Oxidative stress-induced circSOD2 inhibits osteogenesis through sponging miR-29b in metabolic-associated fatty liver disease

Abstract

BACKGROUND

Metabolic-associated fatty liver disease (MAFLD) is characterized by lipid accumulation in hepatocytes and is closely associated with oxidative stress. Increasing clinical evidence indicates that MAFLD is linked to bone metabolic disorders, including osteoporosis. Recent studies indicate that the expression profiles of liver circular RNAs (circRNAs) are altered in MAFLD. However, the effects of these changes on bone metabolism remain poorly understood.

AIM

To investigate the effects and mechanism of differently expressed circRNAs secreted by the liver on osteogenic differentiation in MAFLD.

METHODS

RNA sequencing was performed to identify highly expressed circRNAs in the liver, validated by quantitative real-time reverse transcription polymerase chain reaction, and localized using fluorescence in situ hybridization (FISH). A mouse model induced by a high-fat diet was used to simulate MAFLD.

RESULTS

CircSOD2 was significantly upregulated in liver tissues and primary hepatocytes from subjects with MAFLD. CircSOD2 was induced by oxidative stress and attenuated by antioxidants in the mouse model. In addition, circSOD2 was delivered from hepatocytes to bone marrow mesenchymal stem cells (BMSCs). Furthermore, circSOD2 inhibited the osteogenic differentiation of BMSCs and in vivo bone formation by sponging miR-29b. Moreover, miR-29b inhibition reversed the stimulatory effect of circSOD2 silencing on osteogenic differentiation of BMSCs and in vivo bone formation. Mechanistically, the interaction between circSOD2 and miR-29b confirmed through a luciferase reporter assay and the co-localization in the cytoplasm evidenced by FISH indicated that circSOD2 acted as a sponge for miR-29b.

CONCLUSION

This study provides a novel mechanism underlying the liver-bone crosstalk, demonstrating that circSOD2 upregulation in hepatocytes, induced by oxidative stress, inhibits osteogenic differentiation of BMSCs by sponging miR-29b. These findings offer a better understanding of the relationship between MAFLD and osteoporosis.

Key Words: Circular RNAs; MicroRNAs; Oxidative stress; Osteogenesis; Non-alcoholic fatty liver disease; Osteoporosis

Core Tip: Emerging evidence indicates that metabolic-associated fatty liver disease (MAFLD) closely associated with oxidative stress, is linked to bone metabolic diseases. Altered expression profiles of liver circular RNAs (circRNAs) have been observed in MAFLD. This study explores the upstream and downstream mechanisms of circSOD2 upregulation in osteogenesis by identifying differentially expressed circRNAs in MAFLD livers. Results demonstrated that circSOD2 upregulation in hepatocytes induced by oxidative stress inhibits osteogenic differentiation of bone marrow mesenchymal stem cells by sponging miR-29b. These findings provide a better understanding of the relationship between MAFLD and osteoporosis from the perspective of circRNAs.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is characterized by the accumulation of lipids in hepatocytes and is closely linked with insulin resistance, dysregulated lipid metabolism, inflammation, and oxidative stress[1]. Increasing evidence indicates that NAFLD alters the secretion of endocrine factors from the liver, including cytokines, hepatokines, and inflammatory mediators, which affects metabolic processes in extrahepatic organs and tissues[2]. Recently, experts have advocated for the term metabolic-associated fatty liver disease (MAFLD) as a more accurate descriptor of this pathological condition[3]. MAFLD encompasses a broad spectrum of conditions, from simple steatosis to non-alcoholic steatohepatitis, fibrosis, cirrhosis, and hepatocellular carcinoma. It has become the most common cause of chronic liver disease globally, affecting over 25% of the population[4]. The substantial public health and economic burden of MAFLD have spurred intense research interest in its pathogenesis and impact.

Over the past decades, MAFLD has been linked to an increased risk of cardiovascular diseases, diabetes, and bone metabolic diseases[5]. Maintaining the metabolic balance between bone formation and absorption is critical for preserving bone health and bone mass[6]. Osteogenic differentiation of mesenchymal stem cells (MSCs), an important stage of bone formation, is precisely regulated by growth factors, hormones, and cytokines[7]. Disruption of this balance, such as through impaired osteoblast differentiation, can lead to bone metabolic diseases. Osteoporosis is characterized by low bone mineral density (BMD) and deterioration of bone microarchitecture, resulting in increased bone fragility and fracture risk[8]. The relationship between MAFLD and low BMD or osteoporosis remains controversial despite growing clinical research. For instance, a large-scale cross-sectional study by Li et al[9] reported a significantly higher prevalence of osteoporotic fractures in middle-aged and older adult men with MAFLD in China. Conversely, a meta-analysis of five cross-sectional studies with significant heterogeneity found no clear association between BMD and MAFLD[10]. However, two recent cohort studies with large sample sizes identified a positive association between MAFLD and both low BMD and osteoporosis[11,12]. Additionally, two updated meta-analyses revealed that MAFLD is linked to low BMD and an increased risk of osteoporosis or osteoporotic fractures irrespective of sex[13,14]. Although there is growing clinical evidence implicating MAFLD in the pathogenesis of osteoporosis, basic research on the underlying mechanisms remains limited.

Circular RNAs (circRNAs) are a novel class of noncoding RNAs generated through back-splicing of precursor mRNAs and are characterized by covalently closed continuous loops without a terminal 5'cap and 3′polyadenylate tail[15]. circRNAs are highly stable, conserved, and tissue-specific[16]. They regulate various cellular processes, including cell proliferation and differentiation, through diverse mechanisms[17]. Recent studies have highlighted the regulatory roles of circRNAsin the osteogenic differentiation of MSCs[18] and secondary osteoporosis[19]. Additionally, there is growing evidence suggesting that circRNAs are abundantly expressed in the liver, and their altered expression is correlated with the severity of MAFLD[20,21]. To the best of our knowledge, no studies have investigated the effects of circRNA alteration induced by MAFLD on bone metabolism.

Thus, we hypothesized that circRNAs secreted by hepatocytes via endocrine pathways mediate the effects of MAFLD on bone metabolism. This study aimed to investigate the impact and mechanisms of differentially expressed circRNAs in hepatocytes on the osteogenic differentiation of MSCs in the context of MAFLD, thereby elucidating the relationship between MAFLD and osteoporosis.

MATERIALS AND METHODS

MAFLD subjects and sample collection

This study was conducted following protocols approved by the Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine. Written informed consent was obtained from all adult participants. Surgically resected liver tissues and serum samples were collected from patients undergoing hepatic hemangioma surgery at the hospital. Patients diagnosed with MAFLD were categorized as the MAFLD group (n = 5), while those with simple hepatic hemangioma were assigned to the control group (n = 5). Liver histology was independently evaluated by two blinded pathologists according to the 2018 Practice Guidance of the American Association for the Study of Liver Diseases. Other liver disease etiologies, including viral, autoimmune, cholestatic, genetic, alcoholic, and drug-induced diseases, were excluded.

High-fat diet-induced MAFLD mouse model

All animal experiments were approved by the Institutional Animal Care and Use Committee of the Institute of Health Sciences. Healthy 8-week-old male C57BL/J6 mice were purchased from the Laboratory Animal Center of Zhejiang University. The mice were maintained under a standard 12 hours/12 hours light/dark cycle and randomly allocated into two groups. The MAFLD group was fed with high-fat diet (HFD, 60% kcal from fat, D12492, Research Diet, United States) for 24 weeks, while the control group was fed a normal diet. The investigators were blinded to the group allocation.

RNA sequencing

Total RNAs were extracted using the TRIzol reagent. After isolation and purification, approximately 5 μg of total RNA was subjected to ribosomal RNA depletion using a Ribo-Zero™ rRNA Removal Kit (Illumina Inc.) and subsequently transcribed into fluorescent cRNA using the Arraystar Super RNA Labeling Kit according to the manufacturer’s instructions. The labeled cirRNAs were hybridized onto Arraystar Human circRNA Arrays V2 and incubated at 65 °C for 17 hours in an Agilent Hybridization Oven. The hybridized arrays were washed and scanned using an Agilent Scanner G2505C. Data extraction was performed with Agilent Feature Extraction software, and data processing was performed using the R software limma package. In pairwise comparisons, circRNAs with P < 0.01 and an absolute fold change > 2 were classified as differentially expressed. Validation of candidate circRNA expression was conducted via quantitative real-time polymerase chain reaction (qRT-PCR).

Cell culture and osteogenic induction

Primary hepatocytes and bone marrow mesenchymal stem cells (BMSCs) were cultured in RPMI1640 and Dulbecco's Modified Eagle Medium (DMEM), respectively, both supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were incubated at 37 °C, 5% CO₂, and 95% humidity. For co-culture, BMSCs were seeded in the upper chamber, and hepatocytes were seeded in the lower chamber of a 24-well transwell plate. For osteogenic induction, BMSCs were treated with osteogenic medium (DMEM medium supplemented with 0.1 μM dexamethasone, 50 mM ascorbic acid, and 10 mM β-glycerophosphate) after cells reached 70%-80% confluence (Chemicals and reagents were listed in the Supplementary material).

Cell transfection

The small interfering RNA targeting circSOD2 (si-circSOD2) and adenoviral circSOD2 (Ad-circSOD2) were commercially constructed by RiboBio (Guangzhou, China). Corresponding negative controls (si-NC and Ad-NC) were included. The cells were transfected with si-circSOD2 and Ad-circSOD2 in Opti-MEM™ I reduced serum medium using Lipofectamine™ 3000 (Invitrogen) according to the manufacturer’s instructions.

Alkaline phosphatase staining and alkaline phosphatase activity assay

After osteogenic induction for 7 days, alkaline phosphatase (ALP) staining was performed using a commercial kit, according to the manufacturer’s instructions. For ALP activity assay, cells were lysed with 0.5% Triton X-100 for 1 hour at 4 °C. The lysates were centrifuged, and ALP activity in the supernatant was measured colorimetrically using an ALP assay kit (Jiancheng, Nanjing, China). Total protein was quantified using the BCA Protein Assay Kit (Thermo Fisher Scientific Inc.). ALP activity was normalized to the total protein and expressed as nmol/minute/mg protein.

Alizarin red staining and quantitative detection

After osteogenic induction for 14 days, BMSC mineralization was evaluated using Alizarin Red staining. The cells were fixed with 4% paraformaldehyde for 30 minutes and then stained with 40 mM alizarin red S (pH 4.0) for 15 minutes at room temperature. After rinsing with distilled water to completely remove the unbound stain, the cells were visualized and imaged using a light microscope and digital camera. After drying, staining was eluted with 10% hexadecylpyridinium chloride for 15 minutes, and absorbance was measured at 560 nm (OD560).

Hematoxylin & eosin staining and Oil Red O staining

Liver tissues were clipped into appropriate size and fixed in 4% paraformaldehyde. After dehydration and transparency, the samples were embedded in paraffin wax and sliced into 5 μm sections. The sections were dewaxed and stained with hematoxylin & eosin (H&E). Oil Red O staining was performed using an Oil Red O dye kit (Beyotime, Shanghai, China) according to the manufacturer's protocol.

Measurement of intracellular reactive oxygen species

Intracellular reactive oxygen species (ROS) levels were measured using a ROS assay kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions. Hepatocytes were incubated in 2 mL of serum-free DMEM supplemented with 10 μM 2,7-Dichlorodihydrofluorescein diacetate (DCFH-DA) at 37 °C for 20 minutes. After removal of the DCFH-DA-containing medium, the hepatocytes were detached with trypsin, centrifuged, and resuspended. Fluorescence levels were quantified using flow cytometry at excitation and emission wavelengths of 488 and 525 nm, respectively.

RNA isolation, cDNA synthesis, and qRT-PCR

Total RNA was extracted using TRIzol reagent (Invitrogen) following the manufacturer's protocol, and RNA concentration was determined using a spectrophotometer (Nanodrop, Thermo Scientific). Complementary DNA (cDNA) was synthesized with PrimeScript RT Master Mix (Takara Bio, Otsu, Japan). qRT-PCR was performed on an ABI Prism 7500 system (Applied Biosystems, Foster City, CA, United States) using SYBR Green QPCR Master Mix (Takara Bio). The 10 μL reaction system included 2 μL cDNA and 5 μL of SYBR green (Takara Bio). Cycling conditions were as follows: 40 cycles of denaturation at 95 °C for 5 seconds and amplification at 60 °C for 24 seconds. For circRNAs, total RNAs were incubated with or without 3 U/μg of RNase R (Epicentre, San Diego, CA, United States) at 37 °C for 20 minutes. Then, RNAs were subsequently purified using the RNeasy MinElute Cleanup Kit (Qiagen). All reactions were performed in triplicate and normalized to GAPDH or U6. Relative expression level was calculated using the 2^-ΔΔCtmethod (Primers were listed in the Supplementary material).

Fluorescence in situ hybridization assay

Cells were fixed with 4% paraformaldehyde for 15 minutes and permeabilized with 0.5% Triton X-100 for 5 minutes at room temperature. Fluorescence-labeled probes specific for circSOD2 were incubated with cells overnight at 37 °C. FAM-labeledcircSOD2probes and Cy3-labeled miR-29bprobes were designed and synthesized by RiboBio (Guangzhou, China). Signal detection was performed using a fluorescence in situ hybridization (FISH) Kit (RiboBio, Guangzhou, China) according to the manufacturer’s instructions. The images were acquired on a Nikon A1Si Laser Scanning Confocal Microscope (Nikon Instruments, Inc., Tokyo, Japan).

RNA immunoprecipitation

RNA immunoprecipitation (RIP) was performed using a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, United States). HEK-293T cells were transfected with argonaute-2 (AGO2) plasmid or empty vector (HanBio, Shanghai, China) using PEI transfection reagent (Sigma-Aldrich, St. Louis, MO, United States). After 48 hours, approximately 1 × 10⁷ cells were pelleted by centrifugation and then resuspended in 100 μL of RIP Lysis Buffer containing RNase inhibitors and protease inhibitors. The cell lysates were incubated with rabbit anti-IgG or anti-AGO2 (Millipore, Boston, MA, United States) antibodies overnight at 4 °C. The lysates were then treated with proteinase K buffer, and the immunoprecipitated RNA was extracted using the RNeasy MinElute Cleanup kit (Qiagen), followed by reverse transcription using the PrimeScript RT Master Mix (TaKaRa, Tokyo, Japan). CircSOD2 Levels were detected by RT-qPCR.

RNA antisense purification

RAP experiments were performed using an RNA antisense purification (RAP) technology kit (BersinBio, Guangzhou, China) according to the manufacturer's protocol. The hepatocytes were collected, lysed, and sonicated. C-1 magnetic beads were incubated with circSOD2 probes at 25 °C for 2 hours to create probe-coated beads. These probe-coated beads were left in contact with the cell lysates to pull down circSOD2 at 4 °C overnight. RNA complexes with the beads were extracted using the RNeasy Mini kit (QIAGEN, Dusseldorf, Germany) and analyzed by qRT-PCR.

Dual luciferase reporter assay

HEK-293T cells were seeded onto 96-well plates and cultured to reach 60%-70% confluence before transfection. Luciferase reporter plasmids and microRNAs (miRNAs) mimics were generated by GeneChem (Shanghai, China). Wild-type or mutant circSOD2 and miR-29b fragments were inserted into the Xba1 restriction site of the firefly luciferase-Renilla luciferase vector (hFLuc-XbaL-hRLuc) to generate plasmids (Luc-circSOD2 WT or Luc-circSOD2 Mut) and (Luc-miR-29b WT or Luc-miR-29b Mut). The cells were co-transfected with these luciferase reporter plasmids and different miRNA mimics using PEI transfection reagents (Sigma-Aldrich, St. Louis, MO, United States). After 48 hours, luciferase activity was evaluated using a Luciferase Assay Reagent (Yeason, Shanghai, China). Relative luciferase activity was determined as the ratio of firefly to Renilla luciferase activity.

Rat calvarial critical size defect model

Six-week-old Sprague Dawley rats (250 ± 50 g) were randomly distributed into different groups: Blank (Con), overexpression (AD-circSOD2), overexpression control (AD-NC), knockdown (Sh-circSOD2), knockdown control (Sh-NC) rescue (Sh-circSOD2 + antagomiR-29b), and rescue control (Sh-circSOD2 + antagomiR-NC). After anesthesia, right-side full-thickness critical-size calvarial defects (5 mm diameter) were created using a fine surgical burr under copious sterile saline irrigation. BMSCs were infected with an adenovirus to stably overexpress circSOD2 or lentivirus to stably knockdown circSOD2 and then incubated in an osteogenic medium for 7 days. Differentiated BMSCs (1 × 10⁶ cells/mL) were seeded onto collagen I hydrogel. The rat calvarial defects were injected with 8 mL collagen I hydrogel with differentiated BMSCs. After 8 weeks, the animals were euthanized using CO₂ asphyxiation followed by cervical dislocation. The calvarial samples were fixed in paraformaldehyde and scanned using a high-resolution μCT (Skyscan, Aartselaar, Belgium), and in vivo bone formation was analyzed via histomorphometry.

Statistical analysis

The data are presented as mean ± SD. Statistical analyses were performed using Student’s t-test unless otherwise specified. A P value < 0.05 was considered statistically significant. Each experiment was independently repeated at least three times.

RESULTS

circSOD2 is upregulated in MAFLD

To establish a circRNA profiling database, we first performed RNA sequencing on liver tissues from three patients with MAFLD and three control subjects. A heat map revealed 42 differentially expressed circRNAs between the MAFLD and control samples, with 31 circRNAs upregulated and 11 downregulated (Figure 1A). Among the top 10 upregulated circRNAs confirmed by qRT-PCR, circSOD2 (hsa_circ_0004662) shown the highest upregulation in MAFLD liver samples (Figure 1B). Additionally, circSOD2 was highly expressed in primary hepatocytes and serum from MAFLD subjects compared to controls (Figure 1C and D). Thus, circSOD2 was selected as our target circRNA.

Figure 1 circSOD2 is upregulated in metabolic-associated fatty liver disease. A: Heat map depicting differentially expressed circRNAs in the liver from three clinical subjects with metabolic-associated fatty liver disease (MAFLD) and three control subjects after RNA sequencing; B: Validation of the top 10 upregulated circRNAs in MAFLD livers, including circSOD2 (hsa_circ_0004662), confirmed using quantitative real-time reverse transcription polymerase chain reaction (n = 5); C and D: CircSOD2 was highly expressed in primary hepatocytes and serum subjects with MAFLD (n = 5); E: RNase R digestion assay showing that circSOD2 was more resistant to degradation by RNase R; F: Fluorescence in situ hybridization assay showing that circSOD2 was predominantly located in the cytoplasm. ^aP < 0.05; ^bP < 0.01; ^cP < 0.001; NS: Not significant; Con: Control; NAFLD: Nonalcoholic fatty liver disease.

Previous studies have demonstrated that circSOD2 is derived from SOD2 pre-mRNA and contains three exons[22]. Compared to linear RNA, circSOD2 exhibited greater resistance to degradation by RNase R (Figure 1E), indicating enhanced stability. Moreover, we observed that circSOD2 was predominantly located in the cytoplasm according to the FISH assay (Figure 1F).

circSOD2 is induced by oxidative stress

Thereafter, we explored the reason why circSOD2 expression was upregulated in NAFLD that is closely associated with increased oxidative stress. We examined the level of lipid peroxidation metabolite malondialdehyde (MDA) and antioxidant SOD2. As expected, MDA concentration was increased, whereas SOD2 concentration was decreased in the liver of NAFLD subjects (Figure 2A and B). A mouse HFD model was successfully established to simulate MAFLD, as evidenced by H&E and Oil red O staining (Figure 2C and D). Similarly, we found that MDA was upregulated and SOD2 was downregulated in the livers of HFD mice. However, dietary Vitamin C (Vit C) supplementation reversed these changes (Figure 2E and F). In alignment with the patient data, circSOD2 expression was significantly higher in the liver of HFD mice compared to normal diet-fed mice. Notably, Vit C supplementation reduced circSOD2 expression in the livers of HFD mice (Figure 2G).

Figure 2 circSOD2 is induced by oxidative stress. A and B: Levels of malondialdehyde (MDA), a lipid peroxidation metabolite, and antioxidant SOD2 in livers of metabolic-associated fatty liver disease (MAFLD) and control subjects (n = 5); C and D: Representative hematoxylin & eosin and Oil red O staining images confirm the successful establishment of a high-fat diet (HFD) mice model to simulate MAFLD; normal diet mice served as negative control (Con); E-G: MDA, SOD2, and circSOD2 Levels in the livers of Con, HFD, and HFD supplemented with Vitamin C groups (n = 5); H and I: Reactive oxygen species and circSOD2 Levels in primary hepatocytes from Con, HFD, and HFD were pretreated with N-acetyl L-cysteine groups (n = 5). ^aP < 0.05; ^bP < 0.01; NS: Not significant; Con: Control; NAFLD: Nonalcoholic fatty liver disease; MDA: Malondialdehyde; HFD: High-fat diet; VitC: Vitamin C; NAC: N-acetyl L-cysteine.

Intracellular ROS levels were also assessed in primary mouse hepatocytes. We found that ROS levels in HFD hepatocytes were increased compared to that in control hepatocytes, while antioxidant N-acetyl L-cysteine (NAC) pretreatment abrogated the increase of ROS levels in HFD hepatocytes (Figure 2H). Correspondingly, circSOD2 expression in HFD hepatocytes was also increased compared to that in control hepatocytes, while NAC pretreatment abrogated the increase of circSOD2 expression in HFD hepatocytes (Figure 2I). Collectively, these findings suggest that circSOD2 expression is induced by oxidative stress and attenuated by antioxidants.

circSOD2 is delivered from hepatocytes to BMSCs

circSOD2 expression was analyzed across various organs and tissues, revealing its predominant expression in the liver, serum, and bone marrow of HFD mice (Figure 3A). CircSOD2 expression in bone marrow and BMSCs of HFD mice was significantly higher than that in bone marrow and BMSCs of normal diet mice. However, supplements with Vit C abrogated the increase of circSOD2 expression in bone marrow and BMSCs of HFD mice (Figure 3B and C). To investigate whether hepatocytes deliver circSOD2 to BMSCs, circSOD2 expression was measured in BMSCs co-cultured with hepatocytes. BMSCs co-cultured with HFD hepatocytes exhibited elevated circSOD2 Levels compared to control hepatocytes. However, HFD hepatocytes pretreated with NAC abrogated the increase of circSOD2 expression in co-culturedBMSCs (Figure 3D).

Figure 3 circSOD2 is delivered from hepatocytes to bone marrow mesenchymal stem cells. A: Expression pattern of circSOD2 in various organs and tissues of high-fat diet (HFD) mice (n = 5); B and C: Expression levels of circSOD2 in bone marrow and bone marrow mesenchymal stem cells (BMSCs) of normal diet mice (Con), HFD mice, and HFD mice supplemented with Vitamin C (n = 5); D: Expression levels of circSOD2 in mice BMSCs after cocultured with hepatocytes from Con, HFD, and HFD pretreated with N-acetyl L-cysteine groups (n = 5). ^aP < 0.05; ^bP < 0.01; Con: Control; MDA: Malondialdehyde; HFD: High-fat diet; VitC: Vitamin C; NAC: N-acetyl L-cysteine.

circSOD2 inhibits osteogenesis

Given the upregulation of circSOD2 in BMSCs from HFD mice, we investigated its role in regulating osteogenic differentiation. BMSCs were incubated in an osteogenic medium following transfection with AD-circSOD2 or si-circSOD2. We found that osteogenic markers, including ALP, runt-related transcription factor 2 (RUNX2), osteocalcin (OCN), and collagen type 1 alpha 1 (COL1A1), were suppressed by circSOD2 overexpression and enhanced by circSOD2 knockdown during osteogenic differentiation of BMSCs (Figure 4A-D). Similarly, ALP activity, as evidenced by ALP staining and calcium deposition, assessed by Alizarin Red staining, followed the same trend (Figure 4E-G). Furthermore, a rat model of calvarial critical-size defects was used to confirm the effect of circSOD2 on osteogenesis. Micro-computed tomography (micro-CT) revealed that circSOD2 overexpression inhibited, while its silencing enhanced in vivo bone formation in rats (Figure 4H). Further analysis showed that calvarial trabecular bone volume per tissue volume (BV/TV) and BMD decreased in response to circSOD2 overexpression and increased in response to circSOD2 knockdown (Figure 4I and J).

Figure 4 circSOD2 inhibits osteogenesis. A-D: MRNA expression of osteogenic markers, including alkaline phosphatase (ALP), runt-related transcription factor 2, osteocalcin, and collagen type 1 alpha 1 in bone marrow mesenchymal stem cells (BMSCs) incubated with osteogenic medium for 7 days after transfection of AD-circSOD2 or si-circSOD2; E: ALP staining at 7 days and Alizarin Red staining at 14 days, demonstrating ALP activity and calcium deposition, respectively; F and G: Quantification of ALP activity and calcium deposition; H: Representative micro-computed tomography imaging of bone formation in rat calvarial critical-size defect mode at 8 weeks. Rats (n = 25) were randomly assigned to five groups: Blank (Con), overexpression (AD-circSOD2), overexpression control (AD-NC), knockdown (Sh-circSOD2), knockdown control (Sh-NC); I: Trabecular bone volume per tissue volume from histomorphometric analysis in the calvarial model (n = 5); J: Quantitative analysis of bone mineral density in the calvarial rat model (n = 5). ^aP < 0.05; ^bP < 0.01; ALP: Alkaline phosphatase; RUNX2: Runt-related transcription factor 2; OCN: Osteocalcin; COL1A1: Collagen type 1 alpha 1; Con: Control; BV/TV: Bone volume per tissue volume.

circSOD2 serves as a sponge for miR-29b

Next, we investigated whether circSOD2 could bind to miRNAs. We observed that circSOD2 was notably enriched in RIP with an antibody against argonaute 2 (AGO2) that binds miRNAs (Figure 5A), indicating that circSOD2 may function as a miRNA sponge. By cross-referencing the TargetScan, RNAbybrid and miRanda databases, we identified 12 candidate miRNAs as potential targets of circSOD2 (Figure 5B). An antisense RNA probe specific for circSOD2 was designed and validated (Figure 5C). Three miRNAs were identified to be associated with circSOD2 using RAP (Figure 5D). Among these, only miR-29b mimic showed a significant reduction in relative luciferase activity, indicating miR-29b binding to circSOD2 (Figure 5E). Therefore, miR-29b was selected for further analysis. Subsequently, the dual-luciferase reporter assay showed that transfection of miR-29b mimic led to a significant reduction in relative luciferase activity of wild-type circSOD2 (Figure 5F). Moreover, FISH assay indicated the co-localization of circSOD2 and miR-29bin the cytoplasm (Figure 5G). In contrast with circSOD2 expression, miR-29b expression was significantly decreased in BMSCs from HFD mice compared to controls (Figure 5H). Collectively, these findings suggest that circSOD2 serves as a sponge for the downstream target miR-29b.

Figure 5 circSOD2 sponges miR-29b. A: Reverse transcription-quantitative polymerase chain reaction (qRT-PCR) analysis of argonaute-2-bound circSOD2 using RNA immunoprecipitation; B: Venn diagram illustrating 12 potential target miRNAs of circSOD2 through cross-referencing TargetScan, RNAbybrid, and miRanda databases; C: Quantification of the efficiency of the circSOD2 probe for RNA antisense purification analysis using qRT-PCR; D: Quantification of the circSOD2-bound miRNAs using qRT-PCR; E: Relative luciferase activity of circSOD2 Luciferase reporter plasmids co-transfected with candidate miRNA mimics in HEK-293T cells; F: Quantification of relative luciferase activity using dual luciferase reporter assay after co-transfection of Luc-circSOD2 WT or Mut with miR-29b mimic or negative control in HEK-293T cells; G: Fluorescence in situ hybridization assay images demonstrating co-localization of circSOD2 and miR-29b in the cytoplasm; H: Decreased expression of miR-29b in bone marrow mesenchymal stem cells from high-fat diet mice (n = 5). ^aP < 0.05; ^bP < 0.01; NS: Not significant; NC: Negative control; AGO2: Analysis of argonaute-2; RIP: RNA immunoprecipitation; WT: Wildtype; Mut: Mutation; Con: Control.

circSOD2 inhibits osteogenesis through miR-29b

Given the pivotal role of miR-29b, a downstream target of circSOD2, in osteogenic differentiation[23] and bone formation[24], we hypothesized that circSOD2 exerts its inhibitory effects on osteogenic differentiation and bone formation through miR-29b. To test the hypothesis, BMSCs were incubated in an osteogenic medium after co-transfection with si-circSOD2 with or without miR-29b inhibitor. CircSOD2 knockdown significantly stimulated osteogenic markers, including ALP, RUNX2, OCN, and COL1A1, however, miR-29b silencing reversed the stimulatory effect of circSOD2 knockdown on these markers (Figure 6A-D). Likewise, the change pattern of ALP activity and calcium deposition was consistent with osteogenic markers in response to miR-29b inhibition (Figure 6E and F). Furthermore, micro-CT demonstrated that miR-29b silencing abrogated the stimulatory effect of circSOD2 knockdown on in vivo bone formation in rats (Figure 6G). In detail, miR-29b silencing abrogated the stimulatory effect of circSOD2 knockdown on BV/TV and BMD (Figure 6H and I). Taken together, these findings suggest that circSOD2 inhibits osteogenesis via miR-29b.

Figure 6 miR-29b inhibition reversed the effect of circSOD2 silencing on osteogenic differentiation of bone marrow mesenchymal stem cells. A-F: The mRNA expression of osteogenic markers, including alkaline phosphatase (ALP) (A), Runt-related transcription factor 2 (B), osteocalcin (C), and collagen type 1 alpha 1 (D), and ALP activity (E), calcium deposition (F) in bone marrow mesenchymal stem cells incubated with osteogenic medium for 7 days after cotransfection of si-circSOD2 with or without miR-29b inhibitor; G: Representative micro-computed tomography imaging of bone formation at 8 weeks in a rat calvarial critical size defect model. Rats (n = 20) were randomly assigned to four groups: Blank (Con), knockdown (Sh-circSOD2), rescue (Sh-circSOD2 + antagomiR-29b), and rescue control (Sh-circSOD2 + antagomiR-negative control); H: Trabecular bone volume per tissue volume in the rat calvarial model quantified via histomorphometric analysis (n = 5); I: Quantitative analysis of calvarial bone mineral density in the rat model (n = 5). Data indicate the mean ± SD from at least three independent experiments. Representative images are from three independent experiments. ^aP < 0.05; ^bP < 0.01; BMD: Bone mineral density; NC: Negative control; ALP: Alkaline phosphatase; RUNX2: Runt-related transcription factor 2; OCN: Osteocalcin; COL1A1: Collagen type 1 alpha 1; BMSCs: Bone marrow mesenchymal stem cells; BV/TV: Bone volume per tissue volume; Con: Control.

DISCUSSION

The liver, as the central organ of metabolism, communicates with other organs through endocrine signaling in the whole body. The crosstalk between the liver with other tissues such as adipose tissue, heart, and brain has been extensively studied[25,26]. However, the interaction between the liver and bone remains poorly understood. To our knowledge, this study is the first to explore the liver-bone crosstalk from the perspective of circRNAs. We screened and verified differentially expressed circRNAs in MAFLD livers, focusing on the upstream and downstream mechanisms of circSOD2 upregulation during osteogenesis. Our finding revealed that circSOD2 upregulation was induced by oxidative stress in hepatocytes. Moreover, circSOD2 was delivered from hepatocytes to BMSCs via blood circulation. Furthermore, circSOD2 inhibited osteogenic differentiation of BMSCs and bone formation by sponging miR-29b.

Dysregulated circRNAs have been implicated in various diseases, including cancer and MAFLD[27]. Due to high stability and easy detection, they are increasingly investigated as promising disease biomarkers and therapeutic targets[28]. Recent studies have revealed alterations in circRNA expression in patients and animal models of MAFLD, primarily focusing on their role in MAFLD[29-32]. However, their impact on the metabolism of extrahepatic tissues, including bone, remains largely unexplored.

CircSOD2, with a mature spliced sequence of 462 bp, is located on the reverse strand of human chromosome 6 and is ubiquitously expressed in many tissues and cells. Recent evidence has shown thatcircSOD2 was highly expressed in smooth muscle cells[22], M1 macrophages[33], osteoarthritic cartilage[34], and hepatocellular carcinoma tissues[35]. Consistent with these findings, our results showed that circSOD2 was highly expressed in hepatocytes, liver, serum, and bone marrow. Notably, circSOD2 expression was upregulated in the liver and serum of patients with MAFLD and HFD mice compared to controls. Given its stable expression and detectability in blood, our results suggest that circSOD2 may represent a promising and noninvasive biomarker for MAFLD, offering a viable alternative to puncture liver biopsy.

Extensive studies have shown that oxidative stress is a pivotal factor in the pathophysiology of MAFLD. Chronic inflammation and lipid metabolism disorders are closely related to an imbalance between oxidants and antioxidants, leading to lipid peroxidation and mitochondrial dysfunction in MAFLD[36]. Our findings revealed increased MDA levels and decreased SOD2 Levels in patients with MAFLD and HFD mice. Further investigation into the upstream regulation of circSOD2 revealed that its expression was upregulated by oxidative stress but was mitigated by Vit C or NAC pretreatment. These findings align with growing interest in antioxidants and antioxidative drugs as potential strategies to mitigate MAFLD development and progression.

Clinical evidence increasingly supports a strong association between low BMD or osteoporosis and MAFLD in both male and female adults[9,11,37,38]. This relationship is not restricted to adults, highlighted by a recent meta-analysis reporting that MAFLD is significantly linked to reduced whole-body BMD in children and adolescents[39]. Despite this evidence, the causal relationship is still not defined, not to mention molecular mechanism. Interestingly, we found that hepatocytes delivered circSOD2 to BMSCs, as confirmed through cell co-culture experiments, indicating that circSOD2 mediates the crosstalk between the liver and bone via endocrine signaling. Gain- and loss-of-function analyses further revealed that circSOD2 overexpression inhibited osteogenic differentiation of BMSCs and bone formation, whereas circSOD2 silencing enhanced these processes. Our findings provide cellular and molecular evidence to support the mainstream view that MAFLD is a significant risk factor for low BMD or osteoporosis. Furthermore, our results suggest that targeting circSOD2 represents a promising therapeutical approach to prevent the deleterious effects of MAFLD on bone formation and prevent osteoporosis in patients with severe MAFLD.

CircRNAs regulate cell behavior through diverse mechanisms, such as interaction with miRNAs or proteins and modulation of transcription and translation[17,40]. miRNAs, a class of small single-stranded non-coding RNAs, regulate gene expression through mRNA degradation or translational inhibition[41]. Several studies have reported that circSOD2 acts as a sponge for several miRNAs, such as miR-206[22], miR-224-5p[34], miR-502-5p[35], and miR-1296[33]. Similarly, we found an interaction between circSOD2 and miR-29b using a luciferase reporter assay and observed their co-localization in the cytoplasm, as evidenced by the FISH assay. These findings indicate that circSOD2 functions as a competitive RNA, suppressing the activity of miR-29b. Previous studies, including our own, have shown that miRNAs play a crucial role in the osteogenic differentiation of MSCs[7,42]. Furthermore, in vitro and in vivo rescue experiments indicated that circSOD2 inhibited the osteogenic differentiation of BMSCs and bone formation through miR-29b. Our findings were consistent with previous reports that miR-29b promoted osteoblast differentiation by targeting anti-osteogenic factors[43] and biomaterials-mediated intracellular delivery of miRNA-29b enhanced osteogenic differentiation of MSCs[23] and bone regeneration[2]. Given that miR-29b has been identified to regulate the Wnt/β-catenin pathway in prior research[44,45], we speculated that the circSOD2/miR-29b axis also regulates osteogenic differentiation via the Wnt/β-catenin pathway.

The present study had certain limitations. Firstly, the downstream targets and signaling pathways of miR-29b were not investigated. Secondly, the expression of miR-29b in human samples and its relationship with MAFLD was not investigated. Thirdly, the therapeutic effect of targeting circSOD2 was not investigated using an animal model of osteoporosis. Finally, we were unable to establish an animal model of MAFLD simultaneously accompanied by osteoporosis to study this relationship.

CONCLUSION

In summary, this study demonstrates a novel mechanism underlying the crosstalk between the liver and bone. We showed thatcircSOD2 upregulation, induced by oxidative stress in the context of MAFLD, inhibits osteogenesis by sponging miRNA-29b. These findings offer insights into the relationship between MAFLD and osteoporosis, suggesting that MAFLD may contribute to the pathogenesis of osteoporosis. Targeting circSOD2 represents a potential strategy for preventing osteoporosis in patients with MAFLD, which warrants further investigation in future studies.