Human chondrocyte CHON-001 cells were purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, China) and incubated in the Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Carlsbad, CA, United States), containing 10% FBS, 100 units/ml penicillin, and 100 mg/ml streptomycin under 5% CO₂ at 37°C. CHON-001 cells were stimulated with LPS (0, 1, 2.5, 5, and 10 µg/ml) for 12 h to construct an OA in vitro model (Hu and Li, 2019; Sui et al., 2019).

Valproic acid (VPA; Sigma-Aldrich) stock solution was dissolved with DMSO and diluted in the culture medium to the final test concentrations (0, 0.5, 1, 2, and 5 mM) for CHON-001 cell incubation.

Cells with over 80% confluence were seeded into a six-well plate and transfected with negative control (NC, 5′-CTC​CCT​TCT​CTT​CTC​CCG​TCT​T-3′), miR-302d-3p mimic (5′-TAAGTGCTTCCATGTTTGAGTGT-3′)/inhibitor (5′-ACA​CTC​AAA​CAT​GGA​AGC​ACT​TA-3′), scramble siRNA (si-con, 5′-GAT​CGT​TCC​AGT​ACG​AAG​TCA​TGG-3′), si-ITGB4 (5′-GAG​GGT​GTC​ATC​ACC​ATT​GAA​CTC-3′), and pcDNA3.1-ITGB4 vector using Lipofectamine 2000. All the abovementioned bioagents were provided by GenePharma Co. (Shanghai, China). Chondrocytes were then cultured 48 h for future investigation in the presence or absence of VPA.

Total RNA of transfected chondrocytes was extracted with the TRIZOL reagent (Invitrogen, Carlsbad, CA, United States). Oligo (dT) primers were utilized to conduct the reverse transcription. The real-time quantitative PCR assay was performed with TaqMan one-step PCR Master Mix (Applied Biosystems, Foster City, CA, United States) and SYBR Premix Ex Taq (Invitrogen) on ABI7900HT real-time PCR system. For miRNA, the cDNA was synthesized by the TaqMan MicroRNA reverse transcription kit (Applied Biosystems). All reaction conditions were set up as follows: 95°C for 10 min, then 39 cycles consisting of predenaturation at 95°C for 10 s, denaturation at 60°C for 45 s, and extension at 72°C for 30 s, followed by final extension at 72°C for 5 min. Relative expression levels were calculated using the 2^−ΔΔCt method, with specific controls, GAPDH for ITGB4 and U6 for miR-302d-3p.

Primers were listed as follows:

  miR-302d-3p

5 µg/ml LPS was used to treat CHON-001 cells for 5 h after 48 h transfection. Protein extraction was then implemented in RIPA buffer (Beyotime, Beijing, China) supplemented with 1% protease inhibitor cocktail. The quantification of proteins was detected with the BCA method, and separated proteins were denatured at 95°C for 5 min. Electrophoresis was conducted in 12% SDS-PAGE with an equal amount of protein (20 μg) per tank, and the protein bands were transferred onto PVDF membranes. Afterward, these membranes were sealed in 5% skimmed milk powder for 60 min and incubated with primary antibodies at 4°C overnight. After rising in TBST, PVDF membranes were incubated with secondary antibody at ambient temperature for 60 min. All antibodies were as follows: Beclin 1 (cat. no. 3738; 1:1,000, CST), LC3-I/II (cat. no. 23214; 1:1,000, CST), p62 (cat. no. 5114; 1:1,000, CST), p-PI3K (cat. no. 4228; 1:1,000, CST), PI3K (cat. no. 4292; 1:1,000, CST), p-AKT (cat. no. 9271; 1:1,000, CST), AKT (cat. no. 9272; 1:1,000, CST), ITGB4 (cat. no. 4707; 1:1,000, CST), GAPDH (cat. no. 8884; 1:1,000, CST), and the HRP-conjugated anti-rabbit secondary antibody (cat. no. 7074; 1:2,000, CST). Protein signals were developed with ECL (EMD Millipore), and all immunoblots were scanned by Quantity One software (Bio-Rad, Hercules, CA, United States). The results of two repeat western blot assays were showed in Supplementary Figure 1.

Cell viability was assessed by using a CCK-8 kit (Takara, Japan) according to the manufacturer’s protocols. Transfected chondrocytes exposed to VPA or not were firstly seeded into a 96-well plate. After 0, 24, 48, and 72 h incubation, 10 µL of CCK-8 cocktail was added to each well to continuously maintain cells for other 1.5 h at 37°C with 5% CO₂. The absorbance values were measured at 450 nm under a microplate reader.

The proportion of apoptosis was examined with the Annexin V-FITC/PI double staining kit. Chondrocytes (2 × 10⁵) were collected in a 5 ml tube and centrifuged at 2,000 xg at 4°C. The supernatant was discarded, and cells were resuspended using 1× binding buffer. Next, 100 µL of supernatant, 5 µL of Annexin V-FITC, and 5 µL of PI/RNase were mixed in tubes and incubated in the dark at room temperature for 5 min. Finally, 100 µL of PBS was added and flow cytometry machine detection was performed within 1 h. The FITC channel screens GFP-positive cells for Annexin V-APC/7-AAD fluorescence detection by GFP green fluorescence. The red fluorescence of Annexin V-APC was viewed at an excitation wavelength of 633 nm and a maximum emission wavelength of 660 nm. The excitation wavelength was 546 nm, and the emission wavelength was 647 nm for the 7-AAD red fluorescence. Single-color stains for FITC Annexin and PI and unstained cells were included in all experiments as positive and negative controls. The quadrants were as follows: Q1, necrotic cells; Q2, late apoptotic cells; Q3, viable cells; Q4, apoptotic cells.

Based on the bioinformatics analyses on the TargetScan, StarBase, and miRDB, the binding sites were predicted between miR-302d-3p and ITGB4. To validate the correlation, the 3′-UTR of ITGB4 sequences was amplified and inserted into the pLUC vector to construct the ITGB4 wild type (WT). Similarly, ITGB4 mutant (MUT) was generated by the same steps with mutant sequences of binding sites. In order to examine the luciferase activity, CHON-001 cells were transfected with ITGB4 WT or ITGB4 MUT and miR-NC, miR-302d-3p mimic, or inhibitor using Lipofectamine 2000. After 48 h transfection, lysis buffer was used to lyse cells, and the Renilla and Firefly luciferase activities were measured using a Dual-Luciferase Reporter Assay kit (Promega Corporation, Madison, WI, United States). The ratio of Firefly luciferase activity to Renilla luciferase activity was utilized to reflect the association between miR-302d-3p and ITGB4.

The interaction between miR-302d-3p and ITGB4 was further ascertained using Magna RIPTM RNA Binding Protein Immunoprecipitation Kit (Millipore, Bedford, MA, United States). CHON-001 cells transfected with miR-NC or miR-302d-3p mimic were lysed in precooled RIP buffer. The cell lysate was then hatched with Protein A/G magnetic beads and continued to be incubated with anti-Ago2 (Abcam) or anti-IgG (Abcam). After washing, the protein and DNA in the immunoprecipitated complex were removed and qRT-PCR was conducted to evaluate the enrichment level of ITGB4.

OA clinical specimens, accession numbers GSE55457 and GSE12021, were collected from the GEO database and used for the identification of DEGs. Autophagy-related genes were downloaded from the HADb repository (http://www.autophagy.lu/). KEGG analysis was performed on the DAVID website. Drugs related to OA were determined with the CTD website. Statistical analysis was applied with SPSS19.0 and GraphPad Prism 8.0, and comparisons were determined using Student’s t-test or ANOVA followed by Tukey’s post hoc test. All variables were presented as mean ± standard deviation (SD), which were obtained from three independent experiments. p < 0.05 indicates significant difference.

To clearly explore the pathogenesis of OA, OA clinical samples retrieved from the GEO database (accession numbers: GSE55457 and GSE12021) were utilized to identify the DEGs in OA according to the following criteria: log (fold change) ≥1 and p < 0.05. A total of 690 DEGs were identified based on GSE55457, including 211 upregulated mRNAs and 479 downregulated mRNAs. Meanwhile, a total of 729 DEGs (212 upregulated genes and 517 downregulated genes) were achieved on the basis of GSE12021. Afterward, we accessed the HADb repository and downloaded 222 autophagy-related genes. In the overlapping part of 222 autophagy-related genes and DEGs achieved from GSE55457 and GSE12021, nine common genes were obtained (Figure 1A), including MAP2K7, MYC, VEGFA, BCL2L1, CDKN1A, NAMPT, ATG16L1, HDAC6, and ITGB4. These nine common genes were analyzed in the DAVID database for KEGG enrichment analysis and ten meaningful pathways were obtained under the condition of p < 0.05 (Figure 1B). The pathway with the most enriched genes was the PI3K-AKT signaling pathway. For genes enriched in the PI3K-AKT signaling pathway, a review of related articles demonstrated that only ITGB4 has yet to be studied in OA. Thus, we finally selected ITGB4 as a research objective. As shown in Figures 1C,D, the expression of ITGB4 in OA tissues was remarkably lower than normal controls.

Next, we employed TargetScan, StarBase, and miRDB prediction tools to predict the upstream miRNAs of ITGB4. We took the intersection of potential miRNAs obtained from these three prediction sites and ultimately obtained 15 candidate miRNAs (Figure 1E). Given the previous literature, except for miR-302d-3p and miR-373-3p, other miRNAs of ITGB4 were not studied in OA (Wang et al., 2019; Zhu and Jiang, 2019). Considering that ITGB4 was decreased in OA, we finally selected miR-302d-3p with high expression in OA (Li et al., 2018).

The CTD website was subsequently used to query the drugs related to OA, autophagy, and ITGB4 expression, owing to the lack of effective agents in OA. A total of 21 common drugs were identified (Figure 1F). Through comprehensive literature analysis, VPA was selected for further research. Accordingly, we realized that miR-302d-3p and ITGB4 may be crucial therapeutic targets for OA treatment and hypothesized that VPA might regulate the progression of OA through mediating miR-302d-3p/ITGB4 signal axis.

To validate the hypothesis, we firstly mimic human OA using LPS-treated CHON-001 cells. CCK-8 assay revealed that LPS inhibited cell viability in a dose-dependent manner (Figure 2A). 5 µg/ml LPS was selected for further experiments. Next, the effect of VPA on cell viability was examined by CCK-8 assay. Results in Figure 2B manifested that the treatment of VPA (0, 0.5, 1, 2, and 5 mM) significantly relieved the LPS-stimulated injury in CHON-001 cells, and the protective effect reached its peak at 2 mM; thus, we selected 2 mM for further experiments. LPS treatment induced an increased apoptosis rate in CHON-001 cells, which was attenuated after VPA treatment (Figures 2C,D). The levels of autophagy-related markers (Beclin 1, LC3-I, LC3-II, and p62) were measured using western blot. Results revealed that the level of Beclin 1 and the conversion of LC3-I to LC3-II were inhibited in LPS-treated CHON-001 cells, and p62 was significantly increased (Figures 2E,F). Therefore, we speculated that autophagy was suppressed in LPS-treated chondrocytes. The VPA stimulation overturned the suppressive effect of LPS on the autophagy of chondrocytes (Figures 2E,F). Overall, these findings suggested that VPA might protect chondrocytes against LPS-induced injury.

The biological function of miR-302d-3p/ITGB4 axis on LPS-treated chondrocytes was conducted in the following experiments. Before that, we first verified the relationship of ITGB4 and miR-302d-3p. The complementary region sequences between 3′-UTR ITGB4 and miR-302d-3p are presented in Figure 3A. Accordingly, the luciferase reporter vectors ITGB4 WT and ITGB4 MUT were constructed and utilized to evaluate the luciferase activity using a dual-luciferase report assay. Relative luciferase activity of the ITGB4 WT group was significantly decreased after the transfection of miR-302d-3p mimic, whereas miR-302d-3p inhibitor resulted in an increased level of luciferase activity. However, there were no changes in the ITGB4 MUT group (Figure 3A). To further ascertain the interaction between miR-302d-3p and ITGB4, RIP assay was also implemented. Results showed that overexpression of miR-302d-3p induced the copious enrichment of ITGB4 in the Ago2 immunoprecipitation complex, indicating the target interaction between miR-302d-3p and ITGB4 (Figure 3B). The interplay of miR-302d-3p and ITGB4 was analyzed through qRT-PCR and western blot. The expression of ITGB4 was inhibited by miR-302d-3p overexpression but was promoted after miR-302d-3p knockdown (Figures 3C–F). CHON-001 cells transfected with pcDNA3.1-ITGB4 showed an increased ITGB4 expression but a decreased level of ITGB4 after si-ITGB4 transfection (Figures 3C–F). Besides, as expected, the promoting effect of pcDNA3.1-ITGB4 or inhibitory impact induced by si-ITGB4 on ITGB4 expression was attenuated due to the transfection of miR-302d-3p mimic or inhibitor, respectively (Figures 3C–F). Collectively, all results demonstrated that ITGB4 may be directly targeted by miR-302d-3p.

Next, CCK-8 assay revealed that miR-302d-3p mimic or si-ITGB4 dramatically inhibited the cell viability of LPS-treated CHON-001 cells (Figures 4A,B). Conversely, cell viability of CHON-001 cells stimulated by LPS was elevated after the transfection of miR-302d-3p inhibitor or pcDNA3.1-ITGB4 (Figures 4A,B). Moreover, the cotransfection of miR-302d-3p mimic and pcDNA3.1-ITGB4 or miR-302d-3p inhibitor and si-ITGB4 reversed the corresponding impacts of individual transfection of miR-302d-3p mimic, pcDNA3.1-ITGB4, miR-302d-3p inhibitor, and si-ITGB4 (Figures 4A,B). The opposite trends were exhibited in flow cytometry assays. In LPS-stimulated chondrocytes, miR-302d-3p overexpression significantly promoted apoptosis rate, while upregulation of ITGB4 reduced the apoptosis rate. Cotransfection of miR-302d-3p mimic and pcDNA3.1-ITGB4 reversed the miR-302d-3p mimic-induced promoting effect or ITGB4 overexpression-stimulated suppressive role to the apoptotic ability of CHON-001 cells after LPS treatment (Figure 4C). Furthermore, apoptosis rate was attenuated by miR-302d-3p inhibitor but increased due to ITGB4 knockdown, which were all overturned by the cotransfection of miR-302d-3p inhibitor and si-ITGB4 (Figure 4D). Therefore, we concluded that the miR-302d-3p/ITGB4 axis may affect LPS-treated chondrocytes behaviors, including cell viability and apoptosis.

In order to explore whether VPA is involved in the regulatory action of miR-302d-3p/ITGB4 axis in OA, we detected the expression of miR-302d-3p and ITGB4 in CHON-001 cells stimulated by LPS or/and VPA. The expression of miR-302d-3p was significantly increased by LPS. VPA treatment restored the expression of miR-302d-3p to the normal level (Figure 5A). Conversely, the mRNA and protein levels of ITGB4 were significantly decreased in LPS-treated CHON-001 cells, which was reversed by the stimulation of VPA (Figures 5B–D). Together, it can be assumed that VPA may exert a protective role via miR-302d-3p/ITGB4 in OA.

The following experiments were designed and implemented to further clarify the interaction of VPA and miR-302d-3p/ITGB4 axis in OA cells. As shown in Figure 6A, the stimulatory effect of VPA was partially abolished by the miR-302d-3p overexpression but enforced by transfection of pcDNA3.1-ITGB4 in LPS-induced chondrocytes. Apoptosis assay manifested that overexpression of miR-302d-3p reversed the suppressive role of VPA on apoptotic capability, while overexpression of ITGB4 significantly strengthened the inhibitory impact induced by VPA treatment in an in vitro OA model (Figures 6B,C). The potential relevance of VPA and miR-302d-3p/ITGB4 in autophagy of OA cells was performed by western blotting. Compared with LPS + VPA group, upregulation of miR-302d-3p decreased the level of Beclin 1, retarded the conversion of LC3-I to LC3-II, and elevated p62 levels. However, in CHON-001 cells treated by LPS and VPA, ITGB4 overexpression promoted the level of Beclin 1 and accelerated the conversion from LC3-I to LC3-II but repressed p62 (Figure 6D). VPA could stimulate cell viability and autophagy but attenuate apoptosis rate in OA cells through regulating miR-302d-3p/ITGB4.

Considering that ITGB4 was selected as a gene enriched in the PI3K-AKT pathway, thus, western blot assays were conducted to decipher the effect of VPA/miR-302d-3p/ITGB4 on the activity of the PI3K-AKT pathway in OA cells. Figures 7A,B showed that compared with the LPS group, the levels of p-PI3K and p-AKT were markedly decreased in the LPS + VPA group. In OA cells induced by LPS, miR-302d-3p mimic rescued the inhibitory effect of VPA treatment on p-PI3K and p-AKT levels, whereas overexpression of ITGB4 intensified the suppressive role of VPA stimulation on the PI3K-AKT pathway. Thus, we speculated that VPA may inactivate the PI3K-AKT pathway possibly via regulating miR-302d-3p/ITGB4 in LPS-stimulated CHON-001 cells.