Beyond miRNA Cargo Profiles: Anti‐Inflammatory Roles of Extracellular Vesicle‐Enriched miRNAs Derived From Human Intervertebral Disc Cells Unveiled by Functional Testing

ABSTRACT

Intervertebral disc (IVD) degeneration is a leading cause of chronic low back pain and a major contributor to global disability. Understanding the molecular mechanisms underlying this condition is essential for developing targeted therapies. Among these mechanisms, microRNAs (miRNAs) have emerged as critical post‐transcriptional regulators of gene expression in IVD cells, influencing key processes such as extracellular matrix (ECM) homeostasis, inflammatory signalling and cellular senescence. Extracellular vesicles (EVs), which transport miRNAs between cells, represent a promising avenue for therapeutic intervention. However, the composition of their miRNA cargo across different stages of disc degeneration remains inadequately characterised. We isolated EVs from primary human IVD cells derived from non‐degenerate, mildly degenerate, and severely degenerate tissues, and performed small RNA sequencing to profile their miRNA content. Bioinformatic analyses revealed enrichment in pathways related to ECM‐receptor interaction, focal adhesion, inflammation and cell cycle regulation. Notably, let‐7b‐5p and miR‐100‐5p were among the most abundant miRNAs and were significantly lower in EVs from degenerate discs. Functional assays demonstrated that transfection of IVD cells with let‐7b‐5p or miR‐100‐5p mimics individually suppressed IL‐1ß expression at both mRNA and protein levels, confirming their anti‐inflammatory roles. Strikingly, co‐delivery of both miRNAs enhanced suppression of pro‐inflammatory mediators, reduced senescence‐associated p16 expression and upregulated TIE2 mRNA, indicating synergistic effects in promoting a regenerative cell phenotype. These findings highlight the regulatory roles of EV‐enriched let‐7b‐5p and miR‐100‐5p in modulating inflammation and senescence in IVD cells, and underscore the potential of miRNA‐loaded EVs as cell‐free regenerative therapies for disc degeneration.

1. Introduction

Intervertebral disc (IVD) degeneration is a major contributor to chronic low back pain (LBP), the leading cause of disability worldwide, particularly affecting aging populations in developed countries (Ferreira et al. 2023). In addition to its impact on quality of life, IVD degeneration imposes a substantial economic burden through increased healthcare demands and reduced workforce productivity (Ferreira et al. 2023; Vos et al. 2020). The IVD is a complex fibrocartilaginous tissue comprised of two distinct regions: the central, gelatinous nucleus pulposus (NP) and the surrounding, collagen‐rich annulus fibrosus (AF). The NP is characterised by chondrocyte‐like cells that produce high levels of proteoglycans [aggrecan (ACAN)] and Type II collagen, which are essential for the disc's ability to resist compressive loads. IVD degeneration is typically initiated in the NP, where the loss of this specific cellular phenotype leads to decreased matrix synthesis, dehydration and eventual structural failure. Consequently, preserving or restoring the expression of specific NP phenotypic markers, such as TEK receptor tyrosine kinase (TIE2), ACAN, Paired Box 1 (PAX1) and SRY‐Box Transcription Factor 9 (SOX9), is a primary objective of regenerative therapies. TIE2, in particular, marks a sub‐population of NP progenitor cells with high regenerative potential, making its upregulation a key indicator of therapeutic efficacy. Epidemiological studies and magnetic resonance imaging analyses have underscored the progressive nature of the degenerative condition, highlighting the urgent need for innovative therapeutic strategies (Wu et al. 2020; Fujii et al. 2023).

Extracellular vesicles (EVs) have emerged as promising cell‐free tools in regenerative medicine, especially for degenerative diseases such as IVD degeneration (Piazza et al. 2020; Zeng et al. 2022; DiStefano et al. 2022). EVs are lipid bilayer‐bound particles secreted by cells that carry bioactive cargo, including microRNAs (miRNAs), proteins and lipids (Welsh et al. 2024). Their ability to mimic the functional properties of their parent cells makes them ideal candidates for delivering targeted molecular therapies to damaged tissues (Zeng et al. 2022; Wu and Sun 2021). Among these cargo molecules, miRNAs are key regulators of gene expression and reflect the homeostatic state of tissues (Wang et al. 2022; Cazzanelli and Wuertz‐Kozak 2020; Jiang et al. 2021).

Recent studies have shown that EVs derived from human mesenchymal stem cells (MSCs), enriched with anti‐inflammatory miRNAs, can suppress pro‐inflammatory cytokine expression in IVD cells, potentially slowing degeneration (Piazza et al. 2020). Specific miRNAs such as miR‐155‐5p and miR‐27a have been implicated in activating inflammatory pathways, while their inhibition leads to reduced cytokine levels (Cao and Chen 2017; Cazzanelli et al. 2024). Other miRNAs, including miR‐149‐5p and miR‐99a‐3p, have been shown to inhibit matrix metalloproteinases (MMPs) and prevent extracellular matrix (ECM) degradation (Zhao et al. 2021). Furthermore, combinations of miRNAs like miR‐181a‐5p, miR‐92a‐3p, miR‐21‐5p and miR‐186‐5p delivered via EVs have demonstrated the ability to reduce senescence markers and inflammatory mediators in both human cell cultures and animal models (Zhang et al. 2025). These findings suggest that profiling miRNAs associated with tissue health could help identify therapeutic candidates capable of mitigating degeneration, chronic pain and functional impairment (Wang et al. 2015; Genedy et al. 2024). Engineered EVs carrying specific miRNAs offer a controlled approach to modulate cellular responses and promote tissue regeneration in the IVD microenvironment (Rai et al. 2024; Menjivar et al. 2024; Tang et al. 2021).

Despite these promising developments, significant gaps remain in our understanding of the miRNA cargo within EVs derived directly from human IVD cells. Most existing research has focused on MSC‐derived EVs, with limited investigation into the miRNA profiles of EVs from IVD cells themselves (DiStefano et al. 2022; Nazari‐Shafti et al. 2020). Given the role of EV‐associated miRNAs in regulating biological processes relevant to IVD degeneration, a comprehensive profiling of these molecules is essential. To date, no study has systematically examined the miRNA content of EVs from human IVD cells isolated from tissues at varying stages of degeneration using standardised culturing, purification, sequencing and bioinformatic analysis protocols. Additionally, functional studies evaluating the roles of these miRNAs in IVD health are sparse, and no prior work has employed a function prediction‐guided screening strategy to assess their therapeutic potential.

Therefore, the objectives of this study were to: (1) characterise the miRNA profiles of EVs derived from human IVD cells across different grades of tissue degeneration, and (2) evaluate the therapeutic potential of key miRNAs using a targeted, function prediction‐guided screening approach.

2. Materials and Methods

2.1. Study Workflow

Total RNA was extracted from purified EVs, and small RNA was purified from the bulk RNA. Small RNA sequencing was conducted after the quality test. Bioinformatics was used to predict the function of detected miRNAs. Five miRNAs of interest were selected for further functional assessments (Figure 1A). The functional assessments comprised the evaluation of the regulatory roles and the expression of selected miRNAs at the gene and protein levels (Figure 1B). The expression of NP markers, inflammatory mediators and senescence markers was evaluated to determine the effect on gene expression level (Figure 1B(a)). The production of inflammatory mediators was evaluated to validate the anti‐inflammatory effects of selected miRNAs at the protein level (Figure 1B(b)).

FIGURE 1.

Schematic overview of the study workflow. (A) Study workflow includes extracellular vesicle (EV) purification, RNA extraction and small RNA purification, small RNA quality test, small RNA sequencing, bioinformatic analysis, selecting the miRNAs of interest and functional assessments. (B) Functional assessments including (a) gene expression of NP cell phenotypic markers, inflammatory mediators and senescence markers. (b) Protein expression of inflammatory mediators. SEC, size‐exclusion chromatography.

2.2. Sample Preparation

Tissue isolation and cell culture: This study was conducted in strict compliance with ethical protocols approved by McGill University's Institutional Review Board (IRB #00010120). IVD tissue was obtained from consenting organ donors and surgical patients. Discs were graded by visual Thompson grade and grouped as non‐degenerate (Non‐deg) (Grades 1–2), mildly‐degenerate (Mildly‐deg) (Grade 3) and degenerate (Deg) (Grades 4–5). The Non‐deg group (organ donors, 25.8 ± 16.4 years) is significantly younger (p = 0.0343) than the Mildly‐deg group (5 organ donors and 1 spinal surgery donor, 55.7 ± 12.6 years), but has no significant age difference compared to the Deg group (1 organ donor and 18 spinal surgery donors, 42.6 ± 13.6 years). Sex was evenly distributed across groups (p = 0.0748). NP and inner AF tissue were retrieved as previously reported (Li et al. 2024). Following enzymatic isolation, primary IVD cell viability was assessed using Trypan Blue exclusion. Only cell populations exhibiting >90% viability were utilised for downstream culture. Isolated cells were washed and resuspended in DMEM (2.25 g/L glucose) supplemented with 10% (v/v) FBS, 50 µg/mL gentamicin, 2 mM GlutaMAX and 50 mM HEPES, and incubated at 37°C, 5% CO₂. Cells were seeded at a density of 1 million cells/T75 flask (approximately 13,333 cells/cm²). For organ‐donor cultures, NP and inner AF cells were combined 1:1; surgical samples were from mixed NP/inner AF tissue. Media was refreshed once per week until approximately 80% confluence. Cells were passaged to Passage 1.

Passage 1 cells were expanded 72–96 h to approximately 90% confluence, then washed 3× for 1 min with PBS at 37°C. Cells were switched to RoosterCollect‐EV media, supplemented with 1% EV‐depleted FBS. Conditioned media was collected after 4 days, replaced with fresh EV‐collection media and collected again after an additional 4 days. The workflow specifies pooling conditioned media from Days 0 to 4 and 5 to 8 (Li et al. 2024).

EV preparation: The EV samples investigated in this study were generated from the same donor cohort and isolation procedures as described in our previous proteomic analysis (Li et al. 2024). Conditioned media was collected after two consecutive 4‐day cultures, centrifuged at 500 × g for 5 min and 8400 rpm for 30 min to remove cells and apoptotic bodies. The supernatant was then concentrated and subjected to size‐exclusion chromatography using qEV columns (Izon Science). Fractions 7–9, corresponding to the EV‐enriched peak, were pooled. Particle concentration and size distribution were quantified via nanoparticle tracking analysis, and morphology was validated by transmission electron microscopy as previously characterised. Samples from additional donors were included in this study for the miRNA functional assessments. Donor and patient demographic information are presented in Table 1.

TABLE 1.

Donor and patient information and applications.

Donor	Age	Sex	Cause of death /disc pathology	IVD levels	Thompson grade	Experiments
						EV purification miRNA extraction and quality test	Small RNA sequencing	qPCR	Luminex cytokine multiplex
Organ donors
1	18	M	Trauma	L3‐S1	1, Non‐degenerate	✔	✔
2	19	M	Drug overdose	L3‐S1	1, Non‐degenerate	✔	✔
3	55	M	Fall	L3‐4	2, Non‐degenerate	✔	✔
4	21	M	Motor vehicle accident	L2‐5	1, Non‐degenerate	✔	✔
5	16	M	Drug overdose	L2‐3 and L4‐S1	1, Non‐degenerate	✔
6	57	F	Anoxia	T12‐L2 and L3‐4	2–3, Mildly‐degenerate	✔	✔
7	60	F	Brain haemorrhage	L3‐S1	3, Mildly‐degenerate	✔	✔
8	54	F	Stroke	T11‐L2	1 and 3, Mildly‐degenerate	✔	✔
9	63	M	Cranial trauma	L2‐5	2‐3, Mildly‐degenerate	✔	✔
10	68	M	Brain haemorrhage	L4‐S1	3, Mildly‐degenerate	✔
11	78	M	Unknown	L3‐5	3‐4, Degenerate	✔	✔
Surgical samples
1	32	M	Simple herniation	L4‐S1	3, Mildly‐degenerate	✔
2	32	M	Spondylosis	L5‐S1	3‐4, Degenerate	✔	✔
3	24	M	Simple herniation	L4‐5	4, Degenerate	✔	✔
4	45	F	Simple herniation	L4‐S1	4, Degenerate	✔
5	57	F	Herniation with Type 1 modic changes	L4‐S1	4, Degenerate			✔
6	51	M	Degenerative disc disease	L4‐S1	4, Degenerate			✔	✔
7	56	M	Simple herniation	L4‐5	4, Degenerate			✔	✔
8	23	F	Degenerative disc disease	L5‐S1	4, Degenerate			✔
9	33	F	Simple herniation	L4‐S1	4, Degenerate			✔	✔
10	49	F	Simple herniation	L4‐S1	4, Degenerate			✔	✔
11	39	F	Simple herniation	L4‐5	4, Degenerate			✔
12	34	F	Simple herniation	L4‐S1	4, Degenerate			✔
13	44	M	Degenerative disc disease	L4‐S1	4, Degenerate			✔
14	41	F	Degenerative disc disease	L4‐S1	4, Degenerate			✔
15	43	F	Simple herniation	L4‐5	4, Degenerate			✔
16	34	M	Degenerative discopathy	L4‐5	4, Degenerate			✔
17	28	F	Simple herniation	L5‐S1	4, Degenerate			✔
18	41	F	Degenerative disc disease	L5‐S1	4, Degenerate			✔
19	58	M	Degenerative disc disease	L4‐S1	4, Degenerate			✔
Summary
Group		Non‐degenerate		Mildly‐degenerate		Degenerate
Age (mean ± SD)		25.8 ± 16.4		55.7 ± 12.6		42.6 ± 13.6
Sex distribution		Female	Male	Female	Male	Female		Male
		0	5	3	3	11		8
		0.0%	100.0%	50.0%	50.0%	57.9%		42.1%
Overall		Female 14/30 = 46.7%; Male 16/30 = 53.3%

Note: Age differences across groups were compared using the Welch's ANOVA test with Dunnett's T3 multiple comparisons. Non‐degenerate versus Mildly‐degenerate, p = 0.0343; Non‐degenerate versus Degenerate, p = 0.1981; Mildly‐degenerate versus Degenerate, p = 0.1530. Sex distributions across groups were compared using the Fisher–Freeman–Halton exact test (two‐sided); p = 0.0748.

Abbreviations: EV, extracellular vesicle; SD, standard deviation.

MiRNA purification: EV samples (approximately 1.84 × 10¹⁰ particles/mL, 50 µL) from 15 donors (4–6 biological replicates) were used. EVs were purified using size‐exclusion chromatography purification in phosphate‐buffered saline (PBS) and Fractions 7–9 were pooled. RNA was extracted and miRNA was purified from the EV samples with a miRNeasy micro kit (Cat. #217084, QIAGEN, Toronto, ON, Canada) following the manufacturer's instructions. MiRNAs were stored at −80°C for further analysis.

MiRNA quality test: The quantity and quality of miRNAs were evaluated with a Qubit RNA High Sensitivity Assay Kit (ThermoFisher Scientific, Saint‐Laurent, QC, CA) and the Agilent automated electrophoresis portfolio, including the Agilent 2100 Bioanalyzer system and Small RNA Analysis kit with RNA 6000 Pico Chips (Agilent Technologies, Mississauga, ON, Canada). All miRNA samples passed quality control.

2.3. Small RNA Sequencing and Data Processing

Library preparation and quality test: Small RNA sequencing libraries were prepared using QIAseq miRNA Library Kit [12‐nt unique molecular identifiers (UMIs)] (REF. #1103677, LOT. #76602338, QIAGEN) following the manufacturer's instructions. Sequencing was performed on a NextSeq 500 system (Illumina, San Diego, CA, USA), obtaining around 10 million reads per sample. Libraries were quantified using a Qubit dsDNA Quantification High Sensitivity Assay Kit (ThermoFisher Scientific, Toronto, ON, Canada) and assessed by Agilent 2100 Bioanalyzer using an Agilent High Sensitivity DNA Kit (Agilent Technologies). All libraries passed the quality test.

Data processing: Raw FASTQ files were analysed using QIAGEN's own analytical pipeline RNA‐seq Analysis Portal 5.1 (https://geneglobe.qiagen.com/ca/analyze) using the provided Homo sapiens (GRCh38.103) genome and miRBase v22.0 database. Samples passed quality control if >1% of UMI reads were annotated to miRBase v22.0 records. Eleven out of 15 samples (3–4 donors per group) passed the quality control. Raw data of miRNA expression in count per million (CPM) of the 11 samples were exported for further analysis.

2.4. Pathway and Annotation Analysis for miRNA Function Prediction

Function prediction: The enriched pathway and annotation analyses were performed by using DIANA‐miRPath v4.0, a web server that offers numerous analysis options, including Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analyses, for the target‐based functional analysis of miRNAs (Tastsoglou et al. 2023; Vlachos et al. 2015). We performed miRNA‐centric analysis for KEGG and GO pathways and annotations, and included only experimentally supported human miRNA targets on protein‐coding transcripts by using the TarBase v8.0 database to avoid false‐positive predictions. We applied direct TarBase targets without using a secondary target annotation source when no significant interactions were found for certain miRNAs and did not include long non‐coding RNA targets. The miRNA annotation (or miR‐ID) was based on miRBase v22.1 database. We used the Pathways Union merging method combined with the Classic Analysis testing method to assess the combined action of selected miRNAs. False discovery rate (FDR) correction and a p value threshold of ≤0.05 were applied. Identical or duplicate reads in low‐abundance miRNAs are a common observation in small RNA sequencing, particularly when miRNA concentrations are low. This phenomenon can arise from technical biases, such as those introduced during the PCR amplification step (Fu et al. 2018). We observed increased identical reads in miRNAs with CPM ≤ 300. To minimise false positive results, we focused the functional prediction on miRNAs with CPM > 300. Evaluation metrics and algorithmic process are described in the Supporting Information.

2.5. Cell Culture and Induction of an Inflammatory Microenvironment

Cell culture: Human IVD cell isolation was performed as previously published (Li et al. 2024; Gawri et al. 2014). We used cells from surgical samples for functional testing, which were obtained from a mixture of NP and inner AF tissue. Isolated cells were seeded at one million cells per T75 cell culture flask (Sarstedt, TC Flask T75, Stand, Vent. Cap, Germany). Cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) (Sigma–Aldrich, Oakville, ON, Canada) containing 2.25 g/L glucose. The glucose level was selected for consistency with the EV collection media (Li et al. 2024). The media was further supplemented with 10% (v/v) fetal bovine serum (FBS), 50 µg/mL gentamicin, 2 mM GlutaMAX and 50 mM HEPES (ThermoFisher Scientific, Toronto). The cultures were maintained at 37°C in a 5% CO₂ environment for 1–2 weeks. The culture media was refreshed twice per week until cells reached approximately 80% confluence in Passage 1. The cells then passaged and reseeded into a 6‐well‐plate (Sarstedt, Germany) at a density of 240,000 cells/well with 3 mL/well media and maintained in the same condition overnight. To minimise the effects of isolation and expansion in monolayer cultures, we only used cells at Passage 1 for expansion and Passage 2 for functional testing in this study.

Toll‐like receptor (TLR) activation: Synthetic diacylated lipopeptide Pam2CSK4 (Pam2) (Cat. #tlrl‐pm2s‐1, InvivoGen, San Diego, CA, USA), a potent activator of TLR2 and the pro‐inflammatory transcription factor NF‐κB, was used to induce an inflammatory microenvironment (Krock et al. 2017; Mannarino et al. 2021; Mannarino et al. 2023). Monolayer human IVD cells were rinsed with 1 mL/well sterile PBS three times. Pam2 (100 ng/mL) was applied in 3 mL/well serum‐free media (DMEM (Sigma–Aldrich) containing 2.25 g/L glucose, supplemented with 50 µg/mL gentamicin, 2 mM GlutaMAX and 50 mM HEPES (ThermoFisher Scientific, Toronto) and incubated for 42 h at 37°C in a 5% CO₂ environment (Mannarino et al. 2021).

2.6. Gene Expression Assessment

MiRNA transfection: Human miRCURY LNA miRNA Mimics (QIAGEN), let‐7a‐5p, let‐7b‐5p, let‐7f‐5p, miR‐16‐5p and miR‐100‐5p, were prepared following the manufacturer's instructions. Single‐miRNA treatment (5 miRNA mimics, respectively) and miRNA combination treatment (the combinations of two miRNAs of let‐7b‐5p, miR‐16‐5p, miR‐100‐5p and the combinations of all of them, respectively) were prepared at a working concentration of 5 nM for each miRNA mimic (Gharakhyli et al. 2023) in the solution using the above‐mentioned serum‐free complete media. Human IVD cells were rinsed with 1 mL/well sterile PBS for three times and transfected with 3 mL/well miRNA mimic solution for 24 h using 4 µL/mL HiPerFect transfection reagent (QIAGEN) according to the manufacturer's protocols (Figure 1B(a)).

Quantitative polymerase chain reaction (qPCR): IVD cells were lysed in TRIzol reagent (ThermoFisher Scientific, Toronto), and RNA extraction was performed following the manufacturer's instructions. Reverse transcription and qPCR procedures were conducted as previously described (Li et al. 2022; Krock et al. 2014; Cherif et al. 2019). Briefly, complementary DNA (cDNA) synthesis was conducted with a High‐Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific, Toronto) using Applied Biosystems Veriti 96‐Well Fast Thermal Cycler (ThermoFisher Scientific, Toronto). QPCR was performed in PROGENE 96‐Well Half‐Skirt ABI Fast PCR Plates (UltiDent Scientific, Montreal, QC, Canada) with PROGENE Adhesive Routine PCR Sealing Film (UltiDent Scientific) using an Applied Biosystems StepOnePlus Real‐Time PCR System (ThermoFisher Scientific, Toronto) with PowerUp SYBR Green Master Mix (ThermoFisher Scientific, Toronto). Primers are presented in Table 2. The melting curves were examined to exclude the risks of false amplifications. Data were normalised to the geometric mean of the reference genes (beta‐ACTIN and 28S rRNA) (Vandesompele et al. 2002) and presented in fold‐changes calculated with the 2^−ΔΔCt method (Livak and Schmittgen 2001).

TABLE 2.

qPCR primer list.

Target gene	Name/gene ID	Forward primer sequence (5′‐3′)	Reverse primer sequence (5′–3′)	Primer bank ID	Reference (PMID)
beta‐ACTIN (Reference gene)	ACTB/ID: 60	GTC TTC CCC TCC ATC GTG G	AAT CCT TCT GAC CCA TGC C		32853438
RNA28SN5 (Reference gene)	28S rRNA/ID: 100008589	GTT GAC GCG ATG TGA TTT CTG CC	CAT AGT TAC TCC CGC CGT TTA CCC		15958182
FOXF1	FOXF1/ID: 2294	GCG GCT TCC GAA GGA AAT G	CAA GTG GCC GTT CAT CAT GC	110735444c1
PAX1	PAX1/ID: 5075	TCG CTA TGG AGC AGA CGT ATG	GCT GCC GAC TGA TGT CAC A	380036025c1
TEK	TIE2/ID: 7010	TTA GCC AGC TTA GTT CTC TGT GG	AGC ATC AGA TAC AAG AGG TAG GG		33322051
SOX9	SOX9/ID: 6662	AGC GAA CGC ACA TCA AGA C	CTG TAG GCG ATC TGT TGG GG	182765453c1
ACAN	ACAN/ID: 176	TCG AGG ACA GCG AGG CC	TCG AGG GTG TAG CGT GTA GAG A		11237658
IL‐1ß	IL‐1ß/ID: 3553	AAG CTT GGT GAT GTC TGG TC	ACA AAG GAC ATG GAG AAC ACC		32853438
IL‐6	IL‐6/ID: 3569	TGA ACC TTC CAA AGA TGG CTG	CAA ACT CCA AAA GAC CAG TGA TG		28436958
CXCL8	IL‐8/ID: 3576	TCC TGA TTT CTG CAG CTC TG	GTC TTT ATG CAC TGA CAT CTA AGT TC		28436958
CDKN2A	p16/ID: 1029	CTG CCC AAC GCA CCG AAT A	GCT GCC CAT CAT CAT GAC CT		28436958
TP53	p53/ID: 7157	ACA AGG TTG ATG TGA CCT GGA	TGT AGA CTC GTG AAT TTC GCC	40804465c3

Note: Cycling number: 40; annealing temperature: 60°C for all primers.

In addition, our recent single‐cell RNA‐sequencing study has shown that NP and inner AF cells, both of which exhibit a chondrocyte‐like phenotype, share similar transcriptional profiles (Cherif et al. 2022). Consequently, commonly used NP phenotypic markers, such as TIE2 and ACAN, are also robustly expressed in the inner AF cells. Therefore, to simplify experimental design and maintain consistency, these NP markers were exclusively assessed in the qPCR analyses to evaluate the phenotypic rescue capacity of the miRNA mimic treatments in this study.

2.7. Protein Expression Assessment

MiRNA transfection: Human IVD cells were transfected with let‐7b‐5p, miR‐100‐5p or a combination of let‐7b‐5p and miR‐100‐5p (5 nM of each miRNA in serum‐free complete media). The culture media were spiked with the same amount of miRNA mimics after 24 h (Figure 1B(b)). Conditioned media were collected afterwards, centrifuged at 1500 rpm for 5 min to remove cellular debris, and stored at −80°C for further analysis.

Inflammatory mediator analysis: Fifteen inflammatory mediators of interest were analysed using the custom Human ProcartaPlex Mix&Match kit (#PPX‐15‐MXCE7EV) (ThermoFisher Scientific, Toronto) and measured with the Luminex xMAP system (Serial #MAGPX11188004) (ThermoFisher Scientific, Toronto) according to the manufacturer's instructions as previously described (Patron et al. 2024). The samples were pre‐diluted 1:3 with DMEM media. Standard curves for each analyte were generated with a 4‐parameter logistic (4‐PL) algorithm. Both median fluorescence intensity (MFI) corrected with background and the final concentration (pg/mL) were reported by the ThermoFisher ProcartaPlex analysis platform. The concentrations of all analytes were within the detection range with all biological replicates in at least one sample group. Few concentrations were out of the detection range and were extrapolated using the associated standard curves for further analysis.

2.8. Statistical Analysis and Visualisation

Statistical analyses were performed by GraphPad Prism 10 [version 10.3.1 (464)] (GraphPad Software, Inc., La Jolla, CA, USA). Shapiro–Wilk normality tests were conducted for all quantitative data (Ghasemi and Zahediasl 2012). Two‐tailed paired Student's t test, Welch's analysis of variance (ANOVA) test and one‐way ANOVA test were used for the parametric data accordingly. Friedman test was used for the nonparametric data. p < 0.05 was considered statistically significant. Data are presented as mean ± standard error of the mean (SEM). All assessments were conducted with three or more independent experiments, indicated by “n” in the figure legends. Venn diagrams were generated by Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/).

3. Results

3.1. The miRNA Cargo Profile and Distribution in EVs From Human IVD Cells

Small RNA sequencing analysis was performed to establish the landscape and evaluate miRNA abundance in IVD cell‐derived EVs from tissues of different degrees of degeneration. Three groups were created of Non‐deg, Mildly‐deg and Deg tissues, respectively. A total of 2632 miRNAs (including isoforms) were identified across the samples and groups, and 14.9% (393 miRNAs) of them were detected with reliable abundance values (CPM). Of the 393 miRNAs, 184 (46.8%) were shared between the groups (Figure 2). Forty‐eight miRNAs (12.2%) were detected only in the Non‐deg and Mildly‐deg groups, 26 miRNAs (6.6%) were detected only in the Non‐deg and Deg groups, and nine miRNAs (2.3%) were detected only in the Mildly‐deg and Deg groups (Figure 2). In addition, 68 miRNAs (17.3%) were exclusively detected in the Non‐deg group, 38 miRNAs (9.7%) were detected solely in the Mildly‐deg group and 20 miRNAs (5.1%) were only detected in the Deg group (Figure 2). In summary, less than 50% of miRNAs were shared by all three groups, with the highest number of unique miRNAs found in the Non‐deg samples.

FIGURE 2.

miRNA distribution in extracellular vesicles (EVs) isolated from human intervertebral disc (IVD) cells. The Venn diagram represents overlapping and unique miRNAs in EVs from Non‐deg (green), Mildly‐deg (orange) and Deg (blue) human IVD samples.

3.2. Comparisons of the Pathways and Functional Annotations of the miRNAs Exclusively Detected in One Group

The relative expression level of miRNAs exclusively detected in one group was low, ranging between 54 and 714 CPM (Figure 3A). Overall, KEGG analysis identified five common pathways in the three groups (Figure 3B and Table 3). The enrichment of pathways such as regulation of the actin cytoskeleton, Hippo signalling, focal adhesion, transforming growth factor (TGF)‐beta and Wnt signalling indicates that the miRNAs are involved in key processes governing cell structure, survival, inflammation and tissue regeneration. In addition, five pathways were exclusively detected in the Deg group, and the miRNA cargo of EVs from the Deg samples was associated with the most distinctive KEGG pathways (Table 3). The enrichment of pathways such as cell cycle arrest, p53 signalling, vascular endothelial growth factor (VEGF) signalling, apoptosis and cellular senescence indicates a cellular landscape marked by stress, aging and degeneration. These pathways are commonly activated in response to DNA damage, oxidative stress and inflammatory stimuli, all of which are characteristic of IVD degeneration. Their presence suggests that the miRNA cargo within EVs may either reflect a degenerative state or actively modulate these processes, offering potential therapeutic avenues to counteract cell loss, inflammation and functional decline. A Venn diagram of unique and overlapping GO terms with biological function (BP), cellular component (CC) and molecular function (MF) are in Figure S1 and Tables S1–S3.

FIGURE 3.

miRNAs and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with the uniquely identified miRNAs. (A) Heatmap demonstrating the expression of miRNAs uniquely detected in the (a) Non‐deg, (b) Mildly‐deg and (c) Deg samples. The yellow‐blue colour strip represents count per million (CPM) mean: expression is depicted from high in yellow to low in blue, and rows represent DEMs. n = 3–4. (B) Venn diagram presenting the distribution of the Top 10 KEGG pathways associated with musculoskeletal disease, shared and exclusively detected in Non‐deg (green), Mildly‐deg (orange) and Deg (blue) samples.

TABLE 3.

Shared and unique Top 10 musculoskeletal disease‐associated KEGG pathways of miRNAs detected in the Non‐deg, Mildly‐deg and Deg samples.

Groups	Quantity of shared pathways	Pathways
Detected in the Non‐deg, Mildly‐deg and Deg groups	5	Regulation of actin cytoskeleton
		Hippo signalling pathway
		Focal adhesion
		TGF‐beta signalling pathway
		Wnt signalling pathway
Detected in the Non‐deg and Mildly‐deg groups	2	ECM‐receptor interaction
		PI3K‐Akt signalling pathway
Exclusively detected in the Non‐deg group	3	mTOR signalling pathway
		AMPK signalling pathway
		TNF signalling pathway
Exclusively detected in the Mildly‐deg group	3	FoxO signalling pathway
		Adherens junction
		MAPK signalling pathway
Exclusively detected in the Deg group	5	Cell cycle
		p53 signalling pathway
		VEGF signalling pathway
		Apoptosis
		Cellular senescence

3.3. Comparisons of the Pathways and Functional Annotations of the Shared miRNAs Detected in Two of the Three Groups

As with the uniquely detected miRNAs, the shared miRNAs present in two of the three groups were expressed at low levels (Figure 4) across all pairwise comparisons: Non‐deg and Mildly‐deg groups (Figure 4A; 54–546 CPM), Non‐deg and Deg groups (Figure 4B; 54–571 CPM) and Mildly‐deg and Deg groups (Figure 4C; 78–367 CPM). KEGG analysis consistently highlighted focal adhesion and phosphatidylinositol 3‐kinase (PI3K)‐Akt signalling pathway in the comparisons (Table 4). The pathways integrate cell‐ECM mechanotransduction with survival and anabolic control, supporting proteoglycan and collagen synthesis while constraining inflammatory cascades and apoptosis. Their recurrence suggests common, stage‐spanning activities through which EV‐miRNAs modulate disc homeostasis. The unique pathway and annotation results are described in the Supporting Information (Tables S4–S12). Briefly, the pairwise analyses reveal that shared pathways focal adhesion and PI3K‐Akt signalling pathway span stages, early Non‐deg‐anchored shifts in growth control and matrix signalling, Mildly‐deg‐anchored stress and phenotype reprogramming, and Deg‐anchored cytoskeletal and mechanical rewiring with metabolic and junctional liabilities. This framework pinpoints actionable pathways for EV‐miRNA interventions aimed at restoring mechanotransduction balance, sustaining anabolic capacity, and curbing inflammation, apoptosis and neovascular cues.

FIGURE 4.

The relative expression of miRNAs shared between two of the three groups. The Heatmap shows differentially expressed miRNAs comparing (A) Non‐deg and Mildly‐deg, (B) Non‐deg and Deg and (C) Mildly‐deg and Deg samples. The yellow‐blue colour strip represents count per million (CPM) mean: expression is depicted from high in yellow to low in blue, and rows represent DEMs. n = 3–4.

TABLE 4.

Shared and unique Top 10 musculoskeletal disease‐associated KEGG pathways of miRNAs detected in two of the three groups.

Groups	Quantity of shared pathways	Pathways
Shared across Non‐deg/Mildly‐deg, Non‐deg/Deg and Mildly‐deg/Deg	2	Focal adhesion
		PI3K‐Akt signalling pathway
Non‐deg/Mildly‐deg and Non‐deg/Deg	5	Cell cycle
		p53 signalling pathway
		Hippo signalling pathway
		TGF‐beta signalling pathway
		Ubiquitin mediated proteolysis
Non‐deg/Mildly‐deg and Mildly‐deg/Deg	3	FoxO signalling pathway
		MAPK signalling pathway
		Wnt signalling pathway
Non‐deg/Deg and Mildly‐deg/Deg	1	Regulation of actin cytoskeleton
Non‐deg/Mildly‐deg	0	None
Non‐deg/Deg	2	AMPK signalling pathway
		Adherens junction
Mildly‐deg/Deg	4	Apoptosis
		VEGF signalling pathway
		mTOR signalling pathway
		Tight junction

3.4. The Differential Expression and Comparisons of the Pathways and Functional Annotations of the miRNAs Shared Between All Groups

Although Figures 3 and 4 highlight stage‐specific miRNAs, the most robust homeostatic regulators are likely those conserved across all stages. Across the 184 miRNAs shared by all groups, the expression level ranged from 54 to 227,602 CPM. KEGG analysis converged on pathways governing senescence control, survival/anabolism and mechanotransduction (Figure 5A). Consistent with this, GO‐BP highlighted cell cycle/protein phosphorylation and ECM organisation (Figure S2A); GO‐CC emphasised focal adhesions, cytoskeleton and ECM/collagen compartments (Figure S2B); and GO‐MF underscored integrin/cadherin/collagen binding alongside kinase and ubiquitin‐transferase activities (Figure S2C).

FIGURE 5.

Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis and relative expression of miRNA shared between all three groups. (A) KEGG analyses showing the Top 10 musculoskeletal disease‐associated pathways, p value threshold of 0.05. The x‐axis represents the logarithmic scale of the false discovery rate (FDR) [−Log₁₀ (FDR)], and the y‐axis describes the enriched KEGG pathways. (B) Heatmaps displaying miRNAs within different count per million (CPM) ranges: (a) CPM ≥ 25 k, (b) 25k > CPM ≥ 5k, (c) 5k > CPM ≥ 1k, (d) 1k > CPM ≥ 500 and (e) 500 > CPM. The yellow‐blue colour strip represents CPM mean: expression is depicted from high in yellow to low in blue, and rows represent DEMs. n = 3–4.

Ten miRNAs were expressed at very high levels exceeding 25,000 CPM. The Top 10 abundant miRNAs are let‐7b‐5p, let‐7a‐5p, let‐7i‐5p, miR‐21‐5p, miR‐16‐5p, miR‐199a‐3p, miR‐125b‐5p, let‐7f‐5p, let‐7c‐5p and miR‐100‐5p (Figure 5B). KEGG analysis supports the shared core: cell cycle, forkhead box O (FoxO), Hippo and p53 signalling pathways, indicating a strong control of senescence checkpoints, survival and mechanotransduction. The added terms: proteoglycans in cancer, adherens junction and oocyte meiosis expand their predicted functions towards broad stress‐response and adhesion remodelling. GO‐BP/CC/MF reinforce a transcription‐centric program anchored in the nucleus/nucleoplasm and coupled to cytoskeletal compartments (Figure S3A).

Fifteen miRNAs were expressed at high levels with a CPM range of 5000–25,000 (Figure 5B). KEGG analysis largely supported the shared backbone: focal adhesion, PI3K‐Akt, Hippo, actin cytoskeleton, TGF‐beta, Wnt, FoxO signalling pathways and uniquely adds mitogen‐activated protein kinase (MAPK), adherens junction, and ubiquitin‐mediated proteolysis, linking survival/anabolism to junctional rewiring and proteostasis. GO terms concentrate on ECM organisation and focal/adherens junction locales with integrin/cadherin/collagen binding and kinase/ubiquitin activities (Figure S3B). Forty‐one miRNAs were expressed at moderate levels with a CPM range of 1000–5000 (Figure 5B). KEGG retains core pathways and uniquely introduces ECM‐receptor interaction, MAPK and ubiquitin‐mediated proteolysis, emphasising matrix‐receptor crosstalk and stress adaptation. GO highlights basement membrane/ECM compartments and SMAD/actin/collagen binding (Figure S3C). Nineteen miRNAs were expressed at low levels with a CPM mean range of 500–1000 (Figure 5, B). KEGG overlaps with the core and uniquely adds ECM‐receptor interaction, adherens junction, MAPK and VEGF signalling pathways, pointing to adhesion remodelling and angiogenic cues. GO emphasises cytoskeleton/focal adhesion/ECM locales with SMAD/actin and ubiquitin/kinase interfaces (Figure S3D). Ninety‐nine miRNAs were expressed at very low levels with a CPM range of 54–500 (Figure 5, B). KEGG again aligns with the shared backbone and adds ECM‐receptor interaction, adherens junction and MAPK signalling pathways, consistent with low‐abundance modulators of adhesion‐cytoskeleton‐inflammation coupling. GO terms support the functions (Figure S3E).

All abundance groups converge on cellular‐senescence control and survival/anabolism and on mechanotransduction/adhesion‐ECM circuitry. Together, the data support a model in which EV‐miRNAs cooperatively sustain matrix homeostasis, limit apoptosis and senescence, and dampen inflammatory and angiogenic drivers across degeneration stages.

3.5. Selection of Five miRNAs for Functional Testing

We focused on miRNAs with high abundance across samples, using the average abundance of all groups as a criterion for selection (Table 5). Let‐7b‐5p and let‐7a‐5p were the two most abundant miRNAs identified in our analysis and are known for anti‐inflammatory effects in various models (Brennan et al. 2017; Xiang et al. 2017; Tan et al. 2019; Kumar et al. 2011; Xie et al. 2021). Notably, let‐7b‐5p exhibited a significantly lower expression in Deg samples than in Non‐deg samples (p = 0.0243). Let‐7f‐5p showed consistent expression across all groups, as indicated by its small standard deviation of CPM values. MiR‐16‐5p supplementation has shown mixed effects on inflammatory responses in osteoarthritis and spinal cord injury studies (Zhao et al. 2020; Lei et al. 2019; Yang et al. 2022), and it showed a decreasing trend in expression from Non‐deg to Deg samples (p = 0.0390). In addition, miR‐100‐5p was evenly expressed across the groups and was selected because it is known primarily as a tumour suppressor in various cancers and could potentially play a role in the regulation of cellular senescence (Ye et al. 2020; Zhang et al. 2022). In summary, our selected miRNAs include let‐7b‐5p and let‐7a‐5p, which are in high relative abundance and known for anti‐inflammatory effects; let‐7f‐5p and miR‐16‐5p, which are linked to inflammatory responses; and miR‐100‐5p, which is implicated in the regulation of cellular senescence (Figure 6A and Table 5).

TABLE 5.

Top 10 abundant shared miRNAs in the Non‐deg, Mildly‐deg and Deg samples.

Abundance rank	Overall	Non‐deg group	Mildly‐deg group	Deg group
1	let‐7b‐5p	let‐7b‐5p	let‐7b‐5p	let‐7a‐5p
2	let‐7a‐5p	let‐7a‐5p	let‐7a‐5p	let‐7b‐5p
3	let‐7i‐5p	let‐7i‐5p	let‐7i‐5p	let‐7i‐5p
4	miR‐21‐5p	miR‐21‐5p	miR‐21‐5p	miR‐21‐5p
5	miR‐16‐5p	miR‐16‐5p	miR‐16‐5p	miR‐199a‐3p
6	miR‐199a‐3p	miR‐125b‐5p	miR‐199a‐3p	let‐7f‐5p
7	miR‐125b‐5p	miR‐199a‐3p	let‐7f‐5p	miR‐125b‐5p
8	let‐7f‐5p	let‐7f‐5p	miR‐125b‐5p	miR‐16‐5p
9	let‐7c‐5p	let‐7c‐5p	let‐7c‐5p	let‐7c‐5p
10	miR‐100‐5p	miR‐196a‐5p	miR‐100‐5p	miR‐100‐5p

Note: Bold font indicates miRNAs of interest.

FIGURE 6.

The relative expression and regulatory effects of five selected miRNAs. (A) Count per million (CPM). Regulatory effects of single miRNA mimic treatment on NP cell phenotypic marker expression: under (B) non‐inflammatory conditions and (C) inflammatory conditions. Regulatory effects of single miRNA mimic treatment on inflammatory mediators and senescence marker expression: under (D) non‐inflammatory conditions and (E) inflammatory conditions. Regulatory effects of miRNA mimic combination treatment on (F) NP cell phenotypic markers and (G) inflammatory mediators and senescence markers. Data were normalised to vehicle‐treated cells. All values are presented as mean ± SEM. The data in (A) was analysed by Welch's analysis of variance (ANOVA): p < 0.05. n = 3–4. The data in (B) (SOX9, IL‐6 and IL‐8) and (C) (p16) are non‐parametric and were analysed by Friedman test with Dunn's multiple comparisons test as indicated in Methods. *, ** and *** indicate a statistical significance of p < 0.05, p < 0.01 and p < 0.001. All other data were analysed by one‐way ANOVA with Dunnett's T3 multiple comparisons test. n = 10.

3.6. The Regulatory Effect of the Selected miRNAs in Human IVD Cells

3.6.1. The Regulatory Effects of Single‐miRNA Treatment on Cell Phenotype

Each of the five selected miRNA mimics was applied to IVD cells under non‐inflammatory and inflammatory conditions (Figure S4A), and the expression of NP markers was assessed. MiR‐100‐5p significantly upregulated TIE2 expression (2.20 ± 0.42, p = 0.0201) under non‐inflammatory conditions (Figure 6B). PAX1 expression was significantly upregulated by miR‐100‐5p (1.38 ± 0.11, p = 0.0125), miR‐16‐5p (1.28 ± 0.06, p = 0.0035) and let‐7a‐5p (1.21 ± 0.05, p = 0.0079). No significant change was observed in the expression of the NP markers Forkhead Box F1 (FOXF1), SOX9 and ACAN under non‐inflammatory conditions. PAX1 expression was significantly upregulated by let‐7a‐5p (1.31 ± 0.08, p = 0.0074) and let‐7b‐5p (1.35 ± 0.11, p = 0.0308) under inflammatory conditions. In addition, no significant change was observed in the expression of NP markers TIE2, FOXF1, SOX9 and ACAN under inflammatory conditions (Figure 6B).

3.6.2. The Regulatory Effect of Single‐miRNA Treatment on Inflammatory Mediators and Senescence Markers

Inflammatory mediator [Interleukin (IL)‐1 beta (IL‐1β), IL‐6 and IL‐8] and senescence marker (p16 and p53) expression was very low under non‐inflammatory conditions, and no significant change was observed in their expression following miRNA treatment (Figure 6C). Their expression was as expected elevated under inflammatory conditions (Figure S4A), where miR‐100‐5p (0.59 ± 0.09, p = 0.0382), let‐7a‐5p (0.49 ± 0.08, p = 0.0107) and let‐7b‐5p (0.53 ± 0.08, p = 0.0144) significantly downregulated IL‐1β. Let‐7b‐5p significantly downregulated IL‐6 (0.77 ± 0.07, p = 0.0094) and p16 (0.70 ± 0.13, p = 0.0461). No significant change was observed in the expression of IL‐8 and p53 (Figure 6C).

Having established that let‐7b‐5p predominantly targets inflammatory pathways (reducing IL‐1β, IL‐6) while miR‐100‐5p exhibits anabolic potential (upregulating TIE2), we hypothesised that a combinatorial approach could address both the catabolic and inflammatory aspects of IVD degeneration simultaneously. Consequently, we selected miR‐100‐5p, let‐7b‐5p and miR‐16‐5p to construct a combination treatment matrix. The selection was based on their distinct regulatory effects on key NP markers, inflammatory mediators and senescence markers under inflammatory and non‐inflammatory conditions.

3.6.3. The Regulatory Effects of miRNA Combination Treatment on Cell Phenotype

To explore their combined effects, miRNA mimics were combined in four treatment groups: (1) miR‐100‐5p + let‐7b‐5p, (2) miR‐16‐5p + let‐7b‐5p, (3) miR‐100‐5p + miR‐16‐5p and (4) miR‐100‐5p + miR‐16‐5p + let‐7b‐5p. These treatments were applied to IVD cells under inflammatory and non‐inflammatory conditions, and their effects on NP markers, inflammatory mediators and senescence markers were assessed. MiR‐100‐5p + let‐7b‐5p significantly upregulated the expression of TIE2 (1.63 ± 0.13, p = 0.0005), FOXF1 (1.60 ± 0.17, p = 0.0060), PAX1 (1.37 ± 0.12, p = 0.0211) and ACAN (1.43 ± 0.15, p = 0.0308) under non‐inflammatory conditions. No significant change was observed in the expression of the NP marker SOX9 under non‐inflammatory conditions. MiR‐100‐5p + let‐7b‐5p significantly upregulated TIE2 expression (1.64 ± 0.16, p = 0.0089) under inflammatory conditions. MiR‐100‐5p + miR‐16‐5p + let‐7b‐5p significantly upregulated FOXF1 (1.50 ± 0.15, p = 0.0173) and PAX1 (1.27 ± 0.09, p = 0.0350) expression under inflammatory conditions. No significant changes were observed in the expression of the NP markers SOX9 and ACAN under inflammatory conditions (Figure 6D).

3.6.4. The Regulatory Effects of miRNA Combination Treatment on Inflammatory Mediators and Senescence Markers

Under inflammatory conditions, miR‐100‐5p + let‐7b‐5p significantly downregulated the expression of IL‐1β (0.64 ± 0.12, p = 0.0338), IL‐8 (0.69 ± 0.07, p = 0.0487) and p16 (0.63 ± 0.07, p = 0.0079). MiR‐16‐5p + let‐7b‐5p significantly downregulated the expression of IL‐1β (0.55 ± 0.09, p = 0.0107), IL‐6 (0.69 ± 0.05, p = 0.0024), IL‐8 (0.69 ± 0.06, p = 0.0024) and p16 (0.60 ± 0.06, p = 0.0167). MiR‐100‐5p + miR‐16‐5p significantly downregulated the expression of IL‐6 (0.78 ± 0.04, p = 0.0019), IL‐8 (0.73 ± 0.07, p = 0.0045) and p16 (0.63 ± 0.08, p = 0.0113). In addition, miR‐100‐5p + miR‐16‐5p + let‐7b‐5p significantly downregulated the expression of IL‐1β (0.51 ± 0.08, p = 0.0200), IL‐6 (0.73 ± 0.04, p = 0.0007), IL‐8 (0.72 ± 0.07, p = 0.0058) and p16 (0.69 ± 0.10, p = 0.0394) (Figure 6E). No significant changes were observed in the expression of p53 (Figure 6E) under inflammatory conditions and all the markers under non‐inflammatory conditions (Figure S4B).

Overall, the results suggest that the combination treatment with miR‐100‐5p + let‐7b‐5p has promising regulatory effects on maintaining IVD cell phenotype with anti‐inflammatory and anti‐senescence effects in inflammatory conditions.

3.7. The Anti‐Inflammatory Effects of miRNA Treatment at the Protein Level

We quantified the expression of 15 inflammatory mediators under inflammatory conditions (Figure S5). Our results showed that the production of three cytokines [IL‐6, IL‐1 alpha (IL‐1α) and tumour necrosis factor alpha (TNF‐α)] was significantly reduced by miR‐100‐5p, let‐7b‐5p and their combination treatment. In addition, the production of two cytokines [IL‐1β and interferon gamma (IFN‐γ)] was significantly reduced by the combination treatment of miR‐100‐5p and let‐7b‐5p. A decreasing trend for IL‐21 was observed following all treatments; however, no statistically significant difference was detected. Anti‐inflammatory mediators were also assessed. IL‐10 expression was significantly increased by miR‐100‐5p, let‐7b‐5p and their combination treatment compared with the control group (Figure 7A). An increasing trend of IL‐4 production was observed following all treatments; however, no statistically significant difference was detected (Figure 7A). Quantitative data are presented in Table S13.

FIGURE 7.

Protein expression of inflammatory mediators post‐miRNA treatment. (A) cytokine expression, (B) chemokine expression and (C) growth factor expression. Concentrations are presented as mean ± standard error of the mean (SEM) in pg/mL. All data were analysed by one‐way analysis of variance (ANOVA) with Tukey's multiple comparisons test. *, **, *** and **** indicate a statistical significance of p < 0.05, p < 0.01, p < 0.001 and p < 0.0001, compared with the control group, respectively. n = 8.

In addition, chemokine CXCL10 expression was significantly decreased by miR‐100‐5p, let‐7b‐5p and their combination treatment compared with the control group. Furthermore, the production of three chemokines (IL‐8, CXCL1 and CCL24) was significantly reduced by the combination treatment of miR‐100‐5p and let‐7b‐5p. A decreasing trend for CXCL5 was observed following all treatments; however, no statistically significant difference was detected. The expression of anti‐inflammatory chemokine CCL22 was significantly increased by miR‐100‐5p, let‐7b‐5p and their combination treatment compared with the control group (Figure 7B). The expression of growth factor brain‐derived neurotrophic factor (BDNF) was significantly decreased by miR‐100‐5p, let‐7b‐5p and their combination treatment compared with the control group (Figure 7C). Quantitative data are presented in Table S13.

4. Discussion

This work combines insights from profiling miRNA cargo in EVs with functional prediction analyses to identify miRNAs with specific roles in IVD maintenance and degeneration. Previous studies emphasised the significance of miRNAs in IVD degeneration (Cazzanelli and Wuertz‐Kozak 2020; Jiang et al. 2021; Ji et al. 2018; Chen et al. 2023) and their emerging roles in EV‐mediated cell communication, which can potentially be leveraged for regenerative medicine (Tang et al. 2021; Munir et al. 2020; Teo et al. 2023Teo et al. 2023; Muskan et al. 2024). However, while studies have profiled miRNAs in IVD tissues (Ohrt‐Nissen et al. 2013; Zhao et al. 2013; Sherafatian et al. 2019), there is a lack of systematic investigation into miRNA cargo in EVs derived from human IVD cells, especially across different stages of degeneration. This gap motivates our study's objectives: to profile miRNA cargo in EVs from human IVD cells representing various degeneration grades and to explore the therapeutic functions of selected miRNAs through function prediction‐guided screening.

Our results showed that 46.8% of the miRNA cargo was shared among Non‐deg, Mildly‐deg and Deg samples. This is much less than the proportion of shared protein cargo from the same EV samples, which was 88.6% (Li et al. 2024), suggesting a more dynamic role for miRNAs in response to degeneration‐specific cellular changes. Specific miRNAs, such as miR‐100‐5p and let‐7b‐5p, were identified as highly enriched in EVs from IVD cells. These miRNAs demonstrated distinct regulatory functions, including the modulation of NP phenotypic markers, inflammatory mediators and senescence markers. For instance, miR‐100‐5p increased TIE2 expression, which is essential for cell survival, proliferation, cellular maintenance and regenerative responses in IVD cells (Sakai et al. 2012; Wangler et al. 2019; Soma et al. 2023). At the same time, let‐7b‐5p reduced IL‐1β expression, a pro‐inflammatory cytokine implicated in inflammatory responses, cell apoptosis, cellular senescence and ECM breakdown within degenerated IVD tissues (Hoyland et al. 2008; Wang et al. 2020). Moreover, a combination of these miRNAs demonstrated synergistic therapeutic effects on promoting IVD cell phenotypic marker expression and reducing inflammatory mediator and senescence marker expression, suggesting potential for combined miRNA therapies in future applications.

4.1. Comparison of EV‐Enriched and Tissue Lysate‐Enriched miRNAs in Human IVD

Previous studies have profiled miRNAs from human IVD tissue (Table 6). Ohrt‐Nissen et al. (2013) compared miRNA profiles of Deg NP and AF tissues and highlighted a high abundance of growth factor‐associated signalling pathways influenced by miRNAs, including TGF‐beta, platelet‐derived growth factor, insulin‐like growth factor, epidermal growth factor signalling pathways, which are critical for cellular processes in IVD homeostasis. Zhao et al. (2013) profiled and compared miRNAs in Non‐deg and Deg NP tissue, linking the dysregulated miRNAs to anti‐apoptosis, cell proliferation, cell survival and angiogenesis‐associated pathways, suggesting their role in the homeostasis and pathogenesis of IVD degeneration. In addition, Sherafatian et al. (2019) conducted a meta‐analysis of three miRNA datasets of Non‐deg and Deg NP tissue, identifying commonly dysregulated miRNAs associated with anti‐apoptosis and cell proliferation‐associated pathways.

TABLE 6.

Comparison of EV‐enriched and tissue lysate‐enriched miRNAs in human IVD.

Study	Sample source	miRNA expression		Highly‐associated pathways
(Ohrt‐Nissen et al. 2013) (PMID: 23586579)	Human degenerate NP tissue	miRNAs in high‐abundance (Top 10): miR‐342‐3p miR‐223 miR‐668 miR‐627 miR‐483‐3p miR‐616* miR‐135a* miR‐1253 miR‐135b* miR‐561		Top canonical pathways: TGF‐beta signalling pathway, Platelet‐derived growth factor (PDGF) signalling pathway, Insulin‐like growth factor (IGF) signalling pathway, Epidermal growth factor (EGF) signalling pathway, Actin cytoskeletal pathway
	Human degenerate annulus fibrosus (AF) tissue	miRNAs in high‐abundance (Top 10): miR‐342‐3p miR‐223 miR‐154 miR‐433 miR‐483‐3p miR‐135a* miR‐21* miR‐622 miR‐185* miR‐135b*
(Zhao et al. 2013) (PMID: 24173697)	Human degenerate NP tissue	(Compared to non‐degenerate NP tissue) Upregulated miRNAs: hsv2‐miR‐H6* miR‐3680 miR‐147b miR‐920 ebv‐miR‐BART21‐3p hsv1‐miR‐H3 miR‐130b* miR‐4275 miR‐299‐5p miR‐146b‐3p miR‐764 miR‐3663‐5p miR‐18b* miR‐377* miR‐2355‐5p miR‐3681 miR‐675* miR‐3126‐5p miR‐3654 miR‐200c miR‐363* miR‐2113 miR‐1909 miR‐4297 miR‐1322	(Compared to non‐degenerate NP tissue) Downregulated miRNA: miR‐10a* hsv2‐miR‐H3 miR‐3138 miR‐664 ebv‐miR‐BART17‐5p miR‐532‐5p miR‐518c* miR‐182 miRPlus‐l152* miR‐203 miR‐200b* miR‐139‐3p miR‐25* miR‐H6‐5p miR‐676 miRPlus‐l107* miR‐148a miR‐1266 miR‐3065‐3p miR‐493* miR‐411* miR‐331‐3p miR‐3065‐5p miR‐34a miR‐516b miR‐3128	Top KEGG pathways: Anti‐apoptosis/pro‐cell proliferation‐associated pathways: Focal adhesion, Regulation of actin, Adherens junction, Wnt signalling pathway, Cell proliferation/survival/angiogenesis‐associated pathways: Mitogen‐activated protein kinase (MAPK) signalling pathway, Epidermal growth factor receptor (EGFR; ErbB) signalling pathway, Phosphoinositide 3‐kinase (PI3K)‐Akt signalling pathway,
(Sherafatian et al. 2019) (PMID: 30904808)	Human degenerate NP tissue	(Compared to non‐degenerate NP tissue) Upregulated miRNAs: miR‐199a‐5p miR‐574‐3p miR‐551a miR‐640	(Compared to non‐degenerate NP tissue) Downregulated miRNA: miR‐483	Top KEGG pathways: Anti‐apoptosis/pro‐cell proliferation‐associated pathways: ECM‐receptor interaction, Adherens junction, Proteoglycans in cancer, Hippo signalling pathway, TGF‐beta signalling pathway
Our study	EVs derived from the mixed populations of NP and inner AF cells isolated from non‐degenerate, mildly‐degenerate, and degenerate human IVD tissues	Shared miRNAs in very‐high‐abundance (CPM > 25,000): let‐7b‐5p let‐7a‐5p let‐7i‐5p miR‐21‐5p miR‐16‐5p miR‐199a‐3p miR‐125b‐5p let‐7f‐5p let‐7c‐5p miR‐100‐5p		Top KEGG pathways: Anti‐apoptosis/pro‐cell proliferation‐associated pathways: Hippo signalling pathway, Ubiquitin‐mediated proteolysis, TGF‐beta signalling pathway, Focal adhesion, Regulation of actin cytoskeleton, Wnt signalling pathway, Glycosaminoglycan biosynthesis‐chondroitin sulphate/dermatan sulphate Inflammation/apoptosis‐associated pathways: MAPK signalling pathway, TNF signalling pathway
		Shared miRNAs in high‐abundance (25,000 > CPM > 5000): miR‐196a‐5p miR‐125a‐5p miR‐29a‐3p miR‐26a‐5p miR‐148a‐3p miR‐10a‐5p miR‐92a‐3p miR‐155‐5p let‐7e‐5p miR‐127‐3p miR‐221‐3p miR‐151a‐3p miR‐486‐5p miR‐27b‐3p miR‐10b‐5p		Top KEGG pathways: Anti‐apoptosis/pro‐cell proliferation‐associated pathways: Ubiquitin‐mediated proteolysis, Focal adhesion, Hippo signalling pathway, Regulation of actin cytoskeleton, TGF‐beta signalling pathway, Adherens junction, Wnt signalling pathway Cellular senescence‐associated pathways: FoxO signalling pathway, PI3K‐Akt signalling pathway Angiogenesis‐associated pathways: MAPK signalling pathway

Our findings reveal that the miRNA profiles of EVs derived from IVD cells differ significantly from those of tissue lysates. We did not identify any overlap between miRNAs with an abundance above 5000 CPM in our study with the high‐abundance or dysregulated miRNA described in the studies mentioned above (Ohrt‐Nissen et al. 2013; Zhao et al. 2013; Sherafatian et al. 2019). However, pathways associated with anti‐apoptosis, cell proliferation, angiogenesis, inflammation and cellular senescence were also identified in our samples. These differences may indicate the selective packaging of miRNAs into EVs, aligning with studies demonstrating that EVs actively transport specific miRNAs (Guduric‐Fuchs et al. 2012; Groot and Lee 2020; Martins‐Marques et al. 2022), likely reflecting the regulatory needs of target cells under specific physiological and pathological conditions. In addition, when interpreting the different miRNA profiles between EV and tissue samples, parameters such as tissue source and technical assessment should be taken into consideration. We purified miRNAs from EVs derived from a mixed population of NP and inner AF cells isolated from human IVD tissues with different degrees of degeneration, while the other studies purified miRNAs directly from Non‐deg and Deg human NP and AF tissues and not from EVs, respectively. In addition, we evaluated EV‐enriched miRNA profiles using the small RNA sequencing technique, while the other studies used microarray assays. Future studies comparing miRNA profiles between EV and cell/tissue lysate samples should use the same culture system and analytical technique and method to minimise systematic variations.

4.2. Comparison of Human IVD Cell‐Derived EV‐Enriched miRNAs and NP Stem Cell‐Derived Exosome‐Enriched miRNAs

A previous study reported relative expression of miRNAs from exosomes derived from human NP‐derived stem cells (NPSC) isolated from Non‐deg and Deg NP tissues (Table 7) (Zhuang et al. 2022). We detected all ten significantly dysregulated miRNAs that were reported and identified different relative expression patterns. Particularly, the authors detected higher expression of miR‐100‐5p compared with that of let‐7b‐5p. Their relative expression significantly increased in the Deg samples. We detected the highest average expression of let‐7b‐5p over all detected miRNAs. The average expression of miR‐100‐5p was ranked tenth among all detected miRNAs in our samples. In addition, we found that the relative expression of let‐7b‐5p significantly decreased in the Deg samples compared with that of the Non‐deg samples, which showed the opposite expression pattern to the reported data. We did not detect a significant difference in the relative expression of miR‐100‐5p among our samples, which is also different from the reported data (Zhuang et al. 2022). The remaining eight reported miRNAs were also detected in all samples with various expression levels, except for miR‐107, which was detected only in the Non‐deg and Mildly‐deg samples in our study.

TABLE 7.

Comparison of human IVD cell‐derived EV‐enriched miRNAs and NP stem cell‐derived exosome‐enriched miRNAs.

	NP stem cell‐derived exosomes (Y. 66) (PMID: 35111808)		IVD cell‐derived EVs (Our study)
miRNA	Detected in Non‐deg samples (TPM, × 103)	Detected in Deg samples (TPM, × 103)	CPM mean (× 103)	Distribution
let‐7b‐5p	12.1	60.8	178.2	Detected in all samples	Very‐high‐abundance group (CPM ≥ 25k)
miR‐100‐5p	48.8	156.5	25.2
miR‐10a‐5p	3.9	0.7	7.8		High‐abundance group (25k > CPM ≥ 5k)
miR‐10b‐5p	5.9	2.3	5.2
miR‐101‐3p	0.5	0.2	2.3		Intermediate‐abundance group (5k > CPM ≥ 1k)
miR‐103a‐3p	0.4	0.1	2.7
miR‐106b‐3p	0.07	0.3	0.4		Very‐low‐abundance group (500 > CPM)
miR‐10a‐3p	9.7	30.5	0.2
miR‐11400	3.4	11.3	CPM not calculatable		N/A
miR‐107	0.03	0.00001	0.2	Detected only in Non‐deg and Mildly‐deg samples	Very‐low‐abundance group (500 > CPM)

Abbreviations: CPM, counts per million; TPM, transcripts per million.

NP cells and NPSC are both found within the NP region of IVDs. They share a common origin and exhibit both similar and distinct biological characteristics. NPSCs and NP cells both express MSC surface markers CD73, CD90 and CD105, but NPSCs express higher levels of these markers (Chen et al. 2016), indicating a greater stemness. In addition, NPSCs lack the mature NP cell phenotypic marker CD24 (Guan et al. 2014). These similarities and differences between NPSCs and NP cells may explain the specific miRNA profile and relative expression pattern in EVs derived from each cell source.

4.3. miRNA Regulatory Effects on Cell Phenotype and Homeostasis in Human IVD Tissues

Previous studies showed a regulatory role in improving human IVD cell phenotypic markers, cellular homeostasis and tissue integrity. It was reported that the relative expression of miR‐129‐5p was significantly reduced in NP from Deg compared with Non‐deg tissue. Delivery of miR‐129‐5p via MSC‐derived EVs improved cell viability, reduced cell apoptosis and decreased ECM degradation in IL‐1β‐induced NP cells (Cui and Zhang 2021). We detected miR‐129‐5p only in EVs derived from Non‐deg samples. However, its relative expression was very low (CPM = 54). Similarly, a previous study showed that the relative expression of miR‐499a‐5p was reduced in NP from Deg compared with Non‐deg tissues. Overexpression of miR‐499a‐5p decreased the levels of catabolic enzymes MMP 3 and 13 and increased the synthesis of ECM compositions Type II collagen and ACAN in TNF‐α‐induced NP cells (Sun et al. 2019). However, we detected a weak signal of miR‐499a‐5p in all EV samples, and its relative expression was too low to be calculable using the CPM method. In addition, Zhuang et al. (2022) reported that let‐7b‐5p inhibited cell proliferation, migration and matrix synthesis while promoting apoptosis in human AF cells isolated from Deg IVD tissue. We did not detect any significant regulatory effect of let‐7b‐5p on the cell phenotypic and homeostatic markers. The differences between our results and those of previous studies indicate that selective packaging systems may exist in human IVD cells. IVD tissue‐enriched miRNAs with therapeutic potentials are not necessarily selected and packaged into IVD cell‐derived EVs for cell communication, cargo transportation and/or signalling transduction.

We observed that the EV‐enriched miR‐100‐5p demonstrated an upregulatory effect on TIE2 under non‐inflammatory conditions. When combined with let‐7b‐5p, the combination treatment upregulated TIE2 and FOXF1 under non‐inflammatory conditions and TIE2 under inflammatory conditions. These findings align with prior reports that miR‐100‐5p supports cellular homeostasis by engaging growth and survival pathways: it promotes myoblast proliferation while restraining differentiation via Trib2/mTOR/S6K signalling (Wang et al. 2023) and, by targeting homeodomain‐interacting protein kinase 2, it activates PI3K/AKT signalling in endothelial cells (Zheng et al. 2021).

4.4. Regulation of Inflammatory Response of miRNAs on Human IVD and Other Musculoskeletal Cells

MiRNAs play pivotal roles in regulating inflammatory responses in IVD and other musculoskeletal tissues, with effects varying by miRNA type and cellular context. For instance, Cao et al. (2024) reported that inhibition of let‐7b‐5p reduced pro‐inflammatory mediators like IL‐6, IL‐8 and MMP10 in chondrocytes, suggesting an inflammatory role. However, Palamà et al. (2023) found no significant change in IL‐6, IL‐8 or COX‐2 levels with let‐7b‐5p mimic treatment in IL‐1β‐induced human articular chondrocytes. In our study, the let‐7b‐5p mimic significantly downregulated the pro‐inflammatory mediators IL‐1β at the gene expression level and IL‐6, IL‐1α, TNF‐α, CXCL10 and BDNF at the protein level. In addition, the let‐7b‐5p mimic significantly upregulated the anti‐inflammatory mediators CCL22 and IL‐10 at the protein level. Our findings indicate an anti‐inflammatory role of let‐7b‐5p. These conflicting findings suggest that let‐7b‐5p's effects may depend on cell type and microenvironment, warranting further investigation.

Similarly, miR‐100‐5p has demonstrated anti‐inflammatory properties. It was shown to inhibit reactive oxygen species production and apoptosis in human chondrocytes (Li et al. 2021), while Gao et al. (2021) reported that exosomal miR‐100‐5p from human umbilical cord MSCs attenuated TNF‐α, IL‐1β, IL‐6 and IL‐8 in mouse eosinophils. In our study, the miR‐100‐5p mimic significantly downregulated IL‐1β gene expression level and IL‐6, IL‐1α, TNF‐α, CXCL10, BDNF, alongside a significant upregulation of CCL22 and IL‐10 at the protein level, further supporting its anti‐inflammatory property.

The combination treatment of let‐7b‐5p and miR‐100‐5p significantly downregulated 10 out of 12 proinflammatory mediators and significantly upregulated 2 out of 3 anti‐inflammatory mediators, indicating strong anti‐inflammatory properties. The expression of IL‐1β, IL‐8, CXCL1, CCL24 and IFN‐γ was significantly downregulated by the combination treatment, but not by the single‐miRNA treatment. No statistically significant difference was observed in the multiple comparisons between the three treatment groups, which suggests there was no added effect from the combination compared with the single‐miRNA treatment. More investigations are needed to further evaluate the potential of the added effects of combination treatment.

4.5. Regulation of Cellular Senescence of miRNAs on Human IVD and Other Musculoskeletal Cells

The regulation of cellular senescence of let‐7b‐5p is multifaceted. It was shown that let‐7b‐5p decreased the expression of p53 at the gene and protein expression levels in the human acute myeloid leukaemia cells (K562 and HL‐60), which was reversed when a let‐7b‐5p inhibitor was used (Dong et al. 2021). In contrast, Cao et al. (2024) demonstrated that inhibiting let‐7b‐5p reversed the increased cellular senescence in chondrocytes, indicating a pro‐senescence role of let‐7b‐5p. Our study showed that let‐7b‐5p or the combination of let‐7b‐5p and miR‐100‐5p significantly downregulated the expression of the cyclin‐dependent kinase inhibitor p16, which is strongly associated with reducing cellular senescence and the progression of tissue degeneration in IVDs. We did not detect significant dysregulation of p53 under single‐miRNA or combination treatment. The function of let‐7b‐5p appears to be context‐dependent, in addition, although miRNA inhibitors are designed to specifically bind and silence a specific miRNA, off‐target effects have been reported that could explain the differences. More investigations are required to determine the regulatory roles of let‐7b‐5p and miR‐100‐5p on cellular senescence in human IVD cells, as such miRNAs could be particularly valuable in cell‐free therapies aimed at managing cellular aging processes in IVD degeneration and other degenerative musculoskeletal diseases.

Taken together, miR‐100‐5p and let‐7b‐5p displayed considerable ECM preservation properties, anti‐senescence and anti‐inflammatory effects in IVD cells by modulating transcriptional regulation and protein production. Their distinct yet complementary roles underscore the therapeutic potential of targeting these miRNAs to preserve cell phenotype, cellular homeostasis, and decrease inflammatory microenvironment and cellular senescence in IVD degeneration. Our study demonstrates the potential of EV‐enriched miRNAs to modulate degenerative pathways in human IVD cells. Our findings open avenues for developing EV‐enriched miRNA‐based, cell‐free therapies targeting IVD degeneration and related disorders. By focusing on miRNAs like miR‐100‐5p and let‐7b‐5p, our study not only elucidates the biological roles of these molecules in degenerative diseases but also supports further research into the combinatorial use of miRNAs for therapeutic applications from the EV perspective.

4.6. Shifting Landscapes: Loss of Homeostasis and Gain of Dysfunction

Our comparative analysis reveals a distinct shift in the miRNA landscape that mirrors the pathological progression of the tissue: (1) loss of protective function: the significant reduction of let‐7b‐5p in degenerate EVs supports a model where Non‐deg IVD cells constitutively secrete EVs enriched with “anti‐inflammatory brakes” to maintain microenvironmental homeostasis. The loss of this signal in degeneration acts as a permissive factor for chronic inflammation. (2) Gain of toxic function: conversely, the miRNAs uniquely identified in the Deg group were associated with KEGG pathways such as p53 signalling, Cell cycle arrest and Apoptosis. This suggests that EVs from degenerate cells may actively propagate stress signals to bystander cells, which is a phenomenon analogous to the senescence‐associated secretory phenotype (SASP). These “toxic” EVs could potentially prime neighbouring healthy cells for apoptosis or senescence, creating a feed‐forward loop of tissue degradation. Therefore, therapeutic strategies must likely involve both the replacement of lost protective factors (like let‐7b‐5p/miR‐100‐5p) and the inhibition of the unique degenerative cargo.

4.7. Limitations

Although our size‐exclusion chromatography‐based isolation protocol is highly effective at separating EV cargo (Fractions 7–9) from soluble proteins and non‐vesicular ribonucleoprotein complexes (Fractions 10+), we acknowledge the inherent challenge in achieving absolute purity in EV preparations. Although our previous characterisation of these samples confirmed the presence of classical EV markers (CD81, ANXA5 and FLOT1) and the absence of cellular contaminants (TOM20 and Golgi 58k), it is possible that a fraction of the detected miRNAs is associated with co‐isolated non‐vesicular particles or supermeres (Zhang et al. 2021). However, comparative studies have shown that size‐exclusion chromatography provides a higher ratio of vesicular to non‐vesicular RNA compared to precipitation methods (Yang et al. 2021). Therefore, we refer to these as EV‐enriched miRNAs to reflect this nuance.

A limitation of our functional validation is the absence of a non‐targeting “scramble” miRNA control. Although all data were normalised to a mock‐transfection control (reagent only) to account for delivery‐related effects, we cannot unequivocally rule out non‐specific responses to the introduction of exogenous RNA. However, the specificity of the observed biological effects, such as the selective upregulation of TIE2 by miR‐100‐5p but not by the equally abundant let‐7b‐5p, suggests a sequence‐specific mechanism rather than a generic innate immune response to cytosolic RNA.

5. Conclusion

The selective packaging of miRNAs in IVD cell‐derived EVs, particularly miR‐100‐5p and let‐7b‐5p, illustrates their relevance in modulating key cellular pathways. Our findings validated their functions in improving IVD cell phenotypic marker expression and homeostasis, counteracting the inflammatory microenvironment and potentially reducing cellular senescence. Our study highlights the critical role of EV‐enriched miRNAs in these aspects. EV‐enriched miRNAs hold promise for developing cell‐free regenerative therapies, particularly for treating IVD degeneration and potentially other musculoskeletal disorders. Further exploration of miRNA combinatorial therapies may provide novel insights into enhancing therapeutic efficacy and specificity.

Author Contributions

Conceptualisation: Li Li and Lisbet Haglund. Methodology: Li Li and Hadil Al‐Jallad. Validation: Li Li, Aiwei Sun, Saber Ghazizadeh, Hosni Cherif and Lisbet Haglund. Formal analysis: Li Li. Investigation: Li Li. Resources: Kirby Upshaw, Saleh Alfaisali, Jean Ouellet, Peter Jarzem and Lisbet Haglund. Data curation: Li Li and Aiwei Sun. Writing – original draft preparation: Li Li. Writing – review and editing: Li Li, Aiwei Sun, Saber Ghazizadeh, Hadil Al‐Jallad, Kirby Upshaw, Saleh Alfaisali, Jean Ouellet, Peter Jarzem, Hosni Cherif and Lisbet Haglund. Visualisation: Li Li. Supervision: Hosni Cherif and Lisbet Haglund. Project administration: Lisbet Haglund. Funding acquisition: Lisbet Haglund. All authors have read and agreed to the current version of the manuscript.

Funding

This work was supported by the Canadian Institutes of Health Research under grant #PJT‐178111; Le Réseau de Recherche en Santé Buccodentaire et Osseuse major infrastructure grant; ThéCell and Le Fonds de Recherche du Québec‐Santé doctoral training award (#272079) to L.L.

Ethics Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of McGill University (IRB #00010120).

Consent

Informed consent was obtained from all subjects involved in the study.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information: jex270127‐Sup‐0001‐SuppMat.docx

Data Availability Statement

All data generated or analysed during this study are included in the manuscript and Supporting Information files. The raw data and materials used to support the findings of this study are available from the corresponding author upon request.