The clinical role of lncRNA PRKG1-AS1 in lumbar disc degeneration and its mechanism in regulating inflammation and ferroptosis via miR-218-5p

Abstract

Background

Lumbar disc degeneration (LDD) is a common spinal disorder that predisposes patients to lumbar disc herniation (LDH) and causes chronic low back pain. Its pathogenesis remains incompletely understood. This study investigated the clinical relevance of PRKG1-AS1 in LDD and explored its regulation of LDD progression through miR-218-5p, providing experimental support for LDD targeted treatment.

Methods

This study enrolled 128 LDD patients and 102 healthy individuals to measure serum levels of PRKG1-AS1 and analyze its associations with clinical parameters. A degeneration model of human nucleus pulposus cells (hNPCs) was established using TNF-α induction. The effects of PRKG1-AS1 on cell proliferation, metabolic balance, ferroptosis, inflammation, and oxidative stress were assessed through transfection experiments. Additionally, dual-luciferase reporter assays confirmed the targeted binding interactions.

Results

Serum PRKG1-AS1 showed strong diagnostic value for LDD. Its expression was closely related to disease severity and patient functional status, making it an independent risk factor for progression from LDD to LDH. Overexpressing PRKG1-AS1 significantly improved the TNF-α-induced degenerative phenotype, enhanced hNPCs proliferation, restored metabolic balance, reduced ferroptosis, and alleviated inflammation and oxidative stress damage. Dual-luciferase assays confirmed that PRKG1-AS1 directly binds to miR-218-5p, and miR-218-5p targets CUL3. PRKG1-AS1 exerted a protective effect in LDD progression via the miR-218-5p/CUL3 axis.

Conclusion

Serum PRKG1-AS1 may serve as a promising biomarker for early detection and prediction of LDD progression. It modulates inflammation and ferroptosis in nucleus pulposus cells through the PRKG1-AS1/miR-218-5p/CUL3 axis, thereby inhibiting the LDD development.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13018-026-07004-x.

Background

Lumbar disc degeneration (LDD) is the most prevalent degenerative disorder of the spinal column. Progressive LDD may lead to lumbar disc herniation (LDH), which is associated with chronic low back pain, radicular leg pain, and neurological deficits. These symptoms collectively impair patients’ functional capacity and quality of life [1]. Although MRI is widely used in clinical practice to evaluate LDD, it lacks sufficient sensitivity to detect molecular and microstructural changes that appear before obvious morphological abnormalities [2]. Furthermore, the pathogenesis of LDD remains unclear, and there is currently no cure. Clinical treatment mainly focuses on relieving symptoms [3]. Therefore, exploring the mechanisms and key targets of LDD progression is highly important for optimizing clinical diagnosis and treatment.

Long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) are core components of gene expression regulatory networks. Their abnormal expression is closely associated with musculoskeletal diseases, providing new directions for disease diagnosis and treatment [4]. The lncRNA PRKG1-AS1 is highly expressed in various malignant tumors, including lung adenocarcinoma and oral squamous cell carcinoma [5, 6]. It is not only an independent risk factor for evaluating patient prognosis, but also participates in tumor progression by regulating malignant cell phenotypes, vascular development, and the ceRNA network [7, 8]. In aged skeletal muscle, PRKG1-AS1 expression is increased, and knocking it down enhances muscle cell viability and upregulates genes related to muscle differentiation [9], suggesting a conserved role in regulating cellular function. Bioinformatics analysis revealed that PRKG1-AS1 is a core molecule in the LDD ferroptosis-related regulatory network and is lowly expressed in TNF-α-induced nucleus pulposus cells [10]. Given its key regulatory role in various diseases and potential association with LDD pathogenesis, exploring the clinical significance and regulatory mechanism of PRKG1-AS1 in LDD is expected to provide new targets for disease diagnosis and treatment.

As a key regulatory miRNA in musculoskeletal disorders, miR-218-5p has been functionally validated in multiple pathological contexts. In osteoarthritis (OA), miR-218-5p is significantly up-regulated in cartilage tissue and chondrocytes. It promotes the secretion of pro-inflammatory cytokines and disrupts bone metabolic homeostasis by directly targeting the PI3K/Akt/mTOR pathway [11–13]. Notably, inhibiting miR-218-5p has been shown to improve TNF-α-induced chondrocyte dysfunction [14]. In rheumatoid arthritis (RA), miR-218-5p is also overexpressed in synovial tissue and fibroblast-like synoviocytes. It contributes to abnormal cell proliferation, elevated oxidative stress, and dysregulated bone remodeling by modulating the Wnt/β-catenin and JAK2/STAT3 signaling axes [15, 16]. Conversely, in postmenopausal osteoporosis, miR-218-5p plays a protective role by enhancing osteogenic differentiation of bone marrow-derived mesenchymal stem cells [17, 18], highlighting its context-dependent but consistently pivotal role in skeletal homeostasis. Bioinformatic analyses strongly predict a sequence-specific interaction between miR-218-5p and PRKG1-AS1. However, experimental validation of their functional regulation and potential involvement in LDD pathogenesis has not been reported.

Based on the above evidence, the study examined the expression characteristics of PRKG1-AS1 in the serum of LDD patients. The correlation between PRKG1-AS1 expression and clinical indicators as well as disease progression was analyzed. This study also explored the molecular mechanism by which PRKG1-AS1 targets miR-218-5p to regulate the biological function of nucleus pulposus cells using cell models. The aim was to provide an experimental basis for LDD assessment and targeted therapy.

Methods

Research subjects and grouping

A total of 128 patients diagnosed with LDD at our institution from March 2022 to June 2024 were enrolled as the patient cohort. The included patients were aged 18 years or older. The patient experienced chronic lower back pain lasting three months or more, or symptoms related to nerve root involvement. Inclusion required confirmation of LDD via lumbar MRI, and exclusion of prior spinal surgery or immunomodulatory therapy. All participants underwent lumbar spine MRI using a 1.5 T or 3.0 T scanner with a phased-array spine coil. Standardized sagittal T1-weighted, sagittal, and axial T2-weighted sequences were obtained. Slice thickness was 4 mm with 0.4 mm inter-slice gap. The index lumbar level involved in degeneration was determined by consensus reading of lumbar MRI scans by two senior radiologists following the Pfirrmann classification criteria. The Pfirrmann grade was assessed for each lumbar intervertebral disc from L1–L2 to L5–S1 on sagittal T2-weighted images. The most severely degenerated disc level was designated as the index level for each patient. In cases of equal severity across multiple levels, the most symptomatic level (corresponding to the patient’s reported pain location and neurological findings) was selected as the index level. Inter-rater reliability between the two radiologists was assessed using Cohen’s kappa coefficient (κ = 0.82), indicating substantial agreement. Discrepancies were resolved by consensus discussion. Disease severity was stratified using the Pfirrmann classification system: 32 patients were Grade II, 40 Grade III, 35 Grade IV, and 21 Grade V. Among these patients, 67 met additional diagnostic criteria for LDH, defined as radicular symptoms (including lower-limb radiating pain and paresthesia) combined with MRI-confirmed structural abnormalities such as disc protrusion and compression of the nerve root or dural sac.

Concurrently, 102 asymptomatic volunteers undergoing routine annual health screening at the same hospital during the same study period (March 2022 to June 2024) served as the healthy control cohort. Controls were age- and sex-matched to the patient cohort. All controls underwent lumbar MRI to exclude occult disc degeneration (Pfirrmann grade I) and other spinal pathologies. All controls had no spinal pathology, musculoskeletal disorders, or systemic inflammatory conditions known to affect the biomarkers under study. None reported chronic lower back pain, bilateral leg discomfort, or prior diagnosis of degenerative disc disease, and none had prior spinal surgery or trauma. Controls were recruited via consecutive sampling from the health screening registry and provided written informed consent prior to enrollment.

Exclusion criteria applied uniformly across both groups and encompassed coexisting severe spinal pathologies (e.g., spinal stenosis, spondylolisthesis, spinal tumors, infections, or trauma), prior lumbar spinal intervention or trauma within the preceding six months, pregnancy or breastfeeding, active autoimmune (e.g., rheumatoid arthritis, ankylosing spondylitis) or primary neurological conditions, systemic inflammatory conditions known to influence the biomarkers under investigation, and ongoing treatment with corticosteroids or other immunosuppressive agents. This was an observational cross-sectional study. All participants provided written informed consent, and their clinical records were fully documented. The study protocol was approved by the Institutional Review Board of The Fourteenth Medical Center, Chinese PLA General Hospital, and was performed in accordance with the principles of the Declaration of Helsinki.

Collection of clinical data and serum samples

Data were collected using standardized questionnaires and retrospective review of electronic medical records. Variables included demographic characteristics (age, sex, height, weight, tobacco and alcohol use), metabolic conditions (e.g., diabetes mellitus and dyslipidemia), and symptom-related features, especially the anatomical location of pain. The index lumbar level involved in degeneration was determined by consensus reading of lumbar MRI scans by 2 senior radiologists. Clinical functional status was assessed using 3 validated tools: the Visual Analogue Scale (VAS) for low back pain intensity, the Japanese Orthopedic Association (JOA) score for neurological impairment and functional limitation, and the Oswestry Disability Index (ODI) for activity-related disability. For patients with concurrent radicular leg pain, VAS scores were recorded separately for low back pain and leg pain; the low back pain VAS score was used for primary correlation analyses. To ensure inter-rater reliability, all assessments were performed independently by 3 trained clinical researchers following a standardized protocol; the final scores were the arithmetic mean of the 3 evaluations.

Venous blood samples (3 mL, fasting) were collected from LDD patients on the second day of hospitalization and from healthy controls during their scheduled morning health check-up. A portion of each sample was immediately processed for routine hematological and inflammatory biomarker analysis, including white blood cell count (WBC), C-reactive protein (CRP), and serum ferritin, using a fully automated hematology analyzer. CRP and ferritin were selected as representative markers of systemic inflammation and iron metabolism status, respectively, given their established clinical utility and relevance to LDD pathophysiology (chronic inflammation and ferroptosis). All assays were performed using the same batch of reagents by a single laboratory technician to minimize inter-batch variability. CRP and ferritin were measured simultaneously with serum PRKG1-AS1 at a single time point. As acute-phase reactants, their levels may fluctuate with recent inflammatory events. We did not collect longitudinal measurements to assess temporal stability or within-subject variation. The remaining blood was allowed to clot at room temperature for 30 min, then centrifuged at 3,000 × g for 10 min at 4 °C. The resulting supernatant (serum) was aliquoted and stored at − 80 °C for subsequent molecular assays.

Cell culture and degeneration model construction

Human nucleus pulposus cells (hNPCs; Sunncell, China) were cultured in DMEM/F12 (Procell, China) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Solarbio, China), 100 U/mL penicillin, and 100 µg/mL streptomycin (Solarbio, China) under standard conditions (37 °C, 5% CO₂, humidified atmosphere). Cells were subcultured when they reached approximately 80% confluence. Only low-passage (P3–P5), exponentially growing hNPCs were used for experiments.

To establish an inflammatory degeneration model, confluent hNPCs (80–90% density) were thoroughly rinsed with pre-warmed phosphate-buffered saline (PBS), then incubated with fresh complete medium containing 25 ng/mL human tumor necrosis factor-alpha (TNF-α; Invitrogen, USA). After 48 h of stimulation, cells were harvested for downstream molecular and functional assays.

Cell transfection

The PRKG1-AS1 overexpression plasmid (oe-PRKG1-AS1), miR-218-5p mimics and inhibitors, and their corresponding negative controls (NCs) were custom-designed and synthesized by Ribobio (Guangzhou, China). hNPCs were uniformly seeded into 6-well plates at a predetermined optimal density for transfection. When cell confluence reached approximately 60%, transfection was performed using the TransIT-X2™ Dynamic Delivery System (Mirus Bio, USA), strictly following the manufacturer’s instructions for reagent ratios and procedures. Briefly, nucleic acid–transfection reagent complexes were prepared by gentle mixing and allowed to equilibrate at room temperature for 30 min before application. The mixture was then added dropwise to the cultured cells, followed by gentle rocking of the plate to ensure even distribution across each well. Cells were harvested 48 h after transfection for subsequent molecular characterization and functional evaluation.

Cell proliferation assay

Cell proliferative capacity was assessed using the Cell Counting Kit-8 (CCK-8) assay. After transfection, cells from each experimental group were seeded into 96-well plates at a density of 2 × 10⁴ cells per well and allowed to adhere overnight under standard culture conditions (37 °C, 5% CO₂). At designated time points—0, 24, 48, 72, and 96 h after seeding—10 µL of CCK-8 solution (Yeasen, China) was added to each well. Plates were then incubated at 37 °C for 2 h in the dark. The optical density (OD) value at 450 nm was measured using a BioTek ELx800 microplate reader (BioTek, USA).

Gene expression detection

Total RNA was extracted from serum samples and cultured cells using TRIzol^® reagent (Invitrogen, USA), followed by strict quality control to evaluate RNA concentration, purity, and integrity. High-quality RNA was reverse-transcribed using the HiScript^® IV 1st Strand cDNA Synthesis Kit (with gDNA removal module) (Vazyme, China) according to the manufacturer’s instructions.

Quantitative PCR (qPCR) was performed on an FQD-96 C real-time fluorescence detection system (Bioer, China) using SupRealQ™ Ultra Hunter SYBR Green qPCR Master Mix (U+) (Vazyme, China). For lncRNA and mRNA targets (including PRKG1-AS1, COL2A1, ACAN, TEK, GPX4, ACSL4, SLC7A11, NRAS, PUM2, BIRC6, CUL3, and DDX6), GAPDH served as the endogenous control. U6 snRNA was used for normalization of miR-218-5p expression. Relative expression was calculated using the comparative 2^−ΔΔCt method. All gene-specific primers were custom-designed and synthesized by GENEWIZ (Suzhou, China).

Detection of free ferrous ion

The ferrous ion content in each group of cells was detected using a Ferrous ion assay kit (Solarbio, China). After cell collection, 1 mL of extraction solution was added, and cells were sonicated in an ice bath (power: 200 W; sonication time: 5 s; interval: 5 s; total sonication time: 5 min). Cells were then centrifuged at 10,000 × g for 10 min at 4 °C to obtain the supernatant. Standard samples, sample supernatants, and blank controls were sequentially added to a 96-well plate. The corresponding volumes of reaction solution were added to each well. The plate was gently mixed and incubated at 37 °C for 20 min. The absorbance of each well was measured at 593 nm using a BioTek ELx800 microplate reader preheated for more than 30 min. The ferrous ion concentration in the sample was then calculated using the standard curve fitting equation.

Inflammation and oxidative stress essay

Cells from each experimental group were harvested and centrifuged at 3,000 × g for 5 min at 4 °C; the supernatant was carefully collected, and the cell pellet was discarded. A volume of 100 µL of supernatant was transferred to designated wells of a 96-well ELISA plate and incubated at 37 °C for 90 min. Secreted pro-inflammatory mediators, including interleukin-1β (IL-1β), interleukin-6 (IL-6), and monocyte chemoattractant protein-1 (MCP-1), were performed using commercial ELISA kits (SpBio, China).The release levels of oxidative stress markers ROS and GSH (CoiBo, China) were also measured. After antibody binding and chromogenic reaction, the OD value at 450 nm was measured using a BioTek ELx800 microplate reader.

Target gene screening

Potential target genes of miR-218-5p were computationally predicted using multiple public resources, including TargetScan, miRTarBase, miRDB, and StarBase. Overlapping genes across all 4 databases were selected for further study. Subsequently, a protein–protein interaction (PPI) network was generated using the STRING platform, and the 5 genes with the greatest connectivity (i.e., highest degree centrality) were designated as hub genes.

Verification of targeted binding relationship

Predicted binding sites between PRKG1-AS1 and miR-218-5p, as well as between CUL3 and miR-218-5p, were identified using the StarBase platform. Wild-type (wt) fragments of PRKG1-AS1 and CUL3 (each containing the predicted miR-218-5p binding regions) and their corresponding site-directed mutant (mut) versions using a commercial kit (Macklin, China), were individually cloned into the pmiR-GLO dual-luciferase reporter vector (YouBio, China). hNPCs were plated in 24-well plates and transfected at ~ 70% confluence with the reporter constructs together with miR-218-5p mimics, inhibitors, or corresponding NCs. Luciferase activity was measured 48 h after transfection using a dual-luciferase assay system (Solarbio, China) on a multimodal microplate reader (Myrevity, USA). Renilla luciferase signal was used as an internal control to normalize firefly luciferase readings across samples.

Statistical analysis

Statistical analyses were performed using GraphPad Prism software. Normality and homogeneity of variance were assessed using the Shapiro-Wilk test and Levene’s test, respectively. All continuous variables were confirmed to follow a normal distribution with homogeneous variances. Accordingly, parametric tests were applied: continuous variables were assessed with either unpaired T-test (for two-group comparisons) or one-way/two-way ANOVA (among multiple groups) followed by Tukey’s post hoc test. Categorical data were analyzed using the chi-square test. Correlations between continuous variables were examined using the Pearson correlation coefficient. Diagnostic performance was assessed using receiver operating characteristic (ROC) curve analysis, and multivariate logistic regression modeling was used to identify independent risk factors. A two-sided P value < 0.05 was considered statistically significant.

Results

Baseline clinical characteristics of study participants

There were no significant differences between the two groups in age, sex, BMI, smoking status, alcohol consumption, or histories of diabetes and hyperlipidemia (P > 0.05), confirming baseline comparability and supporting reliable downstream analyses. These variables were included as covariates in all multivariate regression analyses to minimize residual confounding. Laboratory assessments revealed markedly elevated serum concentrations of CRP and ferritin in the LDD cohort compared with controls (P < 0.001). Whether this elevation represents a contributing factor to LDD pathogenesis or a secondary response to degenerative tissue damage cannot be determined from the current cross-sectional data. Clinical symptom distribution showed unilateral low back pain in 79 patients (61.72%) and bilateral involvement in 49 (38.28%). The most frequently affected spinal level was the L4/L5 lumbar segment, identified in 71 cases (55.47%, Table 1).

Table 1.

Clinical baseline characteristics across study cohorts

Parameters	Healthy control (N = 102)	LDD patient (N = 128)	P value
Age (years)	50.18 ± 9.90	52.34 ± 10.17	0.107
Gender (Male, %)	64 (62.75)	91 (71.09)	0.180
BMI (kg/m2)	21.82 ± 2.25	22.40 ± 3.19	0.119
Smoking History (Yes, %)	37 (36.27)	59 (46.09)	0.134
Drinking History (Yes, %)	45 (44.12)	61 (47.66)	0.593
Diabetes History (Yes, %)	41 (40.20)	60 (46.88)	0.311
Hyperlipoidemia History (Yes, %)	39 (38.24)	55 (42.97)	0.468
WBC (×109/L)	6.26 ± 1.31	6.41 ± 1.32	0.391
CRP (mg/L)	3.99 ± 1.75	5.76 ± 2.35	< 0.001
Ferritin (ng/mL)	118.00 ± 52.46	143.80 ± 53.95	< 0.001
Pain Location n, %			–
Unilateral	–	79 (61.72)
Bilateral	–	49 (38.28)
Lumbar Segment n, %			–
L4/L5	–	71 (55.47)
L5/S1	–	57 (44.53)

Notes LDD, lumbar disc degeneration; BMI, body mass index; WBC, white blood cell count; CRP, C-reactive protein; P < 0.05 indicated that there was a statistically significant difference in the data results

The expression characteristics and diagnostic value of serum PRKG1-AS1

Serum PRKG1-AS1 levels were significantly lower in LDD patients than in healthy controls (P < 0.001, Fig. 1A). ROC analysis showed that serum PRKG1-AS1 distinguished LDD patients from controls with an AUC of 0.851, with 77.34% sensitivity and 83.33% specificity (P < 0.001, Fig. 1B), indicating its potential as a diagnostic biomarker for LDD.

Fig. 1.

The serum PRKG1-AS1 expression in healthy control and LDD groups (A). The diagnostic efficiency of serum PRKG1-AS1 for LDD via ROC curve (B). The serum PRKG1-AS1 expression in LDD patients with different Pfirrmann grades (C). The correlation between serum PRKG1-AS1 expression and VAS (D), JOA (E), and ODI (F) scores in LDD patients. *P < 0.05; **P < 0.01; ***P < 0.001

Across Pfirrmann grades II–V, serum PRKG1-AS1 expression gradually decreased with increasing disc degeneration severity, and pairwise comparisons between all adjacent grades were statistically significance (P < 0.05, Fig. 1C).

Correlation analyses revealed strong associations between serum PRKG1-AS1 and clinical outcomes: it was inversely correlated with low back pain intensity (VAS score for axial pain; r = − 0.679, P < 0.001, Fig. 1D), positively correlated with functional status (JOA score; r = 0.552, P < 0.001, Fig. 1E), and inversely correlated with disability severity (ODI score; r = − 0.571, P < 0.001, Fig. 1F).

Association between serum PRKG1-AS1 expression and progression from LDD to LDH

To control for potential confounding, multivariate logistic regression was performed with adjustment for age, sex, BMI, smoking status, alcohol consumption, diabetes mellitus, and dyslipidemia. Multivariate logistic regression identified both elevated CRP (OR = 3.150, P = 0.001) and increased serum ferritin (OR = 2.085, P = 0.023) as independent predictors of LDD occurrence, alongside reduced PRKG1-AS1 expression (OR = 0.128, P < 0.001) (Supplementary Material 1). Notably, PRKG1-AS1 exhibited the strongest effect size (lowest OR) among all significant predictors.

Serum PRKG1-AS1 levels were significantly lower in patients with LDH than in those with uncomplicated LDD (P < 0.001, Fig. 2A). Multivariate logistic regression, (adjusted for the aforementioned covariates) identified several independent predictors of LDH in LDD patients: elevated CRP (OR = 3.039, P = 0.044), higher VAS pain scores (OR = 4.402, P = 0.006), lower JOA functional scores (OR = 0.272, P = 0.018), greater ODI disability scores (OR = 3.441, P = 0.022), and reduced serum PRKG1-AS1 expression (OR = 0.148, P < 0.001). Notably, low PRKG1-AS1 showed the strongest association with LDH progression. Additionally, higher BMI (OR = 2.855, P = 0.055) and increased serum ferritin (OR = 2.430, P = 0.098) showed trends toward association with LDH risk but did not reach statistical significance (Fig. 2B). In the multivariate logistic regression model predicting LDH progression, this study included 16 covariates with 67 LDH events, yielding an events-per-variable (EPV) ratio of approximately 4.2. Although this falls below the conventional threshold of 10 EPV, several measures were taken to mitigate overfitting concerns and validate estimate stability. First, collinearity diagnostics confirmed minimal multicollinearity among predictors (all variance inflation factors < 1.32; Supplementary Material 2). Second, bootstrap resampling (1,000 iterations) demonstrated that the regression estimates—particularly for PRKG1-AS1 and other significant predictors—remained consistent with the original model, supporting the robustness of the findings (Supplementary Material 3).

Fig. 2.

The serum PRKG1-AS1 expression in LDD and LDH groups (A). Prediction of risk factors associated with LDD progression to LDH via logistic regression model and forest plot (B). ***P < 0.001

Impact of PRKG1-AS1 overexpression on biological functions of cellular models of LDD

In the TNF-α–induced hNPCs degeneration model, PRKG1-AS1 expression was significantly suppressed (P < 0.001). Conversely, forced overexpression of PRKG1-AS1 strongly restored its levels (P < 0.001, Fig. 3A).

Fig. 3.

The effects of PRKG1-AS1 overexpression in TNF-α-induced hNPCs model on PRKG1-AS1 expression (A), cell proliferation (B), the mRNA levels of degeneration-related genes COL2A1, ACAN, and TEK (C), Fe²+ concentration (D), the mRNA levels of ferroptosis-related markers GPX4, ACSL4, and SLC7A11 (E), the relative concentrations of inflammatory cytokines IL-1β, IL-6, and MCP-1 (F), as well as the relative levels of oxidative stress markers ROS and GSH (G). **P < 0.01; ***P < 0.001

TNF-α exposure severely impaired hNPCs proliferative capacity (P < 0.001), whereas PRKG1-AS1 upregulation significantly rescued this defect (P < 0.001, Fig. 3B). At the transcriptional level, PRKG1-AS1 overexpression counteracted TNF-α–induced dysregulation of key extracellular matrix and degeneration-associated genes: it restored COL2A1 and ACAN mRNA expression and inhibited abnormal TEK upregulation (P < 0.001, Fig. 3C).

TNF-α triggered intracellular Fe²⁺ accumulation (P < 0.001), an effect effectively reversed by PRKG1-AS1 overexpression (P < 0.001, Fig. 3D). Detection of ferroptosis-related indicators showed that TNF-α downregulated GPX4 and SLC7A11 mRNA while increasing ACSL4 expression (P < 0.001), whereas PRKG1-AS1 overexpression normalized these changes (P < 0.01, Fig. 3E).

TNF-α also induced a strong pro-inflammatory response in hNPCs, as shown by significant increases in IL-1β, IL-6, and MCP-1 secretion (P < 0.001). PRKG1-AS1 overexpression significantly attenuated this cytokine surge (P < 0.001, Fig. 3F). Furthermore, TNF-α exacerbated oxidative stress—reflected by increased ROS production and decreased GSH levels (P < 0.001)—whereas PRKG1-AS1 overexpression reduced oxidative damage by inhibiting ROS accumulation and enhancing GSH synthesis (P < 0.01, Fig. 3G).

Targeted regulatory interaction between PRKG1-AS1 and miR-218-5p

qPCR analysis showed that serum miR-218-5p levels were significantly higher in LDD patients than in healthy controls (P < 0.001, Fig. 4A), and further increased in LDH patients compared with those with isolated LDD (P < 0.001, Fig. 4B). A strong inverse correlation was observed between serum PRKG1-AS1 and miR-218-5p expression in the LDD group (r = − 0.719, P < 0.001, Fig. 4C).

Fig. 4.

The serum miR-218-5p expression in healthy control and LDD groups (A), as well as in LDD and LDH groups (B). The correlation between serum PRKG1-AS1 and miR-218-5p expression in LDD patients (C). The predicted binding sites between PRKG1-AS1 and miR-218-5p (D). The verification of targeted binding relationship between PRKG1-AS1 and miR-218-5p via dual-luciferase reporter assay (E). ***P < 0.001

Bioinformatic prediction identified a putative binding site for miR-218-5p in the PRKG1-AS1 sequence (Fig. 4D). To experimentally validate this interaction, dual-luciferase reporter assays were performed. Co-transfection of the wt-PRKG1-AS1 reporter construct with miR-218-5p mimic led to a marked reduction in luciferase activity (P < 0.001), whereas co-transfection with miR-218-5p inhibitors significantly increased it (P < 0.001). In contrast, mutation of the predicted binding site (mut-PRKG1-AS1) abolished these effects; neither mimics nor inhibitors significantly altered luciferase activity (P > 0.05, Fig. 4E).

PRKG1-AS1 modulates biological functions of cellular models of LDD via miR-218-5p

TNF-α stimulation induced strong upregulation of miR-218-5p in hNPCs (P < 0.001). In this degeneration model, PRKG1-AS1 overexpression effectively suppressed this elevation (P < 0.001), whereas subsequent transfection with miR-218-5p mimics restored high miR-218-5p expression (P < 0.001, Fig. 5A).

Fig. 5.

The effects of co-overexpression PRKG1-AS1 and miR-218-5p in TNF-α-induced hNPCs model on miR-218-5p expression (A), cell proliferation (B), the mRNA levels of degeneration-related genes COL2A1, ACAN, and TEK (C), Fe²+ concentration (D), the mRNA levels of ferroptosis-related markers GPX4, ACSL4, and SLC7A11 (E), the relative concentrations of inflammatory cytokines IL-1β, IL-6, and MCP-1 (F), as well as the relative levels of oxidative stress markers ROS and GSH (G). **P < 0.01; ***P < 0.001

Functionally, PRKG1-AS1 overexpression enhanced hNPCs proliferation in the TNF-α–treated setting; however, this pro-proliferative effect was abolished by co-overexpression of miR-218-5p mimics (P < 0.001, Fig. 5B). Similarly, while PRKG1-AS1 restored physiological mRNA levels of matrix-related genes—reversing TNF-α–induced suppression of COL2A1 and ACAN and induction of TEK—these beneficial effects were lost when miR-218-5p was co-overexpressed (P < 0.001, Fig. 5C).

Regarding ferroptosis regulation, co-overexpression of PRKG1-AS1 and miR-218-5p restored intracellular Fe²⁺ accumulation (P < 0.01, Fig. 5D) and reversed the protective transcriptional profile: GPX4 and SLC7A11 mRNA were significantly downregulated, while ACSL4 was upregulated (P < 0.01, Fig. 5E).

Moreover, the anti-inflammatory and antioxidant effects conferred by PRKG1-AS1 were significantly counteracted by miR-218-5p overexpression: IL-1β, IL-6, and MCP-1 secretion increased (P < 0.001, Fig. 5F), and oxidative stress reappeared, as shown by elevated ROS and depleted GSH (P < 0.01, Fig. 5G).

Identification and experimental validation of target genes of miR-218-5p

Bioinformatic screening across four databases—TargetScan (1,102), miRTarBase (831), miRDB (1,084), and StarBase (1,357)—predicted potential targets of miR-218-5p. Integration of these predictions yielded 82 high-confidence candidate genes. Subsequent PPI network analysis using the STRING database identified the top five hub genes based on node degree centrality: NRAS, PUM2, BIRC6, CUL3, and DDX6 (Fig. 6A). qPCR validation in serum samples showed that NRAS, PUM2, BIRC6, and DDX6 mRNA levels were significantly higher in LDD patients than in controls, whereas CUL3 expression was markedly lower (P < 0.001, Fig. 6B). Notably, CUL3 downregulation is consistent with the canonical inhibitory mechanism of miRNA–target interaction. Further stratification showed that CUL3 expression was even lower in LDH patients than in those with uncomplicated LDD (P < 0.001, Fig. 6C).

Fig. 6.

The prediction of the target genes of miR-218-5p via TargetScan, miRTarBase, miRDB, and StarBase databases and the screening of the top 5 core genes with the highest node degrees through the String database (A). The mRNA levels of candidate target genes NRAS, PUM2, BIRC6, CUL3, and DDX6 in healthy control and LDD groups (B). The serum CUL3 expression in LDD and LDH groups (C). The correlation between CUL3 and miR-218-5p expression in LDD patients (D). The predicted binding sites between CUL3 and miR-218-5p (E). The verification of targeted binding relationship between CUL3 and miR-218-5p via dual-luciferase reporter assay (F). ***P < 0.001

A strong inverse correlation was observed between serum CUL3 and miR-218-5p levels in LDD patients (r = − 0.737, P < 0.001, Fig. 6D). Sequence analysis confirmed a conserved binding site for miR-218-5p in 3′UTR of CUL3 (Fig. 6E). Dual-luciferase assays provided functional evidence: co-transfection of wt-CUL3 reporter with miR-218-5p mimics suppressed luciferase activity (P < 0.001), while miR-218-5p inhibitors enhanced it (P < 0.001). In contrast, mutation of the predicted seed-binding region (mut-CUL3) abolished responsiveness to either mimics or inhibitors (P > 0.05, Fig. 6F).

Discussion

LDD is a prevalent, progressive spinal disorder and a leading cause of low back pain, imposing personal and societal burdens [19]. Current diagnosis relies on imaging and clinical assessment, which are insensitive to early degeneration and prone to overdiagnosis. While MRI effectively identifies advanced structural abnormalities in symptomatic patients, it lacks sensitivity for detecting molecular- and microstructural-level alterations that precede gross morphological changes [2]. Consequently, a substantial proportion of patients with early-stage degeneration or equivocal imaging findings may be underdiagnosed or inappropriately reassured [20]. Treatments are largely symptomatic and do not halt disease progression [21]. Identifying molecular targets that serve as both diagnostic biomarkers and degeneration regulators is therefore essential for enabling earlier, more precise interventions.

Recent studies have confirmed that non-coding RNAs play a central regulatory role in musculoskeletal diseases by modulating key processes such as extracellular matrix (ECM) metabolism and inflammatory responses [22, 23]. Specifically, miRNAs orchestrate proliferation and differentiation of stromal cells involved in ECM composition and regulate cytokine expression in tendon injuries [23], while their aberrant expression in osteoarthritic cartilage contributes to joint homeostasis disruption [24]. A recent meta-analysis further demonstrated that circulating miRNAs exhibit high diagnostic accuracy for osteoarthritis, with pooled sensitivity of 0.81 and specificity of 0.85, supporting their potential as non-invasive biomarkers in musculoskeletal disorders [25]. Among these, lncRNAs such as ZFAS1, HOTAIR, and OIP5-AS1 show significant potential in the clinical diagnosis and treatment of LDD [26]. They are closely linked to disease risk, severity grading, and inflammation levels [27–29], providing a novel molecular perspective for disease evaluation. Beyond lncRNAs and miRNAs, small interfering RNAs (siRNAs) and circular RNAs (circRNAs) have emerged as additional therapeutic modalities in musculoskeletal disorders: siRNAs can identify molecular targets and modulate inflammatory cytokines in rheumatoid arthritis [30], regulate metabolic processes and growth factors in osteoporosis [31], and influence tendon structural protein gene expression [32]; meanwhile, circRNAs represent a promising loop-structured RNA platform for osteoporosis therapy exploration [22]. Similarly, lncRNA CRNDE has been identified as a serum biomarker for delayed fracture healing, exerting protective effects on osteoblast viability through sponging of miR-29a-3p [33]. Notably, PRKG1-AS1 is involved in cell survival and functional regulation [7–9], and bioinformatics analyses identify it as a key molecule in the LDD-related ferroptosis regulatory network [10], although its specific clinical relevance in bone diseases remains understudied. Clinical sample analysis in this study revealed that PRKG1-AS1 was downregulated in the serum of LDD patients. Importantly, although inflammation (elevated CRP) and iron dyshomeostasis (elevated ferritin) were confirmed as independent risk factors for LDD occurrence through multivariate analysis, PRKG1-AS1 demonstrated the strongest predictive effect and added diagnostic value beyond these conventional markers, with its diagnostic accuracy, sensitivity, and specificity for LDD were all at higher levels, indicating its potential as an early screening biomarker. These findings align with emerging evidence that non-coding RNAs serve as robust diagnostic tools in bone-related diseases; for instance, miR-106a-5p has been shown to diagnose postmenopausal osteoporosis with high accuracy and is negatively correlated with serum ferritin levels [34], while miR-217 participates in postmenopausal osteoporosis progression by targeting OPG to regulate the RANKL/RANK pathway [35]. In clinical practice, serum PRKG1-AS1 measurement could be integrated into existing workflows to address specific diagnostic gaps: (i) screening asymptomatic individuals with occupational risk factors or genetic predisposition to identify subclinical degeneration; (ii) monitoring disease progression in patients with mild Pfirrmann grades (II–III) where MRI changes are subtle; and (iii) refining surgical candidacy assessments by identifying LDD patients at elevated risk of LDH progression who may benefit from earlier intervention. Its minimally invasive nature and quantitative output offer logistical advantages over repeated MRI surveillance. Furthermore, Pfirrmann grading is a primary measure of degeneration severity [36]. Serum PRKG1-AS1 expression was significantly correlated with clinical indicators such as the VAS score, ODI index, and JOA score, indicating that it can objectively reflect disease severity and patient functional status [37]. Notably, the VAS score in this study specifically assessed low back pain intensity rather than leg pain. This distinction is clinically relevant because axial low back pain and radicular leg pain in LDH may involve distinct pathophysiological mechanisms: the former primarily reflects discogenic inflammation and mechanical disruption, whereas the latter is predominantly driven by nerve root compression and radicular inflammation. The strong inverse correlation between PRKG1-AS1 and low back pain severity suggests that this biomarker may primarily mirror disc-intrinsic degenerative processes rather than nerve root involvement. Future studies should separately evaluate the association between PRKG1-AS1 and leg pain severity to further refine its clinical applicability. Further multi-dimensional analysis showed that serum PRKG1-AS1 expression was significantly correlated with these measures, suggesting it can objectively reflect disease severity and patient functional status. Importantly, low PRKG1-AS1 expression was an independent risk factor for progression from LDD to LDH. LDH is a progressive stage of LDD involving disc protrusion that can compress nerve roots or the dural sac [38], providing a new tool for predicting disease progression and guiding personalized treatment.

Pathologically, elevated serum CRP and ferritin in LDD patients confirm inflammation and iron dyshomeostasis are core drivers of hNPC dysfunction and disc degeneration [39]. CRP, as an acute-phase reactant, reflects the persistent low-grade systemic inflammation in LDD patients. Beyond being a biomarker, CRP is induced by IL-6 and TNF-α in the liver, forming a positive feedback loop that amplifies the inflammatory cascade [40]. In the degenerative disc microenvironment, TNF-α and IL-1β directly induce matrix-degrading enzymes (MMPs and ADAMTS) in nucleus pulposus cells and promote inflammatory infiltration [41]. The functional state of hNPCs—key to disc structural stability and ECM metabolism—directly affects disc health [42]. TNF-α, a pro-inflammatory factor highly expressed in the LDD degenerative microenvironment, induces pathological features in hNPCs including tissue degeneration and inflammatory infiltration during disease progression [43]. This study used it to establish a degeneration model. The proliferation capacity of hNPCs and balanced expression of key molecules involved in ECM synthesis and degeneration markers directly determine intervertebral disc function. TNF-α suppressed hNPC proliferation, reduced expression of ECM synthesis markers (COL2A1, ACAN), and increased expression of the degeneration marker TEK, highlighting pathological decline in nucleus pulposus cell function and disrupted ECM homeostasis. Overexpression of PRKG1-AS1 reversed these changes, indicating its potential to protect hNPC function and maintain ECM balance. Ferroptosis is a key regulatory mechanism in LDD progression [44]. TNF-α triggers free iron accumulation in hNPCs, downregulates the ferroptosis inhibitors GPX4 and SLC7A11, and upregulates the pro-ferroptotic factor ACSL4, activating the ferroptosis pathway [45]. PRKG1-AS1 overexpression effectively inhibits this process. Serum ferritin elevation not only indicates increased iron stores but also reflects intracellular iron dyshomeostasis. Free iron catalyzes lipid peroxidation via the Fenton reaction, initiating ferroptosis. Notably, ferritin itself is regulated by NF-κB signaling and participates in inflammatory responses, providing a pathological basis for the synergistic effects of CRP and ferritin as independent risk factors [46]. Inflammation and oxidative stress drive accelerated disc degeneration. TNF-α stimulation increases pro-inflammatory cytokines (IL-1β, IL-6), elevates ROS, and depletes GSH. PRKG1-AS1 significantly inhibits these changes, exerting anti-inflammatory and antioxidant effects. It is worth noting that inflammation and ferroptosis are not independent processes but interact closely: TNF-α activates NF-κB, which can transcriptionally regulate SLC7A11 expression, while ferroptosis releases damage-associated molecular patterns that further activate inflammasomes, forming an “inflammation-ferroptosis” vicious cycle [47–49]. The dual suppressive effect of PRKG1-AS1 on inflammatory cytokines and ferroptosis markers in this study suggests its protective role may involve interrupting this cycle, offering a new perspective for understanding the multi-mechanism coordinated pathology of LDD. Consistent with its conserved role in regulating cell function [7–9], this study proposes that PRKG1-AS1 protects against LDD through two mechanisms: inhibiting ferroptosis to prevent iron-dependent hNPC damage, and restoring the inflammation–oxidative stress balance to maintain cellular homeostasis.

Further mechanistic investigation focuses on the ceRNA network—a key mechanism by which lncRNAs regulate gene expression and a critical player in skeletal and muscle disorders [50]. This study identified a significant negative correlation between serum PRKG1-AS1 and miR-218-5p levels in LDD patients, validated by complementary binding sites and confirmed by dual-luciferase assay. Given miR-218-5p’s conserved role in promoting lesions via inflammatory pathway activation and bone metabolism dysregulation in OA and RA [11–16], and its persistently elevated serum levels in LDD and LDH patients in this study, we hypothesize it drives similar pathology in LDD. Notably, the therapeutic potential of miRNA-based interventions in intervertebral disc degeneration has been highlighted by recent studies showing that miR-204-5p protects against nucleus pulposus cell apoptosis by targeting the SSRP1/NF-κB pathway [51], and that lncRNA HCG18 promotes spinal tuberculosis-related disc destruction by modulating the hsa-miR-146a-5p/TGF-β1/SMADs axis [52]. Mechanistic studies revealed that in a degenerative model of hNPCs, co-transfection of miR-218-5p mimics and PRKG1-AS1 effectively counteracted the protective effect of PRKG1-AS1 overexpression on hNPC degeneration, indicating that PRKG1-AS1 alleviates LDD degenerative progression by targeting and adsorbing miR-218-5p. To clarify downstream regulation, this study identified CUL3—the core target of miR-218-5p —through cross-database validation and PPI network analysis. As a conserved CULLIN family protein, CUL3 regulates cell cycle progression, signal transduction, and stress responses. CUL3 dysfunction is linked to metabolic disorders and neuromuscular degeneration [53], and it mediates apoptosis and inflammation in hNPCs, contributing critically to intervertebral disc degeneration [54]. Clinical analysis showed reduced serum CUL3 expression in LDD patients—negatively correlated with miR-218-5p levels—and direct targeting by miR-218-5p, confirmed by dual-luciferase assay. We proposed that PRKG1-AS1 relieves the post-transcriptional inhibition of CUL3 by competitively binding miR-218-5p, thereby improving hNPC proliferation disorder, inhibiting ferroptosis, reducing inflammation and oxidative stress damage, and ameliorating the degenerative phenotype of hNPCs. This establishes a complete regulatory axis of PRKG1-AS1/miR-218-5p/CUL3 that participates in the pathological progression of LDD. Beyond its diagnostic utility, the PRKG1-AS1/miR-218-5p/CUL3 axis presents a rational target for therapeutic intervention. The central pathological processes regulated by this axis—ferroptosis and inflammatory amplification—are modifiable cellular events. PRKG1-AS1 overexpression in the cellular model effectively rescued degenerative phenotypes, suggesting that restoring PRKG1-AS1 levels, or pharmacologically mimicking its downstream effects (e.g., CUL3 stabilization or miR-218-5p inhibition), could represent disease-modifying strategies. However, several translational hurdles remain: lncRNA-based therapeutics face challenges in targeted delivery, stability, and off-target effects in vivo; furthermore, the optimal timing of intervention—whether at the pre-symptomatic stage or during active degeneration—requires prospective validation. Future studies should prioritize in vivo models to evaluate the efficacy and safety of axis-targeted therapies, and explore small-molecule or oligonucleotide approaches that can recapitulate the protective effects observed in cell culture.

This study used single-center clinical samples; thus, larger, multi-center validation is needed to confirm findings. Additionally, while we have outlined potential clinical applications and therapeutic directions for PRKG1-AS1, these remain hypothetical pending prospective validation. The cost-effectiveness of incorporating serum PRKG1-AS1 into routine clinical practice, relative to standard MRI protocols, has not been assessed. Although we adjusted for major confounders including age, sex, BMI, smoking, alcohol consumption, diabetes, and dyslipidemia in multivariate models, residual confounding from unmeasured variables (e.g., occupational physical loading, genetic predisposition, or dietary iron intake) cannot be entirely excluded. Future studies should incorporate comprehensive lifestyle questionnaires and genetic risk scores to further refine risk estimates. Furthermore, the relatively wide confidence intervals for some covariates (e.g., BMI, ferritin) reflect reduced precision inherent to limited EPV. Future prospective studies with larger event counts should employ penalized regression techniques (e.g., Firth’s bias-reduced logistic regression or LASSO) to further optimize model performance and validate PRKG1-AS1 as an independent predictor of LDD progression. As a cross-sectional study, we cannot establish causality or temporal order for the observed associations between CRP, ferritin, and LDD. Whether elevated inflammatory and iron markers are causative factors or secondary to degenerative changes remains to be verified in prospective cohort studies combined with causal inference methods such as marginal structural models or instrumental variable analysis. Moreover, although this study specified that VAS scores were assessed for low back pain, we did not systematically analyze the independent correlation between serum PRKG1-AS1 and leg pain severity. Given that axial pain and radicular pain may represent different pathophysiological processes in LDD and LDH, future research should evaluate whether PRKG1-AS1 exhibits differential associations with these distinct pain components, which would help to more precisely define the clinical scope of this biomarker. Cell models recapitulate key pathological features but cannot fully mimic the native intervertebral disc microenvironment. Therefore, in vivo validation using animal models is required to define the physiological role of the regulatory pathway and its crosstalk with other signaling pathways.

Conclusion

Using clinical samples, this study initially identified serum PRKG1-AS1 as a potential biomarker for early diagnosis, disease evaluation, and prediction of LDD progression. Additionally, through cell models and molecular experiments, this study preliminarily demonstrated that the PRKG1-AS1/miR-218-5p/CUL3 axis regulates inflammation and ferroptosis in nucleus pulposus cells through targeted mechanisms, thereby slowing the progression of LDD. This provides a new theoretical basis for early detection and targeted treatment of LDD.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JQ X, ZY M, L W conceived and designed the experiments. ZC L, TH W, JH L, YF Z performed the experiments. JQ X, ZY M, L W, XH Z, B X did data analysis. JQ X, ZY M, L W drafted the manuscript. All authors participated in the revision of the manuscript and approved the submitted version.

Funding

The authors did not receive support from any organization for the submitted work.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All participants provided written informed consent, and their clinical records were fully documented. The study protocol received formal ethical clearance from the Institutional Review Board of The Fourteenth Medical Center, Chinese PLA General Hospital and performed in line with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.