HO is often divided into three categories: traumatic, neurogenic, and genetic. The prevalence of traumatic-induced HO following burn injury has been reported to range from 0.2 to 4%, and up to 90% following the total hip joint arthroplasty or acetabular fractures (Cipriano et al., 2009; Maender et al., 2010; Rath et al., 2013; Medina et al., 2014; Medina et al., 2015). The predilection age of trauma-induced HO is 20–40 years old. Approximately half of HO occurs at this age. However, the other half of the HO could present dispersedly from infancy to late adulthood (Ackerman, 1958; Elmas and Shrestha, 2017; Xu et al., 2017; Meyers et al., 2019; Kaliya-Perumal et al., 2020). The morbidity of heterotopic ossification following central neurologic injury has been calculated to range from 10 to 53% (Teasell et al., 2010). Most studies regard traumatic brain-injured patients and spinal cord injured patients as the same category. And the prevalence of genetic HO, including fibrodysplasia ossificans progressiva (FOP), progressive osseous heteroplasia (POH), and Albright’s hereditary osteodystrophy (AHO) (Shore and Kaplan, 2010), is extremely rare, affecting 1 in 2,000,000 people (Baujat et al., 2017). However, genetic HO is consensually regarded as the most severe HO disease in humans (Qi et al., 2017; Kaliya-Perumal et al., 2020). Male sex, the amount, and the type of motion could also raise the risk of HO. Men are slightly more vulnerable to HO with a sex ratio of 3:2 (Meyers et al., 2019), perhaps due to the various muscle mass, differential level of physical activity, repetitive mechanical stress working as “microtrauma”, and distinct hormonal signaling pathways affecting osteogenesis (Ranganathan et al., 2015; Ko et al., 2016; Malca et al., 2018; Dowdell et al., 2020; Rüdiger et al., 2020).

The typical clinical features of HO include the limited range of motion around the involved joint, complete bony ankylosis in severe cases, and deformity in the cervical spine, elbow, shoulder, fingers, jaw exostosis, or temporomandibular joint ankylosis (TMJA) (Zhao et al., 2020). HO could occur almost anywhere in the body, as long as it is associated with the periosteum. Typically, HO initiates away from the periosteum, and then fuse to the periosteum as a secondary feature (Meyers et al., 2019). But it is rare to observe HO in some anatomic tissues, such as the viscera or the diaphragm. This might be due to the lack of pluripotent stem cells in these sites or because these sites are not mechanically stimulated as often as the peri-articular areas prone to heterotopic ossification. Moreover, HO can only be detected as an asymptomatic finding on a radiograph. It is quite challenging to identify the potential biomarkers for early disease detection and monitoring, let alone the symptom present with complications that usually confound diagnosis (Crowgey et al., 2018). There are several ways to classify HO diseases. Four levels of classification for HO around the hip were set by Brooker to indicate the severity (Brooker et al., 1973). The Hastings and Graham classification system classifies HO at the elbow into three grades based on clinical and radiographic data (Hastings and Graham, 1994).

The presentations of genetic HO are more serious than traumatic-induced HO. Almost all FOP patients reported to date were caused by Acvr1 mutation, and showing abnormality early. Acvr1 gene locates on chromosome 2 (2q23-24) and encodes a bone morphogenetic protein (BMP) type 1 receptor, which is generally considered to be the major regulator in HO pathophysiology (Wang et al., 2016; Haupt et al., 2019; Meyers et al., 2019; Pearson et al., 2019; Stanley et al., 2019; Botman et al., 2020; Kaliya-Perumal et al., 2020). Acvr1 mutation results in abnormally enhanced sensitivity of this receptor to BMPs, allowing for overexcitation of the BMP/SMAD pathway and heterotopic ossification. The typical feature of FOP is multiple skeletal deformities, involving fingers, toes, and cervical spine, and eventually resulting in pain, movement, and function limitation. POH is a genetic HO caused by inactivating mutations in the GNAS1 gene, which result in decreased expression or function of the alpha subunit of the stimulatory G protein (Gsα) of adenylyl cyclase (Zhang et al., 2018). POH is characterized by intramembranous and cutaneous ossification, and could occur on the ear or fingers as an atypical phenotype (Kaplan et al., 1994; Zhang et al., 2018).

However, HO may be alleviated by physical intervention for traumatic-induced patients such as immobilization or Long-term bedridden. Doctors routinely use immobilization for extremity trauma patients (Kunz et al., 2014). But the mechanism that how immobilization protects the injury site reduces pain and improves healing remains unknown (Huber et al., 2020). Conversely, heterotopic ossification may become more severe in patients with insufficient immobilization and bed rest after fracture injury or joint surgery.

One of the most significant differences between pathological heterotopic ossification and physiological osteogenesis is the distinct cellular source. The cellular origin of physiological osteogenesis is the differentiation of preosteoblast, but the precursor cellular origin of pathological heterotopic ossification has not been fully investigated. Table 1 summarizes the cell types that contribute to heterotopic ossification based on currently published studies. In general, the cellular origin of pathological osteogenesis is not limited to the osteoblast lineage, but potentially results from the pluripotent differentiation of a diverse range of stem cells.

To be more specific, Ctsk was previously found to be able to label osteoclasts and periosteum stem cells. Recently, a subgroup of tendon-derived progenitor cells (TDPCs) was also found to be labeled by Ctsk (Feng et al., 2020). TDPCs, as stem cells in tendon tissue, are capable of multidirectional differentiation and would differentiate towards osteogenesis under certain conditions resulting in heterotopic ossification. In addition, mesenchymal stem cells in tendon areas could also be activated to osteogenic differentiation, which can be labeled by Nfatc1-Cre, Prx1-Cre, and Dermo1-Cre. It is possible that some other cells with proliferative capacity may also shift to osteogenic differentiation in some conditions. For example, perivascular cells (Gli1-Cre), PS⁺ and SP7⁺ cells from peripheral nerves, and muscle satellite cells (Tie2-Cre/VE-Cadherin-Cre) all contribute to HO. In conclusion, the cellular origin of HO is relatively complicated, and a variety of cells have the potential to shift to osteogenic differentiation in response to some specific stimulus, which in turn promotes HO formation.

Inflammation serves as an important microenvironmental alteration in the development of heterotopic ossification. Trauma leads to a state of local and systemic inflammation, resulting in elevated inflammatory cytokines, such as TNFα, IL-1β, IL-6, and MCP-1, which could cause abnormal activation of mesenchymal stem cells in the soft tissues (Sung Hsieh et al., 2017). Inflammation-associated cells, such as macrophages and mast cells, also accumulate at the site of trauma-induced heterotopic ossification and promote heterotopic ossification (Convente et al., 2018). Lymphoid tissues also contribute to the cellular niche in Heterotopic Ossification (Loder et al., 2016). The main role of inflammation is to turn MSCs, such as normal fibroblast lineage, into the osteogenic lineage, initiating the onset of heterotopic ossification.

The hypoxic state of local tissues after trauma may also initiate heterotopic ossification. Regional tissue hypoxia causes the activation of Hypoxia-inducible factors (HIFs), consisting of 1 of 3 α subunits bound to HIFβ (Meyers et al., 2019). HIFs could increase the production of pro-angiogenic cytokines such as VEGF, facilitating localized pathological bone tissue formation (Dilling et al., 2010; Hwang et al., 2019). The inhibition of HIFs could attenuate HO formation in experimental models (Agarwal et al., 2016).

Most of the fundamental research on heterotopic ossification is presently based on traumatic and genetic mouse models. In general, hyperactivation of bone morphogenetic protein (BMP) and consequent cascading activation of activin type-1 receptor (ACVR1) is thought to lead to abnormal endochondral osteogenesis, resulting in heterotopic ossification. The dysregulation of Hedgehog (Hh) signaling also contributes to many HO. However, recent studies have suggested that this pathological osteogenic process may share similar biological mechanisms with physiological osteogeneses, such as RUNX2, a classical osteogenic transcription factor (Kim et al., 2020). CK2/HAUSP pathway is a critical regulator of RUNX2 stability because Casein kinase 2 (CK2) phosphorylates RUNX2 and recruits the deubiquitinase herpesvirus-associated ubiquitin-specific protease (HAUSP) to stabilize RUNX2 away from ubiquitin-dependent proteasomal degradation. Meanwhile, regional osteoclast activities are also enhanced during the formation of heterotopic ossification, as the formation of the bone marrow cavities depends on a dynamic balance between osteogenesis and bone resorption. Furthermore, osteogenic-osteoclastic crosstalk, such as the transforming growth factor-beta (TGF-β) released after augmented osteoclastic activity that recruits mesenchymal stromal/progenitor cells (MSPCs) in the HO microenvironment for bone remodeling activities, also plays an important role in heterotopic ossification (Wang et al., 2018). PDGF-BB concentration was also increased during HO progression. Therefore, the bone formation process of heterotopic ossification is different but correlated to that of normal physiological osteogenesis.

Some proteins that affect bone morphology and bone development also influence the formation of heterotopic ossification. Bone morphogenetic proteins (BMPs) are required for multiple developmental processes, including bone and cartilage formation (Kaliya-Perumal et al., 2020). BMPs bind to ACVR1, which locates on the cell membrane surface phosphorylating SMAD1/5/9(8). Phosphorylated SMAD1/5/9(8) combine with SMAD4 and import into the nucleus, regulating transcription that drives endochondral ossification (Nosho et al., 2020). When BMP receptors bonded with Activin A, SMAD2/3 is activated to regulate inflammation (Rautela et al., 2019). The occurrence of FOP is also associated with the R206H mutant substitution of Acvr1, enhancing the response to various BMP ligands (Alessi Wolken et al., 2018). Retinoic acid receptors (RARs) are morphogens that impact both osteogenesis and chondrogenesis. There is a hypothesis that RAR agonism could impede HO formation by preventing the differentiation of prechondrogenic cells, and was partly tested in a subcutaneous rBMP2-induced HO model in mice (Cash et al., 1997; Shimono et al., 2010; Riedl et al., 2020). The Hedgehog (Hh) pathway also plays an important role in HO. Hh protein inhibits the GPCR-like protein Smoothened (SMO) by binding to the Patched (PTCH1) receptor, leading to SMO aggregation in cilia and phosphorylation of the cytoplasmic tail. SMO mediates downstream signaling and induces GLI protein detachment from SUFU. GLI1 and GLI2 proteins translocate to the nucleus to activate the transcription of Hh target genes (Regard et al., 2013; Feng et al., 2020). From this viewpoint, it can be inferred that biomolecules such as microRNAs, LncRNAs, and exosomes could also regulate heterotopic ossification by influencing some specific key proteins that regulate bone morphology and development, but this remains to further study.

Heterotopic ossification can be modulated by mechanical signals. It is generally acknowledged that mechanical stress stimulation serves an important function in the physiological osteogenesis process. Osteocytes can sense local mechanical cues and thus induce bone formation, disuse-induced bone loss, and skeletal fragility (Qin et al., 2020). The primary mechanosensors in osteocytes include osteocyte cytoskeleton, dendritic processes, integrin-based focal adhesions, connexin-based intercellular junctions, primary cilium, ion channels, and extracellular matrix (Uda et al., 2017). It is now generally accepted that the traditional regulation of bone metabolism is deeply affected by mechanical stimulation signals. Current studies suggest that heterotopic ossification, a pathological osteogenic process, is modulated by mechanical signals as well. Mechanical stress initiates osteogenic differentiation of mesenchymal stem cells (MSCs) in soft tissue. Stem cell fate of MSCs shifts from favoring lipogenic cells to osteogenic cells under mechanical loading (Figure 1).

In the genetic-induced heterotopic ossification murine model, Acvr1 mutant cells change the local microenvironment, resulting in the skewing of the threshold for mechanical stimuli and becoming more sensitive to the fate of chondral/osteogenic lineages (Haupt et al., 2019). Stanley’s study revealed that mechanistic signalings of Acvr1 mutant cells in the soft matrix resemble that of non-mutant cells in the hard matrix, and are dependent on RhoA and YAP1 signaling (Stanley et al., 2019). Huber’s study found that mechanical stress can be transmitted to mechanosignaling receptors on heterotopic ossified mesenchymal progenitor cells through the extracellular matrix and cell adhesion, such as through focal adhesion kinase signaling and nuclear translocation of the transcription coenzyme TAZ, which regulates the progression of heterotopic ossification (Huber et al., 2020). However, the specific mechanism of Acvr1 in the mechanical signaling process is not clear, and no literature suggests a direct action in the mechanical signaling cascades. Because mutations in Acvr1 result in increased sensitivity to BMP, it is reasonable to believe that the Acvr1 response to mechanical stimulation is BMP-dependent.

Early studies have found that BMP-2, 4, 6, and 7 are differentially expressed depending on the mechanical stimulation (Rui et al., 2011). However, how BMPs can sense mechanical signals has been unclear for a long time, and only recently some studies have made advances. BMP-2 signaling senses mechanical signs because of the cross-talk with YAP/TAZ at the transcriptional level. In C2C12 cells, it was shown that Smad1/5/8 can be phosphorylated and translocated into the nucleus in the presence of BMP-2 signaling alone. However, activation of osteogenic genes requires cytoskeletal tension-induced nuclear accumulation of YAP/TAZ. BMP-2 signaling responds to mechanical cues by sensing nucleocytoplasmic shuttling of YAP/TAZ (Wei et al., 2020).

YAP and TAZ (also known as WWTR1) are two protooncogene proteins that are widely known as mechanosensors and mechanotransducers in various cell types (Dupont et al., 2011). The link between YAP/TAZ and mechanical signals is extensively explored in physiological osteogenesis as well as in osteogenic lineage. YAP/TAZ translocates from the cytoplasm to the nucleus depending on ECM stiffness in MSCs (Panciera et al., 2017), and mechanical niches trigger YAP/TAZ translocation contributing to osteoblastogenesis (Xiong et al., 2018). MST1/2 complexes with the scaffolding protein MOB kinase activator 1 (MOB1) to phosphorylate many proteins involved in chromatin condensation, apoptosis, and proliferation regulation, including cytoplasmic large tumor suppressor kinases 1 and 2 (LATS1 and LATS2). Activated LATS1/2, in turn, binds to YAP/TAZ and phosphorylates its serine, resulting in its retention in the cytoplasm and non-entry into the nucleus for function (Kovar et al., 2020). This part of YAP/TAZ pathway can interact with multiple signaling pathways at different levels, such as Hippo. In the process of heterotopic ossification, mesenchymal stem cells in soft tissues could be activated for osteogenic differentiation and become osteoblast rather than fibroblast after mechanical stimulation by YAP/TAZ conduction. Moreover, once MSC pluripotent differentiation leads to the initiation of the osteogenic procedure, mechanical stimulation further promotes the proliferation and differentiation of osteoblasts, resulting in enhanced heterotopic ossification (Yu et al., 2018). Simultaneously, osteoclast, as well as bone resorption activity, can also be affected by mechanical stresses, and even osteoclast-osteoblast crosstalk based on PIEZO1 could occur in response to mechanical forces (Wang et al., 2020). However, it is still unclear whether these osteoclast and osteoblast characteristics of normal bone tissue are completely identical in heterotopic ossification.

LRP5/6 is a key receptor in the Wnt signaling pathway. Wnt signaling plays a central role in the mechanotransduction of bone. But the mechanisms by which wnt signaling senses mechanotransduction signals specifically may be multi-pathway and multi-level. YAP/TAZ is still an important part of the Wnt pathway to sense mechanical signals. At the cell membrane, YAP/TAZ binds to Axin on LRP6, allowing the recruitment of β-transducin repeatase containing E3 ubiquitin-protein ligase (BTRC) to the β-catenin disruption complex (Azzolin et al., 2014). In the cytoplasm, YAP/TAZ binds to the cytoplasmic Wnt signaling transducer disheveled segment polarity protein 1 (DVL1) and inhibits its phosphorylation, thereby abrogating its translocation to the nucleus (Barry et al., 2013). Serine phosphorylated YAP and TAZ can also bind directly to β-catenin (Zhou et al., 2017). In addition, it can also function as a transcriptional co-activator. How the Wnt pathway specifically senses mechanical signals in bone metabolism has not been completely understood, but there is no doubt that the wnt pathway plays an important role in the biomechanics of bone.

In addition, mTORC1 signaling pathway serves as a mechanosensor modulating HO. Rodgers found that mTORC1 could activate quiescent stem cells into an “alert state” thus responding quickly to injury and stress conditions (Rodgers et al., 2014). The activation of mTORC1 promotes chondrogenesis and osteogenesis. Several studies have demonstrated that mechanical loading could activate the mTORC1 signaling pathway via inducing the phosphorylation of p70 S6 kinase (Lin and Liu, 2019). Chen found mechanical loading modulated HO of the tendon through the mTORC1 signaling pathway, furthermore, low elongation mechanical loading attenuated HO, while high elongation mechanical loading accelerated HO in vivo (Chen et al., 2017). Stimulated by mechanistic signaling, mTORC1 activates Sirtuin 1 (Sirt1) in the nucleus. Sirt1 is a histone deacetylase that acts as a novel bone regulator and represses the expression of sclerostin gene SOST, which is usually regarded as a strong negative regulator of osteoblast differentiation and bone formation (Liu et al., 2019). SOST inhibits β-catenin and osteogenic gene expression after binding to LRP5/6. Therefore, rapamycin, a selective mTORC1 signaling pathway inhibitor, is a potential therapeutic agent for heterotopic ossification.

Mechanical interventions may affect HO formation by altering stem cell fate. Stem cells are able to sense their mechanical environments through various mechanosensors, including the cytoskeleton, focal adhesions, and primary cilia (Chen and Jacobs, 2013). The cytoskeletal tension could be generated by the interacts between myosin and actin, which is important for mechanically induced osteogenesis of stem cells. Focal adhesion is formed by the adapter proteins linking the cytoskeleton to integrins. Forces are transmitted based on these intact focal adhesions (Nardone et al., 2017). The primary cilium is a single, non-motile, antenna-like transmembrane structure, acting as a microdomain to promote biochemical signaling (Pala et al., 2017). Joint immobilization could reduce mechanotransduction signaling (Kunz et al., 2014). In the immobilized murine model, the fate of mesenchymal progenitor cells was altered. Mobile MPCs expressed more genes related to osteogenesis and chondrogenesis, such as Sox9, Runx2, Spp1, and differentiated more into osteogenic cells; immobile MPCs expressed more genes related to lipogenesis, e.g. Fabp4, Pltp, Lrp1, and differentiated more into lipogenic cells (Huber et al., 2020). In the osteogenic-lipogenic fate shifting of MSCs caused by mechanical stimulation, sclerostin signaling potentially serves as a significant regulator. Unloading makes the expression of the sclerostin increase, which downregulates two key osteogenic procedures: Wnt/β‐catenin signaling and YAP/TAZ transcriptional activity. The crosstalks between Wnt/β‐catenin and PPARγ influence the physiological balance between osteogenesis and adipogenesis (Benayahu et al., 2019). As the MSCs are mechanically stimulated and favor osteogenic differentiation, heterotopic ossification becomes severe. Conversely, when they favor lipogenic differentiation, the amount of heterotopic ossified bone decreases. Therefore, joint immobility after injury promotes adipogenesis rather than osteogenesis, leading to reduced HO formation. And the use of pharmacologic inhibitors altering mechanical signaling may prove to be an effective therapy that spontaneously induces adipogenesis at sites prone to osteogenesis. The accumulation of fatty tissue in the joint near the site of injury is much less severe than HO, leading to a more favorable outcome (McTighe and Chernev, 2014).

Mechanical loading has also been demonstrated to cause stem cell fate shift at the cellular level (Figure 2). Mechanical loading appears to favor osteogenesis whereas unloading conditions seem to promote adipogenesis. Passive stimuli including stiffness and viscoelasticity, as well as active stimuli including tensile/compressive stress and fluid shear stress, can affect cells through the extracellular matrix (Benayahu et al., 2019). Mechanical signals are conducted from the extracellular matrix through the cytoskeleton to regulate intracellular actions. Some important signaling pathways interact with mechanistic signals. For example, Wnt ligand binding to low‐density lipoprotein receptor‐related protein 5/6 (LRP5/6) coreceptors results in the translocation of β‐catenin to the nucleus and the enhanced transcription of genes that govern osteogenesis, and its interaction with the Hippo pathway that governs the activity of YAP/TAZ, which is regarded as an important mechanistic signaling transcription factor (Benayahu et al., 2019; García de Herreros and Duñach, 2019). Even cells that already have terminally differentiated into the myogenic lineage may be reconverted to the osteoblast lineage under certain conditions: C2C12, a myoblast cell line, can be converted to osteoblasts under the combined effect of BMP and mechanical stimulation (Wei et al., 2020). Although these studies demonstrate that cells of other lineages are capable of osteogenic differentiation, it is not clear whether the same phenotype occurs in vivo, resulting in heterotopic ossification.

Beyond biological experiments, a significant influence of local loading on the formation of heterotopic ossification has been found through the mechanobiological algorithm system. By designing a computational model of physiology that takes into account both mechanical and biological factors, Rosenberg found that modifications to the mechanical environment significantly alter the shape and production of heterotopic bone. Adjustment of load orientation, skin material characteristics, and location of maximum trauma resulted in four characteristic HO types. Simulation of negative pressure dressings and tourniquet application also served to highlight the behavioral characteristics of HO (Rosenberg and Bull, 2018). Still, the mechanobiological algorithm system needs further development to make it more compatible with the real world.

These basic studies suggest that mechanical signals contribute to the formation and development of heterotopic ossification, not only initiate heterotopic ossification through the activation of pluripotent differentiation of MSCs, but also influence the osteogenic program during HO by affecting osteocytes, osteoblasts, and osteoclasts. However, there are only a few studies related to mechanical stimulation and heterotopic ossification. Representative basic studies have only applied fixed models for attenuated mechanical stimulation, but elaborate force-added models also need to be investigated. Relevant clinical studies are even more lacking. Further studies in this direction would have guiding values for the development of new drug targets for the treatment of HO, as well as for the development of more effective clinical methods of physical therapy and prophylaxis for HO.

In summary, the effects of mechanics on heterotopic ossification could be considered from early, middle, and late stages, respectively. In the early stage of HO, mechanical stimulation may activate pluripotent differentiation of MSCs in soft tissues, e.g., mTORC1 could activate quiescent stem cells into an “alert state”, and promote chondrogenesis and osteogenesis, leading to HO initiation. Mechanical stimulation can alter stem cell fate, causing chromatin regions around osteogenic genes to open. This results in more expression of osteogenic-related proteins and promotes stem cell differentiation toward osteogenesis. Clinically, early post-trauma immobilization can attenuate or even prevent heterotopic ossification. In the middle stage of HO, which means heterotopic ossification has been triggered and pathologic ossification is in the process of formation. Since physiological osteogenesis and pathological osteogenesis have some commonalities, they both require stem cells to differentiate into osteoblasts, and the eventual ossifications are dependent on the function of osteoblasts performing osteogenic functions. Many fundamental signaling pathway, such as CK2/HAUSP/RUNX2 are necessary for both physiologic bone formation and HO. It can be assumed that the effects of mechanics on HO may be similar to that on the osteogenesis process. From the clinical perspective, patients at this stage may still need as much bed rest as possible to avoid stress on the trauma site and to prevent pathological osteogenesis. Conversely, for the late stage, prolonged immobilization may instead lead to local tissue inflammation and hypoxia, both of which are risk factors for heterotopic ossification, and may lead to tissue ischemia and necrosis along with malfunctioning. Therefore, for patients potentially suffering from heterotopic ossification in the initial stages of injury, early and adequate immobilization is essential to avoid stress on the injured area. For those patients who have been adequately immobilized after trauma, appropriate rehabilitation exercises are recommended in the late stages to prevent heterotopic ossification as well as promote functional recovery.