Extracellular vesicles released by senescent myoblasts affect recipient cells via miRNA-target interactions

Senescent myoblasts release factors that impair myotube differentiation

Senescent cells are known to release SASP factors that can influence their cellular environment [7]. We have previously established a model of myoblast senescence using the chemotherapeutic agent bleomycin, which acts by rapidly causing single- and double-stranded DNA breaks [17]. Treating C₂C₁₂ myoblasts with 14 μM bleomycin for 12h induces a senescence phenotype 48h after the damaging stimulus is removed, as evidenced by prominent SA-β-gal staining (Figure 1A), cell cycle arrest (Figure 1B), and elevated Cdkn1a mRNA expression (Figure 1C). To assess if these myoblasts produce SASP-like effects on neighbouring cells, we utilized transwell inserts with a 0.4-μm membrane pore which permits the exchange of extracellular factors between cell populations without direct cell-cell interaction (Figure 1D). Senescent (SEN) or non-senescent (NSEN) myoblasts cultured onto 0.4-μm transwell inserts were co-incubated with untreated myoblasts for 6 days in low-serum conditions to stimulate myotube formation. Over this period, NSEN inserts were replaced every 48h to minimize transwell myoblast fusion. After 6 days of differentiation, myotubes that were co-incubated with SEN myoblasts had elevated senescent gene expression, including Trp53 (+18%; p < 0.05; Figure 1E) and Cdkn1a (+19%; p < 0.05; Figure 1F) mRNA. Notably, these myotubes also exhibited a 27% decrease in diameter (p < 0.05; Figure 1G–1H), a 30% reduction in surface area (p < 0.05; Figure 1I), and a 50% decrease in myonuclear index (p < 0.05; Figure 1J) relative to myotubes co-incubated with NSEN myoblasts. Taken together, these findings suggest that senescent myoblasts produce a secretome that elicits a SASP-like effect on adjacent cells, ultimately increasing senescent gene expression and impairing myotube formation.

Figure 1. Transwell culture with senescent myoblasts impairs myotube formation. (A) Representative phase contrast microscopy images of myoblasts stained for SA-β-gal at pH 6.0, captured 48h after dosing with vehicle (VEH) or bleomycin (BLEO). (B) Flow cytometry histograms of myoblasts stained with propidium iodide 48h after treatment. The fluorescent intensity of propidium iodide determines the cell cycle phase. (C) Relative mRNA expression of Cdkn1a (p21) following VEH or BLEO treatment. (D) Schematic diagram of myotubes exposed to SEN/NSEN myoblasts in a 0.4-μm transwell insert. This allows for intercellular communication exclusively through factors released into the cell culture media. Relative mRNA expression of senescence genes (E) Trp53 and (F) Cdkn1a in myotubes after 6 days of 0.4-μm transwell culture with NSEN or SEN myoblasts. (G) Representative immunofluorescence microscopy images of myotubes after co-culture with NSEN or SEN myoblasts. Myotubes are stained for MyHC (green) and nuclei (DAPI; blue). Graphical quantification of (H) myotube diameter, (I) myotube surface area, and (J) myonuclear index.

Significantly altered miRNA expression profile within senescent myoblast-derived EVs

Recent work has implicated extracellular vesicles (EVs) as having a strong effect on C₂C₁₂ myoblast differentiation [18]. Given the pore size in the transwell inserts only permits the exchange of particles smaller than 400 nm, we hypothesized that small EVs are implicated in the observed intercellular signalling. EVs were isolated from the conditioned media (CM) of SEN or NSEN myoblasts after 48h of culturing (Figure 2A). Following isolation, the samples were visually examined through transmission electron microscopy, which morphologically identified the presence of EVs (Figure 2B). Isolates also had positive protein markers of vesicle biogenesis (Alix, TSG101, Syntenin), tetraspanins (CD9, CD63), and vesicle membranes (HSP70) (Figure 2C). The contaminant marker Calreticulin was undetectable via immunoblotting, indicating that samples contained very minimal apoptotic bodies. Particle size evaluation using nanoparticle tracking analysis revealed that the median diameter was ~100 nm, and most particles were within the range of 30 – 200 nm and therefore consistent with expected EV diameters (Figure 2D).

Figure 2. MiRNA expression profile of EVs released by senescent myoblasts. (A) Schematic diagram of the isolation of extracellular vesicles from the conditioned media of VEH- or BLEO-treated myoblasts. (B) Transmission electron micrographs of isolated EVs. Scale bars represent 100 nm and 50 nm. (C) Representative western blots of EV-specific markers (Alix, TSG101, Syntenin, CD9, CD63, HSP70) and endoplasmic reticulum contamination markers (Calreticulin) measured in the cell lysate (CL), post-2000g spin discard (2K), post-20,000g spin discard (20K), and the concentrated EV pellet (100K). Blots have been cropped to improve clarity and conciseness of the results. (D) EV size and concentration distribution, as measured by nanoparticle tracking analysis. (E) Principal component analysis (PCA) of SEN-derived EVs (SEV; cyan) and NSEN-derived EVs (NSEV; orange). (F) Volcano plot of differentially expressed (DE) miRNAs based on adj. p < 0.05 and FC ≥ |1.5|. Red dots indicate significant DE miRNAs, while green dots are non-significant with FC ≥ |1.5| and adj. p > 0.05. (G) Heatmap showing all DE miRNA (adj. p < 0.05 and FC ≥ |1.5|) with hierarchal clustering of SEV and NSEV samples.

EVs, among other factors, are packaged with small, non-coding RNAs termed microRNAs (miRNAs), which are known to target and post-transcriptionally regulate messenger RNA (mRNA) [14, 15]. Therefore, we assessed the miRNA expression profile of senescent-derived EVs (SEV) and non-senescent-derived EVs (NSEV). Total RNA isolated from EVs was analyzed with a miRNA 4.0 GeneChip array. The principal component analysis of normalized expression data revealed separate clustering of the SEV and NSEV miRNA profiles, with a cumulative variance of 41% across the first two principal components, suggesting distinct genetic profiles between the groups (Figure 2E). We identified 1906 mouse miRNA transcripts, and differential expression analysis (DEA) revealed 22 significant (adj. p < 5% and FC ≥ |1.5|) miRNAs (Figure 2F). A heatmap displays the differentially expressed (DE) miRNAs, with 10 upregulated and 12 downregulated in the SEV group relative to NSEV (Figure 2G). To the best of our knowledge, the following miRNAs have not been previously linked to cellular senescence: [upregulated in SEVs] miR-5112, miR-301a-3p; [downregulated in SEVs] miR-669k-5p, miR-669f-5p, miR-466j, miR-466m-5p, miR-669m-5p, miR-466f-5p, miR-466h-5p, miR-3470b, miR-3535, miR-351-5p. Other DE miRNAs have been linked to cell cycle regulation (miR-344d-3p and miR-30c-2-3p) or aging (miR-1946b), but not directly to senescence [19–21]. As miRNAs tend to be evolutionarily conserved across several species, we also performed DEA on all miRNA transcripts present on the array (Supplementary Figure 1). Prior to any statistical comparisons, we filtered out other species’ miRNAs if they had a nucleotide sequence found in any mouse miRNA. We observed that several of the DE mouse miRNAs were also significantly expressed in other species, especially miR-22, which was upregulated across 12 different species. Taken together, these data highlight a distinct miRNA profile in EVs released by senescent myoblasts, with several of these miRNAs being implicated in senescent gene regulation.

Senescent myoblasts secrete EVs that contain miRNAs implicated in cell cycle regulation, apoptotic signalling, and autophagy

Next, we sought to identify the biological processes of the genes targeted by the DE miRNA. These genes were identified using the miRTarBase, miRecords, and TarBase databases, and only validated targets were included in the analysis. Gene ontology (GO) functional annotation of these genes was conducted, and the biological processes (BP) were clustered into groups using the clusterProfiler package in R. A network mapping of the GO BP terms revealed several biological functions associated with cellular senescence, with group themes consisting of cell cycle regulation, autophagy, and apoptotic signalling (Figure 3). Interestingly, there were several BP annotations associated with DNA damage and double-stranded break repair, suggesting that SEVs contain miRNAs that are relevant not only to the senescent condition, but also the stimulus that induced it. Furthermore, SEV miRNAs seem to be implicated in leukocyte regulation, suggesting that they may have SASP-like effects on these immune cells. Taken together, these findings revealed that EVs released by senescent myoblasts are packaged with meaningful miRNA cargo that reflects the current state of the cell and that these SEVs have the potential to be potent SASP members.

Figure 3. Network mapping of biological processes implicated by senescent EV miRNAs. Validated gene targets of differentially expressed miRNAs were analyzed for gene ontology (GO) terms related to biological processes. GO terms were clustered into groups and mapped using the R function emapplot in the enrichplot package. Node size is proportional to the number of gene targets found in each term, and node colour is related to the adjusted p-value.

We then opted to examine the target genes for the upregulated and downregulated miRNAs separately. Upset plots visually display overlapping gene targets between downregulated miRNAs (Figure 4A) and upregulated miRNAs (Figure 4B). Notably, there were significantly more gene interactions between the upregulated miRNAs than the downregulated miRNAs, most likely due to the identification of more validated gene targets in the former. Several miRNA pairings targeted the same gene sets, such as miR-669m/miR-466m/miR-466j/miR-466h (79 intersections) or miR-34a/miR-301a (1301 intersections). These data imply that a redundancy in gene regulation exists in miRNAs that changed in the same direction within SEVs.

Figure 4. Visualization of miRNA gene target intersections. Upset plots depicting the number of shared gene targets between all the (A) downregulated miRNAs and (B) upregulated miRNAs. The vertical bars display the size of each intersection, while the horizontal bars represent the size of each miRNA set. The connected dots below the vertical bars indicate which sets of miRNAs are included in that intersection.

An interesting finding is the overlap in gene targets between miR-34a-5p, miR-301a-3p, and miR-22-3p (Figure 4B; 786 intersections). While both miR-34a and miR-22 have been extensively associated with cellular senescence, miR-301a has not, suggesting this may be a novel senescence-associated miRNA [22-24]. To examine this further, we mapped out the GO BP term clusters of the three upregulated miRNAs (Supplementary Figure 2). The network revealed significant overlap in biological functions between the genes targeted by miR-34a, miR-301a, and miR-22, with plotted GO terms displaying the proportion of validated genes regulated by each miRNA (pie-chart nodes). Importantly, key biological processes associated with senescence, autophagy, and muscle cell differentiation were all observed. KEGG enrichment analysis revealed that the validated gene targets of miR-301a were the most associated with cell cycle signalling pathways (Supplementary Figure 3), providing further evidence that this miRNA might be implicated in senescent gene signalling.

Shared regulation of top targeted genes by divergently expressed miRNAs

To identify the genes most likely to be affected by SEV miRNAs, we plotted the interactions between all DE miRNAs and the topmost targeted genes (defined as any gene targeted by more than 5 DE miRNAs; Figure 5). Network visualization was conducted using miRnet.ca, and miRNAs were coloured based on expression (upregulated – red; downregulated – blue), with node size corresponding with the degree of connectivity to displayed target genes. We identified several genes of interest, including Pten, which was regulated by the greatest number of DE miRNAs. Another topmost targeted gene of interest was Mdm2, a member of the senescence-associated super-enhancer (SASE) family that acts to suppress apoptosis by reducing p53 activity in response to DNA damage [25].

Figure 5. Network mapping of DE miRNAs and their topmost targeted genes. A visualization of upregulated (red) and downregulated (blue) miRNAs and the target genes they interact with. Node size is proportional to the degree value of connectivity. The network was degree filtered with a cut-off of 5 to display the top miRNA-targeted genes.

Notably, every gene identified in the network map was regulated by at least one upregulated and one downregulated miRNA. To examine this further, we evaluated the overlap in gene regulation by miRNAs that were divergently changing in SEVs (i.e., how many gene targets do an upregulated and downregulated miRNA share). An Upset plot depicting only the intersections between divergently expressed miRNAs displays the overlaps in gene regulation between upregulated and downregulated miRNAs, with the most intersecting being miR-34a/miR-3470b (62 intersections; Supplementary Figure 4). This finding was unexpectedly high, as the greatest overlap between all downregulated miRNA targets was only 79 intersections (Figure 4A). Altogether these observations support the notion that a change in the expression of a given gene cannot be confidently inferred by examining the directional change of its upstream targeting miRNA.

Senescent-derived EVs promote anti-apoptotic gene signalling

Given the potential for SEV-mediated regulation of gene signalling, we next sought to experimentally determine if EV treatment can affect the genetic expression of recipient cells. Isolated NSEVs and SEVs were added to C₂C₁₂ myoblasts for a total of 48h, after which the cells were washed and collected for RT-qPCR (Figure 6A). There was no effect of SEV transfer (SEV–T) on senescence markers Trp53 (Figure 6B) or Cdkn1a (Figure 6C) relative to NSEV transfer (NSEV–T). However, the expression of two of the most targeted genes by DE miRNAs did change; an elevation in Pten (+52%; p < 0.05; Figure 6D) and Mdm2 (+56%; p < 0.05; Figure 6E) mRNA was observed in myoblasts co-incubated with SEVs. Interestingly, SASP-related genes Il33 and Cyr61 decreased with SEV–T (−52%, −26%; p < 0.05; Figure 6F–6G). These findings demonstrate that SEVs released from myoblasts are capable of regulating genes targeted by their packaged miRNAs. Furthermore, SEVs do not elicit a senescent genotype but rather promote anti-apoptotic and anti-SASP signalling.

Figure 6. Co-culture transfer experiments of senescent-derived EVs onto myoblasts. (A) Schematic diagram of myoblasts cultured with EVs derived from senescent (SEV) or non-senescent (NSEV) myoblasts. Relative mRNA expression of senescence genes (B) Trp53 and (C) Cdkn1a within myoblasts 48h following the transfer of NSEV or SEV. Relative mRNA expression of top DE miRNA gene targets (D) Pten and (E) Mdm2. Relative mRNA expression of SASP genes (F) Il33 and (G) Cyr61. Data are means ± SD. * denotes a significant difference from the NSEV-T group (p<0.05). SEV-T: Senescent EV-treated myoblasts; NSEV-T: Non-senescent EV-treated myoblasts.

SEV miRNA gene targets are differentially expressed in aged mouse muscle

Lastly, we sought to investigate whether senescent EV-containing miRNAs act on genes that are altered in aging skeletal muscle. RNA-Seq data from the gastrocnemius muscle of adult (8-month) and geriatric (28-month) C57BL/6JRj mice were obtained from previous work [26]. Börsch and colleagues (2021) identified 2302 genes that were differentially expressed (FDR < 5% and FC ≥ |1.5|) in aged skeletal muscle (Figure 7A). Of these, 700 differentially expressed genes (DEG) are also validated gene targets of 2 or more SEV DE miRNAs as found in our dataset, representing a significant overlap (p = 4.52 × 10⁻¹⁴). Together, these observations imply that a meaningful relationship exists between the aging-related genetic alterations in skeletal muscle and the miRNAs released by senescent cells. We next performed KEGG enrichment analysis on the 700 overlapping DEG and identified the top 10 significant pathways (Figure 7B). Notable amongst these were pathways related to cellular senescence, autophagy, and FoxO signalling. Heatmaps display the significant DEG (between 8mo and 28mo gastrocnemius muscle) implicated in cellular senescence and apoptosis, with only the genes targeted by two or more SEV miRNAs being displayed (Figure 7C). Despite only 14.6% (700/4800) of SEV miRNA targets being altered in muscle during aging, these genes are heavily implicated in pathways related to MuSC function, muscle differentiation, autophagy, apoptosis, and cellular senescence [6, 27, 28].

Figure 7. Association of SEV-derived miRNAs with the transcriptomic profile of geriatric mouse muscle. RNA-seq data of gastrocnemius muscle from young (8mo; n = 8) and geriatric (28mo; n = 9) C57BL mice. (A) Volcano plot of differentially expressed genes (DEG) based on adj. p < 0.05 and FC ≥ |1.5|. Red dots indicate significant DEG, while purple dots are significant DEG that are validated gene targets of SEV miRNA. Green dots are non-significant with FC ≥ |1.5| and adj. p > 0.05. A Venn diagram illustrates the overlap (purple) between aged gastrocnemius DEG and genes targeted by two or more DE miRNAs within SEV. (B) KEGG enrichment analysis of overlapping geriatric DEG and SEV miRNA targets. The top 10 significant pathways are plotted in descending order by adj. p-value. (C) Heatmap showing gene expression data from 8mo and 28mo gastrocnemius muscle for two significant KEGG pathways, cellular senescence (mmu04218) and apoptosis (mmu04210). All genes displayed are significant (FDR < 0.05 and FC ≥ |1.5|) and validated targets of 2 or more SEV miRNA. RNA-seq data was obtained from publicly available data (GSE145480) [26].

Cell culture

C₂C₁₂ myoblasts (American Type Culture Collection) were cultured in a humidified, 5% CO₂ incubator at 37° C with growth medium (GM) composed of Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Invitrogen). Bleomycin sulfate (Cat. #13877; Cayman Chemical) was utilized to induce senescence as established in previous work from our group [17]. Briefly, bleomycin was dissolved in nitrogen-purged dimethyl sulfoxide (DMSO; Thermo Fisher Scientific) and diluted to 14 μM in GM. At ~50-60% confluence, C₂C₁₂ myoblasts were treated with bleomycin (BLEO) or vehicle (VEH; isovolumetric nitrogen-purged DMSO) for 12h. Cells were then washed twice with sterile 1X phosphate-buffered saline (PBS; Gibco) and replaced with fresh GM or differentiation media (DM), based on experimental design. A senescent phenotype was confirmed in this model using senescence-associated β-galactosidase staining, cell cycle flow cytometry, Cdkn1a mRNA expression and γH2AX immunocytochemistry [17]. Differentiation was induced with differentiation medium composed of DMEM supplemented with 2% horse serum (Invitrogen) and 1% penicillin/streptomycin; DM was changed every 48h. Experiments were conducted with cells at passages 6-8, and technical duplicates were averaged to create each N, with a total N = 3 per condition. The C₂C₁₂ cell line was authenticated prior to use, and cells were routinely examined for mycoplasma contamination throughout the study.

Senescence-Associated β-galactosidase activity

To histologically identify senescent myoblasts, treated cells were stained for Senescence-Associated β-galactosidase (SA-β-gal) 48h after treatment. Briefly, the myoblasts were fixed with 1X fixative solution (Cell Signalling Technologies) for 10 min. After washing with PBS, β-galactosidase staining solution (pH 6.0, Cell Signalling Technologies) was added, and the plates were stored in a dry, 37° C incubator for 16h. Images were captured through phase-contrast microscopy on the Eclipse TI2 microscope (Nikon), 20X objective, and analyzed with NIS-Elements (V5.30.01, Nikon).

Flow cytometry

Myoblasts were treated with BLEO or VEH for 12h, then cultured in GM for 48h before being washed twice, collected, and filtered through a 35-μm cell strainer into flow tubes (Corning). Cells were then pelleted, and ice-cold 70% ethanol was added dropwise under gentle agitation. Myoblasts were fixed for 30 min at 4° C and then washed twice with 1 × PBS before being resuspended in 500 μL of propidium iodide (PI) solution (5 μg/mL of PI and 20 μg/mL of RNase A in 1×PBS). Cell cycle kinetics were determined based on PI fluorescence using a Cytoflex LX (Beckman Coulter) and quantified with FlowJo (V10.8.1, BD Biosciences). Staining dyes used for flow cytometry are detailed in Table 1.

Transwell co-culture

Myoblasts were seeded onto a transwell insert with a 0.4-μm membrane pore (Corning, Falcon) and treated with BLEO or VEH (as stated above) to generate senescent (SEN) or non-senescent (NSEN) myoblasts. The insert was placed into a well that was seeded with untreated myoblasts 12h earlier, which allowed for intercellular communication between SEN or NSEN myoblasts and the untreated cells, without permitting direct cell-cell interactions. Untreated myoblasts were then allowed to proliferate to ~90% confluence before changing to a differentiation medium (DM) composed of DMEM supplemented with 2% horse serum (Invitrogen) and 1% penicillin/streptomycin. Every 48h, DM was changed and the inserts were replaced to prevent over-confluence. Differentiation was induced after 6 days of culturing with DM, and myotubes were collected for ICC or RT-qPCR.

RNA isolation and real-time quantitative PCR (RT-qPCR)

Myoblasts or myotubes were collected with Trizol reagent (Invitrogen), and RNA was purified using the E.Z.N.A. Total RNA Kit (Omega Bio-Tek) as per manufacturer instructions. Isolated RNA concentration and purity were measured with the Nanodrop 1000 spectrophotometer (Thermo Fisher Scientific), after which 1 μg of RNA was reverse transcribed into cDNA using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific). Gene amplification was performed using TaqMan assays (Table 2) and Taqman Fast Advanced Master Mix (Thermo Fisher Scientific). Cq values were obtained in triplicate and relative gene expression data were normalized to the geometric mean of reference genes Gapdh and Tbp using the comparative 2^-ΔΔCq method [55]. A no-template control (NTC) was used for each probe to confirm the absence of contamination or non-specific amplification.

Immunocytochemistry

Differentiated myotubes were washed with 1X PBS and fixed with 4% paraformaldehyde (PFA) for 10 min. Myotubes were permeabilized for 20 min with 0.1% Triton-X in PBS before blocking with 2% bovine serum albumin (BSA) in PBS for 1h. Myosin heavy chain (MyHC) supernatant was incubated overnight at 4° C (Table 1). Myotubes were then washed twice with 1X PBS and incubated with Alexa Fluor^® 488 goat anti-mouse secondary antibody (Invitrogen) at 1:500 in PBS for 2h at room temperature. After two washes, DAPI (4’,6-diamidino-2-phenylindole; Sigma-Aldrich) was added for 5 min to label nuclei. Myotubes were imaged at 20X magnification using an Eclipse TI2 microscope (Nikon) and analyzed with NIS-Elements (V5.30.01, Nikon). Images were quantified to assess myotube diameter (μm), myotube surface area (MyHC+ μm² per image mm²), and myonuclear index (%).

Extracellular vesicle isolation

Extracellular vesicles (EVs) were isolated from the CM of senescent myoblasts in line with the MISEV recommendations for cell culture-conditioned medium [9]. Myoblasts were treated with VEH or BLEO for 12h and then washed twice with PBS. Cells were then incubated for 48h with DMEM supplemented with 10% exosome-depleted FBS (Cat. #A2720801; ThermoFisher Scientific) and 1% penicillin/streptomycin (GM-EV). 5 mL of conditioned media, controlled between experimental conditions, was collected and centrifuged at 2000 g for 10 min to pellet cell debris. Samples underwent another centrifugation at 20,000 g for 1h at 4° C to remove larger EVs (i.e., microvesicles and apoptotic bodies). The supernatant was collected and filtered through a 0.4-μm pore before being concentrated by ultracentrifugation at 100,000 g for 16h at 4° C. EV concentrates were washed twice with 1X PBS prior to final suspension. Ultracentrifugation was performed with the Optima^TM MAX-XP (Beckman Coulter) in appropriate polypropylene microtubes (Cat. #357448; Beckman Coulter) and the presence of EVs was confirmed with western blotting, transmission electron microscopy, and nanoparticle tracking analysis as previously described [56]. In accordance with MISEV2023 guidelines, the general term “EV” will be used throughout this manuscript in place of more common terms such as exosome or exosome-like vesicles [9].

Protein extraction and immunoblotting

Isolated EVs were lysed with ice-cold RIPA buffer (Sigma-Aldrich) supplemented with Pierce protease inhibitor (Thermo Fisher Scientific) and phosphatase inhibitor (Sigma-Aldrich). Samples were sonicated and then centrifuged at 14,000 g for 15 min to pellet debris. Protein concentrations were determined using a bicinchoninic acid assay (BCA; Thermo Fisher Scientific), and samples were prepared in 4X Laemmli Buffer (Bio-Rad). Samples were then loaded into each lane of a 4 – 15% TGX precast gel (Bio-Rad) and underwent SDS-PAGE. Following electrophoresis, proteins were transferred to a nitrocellulose membrane and stained with Ponceau S solution (Sigma-Aldrich) to assess equal loading between samples. Ponceau solution was removed with 3 x 5-min washes with Tris-buffered saline with 0.1% Tween-20 (TBS-T), and membranes were blocked with 5% BSA in TBS-T for 1h. Membranes were then incubated overnight at 4° C with primary antibodies diluted in 5% BSA in TBS-T (Table 1). The next day, membranes were again washed with TBS-T and incubated with the appropriate HRP-conjugated secondary antibody (Thermo Fisher Scientific) diluted at 1:10000 in 5% BSA in TBS-T for 2h. Bands were visualized upon the addition of ECL solution (Bio-Rad). Images were captured with the ChemiDoc MP Imaging System (Bio-Rad) and quantified using Image Lab software (V6.0.1, Bio-Rad).

Transmission electron microscopy

Following EV isolation and concentration, EV pellets were diluted 1:20 into nuclease-free water and 5 μL was added to a Cu/Pd grid (200 mesh with Formvar). Afterwards, grids were air-dried for 10 min then gently dried with filter paper before being stained with 1% uranyl acetate for 1 min. Samples were visualized with a JEOL JEM 1200EX TEMSCAN transmission electron microscope with an HV of 80kV. The presence of EVs was verified with images taken at 40,000X – 150,000X magnification.

Nanoparticle tracking analysis

Isolated EV samples were diluted 1:500 in ultrapure water (DNase/RNase-free and 0.1-μm membrane-filtered; Invitrogen) and analyzed for size and concentration, as detailed previously [56]. Before and after each sample was measured, approximately 5 mL of ultrapure water was flushed through the cell. Diluted EV samples were injected, and particle size and concentration were measured at 11 different positions, with three readings per position. Pre-acquisition parameters: temperature – 23° C; camera sensitivity – 85; shutter time – 120; and frame rate – 30 fps. Post-acquisition parameters: minimum brightness – 30; minimum size – 5 pixels; and maximum size – 200 pixels. Samples were analyzed with a ZetaView PMX 110 (Particle Metrix).

miRNA microarray

Total RNA was extracted from EVs isolated from 10 ml of CM for each sample. Two technical replicates were combined (5 mL CM each) to produce each sample, and after ultracentrifugation, 50 μL aliquots of purified EVs were stored at -80° C until processed for RNA isolation. To promote the recovery of small RNAs, the MagMAX mirVana Total RNA Isolation Kit (Cat. #A27828; Applied Biosystems) was utilized in conjunction with the KingFisher Apex purification system (ThermoFisher Scientific). RNA concentration and purity were measured using the Qubit 4 fluorometer (Invitrogen) and Nanodrop spectrophotometer (Thermo Fisher Scientific). Total RNA was then labelled using FlashTag Biotin HSR for miRNA microarrays (Applied Biosystems) to prepare samples for hybridization.

Biotin-labelled samples were combined with a hybridization master mix and incubated at 99° C for 5 min, then 45° C for 5 min. Sample cocktail was then injected into a miRNA 4.0 GeneChip cartridge array (Affymetrix) and placed into hybridization ovens to incubate at 48° C and 60 rpm for 18h. Arrays were then washed with a GeneChip Fluidics Station 450 and scanned using a Gene Chip Scanner 3000 7G.

Extracellular vesicle transfer experiments

EVs were collected from SEN and NSEN myoblasts via ultracentrifugation, as described above. After EV isolation, a small fraction was lysed to determine the protein concentration, and samples were normalized to have the same approximate number of particles (2 × 10⁸ particles/mL). Myoblasts cultured in GM-EV were then treated with the isolated NSEVs or SEVs every 24h for a total of 48h before being collected for RT-qPCR. NSEV-T and SEV-T designate myoblasts after NSEV treatment and SEV treatment, respectively.

RNA-Seq data sets of mouse gastrocnemius muscle

The gastrocnemius muscle of male C57BL/6JRj mice at 8 months-old (8mo; n=8) and 28 months-old (28mo; n=9) was collected and sequenced as part of a previous study [26]. RNA-Seq data were obtained from the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE145480. Data were filtered to remove other time points, and significance was set at a false discovery rate (FDR) < 0.05 and fold change ≥ |1.5| to identify differentially expressed genes (DEG). Functional enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database was performed on DEG that were also validated gene targets of 2+ SEV miRNAs. Data processing and visualization were executed as stated below.

Data processing and statistical analysis

Microarray data analysis

After the miRNA 4.0 arrays were scanned, the raw intensities in a CEL format were exported and subsequent data processing and analysis were performed in RStudio with R version 4.3.2. CEL files and the sample and data relationship format (SDRF) file were read into R using the oligo package [57], and chip quality control was evaluated through the arrayQualityMetrics package [58] and raw intensity/pseudoimage chip visualization. MiRNA expression was preprocessed using robust multichip averaging (RMA), which performs background subtraction, quantile normalization, and summarization of probe-level data [59]. Low intensity probes were filtered out, and probe annotation was performed using the miRNA 4.0 annotations file (Affymetrix). Only mouse miRNAs were filtered in, and other species or small RNA types were removed prior to model fitting. Subsequently, linear modelling was performed to find the differentially expressed (DE) mouse miRNAs between SEN and NSEN-derived EVs using the limma package [60]. Multiple test corrections were applied using the Benjamini-Hochberg method to estimate the FDR, and we set significance thresholds to an FDR < 5% and fold change ≥ |1.5| to determine DE miRNAs [61]. For cross-species validation of DE mouse miRNAs (Supplementary Figure 1), additional miRNAs from other species that contained a nucleotide sequence not found in any mouse miRNA were added to the analysis. The multiMiR package was employed to identify validated targets of upregulated and downregulated miRNAs from several external databases (miRTarBase, miRecords, and TarBase) [62]. Identified gene targets were filtered to remove replicates before undergoing gene ontology (GO) analysis and functional enrichment analysis using the KEGG database through the clusterProfiler package [63]. GO and KEGG plots were ordered by adj. p and filtered for related terms. GO biological process network mapping was generated from the multiMiR-identified gene targets after reducing redundant GO terms using the simplify function of the clusterProfiler package and filtering for terms with FDR < 5 x10^-5. Upset plots were generated with the UpSetR package using discrete upregulated and downregulated miRNA targets, and plots were filtered to show intersections between 2 or more miRNAs to best display overlaps in miRNA regulation of gene targets [64]. The MiRNet 2.0 web tool (https://www.mirnet.ca) was used to generate a network visual between all DE miRNAs and their target genes [65]. The degree filter was set to a cutoff of 5 to limit network size and display only the gene targets that connected to more than 5 miRNAs. Cytoscape (V3.10.2) was then used to visualize the miRNA-gene interaction network. The probability of gene list overlap was calculated with a hypergeometric distribution. A full list of the packages used for RNA-Seq and miRNA microarray data processing and analysis can be found in the supplementary file (Supplementary Table 1).

Experimental data analysis

For the statistical comparison of two groups, unpaired two-tailed Student’s t-tests were employed using RStudio with R version 4.3.2. Normality was evaluated with the Shapiro-Wilk test, and in the case of non-normal data, a non-parametric test was utilized. Statistical significance was accepted at p < 0.05, and significance symbols are defined in figure captions. Data are presented as means ± SD. Bar graphs were generated with Prism (V9.5.1, Graphpad Software LLC).

Data availability

The data that support the findings of this study are openly available in Gene Expression Omnibus (GEO) at https://www.ncbi.nlm.nih.gov/geo/, reference number GSE297375.