Cardiac-derived extracellular vesicles carrying miR-4433b-3p accelerate cognitive decline and brain aging by suppressing TP53INP2-mediated neuronal autophagy in mice

Abstract

Brain aging is not an independent process, yet how systemic aging drives neural decline remains unclear. Here, we identified a circulating miR-4433b-3p, packaged within extracellular vesicles (EVs), as a trans-organ effector bridging cardiac aging with central nervous system (CNS) decline. Small RNA sequencing and human cohort validation revealed selective enrichment of miR-4433b-3p in aged plasma EVs (Op-EVs), correlating with blood biomarkers of brain aging. Source tracing in mice identified the aged heart as the major origin of miR-4433b-3p-laden EVs. Functionally, aged cardiac EVs (Oc-EVs) accumulated in the hippocampus, impaired memory and induced neuronal senescence. Mechanistically, miR-4433b-3p suppressed TP53INP2, a facilitator of autophagic flux, leading to disrupted autophagosome maturation. Restoring TP53INP2 or inhibiting miR-4433b-3p rescued neuronal autophagy and improved cognition. Collectively, these findings uncover a heart-brain axis by EV-mediated miRNA signaling, positioning cardiac EV-miR-4433b-3p as a circulating biomarker and potential therapeutic target for age-related cognitive decline.

Cardiac aging signals propagate to the brain via miR-4433b-3p–enriched extracellular vesicles that disrupt neuronal autophagy and accelerate brain aging in mice.

Introduction

Aging is a systemic and multifactorial biological process characterized by progressive functional decline across multiple organ systems¹. Among these, the brain is particularly vulnerable, owing to its cellular complexity and central role in cognition, memory, and emotional regulation—capacities frequently impaired during aging². As globally populations age, brain aging has emerged as a pivotal risk factor for neurodegenerative disorders, such as Alzheimer’s disease (AD) and Parkinson’s disease (PD)^3,4. However, the molecular mechanism of brain aging—particularly those involving inter-organ communication—that underlie age-related neural deterioration remain poorly understood.

Historically, studies of brain aging have emphasized cell-intrinsic drivers, including genomic instability, oxidative stress, mitochondrial dysfunction, and neuroinflammation^5,6. However, accumulating evidence suggests that the aging brain is dynamically influenced by peripheral signals⁷. Circulating factors originating from remote organs, including the heart, liver, and skeletal muscle, have been shown to modulate neurogenesis and synaptic function⁸. Classic parabiosis experiments, wherein the vasculature of young and aged mice is surgically joined, have demonstrated that exposure to aged blood impairs hippocampal plasticity and cognition in young animals—highlighting the presence of pro-aging factors in systemic circulation^9,10. Nevertheless, the molecular identity, origin, and mechanistic contribution of such circulating signals to central nervous system (CNS) decline remain largely unresolved.

Extracellular vesicles (EVs)—lipid bilayer-enclosed nanoparticles enriched in proteins, lipids, and nucleic acids—have emerged as potential mediators of peripheral-to-central communication¹¹. EV composition shifts with age, and increased permeability of the blood-brain barrier (BBB) may permit greater entry of peripheral EVs into the brain, where they can interface with neural cells¹². Of particular interest are EV-associated microRNAs (miRNAs), a group of small noncoding RNAs (about 22 nucleotides) that regulate gene expression post-transcriptionally and are implicated in both aging and neuroinflammation^11,13. However, whether age-associated EV miRNAs actively impair neuronal function and contribute to cognitive decline remains unknown.

In this study, we investigated the role of aged cardac EVs (Oc-EVs) in driving neuronal dysfunction and cognitive decline during aging. Small RNA sequencing, combined with in vitro and in vivo validation, demonstrated that Oc-EVs regulate neuronal autophagy through miRNA-mediated mechanisms. Together, these findings delineate a cardiac-brain communication axis and provide molecular insights into how peripheral, heart-specific aging shapes CNS deterioration.

Results

Aged plasma EVs induce cognitive impairments and hippocampal neuronal senescence

To investigate the potential role of EVs in brain aging, we first isolated Op-EVs from 21-month-old mice (Fig. 1A). Comprehensive characterization of isolated plasma EVs using NTA (Fig. S1A), TEM (Fig. 1B), and western blot (WB) (Fig. S1B) confirmed typical vesicle morphology, size distribution, and marker enrichment (ALIX, TSG101, and CD63), with no detectable calnexin and Golgi contamination (GM130)¹⁴.

Fig. 1. Exogenous old-mouse-plasma EV intravenous injection via tail vein inducing cognitive deficits and neuronal senescence in middle-aged mice.

A Experimental timeline. The 10-month-old mice are given tail vein injections of plasma-derived extracellular vesicles (Op-EVs) from 21-month-old donors every other day for 28 d, followed by behavioral assessments and subsequent brain tissue collection for pathological analysis. Created with BioRender.com. B Transmission electron microscopy (TEM) analysis revealing characteristic morphology of EVs. C In vivo imaging of DiR-labeled aged EVs demonstrating accumulation in the brain and peripheral organs. D–J Behavioral assessments following tail vein injection of PBS or Op-EVs (n = 8 per group). D Escape latency during the training phase of the Morris water maze (MWM). E Time spent in the target quadrant and platform crossings in probe trial. F Alternation percentage in the Y-maze test. G Discrimination index for novel object recognition (NOR). H, I Trajectories and quantification of center zone activity in the open field test (OFT). J Immobility time in the tail suspension test (TST). K, L Nissl staining of hippocampal neurons assessing neuronal survival and quantification of hippocampal neuronal density. M Representative images of SA-β-gal staining in hippocampus. N H&E staining of hippocampal sections. O Western blot (WB) and quantification of p21 and TNF-α expression in hippocampus. P Immunofluorescence images of p21 in hippocampal neurons. Data are presented as mean ± SEM from three independent biological replicates (n = 3). Statistical significance is determined using a two-tailed unpaired Student’s t test based on P < 0.05 (*) and P < 0.01 (**).

To assess the biological effects of these Op-EVs, 10-month-old recipient mice, selected based on preliminary experiments comparing 7–12 months old mice for cognitive decline and neuronal senescence (Fig. S1C, D), received intravenous injections of Op-EVs. PKH67-labeled EVs showed efficient brain penetration based on confocal microscopy (Fig. S1E), while longitudinal in vivo imaging of DiR-labeled EVs revealed systemic biodistribution, with significant accumulation in the brain, liver, kidney, and heart persisting for 4 weeks (Fig. 1C).

Behavioral tests were conducted to evaluate cognitive and emotional outcomes, including Morris water maze test (MWM), Y-maze, novel object recognition test (NOR), open field test (OFT), and tail suspension test (TST). In the MWM training phase, Op-EV-treated mice required significantly more time to locate the hidden platform, indicating impaired spatial learning (Fig. 1D). In the probe trial, these mice spent less time in the target quadrant, further supporting spatial memory deficits (Fig. 1E). Y-maze test exhibited a significant reduction in spontaneous alternation percentage (Fig. 1F), while NOR showed decreased the exploration ratio, further supporting deficits in recognition memory (Fig. 1G). Op-EV-treated mice exhibited anxiety- and depression-like behaviors, characterized by reduced central exploration in the OFT and prolonged immobility in the TST (Fig. 1H–J). Together, these behavioral changes indicated that the Op-EVs impaired both cognitive function and emotional regulation in mice.

To determine whether these cognitive changes were accompanied by neuropathological alterations, we systematically evaluated neuronal survival, cellular senescence, and tissue architecture. Nissl staining revealed a significant decrease in the number of surviving hippocampal neurons in Op-EV-treated mice, indicating compromised neuronal integrity and reduced neuronal viability¹⁵ (Fig. 1K–L). Given that neuronal loss is often linked to cellular senescence during aging, we next assessed SA-β-gal activity. A marked increase in SA-β-gal-positive cells was observed in the hippocampus of Op-EV-injected mice, suggesting enhanced accumulation of senescent neurons (Fig. 1M). Consistent with these cellular alterations, H&E staining further demonstrated pronounced structural deterioration, including reduced neuronal density and disrupted cytoarchitecture¹⁶, indicating that Op-EVs contributed to hippocampal neurodegeneration (Fig. 1N). Molecular analysis showed upregulation of senescence marker p21 and the pro-inflammatory cytokine TNF-α (Fig. 1O, P), indicating associations among Op-EVs exposure, neuronal senescence, and inflammation. These findings supported that Op-EVs induced hippocampal injury at multiple levels, ranging from neuronal loss and enhanced cellular senescence to overt structural degeneration.

MiR-4433b-3p is enriched in age-dependent cardiac-derived EVs and correlates with brain aging biomarkers

To identify age-associated miRNAs that may mediate the effects of Op-EVs on brain aging, sRNA-Seq was performed on plasma EVs from six elderly (>80 years) and six young (18–30 years) donors (Fig. 2A). Based on this analysis, we focused on a total of 77 differentially expressed miRNAs, which were visualized in the heatmap and volcano plot (Fig. 2B, C).

Fig. 2. Cardiac-derived EVs enriched in miR-4433b-3p correlated with brain aging biomarkers.

A Schematic of experimental design. Plasma-derived EV miRNAs sequenced from young (n = 6, ≤ 30 years) and elderly (n = 6, ≥ 80 years) individuals. Created with BioRender.com. B Heatmap of differentially expressed exosomal miRNAs between young and elderly plasma samples (fold change > 1.5). C Volcano plot illustrating significantly upregulated exosomal miRNAs in the elderly individuals based on sRNA sequencing. D Plasma samples collected from 197 individuals aged 30 to 80 years. Following EV isolation, miRNAs are amplified, and the relative expression levels of brain aging markers across age groups are quantified using single molecule micro-wells immunoanalyzer. A series of correlation analyses are subsequently performed. Created with BioRender.com. E qRT-PCR performed on the cohort to validate differentially upregulated miRNAs identified by sequencing. F Quantification of miR-4433b-3p abundance across age groups. G, H Expression levels of plasma brain aging biomarkers across groups stratified by EV miR-4433b-3p expression (Low, < 26.06868352; Middle, 26.20805039-82.03069673; High, > 83.79900493). Statistical analyses are performed by one-way ANOVA with Turkey’s multiple comparison test. Ns, not significant; P < 0.05 (*), P < 0.01 (**), P < 0.001 (***), and P < 0.0001 (****). I Spearman correlation between levels of plasma EV miR-4433b-3p and brain aging biomarkers. Numbers indicate Spearman’s correlation coefficients (r_s). Corresponding p values are provided in Table S2. J RT-PCR analysis of miR-4433b-3p of plasma EVs from control mice of 3-month-old and aged mice of 21-month-old (n = 6 per group). K qRT-PCR analysis of miR-4433b-3p of EVs based on tissue lysates from brain, heart, kidney, liver, lung, and spleen of control mice of 3-month-old and aged mice of 21-month-old (n = 6 per group). Statistical analyses are performed by two-way ANOVA with Turkey’s multiple comparison test. Ns, not significant; P < 0.01 (**) and P < 0.0001 (****).

To validate and quantify the relevance of the candidate miRNAs to aging, we further analyzed plasma EV samples from a cohort of 197 individuals aged 20 to 87 years (Fig. 2D). Plasma EVs of individuals were characterized using TEM (Fig. S1F), NTA (Fig. S1G), and WB (Fig. S1H), indicating high plasma EV purity¹⁷.

Initially, qRT-PCR was performed on selected upregulated miRNAs derived from small RNA sequencing analysis, comparing individuals aged 20–40 and 60–80 years. Among these miRNAs, miR-4433b-3p showed the most pronounced and consistent age-related increase, thus was selected for further investigation (Fig. 2E). We then examined the abundance of miR-4433b-3p across the entire cohort. Plasma EV-derived miR-4433b-3p showed a robust age-dependent increase (Fig. 2F), with abundance levels significantly higher in older adults ( > 65 years old; 153.3 ± 89.4) compared to young adults (20-45 years old; 43.2 ± 28.7; p < 0.001; Table S1). To determine whether plasma EV-associated miR-4433b-3p was encapsulated within EVs rather than externally associated, RNase and protease protection assays demonstrated that miR-4433b-3p was predominantly localized within the EV lumen (Fig. S1J).

To explore the clinical relevance of this miRNA, we examined its association with key neurodegeneration-related biomarkers, including neurofilament light chain (NfL), amyloid-beta 42 (Aβ42), amyloid-beta 40 (Aβ40), and phosphorylated tau 181 (pTau181). Stratification of participants into tertiles based on miR-4433b-3p expression revealed significantly elevated levels of NfL and Aβ40 in individuals in the highest tertile compared to those in the lowest tertile (Fig. 2G, H and S1K). For instance, NfL levels were increased from 4.3 ± 3.2 pg/mL in younger individuals to 49.3 ± 19.4 pg/mL in older individuals (p < 0.001; Table S1). Additionally, Spearman correlation analysis revealed a strong positive association between plasma EV-derived miR-4433b-3p levels and NfL, a well-established marker of neuroaxonal injury. Modest but statistically significant associations were also observed between miR-4433b-3p and Aβ40 as well as Aβ42, whereas no significant correlation was detected with pTau181 (Fig. 2I).

To confirm whether the age-associated elevation of plasma EV miR-4433b-3p observed in mice mirrored the trend in humans, we performed qRT-PCR analysis on plasma EVs isolated from 3-month-old and 21-month-old mice (n = 6 per group). The miR-4433b-3p levels were significantly higher in aged mice, consistent with the pattern observed between young and elderly human donors (Fig. 2J). To explore the potential peripheral tissue sources responsible for the elevated miR-4433b-3p expression in Op-EVs, EVs were isolated from major organs, including the heart, liver, spleen, lung, kidney, and brain, from 3-month-old (young) and 21-month-old (aged) mice (n = 6 per group). As an initial qualitative screening step, canonical EV markers CD63 and ALIX were assessed to verify the presence of EV-associated proteins and to identify potential age-associated differences among organ-derived EV preparations. Notably, EV preparations derived from the aged heart exhibited higher levels of CD63 and ALIX compared with those from young mice (Fig. S2A). Because EV marker abundance alone cannot distinguish between increased EV number and altered protein loading per vesicle, we next performed NTA to directly quantify EV concentration. NTA revealed a significant age-associated increase in EV concentration specifically in heart-derived EVs, whereas EVs derived from other organs showed no significant differences between young and aged mice (Fig. S2B). Importantly, qRT-PCR profiling demonstrated that miR-4433b-3p was predominantly enriched in EVs derived from the heart of aged mice (Fig. 2K). Together, these results suggest that heart could serve as key contributors to the age-associated increase in circulating EV miR-4433b-3p. Overall, our findings provide mechanistic evidence linking cardiac senescence to CNS decline, suggesting that aging-associated cardiac EVs carrying miR-4433b-3p may mediate heart-brain communication and contribute to brain aging.

Aging-associated cardiac EVs suppress neuronal autophagy in vivo and in vitro

To explore how Oc-EVs contributed to brain aging, differential expression analysis coupled with KEGG pathway enrichment revealed significant upregulation of age-associated miRNAs targeting genes involved in autophagy, implying that defective autophagic flux could be a major downstream consequence of Oc-EV signaling (Fig. 3A).

Fig. 3. Autophagy-related miRNAs enriched in aged cardiac EVs contributing to brain autophagic dysfunction.

A KEGG enrichment analysis revealing enrichment of autophagy-related pathways among differentially expressed miRNAs. B qRT-PCR analysis of Cdkn1a and Tnf-α of heart from control mice of 3-month-old and aged mice of 21-month-old (n = 3 per group). C Study design. Old cardiac derived EVs (Oc-EVs) are isolated from the hearts of aged (21-month-old) mice and intravenously injected into 10-month-old recipient mice via the tail vein every other day for 28 d. Mice in the Oc-EVs group receive 200 µg protein-equivalent EVs diluted in PBS per injection. Hippocampal tissues are collected 24 h after the final injection for subsequent autophagy-related analyses. Created with BioRender.com. D, E WB analysis and quantification of autophagy markers under different treatments. F qRT-PCR analysis of Sqstm1 mRNA expression in the hippocampus of control and Oc-EV-treated mice. Ns, no significant difference. G TEM analysis of mouse brain tissue after Oc-EV injection. Red arrows indicate autophagosomes. H Autophagosomes quantified in ≥ 10–15 neurons per mouse from three independent mice. Data are presented as mean autophagosomes per neuron ± SEM. I Immunofluorescence images of p62 in hippocampal neurons. Data are presented as mean ± SEM from three independent biological replicates (n = 3). Statistical significance is determined using a two-tailed unpaired Student’s t test. Ns, not significant; P < 0.05 (*), P < 0.01 (**), and P < 0.0001 (****).

Based on KEGG pathway enrichment analysis indicating autophagy-related alterations, we first confirmed cardiac aging in 21-month-old mice, as evidenced by elevated expression of p21 and TNF-α, markers of senescence and inflammation, compared with 3-month-old controls (Fig. 3B). Oc-EVs isolated from aged heart were intravenously injected into 10-month-old recipient mice via the tail vein to evaluate hippocampal autophagic activity (Fig. 3C). Confocal imaging of PKH67-labeled Oc-EVs confirmed their efficient penetration into the brain, supporting their ability to deliver cardiac-derived molecular cargo to neural tissue (Fig. S3A). Having established their brain accessibility, we next examined whether Oc-EVs exposure affected neuronal autophagy. WB analysis revealed a significant decrease in LC3BII, the lipidated form of LC3 that was associated with autophagosome membranes and reflected autophagosome abundance, together with a pronounced accumulation of p62/SQSTM1, a selective autophagy adapter protein that was normally degraded during autophagic flux (Fig. 3D, E). To determine whether the increased p62 resulted from altered transcription rather than defective degradation, we assessed Sqstm1 mRNA levels. The results of qRT-PCR analysis showed no significant change in Sqstm1 mRNA levels in the hippocampus following Oc-EV administration (Fig. 3F), indicating that the increased p62 protein abundance was unlikely to result from transcriptional upregulation. At the ultrastructural level, TEM revealed a marked reduction in the number of autophagosomes within hippocampal per neuron, indicative of compromised autophagic flux (Fig. 3G, H). Consistently, immunofluorescence staining of the hippocampal cornu ammonis 1 (CA1) and dentate gyrus (DG) regions revealed pronounced accumulation of p62 in Oc-EVs-treated mice (Fig. 3I), corroborating the in vivo impairment of autophagic flux.

In vitro co-culture of primary neurons with PKH67-labeled Oc-EVs confirmed their uptake (Fig. 4A). Primary hippocampal neurons derived from fetal mice and co-cultured with Oc-EVs displayed similar increasing patterns in p21 and TNF-α expression, indicating neuron senescence induced by Oc-EVs (Fig. 4B, C). To directly measure autophagic flux, we employed a tandem pmCherry-EGFP-LC3B reporter¹⁸. Because EGFP fluorescence is quenched in the acidic lysosomal environment, whereas pmCherry remains stable, red-only puncta represent autolysosomes, while yellow puncta indicate autophagosomes. In primary neurons exposed to Oc-EVs, a significant reduction was observed in red-only puncta, indicating diminished autophagosome clearance (Fig. 4D). WB analysis revealed decreased LC3BII levels and pronounced accumulation of p62. Upon lysosomal inhibition with BafA1, LC3BII accumulation was markedly attenuated in Oc-EV-treated neurons compared with BafA1-treated control neurons, supporting impaired autophagic flux primarily at the level of autophagosome formation (Fig. 4E).

Fig. 4. Oc-EVs internalized by neurons and suppress autophagic flux.

A PKH67 labeled Oc-EVs enter primary neuron. B Relative expression of Cdkn1a and Tnf-α mRNA in different treatment groups based on qRT-PCR. C WB analysis and quantification of p21 and TNF-α in primary neurons. D Representative image and quantification of autophagic flux in primary neurons transfected with pmCherry-EGFP-LC3B after treatment with PBS or Oc-EVs. Autophagic flux is evaluated by counting yellow- (autophagosomes) and red-only (autolysosomes) puncta. E WB analysis and quantification of autophagy-related proteins in primary neurons under the indicated treatments. Bafilomycin A1 (BafA1) is used to assess autophagic flux. F Representative p62 immunofluorescence images of HT22 cells under the indicated treatments, including autophagy induction and flux inhibition conditions. G TEM analysis showing autophagosomes in HT22 treated with Oc-EVs. Red arrows indicate autophagosomes. H Quantification of autophagosomes per neuron based on TEM images of hippocampal neurons. Autophagosomes are quantified using more than 30 cells per experiment from three independent biological replicates. Data are presented as mean autophagosomes per cell ± SEM. I Representative image and quantification of autophagic flux in HT22 transfected with pmCherry-EGFP-LC3B under the indicated treatments. Data are presented as mean ± SEM from three independent biological replicates (n = 3). Statistical significance is determined using a two-tailed unpaired Student’s t test (for panels B–D) or two-way ANOVA followed by Tukey’s multiple comparisons test, as appropriate. P < 0.05 (*), P < 0.01 (**), and P < 0.0001 (****).

To further validate these findings in vitro, Oc-EVs were co-cultured with the mouse hippocampal neuronal cell line HT22. We first assessed p62 accumulation under conditions of autophagic modulation. Immunofluorescence analysis revealed that Oc-EV treatment increased p62 levels compared with control cells. Importantly, BafA1 treatment further enhanced p62 accumulation in control cells but produced only a modest additional effect in Oc-EV-treated cells, suggesting that autophagic degradation was already compromised by Oc-EVs exposure. In contrast, rapamycin reduced p62 accumulation, partially reversing the Oc-EV-induced phenotype (Fig. 4F). To determine whether this impairment was associated with altered autophagosome formation, TEM analysis was performed. Oc-EV treatment significantly reduced the number of autophagosomes per cell compared with controls (Fig. 4G, H), indicating defective autophagosome generation rather than excessive accumulation. Next, we employed the tandem mCherry-EGFP-LC3B reporter to directly evaluate autophagic flux. In control cells, BafA1 treatment led to accumulation of yellow puncta, consistent with blocked autophagosome-lysosome fusion. However, Oc-EV-treated cells exhibited an overall marked reduction in LC3B puncta, and BafA1 failed to restore autophagosome accumulation to control levels (Fig. 4I), supporting that Oc-EVs impaired autophagosome formation and consequently disrupted autophagic flux. Collectively, these results demonstrated that Oc-EVs induced autophagic dysfunction primarily by suppressing autophagosome generation, thereby leading to impaired autophagic flux in neuronal cells.

Inhibition of miR-4433b-3p restores autophagy and cognitive function impaired by Oc-EVs

To assess whether miR-4433b-3p contributed functionally to the cognitive impairments, we stereotactically delivered a specific antagomir to the hippocampus of mice previously treated with Oc-EVs (Fig. 5A and S3B). Prior to functional assessment, RNase and protease protection assays were performed on Oc-EVs to confirm the sub-vesicular localization of miR-4433b-3p, confirming its encapsulation within Oc-EVs (Fig. S3C).

Fig. 5. Inhibition of miR-4433b-3p ameliorating cognitive deficits and restoring neuronal autophagy impaired by Oc-EVs.

A Experimental timeline. Created with BioRender.com. B–H Behavioral assessments in mice injected via tail vein with Oc-EVs and treated with stereotactic injection of miR antagomir and NC (n = 8 per group). B MWM training latency. C Time in target quadrant and platform crossings in the probe trial. D Alternation percentage in the Y-maze test. E Discrimination index for NOR. F, G Trajectories and quantification of center zone activity in the OFT. H Immobility time in the TST. I qRT-PCR measurement of miR-4433b-3p expression in brain tissues across different experimental groups (n = 3 per group). J Identification of neuronal survival base on Nissl staining in the hippocampus and quantification of neuronal density. K TEM of mouse brain tissue after antagomir injection; red arrows indicate autophagosomes. Red arrows indicate autophagosomes. Autophagosomes are quantified using 10-15 neurons per mouse from three independent mice. Data are presented as mean autophagosomes per neuron ± SEM. Statistical analyses are performed by two-way ANOVA with Turkey’s multiple comparison test. P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

Mice receiving the miR-4433b-3p antagomir exhibited a marked restoration of cognitive and affective functions (Fig. 5B–H). In the MWM, antagomir-treated mice showed a significantly reduced escape latency and spent more time in the target quadrant during probe trials, indicating enhanced spatial learning and memory (Fig. 5B, C). Consistently, both the Y-maze tests and NOR demonstrated improved short-term memory performance, as reflected by an increased discrimination index and higher spontaneous alternation rate, respectively (Fig. 5D, E). Furthermore, behavioral assays of emotional state revealed that antagomir treatment elevated the duration and frequency of central zone exploration in the OFT (Fig. 5F, G) and reduced immobility time in the TST (Fig. 5H), signifying attenuated anxiety- and depression-like behaviors.

To determine whether miR-4433b-3p mediated Oc-EV-induced neuronal dysfunction, we examined the effects of its inhibition in vivo. Antagomir administration effectively reduced hippocampal miR-4433b-3p levels (Fig. 5I), confirming successful knockdown. SA-β-gal staining revealed a marked reduction in senescent cells, while Nissl staining demonstrated improved neuronal survival in the hippocampus (Fig. 5J and S3D), indicating partial restoration of neuronal integrity. Given thar Oc-EVs impaired autophagy, TEM analysis further assessed increased number of autophagosomes following antagomir treatment (Fig. 5K), suggesting recovery of autophagosome formation. Consistently, WB analysis demonstrated reduced p62 accumulation and normalization of LC3BII levels (Fig. S3E), and immunofluorescence staining of the CA1 and DG regions confirmed decreased p62 deposition in antagomir-treated mice (Fig. S3F), supporting restoration of autophagic flux in vivo. Collectively, these findings functionally validated miR-4433b-3p as a critical mediator of Oc-EV-induced neuronal dysfunction and demonstrated that its inhibition could reverse both behavioral deficits and autophagy impairment.

Impaired autophagic flux and neuronal senescence are driven by TP53INP2 downregulation upon exposure to Oc-EVs

To further explore the functional relevance of miR-4433b-3p, we sought to identify downstream target genes that were not only autophagy regulators but also showed age- or disease-associated expression changes in the brain. Among the age-related changes, we focused on TP53INP2, a known autophagy-regulating protein involved in facilitating LC3B nuclear export and autophagosome membrane formation. Previous transcriptomic studies have shown that TP53INP2 was downregulated in aged oligodendrocytes and in demyelinating lesions in multiple sclerosis (MS)¹⁹, as well as in PD postmortem brains²⁰. Notably, TP53INP2 was predicted to be highly enriched in the CNS based on tissue specificity index analysis, yet its role in brain aging and autophagy regulation has not been clearly defined. Based on our sequencing data and in silico target predictions using miRBase and TargetScan, miR-4433b-3p was identified as a candidate upstream regulator of TP53INP2. Dual-luciferase reporter assays confirmed that miR-4433b-3p directly bound to the 3′ untranslated region (3′-UTR) of TP53INP2 and significantly suppressed its expression (Fig. 6A). To assess the physiological relevance of this interaction in vivo, we examined TP53INP2 expression in the hippocampus of mice subjected to miR-4433b-3p antagomir treatment, supporting the functional relevance of this miRNA-target axis in the brain (Fig. S4A). Consistently, in vitro co-culture of HT22 neurons with Oc-EVs resulted in a marked reduction of TP53INP2 protein levels, as demonstrated by both WB and immunofluorescence analyses, indicating that Oc-EV-derived miR-4433b-3p regulated its expression (Fig. 6B, C). These results corroborate the dual-luciferase findings and established TP53INP2 as a direct and functionally relevant target of miR-4433b-3p in neurons.

Fig. 6. miR-4433b-3p exacerbating Oc-EVs-induced neuronal autophagy impairment by suppressing TP53INP2.

A Dual-luciferase reporter assay confirming TP53INP2 as the target gene of miR-4433b-3p (n = 9 per group). B WB analysis and quantification of TP53INP2 after co-culture with Oc-EVs. C Immunofluorescence images of TP53INP2 in HT22 cells co-cultured with Oc-EVs. D WB analysis and quantification of autophagy-related proteins after mimic or inhibitor transfection. E Immunofluorescence staining of TP53INP2. F TEM analysis and quantification of autophagosomes (indicated by red arrows). Quantification of autophagosomes per neuron based on TEM images of hippocampal neurons. Autophagosomes are quantified using more than 30 cells per experiment from three independent biological replicates. Data are presented as mean autophagosomes per cell ± SEM. G Testing rapamycin rescues of miR-4433b-3p mimic-induced autophagy impairment. Rapamycin treatment is applied to determine whether pharmacological activation of autophagy could rescue the autophagy impairment induced by miR-4433b-3p mimics. H Testing the protective effect of miR-4433b-3p inhibition based on intact autophagic flux. BafA1 is applied to miR-4433b-3p inhibitor-treated cells to determine whether the protective effect of miR-4433b-3p inhibition is dependent on functional autophagic flux. Data are presented as mean ± SEM from three independent biological replicates. Statistical analyses are performed using a two-tailed unpaired Student’s t test (for panel B) or two-way ANOVA followed by Tukey’s multiple comparisons test, as appropriate. P < 0.05 (*), P < 0.01 (**), P < 0.001 (***), and P < 0.0001 (****).

We next explored whether miR-4433b-3p regulated autophagic activity through directly influence of TP53INP2 expression in senescent neurons. Both miR-4433b-3p mimics and inhibitor were transfected into Oc-EVs-treated cells (Fig. S4B). The results showed that miR-4433b-3p mimics effectively suppressed TP53INP2 and LC3BII expression and promoted p62 accumulation, collectively indicating impaired autophagic degradation, while miR-4433b-3p inhibitor alleviated autophagic dysfunction (Fig. 6D and S4C). Consistently, immunofluorescence showed decreased TP53INP2 in mimic-transfected cells and an increase in inhibitor-transfected cells (Fig. 6E). TEM analysis revealed that autophagosome numbers were significantly lower in per cell overexpressing miR-4433b-3p, and higher in cells treated with inhibitors, suggesting that miR-4433b-3p suppressed autophagic activity (Fig. 6F).

Pharmacological interventions further validated the role of miR-4433b-3p in regulating autophagic flux via TP53INP2. Treatment with rapamycin in cells transfected with miR-4433b-3p mimics restored TP53INP2 expression and enhanced autophagic activity, as indicated by increased LC3BII levels and decreased p62 accumulation (Fig. 6G and S4D). In contrast, treatment with BafA1 in miR-4433b-3p inhibitor-transfected cells reversed the inhibitor-induced upregulation of TP53INP2, reduced LC3BII levels, and promoted p62 accumulation, consistent with impaired autophagosome degradation (Fig. 6H and S4E).

Taken together, these results indicated that miR-4433b-3p was identified as a negative regulator of autophagy in senescent neurons through suppression of TP53INP2, highlighting that modulation of autophagy could either mitigate or worsen this dysfunction depending on the direction of intervention.

TP53INP2 overexpression reverses miR-4433b-3p-induced autophagy deficiency and synaptic damage

To assess whether TP53INP2 acted as a functional downstream effector of miR-4433b-3p, we conducted parallel experiments using AAV-mediated overexpression of TP53INP2 (Fig. 7A). Mice injected with AAV-TP53INP2 exhibited behavioral profiles highly similar to those observed in the miR-antagomir group. WB analysis confirmed successful overexpression of TP53INP2 in hippocampal tissues (Fig. 7B and S4F). Spatial learning and memory were significantly improved, as shown by reduced latency and enhanced target quadrant occupancy in the MWM (Fig. 7C, D). Y-maze analysis showed enhanced spontaneous alternation and NOR revealed increased novel object preference (Fig. 7E, F), indicating improved cognitive flexibility. Emotional behavior was also normalized, as reflected by increased central exploration in the OFT (Fig. 7G, H) and reduced immobility time in the TST (Fig. 7I).

Fig. 7. Restored TP53INP2 ameliorating cognitive deficits and neuronal axonal damage caused by cardiac EVs from aged mice.

A Experimental timeline. Created with BioRender.com. B WB analysis and quantification verifying the efficiency of adenovirus-mediated overexpression of TP53INP2 in the brain. C–I Behavioral assessments in mice treated with AAV-TP53INP2 or NC after Oc-EVs injection (n = 8 per group). C MWM training latency. D Time spent in the target quadrant and platform crossings in probe trial. E Alternation percentage in the Y-maze test. F Discrimination index for NOR. G, H Trajectories and quantification of center zone activity in the OFT. I Immobility time in the TST. J SA-β-gal staining of hippocampal sections. K Representative Golgi-staining image of dendrites with spines of hippocampus and quantification of dendritic spine density. Statistical analyses are performed by two-way ANOVA with Turkey’s multiple comparison test based on P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

Then, SA-β-gal staining revealed a reduction in senescent cell burden in the hippocampus of AAV-TP53INP2-treated mice, compared to NC (Fig. 7J). H&E staining further demonstrated that stereotactically delivered AAV-TP53INP2 alleviated structural deterioration in the hippocampus, as evidenced by improved neuronal density, indicating that TP53INP2 overexpression effectively mitigated hippocampal neurodegeneration compared to the NC (Fig. S4G). Consistently, immunofluorescence analysis revealed reduced p62 accumulation in the hippocampus of AAV-TP53INP2-treated mice, indicating restoration of autophagic function (Fig. S4H). Golgi staining further demonstrated a significant increase in dendritic spine density (Fig. 7K), indicative of synaptic structural restoration. Together, these findings established TP53INP2 as a critical downstream target of miR-4433b-3p, demonstrating that its restoration was sufficient to reverse Oc-EV-induced autophagy defects and associated neurobehavioral impairments.

Discussion

In this study, we elucidated a novel mechanism by which Oc-EVs act as systemic effectors of brain aging. Specifically, miR-4433b-3p was enriched in Oc-EVs and mechanistically impaired neuronal autophagy by targeting and downregulating TP53INP2, a critical regulator of autophagosome maturation²¹. Loss of TP53INP2 disrupted autophagic flux, promoted p62 accumulation, and accelerated neuronal senescence. Two treatment strategies for Oc-EV-treated mice, including hippocampal delivery of miR-4433b-3p antagomirs and AAV-mediated overexpression of TP53INP2, effectively mitigated cognitive decline, highlighting potential clinical targets (Fig. 8). Collectively, our findings revealed that the accumulation of miR-4433b-3p in Oc-EVs impaired autophagy by suppressing TP53INP2.

Fig. 8. Overexpression of cardiac EV-derived miR-4433b-3p accelerating brain aging by targeting TP53INP2 and impairing autophagy.

Schematic illustration of the proposed mechanism. Aging-related cardiac EVs are enriched with miR-4433b-3p, which is transported through the circulation and taken up by recipient brain neurons. This miRNA directly targets and suppresses TP53INP2, a key regulator of autophagic flux, thereby disrupting autophagosome maturation and impairing autophagic homeostasis. The resulting autophagy deficiency promotes neuronal senescence, neuroinflammation, and dendritic spine loss, ultimately leading to cognitive decline and the progression of brain aging. Created with BioRender.com.

Building on the central discovery, our study showed that aging is not a localized process but a systemic phenomenon mediated by circulating EVs. Our results revealed that Oc-EVs are not merely biomarkers but act as functional effectors that cross the BBB, accumulate in the hippocampus, and elicit cognitive impairments. More importantly, epidemiological evidence supports this association between cardiovascular and brain aging: a population-based study from the 1946 British Birth Cohort demonstrated that mid- and late- life cardiovascular risk is strongly correlated with increased brain-predicted age difference—a neuroimaging biomarker of accelerated brain aging²². These observations support the concept of inter-organ aging communication and identify vesicle-mediated molecular transfer as a key route for systemic aging signals to affect brain function^23,24. Critically, this EV accumulation is facilitated by age-related changes in the brain microenvironment, i.e., age-associated deterioration of the BBB²⁵, combined with increased expression of vesicle-binding receptors²⁶ on senescent neural cells and the pro-inflammatory CNS environment, collectively promote the preferential accumulation of cardiac EVs in the aging brain.

Previous studies have increasingly recognized EVs as long-range messengers mediating inter-organ communication in aging and diseases. Plasma-derived EVs have been shown to cross the BBB²⁷, activate astrocytes and microglia, and thereby exacerbate neuroinflammation. Gut-derived bacterial vesicles can deliver small non-coding RNAs into host neurons to regulate gene expression and influence cognition²⁸, while lung-derived EVs released during viral infection have been shown to reach the brain and trigger neuroinflammation⁸. Mesenchymal stem cell (MSC)-derived EVs carry bioactive lipids that facilitate immune signaling and intercellular communication¹⁷. While most research has focused on gut-, lung-, and immune system-derived EVs, the potential contribution of heart-derived EVs to brain aging remains largely unexplored.

Here, we propose and experimentally validate a novel cardiac-to-brain signaling axis, in which Oc-EVs enriched with miR-4433b-3p cross the BBB, accumulate in the hippocampus, and induce neuronal autophagy impairment and cognitive decline. This finding extends the paradigm of peripheral EV signaling beyond inflammatory or metabolic organs, identifying the aging heart as an active endocrine organ that contributes to CNS aging through vesicle-mediated molecular transfer.

To mechanistically explore the effects of Op-EVs, we identified miR-4433b-3p as a key aged plasma-enriched EV cargo based on sRNA-Seq analysis. To trace the systemic origin of miR-4433b-3p, we analyzed EVs isolated from aged mouse organs. The miR-4433b-3p was predominantly enriched in cardiac-derived EVs, suggesting that heart was the likely source of aging signals. This discovery expands the model of brain aging to include active contributions from heart, highlighting the importance of cardiac EV cargo in coordinating multi-organ aging trajectories. Given the increased BBB permeability during aging²⁹ and the stability of EV cargo, our results support a model in which the aged heart actively contributes to CNS aging via vesicle-mediated molecular transfer.

Mechanistically, dual-luciferase reporter assays confirmed that miR-4433b-3p directly targets TP53INP2, a protein involved in LC3BII-mediated autophagosome maturation. Functionally, downregulation of TP53INP2, either via miR-4433b-3p overexpression or exposure to Oc-EVs, led to impaired autophagic flux and increased cellular senescence, characterized by decreased LC3BII levels and p62 accumulation³⁰. These results establish TP53INP2 as a key downstream effector of miR-4433b-3p in autophagy regulation and cellular homeostasis. Given the complexity of miRNA-mediated gene regulation, TP53INP2 expression during aging is likely subjected to additional layers of post-transcriptional control, which warrant further investigation.

To evaluate the relevance of this mechanism in vivo, we stereotactically delivered a miR-4433b-3p antagomir into the hippocampus of Oc-EV-treated mice. This intervention reversed cognitive impairments and normalized behavioral performance, including improvements in memory, exploratory behavior, and depressive-like symptoms. These effects were further validated through AAV-mediated overexpression of TP53INP2, which restored dendritic spine density and reduced hippocampal senescence. These findings confirm that the miR-4433b-3p-TP53INP2 axis is both necessary and targetable in the aging brain.

Translating mechanistic insights from aging research into clinically actionable biomarkers and interventions for cognitive decline remains an urgent priority. To establish the systemic relevance of miR-4433b-3p, we analyzed aged plasma EVs in a human cohort and observed a significant age-associated increase in EV-derived miR-4433b-3p levels. Notably, miR-4433b-3p showed a robust association with NfL and a modest but statistically significant association with Aβ40, both of which are linked to brain aging. MiR-4433b-3p was previously identified as part of a serum-based 3-miRNA panel predictive of cerebrovascular disorders³¹, further supporting its clinical relevance as a circulating biomarker for age-related brain pathology. These findings establish the heart as a major source of aging signal that contributes to CNS aging, expanding the model of brain aging to include active peripheral input. Nevertheless, clinical translation must contend with key challenges—such as the heterogeneity of human aging, comorbidities, and pharmacological interactions—that are not fully recapitulated in current models. Developing clinically relevant platforms and prospective human studies will be essential to evaluate the diagnostic and therapeutic potential of EV miR-4433b-3p in age-related cognitive disorders.

While our results establish a mechanistic link between Oc-EVs and brain aging, several questions remain unanswered. At the cellular level, the specificity of EV-miR-4433b-3p signaling within the brain requires further clarification. Although pronounced effects were observed in neurons, the potential involvement of microglia and astrocytes remains to be determined. Importantly, from an experimental design perspective, PBS was used as a control in EV injection experiments; although this design allows assessment of the pathological effects of aged EVs, the inclusion of EVs derived from young mice would provide a more rigorous comparison to define age-specific effects. At the level of inter-organ communication, although our multi-omics and cross-species analyses indicate that the aged heart is a major contributor of miR-4433b-3p-enriched EVs, the upstream triggers of their release remain unclear. Elucidating whether cardiac senescence, metabolic remodeling, or inflammatory signaling drives EV secretion will be important for understanding inter-organ communication in aging. Mechanistically, an additional key limitation lies in the incomplete characterization of the TP53INP2 regulatory axis. While rapamycin functionally rescued the autophagy-suppressive phenotype induced by miR-4433b-3p mimics, we did not directly assess mTOR pathway activity or TP53INP2 subcellular localization, which are critical to establishing a causal mechanistic link. In addition, AAV-mediated TP53INP2 restoration was used as a functional approach to evaluate phenotypic rescue and may not fully reflect endogenous regulation. Future studies are needed to define the precise upstream signaling regulatory mechanisms governing TP53INP2 in aging. Finally, although we systematically evaluated the association between cardiac EV-derived miR-4433b-3p and circulating brain aging biomarkers in elderly individuals, future longitudinal and integrative studies—combining cerebrospinal fluid profiling, brain imaging, and regional transcriptomic mapping—are essential to translate miR-4433b-3p from an experimental effector to a clinically actionable biomarker and therapeutic target.

In summary, we have identified a novel cardiac EV-mediated mechanism by which peripheral aging impairs neuronal function and cognitive health. Specifically, Oc-EV-derived miR-4433b-3p suppresses TP53INP2, disrupts autophagy, and promotes neurodegeneration. This EV-mediated mechanism links heart aging to central neurodegeneration, highlighting miR-4433b-3p as both a biomarker and a potential therapeutic target. Our findings provide novel insights into the cardiac communication networks that orchestrate brain aging, suggest that modulating EV cargo may provide a promising strategy to counteract age-associated cognitive decline.

Methods

Clinical participants

The current study was conducted on a cohort of 197 patients from Shandong Provincial Hospital between June 2023 and September 2024. The patients were classified into three age groups: Group I of young adults (n = 62; 20–40 years old), Group II of middle-aged adults (n = 64; 41–60 years old), and Group III of senior adults (n = 71; age > 60 years old). The proportion of male was 31.4%, 30.9%, and 36% in these three groups, respectively. Individuals with cancer, HIV, COVID 19, any form of brain illness, or positive results for hepatitis B or C surface antigen antibodies, were excluded in this study. Demographic details of the participants were provided in Table S1. Peripheral blood was collected from each participant and placed into EDTA-coated anticoagulant tubes and processed within 4 h of collection. Plasma was separated by centrifugation at 1150 × g for 15 min, aliquoted into 500 μL portions, and stored at −80 °C until further analysis. For brain aging biomarker measurement, frozen plasma samples were thawed at 4 °C and then centrifuged at 10,000 × g for 10 min to remove residual cell debris. The resulting supernatant was then divided into 75 μL aliquots for quantification of brain aging markers using a single-molecule detection system (AXL-2000, Lychix, Suzhou, China).

This study was approved by the Ethics Committee of the Provincial Hospital affiliated with Shandong First Medical University for Biomedical Research Involving Human Subjects (NSFC No. 2022-511). All ethical regulations relevant to human research participants were followed. All procedures were conducted in accordance with the ethical principles of the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment.

Experimental animals

The 10- and 21-month-old wild-type (WT) C57BL/6 J mice were purchased from Jinan Xingkang Bio-tech (Jinan, China) and housed in a temperature-controlled facility under a 12-h light/12-h dark cycle with ad libitum access to sterile food and water. The animals were allowed to acclimate for 1 week prior to the initiation of any experimental procedures. Mice were deeply anesthetized using isoflurane delivered via a small animal anesthesia system (induction: 3–5%, maintenance: 1–1.5%, oxygen flow rate: 0.2–0.5 L/min). Blood samples were collected through cardiac puncture and placed into EDTA-coated anticoagulant tubes. Mice were then euthanized by continued isoflurane overdose, and tissue including the brain, kidney, liver, lung, spleen, and heart were harvested. Plasma was isolated by centrifuging the blood samples at 1150 × g for 15 min and stored at −80 °C until further use. All experimental procedures were approved by the Animal Care and Use Committee of Shandong Provincial Hospital affiliated to Shandong First Medical University and compiled with institutional guidelines and all relevant ethical regulations for animal use.

Animal treatment

Three plans were conducted to treat animals. The 10-month-old C57BL/6 J mice were used as recipients of aged plasma EVs (Op-EVs). This age was chosen based on preliminary experiments comparing mice of 7 to 12 months old for cognitive function and neuronal senescence markers (see Results), as well as alignment with human middle-age brain aging.

Plan A

At the age of 10 months, mice (male-to-female ratio 1:1) were randomly assigned to two groups (n = 8 per group): the Op-EVs group and the PBS control group. Op-EVs were isolated from 21-month-old donor mice and quantified using a BCA Protein Analysis Kit (PC0020, Solarbio, China). Mice in the Op-EVs group received tail vein injections of 200 µg protein-equivalent EVs diluted in PBS every other day for a total of 4 weeks. Mice in the PBS control group were administered an equal volume of PBS following the same schedule as that for the Op-EVs group.

Plan B

At the age of 10 months, mice (male-to-female ratio 1:1) were randomly assigned to four groups (n = 8 per group): PBS control, Oc-EVs, Oc-EVs+ miR-antagomir-NC, and Oc-EVs+ miR-antagomir (GenePharma, China). Mice in the Oc-EVs+ miR-antagomir and Oc-EVs+ miR-antagomir-NC groups first received bilateral stereotactic injections of 2.0 μL of 50 μM miRNA antagomir or negative control solution into the dentate gyrus of the hippocampus using a small high-precision stereotaxic instrument (SA-100, Yuyan, China). Mice were intraperitoneally injected with tribromoethanol (1.25%, 0.02 mL/g body weight) (JT0781, JITIAN BIO, China). A midline scalp incision was made, and a craniotomy was drilled above each injection site. The stereotactic coordinates were as follows: anteroposterior (AP) −2.0 mm from Bregma, mediolateral (ML) ± 1.0 mm, and dorsoventral (DV) −2.0 mm³². The injection rate was maintained at 0.2 μL/min, and the needle was maintained in place for an additional 5 min post-injection to ensure full absorption before withdrawal.

Following post-surgical recovery, mice in the Oc-EVs, Oc-EVs+ miR-antagomir-NC, and Oc-EVs+ miR-antagomir groups received tail vein injections of 200 µg protein-equivalent Oc-EVs diluted in PBS every other day for 4 weeks. The PBS control group received an equal volume of PBS following the same injection schedule as that for the other three treatment groups.

Plan C

At 10 months, mice (male-to-female ratio 1:1) were randomly divided into 4 groups (n = 8 per group): PBS control, Oc-EVs, Oc-EVs+ AAV-TRP53INP2-NC, and Oc-EVs+ AAV-TRP53INP2. AAV9 serotype viruses containing mouse TRP53INP2, AAV-EGFP, and virus of Scr control were separately generated from Obio Technology, China. Mice were intraperitoneally injected with tribromoethanol (1.25%, 0.02 mL/g body weight) (JT0781, JITIAN BIO, China). Purified AAV9, titers of 3.84 × 10¹² viral genomes per mL (vg/mL), were stereotactically injected into the hippocampus of mice at the following coordinates: AP, −2.0 mm; ML, ± 1.0 mm; DV, -2.0 mm, with a rate of 400 nL/min and a total volume of 1 μL unilaterally. The injection of Oc-EVs into mice was the same as that described in Plan B. After the completion of behavioral testing, all mice were deeply anesthetized with inhaled isoflurane and euthanized prior to brain tissues collection for further analysis.

Plasma and tissue EV extraction and characterization

Blood samples from human participants and mice were centrifuged at 1150 × g for 20 min at 4 °C to obtain plasma. Plasma was aliquoted (500 μL per tube) and stored at −80 °C until EV isolation. EVs were isolated from 500 μL plasma using the Total Exosome Isolation Kit (UR52151, Umibio Biotechnology, China) according to the manufacturer’s instructions³³. Briefly, plasma samples were sequentially centrifuged at 3000 × g for 10 min and 12,000 × g for 10 min at 4 °C to remove cellular debris and insoluble impurities. The supernatant was incubated with pre-cooled Solution A and centrifuged at 12,000 × g and 4 °C for 20 min to eliminate contaminating proteins. Subsequently, Solution B was added, and the mixture was incubated at 4 °C for at least 30 min. EVs were pelleted by centrifugation at 12,000 × g and 4 °C for 15 min, washed once, and resuspended in 200 µL of 1× PBS. The crude EV preparation was further purified using an Exosome Purification Filter (EPF) column by centrifugation at 3000 × g and 4 °C for 10 min. The filtrate containing purified EVs was collected, aliquoted, and stored at −80 °C for downstream analyses.

For tissue-derived EV isolation, major organs—heart, liver, spleen, lung, kidney, and brain—were harvested from 21-month-old and 3-month-old mice and processed using the Tissue Exosome Isolation Kit (UR52161, Umibio Biotechnology, China). Briefly, freshly isolated tissues (about 0.1 g per organ) were placed on ice in sterile dishes and finely minced into approximately 1–3 mm³ fragments. The tissue pieces were transferred to centrifuge tubes, washed with at least 10 volumes of cold 1× PBS, gently mixed, and centrifuged at 300 × g for 10 min to remove blood residues and loosely associated cells. The tissue pellets were then incubated with Solution A2 at 37 °C for 30 min with gentle mixing every 5 min to facilitate EV release from the extracellular space while minimizing cell disruption. The digestion mixture was centrifuged at 8000 × g for 10 min, and the supernatant was further clarified by centrifugation at 12,000 × g and 4 °C for 10 min to remove residual tissue debris. One-quarter volume of Solution B2 was added to the collected supernatant, followed by incubation at 4 °C for at least 1 h. EVs were pelleted by centrifugation at 12,000 × g and 4 °C for 30 min, washed once, and resuspended in 200 μL of 1× PBS. The crude EVs were further purified using an EPF column following the same purification procedure as described for plasma-derived EVs. Total RNA was then extracted from the isolated EVs for further analysis.

For morphological characterization of EVs, 0.02 µm-filtered PBS-suspended exosome samples were deposited onto carbon-coated 200-mesh copper grids and incubated at room temperature for 5 min. After removing excess liquid with filter paper, grids were stained with uranyl acetate for 10 s, dried at room temperature for 30 min, and subsequently imaged using a transmission electron microscope (TEM) (H7700, Hitachi, Japan).

To determine the particle size distribution and concentration, EV samples were diluted in 500 µL of 0.02 µm-filtered PBS and analyzed using a NanoSight NS300 instrument (Malvern Instruments, UK). Measurements were conducted using Nanoparticle Tracking Analysis (NTA) software (version 3.4, build 3.4.4), with the camera level set to 11 and the detection threshold at 5.

Tracking of EVs in vivo and in vitro

To trace the biodistribution and cellular uptake of EVs, both in vivo and in vitro tracking experiments were performed using two fluorescent dyes: DiR and PKH67³⁴ (UR21017 and UR52303, Umibio Biotechnology, China, respectively).

For in vivo imaging, EVs were labeled with DiR by incubating them at 37 °C for 30 min in the dark. After staining, the reaction was quenched with an equal volume of 1% BSA in PBS, and unbound dye was removed through a second round of EV isolation. A total of 200 μg of DiR-labeled EVs was administered via tail vein injection into recipient mice every other day for 4 weeks. Control mice received equivalent volume of PBS. At the endpoint, animals were euthanized under isoflurane anesthesia, and major organs (brain, liver, kidney, and heart) were collected for fluorescent imaging using an in vivo imaging system (IVIS Lumina II, PerkinElmer).

For brain-targeted localization, EVs were labeled with PKH67 by adding the dye to EV suspensions, followed by 1-min vortexing and incubation at 37 °C for 10 min in the dark. The staining reaction was stopped using 1% BSA/PBS, and residual dye was removed by subsequent purification. Then, 200 μg of PKH67-labeled EVs was injected into the tail vein of recipient mice. Four hours after injection, the animals were euthanized by inhalation of isoflurane prior to brain tissues collection for assessment of EV distribution at the cellular level.

For in vitro tracking, PKH67-labeled EVs were co-cultured with cultured neuronal cells under standard conditions, and their internalization was assessed using fluorescence microscopy.

Morris water maze test

The Morris water maze was used to assess spatial learning and memory capability³⁵. Mice were trained in a circular pool (120 cm in diameter) filled with opaque water (22 ± 1 °C) and divided into four quadrants with distinct visual cues. The test consisted of a 5-d training phase followed by a probe trial on day 6. During the training phase, the first day was the platform (10 cm in diameter) visibility period, and from days 2 to 5, the platform was submerged 2 cm below the surface in a fixed quadrant. In each training trial, mice were allowed to swim for up to 60 s. If a mouse located and climbed onto the platform, it was allowed to remain there for 10 s before being removed. On day 6 (the start of probe trial), the platform was removed, and each mouse was released from the quadrant opposite the previous platform location for a 60-s free swim. The time spent and number of crossings made in the target quadrant were recorded to evaluate memory retention. Escape latency and time spent in target quadrant were recorded using an overhead tracking system.

Y-maze test

Y mazes were used to evaluate the spatial memory, learning ability, and exploration behavior in mice³⁶. Mice were acclimated to the testing room for 1 h prior to the test. Each mouse was placed at the center of the maze and allowed to explore freely for 5 min while being recorded using a video tracking system. Spontaneous alternation behavior was defined as successive entries into three different arms of the Y maze, and the alternation percentage was calculated as: [(number of alternations) / (total arm entries - 2)] × 100%. The maze was cleaned with 75% ethanol between trials, and the test was conducted in a quiet, dimly lit environment.

Open field test

The open field test was conducted to assess locomotor activity, exploratory behavior, and anxiety-like responses of mice in a novel environment³⁷. After 30 min of habituation, each mouse was placed in a 50 × 50 × 40 cm arena and allowed to explore freely for 5 min. Behavioral parameters, including the number of entries into the central area, time spent in the center, and total traveled distance, were recorded using a video tracking system and analysis software. The arena was cleaned between trials to avoid olfactory interference.

Novel object recognition test

To assess recognition memory, mice underwent the novel object recognition test³⁸. On day 1, animals were allowed to freely explore an open-field arena for adaptation. On days 2 and 3, two identical objects were placed in the arena, and mice were allowed to explore for 10 min. After a 2-h retention interval in their home cages, mice were returned to the same arena containing one familiar and one novel object. Mice were given 5 min to explore, and only those that spent at least 10 s interacting with the objects were included in the analysis. The novel object recognition index (NOI) was calculated as: NOI = (T _new - T _familiar) / (T _new + T _familiar), where T _new and T _familiar represented the time spent exploring the novel and familiar objects, respectively.

Tail suspension test

Depressive-like behavior was evaluated using the tail suspension test³⁹. Mice were suspended by the tail using adhesive tape fixed approximately 1 cm from the tip of the tail, in a 40 cm-high enclosure. Each trial lasted for 6 min, during which immobility time was recorded over the final 5 min. Mice were considered immobile when they hung passively without active movement.

SA-β-gal staining

SA-β-gal staining was performed using a commercial β-galactosidase detection kit (Solarbio, China) following the established protocol¹⁰. Stained images were captured under a Pannoramic scan.

Nissl staining

Paraffin-embedded brain sections were dewaxed and rehydrated, then stained with Nissl staining solution (GP1036, Servicebio, China) following the established protocol⁴⁰. Nissl staining labels Nissl bodies, which consist primarily of rough endoplasmic reticulum and ribosomes and are abundantly present in neuronal cell bodies, serving as a classical histological marker of neuronal integrity and viability. Nissl bodies appeared as dark blue granules, neuronal nuclei were light blue, and the background was stained pale blue. The number of Nissl-positive neurons was quantified from at least three randomly selected scans in the hippocampal region using ImageJ software (National Institutes of Health, USA).

Hematoxylin and Eosin staining

Paraffin sections were dewaxed and rehydrated through a graded ethanol series, then the sections underwent staining with a hematoxylin and eosin (H&E) HD constant dye kit (G1076, Servicebio, China) and a Masson dye solution set (G1006, Servicebio, China) according to the manufacturer’s instructions⁴¹. The nuclei appeared blue and the cytoplasm pink. Stained sections were scanned using a light microscope (Nikon Eclipse 100), and histological features were evaluated by image analysis.

Transmission electron microscopy

To assess autophagosome number and morphology, mouse brain tissues were fixed in electron microscopy fixative, post-fixed with 1% osmium tetroxide, dehydrated, and embedded in 812 resin. For TEM analysis of cultured cells, cells were collected after the indicated treatments and processed using the same procedures as described for tissue samples to ensure consistency across experimental conditions. Autophagosomes were identified based on established ultrastructural criteria, defined as double-membrane-bound vesicles enclosing cytoplasmic material or organelles, and clearly distinguished from autolysosomes and other single-membrane vesicular structures. Quantification was performed using ImageJ software (National Institutes of Health, USA). Autophagosomes were manually counted in randomly selected cell sections by an investigator blinded to the experimental groups. The number of autophagosomes was quantified per cell section and used for subsequent statistical analysis.

Golgi staining

Golgi staining was performed utilizing a Golgi staining kit (GP1152, Servicebio, China) following the instructions provided by the manufacturer⁴². Dendritic spines were imaged using a light microscope, and spine density and morphology were quantified using ImageJ software (National Institutes of Health, USA).

Cell culture

Timed-pregnant C57BL/6 J mice (embryonic day 17–18) were used for primary neuronal cultures, and hippocampus were dissected in ice-cold HBSS, enzymatically dissociated with 0.25% trypsin-EDTA at 37 °C for 15 min, and gently triturated. Dissociated hippocampal neurons were plated onto poly D-lysine-coated coverslips or dishes at 20,000–100,000 cells cm^-2 in culture medium (A3582901, Thermo Fisher Scientific, US) supplemented with B27 (17504044, Thermo Fisher Scientific, US), 2mMm GlutaMax (35050061, Thermo Fisher Scientific, US), and 100 U/mL penicillin/streptomycin and maintained at 37 °C in 5% CO₂ with half-medium changes every 3–4 d. Experiments were performed for 7–14 d in vitro. Primary neuron was co-cultured with 200 μg protein-equivalent EVs isolated from the heart of 21-month-old mice for 24 h.

The HT22 mouse hippocampal neuronal cell line (CL-0697, Pricella, China) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin solution to prevent bacterial contamination. Cells were maintained at 37 °C in a humidified incubator with 5% CO₂. For experiments, cells were seeded into 6-well plates and allowed to reach approximately 70% confluence. They were co-cultured with Oc-EVs isolated from the heart of 21-month-old for 24 h.

Cell transfection

For cell transfection experiments, 50 pmol of miRNA mimics or miRNA inhibitor (GenePharma, China) were transfected into HT22 cell line using Lipofectamine 3000 (L3000015, Thermo Fisher Scientific, USA) following the manufacturer’s instructions⁴³. Total RNA and protein were extracted 48 h after transfection. The two sequences of miR-4433b-3p mimics were 5’-CAGGAGUGGGGGGUGGGACGU-3’ and 5’-GUCCCACCCCCCACUCCUGUU-3’, and the sequence of miR-4433b-3p inhibitor was 5’-ACGUCCCACCCCCCACUCCUG-3’.

Dual-luciferase reporter assay

The interaction between miR-4433b-3p and 3′UTR of TP53INP2 was predicted using miRDB (http://mirdb.org/) and RNAhybrid (https://bibiserv.cebitec.uni-bielefeld.de/rnahybrid/) databases. Based on the predicted binding site, WT and seed region-mutated (MUT) fragments of TP53INP2 were cloned into the pmirGLO vector (GenePharma, China). HEK-293T cells were seeded in 24-well plates and co-transfected with 500 ng of either TP53INP2-WT or -MUT reporter plasmids and 50 nM has-miR-4433b-3p mimics or negative control mimics (GenePharma, China) using Lipofectamine 3000. In 48 h, luciferase activity was measured with the Dual-Luciferase^® Reporter Assay System (11402ES60, YEASEN, China). Firefly luciferase activity was normalized to Renilla luciferase. All experiments were performed in triplicate, and results were expressed as mean ± standard deviation (SD).

RNase and protease protection assay

To determine whether miRNAs were encapsulated within EVs, an RNase protection assay was performed as previously described⁴⁴. Purified EVs were divided into six treatment groups: (1) untreated control; (2) RNase A (EN0531, Thermo Fisher Scientific, USA); (3) proteinase K (AM2546, Thermo Fisher Scientific, USA); (4) RNase A + proteinase K; (5) Triton X-100 (0.1%); and (6) RNase A + Triton X-100. Specifically, EVs were incubated with RNase A at 37°C for 30 min to degrade surface-exposed RNAs. For membrane disruption, EVs were treated with 0.1% Triton X-100 prior to RNase A digestion. For protease treatment, EVs were incubated with proteinase K at 37 °C for 30 min to digest surface-associated proteins. In the combined treatment group, RNase A and proteinase K were added simultaneously under the same conditions. For membrane disruption, EVs were pretreated with 0.1% Triton X-100 for 10 min at room temperature prior to RNase digestion. Following enzymatic treatments, RNase activity was terminated by adding RNase inhibitor, and proteinase K was inactivated according to the manufacturer’s instructions. Total RNA was subsequently extracted using the miRNA isolation kit, and miR-4433b-3p abundance was quantified by RT-qPCR. Resistance to RNase digestion in the absence of detergent, and sensitivity upon membrane disruption, were interpreted as evidence of luminal localization of miRNA within EVs.

Autophagic flux assay via pmCherry-EGFP-LC3B

To monitor autophagic flux, primary neurons or HT22 cells were transfected with the tandem fluorescent-tagged LC3B plasmid pmCherry-EGFP-LC3B (Research Cloud Biology, China) using Lipofectamine 3000 according to the manufacturer’s instructions. Upon autophagosome-lysosome fusion, the acidic lysosomal environment quenched EGFP fluorescence while preserving pmCherry fluorescence, resulting in red-only puncta that indicated successful autophagic degradation. In contrast, yellow puncta (pmCherry⁺/EGFP⁺) represented autophagosomes⁴⁵. Quantification of puncta was performed using ImageJ software (National Institutes of Health, USA).

RNA isolation and quantitative real-time PCR

Total RNA was extracted from cells and tissues using TRIzol reagent (AG21101, Accurate Biotechnology, China) and exosomal RNA was isolated with the ExoRNeasy Serum/Plasma Maxi Kit (Qiagen, Frankfurt, Germany). cDNA synthesis was performed using Evo M-MLV RT Premix (AG11706, Accurate Biotechnology, China) for mRNA and miRNA First Strand cDNA Kit (Stem-loop) (AG11742, Accurate Biotechnology, China) for miRNA. Quantitative real-time PCR (qRT-PCR) was carried out using the SYBR^® Green Pro Taq HS Premixed qPCR Kit (AG11701, Accurate Biotechnology, China) with gene-specific primers. GAPDH was used as internal control for mRNA normalization, and both U6 and cel-miR-39-3p were used as internal controls for miRNA normalization. Relative expression was calculated using the 2^-ΔΔCT method. Primer sequences were listed in Table S3.

Protein extraction and western blot analysis

Cells or tissues were lysed with RIPA Lysis Buffer with the addition of protease phosphatase inhibitor mixture. The total proteins in each sample were quantified by the BCA Protein Assay Kit (PC0020, Solarbio, China) and separated by a 12% SDS-PAGE. Then, the samples were transferred to the PVDF membrane (ISEQ00010, Merck Millipore, China) through electrophoretic blotting. The membranes were sealed with nonfat milk powder and immunoblotted with primary antibody. GAPDH was used as an endogenous control, and anti-rabbit IgG (1:5000, 30000-0-AP, Proteintech, China) and anti-mouse IgG (1:5000, SA00001-1, Proteintech, China) antibodies were used as secondary antibodies. The primary antibodies used in this experiment included anti-CD9 (1:1000, ab307085, Abcam, USA), anti-TSG101 (1:1000, ab125011, Abcam, USA), anti-ALIX (1:1000, ab186429, Abcam, USA), anti-GM130 (1:2000, ab52649, Abcam, USA), anti-CD63 (1:2000, A5271, Abclonal, China), anti-Calnexin (1:1000, ab22595,Abcam, USA), anti-TP53INP2 (1:1000, PA5324, Abmart, China), anti-LC3B (1:6000, 14600-1-AP, Proteintech, China), anti-p62 (1:2000, ab109012, Abcam, USA), anti-p21 (1:1000, PA5-14456, Invitrogen, USA), anti-TNF-α (1:5000, 60291-1-lg, Proteintech, China), and anti-GAPDH (1:5000, 60004-1-lg, Proteintech, China).

Immunofluorescence staining

Immunofluorescence staining was performed following the previous protocol⁴⁶. The antibodies used in this study included anti-Neun (1:500, ab177487, Abcam, USA), anti-MAP2 (1:500, ab183830, Abcam, USA), anti-TP53INP2 (1:200, LS-C119137, LSBIO, USA), anti-p21 (1:500, GT1032, Invitrogen, USA), Alexa Fluor^® 555 (1:500, ab150078, Abcam, USA), and Alexa Fluor^® 488 (1:500, ab150113, Abcam, USA). Fluorescence intensity and positive cell quantification were performed using ImageJ software (National Institutes of Health, USA).

Small RNA sequencing and analysis

To identify candidate plasma exosomal miRNAs associated with brain aging, small RNA sequencing (sRNA-Seq) was performed on plasma-derived EV. Samples were obtained from six elderly individuals (≥80 years old) and six young donors (≤30 years old) randomly selected from the study cohort. Clinical characteristics of the subjects were summarized in Table S1. Total RNA was extracted from isolated EVs, and sequencing libraries were prepared using the Multiplex Small RNA Library Prep Set for Illumina^® (San Diego, CA, USA) following the manufacturer’s protocol. Sequencing was conducted by Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). Clean reads were filtered based on length distribution to enrich for sRNAs. Reads were aligned to the human reference genome (GRCh38) using Bowtie. Known miRNAs were annotated by alignment to miR Base v20.0, and novel miRNAs were predicted using a combination of miR Evo and miRDeep2. Differential expression analysis between the elderly and young groups was performed using the DESeq R package, and miRNAs with a log₂ (fold change) > 1 and adjusted p-value (FDR) < 0.05 were considered significantly differentially expressed. Subsequent validation in a larger patient cohort confirmed the findings and ensured robust support for the conclusions.

Statistics and reproducibility

Statistical analyses of clinical data were performed using R software (version 3.4.3, The R Foundation) and EmpowerStats (www.empowerstats.com; X&Y Solutions, Boston, MA, USA). Continuous variables were expressed as either median (interquartile range) or mean ± SD, as appropriate. Group comparisons were conducted using the Kruskal-Wallis H test or one-way ANOVA. Correlations between exosomal miRNA levels and plasma brain aging biomarkers were assessed using Spearman correlation and multivariable linear regression. For animal and cell-based experiments, statistical analyses were performed using GraphPad Prism version 10.2 (RRID: SCR_002798). Comparisons between two groups were made using unpaired two-tailed t-tests, while multiple groups were analyzed using one-way or two-way ANOVA followed by appropriate post hoc tests. Differences were considered statistically significant based on P < 0.05.

Biological replicates refer to independent samples derived from different animals or experimental units, whereas technical replicates represent repeated measurements of the same sample. The number of biological replicates (n) and statistical tests used are indicated in the figure legends. All experiments were independently repeated at least 3 times with similar results.

Ethics approval and consent to participate

Human studies were conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Shandong Provincial Hospital, affiliated with Shandong First Medical University (approval No. 2022-511). Written informed consent was obtained from all participants. All animal experiments were reviewed and approved by the Laboratory Animal Management and Ethics Review Committee of Shandong Provincial Hospital (approval No. 2022-511).

Consent for publication

All authors consented for publication.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Q.C. and Z.C. designed and performed the experiments; Q.C. and M.L. analyzed the data and wrote the manuscript; J.H., S.W., Q.C., S.Y., H.C. and Q.Y. performed the clinical studies; Y.W., Y.J., G.L. and Z.L. supervised the study. All authors read and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Gregory Lavieu and Dario Ummarino.

Funding

This work was supported by the National Natural Science Foundation of China (82272414, 82572653), National Natural Science Foundation of China (82502804), the Natural Science Foundation of Shandong Province (ZR2025QC1673), and the Taishan Scholars Program of Shandong Province (tsqnz20240852).

Data availability

The sRNA sequencing data have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE329277 and will be publicly available upon publication. Source data underlying all figures are provided as Supplementary Data.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10332-7.