The HAPLN1 expression in the synovium of RA (n=20) was found by immunohistochemical assays to be significantly higher than that of OA patients (n=17) ( Figure 1A and Supplementary Table S1 for participants’ details). Similarly, the HAPLN1 level in the plasma of RA patients (n=61) was higher than that of OA patients (n=20) and healthy control subjects (HC, n=12) ( Figure 1B and Supplementary Table S2 for participants’ details). These results agree with our earlier mRNA sequencing analysis, which showed that the level of HAPLN1 expression in RA-FLSs is higher than that in OA FLSs ( Figure 1C ) (9).

The AMP-activated protein kinase (AMPK) pathway contributes to cell viability, metabolism, and inflammation during the onset and progression of RA (21, 22). Treatment of RA-FLSs with an AMPK activator, metformin, up-regulates HAPLN1 expression (9). Therefore, changes in AMPK expression could help elucidate the functions of HAPLN1. In this study, the AMPK level in healthy people was found to be significantly higher than OA and RA patients and had a significant positive correlation with the HAPLN1 level in the plasma of RA patients (n = 48; r = 0.693, p < 0.0001) ( Figure 1C ). In addition, the plasma HAPLN1 level negatively correlated with the course of the arthritis disease (n = 46, r = -0.311, p = 0.038). HAPLN1 in RA patients having less than 3 years of disease activity (n = 20) was higher than that in patients having disease symptoms for more than 3 years (n = 41) ( Figure 1D and Supplementary Figure S1 ). A moderate positive correlation between HAPLN1 and the rheumatoid factor was noted (n = 24, r = 0.431, p < 0.05) ( Figure 1E ). However, no correlation between the HAPLN1 level and other indexes of disease activity (such as ESR and CRP) was revealed by the data in this study ( Figures 1F, G ). The elevated AMPK level was consistent with our previous observations (9). It is plausible that HAPLN1 participates in AMPK-regulated metabolic pathways.

To dissect the role of HAPLN1 in determining the viability of RA-FLSs, three small interfering RNA (si-RNA HAPLN1) molecules and an over-expression plasmid vector (HAPLN1^OE) were used to study the proliferation and migration ability of RA-FLSs. All of the tested siRNA molecules in this study effectively down-regulated the mRNA level of HAPLN1, while transfecting RA-FLSs with HAPLN1^OE up-regulated the expression of HAPLN1 ( Supplementary Figure S2 ). Recombinant HAPLN1 (rHAPLN1) at different concentrations was also added to RA-FLSs to study its functions.

Transfection with si-HAPLN1 did not affect the proliferation of RA-FLSs ( Figure 2A and Supplementary Figure S3 ), though a significant increase in the apoptotic ratio of RA-FLSs was revealed by the results of our TUNEL assay ( Figure 2B ). The wound healing assay and the transwell assay showed an increase in the migration ability of RA-FLSs after transfection with si-HAPLN1 ( Figures 2C, D ), and the effect was reversed after the addition of rHAPLN1 (50 ng/ml) ( Figure 2D ). Conversely, after transfection with HAPLN1^OE, RA-FLSs showed enhanced proliferation and a decline in the apoptotic ratio ( Figures 3A, B and Supplementary Figure S4 ). Results of the wound healing assay showed a decrease in the level of migration exhibited by HAPLN1^OE-transfected RA-FLSs ( Figure 3C ), whereas rHAPLN1 confirmed the effects of HAPLN1 on RA-FLSs which showed a significant increase in the proliferation activity ( Figure 4A and Supplementary Figure S5 ) and a reduction in apoptosis, especially during the early phase ( Figures 4B, C ). The wound healing assay and the transwell assay demonstrated that rHAPLN1 inhibited the migration ability of RA-FLSs ( Figures 4D, E ). It is known that cell mobility is elevated in cancer cells as well as in activated RA-FLSs (6). However, our experiments with si-HAPLN1, HAPLN1^OE, and rHAPLN1 demonstrated that HAPLN1 shows an inhibitory effect on the mobility of RA-FLSs although it could activate RA-FLS proliferation.

As the current molecular interaction network of HAPLN1 is barely barren, to get acquaintance with HAPLN1 functions in RA-FLSs from a more comprehensive view, proteomics and transcriptome analyses were done with rHAPLN1 treated or si-HAPLN1 transfected RA-FLSs to further investigate HAPLN1 functions in RA-FLSs.

We used blood samples from 61 RA and 20 OA patients and 12 age- and gender-matched healthy controls (HC) for measuring HAPLN1 levels by ELISA. The mean disease course of RA patients was 6.39 years (ranging from 0.2 to 30.0 years). Synovium samples were collected by arthroscopic surgery done with 20 RA and 17 OA patients and used for immunohistochemical (IHC) staining of HAPLN1. The inclusion and exclusion criteria and general information such as age, gender, disease activity, and disease course reported earlier (9) are summarized in Supplementary Tables S1 , 2 .

Blood samples of HC, OA, and RA patients were centrifuged after standing at room temperature for 2 h, at 1500 g for 10 min to collect the plasma. The HAPLN1 levels were detected by ELISA (RayBiotech, US). Plasma AMPK levels in RA patients were also evaluated by ELISA according to the manufacturer’s protocol (Jianglaibio, China). The SuPerMax 3000FA absorbance microplate reader (Flash Co. Ltd., China) was used to read the optical density (OD) values at 450 nm and concentrations of specific proteins were calculated based on the standard curve.

Synovium samples collected from 20 RA and 17 OA patients for IHC staining of HAPLN1 were prepared as reported earlier (9). Rabbit monoclonal anti-HAPLN1 antibody (Abcam, US) was added as the primary antibody (1:50) and incubated for 2 h at 37°C. Biotin-conjugated goat anti-rabbit antibodies (ZSGB-Bio, China), streptavidin-peroxidase conjugate, and diaminobenzidine were used as the detecting system. IHC-stained sections were semi-quantified under a microscope. The staining intensity was counted as none (0 points), weak positive (1+), moderate positive (2+), and strong positive (3+). The percentage of positive cells was obtained to calculate the H-Score. The range of H score for each slice was between 0 and 300. The formula of the H-Score is as follows (65):

Primary RA-FLSs were acquired from 3 untreated RA patients. Isolation and culture of RA-FLSs were reported as before (9). Briefly, FLSs were isolated by enzyme digestion and subsequently cultured in Dulbecco’s modified essential medium (DMEM) containing 10% fetal bovine serum (FBS, Invitrogen) and antibiotics (penicillin and streptomycin) at 37°C with 5% CO₂. Cells cultured between passages 4 and 9 were used in this study.

RA-FLSs at 60-70% confluency were transfected with siRNAs (Ribobio Company, China) at 50 nM with Lipofectamine™ 3000 reagent (Invitrogen, US). The following siRNA sequences were used: control siRNA (confidential sequence information) and 3 siRNAs of HAPLN1, si-1 (5’-AGGGTAGAGTGTTTCTGAA-3’), si-2 (5’-CCTGGAAAATTCTCGGATA), and si-3 (5’-ACCTCACTCTGGAAGATTA-3’). The three siRNAs effectively silenced HAPLN1 expression ( Supplementary Figure S2A ), and si-1 was selected randomly and used in subsequent studies. The negative control (NC) group denoted control siRNA transfection of RA-FLSs in the experiment.

For HAPLN1^OE RA-FLSs experiments, HAPLN1 over-expression plasmid and its control were constructed and packaged by Ubigene Biosciences (Guangzhou, China). The stbl3 strain plasmid cytomegalovirus vector-infected cells were cultured in LB medium (QDRS Biotec, China) with 100 g/ml of ampicillin under 37°C, 225 rpm for 24 h. The HAPLN1^OE plasmid vector and its negative control (NC) were then isolated with the Genopure Plasmid Maxi Kit (Roche, US). RA-FLS at 60-70% confluency was transfected with HAPLN1^OE vector or its negative control with Lipofectamine™ 3000 reagent (Invitrogen). The effects of HAPLN1^OE plasmid vector are shown in Supplementary Figure S2B .

MTT assay was used to ascertain FLSs viability transfected with si-HAPLN1, HAPLN1^OE, or their corresponding NC, or treated with rHAPLN1 (recombinant human HAPLN1 protein, Abcam, US) at different concentrations (0, 25, and 50 ng/ml). FLSs samples (si-HAPLN1 vs. its negative control, HAPLN1^OE vs. its negative control, or treated with different concentrations of rHAPLN1) digested using 0.25% pancreatin were transferred to 96-well plates with 3-5 × 10³ cells/well. At different time points (24, 48, and 72 h), the viability of the cells was measured using the MTT assay kit (Abcam).

Cell viability after transfection with si-HAPLN1, HAPLN1^OE, or their respective controls, or treated with different concentrations of rHAPLN1, was determined using the Cell Counting Kit-8 (CCK-8, Molecular Technology, Japan) assay.

RA-FLSs transfected with si-HAPLN1, HAPLN1^OE, or their corresponding controls, or treated with rHAPLN1 (0 or 50 ng/ml), were digested and transferred to 6-well plates with 2-3 × 10⁵ cells/well, cultured for 48 h, and stained by the One Step TUNEL Apoptosis Assay Kit (Beyotime, China). The apoptosis rate was calculated under a fluorescence microscope (Leica, Germany) with the excitation wavelength at 550 nm (Cy3) and the emission wavelength at 570 nm (red fluorescence).

FLSs apoptosis treated with rHAPLN1 was measured using the Annexin V-FITC/PI Cell Apoptosis Detection Kit (Vazyme, China) by flow cytometry. After treatment of FLSs with rHAPLN1 (0 or 50 ng/ml) for 48 h in 6-well plates, the cells were collected (3×10⁵/well), washed twice with PBS, re-suspended in 500 µl 1 × binding buffer, mixed with Annexin-V-fluorescein isothiocyanate (FITC, 5 µl) and propidium iodide (PI, 5 µl), and analyzed using a flow cytometer (BD FACSCantoTM II, US). The scatter diagram was distributed as follows: Q3, healthy cells (FITC-/PI-); Q2, apoptotic cells at an advanced stage (FITC+/PI+); and Q4, apoptotic cells at an early stage (FITC+/PI-). The apoptosis rate was calculated as a ratio of apoptotic cells in P2 (Q4 + Q2).

Wound healing assay was conducted to evaluate the migration capacity of FLSs transfected either with si-HAPLN1, HAPLN1^OE, or their corresponding controls, or treated with different concentrations (0, 25, and 50 ng/ml) of rHAPLN1. FLS samples (si-HAPLN1 vs. its negative control, HAPLN1^OE vs. its negative control, or treated with different concentrations of rHAPLN1) were transferred to 6-well plates with 3 × 10⁵ cells/well and cultured with serum free-RPMI 1640 medium. At different time points, the migrating ability of the cells was measured by using the wound healing assay as previously reported (9).

Transwell assay was performed to evaluate the migration capacity of FLSs transfected with si-HAPLN1, HAPLN1^OE, or their respective controls, or treated with rHAPLN1. FLSs in each set of experiments were re-suspended after culturing for 24 h. Transwell assay was conducted as previously described (9).

Total RNA from FLSs transfected with si-HAPLN1, HAPLN1^OE, or their respective controls, or treated with rHAPLN1, was prepared using TRIzol^® Reagent (Thermo Scientific, US) and quantified using Qubit (Thermofisher, US). RNA was reverse transcribed into cDNA using PrimeScript™ RT Master Mix (Takara, Japan). The reaction mixture contained 5 ml of 2 × TB Green Premix Ex Taq II (Takara, Japan), 3 ml of nuclease-free water, 1 ml of cDNA, 0.4 ml of each gene-specific primer, and 0.2 l of ROX reference dye. The qRT-PCR analysis was performed using the Applied Biosystems ViiA™ 7 Real-Time PCR System (Thermofisher, US). Each value represented an average from three independent biological replicates. GAPDH gene expression was used for data standardization. The fold change was calculated using the 2-ΔΔCt method. Primers of GAPDH, AMPK-ɑ, TNF-ɑ, IL-6, TGF-β, ACAN, fibronectin, collagen II, MMP1, MMP3, MMP9, Cyclin-D1, and Ki-67 are given in Supplementary Table S3 .

Total proteins from FLSs transfected with si-HAPLN1, HAPLN1^OE, or their respective controls for 48 h were extracted with Cell Lysis Buffer (Cell Signaling, US). Their concentration was measured using the BCA Protein Assay Kit (Merck, US). Relative changes in HAPLN1, pAMPK-α, IL-6, TNF-α, MMP1, MMP3, and MMP9 protein levels were determined. Expression of β-tubulin was selected as an internal reference. Capillary electrophoresis and Western blot analysis were carried out using reagents provided in the kit in accordance to the instructions provided by the user manual (ProteinSimple WES, US) as previously reported (9). Rabbit anti-HAPLN1 antibody (Abcam, US), rabbit anti-TNF-α, AMPK-α, pAMPK-α, MMP-1, MMP-3, IL-6, Cyclin D1, and β-tubulin specific mAbs (Cell Signaling, US) were used (1:100). Goat anti-rabbit secondary antibodies were provided by the ProteinSimple WES kit and applied as instructed. Data were analyzed using an in-built Compass software SW 4.0. The truncated and full-length target protein intensities (area under the curve) were normalized to that of the tubulin peak. In most of the figures, electropherograms were represented as pseudo-blots, generated using Compass software.

Statistical analysis was performed using GraphPad Prism 8.0 software. All the data were given as mean ± SD. Differences between two groups were evaluated for statistical significance using Student’s t-test. One-way ANOVA with Tukey’s multiple comparisons test was used to evaluate the differences among three or more groups. Correlations were evaluated using Liner regression and correlation test. p < 0.05 was considered statistically significant.

Label-free proteomics study was applied to FLSs transfected with si-HAPLN1 or treated with rHAPLN1 (50 ng/ml) and their controls for 48 h (management of each group is given in Supplementary Table S4 ) by PTMBiolabs, Inc. (Hangzhou, China). Each concentration was tested with 3 biological replicates. Cell samples were processed as reported earlier (66). LC−MS/MS proteomics analysis was performed on an EASY-nLC 1000 ultra-performance liquid chromatography (UPLC) system, followed by MS/MS using Q Exactive Plus (ThermoFisher Scientific, US) coupled online to the UPLC system. The MS/MS data were retrieved by the Maxquant search engine (v1.6.6.0). A human database was searched (Swiss-Prot). The decoy database anti-library was used to reduce the false positive rate (FDR). The FDR was adjusted to < 1%, and the minimum score for modified peptides was set > 40. Proteins with a fold-change ≥1.50 or ≤0.67 between si-HAPLN1, rHAPLN1, and their controls were considered as expression significant. Based on the protein sequence alignment method, the protein domain functions were defined by InterProScan (http://www.ebi.ac.uk/interpro/). Functional annotation enrichment of DEPs was performed by KEGG analysis. The enrichment significance was identified as p < 0.05 in the Fisher’s exact test and q < 0.05 in the Benjamini-Hochberg procedure.

High-throughput RNA sequencing was performed using FLSs after treatment with rHAPLN1 (0 and 50 ng/ml) for 48 h. Each concentration was tested three times. RNA-seq analysis was performed by Seqhealth Technology Co., Ltd (Wuhan, China). Total RNA (2 g) was used for stranded RNA sequencing library preparation using KCTM Stranded mRNA Library Prep Kit for Illumina^® (Seqhealth Co., Ltd. China). PCR products corresponding to 200-500 bps were enriched, quantified, and sequenced with Novaseq 6000 sequencer (Illumina), PE150 model. Raw sequencing data were first filtered by Trimmomatic (v. 0.36). Low-quality reads were discarded. The reads contaminated with adaptor sequences were trimmed. Clean data were mapped to the human reference genome from UCSC (https://genome.ucsc.edu/) using STRA software (v. 2.5.3a) with default parameters. Reads mapped to the exon regions of each gene were counted by feature Counts (Subread-1.5.1; Bioconductor) and then RPKMs were calculated. DEGs between groups were identified using the edgeR package (v. 3.12.1) in R studio software (version 3.6). A p-value cut-off of 0.05 and fold-change cut-off of 2.0 were used to judge the statistical significance of gene expression differences. The volcano plot was drawn with the ggplot2 package in R studio. Heatmaps of pathway enrichment analysis of DEGs were generated using Metascape (http://metascape.org) and a P-value less than 0.05 was statistically significant. KEGG enrichment analysis for DEGs was performed using KOBAS software (v. 2.1.1) with a p-value cut-off of 0.05. To compare transcriptome characteristics of rHAPLN1 with PBS groups, GSEA software (version 4.0.0) was used. Annotated pathway files (c5.go.bp.v7.4.symbols.gmt) were downloaded in the MSigDB database (http://www.gsea-msigdb.org/gsea/msigdb/collections.jsp). Pathways with a P-value less than 0.05 and false discovery rate (FDR) less than 0.2 were significantly enriched.