Introduction
Ankylosing spondylitis (AS) is a chronic inflammatory disease primarily affecting the spine and sacroiliac joints, often leading to progressive bone fusion and spinal rigidity (Mauro et al. 2021). Although the exact pathogenesis of AS remains incompletely understood, research suggests that genetic factors, immune responses, and environmental influences play significant roles in its development (Bilski et al.
2024; Metzgar 2020). Current therapeutic approaches mainly include nonsteroidal anti-inflammatory drugs (NSAIDs), tumor necrosis factor-alpha (TNF-α) inhibitors, and interleukin-17 (IL-17) inhibitors, which can alleviate symptoms and slow disease progression to some extent (Carbo et al. 2018; Ren et al.
2024). However, the efficacy of existing treatments is limited, and they do not offer a cure. Particularly in the advanced stages of AS, many patients still face severe functional impairments and reduced quality of life.
Therefore, further exploration of the underlying molecular mechanisms and the development of novel therapeutic targets are of great importance.
Mesenchymal stem cells (MSCs) are adult stem cells with self-renewal and differentiation capabilities (Mishra et al. 2020). MSCs from AS patients are believed to exhibit enhanced osteogenic differentiation potential, which may be a key factor contributing to pathological ossification in AS, such as ligament ossification and bone bridge formation (Abdolmohammadi et al.
2019; Liu et al. 2023; Hu et al. 2022).
Researchers are investigating multiple potential signaling pathways and molecular mechanisms underlying the increased osteogenic differentiation of MSCs in AS, including the Wnt/β-catenin (Daoussis et al. 2022), BMP (bone morphogenetic protein) (Zheng et al. 2019), and TGF-β (transforming growth factor-β) signaling pathways (Chen et al.
2025), as well as inflammatory cytokines (Petryk and Shevchenko 2020) and genetic factors (Feng et al. 2024).
Circular RNAs (circRNAs) are a novel class of non-coding RNAs characterized by their unique circular structure, which confers high stability and conservation (Xu et al. 2024). Recent studies have increasingly highlighted the regulatory roles of circRNAs in the pathogenesis of various diseases, particularly inflammatory diseases and cancers (Xu et al.
2024; Liu et al. 2022; Bhat et al. 2024).
In the context of AS, certain circRNAs have been found to be closely associated with its pathological processes (Song et al. 2023). For instance, hsa_circ_0003307 is significantly upregulated in AS patients.
Overexpression of hsa_circ_0003307 promotes the activation of the PI3K/AKT pathway and increases the expression of inflammatory cytokines such as TNF-α and TNF-α-induced protein 2 (TNFAIP2), while knockdown of hsa_circ_0003307 reduces their expression (Fang et al. 2022). Additionally, circRNAs are thought to function as miRNA sponges, regulating downstream gene expression and thereby influencing disease progression.
For example, circ_0018168 has been identified as a sponge for miR-330-3p, which targets DKK1. Circ_0018168 may inhibit fibroblast proliferation and osteogenic differentiation by modulating the miR-330-3p/DKK1 axis, thereby affecting AS progression (Zhao et al. 2022).
Our preliminary bioinformatics screening identified circ_0018814 as a candidate circRNA potentially associated with AS, suggesting its potential role in the pathogenesis of AS.
MiRNAs, as crucial regulators of gene expression, also play a pivotal role in the development and progression of AS (Fang and Liu 2023). MiR-944 is a recently studied miRNA, and although its specific function in AS has not been extensively investigated, preliminary studies suggest that miR-944 may influence cell differentiation and immune responses by targeting multiple transcription factors (Chia et al. 2020; Wei et al.
2024). Research has revealed that miR-1270 and miR-944 are downregulated during the osteogenic differentiation of bone marrow mesenchymal stem cells (BMSCs) (Chia et al. 2020).
Circ-DAB1 can interact with miR-1270 and miR-944, upregulating the expression of RBPJ and DAB1, thereby promoting BMSC proliferation and osteogenic differentiation (Chia et al. 2020). KLF4 (Kruppel-like factor 4) is an important transcription factor that has been shown to play a significant role in osteogenic differentiation (Xu et al.
2018). KLF4 not only participates in the osteogenic differentiation of MSCs but may also influence the pathological progression of AS by regulating the expression of bone matrix proteins (Chen et al. 2022).
Therefore, the interaction between miR-944 and KLF4 may be of significant importance in the pathogenesis of AS.
This study investigated the role of the circular RNA circ_0018814 in the osteogenic differentiation of MSCs in AS patients and to elucidate its potential impact on the pathological progression of AS through the regulation of the miR-944/KLF4 axis. By conducting in-depth functional and interaction studies on circ_0018814, miR-944, and KLF4, we hope to uncover novel mechanisms underlying the development and progression of AS and provide new potential therapeutic targets for future AS treatment.
Materials and methods
Bioinformatics analysis
Differentially expressed circRNAs were explored from the GEO dataset GSE203155, and these findings were visualized using a volcano plot. Additionally, the top 20 upregulated and downregulated circRNAs were displayed via a heatmap. Potential target miRNAs of these circRNAs were analyzed using the circinteractome and circBank databases, and the intersection of miRNAs obtained from both databases yielded common miRNAs.
The target genes of miR-944 were predicted using the PITA, RNA22, and microT databases, and the intersection of these predictions identified common genes. GSE203155 was used as an exploratory screening dataset for candidate circRNA identification rather than as confirmatory evidence of differential expression after multiple-testing correction.
Bone marrow samples
All bone marrow samples were obtained from patients undergoing total hip replacement surgery at People’s Hospital of Ningxia Hui Autonomous Region, including patients with AS and those with femoral neck fractures (serving as non-AS controls). The diagnosis of AS patients was based on the 1984 revised New York criteria, and these patients had no history of biologic agent use, no other systemic diseases (e.g., diabetes, malignancies, hepatitis), and no active inflammation in other body regions (e.g., osteoarthritis, infections). The study was approved by the Clinical Ethics Committee of the People’s Hospital of Ningxia Hui Autonomous Region (NXH-leyxy-r046).
All enrolled patients, including those with AS and femoral neck fractures, were informed of the sample usage and provided written informed consent. Bone marrow samples were collected during surgery, transferred to the laboratory in an ice box, and processed in a biosafety cabinet for the isolation of human bone marrow-derived mesenchymal stem cells (MSCs). Circ_0018814 expression shown in Fig.
1C was measured in bone marrow samples from 24 patients with AS and 24 non-AS controls. Baseline clinical characteristics of the study population were summarized in a Supplementary Table 3.
Fig. 1Circ_0018814 is aberrantly overexpressed during osteogenic differentiation of AS-MSCs. A The volcano plot illustrates the differential expression of circRNAs in AS-MSCs compared to non-AS control MSCs within the GSE203155 dataset, revealing 196 upregulated and 401 downregulated circRNAs.
B Heatmap of differentially expressed circRNAs (TOP20) in AS-MSCs and non-AS control MSCs during osteogenic differentiation. hsa_circ_0018814 is indicated by the red box. C The relative expression level of circ_0018814 was detected via qRT-PCR in bone marrow samples from non-AS controls (n = 24) and AS patients (n = 24). D qRT-PCR analysis of circ_0018814 expression in AS-MSCs and non-AS control MSCs at days 0, 3, 7, and 14 post-osteogenic induction.
(E-F) FISH showing the localization and expression levels of circ_0018814 in AS-MSCs and non-AS control MSCs. Nuclei were counterstained with DAPI (blue). Red arrows highlight representative circ_0018814 signals. **P < 0.01 versus non-AS control MSCs group, **P < 0.01 versus HD group, *P < 0.05 versus 0 day group, **P < 0.01 versus 0 day group
Isolation, culture, and identification of MSCs
The bone marrow was diluted with an equal volume of PBS, gently mixed, and then layered over an equal volume of Ficoll separation medium in a new centrifuge tube. After centrifugation at 2000 rpm for 30 min, the MSCs were collected from the intermediate white layer. This layer was carefully aspirated and transferred to a new 15 mL centrifuge tube, mixed with an equal volume of α-MEM culture medium, and centrifuged at 1000 rpm for 5 min.
The supernatant was discarded, and the cell pellet was resuspended in 5 mL of α-MEM medium, followed by another centrifugation step. After discarding the supernatant, the cells were resuspended in 5 mL of α-MEM medium, counted using a hemocytometer, and seeded into 25 cm² culture flasks at a density of 2 × 105 cells/mL (designated as P0). The cells were cultured in a 37 °C incubator, and the medium was replaced after 72 h.
When the cells reached 80–90% confluence, they were passaged at a 1:3 ratio using 0.25% trypsin-EDTA. The medium was replaced every two days, and P3 cells were used for purity assessment. Primary human MSCs were obtained from 5 individual donors.
Donor-derived MSC preparations were considered biological replicates, whereas repeated wells within the same experiment were considered technical replicates.
Morphological identification
P3 cells were observed under a microscope, and 3–5 random fields were selected to assess cell morphology.
Flow cytometry identification
Cells were washed with PBS, trypsinized, and resuspended in complete medium. After centrifugation at 1000 rpm for 5 min, the cells were resuspended in PBS, counted, and adjusted to a density of 2 × 106 cells/mL. A 100 µL aliquot of cell suspension was incubated with specific antibodies (CD105, CD90, CD73, CD45, CD14, and CD34) at a 1:100 dilution in the dark at 4 °C for 30 min.
After washing, the cells were resuspended in PBS containing 2% fetal bovine serum and analyzed by flow cytometry.
Cell transfection
MSCs were infected with lentiviruses (Genechem Biotechnology, China) carrying overexpression vectors or shRNAs specifically targeting Circ_0018814 (The dosing of Lv-si-circ_0018814 is 1 × 108 TU/mL, 100 µL), and incubated at 37 °C overnight. The medium was subsequently replaced with fresh α-MEM to remove residual virus. Transduction efficiency was assessed via real-time PCR, and positive monoclonal cells were selected for subsequent amplification.
For the modulation of miR-944 and KLF4 expression, MSCs were transfected with miR-944 mimics, miR-944 inhibitors, si-KLF4, or pcDNA-KLF4 plasmids (Hanbio Biotechnology, China) using Lipofectamine RNAiMAX reagent (Invitrogen Life Technologies). A negative empty vector or scrambled siRNA served as the control. OPTI-MEM (Invitrogen Life Technologies) was mixed with Lipofectamine RNAiMAX and the corresponding RNAs or plasmids, incubated for 15 min, and subsequently added to the cells.
After an incubation period of 4–6 h, the OPTI-MEM was discarded and replaced with complete α-MEM containing serum. The efficiency of transfection was analyzed via real-time PCR.
CCK-8 assay
Cells were inoculated into 96-well culture plates at 2000 cells/well. Cell viability was assessed at 48 h by adding 10 µL of CCK-8 reagent (CA1210, Solarbio) to each well. After co-incubation at 37 °C for 2 h, optical density (OD) values at 450 nm were quantified using a Microplate reader (BioRad).
EdU assay
Cells were seeded into 24-well plates at a density of 2 × 105 cells/mL. Post-treatment, cell proliferation was evaluated using the EdU Cell Proliferation Kit (C0078S, Beyotime) following the standard protocol. The cells were incubated with EdU working solution at 37 °C for 4 h, then subjected to fixation, washing, and permeabilization.
After applying the additive solution, cell nuclei were stained with DAPI. Imaging was performed via Fluorescence microscopy (Olympus, Tokyo, Japan). The percentage of proliferating cells was calculated as: (EdU-positive cells/DAPI-positive cells) × 100%.
Dual-luciferase reporter assay
The targeting relationship between circ_0018814, KLF4, and miR-944, as well as the binding sites of miR-944 in the 3′ untranslated regions (UTRs) of circ_0018814 and KLF4, were analyzed using bioinformatics software (https://rnasysu.com/encori/). Wild-type (WT) and mutant (MUT) plasmids containing the 3’UTR sequences of circ_0018814 and KLF4 were constructed. MSCs at 70% confluence were transfected with WT or MUT plasmids, along with miR-944 mimic or mimic negative control (mimic NC), using Lipofectamine 2000 (Invitrogen).
After 48 h, luciferase activity was measured using the Dual-Luciferase® Reporter Assay System (Promega, Madison, USA).
RNA immunoprecipitation (RIP)
The RIP experiment was performed using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, MA). Approximately 1 × 107 cells were collected, washed with ice-cold PBS, and lysed in RIP lysis buffer containing protease and RNase inhibitors. After incubation on ice for 10 min, the lysate was centrifuged at 12,000g for 10 min (4 °C), and the supernatant was used for RIP.
The samples were incubated with magnetic beads conjugated with anti-Argonaute2 antibody (experimental group) or IgG (negative control) at 4 °C overnight to capture RNA-protein complexes. The beads were washed five times with RIP wash buffer, and the bound RNA was eluted using the provided lysis buffer. The extracted RNA was reverse-transcribed, and circ_0018814 enrichment was detected by qPCR.
Fluorescence in situ hybridization (FISH)
MSCs were collected, washed with ice-cold PBS, and fixed with 4% paraformaldehyde (PFA) at room temperature for 15 min. After washing with PBS, the cells were permeabilized with 0.1% Triton X-100 for 10 min. Pre-hybridization was performed in hybridization buffer at 37 °C for 1 h.
FAM-labeled circ_0018814 and Cy3-labeled miR-944 probes (GenePharma, Shanghai, China) were hybridized with the cells at 37 °C overnight in the dark. After hybridization, the cells were washed with preheated SSC buffer (2× and 1×) to remove unbound probes. In some experiments, cells were incubated with Rorγt-specific antibody (CST, MA, USA) at 4 °C overnight, followed by incubation with fluorescently labeled secondary antibody at room temperature for 1 h.
Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, KeyGEN, Nanjing, China) for 5 min. The samples were mounted with anti-fade mounting medium and imaged using a fluorescence microscope system (Olympus, Tokyo, Japan). All procedures were performed in the dark, and strict negative controls were included to ensure the accuracy of the results.
Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from cells or tissues using the TRIZOL reagent kit. The extracted total RNA was used as a template to synthesize cDNA using a reverse transcription kit. The reaction mixture (20 µL) contained 1 µg of total RNA, 4 µL of 5× reverse transcription buffer, 1 µL of random primers, 1 µL of dNTP mixture (10 mM), 1 µL of reverse transcriptase, and an appropriate amount of RNase-free water.
The reaction conditions were as follows: 25 °C for 5 min, 42 °C for 60 min, and 70 °C for 10 min. The synthesized cDNA was used as a template for the amplification of target gene fragments. The PCR reaction mixture (25 µL) consisted of 2 µL of cDNA template, 12.5 µL of SYBR Premix Ex Taq II, 1 µL of forward primer, 1 µL of reverse primer, and sterile distilled water to a final volume of 25 µL.
The PCR conditions were as follows: initial denaturation at 94 °C for 10 min; 35 cycles of 94 °C for 30 s, 60 °C for 30 s, and 72 °C for 40 s; and a final extension at 72 °C for 10 min. Each sample was run in triplicate to ensure reproducibility. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and U6 small nuclear RNA (U6) were used as internal reference genes to normalize the expression levels of the target genes.
Relative gene expression levels were calculated using the 2−ΔΔCt method.
Western blot
Cells or tissue samples were washed twice with ice-cold PBS to remove residual medium or blood. An appropriate amount of lysis buffer (containing protease and phosphatase inhibitors) was added, and the samples were lysed on ice for 30 min, with vortexing every 10 min. After lysis, the samples were centrifuged at 12,000×g for 15 min at 4 °C, and the supernatant was collected as the total protein extract.
Protein concentration was determined using the BCA protein assay kit. Fifty micrograms of protein were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). After electrophoresis, the proteins were transferred onto a nitrocellulose membrane using a wet transfer method.
The membrane was blocked with 3% non-fat milk in TBST at room temperature for 1 h. After blocking, the membrane was incubated with rabbit polyclonal antibodies against ALP (ab229126, Abcam, 1:1000), RUNX2 (ab192256, Abcam, 1:1000), COL1A1 (ab138492, Abcam, 1:1000), OPN (ab214050, Abcam, 1:1000), SP7 (ab229258, Abcam, 1:1000), BMP2 (SAB5700607, Sigma-Aldrich, 1:1000) or GAPDH (ab181602, Abcam, 1:1000) at 4 °C overnight. The next day, the membrane was washed three times with TBST for 10 min each to remove unbound primary antibodies.
Subsequently, the membrane was incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit secondary antibody (1:2,000 dilution) at room temperature for 2 h. After incubation, the membrane was washed three times with TBST for 10 min each. The ECL working solution was evenly applied to the membrane, and after 1 min of incubation at room temperature, the chemiluminescent signal was captured using a chemiluminescence imaging system.
GAPDH was used as an internal reference.
Osteogenic differentiation
BMSCs were seeded in 12-well plates and cultured in osteogenic differentiation medium for 14 days. The medium consisted of Dulbecco’s Modified Eagle Medium (DMEM), 10% fetal bovine serum (FBS), 10 nM dexamethasone, 10 nM β-glycerophosphate, 50 µg/mL vitamin C, 100 IU/mL penicillin, and 100 IU/mL streptromycin (Sigma). When the cells reached 70–80% confluency in 24-well plates, the medium was replaced with osteogenic induction medium, which was refreshed every 3 days.
Alkaline phosphatase (ALP) staining
On day 7 of osteogenic induction, the supernatant was discarded, and the cells were washed three times with PBS. Five hundred microliters of 95% ethanol were added to each well for fixation at room temperature for 15 min. After fixation, the ethanol was removed, and the cells were washed three times with PBS for 2 min each to ensure complete removal of ethanol.
An appropriate amount of ALP staining solution was added to each well to cover the cells, and the cells were incubated in the dark at room temperature for 30 min. After incubation, the staining solution was discarded, and the cells were washed twice with distilled water to stop the reaction. The cells were then observed and photographed under an optical microscope.
ALP activity assay
Cells from each group were collected and lysed with cell lysis buffer (containing 1% Triton X-100). The lysates were scraped and transferred to microcentrifuge tubes, which were placed on ice and vortexed every 10 min for a total of 30 min. The lysates were then centrifuged at 18,000 rpm for 10 min, and the supernatant was collected.
Protein concentration was determined using the BCA method. ALP activity was measured according to the instructions provided with the ALP activity assay kit.
Alizarin red S (ARS) staining
After 14 days of osteogenic induction, cells in each group were fixed with 500 µL of 95% ethanol at room temperature for 15 min. The ethanol was then discarded, and the cells were washed three times with PBS. An appropriate amount of ARS staining solution was added to each well to cover the cells, and the cells were incubated at room temperature for 5 min.
Orange-red nodules were observed in the wells. The staining solution was discarded, and the cells were washed three times with distilled water to stop the reaction. The cells were then observed and photographed under an optical microscope.
Induction of AS mouse model and grouping
In the preparation of the immunoadjuvant (DDA) and proteoglycan mixture, 33 mg of DDA was completely dissolved in 3300 µL of PBS, heated in a water bath to 50–60 °C, and then cooled on ice for 15 min. Proteoglycan lyophilized powder was dissolved in PBS and kept on ice. The mixture was filtered through a 0.22 μm membrane for sterilization, and the proteoglycan concentration was verified to ensure consistency.
Each 24–26-week-old BALB/c mouse was intraperitoneally injected with 200 µL of the mixture at weeks 0, 3, and 6. The mice were observed three times a week, and any abnormal changes caused by arthritis were recorded (Shen et al. 2023).
The experimental groups were randomly allocated into a sham group (n = 8), a model group (n = 8), a model + Lv-si-NC group (n = 8), and a model + Lv-si-circ_0018814 group (n = 8). The sham group was injected with an equal amount of DDA only. Two weeks after the third immunization, the latter two groups began receiving injections of Lv-si-NC or Lv-si-circ_0018814 via the tail vein every two weeks.
After sacrifice, spinal lesion tissues were immediately collected from mice in each group. Primary MSCs were further isolated from the lesion-related tissues, and total RNA was extracted for qRT-PCR analysis of circ_0018814 expression to evaluate the in vivo knockdown efficiency of Lv-si-hsa_circ_0018814. This study was approved by the Ethics Committee of People’s Hospital of Ningxia Hui Autonomous Region (NXH-lcyxy-n063).
In this animal study, the investigator responsible for group allocation was aware of the group assignment during randomization. However, the experimenters administering the interventions, the assessors evaluating the outcomes, and the statisticians analyzing the data were blinded to the group assignment throughout the study. The blinding was maintained until the completion of the primary analysis to ensure the objectivity of the results.
Hematoxylin and Eosin (H&E) staining
After the completion of the model induction, the mice were euthanized by cervical dislocation. The spine was dissected using tissue scissors, and bloodstains were washed away with PBS. Some tissues were fixed in 4% paraformaldehyde for 48 h, followed by decalcification in 10% EDTA.
The tissues were dehydrated through a graded ethanol series (50%, 70%, 85%, 95%, and 100%), cleared in xylene twice (5 min each), and embedded in paraffin. Sections of 4 μm thickness were cut using a microtome. The sections were deparaffinized in xylene twice (10 min each), rehydrated through a graded ethanol series (100%, 95%, 85%, and 75%) for 5 min each, and rinsed with distilled water for 5 min.
The sections were stained with 2% hematoxylin for 5 min, differentiated in 1% acid alcohol for 5 s, and blued in ammonia water for 1 min. After rinsing with distilled water for 1 min, the sections were stained with 0.5% eosin for 2 min, rinsed with distilled water for 30 s, dehydrated through 80%, 95%, and 100% ethanol for 30 s, 2 min, and 3 min, respectively, cleared in xylene for 3 min, and mounted with neutral resin. The cervical spine tissue structure was observed and photographed under an optical microscope.
Safranin O and fast green staining
The sections were placed in a 37 °C oven overnight and then transferred to a 60 °C oven for an additional 30 min. The sections were cleared in xylene three times (2 min, 1 min, and 1 min), rehydrated through 100% ethanol, 95% ethanol, and 95% ethanol for 1 min each, and rinsed with distilled water for 2 min to remove the paraffin. The sections were stained with hematoxylin for 10 min, rinsed with running water three times, and washed with distilled water for 2 min.
The sections were differentiated in 1% acid alcohol for 10 s and rinsed with distilled water for 2 min. The sections were then stained with 0.05% fast green for 5 min, dipped in 1% acetic acid for 15 s, and stained with 0.1% safranin O for 5 min. The sections were dehydrated through 95% ethanol, 100% ethanol, and 100% ethanol for 1 min, 15 s, and 1 min, respectively, and cleared in xylene three times (1 min each).
A drop of resin was added to the sections, and a coverslip was gently placed on top to avoid air bubbles. The sections were allowed to dry flat and were observed and photographed under an optical microscope.
Statistical analysis
All in vitro experiments were repeated at least three times independently under identical conditions, and similar results were observed. The raw data generated in this study are available from the corresponding author upon reasonable request. Statistical analysis was performed using GraphPad Prism software, with each experiment independently repeated three times to ensure reliability and reproducibility.
All results were summarized and presented as the mean ± standard error of the mean (SEM). Differences between two groups were evaluated using independent sample t-tests, while comparisons involving three or more groups were analyzed using one-way analysis of variance (ANOVA) followed by Bonferroni’s post hoc test. P < 0.05 was considered statistically significant.
Results
The abnormal overexpression of circ_0018814 during osteogenic differentiation of MSCs in patients with AS
In this study, we investigated the differential expression of circRNAs during osteogenic differentiation of MSCs derived from patients with AS and non-AS controls. Analysis of the dataset GSE203155 identified circ_0018814 as a candidate circRNA associated with osteogenic differentiation in AS-MSCs (Fig. 1A, B)non-AS control MSCs.
Because this dataset served as an exploratory screening dataset, these findings were not interpreted as confirmatory evidence alone. We further demonstrated that circ_0018814 was significantly upregulated in bone marrow samples from patients with AS (Fig. 1C).
To validate these findings, bone marrow samples were collected from AS patients and non-AS control patients with femoral neck fractures, and MSCs were isolated and characterized. The isolated MSCs exhibited negative expression for CD14, CD45, and CD34, while positive expression for CD73, CD90, and CD105 (Supplementary Fig. 1), confirming their identity as MSCs.
Following osteogenic induction for two weeks, qRT-PCR analysis demonstrated a progressive increase in circ_0018814 expression in AS-MSCs, with a 3-fold increase at day 3, 5-fold at day 7, and 8-fold at day 14 (Fig. 1D). In contrast, non-AS control MSCs showed a modest increase in circ_0018814 expression, with only a 1.5-fold increase by day 14.
FISH further confirmed the higher expression of circ_0018814 in AS-MSCs compared to non-AS control MSCs (Fig. 1E, F). Additionally, circ_0018814 was localized in both the nucleus and cytoplasm of both AS-MSCs and non-AS control MSCs.
These results suggest that circ_0018814 is aberrantly overexpressed in AS-MSCs during osteogenic differentiation, potentially contributing to the pathological bone formation observed in AS.
Circ_0018814 promotes osteogenic differentiation of AS-MSCs
To further investigate the role of circ_0018814 in the osteogenic differentiation of AS-MSCs, we performed both knockdown and overexpression experiments. The efficiency of circ_0018814 interference and overexpression was confirmed by qRT-PCR (Fig. 2A).
As shown in the results, the expression level of circ_0018814 was significantly reduced in the Lv-si-circ_0018814 group compared to the Lv-si-NC group, while it was markedly increased in the Lv-circ_0018814 group compared to the Lv-NC group. Western blot analysis revealed that circ_0018814 significantly influenced the expression of key osteogenic markers, including ALP, RUNX2, and Col1a1. In the presence of osteogenic induction medium (OM), the expression levels of these markers were notably higher in the OM + Lv-circ_0018814 group compared to the OM + Lv-NC group.
Conversely, knockdown of circ_0018814 (OM + Lv-si-circ_0018814) resulted in a significant reduction in the expression of these osteogenic markers compared to the OM + Lv-si-NC group (Fig. 2B, E). Consistent with the western blot results, ALP staining and ALP activity assays demonstrated that circ_0018814 overexpression enhanced the osteogenic differentiation of AS-MSCs, as evidenced by increased ALP staining intensity and higher ALP activity in the OM + Lv-circ_0018814 group.
In contrast, circ_0018814 knockdown led to a decrease in ALP staining and activity (Fig. 2F, G). Additionally, Alizarin Red S (ARS) staining was performed to assess the mineralization of the extracellular matrix, a critical step in osteogenic differentiation.
The results showed that circ_0018814 overexpression significantly increased matrix mineralization, whereas knockdown of circ_0018814 reduced mineralization compared to the respective control groups (Fig. 2H). Collectively, these findings suggest that circ_0018814 plays a crucial role in promoting the osteogenic differentiation of AS-MSCs, as evidenced by its positive regulation of osteogenic marker expression, ALP activity, and matrix mineralization.
Fig. 2Circ_0018814 promotes osteogenic differentiation of AS-MSCs. A qRT-PCR validation of circ_0018814 knockdown and overexpression efficiency.
B–E Western blot analysis of osteogenic markers (ALP, RUNX2, Col1a1) in AS-MSCs under circ_0018814 knockdown or overexpression. F–G ALP staining and ALP activity assays in AS-MSCs with circ_0018814 knockdown or overexpression. H ARS staining to assess matrix mineralization in AS-MSCs with circ_0018814 knockdown or overexpression. **P < 0.01 versus Blank group, ##P < 0.01 versus OM + Lv-si-NC group, ##P < 0.01 versus OM + Lv-NC group
Circ_0018814 targets miR-944 and regulates osteogenic differentiation of AS-MSCs
To further investigate the functional mechanism of circ_0018814, we analyzed its potential miRNA targets using the circinteractome and circBank databases, and identified hsa-miR-944 through intersection analysis (Fig. 3A). Although none of the identified miRNAs (including hsa-miR-1265, hsa-miR-548 m, hsa-miR-1248, hsa-miR-553, hsa-miR-556-5p, and hsa-miR-944) have been directly studied in AS, miR-944 was prioritized due to its relatively extensive documentation in other disease contexts.
It has been implicated in critical biological processes such as inflammation, apoptosis, and cell proliferation—pathways highly relevant to AS pathogenesis. In contrast, the other miRNAs lack substantial experimental evidence or reported functional roles, making miR-944 a more rational candidate for further investigation. While no direct evidence demonstrates the role of miR-944 in osteogenic differentiation, literature suggests that its predicted target genes (e.g., STAT3, BMPR2) are involved in bone metabolism-related signaling pathways (Chen et al.
2022; Ning et al. 2025). This supports the possibility that miR-944 may indirectly regulate osteogenesis through downstream mediators, providing a plausible mechanistic link to AS-related bone remodeling.
Luciferase reporter assays confirmed the direct interaction between circ_0018814 and hsa-miR-944. The results showed that transfection with miR-944 mimics significantly reduced luciferase activity in the circ_0018814-wt group, while no significant change was observed in the circ_0018814-mut group, indicating that circ_0018814 directly targets miR-944 through its binding site (Fig. 3B, C).
Additionally, RIP experiments further validated the binding of circ_0018814 and miR-944 within the AGO2 complex, suggesting a functional interaction in the RNA-induced silencing complex (RISC) (Fig. 3D, E). RT-PCR results demonstrated that knockdown of circ_0018814 significantly upregulated miR-944 expression in osteogenically induced AS-MSCs, while overexpression of circ_0018814 significantly suppressed miR-944 expression (Fig.
3F). FISH confirmed that both circ_0018814 and miR-944 were localized in the cytoplasm and nuclear, supporting their interaction in this cellular compartment (Fig. 3G).
Fig. 3Circ_0018814 targets miR-944 to regulate osteogenic differentiation of AS-MSCs. A The miRNA targets of circ_0018814 from the circinteractome and circBank databases were analyzed using Venn analysis.
B Schematic diagram of the predicted binding site between circ_0018814 and miR-944. C Luciferase reporter assay showing the interaction between circ_0018814 and miR-944. D–E RIP assay confirming the binding of circ_0018814 and miR-944 in the AGO2 complex.
F qRT-PCR analysis of miR-944 expression in AS-MSCs with circ_0018814 knockdown or overexpression. G FISH showing the localization and expression levels of circ_0018814 in AS-MSCs and non-AS control MSCs. **P < 0.01 versus Blank group, ##P < 0.01 versus OM + Lv-si-NC group, ##P < 0.01 versus OM + Lv-NC group
To explore the role of miR-944 in the osteogenic differentiation of AS-MSCs, we performed miR-944 knockdown and overexpression experiments. qRT-PCR results revealed that miR-944 mimics significantly upregulated miR-944 expression, while miR-944 inhibitor significantly downregulated its expression (Fig. 4A). Western blot analysis showed that miR-944 overexpression significantly suppressed the expression of osteogenic markers ALP, RUNX2, and COL1A1, whereas miR-944 inhibition significantly enhanced their expression (Fig.
4B, E). ALP staining and activity assays, as well as Alizarin Red S staining, further confirmed that miR-944 overexpression inhibited the osteogenic differentiation of AS-MSCs, while miR-944 inhibition significantly promoted this process (Fig. 4F, H).
In summary, circ_0018814 regulates the osteogenic differentiation of AS-MSCs by targeting miR-944, and miR-944 acts as a negative regulator in this process. These findings provide new molecular insights into the pathological mechanisms of AS and suggest that the circ_0018814/miR-944 axis may serve as a potential therapeutic target for AS.
Fig. 4miR-944 inhibits osteogenic differentiation of AS-MSCs. A qRT-PCR validation of miR-944 knockdown and overexpression efficiency.
B–E Western blot analysis of osteogenic markers (ALP, RUNX2, COL1A1) in AS-MSCs under miR-944 knockdown or overexpression. F–G ALP staining and ALP activity assays in AS-MSCs with miR-944 knockdown or overexpression. H ARS staining to assess matrix mineralization in AS-MSCs with miR-944 knockdown or overexpression. **P < 0.01 versus Blank group, ##P < 0.01 versus OM+mimics-NC group, ##P < 0.01 versus OM+inhibitor-NC group
miR-944 targets KLF4 and regulates osteogenic differentiation of MSCs in AS patients
To elucidate the regulatory mechanism of miR-944 in the osteogenic differentiation of MSCs from AS patients, we first predicted the potential target genes of miR-944 using the PITA, RNA22, and microT databases, and identified KLF4 as a target through intersection analysis (Fig. 5A). To validate this prediction, we performed luciferase reporter assays.
The results showed that miR-944 mimics significantly reduced luciferase activity in the KLF4-wt group compared to the control group, while no significant effect was observed in the KLF4-mut group, indicating that miR-944 directly targets the 3’UTR region of KLF4 (Fig. 5B, C). Further RT-PCR analysis revealed that transfection with miR-944 mimics significantly suppressed KLF4 expression in AS-MSCs, while transfection with miR-944 inhibitor significantly increased KLF4 expression (Fig.
5D). Additionally, overexpression of circ_0018814 significantly upregulated KLF4 expression, while si-circ_0018814 significantly downregulated KLF4 expression (Fig. 5E).
Fig. 5miR-944 targets KLF4 to regulate osteogenic differentiation of AS-MSCs. A The mRNA targets of miR-944 predicted by the PITA, RNA22, and microT databases were analyzed using Venn analysis.
B Schematic diagram of the predicted binding site between miR-944 and KLF4. C Luciferase reporter assay showing the interaction between miR-944 and KLF4. D–E qRT-PCR analysis of KLF4 expression in AS-MSCs with circ_0018814 or miR-944 overexpression or inhibition.
F qRT-PCR validation of miR-944 knockdown and overexpression efficiency. G–J Western blot analysis of osteogenic markers (ALP, RUNX2, COL1A1) in AS-MSCs with KLF4 knockdown or overexpression. (K-L) ALP staining and ALP activity assays in AS-MSCs with KLF4 knockdown or overexpression.
(M) ARS staining to assess matrix mineralization in AS-MSCs with KLF4 knockdown or overexpression. **P < 0.01 versus Blank group, ##P < 0.01 versus OM + si-NC group, ##P < 0.01 versus OM+vector group
To clarify the role of KLF4 in the osteogenic differentiation of AS-MSCs, we conducted KLF4 knockdown and overexpression experiments. qRT-PCR results showed that si-KLF4 significantly reduced KLF4 expression, while pcDNA-KLF4 significantly increased KLF4 expression (Fig. 5F). Western blot analysis further demonstrated that KLF4 overexpression significantly upregulated the expression of osteogenic markers ALP, RUNX2, COL1A1, OPN, SP7 and BMP2, while KLF4 knockdown significantly suppressed their expression (Fig.
5G, J, Supplementary Fig. 2). Functional assays further confirmed the promotive role of KLF4 in the osteogenic differentiation of AS-MSCs.
ALP staining and activity assays revealed that KLF4 overexpression significantly enhanced ALP activity, while KLF4 knockdown significantly inhibited ALP activity (Fig. 5K, L). Moreover, Alizarin Red S staining results showed that KLF4 overexpression significantly increased the mineralization of bone matrix, while KLF4 knockdown significantly reduced mineralization (Fig.
5M). In summary, miR-944 regulates the osteogenic differentiation of MSCs in AS patients by targeting KLF4, and KLF4 plays a critical role in promoting osteogenic differentiation and bone matrix mineralization.
Circ_0018814 promotes osteogenic differentiation of AS-MSCs through the miR-944/KLF4 axis
Downregulation of circ_0018814 in AS-MSCs was achieved, followed by rescue experiments using miR-944 and KLF4. The efficiency of the interference was validated by qRT-PCR, which showed a significant decrease in KLF4 expression in the si-hsa_circ_0018814 group compared to the control. However, the expression of KLF4 was restored to near control levels in both the si-circ_0018814 + miR-944 inhibitor and si-circ_0018814 + pcDNA-KLF4 groups (Fig.
6A). To further examine the effects of the circ_0018814/miR-944/KLF4 axis on MSC viability and proliferation, we have supplemented the manuscript with CCK-8 cell viability assays and EdU cell proliferation assays. The results demonstrate that knockdown of circ_0018814 significantly inhibits MSC viability and proliferation, whereas miR-944 inhibition or KLF4 overexpression partially rescues these inhibitory effects (Supplementary Fig.
3). Western blot analysis further confirmed the impact of the rescue experiments on the expression of osteogenic markers ALP, RUNX2, Col1a1, OPN, SP7 and BMP2. Similar to the qRT-PCR results, the expression levels of these markers were significantly reduced in the si-circ_0018814 group but were rescued to near control levels in both the si-circ_0018814 + miR-944 inhibitor and si-circ_0018814 + pcDNA-KLF4 groups (Fig.
6B, E, Supplementary Fig. 4). ALP staining and ALP activity assays were performed to assess the functional impact of the rescue experiments on the osteogenic differentiation of AS-MSCs.
The results mirrored the previous findings, with ALP staining and activity significantly reduced in the si-circ_0018814 group but restored to near control levels in the rescue groups (Fig. 6F, G). Finally, ARS staining was used to evaluate the effect of the rescue experiments on bone matrix formation in AS-MSCs.
Consistent with the other assays, ARS staining was significantly decreased in the si-circ_0018814 group but was rescued to near control levels in both the si-circ_0018814 + miR-944 inhibitor and si-circ_0018814 + pcDNA-KLF4 groups (Fig. 6H). These results collectively demonstrate that circ_0018814 promotes osteogenic differentiation of AS-MSCs by regulating the miR-944/KLF4 axis, and that the downregulation of circ_0018814 can be effectively rescued by inhibiting miR-944 or overexpressing KLF4.
Fig. 6Circ_0018814 promotes osteogenic differentiation of AS-MSCs through the miR-944/KLF4 axis. A qRT-PCR analysis of KLF4 expression in AS-MSCs with circ_0018814 knockdown and rescue experiments (miR-944 inhibitor or KLF4 overexpression).
B–E Western blot analysis of osteogenic markers (ALP, RUNX2, COL1A1) in AS-MSCs with circ_0018814 knockdown and rescue experiments. F–G ALP staining and ALP activity assays in AS-MSCs with circ_0018814 knockdown and rescue experiments. H Alizarin Red S (ARS) staining to assess matrix mineralization in AS-MSCs with circ_0018814 knockdown and rescue experiments. **P < 0.01 versus control group, ##P < 0.01 versus Lv-si-hsa_circ_0018814 group
Circ_0018814 promotes the progression of AS by enhancing osteogenic differentiation of MSCs in vivo
To further investigate the role of circ_0018814 in the progression of AS in vivo, we established an AS model in BALB/c mice induced by proteoglycan and divided the experiment into four groups: normal control group, model group, model + Lv-si-NC group, and model + Lv-si-circ_0018814 group. Pathological changes in the spine were observed using H&E and safranin O-fast green staining. The results showed that the spine of mice in the normal control group exhibited no lesions, while severe lesions were observed in both the model group and the model + Lv-si-NC group.
In contrast, the spine lesions in the model + Lv-si-circ_0018814 group were significantly alleviated, suggesting that inhibition of circ_0018814 could mitigate the pathological progression of AS (Fig. 7A, B). To evaluate the effects of hsa_circ_0018814 on disease activity and physical function in AS, we measured the spinal movement angles in various groups (Fig.
7C). Figure 7C demonstrates that the mean movement angle was significantly reduced in the AS group compared to the control group. Again, treatment with Lv-si-NC had no significant effect on movement angles in AS, while treatment with Lv-si-hsa_circ_0018814 led to a substantial increase in movement angles.
To further verify the knockdown efficiency of circ_0018814 in vivo, spinal lesion tissues were collected immediately after sacrifice, and primary MSCs were isolated from the lesion-related tissues. qRT-PCR analysis showed that the expression level of circ_0018814 was increased in the AS group and the AS + Lv-si-NC group, and restored in the AS + Lv-si-hsa_circ_0018814 group compared with the control group, indicating that Lv-si-hsa_circ_0018814 effectively reduced circ_0018814 expression in vivo (Fig. 7D). The effects of circ_0018814 on the expression of miR-944 and KLF4 were analyzed using qPCR.
The results demonstrated that compared to the normal control group, the expression of miR-944 was significantly decreased and the expression of KLF4 was significantly increased in both the model group and the model + Lv-si-NC group. In the model + Lv-si-circ_0018814 group, the expression of miR-944 was restored to near-normal levels (Fig. 7E), and the expression of KLF4 was also significantly reduced, indicating that circ_0018814 is involved in the progression of AS by regulating miR-944 and KLF4 (Fig.
7F). Furthermore, Western blot results showed that the expression of osteogenic markers ALP, RUNX2, and Col1a1 was significantly increased in the model group and the model + Lv-si-NC group, while in the model + Lv-si-circ_0018814 group, the expression of these markers was significantly reduced, further confirming that circ_0018814 promotes the pathological progression of AS by enhancing the osteogenic differentiation of MSCs (Fig. 7G and J).
In summary, circ_0018814 significantly promotes the pathological progression of AS in vivo by regulating the miR-944/KLF4 axis and the expression of proteins related to osteogenic differentiation. qRT-PCR analysis of primary MSCs isolated from spinal lesion tissues further confirmed that Lv-si-hsa_circ_0018814 effectively reduced circ_0018814 expression in vivo. However, because the lentiviral constructs were delivered via tail vein injection, these in vivo findings support a disease-related effect of circ_0018814 inhibition but do not by themselves prove an exclusively MSC-specific mechanism in vivo.
Fig. 7Circ_0018814 promotes the progression of AS in vivo by enhancing osteogenic differentiation of MSCs. A, B H&E and safranin O-fast green staining of spinal lesions in the AS mouse model with circ_0018814 knockdown.
Scale bar, 200 μm C Movement angle measurements: Spinal mobility and joint function were quantitatively evaluated using standardized goniometric measurements under blinded assessment. D Movement angle measurements: Spinal mobility and joint function were quantitatively evaluated using standardized goniometric measurements under blinded assessment. E–F qRT-PCR analysis of miR-944 and KLF4 expression in spinal tissues of the AS mouse model with circ_0018814 knockdown.
G–J Western blot analysis of osteogenic markers (ALP, RUNX2, Col1a1) in spinal tissues of the AS mouse model with circ_0018814 knockdown. **P < 0.01 versus control group, ##P < 0.01 versus AS + Lv-si-NC group
Discussion
AS is a chronic progressive inflammatory disease primarily affecting the spine and sacroiliac joints, leading to spinal stiffness and limited mobility (Chen et al. 2021). In recent years, the osteogenic differentiation process of MSCs in AS has garnered significant attention (Papagoras et al.
2021). CircRNAs, a novel class of non-coding RNAs, play crucial roles in various biological processes, including cell differentiation, proliferation, and apoptosis (Zhao et al. 2022; Zou et al.
2023). This study suggests that circ_0018814 is upregulated during the osteogenic differentiation of MSCs in AS. It regulates the expression of KLF4 by targeting miR-944, thereby promoting the osteogenic differentiation of MSCs.
This discovery not only provides a new molecular basis for the pathogenesis of AS but also offers theoretical support for the future development of circRNA-targeted therapeutic strategies.
The regulatory roles of circRNAs in various diseases have been extensively studied (Qian et al. 2025). For instance, in Rheumatoid Arthritis (RA), circ_0000396 promotes the proliferation and inflammatory response of synovial fibroblasts by regulating miR-203 (Wang et al.
2021). In osteoarthritis (OA), circ_0005105 modulates chondrocyte apoptosis and inflammatory responses by targeting miR-26a (Wu et al. 2017).
Circ_0003307 is significantly upregulated in AS patients and positively correlates with disease activity markers (ESR, CRP, BASDAI, and BASFI). Mechanistically, Circ_0003307 promotes inflammation by activating the PI3K/AKT pathway and upregulating pro-inflammatory cytokines (TNF-α and TNFAIP2) in fibroblast-like synoviocytes (Fang et al. 2022).
These studies indicate that circRNAs play a significant regulatory role in inflammatory joint diseases. In this study, we explored the potential involvement of circ_0018814 in regulating AS progression. The results showed that circ_0018814 is significantly upregulated during the osteogenic differentiation of MSCs in AS patients.
Furthermore, the knockdown of circ_0018814 inhibited the osteogenic differentiation of AS-derived MSCs, while its overexpression promoted this process. Recent studies have also provided broader context for pathological bone remodeling, inflammatory regulation, immune-related biological regulation, extracellular-vesicle/non-coding-RNA-mediated intercellular communication, and biomarker research in inflammatory diseases (Zhang et al. 2026; Yu et al.
2026).
CircRNAs function as miRNA sponges, thereby regulating the biological functions of miRNAs. miRNAs are a class of small non-coding RNAs that participate in various biological processes by regulating the expression of target genes (Zheng et al. 2021). In AS, abnormal miRNA expression is closely related to the onset and progression of the disease (Sirufo et al.
2021). For example, miR-21 promotes osteogenic differentiation in AS patients by regulating JAK2/STAT3 Pathway (Zou et al. 2020), while miR-155 inhibits the inflammatory response in AS patients by regulating SOCS3 expression (Jahangir et al.
2023). In this study, we found that circ_0018814 targets miR-944, and their co-localization in the cytoplasm further suggests that circ_0018814 may sponge miR-944. The role of miR-944 in AS has not been fully elucidated, but our results indicate that miR-944 expression inhibits the osteogenic differentiation of MSCs.
This finding provides new insights into the function of miR-944 in AS.
KLF4 is a transcription factor that plays important roles in various biological processes, including cell differentiation, proliferation, and apoptosis (Sousa et al. 2020; Shankman et al. 2015).
The regulatory role of KLF4 in AS-related diseases has been widely studied. For example, in RA, KLF4 inhibits the inflammatory response of synovial fibroblasts by regulating IL-6 expression (Luo et al. 2016).
In OA, KLF4 suppresses chondrocyte apoptosis and inflammatory responses by regulating MMP13 expression (Li et al. 2022). Previous study suggested that KLF4 downregulates FGF21 to activate inflammatory injury and oxidative stress via SIRT1/NF‑κB/p53 signaling in an AS inflammatory injury model.
In this study, the results showed that KLF4 promoted osteogenic differentiation of MSCs. Moreover, we found that circ_0018814 regulates the osteogenic differentiation of MSCs through the miR-944/KLF4 axis. Specifically, circ_0018814 sponges miR-944, thereby upregulating KLF4 expression and promoting the osteogenic differentiation of MSCs.
This discovery not only reveals the regulatory role of KLF4 in AS but also provides a new perspective on the function of the circRNA/miRNA/KLF4 axis in AS.
The role of KLF4 in osteogenic differentiation has been a subject of debate, with reports indicating both promotive and inhibitory effects (Yu et al. 2023; Miyamoto et al. 2025).
This apparent contradiction may stem from the highly context-dependent nature of KLF4’s function, influenced by factors such as cell type, differentiation stage, and the intricate crosstalk with other signaling pathways. In the present study, we demonstrate that circ_0018814 silencing inhibited osteogenic differentiation, whereas miR-944 inhibition or KLF4 overexpression partially reversed this inhibitory effect, as evidenced by decreased ALP activity, mineralized nodule formation, and expression of key markers like RUNX2. This supports a model in which KLF4 acts as a positive regulator of osteogenesis within the specific context of the circ_0018814/miR-944 axis.
Our findings suggest a potential mechanistic explanation for these discrepant reports. KLF4 is a multifaceted transcription factor that can either activate or repress gene expression depending on its interacting partners and post-translational modifications. It is plausible that in different experimental systems, KLF4 is integrated into distinct regulatory complexes, leading to opposite outcomes.
The novel regulation of KLF4 by the circ_0018814/miR-944 ceRNA network, as uncovered here, represents one such specific context that favors its pro-osteogenic function, potentially by modulating its expression levels or altering the cellular milieu in which it operates.
Nevertheless, the precise molecular mechanisms by which KLF4, in turn, transcriptionally regulates downstream osteogenic genes (e.g., whether it directly binds to the promoters of RUNX2 or SP7) remain to be fully elucidated. Future studies employing ChIP-seq and detailed promoter analyses will be crucial to map the direct targets of KLF4 in this pathway and to reconcile its context-specific roles in bone biology.
Despite these findings, this study has some limitations. First, it only explored the regulation of MSC osteogenic differentiation by circ_0018814 through the miR-944/KLF4 axis, without investigating other potential regulatory mechanisms. Future studies should further explore whether circ_0018814 regulates MSC osteogenic differentiation through other miRNAs or signaling pathways.
Additionally, the sample size in this study was relatively small, and future research should expand the sample size to enhance the reliability and generalizability of the results. Finally, this study did not explore the clinical significance of circ_0018814 in AS patients. Future research should analyze clinical samples to investigate the relationship between circ_0018814 expression levels and the clinical characteristics and prognosis of AS patients.
The GEO dataset was used as an exploratory screening resource, and the associated findings should not be interpreted as confirmatory evidence alone. The control group consisted of patients with femoral neck fractures rather than population-based healthy volunteers, and baseline differences may have affected MSC biological behavior. The circular structure of circ_0018814 was not experimentally validated in the present study using assays such as divergent-primer amplification, back-splice junction sequencing, RNase R resistance testing, or transcript stability analysis.
We further confirmed that Lv-si-hsa_circ_0018814 effectively reduced circ_0018814 expression in lesion-related tissue-derived primary MSCs in vivo, which strengthens the association between circ_0018814 knockdown and the observed phenotypic improvement. Nevertheless, because tail vein-delivered lentiviral intervention is not MSC-specific, systemic effects on other cell populations cannot be excluded. Future studies with larger cohorts, more rigorous circRNA validation, and cell-specific in vivo approaches are warranted.
In addition, broader biomarker-oriented investigations integrating molecular and computational strategies across inflammatory diseases may further improve the translational relevance of candidate targets (Yu et al. 2025). The selection of the BALB/c mouse strain for the PG-induced spondylitis model was based on its well-established role as the gold-standard model for this approach.
Unlike the C57BL/6 strain, which is highly resistant to PG-induced disease due to its Th1-dominant immune profile, BALB/c mice exhibit a pronounced Th2 bias that confers exceptional susceptibility to the development of autoimmune spondylitis and arthritis (Finnegan et al. 1999), mirroring several key pathological features of human AS (Shen et al. 2022; Yang et al.
2025). This high sensitivity provided a robust and reproducible in vivo system to interrogate the role of the circ_0018814/miR-944/KLF4 axis in osteogenic differentiation and pathological bone formation.
However, the very attribute that makes this model powerful also defines its primary limitation. The strong Th2-polarized microenvironment of the BALB/c mouse may not fully represent the more heterogeneous immune landscape seen in the human AS patient population, where Th1 and Th17 pathways are also critically involved. Consequently, while our findings definitively establish a pathogenic role for circ_0018814 within this specific model context, the generalizability of our results to other immunological milieus requires further validation.
Future studies employing complementary models, such as HLA-B27 transgenic rats or mice on different genetic backgrounds, will be essential to confirm the broader relevance of this molecular axis in AS pathogenesis and to assess its potential as a universal therapeutic target.
Conclusion
In summary, this study suggests that circ_0018814 promotes MSC osteogenic differentiation in AS by targeting miR-944 to regulate KLF4 expression. This finding provides a potential molecular basis for the pathogenesis of AS and offers theoretical support for the future development of circRNA-targeted therapeutic strategies. Nevertheless, the in vivo mechanistic interpretation should remain cautious because the delivery strategy was not MSC-specific.
Supplementary Information
Below is the link to the electronic supplementary material.
Supplementary Fig. 1. The expression of specific surface markers (CD73, CD90, CD105, CD14, CD45 and CD34) was assessed by flow cytometric analysis in the isolated MSCs
Supplementary Fig. 2. The effect of KLF4 on the expression of OPN, SP7, and BMP2.
(A-D) Western blot analysis of osteogenic markers (OPN, SP7, BMP2) in AS-MSCs with KLF4 knockdown or overexpression. **P < 0.01 vs. Blank group, ##P < 0.01 vs. OM + si-NC group, ##P < 0.01 vs.
OM+vector group.
Supplementary Fig. 3. The effects of the circ_0018814/miR-944/KLF4 axis on MSC viability and proliferation.
(A) The cell viability was detected by CCK8 assay. (B) Quantification of cell proliferation measured by EdU assay. (C) Representative images of EdU staining with EdU (green) and DAPI (blue) shown. **P < 0.01 vs. control group, ##P <.01 vs.
Lv-si-hsa_circ_0018814 group.
Supplementary Fig. 4. Circ_0018814 regulated the expression of ALP, RUNX2, COL1A1, OPN, SP7, and BMP2 by miR-944/KLF4 axis.
(A-D) Western blot analysis of osteogenic markers (OPN, SP7, and BMP2) in AS-MSCs with circ_0018814 knockdown and rescue experiments (miR-944 inhibitor or KLF4 overexpression). (E-J) qRT-PCR analysis of osteogenic markers (ALP, RUNX2, COL1A1, OPN, SP7, and BMP2) in AS-MSCs with circ_0018814 knockdown and rescue experiments
Supplementary Table 1. The results of the differential circRNA expression analysis performed on the GSE203155 dataset were used to generate the volcano plot
Supplementary Table 2. The primer sequences were used for qPCR in this study
Supplementary Material 3