Spinal cord injury dysregulates fibro-adipogenic progenitors miRNAs signaling to promote neurogenic heterotopic ossifications

Neurogenic heterotopic ossifications are intramuscular bone formations developing following central nervous system injury. The pathophysiology is poorly understood and current treatments for this debilitating condition remain unsatisfying. Here we explored the role of miRNAs in a clinically relevant mouse model that combines muscle and spinal cord injury, and in patients’ cells. We found an osteo-suppressive miRNAs response in injured muscle that was hindered when the spinal cord injury was associated. In isolated fibro-adipogenic progenitors from damaged muscle (cells at the origin of ossification), spinal cord injury induced a downregulation of osteo-suppressive miRNAs while osteogenic markers were overexpressed. The overexpression of selected miRNAs in patient’s fibro-adipogenic progenitors inhibited mineralization and osteo-chondrogenic markers in vitro. Altogether, we highlighted an osteo-suppressive mechanism involving multiple miRNAs in response to muscle injury that prevents osteogenic commitment which is ablated by the neurologic lesion in heterotopic ossification pathogenesis. This provides new research hypotheses for preventive treatments.


N
eurogenic Heterotopic Ossifications (NHOs) are ectopic extra-skeletal bone formations, that occur after central nervous system injury, primarily spinal cord injury (SCI), traumatic brain injury and stroke [1][2][3] .NHOs develop at the expense of skeletal muscle tissue, are located around major joints and can impair limb mobility 4,5 .NHOs can induce joint ankylosis, nerve and vascular compressions and severe pain.Surgical resection is currently the most efficient and reliable treatment 5,6 .Up to 30% of SCI and 20% of traumatic brain injury patients develop NHOs, leading to comorbidities and compromising patient's rehabilitation 6,7 .
NHOs development involves a process of endochondral ossification and can lead to a mature bone containing a functional hematopoietic niche [8][9][10] .Fibro-adipogenic progenitors (FAPs) are muscle interstitium resident mesenchymal stromal cells that are major cellular actors in HOs development and specifically in NHOs [11][12][13] .Physiologically, FAPs support myogenic progenitors (MPs) proliferation and myoblastic differentiation upon muscle repair but can also lead to fibrosis under pathological conditions [14][15][16][17] .FAPs directly participate in bone fracture repair, exhibiting previously unsuspected physiological osteogenic capabilities 18 .Muscle regeneration after acute injury is orchestrated by MPs activation, proliferation, differentiation into myoblasts and fusion to form new myofibers [19][20][21] .Lineage tracing experiments have shown that MPs do not contribute to HOs nor NHOs but muscle regeneration seems to be compromised 13,22 .Inflammation is a key mechanism in NHOs pathogenesis.Pro-inflammatory cytokines and macrophages, but not neutrophils, are involved and mandatory to drive FAPs towards an osteogenic fate 23,24 .Central nervous system injury triggers local and systemic inflammatory response 25,26 .Pro-inflammatory cytokines Interleukin-1 and Oncostatin M directly support NHOs pathogenesis in a mouse model and in vitro with patient's muscle progenitors 8,27 .Despite all previous studies, some aspects of NHOs pathogenesis are still unclear, specifically the impact of the central nervous system lesion on FAPs and MPs signalling and fate.
MiRNAs are small non-coding single stranded RNA that posttranscriptionally regulate gene expression by binding and silencing their mRNA target 28,29 .Numerous miRNAs are physiologically involved in bone development and repair processes, they regulate osteoblast proliferation/differentiation or hypertrophic maturation of chondrocytes by targeting key molecular actors such as Runx2, Sox9, Osx, Bmp2, Atf4 or their repressors [30][31][32] .While cellular actors and numerous cytokines have been thoroughly investigated in HOs, the potential involvement of miRNAs is still poorly described.MiR-17-5p and miR-203 regulate HOs in ankylosing spondylarthritis and in a non-neurogenic HO model, respectively 33,34 .In a small cohort of human traumatized muscle samples, miRNA dysregulation was found in patients developing HOs and 2 myomiRs were proposed to contribute to ectopic ossification 35 .Although these studies demonstrate the involvement of miRNAs in HOs formation, their possible role in NHOs remains unclear.
We hypothesized that miRNAs could participate to NHOs formation and studied their expression in our NHOs mouse model combining SCI and acute cardiotoxin (CDTX) muscle injury 23 .In this model, SCI or muscle injury alone do not elicit NHOs, both injuries are needed.To understand why injured muscle fails to regenerate and engages in osteo-chondrogenic processes following SCI, we performed a large miRNAs profiling at early stages, in skeletal muscle tissue.Candidate miRNAs were then studied in mice FAPs and MPs isolated from our model.Dysregulated miRNAs were further studied in vitro in osteogenic differentiation conditions in FAPs isolated from NHOs patients muscle sample and functional assays were performed on human FAPs with the most relevant ones.
In this study, we identified early osteo-suppressive miRNAs downregulation in FAPs concomitant to myogenic miRNAs downregulation in MPs in the NHOs mouse model.We also confirmed that overexpression of these osteo-suppressive miR-NAs could attenuate mineralization and osteogenic markers expression from human FAPs.

SCI induces mineralization and exacerbates osteochondrogenic markers expression in CDTX damaged muscle.
To investigate the role of miRNAs in NHOs, we used our mouse model that combines muscle injury by intramuscular injection of cardiotoxin (CDTX) and concomitant spinal cord injury (SCI) to generate ectopic ossification, or their controls (intramuscular PBS injection and SHAM surgery) 23 (Fig. 1a).We collected gastrocnemius muscle 2, 4 or 7 days later.In injured muscle, the hematoxylin phloxine saffron staining showed classical myofiber necrosis with massive mononuclear cells infiltration at D2, D4 and regenerative myofibers at D7 (Fig. 1b, top and middle panel).No mineralization was observed with Von Kossa staining with CDTX alone (Fig. 1b) nor with SCI alone (Supplementary Fig. 1).Undamaged muscle from SCI mice displayed myofibers atrophy at D7 (Supplementary Fig. 1).SCI combined with CDTX injection induced myofibers necrosis and cell infiltration at D2, D4 and D7 (Fig. 1b).Von Kossa staining showed calcified necrotic fibers (D2, D4) and nodules (D7), consistent with previous findings 23 .Immunostaining analysis showed disrupted laminin staining and impaired muscle regeneration, concomitant to extracellular osteocalcin and collagen III deposition in SCI + CDTX muscles, demonstrating that osteogenic differentiation occurred (Fig. 1c and Supplementary Fig. 1).These results confirm that both muscle and spinal cord injuries are mandatory to the development of NHOs.
The kinetics of signaling pathways activation involved in the early stages of NHOs development is still poorly described, therefore we investigated the expression of osteo-chondrogenic and myogenic markers at the mRNA level in muscle tissue at D2, D4, D7 in PBS + SHAM, PBS + SCI, CDTX + SHAM, CDTX + SCI muscles (Fig. 1d).We observed increased expression of osteo-chondrogenic markers Runx2, Sox9 and Sp7 induced by CDTX injection while CDTX + SCI exacerbated Sox9 expression compared to CDTX alone at D2.Moreover, Runx2, Col10a1 and Tgfb1 expressions were increased at D7 in CDTX + SCI muscle compared to CDTX alone.Surprisingly, Alpl, Vegfa, Atf4 and Bmp2 expression levels were lowered in both CDTX conditions and PBS + SCI at each time point compared to PBS + SHAM.Myf5 myogenic marker expression was significantly lowered in CDTX + SCI muscle compared to CDTX alone at D4 while Myod1 and Myog expression did not change significantly (Fig. 1d).
Myogenic markers and miRNAs involved in muscle regeneration are downregulated by SCI in MPs from injured muscle.While we recently reported that MPs were absent from mineralized nodules in NHOs 14 days post CDTX + SCI injuries, MPs remain a population of interest at the early stages of NHOs pathogenesis since SCI impairs muscle regeneration 13 .MPs cell number drastically decreased 2 days after CDTX injury before a rapid expansion at D4 (Fig. 4a).Interestingly, MPs expression of Alpl and Bmp2 followed the same trend as FAPs with a lower expression in response to CDTX, partially alleviated in MPs from CDTX + SCI muscle (Fig. 4b).MPs also expressed osteochondrogenic markers, Runx2 was expressed similarly in CDTX and CDTX + SCI groups compared to healthy muscle while Sox9 expression was higher in CDTX + SCI MPs compared to MPs from CDTX-injured muscle at D4 (Fig. 4b).Pax7 and Myf5 are markers of MPs quiescence while Myod1 is a marker of MPs activated state.As expected, we observed the activation of MPs in response to acute muscle damage in both CDTX and CDTX + SCI muscles, with a drop in Pax7 and Myf5 expression (Fig. 4c).Interestingly, Pax7 and Myf5 expression levels were lowered in MPs from CDTX + SCI muscle compared to CDTX muscle at D4 while Myod1 expression levels increased steadily in both experimental conditions.Lastly, Myog is involved in myogenic differentiation.We found an activation at D4 in response to CDTX that was blunted in CDTX-SCI mice suggesting a delayed differentiation (Fig. 4c).

Discussion
Although it is currently admitted that NHOs follow an intramuscular osteochondral development, the involvement of the central nervous system injury in the initial physiopathological stage remains unclear.Our results uncover a link between muscle FAPs and MPs miRNAs expression and distant central nervous system injury leading to the onset of NHOs in a clinically relevant mouse model.After muscle injury, we found an increase in osteochondrogenic markers Runx2, Sox9, Col10a1 and Tgfb1.Osteogenic and most osteo-suppressive miRNAs were activated in CDTX-injured muscle tissue while SCI increased the expression of the osteogenic miR-20a-5p in CDTX muscle at D4. Interestingly, the combination of muscle injury and SCI prompted osteogenic miRNAs response while partially inhibiting osteosuppressive miRNAs expression.Among the most SCIdownregulated miRNAs 4 days after injury, we identified suppressors of osteogenesis and chondrogenesis : miR-214-3p, miR-199a-3p, miR-574-3p [37][38][39][40] .These results show that miRNAs landscape in skeletal muscle is profoundly altered in response to muscle damage and that SCI substantially modifies the osteogenic/osteo-suppressive miRNA balance.
Because miRNAs expression in whole muscle may reflect not only active regulations but also profound changes in cell composition, we next investigated miRNAs signature in sorted FAPs, the cells of origin of NHOs.We initially hypothesized that FAPs osteogenic miRNAs would be upregulated as observed in CDTX + SCI muscle, however, their expression levels were lowered at D2 compared to CDTX alone.Similarly, while osteosuppressive miRNAs were upregulated in FAPs after injury, this response was blunted in CDTX-SCI group, including miR-214-3p, whereas osteo-chondrogenic markers were overexpressed.Interestingly, Eisner et al. have previously reported FAPs osteogenic phenotype during muscle inflammation in the onset of traumatic heterotopic ossifications 41 .These results suggest a miRNAs osteo-suppressive mechanism activated by FAPs in response to muscle injury and osteogenic signaling during muscle damage to prevent aberrant FAPs osteogenic commitment, which is impaired by SCI.
Our mouse model is characterized by a SCI-induced failure of muscle regeneration and ectopic ossification.To understand how SCI impairs muscle regeneration and induces NHOs, we next studied MPs.As MPs do not contribute to NHOs formation, we Fig. 5 Conditioned medium from LPS-stimulated-CD14 + macrophages (CM + ) potentiates human FAPs osteogenic differentiation and increases osteochondrogenic markers and miRNAs expression.a Experimental design of human muscle progenitors isolation from NHOs surgical resection.Briefly, human muscles were collected around the ectopic ossification, after surgical removal, mechanically and enzymatically minced, cultured for 2 weeks, and adherent cells were sorted on PDGFRα + CD56 − for FAPs and PDGFRα − CD56 + for MPs from 7-AAD − live single events.b Osteogenic differentiation assay of human FAPs cultured in osteogenic conditions (OB), supplemented with lipopolysaccharides (LPS, 100 ng/mL) and naive (CM − ) or LPSstimulated conditioned medium (CM + ) from macrophages, obtained through CD14 + isolation from human buffy coat, in a 1:10 ratio.Mineralization was observed using alizarin red staining and quantified at 405 nm after acid-SDS extraction.Each dot represents independent biological samples from different NHOs surgery (n = 6-18).c Alkaline phosphatase activity was assessed, 7 days after osteogenic induction with or without CM + by monitoring the optical density at 405 nm after incubation with a commercial substrate at 37 °C (n = 5).d OSX and RUNX2 protein expression levels were quantified by western blot 7 days after osteogenic induction with or without CM + (n = 3-5).Relative RT-qPCR e osteogenic mRNA and f miRNAs expression levels from FAPs control, OB, OB + CM + at 1, 3, 7 and 10 days after osteogenic induction (n = 5) (*P < 0.05, **P < 0.01, ***P < 0.001 represent group effect compared to Ø condition ; # P < 0.05, ## P < 0.01, ### P < 0.001 represent group effect compared to OB condition).Histograms represent mean ± standard deviation (SD).Each dot represents independent biological samples from different NHOs surgery.Statistical differences were calculated using Mann-Whitney U test for simple comparisons or two-way ANOVA with Tukey post hoc correction for multiple comparisons (*P < 0,05, **P < 0,01, ***P < 0,001).anticipated limited modulation of osteogenic and osteosuppressive miRNAs 13 .Our results showed however that MPs osteo-suppressive miRNAs were downregulated 4 days after NHOs induction, as observed for FAPs miRNAs downregulation after 2 days.However, it should be noted that miR-214-3p has well-documented roles in myoblast proliferation and differentiation therefore the expression pattern we found may be totally disconnected from any osteogenic process 42,43 .Myogenic marker Myog and myomiRs were also downregulated in MPs from NHOs developing muscles suggesting altered or delayed myogenesis.Interestingly, SCI-downregulated myogenic miR-1-3p which targets the master regulator of chondrogenesis Sox9, found to be increased in MPs from NHO muscle 44 .
We next studied human FAPs sorted from patient's muscle surrounding NHOs.We used our previously published model where FAPs are cultured with conditioned media from activated macrophages to mimic NHOs muscle environment in vitro 8,27 .We confirmed that the osteogenic differentiation and osteogenic markers expression were significantly increased.Interestingly, osteo-suppressive miR-214-3p was the only miRNA identified in both models that was not overexpressed in mineralizing conditions (CDTX + SCI and OB + CM + ).The functional analysis Fig. 6 Synthetic miRNAs inhibitors and mimics transfections suppress both mineralization and osteo-chondrogenic markers expression in human FAPs.a FAPs viability was assessed by quantifying DAPI intake intensity by dead or apoptotic cells in flow cytometry in five experimental groups: control, lipofectamine RNAimax vector, vector+control oligo, vector+miR-20a-5p inhibitor and vector+miR-214-3p mimic.Inhibitors transfections were performed during 48 h at 10 nM and mimics at 1 nM for both control oligos and specific oligos (n = 5).b Relative RT-qPCR miRNAs expression levels in FAPs after control oligo or miR-20a-5p inhibitor and miR-214-3p mimic transfection (n = 5-6).c Osteogenic differentiation assays of transfected human FAPs cultured in osteogenic conditions (OB).Control inhibitor, miR-20a-5p inhibitor, miR-199a-5p inhibitors were used at 10 nM final and control mimic, miR-146a-5p mimic, miR-214-3p mimic at 1 nM final (n = 12-25).Values were normalized to OB condition alone.Statistical differences were calculated using unpaired t test with Welch correction (*P < 0,05, **P < 0,01, ***P < 0,001).d MTOR, SOX9 and RUNX2 protein expression levels from FAPs were quantified by western blot 7 days after osteogenic induction of control oligo or miR-20a-5p inhibitor transfection (top panel).OSX, SOX9 and RUNX2 protein expression levels were quantified by western blot 7 days after osteogenic induction of control oligo or miR-214-3p mimic transfected FAPs (bottom panel) (n = 5-7).Histograms represent mean ± standard deviation (SD).Each dot represents independent biological samples from different NHOs surgery.Unless stated otherwise, statistical differences were calculated using Mann-Whitney U test for simple comparisons or one-way ANOVA with Tukey post hoc correction for multiple comparisons (*P < 0.05).
revealed that miR-20a-5p and miR-199a-5p inhibitors as well as miR-146a-5p and miR-214-3p mimics reduced FAPs mineralization.Downstream osteo-chondrogenic markers were also decreased at protein level, including the key regulators RUNX2 and OSX.MiR-20a-5p and miR-199a-5p are involved in osteogenic differentiation of human stem cells, respectively regulating BMP/SMAD and HIF1a signaling [45][46][47][48] .On the other side of the spectrum, miR-214-3p and miR-146a-5p have both been described in osteo-suppressive mechanism by targeting ATF4 and SIRT1 37,49 .It is interesting to note that miR-199a-3p and miR-199a-5p are the two arms of the same pre-miRNA but exhibit opposed roles on osteo-chondrogenic process, as well as different abundancy levels in mice FAPs.This suggests that the same pre-miRNA could contain both the stimulation and the inhibition signals, and that each strand has an independent fate after miRNA maturation.As such, investigating their active regulation such as sponging RNAs could be of great interest to decipher their role in NHOs pathogenesis.As transfection of a mix of the four candidates could not improve the osteogenic inhibition, we think that these miRNAs target similar osteogenic pathways in redundant ways.Taken together, our results suggest that FAPs osteogenic differentiation in NHOs is supported by blunted osteo-suppressive miRNAs response linked to SCI concomitant to osteogenic stimuli after muscle injury.Indeed, basal miRNAs expression levels prevent FAPs from undergoing osteogenic differentiation in healthy muscle or control medium, alike the simultaneous upregulation of both osteogenic and osteosuppressive miRNAs in injured/regenerating muscle.Hence, overexpression of osteogenic markers and lack of osteosuppressive miRNAs activation in differentiation and mineralization.
In the present study, we studied isolated FAPs (bone forming cells) and MPs (that fail regenerating) but not macrophages that are key players in NHOs.This is a limitation of our study, but also a deliberate choice, since the complexity of the results leads us to believe that a dedicated study is needed.Our results on whole muscle tissue bring clues and confirm the need for such a study.MiR-214-3p and miR-146a-5p have been reported to regulate M2 macrophage polarization 50,51 while miR-20a-5p targets TGFβ signalling, favouring M1 pro-inflammatory macrophages phenotype 52,53 .As SCI exacerbates inflammation, those miRNAs could represent promising targets to couple osteosuppressive effect and pro-resolution macrophages capacities 54 .Macrophages differentiation and polarization may be studied to better understand how they affect NHOs development and how the SCI could potentially skew miRNAs signaling leading to altered MΦ/M1/M2 activations.The question of a transfer of miRNAs from macrophages to FAPs that could support their osteoblastic differentiation goes beyond the aim of the present study, but further investigations may be of great interest since inflammation is a key mechanism in NHOs.Of further note, in a rat model of traumatic brain injury, injured neurons produced and released osteogenic miRNAs that accelerate bone healing 55 .It should however be acknowledged that despite abundant literature describing miRNAs signaling between cells, the importance of such mechanisms is still debated [56][57][58] .Taken together, our results show for the first time that central nervous system injury substantially modifies the miRNAs landscape in injured muscle as well as in FAPs and MPs.The functional analysis in NHO patient's cells demonstrates that several dysregulated miRNAs alter FAPs osteogenic differentiation.Our results suggest that NHOs pathogenesis is not regulated by a single miRNA but rather by a finely tuned balance between osteogenic and osteosuppressive miRNAs.Muscle damage upregulates both osteogenic and osteo-suppressive miRNAs while central nervous system damage disrupts this equilibrium by lowering osteo-suppressive miRNAs response.More work is needed to determine whether intercellular miRNAs signaling also contributes to NHOs pathogenesis and how macrophages participate in this complex signaling.Six to eight weeks old female C57Bl/6 mice from Janvier Labs (France) underwent spinal cord transection injury (SCI) between T11-T13 or SHAM surgery (control) followed by intramuscular injections of 12 μM cardiotoxin (CDTX, L8102, Latoxan, left hindlimb) and phosphate-buffered saline (PBS, right hindlimb) into each gastrocnemius (150 μL).Mice were euthanized at 0 (naive), 2, 4 or 7 days post-surgery and both gastrocnemius were harvested and either processed for cell isolation or flash frozen in isopentane (M32631, Sigma-Aldrich) and embedded in OCT.Naive mice were also used.The occurrence and volume of NHOs in male or female mice is not different in our model, female mice were used to because urination is blocked by the SCI and triggering it much easier in females, thereby reducing anesthesia duration and infection risk.

Methods
Muscle histology and immunohistochemistry.In all, 8 μm cryosections were dried and stained with Hematoxylin Phloxine Saffron (HPS) (Dako).For Von Kossa staining, cryosections were incubated 30 min at RT in 1% silver nitrate in the dark followed by 15 min incubation in UV chamber and nuclear fast red counterstain.All sections were analyzed using a Pannoramic Midi II slide scanner and Case Viewer software (3D HISTECH Ltd.).
Human ethics statement.All human samples were obtained with the informed consent of the patients, the approval from the people protection committee (CPP approval n°09025, study BANKHO) and the approval from the National Commission for Informatics and Liberties (CNIL approval n°Eyo1066211J).
Alkaline phosphatase activity assay.Alkaline phosphatase activity assays (AP0100, Sigma) were realized according to manufacturer's instructions.
AntimiR and mimic transfection.Functional analyses were performed using miRCURY LNA Power Inhibitors (10 nM final, 339121, Qiagen) and miRCURY LNA Mimics (1 nM final, 339173, Qiagen).Oligonucleotides were incubated 5 min at room temperature with Lipofectamine RNAiMAX (1.5 μL/mL, 13778075, ThermoFisher) in α-MEM prior to cell supplementation.After 48 h, culture medium was changed for osteogenic differentiation medium (see before).Viability was assessed by measuring dapi intensity in flow cytometry.We also performed negative control western blots to avoid the case of RISC complex saturation by analogous sequences leading to non-specific miR-NAs biogenesis dysregulation as described by Khan et al. 60 .
Viability assay.To assess viability of human FAPs after transfection with miRNAs inhibitors or mimics, we stained cells after 48 h of transfection with 5 μg/mL of DAPI for 5 min.Fluorescence was quantified in flow cytometry (CytoFLEX, Beckman Coulter).
Statistics and reproducibility.All graphs and dots represent independent biological samples from different animals, including sorted muscle progenitors populations, or from different human biopsies.Statistical analyses were performed using GraphPad Prism (9.1.1).All plots represent mean ± standard deviation (SD) of different biological samples.Statistical differences were calculated using non-parametric two-sided Mann-Whitney U test for simple comparisons, one-way ANOVA with Tukey post hoc correction for multiple comparisons and two-way ANOVA with Tukey post hoc correction for multiple intra-groups comparisons or Šidák post hoc correction for multiple inter-groups comparisons (one-sided) (*P < 0.05, **P < 0.01, ***P < 0.001).Supplementary reagents.Antibody concentrations and miRNAs sequences are provided in separate tables.
Reporting summary.Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Fig. 2
Fig. 2 Muscle injury triggers osteogenic miRNAs expression while SCI impairs and delays osteo-suppressive miRNAs response after muscle damage.RT-qPCR analysis on whole muscle at day 2 and 4 presenting top ten most overexpressed and underexpressed miRNAs from a CDTX + SHAM compared to PBS + SHAM, b CDTX + SCI compared to CDTX + SHAM.MiRNAs involved in osteogenic, inflammatory, and myogenic processes are highlighted in red (upregulated) or blue (downregulated).A threshold was established to exclude miRNAs expressed later than 32 Cp.Relative miRNAs expression levels measured by RT-qPCR from whole muscle PBS + SHAM, PBS + SCI, CDTX + SHAM, CDTX + SCI at days 2, 4 and 7 for c osteo-suppressive miRNAs, d osteogenic miRNAs and e myomiRs.Histograms represent mean ± standard deviation (SD).Each dot represents independent biological samples from different mouse muscles (n = 6).Statistical differences between PBS + SHAM/ PBS + SCI and CDTX + SHAM/CDTX + SCI were calculated using twoway ANOVA with Tukey post hoc correction for multiple comparisons (*P < 0.05, **P < 0.01, ***P < 0.001).
NHOs mouse model.Animal experiments were conducted in accordance with the Directive 2010/63/EU of the European Parliament and of the Council of September 22, 2010 (2010/63/EU).The experimental procedures were approved by the "C2EA-26" Ethics committee in accordance with French Ministry of Research regulations (#25901-2020060515467494).