVs serve as
   important bone source messengers to preserve brain function in the
   bone‐brain axis (Figure [218]8).

4. Experimental Section

Subjects

   A total of 195 cognitive normal control subjects (67.91 ± 5.79 years
   old) and 117 AD patients (67.72 ± 5.02 years old) were involved in the
   study (Figure [219]1A). All patients were diagnosed by at least two
   experienced doctors from Xiangya Hospital according to the National
   Institute of Aging and Alzheimer's Association (NIA‐AA).^[ [220]^39 ^]
   Cognitive normal controls were recruited from the physical examination
   center of Xiangya Hospital. In addition, a Mini‐Mental State
   Examination (MMSE) ≥ 27 was administered as a standard for cognitive
   normal individuals.^[ [221]^15 , [222]^40 ^] All participants underwent
   dual‐energy X‐ray absorptiometry (DXA) measurement of the lumbar spine
   and total hip in the geriatric outpatient department of Xiangya
   Hospital. According to WHO criteria, BMD classifications of
   osteoporosis, osteopenia, and normal were defined as T‐Score ≤ −2.5,
   −2.5 < T‐Score ≤ −1.0, and T‐score > −1.0, respectively.^[ [223]^16 ^]
   The study was approved by the Ethics Committee of Xiangya Hospital,
   Central South University (institutional review board equivalent)
   (2019030501). Written informed consent was obtained from all
   participants involved in the study.

Plasma Collection and Plasma EVs Extraction from Participants

   Venous blood samples were collected in EDTA‐coated tubes using standard
   venipuncture protocols after an overnight fast. Then plasma was
   collected by centrifugation at 1000 rpm for 10 min at 4 °C. Whole
   plasma EVs were extracted by the ultracentrifugation method. Briefly,
   plasma was sequentially centrifuged at 2000 × g and 10 000 × g for
   30 min to remove dead cells and debris. The supernatant was used to
   concentrate EVs by ultracentrifugation at 100 000 × g for 2 h at 4 °C.
   The EVs pellets were resuspended in PBS and stored at −80 °C until use.

Primary OCY and OCY‐EVs Isolation

   Primary OCY^Young or OCY^Aged were derived from 2‐ or 16‐month‐old
   C57BL/6 male mice according to the published protocol^[ [224]^41 ^]
   with slight modifications. In detail, the femora, tibiae, and humeri
   were aseptically dissected from the mice, and the remaining muscle,
   tendon, and periosteum were removed in α‐MEM (Cat. No. SH30265.01;
   HyClone, Logan, USA) with 2% penicillin and streptomycin (PS; Cat. No.
   P1400; Solarbio, Beijing, China). The bone epiphyses were cut off, and
   the bone marrow was flushed out. The bone was then cut into 1–2 mm
   lengths and washed in α‐MEM with 2% PS. Next, the bone pieces were
   incubated in warmed α‐MEM containing 10 active units per mL of
   collagenase II (Cat. No. 17 101 015; Gibco, Grand Island, USA) for
   30 min at 37 °C, and then washed three times with phosphate‐buffered
   saline (PBS). The incubation step was repeated for another two times.
   Subsequently, the bone pieces were incubated in warmed 5 × 10^−3 m
   ethylenediaminetetraacetic acid (EDTA; Cat. No. E809068; Macklin,
   Shanghai, China) solution at pH = 7.4 for 30 min and then washed three
   times with PBS. The alternate incubation of bone pieces in collagenase
   II and EDTA was conducted for another two times. Finally, the obtained
   bone particles were incubated on plates, which were precoated with 4%
   collagen I (Cat. No. 354 236; Corning, New York, USA) in 0.02 × 10^−6 m
   acetic acid for one hour. α‐MEM with 5% fetal bovine serum (FBS; Cat.
   No. 12664‐025; Gibco), 5% calf serum (CS; Cat. No. A3520502; Gibco),
   and 2% PS was used as primary osteocyte culture medium. The plates were
   kept at 37 °C and 5% CO[2] for seven days until the OCY crawled out
   from the bone slices. In addition, extractions of primary osteoblasts
   and bone marrow macrophages were performed according to the published
   protocols.^[ [225]^42 ^]

   To collect OCY‐EVs, the culture medium was changed to α‐MEM with 10%
   exosome depleted FBS (Cat. No. EXO‐FBS‐50A‐1; System Biosciences, Palo
   Alto, USA) and 1% PS. The cell culture medium was sequentially
   centrifuged at 300 × g for 10 min, 2000 × g for 30 min, and 10 000 × g
   for 30 min at 4 °C, and then the final supernatant was filtered through
   a 0.22‐µm filter (Millipore, Billerica, USA) to remove the residual
   cell debris. EVs were then pelleted by ultracentrifugation at 100 000 ×
   g and 4 °C for 10 h in a Beckman Optima XPN ultracentrifuge (SW 32 Ti
   rotor; k‐factor 204). The EVs pellets were resuspended in PBS and
   stored at −80 °C until use (avoiding multiple freeze‐thaw cycles). The
   protein contents of OCY‐EVs were assessed by bicinchoninic acid (BCA)
   protein assay kit (Cat. No. 70‐PQ0012; MULTI SCIENCES, Hangzhou,
   China). The numbers and sizes of OCY‐EVs were tested by nanoparticle
   tracking analysis (NTA) using ZetaView PMX 110 (Particle Metrix,
   Meerbusch, Germany). The morphologies of OCY‐EVs were detected using a
   Hitachi H‐7650 transmission electron microscope (Hitachi, Tokyo,
   Japan). The efficacy and stability of each batch of OCY‐EVs were
   determined by cellular functional test and NTA analysis, respectively.

In Vivo Distribution of OCY‐EVs

   The OCY^Young‐EVs and OCY^Aged‐EVs were labeled with the lipophilic dye
   DiR iodide (Cat. No. 40757ES25; Yeasen, Shanghai, China) according to
   the manufacturer's instructions for ex vivo fluorescence imaging. After
   incubation, the solution mixture was transferred to Amicon Ultra‐0.5
   Centrifugal filter units (10 kDa; Millipore) and centrifuged at 3000 ×
   g to remove possible excess fluorescent dye. To investigate whether
   OCY‐EVs could reach the brain, DiR‐labeled OCY^Young‐EVs were
   resuspended in PBS and injected in the intramedullary region (6 × 10^7
   particles in 10 µL PBS) of each mouse (4‐month‐old APP/PS1 mouse, n =
   3). The mice in the control group (n = 3) were treated with an equal
   volume of solvent. Twenty‐four hours later, all mice were sacrificed to
   collect the brain, heart, liver, kidneys, femora, and tibiae.
   Fluorescent signals in these tissues were detected immediately and
   quantified with a fluorescence tomography imaging system (FMT‐4000;
   PerkinElmer, USA).

   Furthermore, to compare the metabolic kinetics of OCY‐EVs in the brain
   between APP/PS1 and wild‐type (WT) mice, DiR‐labeled OCY^Young‐EVs or
   OCY^Aged‐EVs were intravenously administered (6 × 10^7 particles in
   50 µL PBS) into each mouse. The mice were sacrificed to collect tissues
   after 6, 24, and 48 h (n = 3 in each group). Fluorescent signals in
   these organs were detected and quantified.

   Similarly, the OCY^Young‐EVs or OCY^Aged‐EVs were also labeled with DiL
   iodide (Cat. No. 40726ES10; Yeasen). After intravenous injection of
   DiL‐labeled OCY‐EVs, the organs were collected at indicated time
   points, fixed in 4% paraformaldehyde for 4 h, and embedded in
   Tissue‐Tek optimum cutting temperature compound (O.C.T. compound; Cat.
   No. 4583; Sakura, Japan) at −20 °C. Due to the particular composition
   of bone tissue, the femora were fixed for 24 h, decalcified in 0.5 M
   EDTA (pH = 7.4) at 4 °C with continuous shaking for 3 days, and then
   embedded in O.C.T. compound at −20 °C. Thirty‐micrometer‐thick slices
   were obtained using a freezing microtome (CryoStar NX50; Thermo, USA),
   and the fluorescence signals of DiL were observed under a fluorescence
   microscope (Carl Zeiss Axio Imager 2; Carl Zeiss, Germany).

Cell Culture and Treatments

   SH‐SY5Y (Cat. No. ZQ0050; Zhong Qiao Xin Zhou Biotechnology, Shanghai,
   China) and HT22 (Cat. No. BFN60808571; BLUEFBIO, Shanghai, China) cells
   were cultured in high glucose DMEM (Cat. No. C11995500BT; Gibco)
   supplemented with 10% FBS and 1% PS at 37 °C and 5% CO[2] in a
   humidified atmosphere. SH‐SY5Y‐APPswe stable cell lines were
   constructed by lentiviral infection with the APPswe mutation (specials:
   Homo sapiens, NM_201 414.3: c.1785G>C p.K595N; c.1786A>C p.M595L),
   which is the most well‐known pathogenic mutation of AD.^[ [226]^43 ^]
   HT22‐APPswe cell lines were constructed by transfection of plasmids
   carrying the APPswe mutation. Corresponding control cell lines were
   transfected with empty lentivirus or plasmids. The successful
   constructions of APPswe cell lines were demonstrated by IF staining
   (Figure [227]S7A,D, Supporting Information) and western blot (Figure
   [228]S7B,C,E,F, Supporting Information) analysis of amyloid precursor
   protein (APP). A total of 5 × 10^4 SH‐SY5Y‐APPswe, SH‐SY5Y,
   HT22‐APPswe, or HT22 cells were seeded in a 24‐well plate with 500 µL
   medium. After 24 h, the cell culture medium was replaced with DMEM with
   OCY^Young‐EVs or OCY^Aged‐EVs (6 × 10^7 particles per mL). Then, the
   cell cultures were collected to detect Aβ levels by enzyme linked
   immunosorbent assay (ELISA), and the cells were analyzed for apoptosis
   by Hoechst 33 342 staining after 48 h.

OCY‐EVs Uptake Assay

   OCY‐EVs were labeled with PKH26 (Cat. No. MINI26; Sigma–Aldrich,
   Darmstadt, Germany) according to the manufacturer's instructions. After
   removing the redundant dye using the above‐described procedures, the
   labeled OCY‐EVs (6 × 10^7 particles per mL) were incubated with SH‐SY5Y
   or HT22 cells at 37 °C for 12 h. Then, the treated cells were washed
   with PBS and fixed with 4% paraformaldehyde for 15 min. After washing
   with PBS, the HT22 cells were incubated with Alexa Fluor™ 488
   Phalloidin (1: 200; Cat. No. A12379; Invitrogen) at 37 °C for 30 min.
   The stained cells on coverslips were washed twice with PBS and mounted
   with the commercial antifade mounting medium with DAPI (Cat. No.
   H‐1200‐10; Vectorlabs, California, USA). After IF staining, a
   fluorescence microscope was used to acquire images.

Hoechst Staining and Apoptosis Analysis

   Cells were stained with Hoechst 33 342 (Cat No.40732ES03; Yeasen) after
   fixation according to the manufacture's protocol. A fluorescence
   microscope was used to acquire images. As the nuclei of apoptotic cells
   were concentrated,^[ [229]^44 ^] apoptotic cells showed a high
   fluorescence intensity by Hoechst staining and were considered
   Hoechst‐positive cells. The percentage of Hoechst‐positive cells was
   analyzed using ImageJ software (NIH).

Animal Experiments

   Animal care and experimental procedures were approved by the Ethical
   Review Board at Xiangya Hospital of Central South University
   (2019030501). APP/PS1 mice (Stock No. 34829‐JAX) and Dmp1^Cre mice
   (Stock No. 02 3047) were imported from Jackson Laboratory. C57BL/6 mice
   were obtained from Hunan SJA Laboratory Animal Co. Ltd.

   EVs reporter (Cd63^fl/+ ) mice and osteocyte specific targeting
   (Sost^CreERT2 ) mice were established using homologous recombination
   CRISPR/Cas9 technology. Cas9 mRNA and gRNA were obtained by in vitro
   transcription. For Cd63^fl/+ mice, a homologous recombination vector
   (donor vector) contained a 3.4 kb 5′ homologous arm,
   loxp‐cc‐mCherry‐polyA‐loxp‐cc‐eGFP, and a 4.0‐kb 3′ homologous arm. For
   Sost^CreERT2 mice, a homologous recombination vector (donor vector)
   contained a 3.0 kb 5′ homologous arm, CreERT2‐Wpre‐pA, and 3.0 kb 3′
   homologous arm. Cas9 mRNA, gRNA, and donor vector were microinjected
   into the fertilized eggs of C57BL/6J mice to obtain F0 generation mice.
   Homologous recombinant F0 generation mice were identified by long
   fragment PCR. To activate CreERT2, 100 mg kg^−1 tamoxifen (Cat. No.
   ST1681; Beyotime) was administered once a day for five consecutive days
   by intraperitoneal manner. The genetic identification results of
   Sost^CreERT2;CD63^fl/+ mice, DMP1^Cre;CD63^fl/+ mice, and APP/PS1 mice
   are shown in Figure [230]S2A–D in the Supporting Information.

   Four‐month‐old early stage AD model mice were intravenously
   administered by OCY^Young‐EVs, OCY^Aged‐EVs, or solvent weekly based on
   the weight of the mouse (2.4 × 10^6 particles per gram, 1.2 × 10^6
   particles per µL) for two months. Six‐month‐old later stage AD model
   mice were intravenously administered by OCY^Young‐EVs, OCY^Aged‐EVs, or
   solvent (4.8 × 10^6 particles per gram, 1.2 × 10^6 particles per µL)
   weekly for two months. Behavior tests were performed after the
   interventions. Then, the brain and femora were harvested for downstream
   measurements.

   An adenovirus carrying small hairpin RNA targeting Rab27a
   (Ad‐Rab27a‐shRNA) was constructed to inhibit the EVs secretion from the
   bone. AD model mice were constructed by injecting Aβ40 into the
   hippocampus. Eight‐week‐old C57BL/6 male mice were randomly divided
   into four groups: Ad‐Mock/Solvent, Ad‐Mock/Aβ40,
   Ad‐Rab27a‐shRNA/Solvent, and Ad‐Rab27a‐shRNA/Aβ40. The mice were
   intramedullary injected by 5 µL (5 × 10^10 IFU mL^−1) of
   Ad‐Rab27a‐shRNA or Ad‐Mock at the bilateral femur under anesthesia.
   After seven days, the mice were placed in a stereotactic device and
   bilaterally administered 3 µL Aβ40 (1 µg µL^−1) or solvent into the
   hippocampus under anesthesia.^[ [231]^45 ^] The Aβ1‐40 monomer solution
   (Cat. No. 2027‐2‐20; ChinaPeptides, Shanghai, China) was incubated at
   37 °C for three days before use. The injection location of the
   hippocampus was −2.0 mm posterior to the bregma, ±1.3 mm lateral to the
   sagittal suture, and 1.7–1.8 mm in depth (Figure [232]S10, Supporting
   Information). Behavior tests were performed two weeks later. After the
   test, the brain and femora were harvested for downstream measurements.

Morris Water Maze

   The apparatus consisted of a white circular pool (130 cm diameter,
   51 cm high) filled with water (30 cm depth at 22 °C) and divided into
   four quadrants. An invisible circular platform (10 cm in diameter) was
   placed in the middle of one of the quadrants and submerged 1 cm below
   the surface. A distal clue was also placed for mice to locate the
   platform. A commercial animal video‐tracking analysis system (AVTAS
   ver4.0; Wuhan, China) was used for the Morris Water Maze (MWM). Animals
   were first subjected to water maze training and four trials from four
   different fixed start positions each day until most of them arrived at
   the platform quickly. Each trial lasted 60 s or until the animal found
   the platform. All animals were left on the platform for 10 s (including
   those who failed to locate it). The interval between each training was
   1 h so that the mice can regain their physical strength. After the last
   training session, the mice were subjected to the probe trail for 60 s.
   The platform was removed from the pool in the probe trial, and the time
   spent in the platform quadrant and the platform crossing time were
   measured.

Y‐Maze

   The apparatus was a symmetrical Y‐Maze; each arm measured 50 × 10 cm
   with 20‐cm‐high sidewalls and was labeled with different shapes. Mice
   were placed in one arm (start arm) of the maze with a closed arm and
   allowed to explore freely for 10 min. The unexplored arm is named the
   novel arm, and the remaining arm is named the familiar arm. After 2 h,
   the mice were placed in the start arm (all three arms were open) and
   the maze was explored for 5 min. The times in each arm were recorded.

Object Location Task and Novel Object Recognition Task

   The apparatus consisted of a square area (30 × 30 cm) with 30‐cm‐high
   sidewalls. Mice were placed into the testing room to acclimate for at
   least 30 min and then allowed to explore the arenas freely for 10 min.
   The arenas were cleaned to minimize olfactory cues before the next use.
   For the training trial, two different objects were secured to the open
   field at 6 × 6 cm away from their respective walls. After a duration of
   20 min, the animals were placed facing the walls in the release corner
   and allowed to investigate the arena and objects freely for 10 min. For
   the object location task (OLT), one object was moved to a new location
   6 × 6 cm away from the walls and not the release corner. The animals
   were placed in the same way and allowed to investigate the arena and
   objects freely for 10 min after 20 min rest. For the novel object
   recognition task (NORT), the stationary object in the OLT was replaced
   with a novel object, while the moving object from the OLT was now the
   familiar object. Twenty minutes later, the animals were placed for the
   third time and allowed to investigate the arena and objects freely for
   10 min. The investigation times for each mouse with each object were
   recorded and analyzed.

IF Staining

   For IF staining, brain tissues were fixed in 4% paraformaldehyde for 24
   h. Subsequently, the samples were dehydrated in 30% glucose with 1%
   polyvinyl pyrrolidone (PVP; Cat. No. 9003‐39‐8; Sigma–Aldrich) at 4 °C
   for ≈24 h until the tissues sank to the bottom, and they were then
   embedded in O.C.T. compound at −20 °C. Ten‐micrometer‐thick slices were
   incubated with different antibodies: SOST (1: 500; Cat. No. AF1589;
   R&D, Minnesota, USA), GFP (1: 50 000; Cat. No. A6290; Abcam, Cambridge,
   England), 6E10 (1: 500; Cat. No. 803 001; Biolegend, San Diego, CA,
   USA), Iba1 (1: 500; Cat. No. ab178847; Abcam), Synaptophysin (1: 500;
   Cat. No. MA5‐14532; Invitrogen), or NeuN (1: 500; Cat. No. MAB377;
   Millipore). Appropriate secondary antibodies (Abcam) were used,
   followed by incubation with DAPI. A total of six sections per brain
   containing the hippocampus and cortex were stained. Fluorescence
   microscopy and Pannoramic Scanner (Pannoramic MIDI; 3D HISTECH,
   Hungary) were used to acquire images.

Image Quantification

   After IF staining, confocal images were captured, and the mean
   intensity of fluorescence, number of immunoreactive cells, and size of
   the plaques were quantified. To quantify 6E10 in the mouse brain, six
   sections (including the hippocampus) were scanned using a Pannoramic
   Scanner. Images were then analyzed by Image Pro‐Plus (MEDIA
   CYBERNETICS), and the background was subtracted by the software for
   fluorescence images before quantification. For Aβ plaques, the number
   of plaques, the size of plaques, and the intensity of 6E10 in six brain
   sections containing the hippocampus and cortex of each mouse were
   measured. For Aβ plaque‐associated Iba1 intensity quantification, Iba1
   intensities were counted within the 100‐µm radius of the 6E10‐positive
   plaques. At least 100 plaques were counted in each group, and Iba1
   intensities were normalized to the plaques. For synapse studies,
   synaptophysin (Syp) intensities in the cortex and hippocampus were
   quantified by Image Pro‐Plus. Because of the great variation in
   synaptic puncta densities by brain region, care was taken to image the
   same areas consistently across all samples. For NeuN‐positive cell
   quantification, six images from the cortex or the scanned images of the
   hippocampus of each mouse were counted manually. All images were
   captured and analyzed blindly using coded slides.

ELISA

   The concentrations of Aβ42 and Aβ40 were measured using commercial
   ELISA kits (Cat. No. DAB142, DAB140B; R&D). All procedures were
   performed in accordance with the manufacturer's instructions.

Western Blot

   Cells or brain tissues were collected and resuspended in modified RIPA
   lysis buffer (Cat. No. P0013B; Beyotime, Shanghai, China) containing
   protease and phosphatase inhibitor mixture. The suspensions were lysed
   on ice for 30 min, sonicated, and then centrifuged at 13 000 rpm for
   10 min at 4 °C. Protein concentrations were estimated using the BCA kit
   (Cat. No. 70‐PQ0012; Multi Sciences). Lysates were separated on 8%,
   10%, 12%, and 15% SDS–polyacrylamide electrophoresis (SDS–PAGE) gels
   (Cat. No. P1200; Solarbio) or 15% Tris‐Tricine‐SDS–PAGE gels (Cat. No.
   P1320; Solarbio). After the separation, proteins were transferred to a
   PVDF membrane (Cat. No. IPVH00010, ISEQ00010; Millipore), and
   nonspecific binding sites were blocked using commercial western
   blocking buffer (Cat. No. SW3010; Solarbio) followed by antibody
   incubation: TSG101 (1: 1000; Cat. No. sc‐7964; Santa Cruz, Dallas,
   Texas, USA), Flotillin‐1 (1: 500; Cat. No. sc‐74566; Santa Cruz), DMP1
   (1: 1000; Cat. No. NBP1‐45525; NOVUS, Colorado, USA), SOST (1: 500;
   Cat. No. AF1589; R&D), P21 (1: 1000; Cat. No. ab188224; Abcam), P16 (1:
   2000; Cat. No. ab211542; Abcam), 6E10 (1: 1000; Cat. No. 803 001;
   Covance), APP (1: 1000; Cat. No. ab32136; Abcam), or β‐actin (1: 1000;
   Cat. No. [233]O10313; TransGen Biotech, Beijing, China). Secondary
   HRP–conjugated antibodies (1: 5000; Solarbio) were used. Western blot
   images were captured using the ChemiDoc XRS+ with Image Lab Software
   (Bio–Rad, California, USA). The western blot bands were quantified
   using ImageJ software.

qRT–PCR

   Bone and bone marrow samples were freshly collected. Total RNA was
   extracted using the TransZol Up Plus RNA Kit (Cat. No. ER501‐01;
   Transgen) and cDNA was synthesized from 1 µg of total RNA using the
   NovoSceipt Plus All‐in‐one 1^st Strand cDNA synthesis kit (Cat. No.
   E047; Novoprotein, Shanghai, China). The cDNA was then amplified with
   2× SYBR Green qPCR Master Mix (Cat. No. [234]B21202; Bimake, Texas,
   USA) in an FTC‐3000 real‐time PCR system (Funglyn Biotech Inc, Toronto,
   Canada). The relative standard curve method (2^−ΔΔCt) was used to
   determine the relative gene expression and Gapdh was used as a
   housekeeping gene for internal normalization. The PCR primers used in
   this study were as follows: mouse‐Rab27a: forward,
   5″‐TGGAAGGGAGTACCTAAGGGAT‐3″, and reverse,
   5″‐CTCAGCCAGGGTTGATGAGATA‐3″ ; mouse‐Gapdh: 5″‐CACCATGGAGAAGGCCGGGG‐3″,
   and reverse, 5″‐GACGGACACATTGGGGGTAG‐3″.

Proteomic Analysis

   OCY‐EVs samples (OCY^Young‐EVs and OCY^Aged‐EVs, n = 3 for each) were
   prepared as described previously,^[ [235]^46 ^] and processed for
   label‐free quantitative proteomic analysis by Jingjie PTM BioLab
   (Hangzhou, China). Differentially enriched proteins were identified
   (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5 or < 0.67, P < 0.05). Enrichment of
   functions and signaling pathway analysis were performed based on Gene
   Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG)
   database at the David site ([236]https://david.ncifcrf.gov).^[ [237]^47
   ^] For functional enrichment analysis, differentially enriched proteins
   (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5) were mapped to terms in the GO
   databases, and then significantly enriched GO terms were searched using
   P < 0.05 as the threshold. GO term analysis was classified into three
   subgroups, namely biological process (BP), cellular component (CC), and
   molecular function (MF). The differentially enriched proteins
   (OCY^Young‐EVs/OCY^Aged‐EVs > 1.5) were also mapped to the KEGG
   database and searched for significantly enriched KEGG pathways at the P
   < 0.05 level.

Statistical Analysis

   Continuous variables are expressed as mean ± SD. The sample size (n)
   for each statistical analysis was detailed in figure legends. The
   Gender, T‐Score Classification, and BMD Classification data of clinical
   subjects were analyzed by the Chi‐Square Test. Spearman Correlation
   Analysis was adopted between the T‐Score and MMSE. The unpaired,
   two‐tailed Student's t‐test was used to analyze the differences between
   two groups. The significance of multiple‐group comparisons was
   performed by one‐way analysis of variance (ANOVA) or Kruskal–Wallis
   test with Bonferroni post hoc correction. For all experiments, P < 0.05
   was considered as significant, and was represented by “*/#”; P < 0.01,
   P < 0.001, and P < 0.0001 were represented by “**/##”, “***/###”, and
   “****/####”. GraphPad Prism software (Version 8.0) and SPSS (Version
   23.0) were used for the above statistical analyses.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   Y.‐L.J. and Z.‐X.W. contributed equally to this work. L.S., H.X.,
   Y.‐L.J., and Z.‐X.W. conceived the project and designed the
   experiments. Y.‐L.J., X.‐X.L., M.‐D.W., Y.‐W.L., Y.‐Y.W., T.‐F.W., and
   Y.Z. performed the experiments. L.S., H.X., B.J., X.‐X.L., Z.‐W.L.,
   C.‐G.H., Y.‐J.T., L.W., Y.‐F.Z., S.‐S.R., J.C., and Z.‐Z.L. contributed
   to the experimental technical consultations. Y.‐L.J. and Z.‐X.W.
   contributed to the data acquisition and analysis. L.S., H.X., Y.‐L.J.,
   and Z.‐X.W. wrote the manuscript. All authors reviewed and revised the
   manuscript.

Supporting information

   Supporting Information
   [238]Click here for additional data file.^ (2.1MB, pdf)

   Supporting Information
   [239]Click here for additional data file.^ (79.9KB, xlsx)

Acknowledgements