Abstract Background Identification of biomarkers to predict the risk of healing delays are of huge clinical interest since 10% of fracture patients progress to delayed or non-union. During endochondral ossification, which takes place in mechanically unstable regions, the bone regenerates through a cartilage intermediate. We previously identified miR-1246, miR-335-5p and miR-193a-5p as fracture-related biomarkers in patient serum, but they appear not to have a functional role in an in vitro model of direct ossification. However, their involvement in other processes related to fracture healing cannot be ruled out and the most common healing process in fracture repair is secondary healing by way of endochondral ossification. Therefore, this study aims to explore the role of miR-1246, miR-335-5p and miR-193a-5p during in vitro endochondral differentiation of human bone marrow-derived mesenchymal stromal cells (BMSCs). Methods The activity of miR-1246, miR-335-5p, and miR-193a-5p was transiently inhibited just before pellet formation and the start of chondrogenic differentiation in human BMSCs (n = 5 donors), serving as a model for early endochondral ossification. The effect of miRNA inhibition was assessed by histology (Safranin O/Fast Green), immunohistochemistry (type II and type X collagen), and gene expression analysis by bulk RNA sequencing and RT-qPCR. Results Inhibition of miR-1246 and miR-335-5p enhanced chondrogenic and hypertrophic differentiation in BMSCs from three out of five donors, while miR-193a-5p inhibition had minimal effect. Donors were categorized as “responders” or “non-responders” based on histological and gene expression profiles. RNA sequencing and RT-qPCR identified differentially expressed genes, including a 1.6 and 1.5-fold upregulation of GDF5 and CCN5 respectively (p < 0.05) and downregulation of SKIL (fold change: 1.3, p = 0.0563) after miR-335-5p inhibition, while the same genes were unaltered by miRNA inhibition in non-responders, suggesting donor-specific responses to miRNA inhibition during early chondrogenesis. Conclusions These results suggest that miR-335-5p and miR-1246 have a regulatory effect on endochondral ossification, and genes regulated by miR-335-5p are involved in TGF-β signalling. The function of these miRNAs in human bone formation and repair should be further investigated to validate their potential role as prognostic markers in fracture healing. Keywords: Biomarkers, Bone healing, Chondrogenesis, Mesenchymal stem cells, MiRNA, Non-coding RNA Graphical Abstract [30]graphic file with name 13287_2025_4589_Figa_HTML.jpg Introduction Fracture non-union and delayed healing affect up to 10% of patients, leading to functional and psychological disability. There are several known risk factors including biological, surgical, and mechanical [[31]1], but predicting the individual risk of developing a non-union is not yet possible. After hematoma formation, which is a rich source of signalling molecules [[32]2–[33]5], bone healing can take place through intramembranous or endochondral ossification [[34]6]. During intramembranous ossification, bone forms within a few days at the periosteal sites by direct differentiation of mesenchymal stromal cells (MSCs) into osteoblasts. On the other hand, during endochondral ossification, bone heals indirectly through formation of a cartilage intermediate. Numerous elements of these processes could be negatively influenced, causing disruptions in the healing process, and their early identification could allow a better prediction of the healing course. Until now, there are no biomarkers predicting the risk of a fracture healing complication [[35]1]. To enhance the management of non-unions or proactively mitigate the likelihood of fracture healing complications, it becomes imperative to identify early biomarkers capable of predicting abnormal healing. MicroRNAs (miRNAs) are small, noncoding RNAs involved in the regulation of gene expression pathways by degradation and translational repression of messenger RNA (mRNA). They influence virtually all cellular processes, including proliferation, differentiation, apoptosis, and cell migration [[36]7–[37]10]. miRNAs have already been discussed as predictive markers since they are indicative of cellular processes and can be assessed non-invasively in blood and other bodily fluids [[38]11, [39]12]. miRNAs have also been investigated as predictive markers for individual outcomes of bone diseases [[40]13]. They have been described to directly regulate osteoblast differentiation, including early regulation of the osteogenic factor Runx2 [[41]14, [42]15]. A recently published work of our group identified miR-193a-5p, miR-1246, and miR-335-5p as differentially expressed in the serum of fracture patients versus healthy controls [[43]16]. However, these miRNAs did not show a role during in vitro direct osteogenic differentiation of bone marrow derived MSCs (BMSCs), as their inhibition did not result in changes in mineral deposition or in osteogenic gene expression levels. Nevertheless, these findings do not exclude a possible role of those miRNAs in other key mechanisms related to fracture healing such as cell recruitment, vascularization, or endochondral ossification. In the latter, typically occurring in the growth plate and during long bone fracture healing, MSCs first differentiate into chondrocytes and build a cartilaginous extracellular matrix; the chondrocytes then become hypertrophic, the callus mineralises and is later remodelled into bone [[44]6, [45]17]. Although several in vitro models have been proposed to recapitulate endochondral differentiation [[46]18, [47]19], the standard MSC chondrogenic pellet culture [[48]20] is a viable approach. Indeed, this model is known not to support the maintenance of a stable cartilage phenotype, but instead it promotes hypertrophy and progression towards endochondral differentiation [[49]21, [50]22]. Supporting this, most MSC chondrogenic pellet mineralize when implanted subcutaneously in mice, unlike chondrocyte pellets [[51]23]. Therefore, in this in vitro study, we evaluated the effects on chondrogenic and hypertrophic differentiation of miR-193a-5p, miR-1246, and miR-335-5p inhibition during pellet formation. This model gave us the opportunity to model early stages of cartilage formation during fracture healing. Methods Cell isolation and expansion Bone marrow aspirates were obtained from patients undergoing orthopaedic surgeries at the Inselspital Bern (Req-2016-00141). The Swiss Human Research Act does not apply to research involving anonymized biological material and/or anonymously collected or anonymized health-related data [[52]24]. Informed written consent, which also covered anonymization of health-related data and biological material, was obtained from all donors. A total of 8 donors were used for subsequent experiments (Table [53]1): n = 3 donors to assess miR-193a-5p, miR-1246, and miR-335-5p expression at day 3 and 7 of chondrogenic differentiation, and n = 5 donors to functionally inhibit miRNA at the start of chondrogenic induction. An average of 9 ml of bone marrow aspirate was obtained for each donor from the thoracic or lumbar spine. Samples were collected in EDTA tubes, shipped to Davos overnight at room temperature, and immediately processed for cell isolation using Ficoll gradient separation. Table 1. Summary of BMSC donors Donor nr Experiment Age, sex #248 miRNA expression F, 48 #257 miRNA expression M, 54 #265 miRNA expression M, 72 #152 miRNA inhibition M, 24 #157 miRNA inhibition M, 33 #162 miRNA inhibition M, 71 #217 miRNA inhibition F, 73 #255 miRNA inhibition F, 74 [54]Open in a new tab Human bone marrow-derived mesenchymal stromal cells were isolated as previously described [[55]25, [56]26]. BMSCs were expanded in minimum Essential Medium alpha (αMEM, Gibco, Thermo Fisher, Zürich, Switzerland) supplemented with 10% foetal bovine serum (FBS; Corning, Corning, NY, USA), 100 U/ml penicillin, 100 µg/mL streptomycin (Gibco, Thermo Fisher) and 5 ng/ml recombinant human basic fibroblast growth factor (bFGF, Fitzgerald Industries International, Acton, MA, USA), with frozen stocks prepared at P1. Each donor is tested for chondrogenic and osteogenic potential using standard protocols to validate potency. Cells were then used at passage 3 for experiments. Pellet culture and chondrogenic induction Induction of chondrogenic differentiation of BMSCs in pellet culture was performed as follows. After cell detachment with a 0.05% trypsin-EDTA solution in PBS, 2 × 10^5 cells/well were seeded in V-bottom 96-well plates (Thermo Fisher) and centrifuged at 400 g for 5 min to facilitate pellet formation. Undifferentiated cells were maintained in chondropermissive medium (CP): Dulbecco’s modified Eagle medium with 4.5 g/l glucose (DMEM, Gibco, Thermo Fisher), 0.11 g/l sodium pyruvate (Sigma-Aldrich, Buchs, Switzerland), 50 µg/ml L-ascorbic acid 2-phosphate sesquimagnesium salt hydrate (Sigma-Aldrich), 100 nM dexamethasone (Sigma-Aldrich), Corning ITS + Premix (6.25 µg/ml human recombinant insulin, 6.25 µg/ml human natural transferrin, 6.25 µg/ml selenious acid, 1.25 mg/ml bovine serum albumin, 5.35 µg/ml linoleic acid), 1% non-essential amino acids (Gibco, Thermo Fisher), 100 µg/mL streptomycin, 100 U/mL penicillin (Gibco, Thermo Fisher). Cells were induced to differentiation using chondrogenic medium (CH): CP + 10 ng/ml TGF-β1 (Fitzgerald Industries International). After pellet formation, medium was refreshed three times per week. Medium samples and pellets were collected at different time points for RNA isolation followed by gene and miRNA expression analysis. Pellets were fixed and embedded for histology at day 28. MiRNA transfection To study the influence of miRNA inhibition during the early stages of chondrogenesis, miRNA antagonists were delivered by electroporation before pellet formation, using the Neon™ Transfection System (Invitrogen, Thermo Fisher). The experiment was performed with BMSCs from five different donors (Table [57]1). After detachment cells were resuspended in R buffer and divided in 7 groups: * CP: pellets cultured in chondropermissive medium (differentiation negative control). * CH: pellets cultured in chondrogenic medium (differentiation positive control). * EP: cells electroporated but not transfected, pellets cultured in chondrogenic medium (electroporation control). * NC: cells electroporated and transfected with miRNA inhibitor-negative control B, pellets cultured in chondrogenic medium (transfection control). * anti-193a-5p: cells electroporated and transfected with hsa-miR-193a-5p inhibitor, pellets cultured in chondrogenic medium. * anti-1246: cells electroporated and transfected with hsa-miR-1246 inhibitor, pellets cultured in chondrogenic medium. * anti-335-5p: cells electroporated and transfected with hsa-miR-335-5p inhibitor, pellets cultured in chondrogenic medium. The groups are further summarized in Table [58]2. Table 2. Summary of the experimental groups for the MiRNA Inhibition experiment Group Short name Medium Electroporation Chondropermissive control CP Chondropermissive No Chondrogenic control CH Chondrogenic No Electroporation control EP Chondrogenic Yes miRNA inhibitor-negative control B NC Chondrogenic Yes hsa-miR-193a-5p inhibitor anti-193a-5p Chondrogenic Yes hsa-miR-1246 inhibitor anti-1246 Chondrogenic Yes hsa-miR-335-5p inhibitor anti-335-5p Chondrogenic Yes [59]Open in a new tab miRNA inhibitors or the inhibitor control (miRCURY LNA miRNA inhibitors, Qiagen, Hilden Germany) were used at a concentration of 350 pmol/10^6 cells and added to the cells resuspended in R buffer (part of the Neon Transfection System, Invitrogen, Thermo Fisher). Electroporation was performed at 990 V and one 40 ms-pulse [[60]27]. After electroporation, cells were seeded for chondrogenic differentiation as described in the previous paragraph. Pellets were maintained in culture for 3, 7, 21, and 28 days and used for RNA sequencing, RT-qPCR, and histological evaluations. Donors were categorized as responders or non-responders to miRNA inhibition according to the histological evaluations and image quantification. RNA isolation Chondrogenic pellets were collected from the V-bottom plates using a microspoon and transferred to Eppendorf tubes each containing 1 ml TriReagent (Molecular Research Center Inc., Cincinnati, OH, USA) and one 5-mm stainless steel ball. Pellets were mechanically homogenized using a MM 400 Mixer Mill (Retsch GmbH, Haan, Germany), and insoluble material was removed by centrifugation at 12,000 g at 4 °C for 10 min. Total RNA was extracted with 1-bromo-3-chloropropane (Sigma-Aldrich). After phase separation, the precipitation of RNA was carried out from the aqueous phase by adding 2-propanol (Sigma-Aldrich). After centrifugation at 12,000 g at 4 °C for 8 min, the RNA pellet was washed with 75% EtOH and finally reconstituted with diethylpyrocarbonate (DEPC)-treated H[2]O. After incubation at 60 °C for 10 min, NanoDrop One (Thermo Fisher) was used to measure the total RNA concentration and purity was evaluated by A260/280 and A260/230 ratio. RNA samples were stored at -70 °C until further analysis for up to 6 months. RNA sequencing To understand the transcriptomic changes induced by miRNA inhibition, RNA sequencing (RNAseq) was performed on day 3 samples from the donors responding to miRNA inhibition. Samples from the CH, NC, and miRNA inhibition groups were selected for this analysis. cDNA library synthesis (Smart-Seq2, Illumina, San Diego, CA, USA) and transcriptome sequencing were performed by the Functional Genomics Center Zürich on a NovaSeq 6000 Sequencing System (Illumina, San Diego, CA, USA), using a single read 100 bp configuration. After sequencing, the raw reads underwent preprocessing to ensure data quality. This involved the removal of adapter sequences, trimming of low-quality ends, and filtering out reads with a Phred quality score below 20, achieved using Fastp (Version 0.23.4) [[61]28]. Following preprocessing, the high-quality reads were pseudo-aligned to the Human reference transcriptome (build GRCh38.p13) based on Gencode release 42. Quantification of gene-level expression was performed using Kallisto (Version 0.46.1) [[62]29]. Differential expression analysis was conducted utilizing the generalized linear model implemented in the Bioconductor package DESeq2 (R version: 4.3.2, DESeq2 version: 1.42) [[63]30]. Genes exhibiting altered expression levels with an adjusted p-value (Benjamini and Hochberg method) below 0.05 were considered differentially expressed. Reverse transcription and qPCR analysis SuperScript™ VILO™ cDNA Synthesis Kit (Thermo Fisher) was used for reverse transcription (RT) from 1000 ng of total RNA, following manufacturer’s protocol. Analysis of chondrogenic differentiation markers and validations of RNA sequencing results were performed by qPCR, using the TaqMan Gene Expression Master Mix (Applied Biosystems, Thermo Fisher) in a QuantStudio 7 Flex Real-Time PCR system (Applied Biosystems, Thermo Fisher). Assay details are reported in Table [64]3. TaqMan gene expression assays (Thermo Fisher) or custom-designed primers and probes (Microsynth AG, Balgach, Switzerland) were used. Results were expressed as 2^−ΔCt. Table 3. qPCR assay details Gene symbol Forward sequence (or assay ID) Reverse sequence Probe sequence^‡ ACAN 5’-AGT CCT CAA GCC TCC TGT ACT CA-3’ 5’-CGG GAA GTG GCG GTA ACA-3’ 5’-CCG GAA TGG AAA CGT GAA TCA GAA TCA ACT-3’ ALPL Hs00758162_m1 CNN5 Hs01031984_m1 COL10A1 5’-ACG CTG AAC GAT ACC AAA TG-3’ 5’-TGC TAT ACC TTT ACT CTT TAT GGT GTA-3’ 5’-ACT ACC CAA CAC CAA GAC ACA GTT CTT CAT TCC-3’ COL1A1 5’-CCC TGG AAA GAA TGG AGA TGA T-3’ 5’-ACT GAA ACC TCT GTG TCC CTT CA-3’ 5’-CGG GCA ATC CTC GAG CAC CCT − 3’ COL2A1 5’-GGC AAT AGC AGG TTC ACG TAC A-3’ 5’-GAT AAC AGT CTT GCC CCA CTT ACC-3’ 5’-CCT GAA GGA TGG CTG CAC GAA ACA TAC-3’ GDF5 Hs00167060_m1 IBSP Hs00173720_m1 ICAM1 Hs00164932_m1 MMP13 5’-CGG CCA CTC CTT AGG TCT TG-3’ 5’-TTT TGC CGG TGT AGG TGT AGA TAG-3’ 5’-CTC CAA GGA CCC TGG AGC ACT CAT GT-3’ PTX3 Hs00173615_m1 RPLP0 ^# 5’-TGG GCA AGA ACA CCA TGA TG-3’ 5’-CGG ATA TGA GGC AGC AGT TTC-3’ 5’-AGG GCA CCT GGA AAA CAA CCC AGC-3’ RUNX2 5’-AGC AAG GTT CAA CGA TCT GAG AT-3’ 5’-TTT GTG AAG ACG GTT ATG GTC AA-3’ 5’-TGA AAC TCT TGC CTC GTC CAC TCC G-3’ SKIL Hs01045418_m1 SLC2A3 Hs00359840_m1 SLC2A5 Hs01086390_m1 SOX9 Hs00165814_m1* SPP1 5’-CTC AGG CCA GTT GCA GCC-3’ 5’-CAA AAG CAA ATC ACT GCA ATT CTC-3’ 5’-AAA CGC CCA AGG AAA ACT CAC TAC C-3’ TGFB3 Hs00234245_m1 TIMP2 Hs00234278_m1 [65]Open in a new tab * TaqMan Gene expression assay (Thermo Fisher). 5’ modification: FAM. 3’ modification: NFQ-MGB ^‡ 5’ modification: FAM. 3’ modification: TAMRA. ^#Reference gene ^# Reference gene Histology on formalin fixed-paraffin embedded pellets At day 28, pellets were collected, washed with phosphate buffered saline (PBS, Sigma-Aldrich) and fixed for 2 h with 4% neutral buffered formalin (Formafix AG, Hittnau, Switzerland). After dehydration, pellets were embedded in paraffin and cut into 5 μm thin sections with the HM 335 S microtome (Microm, Thermo Fisher). Sections were stained with Safranin O and Fast Green to evidence the proteoglycan content within the extracellular matrix. Briefly, after deparaffinization and rehydration, sections were stained for 10 min with Weigert’s Haematoxylin (Roth AG, Arlesheim, Switzerland). The slides were blued in lukewarm tap water for 10 min, then briefly washed in distilled water and incubated for 6 min in 0.02% (v/v) Fast Green FCF (Sigma Aldrich) prepared in 0.01% (v/v) acetic acid (Sigma Aldrich). Fast Green staining was followed by 20 s in 1% (v/v) acetic acid and then 12 min in 0.1% (w/v) Safranin O solution (Chroma-Gesellschaft Schmid GmbH & Co, Münster, Germany). After dehydration, stained slides were mounted using Eukitt mounting medium (Sigma-Aldrich). Sections were imaged with a 10x objective (NA 0.40/∞ 0.17/OFN 26.5) using a BX63F microscope (Olympus, Hamburg, Germany). Immunohistochemistry In addition to Safranin O staining, immunohistochemistry was performed for type II and type X collagen. For type II collagen, the CIIC1 antibody was obtained by the Developmental Studies Hybridoma Bank (University of Iowa, Iowa City, IA, USA) and used at a concentration of 2 µg/ml. For type X collagen, a mouse monoclonal antibody (X53) was used at 5 µg/ml (#14-9771-82, Thermo Fisher). Briefly, sections were deparaffinized and endogenous peroxidase activity was blocked by incubation with 0.3% H[2]O[2] in methanol for 30 min. After washing in 0.1% Tween 20 in PBS (PBS-T), sections were blocked with 5% Normal Horse Serum (Vector Laboratories Inc., Newark, CA, USA). Antigen retrieval was done with 25 mg/ml hyaluronidase and 0.25 U/ml chondroitinase ABC (Sigma-Aldrich) for type II collagen, and with 2 mg/ml hyaluronidase only for type X collagen. Primary antibodies were incubated overnight at 4 °C in a moist chamber. Following washing with PBS-T, the sections were incubated with 2.5 µg/ml of a biotinylated anti-mouse IgG (H + L) secondary antibody (Vector Laboratories #BA-2001) for 30 min at room temperature. Control samples were incubated with the secondary antibody only to check for aspecific and background staining. Detection was performed using the Vectastain Elite ABC kit and ImmPACT DAB solution (Vector Laboratories) according to manufacturer’s instructions. Sections were finally counterstained with Mayer’s Haematoxylin (Fluka), dehydrated, and mounted with Eukitt mounting medium (Sigma-Aldrich). Sections were imaged with a 10x objective (NA 0.40/∞ 0.17/OFN 26.5) using a BX63F microscope (Olympus, Hamburg, Germany). Image analysis Quantification of Safranin O/Fast green, type II collagen, and type X collagen staining was performed using Fiji/ImageJ (NIH, Bethesda, MD, USA). For each stained section, a region of interest (ROI) was manually drawn around the perimeter of the pellet. To isolate specific staining components, the Colour Deconvolution plugin was applied, separating the image into individual stain channels based on their chromogenic profile. Colour thresholding was applied to the selected channels to identify positively stained areas, followed by quantification as a percentage of the total pellet area defined by the ROI. Statistical analysis Statistical analysis was performed using GraphPad Prism v.10 (GraphPad Software, San Diego, CA, USA). Ratio paired t-test or two-way ANOVA followed by Šídák’s multiple comparisons tests were performed to compare gene and miRNA expression levels. Dunnet’s multiple comparison test was used on image quantification to compare results from each group to the transfection negative control (NC). We considered results as statistically significant when p < 0.05. Results MiRNA expression during chondrogenic differentiation The expression of miR-193a-5p, miR-1246, and miR-335-5p in early days of chondrogenic differentiation was explored. At the same time point, the expression of SOX9, RUNX2, and their ratio was assessed to confirm successful chondrogenic commitment. No miRNA was significantly differentially expressed between undifferentiated and differentiated pellets, with a trend towards miR-335-5p downregulation in chondrogenic conditions (Fig. [66]1). Fig. 1. [67]Fig. 1 [68]Open in a new tab miRNA expression profile during first week of chondrogenic induction. (A) Expression of SOX9 and RUNX2 at day 3 and 7 of chondrogenic differentiation, and SOX9/RUNX2 ratio. (B) Expression of miR-193a-5p, miR-1246, and miR-335-5p at day 3 and day 7 of chondrogenic differentiation. Dots of different colors represent different donors, average 2-dCt values. * p < 0.05; ** p < 0.01; *** p < 0.001. CP: chondropermissive medium. CH: chondrogenic medium (chondropermissive + 10 ng/ml TGF-β1) Effect of MiRNA Inhibition on chondrogenic differentiation In three out of five donors (#255, #217, #162), the treatment of human BMSCs with inhibitors of miR-1246 and miR-335-5p resulted in an enhanced chondrogenic and hypertrophic differentiation compared to the NC, as indicated by Safranin-O (Fig. [69]2A), type II collagen (Fig. [70]2D-E), and type X collagen staining (Fig. [71]2F-G). Quantification of type II and type X collagen revealed statistically significant differences, whereas the differences in Safranin-O/Fast green staining are mostly qualitative. Inhibition of miR-193a-5p did not show an effect on chondrogenic differentiation, except for a slight reduction in matrix deposition for one donor (#255). Notably, chondrogenic/hypertrophic differentiation for the two donors that showed higher baseline levels of matrix deposition (#152, #157) was not influenced by miRNA inhibition. Fig. 2. [72]Fig. 2 [73]Open in a new tab Histological evaluation of BMSCs chondrogenic differentiation (day 28) in pellet culture after miRNA inhibition. (A) Representative images from one donor responding to miRNA inhibition, Safranin O/Fast Green staining. (B) Safranin O area quantification, all donors (left), responder donors (middle), non-responder donors (right graph). (C) Fast-Green area quantification, all donors (left), responders (middle), non responders (right graph). (D) Representative images from one donor responding to miRNA inhibition, type II collagen immunohistochemistry. (E) Type II collagen area quantification, all donors (left), responders (middle), non responders (right graph). (F) Representative images from one donor responding to miRNA inhibition, Type X collagen immunohistochemistry. (G) Type X collagen area quantification, all donors (left), responders (middle), non responders (right graph). Scale bar = 200 μm. On the left side of the image panels, the BMSC donor code is reported. CH: chondrogenic positive control. EP: electroporation control. NC: negative control (electroporation + transfection of a negative control miRNA antagonist in chondrogenic medium). *: p < 0.05. **: p < 0.01. ***: p > 0.001. **** p < 0.0001 For this reason, we categorized the donors in “responders” to miRNA inhibition (#255, #217, #162) and “non-responders” (#152, #157). Unexpectedly, the analysis of gene expression revealed no difference in chondrogenic and hypertrophic markers at day 28 after miRNA inhibition, even when separating the donors between responders and non-responders to miRNA antagonist treatments (Fig. [74]3). Between responders and non-responders, the differences observed in histological staining are confirmed by the trends in gene expression levels in non-responder donors, with baseline differences in the expression of COL2A1, ACAN, COL10A1, TIMP2, RUNX2, RUNX3 and ALPL (p < 0.05, Fig. [75]3). Fig. 3. [76]Fig. 3 [77]Open in a new tab Expression of chondrogenic and hypertrophic markers at 28 days after miRNA inhibition. Results are expressed as 2^−ΔCt, with normalization to the reference gene (RPLP0). Non-responder donors: #152 (circle symbol), #157 (square). Responder donors: #255 (triangle), #217 (rhombus), #162 (donut). Two-way ANOVA with Dunnet’s multiple comparison test revealed no difference after miRNA inhibition among non-responders or responder donors. * p < 0.05 To understand the mechanism behind the effect of miRNA inhibition observed for some of the donors, we performed RNA sequencing on samples collected 3 days after miRNA inhibition and the start of chondrogenic differentiation. RNA sequencing was performed on the three responder donors to identify differentially expressed (DE) genes downstream of miRNA inhibition. Subsequently, the expression of DE genes was validated in all the donors. Results showed DE genes and pathway enrichment analysis for all comparisons miRNA inhibition vs. NC (Table [78]4; Fig. [79]4). Table 4. Summary of differentially expressed genes. For each comparison, genes are selected and sorted according to the false discovery rate (FDR) with a threshold of 0.05 miR-193a-5p vs. NC gene_name description entrez_id log2 Ratio pValue FDR PTX3 pentraxin 3 5806 -2.046 0.000 0.000 ENSG00000288894 novel protein - -8.110 0.000 0.000 SOD2 superoxide dismutase 2 6648 -1.563 0.000 0.000 RCAN1 regulator of calcineurin 1 1827 -1.358 0.000 0.001 SPP1 secreted phosphoprotein 1 6696 1.366 0.000 0.001 LRRC17 leucine rich repeat containing 17 10,234 1.553 0.000 0.001 ICAM1 intercellular adhesion molecule 1 3383 -1.236 0.000 0.006 TLR2 toll like receptor 2 7097 -1.576 0.000 0.011 SERPINE1 serpin family E member 1 5054 -1.077 0.000 0.016 SLC2A3 solute carrier family 2 member 3 6515 -0.938 0.000 0.019 ERRFI1 ERBB receptor feedback inhibitor 1 54,206 -1.094 0.000 0.019 SLC2A5 solute carrier family 2 member 5 6518 -1.061 0.000 0.026 RHCG Rh family C glycoprotein 51,458 -2.725 0.000 0.038 miR-335-5p vs. NC gene_name description entrez_id log2 Ratio pValue fdr SKIL SKI like proto-oncogene 6498 -0.643 0.000 0.006 TMED5 transmembrane p24 trafficking protein 5 50,999 -0.658 0.000 0.043 GDF5 growth differentiation factor 5 8200 0.708 0.000 0.046 NAMPT nicotinamide phosphoribosyltransferase 10,135 -0.649 0.000 0.046 OCLN occludin 100,506,658 -0.565 0.000 0.046 S100A4 S100 calcium binding protein A4 6275 0.910 0.000 0.046 RCAN1 regulator of calcineurin 1 1827 -0.858 0.000 0.046 TWF1 twinfilin actin binding protein 1 5756 -0.541 0.000 0.046 CCL20 C-C motif chemokine ligand 20 6364 -2.308 0.000 0.046 IPMK inositol polyphosphate multikinase 253,430 -0.897 0.000 0.046 CCN5 cellular communication network factor 5 8839 0.821 0.000 0.046 ENSG00000279936 TEC - -0.655 0.000 0.046 miR-1246 vs. NC gene_name description entrez_id log2 Ratio pValue fdr TGFB3 transforming growth factor beta 3 7043 0.972 0.000 0.004 SPP1 secreted phosphoprotein 1 6696 0.924 0.000 0.004 IBSP integrin binding sialoprotein 3381 0.924 0.000 0.014 PTX3 pentraxin 3 5806 -0.937 0.000 0.018 [80]Open in a new tab Fig. 4. [81]Fig. 4 [82]Open in a new tab RNAseq analysis of samples at day 3 after miRNA inhibition and the induction of chondrogenic differentiation. n = 3 donors (responders, #255, #217, and #162). A-B) Volcano plot of differentially expressed genes (DEG) and Gene Ontology (GO) enrichment analysis in miR-193a-5p inhibition vs. negative control. C-D) Volcano plot of differentially expressed genes (DEG) and Gene Ontology (GO) enrichment analysis in miR-335-5p inhibition vs. negative control. E-F) Volcano plot of differentially expressed genes (DEG) and Gene Ontology (GO) enrichment analysis in miR-1246 inhibition vs. negative control. In the volcano plots, the threshold for the log2 fold change was set at ± 0.58, corresponding to a 1.5-fold difference, while the threshold for statistical significance was set at a -log10(p-value) of 1.3 (equivalent to p < 0.05). In the GO enrichment analysis plots, bar length indicates the number of genes associated with each pathway, and bar colour represents the p-value, ranging from red (lower p-value) to blue (higher p-value) We selected genes for further validation by RT-qPCR in samples from all donors: CNN5, GDF5, IBSP, ICAM1, PTX3, SKIL, SLC2A3, SLC2A5, SPP1, TGFB3. Among these genes, responder donors showed a 1.6 and 1.5-fold upregulation of GDF5 and CCN5 respectively (Fig. [83]5A and B, respectively, p < 0.05) and a 1.3-fold downregulation of SKIL (Fig. [84]5C, p = 0.0563) after miR-335-5p inhibition, while the same genes were unaltered by miRNA inhibition in non-responders. The trends in SCL2A3 and SLC2A5 expression, identified as downregulated by miR-193a-5p treatment by RNAseq, was also confirmed by qPCR analysis (data not shown). IBSP, PTX3, and TGFB3 gene expression levels, when assessed by RT-qPCR, were not statistically different after treatment with the miR-1246 inhibitor (data not shown). Fig. 5. Fig. 5 [85]Open in a new tab qPCR validation of differentially expressed genes. (A) GDF5. (B) CCN5. (C) SKIL. For each gene, cumulative results (n = 5 donors) are showed, together with donors clustered by response to miRNA inhibitor treatment: responders (#162, #217, #255) and non-responders (#152, #157). * p < 0.05 Discussion The identification of biomarkers to assess the risk of an individual fracture to develop into a non-union is necessary to allow early interventions or changes in surgical strategies. Blood-based biomarkers are ideal to allow a minimally invasive procedure and the blood taking procedure can be combined with other necessary blood tests. Several studies investigated the role of miRNAs as biomarkers for bone diseases. For example, miR-21, miR-23a and miR-24 have been found to be upregulated in the serum of patients who endured a bone fracture, and a similar miRNA profile was detected in osteoporotic bone tissue [[86]31]. Recently we conducted a systematic review with the aim to summarise and structure the findings from literature on miRNAs with a biomarker potential in the context of fracture healing [[87]32]. Some promising candidates were identified; however, we believe that a panel of miRNA involved in different biological processes has a greater potential to be translated to clinical practice. Our previous study focused on screening circulating miRNA in fracture patients versus healthy subjects and to determine those associated to bone formation, leading to the identification of miR-193a-5p, miR-1246, and miR-335-5p as promising candidates [[88]16]. However, functional validation experiments showed these miRNAs do not have a role during direct osteogenic differentiation of human BMSCs in vitro. These results do not exclude an influence of the three miRNAs on other key mechanisms such as chondrogenic and endochondral differentiation, which are main contributors of fracture healing in long bones. Therefore, the aim of this study was to evaluate the effect of miRNA inhibition during early stages of chondrogenic differentiation, i.e. mesenchymal condensation. miRNA antagonists were delivered to cells once just ahead of pellet formation and the start of chondrogenic induction. Histological analysis of pellets after 28 days of culture in chondrogenic conditions revealed an effect of miR-1246 and miR-335-5p inhibition on matrix formation, with an increased deposition of glycosaminoglycans and both type II and type X collagen in the extracellular matrix. However, this effect was only observed for donors that displayed a lower baseline extracellular matrix formation. For donors with a high chondrogenic baseline potential, the positive effect of miRNA inhibition was not observed. Gene expression levels of classical chondrogenic markers failed to explain the mechanisms behind miRNA inhibition effects. Therefore, we performed RNA sequencing at an early time point (day 3) to identify DE genes downstream of miRNA inhibition in responder donors. Gene Ontology analysis revealed that differentially expressed genes between miRNA inhibition groups and the negative control are involved in processes such as extracellular matrix organization, collagen fibril organization and collagen biosynthesis. qPCR validation of RNAseq results validated the differential expression of GDF5, CCN5, and SKIL by miR-335-5p inhibition in responder donors, but not in non-responders. This can explain the differences in the responses observed in our BMSC donor cohort. These three genes are involved in TGF-β and WNT signalling with a known role in chondrogenic differentiation. In particular, GDF5 encodes for a ligand of the BMP and TGF-β superfamily, also known as BMP-14, which binds to the BMPR1B and BMPR2 receptor, is key to cartilage formation and homeostasis, and is predominantly expressed in the superficial zone of articular cartilage [[89]33]. It can be considered as an early marker of joint development [[90]34]. GDF5 has a key role during prechondrogenic condensation [[91]35] and early chondrogenic differentiation [[92]36]. In vitro, it was shown to enhance MSC chondrogenic and hypertrophic differentiation [[93]37]. WISP2 or CCN5 is a member of the WNT1 inducible signalling pathway (WISP) protein subfamily, which is expressed early during chondrogenic differentiation but decreased at later time points during hypertrophy onset [[94]36, [95]38]. CCN5 is highly expressed in osteoarthritis (OA) and in rheumatoid arthritis [[96]39], though it might have a protective role in IL-1β induced chondrocyte inflammation [[97]38]. Finally, the protein encoded by the SKIL gene, also known as SnoN, is a both a target and a negative regulator of TGF-beta signalling through interaction with Smad2 and Smad4 [[98]40–[99]42]. SKIL might mediate a negative feedback mechanism by TGF-β to inhibit hypertrophic maturation of chondrocytes [[100]43]. Overall, miR-335-5p inhibition enhances the progression through cartilage maturation and hypertrophy in donors with a lower baseline differentiation potential. This is in accordance with our previous results [[101]16] which showed a decrease in circulating levels of this miRNA in fracture patients, where endochondral ossification is essential to bone healing. We also previously demonstrated that miR-335-5p is downregulated in both osteogenic and chondrogenic differentiation at day 7 [[102]44]. miR-335-5p is associated with osteoarthritis and its downregulation might improve cartilage formation [[103]45, [104]46], though data from mice and patients indicate that the functional role of this miRNA functional effects may depend on the tissue, joint, species, and OA disease stage [[105]47, [106]48]. While qPCR analysis failed to validate the differential expression of genes identified with RNAseq, it seems that miR-1246 inhibition has similar effects to that of miR-335-5p on chondrogenic differentiation through other mechanisms yet to be determined. Multiple factors can influence the differentiation potential of MSCs and their responsiveness to miRNA modulation. In this study, it is notable that the two donors with the highest chondrogenic potential were also the youngest, while the others were over 70 years old. Although MSC differentiation potential is proposed to decline with age in vivo [[107]49], the extent to which this negative correlation holds true in vitro is debatable. In vitro systems introduce numerous variables, such as FBS batch, culture conditions, and operator variability, that can shadow age-related trends, even when these factors are carefully controlled. Consequently, donor age alone is not a reliable predictor of chondrogenic potential in vitro [[108]27, [109]50]. Moreover, the limited number of donors in this study precludes definitive conclusions regarding the impact of age. Additional patient information, such as comorbidities, is also lacking. While one might speculate that miR-335-5p and miR-1246 inhibition could preferentially benefit older individuals, we prefer to generalize our findings: this inhibition may enhance endochondral differentiation in cells with a reduced chondrogenic capacity, regardless of the underlying cause. Future studies should further investigate the relationship between donor age, MSC differentiation potential, and responsiveness to miRNA modulation. Finally, miR-193a-5p has no apparent effect on chondrogenic differentiation, but its inhibition significantly downregulated SLC2A3 and SLC2A5, hexose transporters involved in cellular uptake glucose and fructose [[110]51]. We cannot exclude that a regulation of this miRNA and the hexose transporter might have an effect later during chondrocyte differentiation, especially for the formation of the glycosaminoglycan-rich extracellular matrix where glucose plays an essential role [[111]52]. Conclusions The transient inhibition of miR-335-5p and miR-1246 during the early stages of human in vitro chondrogenesis in a 3D pellet culture model results in a donor-dependent and potentially age-related increase in cartilage formation and hypertrophic differentiation, as evidenced by increased Safranin-O, type II, and type X collagen staining. This result suggests that miR-335-5p and miR-1246 have a regulatory effect on endochondral ossification. For miR-335-5p, we have identified target genes involved in the TGF-β and WNT signalling pathways. The function of these miRNAs in human bone formation and repair should be further investigated to validate their role as a theranostic marker in fracture healing. Acknowledgements