In the 1970s, circRNAs, a unique closed circular form of non-coding RNA, were first discovered in plant viruses (1) and were considered a by-product of aberrant splicing reactions (2). In 2012, hundreds of circRNAs from different cell types were detected by RNA sequencing technology (3). With the development of high-throughput sequencing technology, several circRNAs have been detected in eukaryotes; however, they have been regarded as splicing by-products without biological functions (4). Unlike linear RNAs, which possess different 5′ and 3′ ends that indicate the start and stop sites of transcription, circRNAs are the product of one or two exons being spliced together via covalent bonding of the 3′ and 5′ ends, resulting in establishment of a covalently closed continuous loop (4–7). Therefore, circRNAs are unlikely to be degraded by RNA exonuclease and are more stable than their linear counterparts (7–9); circRNA functions may be linked to their unique stability (9). Notably, previous studies have indicated that circRNAs can sponge microRNAs (miRNAs/miRs) or functional proteins to regulate specific biological functions at the transcriptional or post-transcriptional level (6,9–11). Therefore, circRNAs may be considered promising diagnostic biomarkers and therapeutic targets for numerous diseases.

Osteonecrosis of the femoral head (ONFH) is a disabling clinical disease that is most common among young adults (12). The mean age at diagnosis is typically <50 years (13). If no early intervention is provided, ~80% of patients progress to femoral head collapse, hip dysfunction and permanent disability (14), and ultimately hip replacement becomes the only treatment option (13). However, postoperative infection, pain, functional rehabilitation, prosthetic replacement and other related complications (15–18) result in vast economic burdens for patients and for society. The mechanism by which ONFH develops remains unclear (19). The majority of experts agree that a lack of blood supply to the femoral head and bone marrow causes death of osteocytes, as explained by the microvascular injury theory, osteoporosis theory, apoptosis theory, osteogenic enhancement and adipogenic weakening theory of bone marrow mesenchymal stem cells (20–24). It has previously been reported that numerous circRNAs are associated with ONFH, which suggested that they may have a critical role in the development and progression of ONFH (25). This review aimed to describe the evidence regarding the biogenesis and biological functions of circRNAs, and to identify their potential mechanistic roles in ONFH.

The manner by which circRNAs are generated is different to how linear RNA transcripts are formed. Notably, the methods for generating circRNAs are either direct splicing or lasso-driven cyclization of precursor RNA, both with classical RNA polymerase-mediated splicing out of introns and connecting of exons via 5′ or 3′ polarities (6,7,9,10). These two methods can form exon circRNAs (exon sequence only, ecircRNAs), intron circRNAs (composed of intron sequences, ciRNAs) and exon-intron circRNAs (exon and intron sequences, EIciRNAs) (Fig. 1); ecircRNAs account for >80% of total circRNAs (26). The 3′ end of the upstream exon of the precursor mRNA is covalently linked to the 5′ end of the downstream exon by a reverse splicing process, after which the intron is cleaved to form an ecircRNA. If introns are retained, then EIciRNAs are formed (27). CiRNAs contain an exclusive 2′-5′ linkage that distinguishes them from ecircRNAs, and their structure relies on 7 nt GU-rich sequences near the 5′splice site and C-rich sequences that are 11 nt in length (28). Another novel model of circRNA biosynthesis has proposed that some RNA-binding proteins that have intron-binding abilities interact with intron-binding sites to promote the formation of circRNAs (29). Common RNA-binding proteins are protein quaking (29) and Drosophila muscle blind protein (30).

Weakened osteogenic differentiation and increased adipogenic differentiation of BMSCs are closely associated with the formation of ONFH (47). miRNAs, which can be negatively regulated by the competitive binding of circRNAs, serve an extremely important role in regulating stem cell differentiation (48). Differential expression of 23 miRNAs has been identified in the BMSCs of steroid-induced ONFH (SONFH) (49). Furthermore, various studies have demonstrated that promoting osteogenic differentiation and inhibiting adipogenic differentiation of BMSCs by regulating the expression level of miRNAs can achieve the goal of effectively treating ONFH (24,50,51).

A total of 231 preferentially expressed circRNAs were detected in the BMSCs of SONFH compared with the control group; 90 were upregulated and 141 were downregulated (25). CircRNAs were also differentially expressed during the osteogenic differentiation of BMSCs, 95% of which were protein-coding genes (52). In addition, silencing circRNA immunoglobulin superfamily member 11 promotes osteoblast differentiation and increases the expression of miR-199b-5p (52), which exerts its role in BMSC osteogenesis through the glycogen synthase kinase 3β/β-catenin signaling pathway (53) and acts as an important therapeutic targets during early-stage ONFH (54). A recent report indicated that circRNA FOXP1 acts as a sponge for several miRNAs and serves a pivotal role in the regulation of MSC differentiation (55). Additionally, Kuang et al (56) demonstrated that circRNA ubiquitin-specific protease 45 can upregulate the expression of phosphatase and tensin homolog through sponging miR-127-5p, which inhibits the protein kinase B pathway and regulates bone mass in rat SONFH. Therefore, circRNAs acting as miRNA sponges may be associated with osteogenic differentiation of BMSCs and ONFH. Although few studies have been performed on adipogenic differentiation, the aforementioned observations provide a good direction for studying circRNA, BMSCs and ONFH.

Osteoblasts are the primary cells that function in bone formation, which are essential for mineralization of bone matrix (57). Osteoclasts secrete proteinases and hydrogen ions to degrade the organic bone matrix and dissolve bone minerals, respectively (57). Maintaining a balance between osteoblasts and osteoclasts is crucial for maintaining normal bone mass (58). Bone cell metabolism is irreversibly destroyed by local circulatory disorders, and leads to the disappearance of osteoblasts, activation of osteoclasts, the eventual destruction of trabecular bone and increased bone fragility (59). This destructive pathway was verified by the observation that osteoclast-related activity is increased in the subchondral bone and necrotic areas of ONFH tissues, whereas osteoblast activity is increased in the sclerotic region (60).

circRNA 19142 (circ19142) and circRNA 5846 (circ5846) act as sponges for miR-7067-5p; this is a network that is confirmed to function in osteoblast differentiation through a circ19142/circ5846-targeted miRNA-mRNA axis (61). Dou et al (62) reported that 19 circRNAs were upregulated and five circRNAs were downregulated during different stages of mouse osteoclast formation. In addition, 260 differentially expressed circRNAs were detected between peripheral blood lymphocytes from postmenopausal patients with osteoporosis and controls (63). It has also been demonstrated that the expression of circRNA 0001275 is markedly increased in postmenopausal patients with osteoporosis; therefore, it was concluded that it may be a potential diagnostic biomarker (64). In summary, it was hypothesized that circRNAs may have an important role in ONFH, which is closely related to the activity of osteoblasts and osteoclasts. Therefore, circRNAs may have a role as diagnostic biomarkers. Identifying circRNAs that are preferentially expressed in peripheral blood lymphocytes, during necrotic collapse or in sclerotic bone tissue may provide a novel direction for studying the relationship between circRNAs and ONFH.

Endothelial cell damage, intravascular coagulation and disordered angiogenesis lead to ischemia, which is considered to be one of the core pathological results of ONFH (20,22). Endothelial cell tube formation is the first step in neovascularization (65), and blood vessels are critical in bone remodeling because they supply nutrients (66). It has been reported that SONFH may cause miRNA alterations in femoral head bone microvascular endothelial cells that are not seen in controls (67). The vascularity of the bone in osteonecrosis is reduced by ~50% (68). Additionally, a reduction in bone strength and vascularity of the bone precedes bone mass reduction and microstructural deterioration (69). Previous studies have suggested that ameliorating vascular endothelial cell proliferation, migration and tube formation may positively promote bone vascularization in the femoral head and prevent ONFH (70,71).

Notably, circRNAs are expressed in endothelial cells and serve a biological role in angiogenesis (72). For example, circRNA 0003575 is upregulated in oxidized low-density lipoprotein-induced human umbilical vein endothelial cells (HUVECs), and promotes HUVEC proliferation and angiogenesis (73). CircRNA 0010729 mediates vascular endothelial cell apoptosis and proliferation by targeting the miR-186/hypoxia inducible factor-1α axis (74). Furthermore, Shan et al (75) reported that circHIPK3 expression is substantially upregulated in endothelial cells during diabetic retinal vasculopathy, and in vitro endothelial cell viability, proliferation, migration, and tube formation are altered by silencing or overexpressing circHIPK3. Acting as an endogenous miR-30a-3p sponge, in vivo silencing of circHIPK3 also attenuates retinal vascular dysfunction (75). Therefore, it was hypothesized that circRNAs may be involved in the endothelial cell damage and disruption of angiogenesis that lead to the formation of ONFH; circRNAs may be considered potential therapeutic targets.

It is generally believed that structural damage of the subchondral bone in post-collapse cases of osteonecrosis contributes to degeneration of articular cartilage (76). Notably, at the beginning of ONFH, the subchondral bone receives reduced nutrition because the blood supply to the area is impaired; that, coupled with degeneration of the cartilage matrix, leads to degeneration of articular cartilage, which can increase instability of the hip joint and accelerate the development of ONFH (77,78). Therefore, prevention and early treatment of hip cartilage damage are beneficial for ameliorating the progression of ONFH (78).

Cartilage inflammation can be enhanced by interleukin-21, which causes degradation of the cartilage in patients with ONFH through the Janus kinase/signal transducers and activators of transcription signaling pathway (79). However, studies on the role of circRNAs in cartilage degeneration are currently more focused on OA (80,81). In addition, it has been reported that intermittent cyclic mechanical tension leads to differential expression of miRNAs, which regulates stromal metabolism and calcification of cartilage endplate chondrocytes via the transforming growth factor-β signaling pathway (82). Furthermore, differentially expressed circRNAs are detected in different regions of cartilage in patients with OA; circRNA 100226 is associated with mechanical tension, and it was demonstrated that circRNA 100226, acting as a sponge for miR-875, promotes the degradation of cartilage matrix by controlling the expression of tumor necrosis factor α (83). Therefore, it was concluded that mechanical stress changes in the early stages and in the collapsed areas of ONFH may lead to alterations in the expression of miRNAs and circRNAs in the corresponding regional cartilage. The molecular mechanism underlying the function of circRNAs in the degradation of ONFH cartilage damage is essential for understanding the pathogenesis of ONFH and exploring novel prospective therapeutic targets.

CircRNAs were discovered decades ago. With advances in research methods, increasing attention has been paid to circRNAs. Notably, it has been reported that circRNAs are involved in the development and progression of various diseases (6,7,46). Although differential expression of circRNAs was observed in BMSCs of ONFH and a possible connection was suggested (25), to the best of our knowledge, no further functional or molecular mechanisms have been investigated. With respect to the potential pathological mechanisms of ONFH, this review describes the potential relationship between BMSCs, osteoblasts, osteoclasts, vascular endothelial cells, articular cartilage and circRNAs (Fig. 2). Addressing these problems may be beneficial to the prevention of ONFH and the development of therapeutic targets for its treatment.

CircRNAs can function as miRNA sponges, regulate gene transcription and interact with RNA-binding proteins (10,11,28,29). Most circRNA studies focus on their role as miRNA sponges, that is, the related mechanisms and functions of the circRNA-miRNA-mRNA axis in human diseases (84). However, a single circRNA has numerous binding sites for one miRNA and can interact with multiple miRNAs; therefore, more extensive research is required to illuminate the functional interactions between circRNAs and miRNAs. In addition, the role of circRNAs in regulating protein translation and binding to RNA-binding proteins in various diseases requires additional research.

In conclusion, circRNAs have a potential role in the initiation mechanisms of ONFH, and they exhibit high stability compared with linear RNAs. Therefore, it may be hypothesized that circRNAs serve a unique role in the formation ONFH and could have a role in ONFH therapy.