Human LF samples

This study has been proved by the Institutional Research Ethics Committee of the Peking University Third Hospital. Informed consent was obtained from all patients who participated in the study. A total of LF sample tissues collected from 24 patients with LFH or lumbar disc herniation at L4/5 level were used in this study, and the details of included patients were listed in Supplementary Table 1. The diagnosis of LFH caused by LF hypertrophy was conducted on the basis of the combination of imaging examination results and clinical symptoms. For the imaging examinations, the magnetic resonance imaging (MRI) was generally used, and the thickness of LF > 4 mm was usually regarded as the diagnostic criteria [1, 20]. For the measurement of LF width on MRI, the width of LF were measured three times by two senior surgeons blind to this research, and the average thickness was used for analysis [20].

Cell cultures

LF samples were acquired from patients who received the surgery for lumbar disc herniation. The LF samples were carefully minced, and then digested using 0.2% type I collagenase (#17100017, Gibco, USA) at 37 °C for 1.5 h. After the digestion, the LF tissues were washed with PBS (#10010023, Gibco, USA) and transferred to cell culture bottle (#707001, NEST, China) containing DMEM with 10% fetal bovine serum) (#A5256701, Gibco, USA), 100 mg/ml streptomycin, and 100 U/ml penicillin (#15140148, Gibco, USA). The samples were then incubated at 37 °C in a 5% CO2 cell incubator. Cells at three to five passages were obtained for subsequent cell experiments. To induce the fibrosis of LF cells, the recombinant protein TGF-β1 (#HY-P70543, MedChemExpress, USA) (10 ng/ml) was used to treat the LF cells. To inhibit the β-catenin signaling pathway, the LF cells were treated with 5 μM MSAB (#HY-120697, MedChemExpress, USA) to promote the degradation of β-catenin protein.

CRISPR/Cas9 system and miR-335-3p mimics transfection

Two single guide RNAs targeting SERPINE2 locus were cloned into the U6-sgRNA-EF1a-Cas9-FLAG-CMV-EGFP-P2A-puro plasmid by GeneChem (Shanghai, China). The details of single guide RNA sequences were as follows: sgRNA-1: 5′-TACGCCGTATCTCATCACCA-3′ and sgRNA-2: 5′-TCAATCAGATTGTGAAGTCG-3′. In brief, the LF cells were transfected by the plasmids or miR-335-3p mimics with the assistance of Lipofectamine 3000 (L3000075) following the manufacturer’s instructions. The sequences of miR-335-3p mimics were listed in Supplementary Table 2.

Lentivirus infection

To overexpress the SERPINE2 in LF cells, we used the lentivirus pSLenti-EF1-EGFP-P2A-Puro-CMV-SERPINE2-3xFLAG-WPRE to infect LF cells (OBiO Technology, China). In details, the LF cells were plated at 1.2 × 105 cells/well in six-well plates containing serum-free culture medium with polybrene (2 µg/ml) at 37 °C in a 5% CO2 atmosphere. Then, LF cells were transduced with control or SERPINE2 lentivirus at a multiplicity of infection of 40 when reaching the 60–70% confluence. Six hours later, the cells were transferred to serum-containing medium and incubated for additional 2 days. The transduction efficiency was evaluated using the western blot to detect the protein level of SERPINE2.

RNA immunoprecipitation (RIP) analysis

The RIP experiment was conducted with the PureBinding®RNA Immunoprecipitation Kit (#P0101, Geneseed, China) with anti-AGO2 (#2897, Cell Signaling Technology, USA), following the manufacturer’s instructions. The IgG (#3900, Cell Signaling Technology, USA) served as a negative control. The AGO2 antibody was recovered with the protein A/G beads. Co-precipitated SERPINE2 and miR-335-3p levels were assessed by qRT–PCR analysis.

Western blot

Samples were lysed by a radioimmunoprecipitation assay buffer (#R0010, Solarbio, China) and their protein concentration were determined by BCA Protein Assay Kit (#PC0022, Solarbio, China). After centrifugation, the lysates were mixed with the loading buffer (#P1040, Solarbio, China) for western blot and then the mixture was heated for 15 min at 99 °C. Then, a total of 20 µg protein was added into the hole of 12% Bis–Tris gel (#LK306, Epizyme, China) and transferred to polyvinylidene fluoride (PVDF) membranes (#88518, Millipore, USA). The membranes were blocked for 1 h at room temperature with 5% non-fat dried milk in TBST, and then incubated for 2 h at room temperature with primary antibody, and for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated secondary antibody (#SE134, #SE131, Solarbio, China). The detailed information for used antibody as follows: anti-COL3A1 (#ab184993, Abcam, UK), α-SMA (#ab7817, Abcam, UK), SERPINE2 (#11303-1-AP, Proteintech, USA), β-actin (#AF5003, Beyotime, China), and HRP-conjugated secondary antibody (#A0208, #A0216, Beyotime, China).

RNA extraction and qRT-PCR assay

Total cellular RNA extraction was performed with the TRIzol method. The first-strand cDNA was obtained from total RNA using the commercial kits (#AG11705, # AG11716, Accurate Biology, China) according to the instructions. The quantitative real-time polymerase chain reaction (qRT-PCR) was conducted with the SYBR Green Supermix (#AG11702, Accurate Biology, China) according to the instructions. GAPDH and U6 were applied as the internal control for the mRNA and miRNA, respectively. The details of primers used in this study were listed in Supplementary Table 2.

Dual luciferase assay

The human embryonic kidney (HEK) 293T cells (#CL-0005, Procell, China) were applied for the luciferase activity analysis. Generally, the HEK 293T cells were plated on 96‐well plates and cultured to 50–60% confluence. The 3′-UTR of SERPINE2 containing putative binding sites for miR-335-3p were cloned into the vector. The wild-type pMIR-REPORT-SERPINE2-3′-UTR and mutant luciferase reporter pMIR-REPORT-SERPINE2-3′-UTR were synthesized by GenePharma (Suzhou, China). A 150 ng vector of SERPINE2 3′-UTR-WT and SERPINE2 3′-UTR -MUT and 60 nM of miR-335-3p and negative control (NC) were transfected. The luciferase activity was examined with Dual Luciferase Reporter Assay kit (#RG029S, Beyotime, China) following the manufacturer’s instructions 48 h after the transfection.

Immunohistochemistry (IHC) assay

Human LF specimens were treated with 4% paraformaldehyde (#P1110, Solarbio, China) for 48 h, and then were embedded in paraffin. Then, we used the microtome (Leica, Germany) to cut the paraffin samples into sections that were 4-μm thick. Next, the sections were then deparaffinized and hydrated. Then, the sections underwent staining using the hematoxylin and eosin (H&E) staining kit (#G1120, Solarbio, China). The Elastic Van Gieson (EVG) staining was conducted by EVG kit (#G1597, Solarbio, China) following the guidelines. An Olympus BX63 microscope (Olympus, Japan) was applied to observe the stained cells.

Immunofluorescence (IF) staining

The LF cells were fixed for 20 min using 4% paraformaldehyde at room temperature. The fixed cells were treated with Triton X-100 (#T8200, Solarbio, China) and then subjected to immunoblocking. The fixed cells were left to incubate overnight at 4 °C with the primary antibodies listed below: anti-β-catenin (#ab32572, Abcam, UK) and anti-SERPINE2 (#11303-1-AP, Proteintech, USA). Then, fluorescein isothiocyanate (FITC)-conjugated secondary antibodies (#ab6717, Abcam, UK) and Cy3-conjugated secondary antibody (#ab6939, Abcam, UK) were used for IF staining. The DAB kit (#ab64238, Abcam, UK) was used for IHC staining, and DAPI (#C0065, Solarbio, China) was used for staining the nuclei in IF staining.

Statistical analysis

All statistical analyses were conducted using the GraphPad Prism 9.0 (La Jolla, USA) and all results were showed as the mean ± standard deviation (SD). The data normality was assessed with the Shapiro–Wilk test. For the comparisons of continuous variables between two groups, the unpaired two-tailed t-tests were used for normally distributed data, and Wilcoxon rank-sum test was used to compare group approaches for nonparametric data. For the comparisons across more than two groups, one-way analysis of variance (ANOVA) tests followed by Tukey’s honestly significant difference (HSD) multiple comparison tests were applied were used. The P < 0.05 was determined to be a significant difference (ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).