Modulating the mesenchymal stromal cell microenvironment alters exosome RNA content and ligament healing capacity

Abstract

Although mesenchymal stromal cell (MSC) based therapies hold promise in regenerative medicine, their clinical application remains challenging due to issues such as immunocompatibility. MSC-derived exosomes are a promising off-the-shelf therapy for promoting wound healing in a cell-free manner. However, the potential to customize the content of MSC-exosomes, and understanding how such modifications influence exosome effects on tissue regeneration remain underexplored. In this study, we used an in vitro system to compare the priming of human MSCs by 2 inflammatory inducers TNF-α and CRX-527 (a highly potent synthetic TLR4 agonist that can be used as a vaccine adjuvant or to induce anti-tumor immunity) on exosome molecular cargo, as well as on an in vivo rat ligament injury model to validate exosome potency. Different microenvironmental stimuli used to prime MSCs in vitro affected their exosomal microRNAs and mRNAs, influencing ligament healing. Exosomes derived from untreated MSCs significantly enhance the mechanical properties of healing ligaments, in contrast to those obtained from MSCs primed with inflammation-inducers, which not only fail to provide any improvement but also potentially deteriorate the mechanical properties. Additionally, a link was identified between altered exosomal microRNA levels and expression changes in microRNA targets in ligaments. These findings elucidate the nuanced interplay between MSCs, their exosomes, and tissue regeneration.

Graphical Abstract

Graphical Abstract.

Significance statement.

Our findings reveal that changes in the MSC environment can significantly alter the therapeutic efficacy of exosomes in ligament recovery. To the best of our best knowledge, this is the first demonstration that alterations in the MSC microenvironment lead to significant changes in the exosome cargo RNA content. We also found a context-dependent correlation between exosomal and parent cell RNA content. The outcomes of our study provide insights into the adaptability and the versatile role of MSC-derived exosomes in tissue regeneration.

Introduction

Mesenchymal stromal/stem cells (MSCs) hold immense potential for clinical applications due to their unique ability to promote tissue repair, immunomodulation, and regeneration.¹ The therapeutic benefits of MSCs arise from their capacity to release growth factors and cytokines, which play a pivotal role in recruiting and activating other cells at the site of injury.² Moreover, MSCs facilitate the formation of new blood vessels, which is crucial for effectively healing damaged tissues.³ Despite the advancements in MSC-based therapies,^4,5 several challenges hinder their broad application in clinical settings. These challenges include but are not limited to immunocompatibility, the influence of target tissue microenvironment, and the migratory capacity.⁶ Immunocompatibility remains a major concern as individual variations among donors and recipients may lead to immune responses and potential rejection of transplanted MSCs. An intriguing alternative approach to exploit the therapeutic attributes of MSCs without using the cells themselves, to avoid problems with immunocompatibility and the other issues noted above, involves using MSC-derived exosomes.^7,8 Exosomes are nanosized extracellular vesicles secreted by cells that harbor diverse bioactive molecules, including cytokines and growth factors, signaling lipids, mRNAs, and miRNAs.⁹ MSC-derived exosomes can act as potent intercellular communicators, modulating cellular behaviors and facilitating tissue repair.¹⁰ However, it remains largely unknown whether altering the microenvironment that primes MSCs influences the contents of exosomes ex vivo, subsequently impacting the healing of tendons and ligaments in vivo.

In this study, we assessed the content of exosomes in vitro after exposure of human bone marrow-derived mesenchymal stem cells (BM-MSCs) to microenvironmental stimuli like a pro-inflammatory cytokine (tumor necrosis factor alpha/TNF-α) and a TLR4 mimetic (CRX-527) and evaluated potency in a rat ligament injury in vivo model. Our results demonstrated that exosomal RNA content (1) is partially determined by the parent MSC in a context-specific manner, (2) is customizable via perturbation of MSC microenvironment, and (3) differentially affects ligament healing. We found under native (unprimed MSCs) conditions, there is a moderate correlation between the RNAs within MSCs and the RNAs within their exosomes. The correlation increased significantly, especially for microRNAs, once MSCs underwent microenvironmental priming (eg, TNF-α or CRX-527¹¹). We showed that MSC-derived exosomes can significantly improve the mechanical properties of the healing ligament after injury. RNA content in exosomes is not only customizable via the type of microenvironmental stimuli of the MSCs but also impacts ligament healing efficiency and mechanical properties. These data suggest a connection between changes in exosomal microRNA abundance and subsequent alterations in the expression of microRNA targets upon in vivo exosome delivery into a ligament, implying a regulatory role of exosomal RNAs in tissue repair and regeneration.

Methods

Cell culture

All protocols were approved by the Health Sciences Institutional Review Board at the University of Wisconsin School of Medicine and Public Health (Protocol 2016-0298). BM-derived MSCs were obtained as remnants from filters after BM harvest of healthy donors.²⁵ BM cells were recovered by rinsing the filters with PBS. Mononuclear cells were separated using Ficol-Paque Plus 1.073. Red blood cells were lysed in ACK lysis buffer, and mononuclear cells were suspended in α-minimum essential medium supplemented with 10% fetal bovine serum (US origin, Hyclone), 1 × nonessential amino acids, and 4 mM l-glutamine. Cells were cultured in 75-cm² filter cap flasks. Attached cells (passage 0) were harvested via TrypLE cell dissociation enzyme (Invitrogen) and then re-plated into new flasks as described previously.^26,27 Cells (passages 4-6) were confirmed as MSCs using flow cytometry.²⁸

Isolation and characterization of exosomes from MSCs

MSCs were grown to confluence in 75 cm² filter cap cell culture flasks, washed with PBS, and the media was replaced with StemPro MSC serum-free media (SFM) CTS (Gibco Life Technologies). Cells were incubated 18-24 hours upon which the conditioned culture media (CM) was collected and exosomes were isolated by differential ultracentrifugation as previously described.²⁹ The CM was centrifuged using a Beckman Coulter Allegra X-15R centrifuge (2000 g at 4°C for 20 minutes) to remove any detached cells, apoptotic bodies, and cell debris. Clarified supernatant CM was then centrifuged in a Beckman Coulter Optima L-80XP Ultracentrifuge (100,000 g at 4°C for 2 hours) with a SW 28 rotor to pellet exosomes. The supernatant was carefully removed, and exosome pellets were re-suspended in PBS and pooled. Exosomes (previously identified as CD146⁺, CD29⁺, CD44⁺, CD63⁺, CD81⁺, and CD105⁺)⁴ were characterized using a Thermo NanoDrop spectrophotometer for protein and RNA concentration. Exosome particle diameter was evaluated via Nanoparticle Tracking Analysis using a Particle Metrix ZetaView (Zen-bio Inc., Research Triangle Park, NC). Results indicated that regardless of preconditioning, exosomes were similar in diameter (mean diameter range: 116.3-120.4 nm (Supplementary Figure S4)) and concentration.

Medial collateral ligament healing model

This study was approved by the University of Wisconsin Institutional Animal Care and Use Committee (Protocol M005785). A total of 40 adult male Wistar rats (300-370 g) were used as an animal model to study the effects of exosomes on ligament healing after bilateral surgical transection.^30,31 A surgically transected rather than torn ligament was chosen in order to create a uniform defect to compare healing. Two experiments were included in this study. For both studies, rats were anesthetized via isoflurane and a 1-cm incision was made over the medial aspect of each stifle. The medial collateral ligament (MCL) was exposed and completely transected at its mid-point. While the ligament was not repaired, a surgical pouch using the muscular layer was constructed over the ligament injury. The first study examined the mechanical response of the healing ligament to exosomes. Treatments (30 μL) were administered bilaterally and locoregionally within the pouch directly over the transected MCL and the pouch was closed by suturing. Treatments to the injured ligaments included: (1) PBS (serving as the injured control), (2) 5 × 10⁹ exosomes (exosome), (3) 5 × 10⁹ exosomes from TNF-α-primed MSCs (TNF), or (4) 5 × 10⁹ exosomes from CRX-527-primed MSCs (CRX). Exosome dose was chosen based on prior results showing that delivery of 1 × 10⁹ exosomes reduced the M1/M2 macrophage ratio but did not improve tendon functionality.⁴ A dose of 5 × 10⁹ exosomes was the maximum dose and volume that could be added to the torn MCL without excessive leakage. The skin was sutured after treatment administration. A second dose of 5 × 10⁹ exosomes was administered to the contralateral MCL, via needle and syringe, 3 days post-injury (when the presence of macrophages is elevated), to examine if an additional dose of exosomes could provide an additive effect on healing. Fourteen days post-injury, MCLs were collected for mechanical testing. As a time, dependent effect within the mechanical results was not observed, the 2 sides were combined for all data analyses.

The second study examined the molecular and biological response of the exosomes on MCL healing. While the same rat MCL injury model and treatments were used (n = 5 rats/treatment), a second dose of exosomes was not administered 3 days post-injury. An additional 4 rats did not undergo MCL transection and the MCLs served as intact controls. On day 14 post-injury (n = 4 rats/treatment) MCLs were dissected, subjectively ranked, imaged, and collected for histology (ipsilateral MCL) and RNA-seq (contralateral MCL). Before ligaments were dissected from the animal, they were assessed for inflammation and vascularity based on a 0-3 ranking system (0 = none, 1 = mild, 2 = moderate, 3 = severe). Ligaments used for histology were carefully dissected and immediately embedded in optimal cutting temperature (OCT) for flash freezing. Tissue was cryosectioned at a 4-μm thickness and used for H&E staining (Supplementary Figure S5). Ligaments used for RNA-seq were carefully dissected, snap-frozen, and stored at −80 °C until RNA isolation and RNA-seq.

Mechanical testing

To test the structural properties of the healing ligament after exosome treatment, day 14 MCLs were mechanically tested. MCLs were dissected, and the surrounding tissue was excised.³² Each MCL was removed with the femoral and tibial insertion sites intact. Ligaments remained hydrated using PBS. MCL length, width, and thickness were measured using digital calipers at pre-load. Width and thickness measurements were obtained at the injury site. The cross-sectional area (assumed to be an ellipse) was then estimated. The femur-MCL-tibia complex was mounted in a custom testing bath and mechanical testing machine. A pre-load of 0.1 N was applied, a cross-section was measured, and each MCL was preconditioned (cyclically loaded to approximately 2% strain for 10 cycles). The ligament was pulled to failure at a rate of 1% strain per second. Failure force, failure stress, ligament stiffness, and Young’s modulus were measured/computed to determine post-treatment MCL mechanical behavior. Failure force was the highest load prior to the failure of the ligament, and Lagrangian stress was calculated by dividing the failure force by the initial cross-sectional area of the ligament. Stiffness was calculated by determining the slope of the most linear portion of the load‐displacement curve. Young’s modulus was calculated by the slope of the linear portion of the stress-strain curve.

RNA sequencing of mRNAs

Total RNAs were isolated from MSCs and MSC-exosomes, as well as rat MCLs (day 14 post-injury) using Trizol (ThermoFisher #15596018) and chloroform phase separations followed by the RNeasy mini protocol (Qiagen #74106) with on-column DNase digestion (Qiagen #79254). One hundred nanograms of total RNA were used to prepare sequencing libraries using the LM-Seq (Ligation Mediated Sequencing) protocol.³³ RNAs were selected using the Next Poly A+ Isolation Kit (NEB #E7490S/L). Poly A+ fractions were eluted, primed, and fragmented for 7 minutes at 85 °C. First stand cDNA synthesis was performed using SmartScribe Reverse Transcriptase (Takara Bio USA #639538), and RNA was removed. cDNA fragments were purified with Ampure XP beads (Beckman Coulter #A63881). The 5ʹ adapter was ligated, and 18 cycles of amplification were performed. These final indexed cDNA libraries were quantified, normalized, multiplexed, and run as single-end reads on the HiSeq 3000 (Illumina, San Diego, CA). RNA-seq reads were mapped to the reference genome (human genome for MSCs and MSC-exosomes; rat genome for rat MCL injury in vivo model) and corresponding annotated protein-coding genes using Bowtie (v0.12.8),³⁴ allowing up to 2-mismatches. The gene-expected read counts and Transcripts Per Million (TPM) were estimated by RSEM (v1.2.3).³⁵ The TPMs were further normalized by EBSeq³⁶ R package to correct the potential batch effect.

Small RNA-seq of microRNAs

Total RNAs were isolated as above. One hundred nanograms of total RNA was used to prepare sequencing libraries using the Multiplex Small RNA Library Prep Set for Illumina (NEB #E7560S). The 3ʹ adapter was ligated and the reverse transcription primer hybridized. The 5ʹ adapter was ligated and cDNA synthesis was performed using ProtoScript II Reverse Transcriptase. PCR amplification was performed with 15 cycles and purified with Qiaquick PCR Purification Kit (Qiagen # 28106). cDNA libraries were then size-selected using AMpure XP beads (Beckman Coulter #A63881). These final indexed cDNA libraries were quantified, normalized, multiplexed, and run as single-end reads on the HiSeq 3000 (Illumina, San Diego, CA). Small RNA sequencing reads were timed to the first 15 bp, and then mapped to the human mature microRNAs (miRbase³⁷) using Bowtie (v0.12.8).³⁴ The microRNA expected read counts and Transcripts Per Million (TPM) were estimated by RSEM.³⁵

Analysis of mRNAs and microRNAs

The EBSeq package³⁶ was used to assess the probability of gene expression (mRNAs) or microRNAs being differentially expressed between any 2 given conditions. We required that DEGs should have FDR < 5% via EBSeq and >2-fold-change of median-by-ratio normalized read counts.

The miRNA targets were predicted via the TargetScan database.²⁴ GO enrichment analysis for microRNA targets was performed by the R package (“allez”).³⁸ The P-values were further adjusted by Benjamini–Hochberg (BH) multiple-test correction.³⁹  P-values of <.05 were considered statistically significant.

The Reactome pathway enrichment analysis for MSCs-exosome enriched mRNAs was performed by the “ReactomePA” R package.⁴⁰

To investigate the relationship between exosomal microRNA abundance and their downstream mRNA targets in rat ligaments, we employed a 2 × 2 contingency table to calculate the statistical significance of this microRNA target enrichment analysis. The contingency table was designed based on the widely accepted notion that an upregulation in microRNA usually corresponds to a downregulation of its mRNA target and vice versa. we examined whether the targets of upregulated or downregulated exosomal microRNAs (CRX-527 or TNF-α primed MSCs vs native MSCs) were more likely to be downregulated or upregulated, respectively, after in vivo exosome delivery.

Specifically, we calculated 4 numbers (namely a, b, c, and d) in the table (taking upregulated microRNAs as an example):

a: the number of instances where the downregulated genes (mRNAs) in rat ligaments after in vivo exosome delivery are also the targets of upregulated microRNA;

b: the number of instances where the downregulated genes (mRNAs) in rat ligaments after in vivo exosome delivery are not the targets of upregulated microRNA;

c: the number of instances where the background genes (|fold-change| < 1 and FDR = 1) are the targets of upregulated microRNA;

d: the number of instances where the background genes (|fold-change| < 1 and FDR = 1) are not the targets of upregulated microRNA;

We calculated the P-values of Fisher’s exact test to interrogate whether the ratio of (a/b) is statistically greater than the ratio of (c/d), asking whether the downregulated mRNAs in rat ligaments are significantly enriched in the targets of upregulated exosomal microRNA.

A similar approach is also applied to calculate the statistical enrichment of whether upregulated mRNAs in rat ligaments are significantly enriched in the targets of downregulated exosomal microRNA.

Nanoscale liquid chromatography coupled to tandem mass spectrometry

Nano LC–MS/MS was used to profile the proteome of MSCs and their exosomes. Digests were desalted using 100 µL Pierce C18 Tips (Thermo Fisher Scientific) according to manufacturer protocol. Eluates in 70%:30%:0.1% Acetonitrile:Water:TFA acid (vol:vol) were dried to completion using speed vac and finally reconstituted for LC–MS/MS profiling. Peptides were analyzed on a newer generation Orbitrap Fusion Lumos Tribrid platform. Chromatography of peptides prior to mass spectral analysis was accomplished using capillary emitter column (PepMap- C18, 2 µM, 100 Å, 500 × 0.075 mm, Thermo Fisher Scientific). NanoHPLC system delivered solvents A: 0.1% (v/v) formic acid, and B: 80% (v/v) acetonitrile, 0.1% (v/v) formic acid at 0.30 µL/minute to load the peptides at 2% (v/v) B, followed by quick 2 minutes gradient to 5% (v/v) B and gradual analytical gradient from 5% (v/v) B to 37.5% (v/v) B over 73 minutes when it concluded with rapid ramp to 95% (v/v) B for a 5-minute flash-out. As peptides eluted from the HPLC-column/electrospray source survey MS scans were acquired in the Orbitrap with a resolution of 120 000 followed by HCD-type MS2 fragmentation into Ion Trap (32% collision energy) under ddMSnScan 1 second cycle time mode with peptides detected in the MS1 scan from 350 to 1600 m/z; redundancy was limited by dynamic exclusion and MIPS filter mode ON.

Lumos acquired MS/MS data files were searched using Proteome Discoverer (ver. 2.5.0.400) Sequest HT search engine against Uniprot Homo sapiens database (UP000005640, 08/11/2020 download, 96 816 total sequences). Static cysteine carbamidomethylation, and variable methionine oxidation plus asparagine and glutamine deamidation, 2 tryptic miss-cleavages, and peptide mass tolerances set at 10 ppm with fragment mass at 0.6 Da were selected. Peptide and protein identifications were accepted under strict 1% FDR cutoffs with high confidence XCorr thresholds of 1.9 for z = 2 and 2.3 for z = 3. Strict principles of parsimony were applied for protein grouping. Chromatograms were aligned for feature mapping, and ion intensities were used for precursor ion quantification using unique and razor peptides. Normalization was performed on the total peptide amount and scaling on all averages.

Results

Independent patterns of exosomal mRNA and protein abundance

Although MSC-exosome cargoes are known to be enriched in proteins and small RNAs,¹² the extent to which MSC-exosomes are enriched in polyadenylated mRNAs and their potential functional relevance remains largely unknown. We performed RNA-seq of exosomes derived from native BM-MSCs and identified 4238 genes that exhibited significant expression in MSC-exosomes, with average normalized TPM values among replicates exceeding 1 (Supplementary Table S1). Polyadenylated mRNAs were found to be enriched in diverse biological functions (Supplementary Table S2), but interestingly, the top 8 enriched pathways pertained to protein translation machinery (Figure 1). We then conducted proteomics profiling of MSC-derived exosomes by nanoscale liquid chromatography coupled to tandem mass spectrometry (nano LC-MS/MS) to determine if there were correlations between the type and abundance of exosomal RNAs and resultant exosomal proteins. Surprisingly, our data indicated no correlation between the abundance of exosomal mRNA and that of exosomal proteins (Supplementary Figure S1A; Spearman’s rank correlation coefficient Rho = 0.018). We then primed MSCs with CRX-527, a synthetic TLR4 agonist, to determine if the RNA or proteomic profile would become more closely correlated in abundance. However consistent with native (unprimed) MSCs, we did not observe any correlation between exosomal mRNA abundance and protein abundance (Supplementary Figure S1B; Rho = 0.055). Thus, our results indicate that exosomal mRNA abundance may be independent of exosomal protein abundance, regardless of microenvironmental stimuli of MSCs by TLR4.

Figure 1.

Top 8 enriched pathways based on Benjamini-Hochberg (BH) adjusted p-values for MSC-exosome encapsulated mRNAs (average TPM > 1 for replicates).

Microenvironmental priming of MSCs leads to the generation of unique exosomal mRNA content

When examining the type of exosomal mRNAs between unprimed MSCs and CRX-527-primed MSCs, 1915 protein-coding genes exhibited differential expression upon CRX-527 treatment (FDR < 5% and >2 fold-change of normalized read counts). There was significant enrichment of the “inflammatory response” gene ontology (GO) term (GO:0002544) among the CRX-527-induced upregulated exosomal mRNAs (Figure 2; adjusted P-value = 3.8E−4). In addition to the inflammatory response GO term, the upregulated exosomal mRNAs enriched various biological functions like positive regulation of fibroblast migration (GO:0010763; adjusted P-value = 6.2E−12), regulation of intrinsic apoptotic signaling pathways (GO:2001242; adjusted P-value = 1.4E−7), and plasminogen activation (GO:0031639; adjusted P-value = 8.9E−7). These results suggest that CRX-527 priming of MSCs alters exosomal mRNAs content and suggests a role of ex vivo microenvironmental stimuli in modulating inflammation and wound healing.

Figure 2.

The illustration of the enrichment of upregulated mRNAs in the inflammatory response gene ontology (GO) term. The heatmap displays the corresponding mRNA expression levels.

To determine whether changes in exosomal mRNA content are specific to particular microenvironmental stimuli or represent a general response to MSC priming, we compared the effects of 2 different priming methods of MSCs: TNF-α and CRX-527 on the extent of overlap in exosomal mRNA content changes. We observed a moderate overlap in the differentially expressed exosomal mRNAs between the 2 MSC priming methods (Figure 3). For instance, out of 180 exosomal mRNAs upregulated in response to CRX-527 priming, only 11 (6.1%) were upregulated in response to TNF-α priming. Regarding downregulated genes, 857 out of 1735 (49.4%) exosomal mRNAs affected by CRX-527 priming of MSCs overlapped with those exosomes affected by TNF-α priming of MSCs. Furthermore, the mRNA content changes associated with each MSC priming method were found to be enriched in distinct biological functions (Figure 3). For example, positive regulation of blood vessel endothelial cell migration and fibroblast migration terms were specifically enriched in CRX-527 primed MSC-exosomes upregulated mRNAs but not enriched in TNF-α primed MSC-exosomes upregulated mRNAs. This observation suggests that exosomal mRNA content is governed by the specific microenvironment of MSCs, with differences noted after TNF-α and TLR4 priming.

Figure 3.

Comparison of mRNA content changes in different microenvironmental perturbations of the parent cells. The Venn diagram shows the overlapping up- or downregulated mRNAs between CRX-527 and TNF-α treatments. The top 5 enriched Gene Ontology (GO) terms based on adjusted P-values for each perturbation are also shown.

Microenvironmental priming of MSCs leads to the modulation of exosomal microRNA content

Recent evidence suggests that microRNA loading of exosomes is a selective and regulated process, rather than a random occurrence.¹³ We sought to determine whether the type of MSC microenvironmental priming influences microRNA content of exosomes. To achieve this, we performed small RNA sequencing to profile short RNA fragments in exosomes derived from native MSCs, CRX-527 primed MSCs and TNF-α primed MSCs. We identified 31 differentially expressed exosomal microRNAs in response to CRX-527 treatment (Figure 4A) and 23 differentially expressed exosomal microRNAs in response to TNF-α treatment of MSCs (Figure 4B), respectively. Notably, none of the differentially expressed exosomal microRNAs overlapped between the CRX-527 and TNF-α treatments, suggesting that exosomal microRNA content is influenced by the specific microenvironmental stimuli used to prime MSCs. “Macrophage differentiation” was the top enriched gene ontology (GO) terms based on the enrichment p-value for the microRNA targets of TNF-α primed MSC-exosomes (Figure 4D). TNF-α is a pro-inflammatory cytokine known to induce M1-like polarization of macrophages. Enriching the “macrophage differentiation” GO term suggests that as molecular cargo, exosomes may transport specific microRNAs to initiate monocyte differentiation into macrophages under pro-inflammatory stimuli such as TNF-α perturbation. Figure 4C and D present the top 3 enriched GO terms based on the most significant P-values for CRX-527 and TNF-α priming, respectively.

Figure 4.

Exosomal microRNA levels are influenced by microenvironmental perturbations of the parent cells. (A) Differentially expressed exosomal microRNAs resulting from CRX-527 treatment in the parent cells (B) Differentially expressed exosomal microRNAs resulting from TNF-α treatment in the parent cells. The top 3 enriched gene ontology (GO) terms based on the best enrichment P-value for the microRNA targets are shown in C and D, respectively.

Correlations in RNA abundance between parent cells and exosomes

We further conducted mRNA and small RNA sequencing on MSCs to investigate whether RNA abundance is correlated between parental MSCs and their exosomes. We observed moderate correlations between exosomes and their parent cells under native conditions, with Spearman’s Rank correlation coefficients of 0.46 for microRNAs and 0.57 for mRNAs, respectively (Figure 5). This suggests that a subset, but not all, exosomal RNAs indeed recapitulate the RNA abundance of their parental MSCs. These correlations increased when MSCs were exposed to microenvironmental stimuli, such as CRX-527 and TNF-α priming, notably for microRNAs (Figure 5). The Spearman’s rank correlation coefficient (Rho) between exosomal and parental MSC microRNA abundance increased from 0.46 (under native conditions) to 0.78 and 0.76 for CRX-527 and TNF-α primed MSCs, respectively. These data indicate that microenvironmental stimuli of MSCs synchronize microRNA profiles of MSCs with their secreted exosomes. A similar trend was observed for mRNAs, albeit less prominently (Figure 5). Thus, MSC-exosomal RNAs are not simply a random subset of RNAs derived from their parental cells, but are regulated and dependent on the microenvironmental stimuli driving exosome generation.

Figure 5.

Context-dependent correlations between exosomal and parental RNA abundance. The Spearman’s rank correlation coefficient (Rho) is indicated.

Locoregional treatment with exosomes affects the mechanical properties of injured ligaments

To determine if native exosomes from unprimed MSCs or exosomes from either CRX-527 or TNF-α primed MSCs show potency on the mechanical properties of acute ligament injury in vivo. We surgically transected medial collateral ligaments (MCL) in rats to mimic an acute MCL injury and then administered a locoregional MSC-exosome treatment within the first 3 days post-injury. The mechanical properties of the MCL were assessed on day 14 after treatment with unprimed MSC-exosomes vs TNF-α-primed MSC-exosomes and CRX-527-primed MSC-exosomes. Treatment with MSC-exosomes from native MSCs significantly increased maximum load (P = 0.038), and stiffness (P = 0.046) compared to untreated PBS controls (Figure 6), suggesting that MSC-exosomes treatment improves the mechanical properties of the MCL and can stimulate healing. No improvement was observed with CRX-527-primed MSCs or TNF-α-primed MSCs. In fact, these exosomes from pro-inflammation primed MSCs did not enhance, and may even have worsened the mechanical properties (Figure 6). These data suggest that microenvironmental stimuli of the MSCs through TNF-α or CRX-527 alters the potency of MSC-exosome treatments.

Figure 6.

Mechanical properties of the healing ligament after treatment with native and pre-conditioned MSC-derived exosomes. Injured rat ligaments were treated with exosomes at early post-injury days (≤3 days). The mechanical properties from 4 types of exosomes (PBS-control, MSC-derived native exosome, TNF-α, and CRX-527 preconditioned MSCs-derived exosomes) treatment were compared. (A) Max load; (B) max stress; (C) stiffness; (D) modulus. The P-values (2-sided) of Wilcoxon Rank-Sum test for native exosome vs PBS-control and native exosome vs pre-conditioned exosomes are indicated. The solid lines indicate P < .05, while the dashed lines indicate P ≥ .05.

To investigate the molecular mechanisms underlying the altered ligament mechanical properties influenced by MSC-exosomes, we conducted RNA-seq analysis on rat MCLs treated locoregionally with native MSC-exosomes, TNF-α-primed MSC-exosomes or CRX-527-primed MSC-exosomes as compared to untreated PBS controls (representing default healing) on day 14 post-injury. We identified 163 differentially expressed genes (DEGs) (fold change > 2 and FDR < 5%) in MCLs between those treated with native MSC-exosomes and PBS controls. The upregulated genes were significantly enriched in multiple GO terms associated with tissue regeneration, such as muscle contraction (adjusted P-value = 7.65E−6), positive regulation of epithelial morphogenesis (adjusted P-value = 0.001), and regulation of axonogenesis (adjusted P-value = 0.02). The top 15 most enriched GO terms for upregulated genes based on adjusted P-values (Supplementary Figure S2A) and an example of the specific upregulated genes enriched in muscle contraction (Supplementary Figure S2B) are shown. Additionally, a wide range of GO terms were enriched in downregulated genes in ligaments due to MSC-exosome treatment, including bone development (adjusted P-value = 0.006) and regulation of angiogenesis (adjusted P-value = 0.007). Detailed information on the enriched GO terms for all upregulated and downregulated genes in MSC-exosome vs PBS control are available (Supplementary Table S3).

Furthermore, we compared the MCL transcriptome after locoregional treatment with native MSC-exosomes and TNF-α-primed MSC-exosomes or CRX-527-primed MSC-exosomes. We identified 163 and 286 differentially expressed genes for CRX-527 and TNF-α-primed MSC-exosomes compared to native MSC-exosomes, respectively. Among the top enriched terms for upregulated genes in MCLs treated with TNF-α primed MSC-exosomes compared to native MSC-exosomes were GO terms related to the regulation of inflammation, including regulation of T-cell chemotaxis (Supplementary Table S4). In contrast, the upregulated genes in MCLs treated with CRX-527 primed MSC-exosomes compared to native MSC-exosome treatment were enriched in terms such as peptide hormone processing and nucleoside biosynthetic process (Supplementary Table S5), which are pathways less related to inflammation.

Locoregional treatment with exosomes reduces inflammation of injured ligaments

Using a rat medial collateral ligament (MCL) healing model, we administered MSC-exosomes in a locoregional manner. After 14 days post-treatment, The MCLs were also macroscopically evaluated for inflammation and vascularity based on a 0–3 scoring system (0 = normal, 1 = mild, 2 = moderate, 3 = severe).¹⁴ While the vasculature of the day 14 post-injured MCL was not significantly influenced by treatments, MCLs treated with native exosomes indicated less inflammation compared to the PBS control (Supplementary Figure S3; P-value = 0.0173; one-sided Wilcoxon rank-sum test).

A connection between the alteration in exosomal microRNA abundance and the subsequent changes in the expression of microRNA targets induced by in vivo exosome delivery in ligaments.

Since the abundance of exosomal microRNAs can be influenced by changes in the microenvironment stimuli used to prime MSCs (Figure 4), we conducted a microRNA-target enrichment analysis to investigate whether changes in MSC-exosomal microRNA abundance directly impact the expression of downstream mRNA targets in rat MCLs. Because microRNAs typically regulate the expression of their targets by downregulating them, we examined whether the targets of up- or downregulated exosomal microRNAs were more likely to be down- or upregulated, respectively, after in vivo exosome locoregional treatment of acute MCL injury. For each differentially expressed exosomal microRNA (eg, exosomes from CRX-527 or TNF-α primed MSCs vs native MSCs), we compared the ratio of microRNA targets among genes that were differentially expressed (>2-fold changes and FDR < 5%) vs genes in the background (control group; < 1-fold change & FDR = 1). Among the differentially expressed exosomal microRNAs (from CRX-527 or TNF-α primed MSCs vs native MSCs), 9 microRNAs exhibited a statistically significant enrichment of microRNA targets (P-value < 0.05, Fisher’s Exact Test) among genes that were differentially expressed in response to the corresponding exosome treatment (Figure 7). For instance, microRNA (miR-20a-5p) was downregulated in exosomes derived from CRX-527 primed MSCs compared to native MSC-exosomes. MiR-20a-5p promotes the repair of cartilage in osteoarthritis.¹⁵ After treating rat MCLs locoregionally, the upregulated mRNAs noted in MCLs treated with CRX-527 exosomes were found to be statistically significantly enriched in miR-20a-5p targets as compared to MCLs treated with native MSC exosomes. Although the specific contribution of miR-20a-5p to ligament healing remains largely unknown, a decrease in miR-20a-5p levels within CRX-527-exosomes, together with an increase in its target mRNAs in rat ligaments, indicates a potentially negative effect of CRX-527 primed MSC-exosome treatments for ligament healing. It also reveals a mechanism whereby modulating the microenvironment of parent cells can influence the microRNA content of exosome cargo and subsequently impact the transcriptomic response to in vivo exosome treatment.

Figure 7.

A connection between differentially expressed exosomal microRNAs and corresponding enriched microRNA-targets in response to exosome treatment. For each figure panel (dashed box), the left figure is the exosomal microRNA which shows a differential pattern (upregulation or downregulation;>2-fold changes and FDR<5%) between pre-conditioned (eg, CRX-527 or TNF-α treatment) and native (untreated) MSCs. The right figure in each dashed box is the percentage of microRNA-targets in differentially expressed mRNAs (downregulated or upregulated) (> 2-fold changes and FDR < 5%) and the background mRNAs (control; < 1-fold change and FDR = 1). The statistical significance of each microRNA-target enrichment analysis (right figure in each dashed box) is calculated by Fisher’s exact test (2 × 2 contingency table: the number of DEGs (mRNAs; up or down) which are overlapped with corresponding microRNA targets in response to exosome (with and without pre-conditioned MSCs) treatment vs the number of background mRNAs (equivalent expressions) which are also overlapped with corresponding microRNA targets.

Discussion

Mesenchymal stromal/stem cells (MSCs) are promising for tissue repair and regeneration,¹ but face challenges like immunocompatibility.¹⁶ An alternative is using MSC-derived exosomes, which carry beneficial molecules for tissue healing,^8,17 bypassing direct cell transplant issues. These exosomes can influence cellular behaviors that are crucial for repair. While we understand that MSCs from different sources produce exosomes with varying characteristics,¹⁸ it is still uncertain how priming MSCs might alter exosome content and its effects on ligament healing. Our study reveals the intricate interplay between MSCs, their exosomes, and tissue regeneration, with a particular focus on ligament injury repair. To our best knowledge, this is the first demonstration that alterations in the MSC microenvironment using inflammatory mediators lead to significant changes in the exosome cargo RNA content. Notably, these alternations in molecular cargo were observable when exosomes were administered in vivo to a rat ligament injury model, underscoring that the exosome cargo is potentially customizable.

Another major discovery in this study is the context-dependent correlation between RNA content in parent cells and exosomes. We noticed a significantly enhanced correlation when the microenvironment of the parent cells was changed, particularly for microRNAs. This finding indicates the potential of exosomes to convey environmental changes or stimuli information from their parent cells to other cells.

In CRX-527 primed mesenchymal stem cell (MSC) exosomes, both miR-21-3p and miR-29c-5p levels are increased. In the context of skin wound healing, miR-21 is linked to the enhancement of fibroblast differentiation, angiogenesis improvement, anti-inflammatory effects, increased collagen production, and wound re-epithelialization.¹⁹ The miR-29 family, including miR-29c, plays a role in regulating extracellular matrix production during tissue repair, and its abnormal levels have been tied to fibrosis.²⁰ Yet, the impact of miR-29c on fibrosis varies greatly depending on the tissue involved. For instance, reduced levels of miR-29c can lead to fibrosis in the lungs and kidneys but not in the liver, heart, or skin.²⁰ This highlights the complex nature of miR-29c’s role in wound healing. It has been known that miR-20a promotes wound healing by regulating the inflammatory response.²¹ However, miR-20a-5p exhibits decreased expression in CRX-527 primed MSC-exosomes, suggesting possible negative impacts on wound healing. Additionally, the let-7 miRNA family is crucial for regulating inflammatory responses.²² Notably, let-7d-3p and let-7a-5p are found to be elevated in CRX-527 and TNF-α primed MSC-exosomes, respectively. A study has identified let-7d-3p as a new biomarker for early sepsis detection,²³ underscoring let-7d-3p’s role in regulating inflammation. This evidence suggests that the microRNAs perturbated by CRX-527 and TNF-α priming have a dual role in wound healing, contributing to both beneficial and detrimental effects (such as promoting inflammation), complicating their contribution to the wound healing process.

The outcomes of our study provide insights into the adaptability and the versatile role of MSC-derived exosomes in tissue regeneration. Specifically, our findings on the customizable nature of exosome cargo and the dynamic RNA content correlation between parent cells and exosomes lay the foundation for developing tailored therapeutic strategies to enhance healing efficiency and mechanical properties in tissue repair.

Moreover, the established link between altered exosomal microRNA levels and changes in microRNA target expression in ligaments unveils novel avenues for manipulating cellular communication for regenerative purposes. This link offers a more nuanced understanding of the interaction between MSCs, their exosomes, and the regeneration of ligament tissues.

The relationship between microRNAs and their mRNA targets is well-established, illustrating that an increase in microRNA levels often leads to a subsequent decrease in the levels of their corresponding mRNA targets. By using established pairs of microRNA and mRNA targets via TargetScan database,²⁴ we can conduct statistical analyses to assess whether modifications in the levels of microRNAs within exosomes result in notable changes in mRNA levels after exosome therapy (Figure 7). However, delineating the connection between exosomal mRNAs and mRNA within rat ligaments poses significant technical difficulties. This complexity arises because the correlations between mRNAs are influenced by translation and post-translational modifications. For instance, introducing a specific mRNA X into a rat ligament does not necessarily mean that mRNA X’s expression will be upregulated in the rat. More commonly, the introduction of a specific mRNA into a biological system can trigger a negative feedback mechanism, such as the system detecting an excess of a particular mRNA and subsequently reducing its expression. The relationships between different mRNAs are generally unclear unless the exosomal mRNAs function as transcription factors (TFs). Therefore, it is challenging to quantitatively assess the impact of exosomal mRNAs on changes in rat ligament mRNA levels. In Figure 2, we show that priming MSCs with CRX-527 leads to an increase in exosomal mRNAs associated with the inflammatory response, as evidenced by 15 specific mRNAs, in comparison to exosomal mRNAs from unprimed MSCs. To further estimate the relative contribution of exosomal mRNAs to ligament mRNA changes, we analyzed the expression levels of these 15 inflammatory response genes in rat ligaments following exosome therapy (comparing exosomes primed with CRX-527 to those from unprimed, native MSCs). Notably, there was no differential expression of these 15 inflammatory genes in the rat ligaments (Supplementary Figure S6), which indicates the complex effects of exosomal mRNAs on changes in ligament mRNAs. Thus, we infer that the influence of exosomal mRNAs, whether from primed or unprimed MSCs, on ligament healing does not necessarily mirror identical mRNA alterations within the ligament. Instead, these exosomal mRNAs changes may provoke a multifaceted response in the ligament.

Although our study focused on early post-injury time points, revealing increased ligament max load and stiffness, we recognize the importance of treatment timing and duration on ligament healing. Future research will delve into optimizing the exosome-based treatment strategy, exploring different time points and durations to ascertain the most efficacious approach. The use of TNF-α and CRX-527 as pro-inflammatory mediators to pre-condition MSCs raises intriguing questions about the influence of alternative pre-conditioning agents. Future studies could explore the use of anti-inflammatory mediators or growth factors to understand their impact on exosome RNA content and the ensuing effects on tissue regeneration.

Conclusion

This study elucidates the intricate relationship among MSCs, MSC-exosomes, and their impact on tissue regeneration. The newfound knowledge on the customizable exosome cargo and the context-dependent RNA content correlation opens up new horizons for developing tailored MSC-derived exosome therapies. By building on these findings, we can further refine and enhance the application of MSC-derived exosomes in regenerative medicine, bringing us a step closer to unlocking their full therapeutic potential.

Author contributions

Concept and experiment design: Connie S. Chamberlain, Peng Jiang. Exosome experiment: Connie S. Chamberlain, William L. Murphy, Ray Vanderby. Animal experiments: Connie S. Chamberlain, William L. Murphy, Ray Vanderby. Proteomics profiling: John A. Kink, Christian M. Capitini, and Peiman Hematti. RNA-seq (mRNA and microRNAs) profiling: Peng Jiang. Technical support: Erika Mueller, Yiyao Li, Stephanie Yopp, Christian M. Capitini

Data analysis: Archana Prabahar & Peng Jiang. Manuscript writing: Connie S. Chamberlain, Archana Prabahar and Peng Jiang.

Funding

This project was supported by the University of Wisconsin Carbone Cancer Center (UWCCC) NIH/NCI P30 CA014520, NIH/NHLBI R01 HL153721 (C.M.C. and P.H.) and DARPA AWD00001593 (PJ).

Conflict of Interest

J.A.K., P.H. and C.M.C. are inventors of patents unrelated to this publication (US Patents 10,166,254 and 11,499,730). C.M.C. reports honorarium from Bayer, Elephas, Nektar Therapeutics, Novartis, and WiCell Research Institute, who had no input in the study design, analysis, manuscript preparation, or decision to submit for publication. The other authors had no potential conflicts of interest to disclose.

Data availability

The raw RNA-seq data and the gene expression data (TPMs and Mapping counts) were submitted to GEO (accession number: GSE245884).