Hyaluronan and Proteoglycan Link Protein 1 (HAPLN1) Promotes Viability of Rheumatoid Arthritis Fibroblast-Like Synoviocytes

To investigate HAPLN1 contribution to the viability of RA-FLSs and identify its potential role in RA pathogenesis. Plasma levels and synovial expression of HAPLN1 were compared between healthy controls, and osteoarthritis (OA) and RA patients. Proliferation and migration of RA-FLSs transfected with siHAPLN1, HAPLN1 OE (over-expression vector) and respective controls or treated with rHAPLN1 were measured by MTT and CCK8 assays as well as wound healing and transwell assays. RT-qPCR and automated WB analysis were used to compare the expression of AMPK-(cid:0), TNF-(cid:0), TGF-β, ACAN, MMPs, Cyclin-D1 and Ki-67 after siHAPLN1 or HAPLN1 OE transfection. Proteomics and mRNA-seq analysis was done to study the differentially expressed proteins/genes after siHAPLN1 or rHAPLN1 treatment. Concept and design: YC and DZL. Experiment performance: YC, BJW, YQW. Acquisition, analysis and interpretation of data: YC, YJC and KSN. Drafting the manuscript: YC, KSN and BJW.


Introduction
Rheumatoid arthritis (RA) is a chronic autoimmune disease in which joint and cartilage are primarily targeted and destroyed. Pannus formation is one of the main pathological manifestations of the disease 1 . During RA development, broblast-like synoviocytes (FLSs) produce several pathogenic mediators like cytokines and proteases, which contribute to the destruction of extracellular matrix (ECM) 2,3 . Therefore, FLSs are considered to be one of the major effector cells in RA.
Hyaluronan and proteoglycan link protein 1 (HAPLN1), also known as cartilage link protein, was originally identi ed from the proteoglycan component extracted from bovine articular cartilage 4 . Physiologically, HAPLN1 maintains the stable aggregation and binding activity of two important ECM macromolecules, hyaluronan acid (HA) and proteoglycan, which contribute to a stable macromolecular structure, and the compression resistance of joints 5 . Indispensable functions of HAPLN1 in the growth of bone, cartilage and heart were well documented 5 . An emerging role for HAPLN1 in oncology was reported recently 5,9,10 .
Originally, the most classic disease reported to be associated with HAPLN1 is juvenile rheumatoid arthritis 5,9,10 . Whereas, studies on HAPLN1 function in rheumatic diseases including RA are very few, and the role of the target genes is also unclear. Mainly through genomics research, HAPLN1 relation with spinal degeneration 12,13 , osteoarthritis 12,13 , ankylosing spondylitis 15 and RA 16 was identi ed.
Interestingly, HAPLN1 showed a signi cant correlation with disease activity in RA 16 and found to be one of the most signi cantly up-regulated genes in RA-FLSs compared to osteoarthritis (OA) FLSs 17 . Based on these observations, it can be inferred that HAPLN1 participates in the development of RA, though its contribution is still unclear.
In cancer-related research, HAPLN1 was shown to have the properties of a oncogene that contributes to an increased susceptibility to lung cancer 18 , aggressiveness of hepatocellular carcinomas 19 , and drug resistance to multiple myeloma 20 . Therefore, we hypothesized that HAPLN1 may increase the viability of RA-FLSs, mimicking the behavior of cancer cells, which may lead to an aggressive pannus formation and joint damage 21 . Material And Methods

Patients' and Samples
We enrolled 12 age and gender matched medical staff as healthy controls. Their blood samples as well as 61 RA and 20 OA patients' samples were used for measuring HAPLN1 levels by ELISA. Synovium samples were collected by arthroscopic surgery done with 20 RA and 17 OA patients and used for IHC staining of HAPLN1. The inclusion and exclusion criteria and general information such as age, gender, disease activity and disease course were reported earlier 21  Enzyme linked immunosorbent assay (ELISA) Blood samples of HC, OA and RA patients were centrifuged after standing at room temperature for 2 h at 1,500 g for 10 min to collect the plasma and HAPLN1 levels were detected by ELISA (RayBiotech, USA). 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) value at 450 nm and concentrations of speci c proteins were calculated based on the standard curve.
Isolation and culture of RA-FLSs were reported as before 17 . Brie y, FLSs were isolated by enzyme digestion and subsequently cultured in Dulbecco's modi ed essential medium (DMEM) containing 10% fetal bovine serum (FBS, Invitrogen) and antibiotics (penicillin and streptomycin) at 37°C with 5% CO 2 .
Cells cultured between passages 4 and 9 were used in this study.

HAPLN1 over-expression vector preparation and transfection
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 µg/ml of ampicillin under 37℃, 225 rpm for 24 h. The HAPLN1 OE plasmid vector and its control were then isolated with Genopure Plsmid Maxi Kit (Roche, USA). RA-FLS at 60 -70% con uency were transfected with HAPLN1 OE vector or its negative control with Lipofectamine™ 3000 reagent (Invitrogen). The effects of HAPLN1 OE plasmid vector were shown in the supplementary gure 2B.

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

TUNEL assay
FLSs transfected with si-HAPLN1, HAPLN1 OE or their corespondent controls, or treated with rHAPLN1 (0 or 50 ng/ml) were digested and transferred to 6-well plates with 2-3 × 10 5 cells/well, cultured for 48 h, and stained by One Step TUNEL Apoptosis Assay Kit (Beyotime, China). Apoptosis rate was calculated under uorescence microscope (Leica, Germany) with the excitation wavelength at 550 nm (Cy3), and the emission wavelength at 570 nm (red uorescence).

Wound healing assay
Wound healing assay was done to evaluate the migration viability of FLSs transfected either with si-HAPLN1, HAPLN1 OE or their corespondent controls, or treated with different concentrations (0, 25 and 50 ng/ml) of rHAPLN1. FLSs 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 × 10 5 cells/well and cultured with serum free -RPMI 1640 medium. At different time points, migration viability was measured by wound healing assay as previously reported 17 . Transwell assay Transwell assay was also 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 experiment were re-suspended after culturing for 24 h. Transwell assay procedure was described earlier 17 .

Quantitative real-time polymerase chain reaction (qPCR)
Total RNA from FLSs transfected with si-HAPLN1, HAPLN1 OE or their respective controls, or treated with rHAPLN1 was prepared using TRIzol® Reagent (Thermo Scienti c, USA) and quanti ed using Qubit (Thermo sher, USA). RNA was reverse transcribed into cDNA using PrimeScript™ RT Master Mix (Takara, Japan). The reaction mixture contained 5 µl of 2 x TB Green Premix Ex Taq II (Takara, Japan), 3 µl of nuclease-free water, 1 µl of cDNA, 0.4 µl of each gene-speci c primer and 0.2 µl of ROX reference dye. The qRT-PCR analysis was performed using Applied Biosystems ViiA™ 7 Real-Time PCR System (Thermo sher, USA). Each value represents an average from three independent biological replicates.

Automated western blot analysis
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, USA) and their concentration was measured using a BCA Protein Assay Kit (Merck, USA). Relative changes in HAPLN1, pAMPK-α, IL-6, TNF-α, MMP1, MMP3 and MMP9 protein levels were determined. Expression of β-tublin was selected as internal reference. Capillary electrophoresis and western blot analysis were carried out using reagents provided in the kit and following instructions in the user manual (ProteinSimple WES, USA) as previously reported 17 . Rabbit anti-HAPLN1 antibody (Abcam, USA), Rabbit anti-TNF-α, pAMPK-α, MMP-1, MMP-3, IL-6, and β-tublin speci c mAbs (Cell Signaling, USA) were used (1:100). Goat anti-rabbit secondary antibodies were provided in 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 tubulin peak (control). In most of the gures, electropherograms are represented as pseudo-blots, generated using Compass software.

Proteomics study
Label-free proteomics study was applied to FLSs transfected by si-HAPLN1, or treated by rHAPLN1 (50 ng/ml) and their controls for 48 hours (for management of each group is seen in table 4) by PTMBiolabs, Inc. (Hangzhou, China). Each concentration with 3 biological replicates. Cell samples processed as reported 23 . 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 Scienti c, USA) 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 antilibrary was used to the reduce false positive rate (FDR). The FDR was adjusted to < 1%, and the minimum score for modi ed 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 signi cant. Based on the protein sequence alignment method, the protein domain functions were de ned by InterProScan (http://www.ebi.ac.uk/interpro/). Functional annotation enrichment of DEGs were performed by Gene Ontology (GO) annotation analysis and KEGG analysis. The enrichment signi cant was identi ed as p < 0.05 in Fisher's exact test and q < 0.05 in Benjamini-Hochberg's procedure.
High-throughput mRNA sequencing High-throughput RNA sequencing was performed using FLSs after treatment with rHAPLN1 (0 and 50 ng/ml) for 48 h. Each concentration was tested thrice. RNA-seq and high throughput sequencing were conducted by Seqhealth Technology Co., Ltd (Wuhan, China). Total RNA (2 µg) 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, quanti ed and sequenced with Novaseq 6000 sequencer (Illumina), PE150 model. Raw sequencing data was rst ltered by Trimmomatic (v. 0.36), and 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 identi ed using the edgeR package (v. 3.12.1) in R studio software (version 3.6). A p-value cutoff of 0.05 and fold-change cutoff of 2.0 were used to judge the statistical signi cance 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) 24 and p value less than 0.05 was considered to be statistically signi cant. Gene ontology (GO) analysis and Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis for DEGs were done uisng KOBAS software (v. 2.1.1) with a p value cutoff of 0.05. To compare transcriptome characteristics of rHAPLN1 and PBS groups, GSEA software (version 4.0.0) was used. Annotated pathway les (c5.go.bp.v7.4.symbols.gmt) were downloaded in the MSigDB database (http://www.gsea-msigdb.org/gsea/msigdb/collections.jsp). Pathways with P value less than 0.05 and false discovery rate (FDR) less than 0.2 were considered to be signi cantly enriched.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 9.0 software. All the data were given as mean ± SD. Differences between two groups were evaluated for statistical signi cance using Student's t-test. One-way ANOVA with Tukey's multiple comparisons test was used to evaluate the differences between three or more groups. Pearson correlation coe cient was calculated using "cor" function in R studio. Correlations were evaluated using liner regression and correlation tests. p < 0.05 was considered as statistically signi cant.

Results
Increased HAPLN1 expression in synovium and plasma of RA patients Our mRNA sequencing analysis showed HAPLN1 expression in RA FLSs was higher than OA FLSs 17 ( Figure 1A). Here, we report HAPLN1 expression in the synovium was signi cantly higher in RA than OA patients ( Figure 1B, participants' information is given in the supplementary table 1). Similarly, HAPLN1 levels in the plasma of RA patients was also higher than OA patients and healthy controls ( Figure 1C, participants' information is given in the supplementary table 2). In addition, plasma HAPLN1 levels have positively correlated with the severity of disease in RA patients (r = -0.311, p = 0.038) (supplementary gure 1A). Interestingly, patients with less severe disease course had higher HAPLN1 levels (supplementary gure 1B). However, no correlation between HAPLN1 level with other indexes of disease activity like ESR and CRP was observed, but it positively correlated with plasma AMPK levels (r = 0.693, p < 0.0001) (supplementary gure 1C & D).

HAPLN1 increased the proliferation but inhibited mobility of RA-FLSs
All the three siRNAs prepared have effectively down-regulated HAPLN1 mRNA expression, while transferring HAPLN1 over-expression vector into RA-FLSs up-regulated its expression (Supplementary gure 2). Using MTT and CCK8 assays, we have shown that si-HAPLN1 transfection has failed to affect the proliferation of RA-FLSs (Figure 2A and supplementary gure 3). However, it has signi cantly increased the apoptotic ratio of RA-FLSs as observed with TUNEL assay ( Figure 2B). Both wound healing and transwell assays showed an increased migration ability of RA-FLSs after si-HAPLN1 transfection ( Figure 2C & D), which could be reversed by adding r-HAPLN1 (50 ng/ml) to the culture medium ( Figure   2D).
Conversely, after transfection with HAPLN1 over-expression vector, RA-FLSs showed an increased proliferation and a decreased apoptotic ratio ( Figure 3A & B and supplementary gure 4). Wound healing assay suggested a decreased level of migration with HAPLN1 OE transfected RA-FLSs ( Figure 3C). Whereas, rHAPLN1 con rmed the effects on RA-FLSs, which signi cantly enhanced the proliferation activity of RA-FLSs ( Figure 4A, and supplementary gure 5) and reduced RA-FLSs apoptosis, especially during early phase ( Figure 4B & C). The wound healing and transwell assays demonstrated that rHAPLN1 inhibited the migration ability of RA-FLSs ( Figure 4D & E).
In order to better understand potential interactions between these molecules, we collected multiple expression data of relative mRNA levels from NC groups and applied Pearson correlation coe cient to identify any potential correlations. Importantly, HAPLN1 level has positively correlated with AMPK-, which is in accordance with its plasma levels, and also with our previous ndings using RA-FLSs 17 . In addition, HAPLN1 levels have positively correlated with in ammatory molecules like MMP1, MMP9, TNF-, and IL-6. Furthermore, HAPLN1 levels showed positive correlations with ECM modulators like TGF-βand Collagen type II, while a negative correlation was observed with the in ammatory molecules MMP1, MMP-3, MMP9, TNF-, and IL-6 (supplementary gure 6).

Proteome and mRNA-seq analysis of HAPLN1 functions in RA-FLSs
In order to understand HAPLN1 functions in RA-FLSs, proteomic analysis of si-HAPLN1 and rHAPLN1 treated RA-FLSs, and transcriptome analysis of rHAPLN1 treated RA-FLSs were done. In the proteomic study, RA-FLSs from NC, si-HAPLN1, and rHAPLN1 groups included 443,973 matched spectra in total, and 4,184 quanti able proteins were identi ed (Table 1). Principal component analysis (PCA) plot indicated a high aggregation between duplicated samples, demonstrating the quantitative repeatability of experiments (supplementary gure 7). Among the identi ed proteins, 14 were up-regulated and 47 were down-regulated after si-HAPLN1 transfection. Besides, 101 proteins were up-regulated, and 82 were downregulated by rHAPLN1 treatment compared to NC ( Figure 6A). Quanti able proteins 4184 Note: 1. Total number of spectra includes number of total spectra and number of secondary spectra generated by mass detection; 2. Number of matched spectra includes the number of effective spectra and the number of spectra matching the theoretical secondary spectra; 3. Peptides include the number of identi ed peptides and the number of peptide sequences resolved by the matching result; 4. Unique peptides include the number of unique identi ed peptides and, the number of unique peptide sequences resolved by the matching result; 5. Identi ed proteins include the number of identi ed proteins; 6.
Quanti able proteins include quantitative protein numbers and the number of proteins quanti ed by speci c peptides. Metascape pathway analysis showed the up-regulated DEGs treated by rHAPLN1 were mainly enriched in GTPase Cycle, cell cycle and regulation of cell division ( Figure 7C), which were con rmed by proliferation and apoptosis studies. Moreover, GSEA analysis were performed to compare rHAPLN1 and PBS groups. The results showed that rHAPLN1 group has positively associated with the pathways of extracellular matrix structural constituents, alpha actin binding, metalloaminopeptidase activity, proteoglycan binding, focal adhesion, regulation of protein exit from endoplasmic reticulum, lipid translocation, regulation of androgen receptor signaling pathway, retrograde axonal transport, dendritic spine development, peptide cross linking and insulin like growth factor receptor signaling pathway (Supplementary gure 11).

Discussion
Here, we con rmed an increased HAPLN1 expression in the synovium and plasma samples from RA patients. Over-expression of HAPLN1 or treatment with rHAPLN1 increased the proliferation but decreased apoptosis of RA-FLSs. However, si-HAPLN1 transfection has failed to show any effect on proliferation, though it induced signi cant apoptosis of RA-FLSs. HAPLN1 was discovered 50 years ago, with a wide range of physiological effects as documented. HAPLN1 interacts with the globular domains of hyaluronic acid and proteoglycans, such as aggrecan, versican andα-trypsin inhibitor in various extracellular matrices to form a stable ternary complex 25 and contributes to the compression resistance and shock absorption of the joints. Thus, HAPLN1 plays an important role in cartilage formation and homeostasis. HAPLN1-de cient mice showed a series of cardiac malformations, including atrial septal and myocardial defects with a signi cant reduction in the level of multifunctional proteoglycans 8 . In addition, HAPLN1 also plays an important role in regulating the development of the central nervous system 26 . In the embryonic development of zebra sh, the expression of HAPLN1 was observed in the paraxial cells at the budding stage. During the somatogenesis stage, HAPLN1 was observed in the condyles, basal plate, inferior notochord and rhombohedral protonode branchial arch 27  Here, a signi cant positive correlation between plasma AMPK and HAPLN1 levels was identi ed. Considering AMPK mediated inhibition of a broad range of in ammatory reactions involving multiple biological pathways, it is plausible that an increased HAPLN1 might be the way, how metformin bene ts RA patients 17 , and also how body reacts to in ammation induced stress responses. However, HAPLN1 OE or rHAPLN1 treated RA-FLSs showed an increased proliferation, which is contradictory to our previous assumption about HAPLN1 as a protective factor. On the other hand, si-HAPLN1 decreased the migration capability while HAPLN1 OE or rHAPLN1 treated RA-FLSs showed an inhibitory effect. At this point, HAPLN1 seemed to be a protective molecule. Thus, it seems there is a dilemma in clarifying the role of HAPLN1 in RA-FLSs viability by functional studies. It's interesting to note that both results could possibly be consistent with the ndings in cancer studies, because an increased HAPLN1 seems associated with more aggressiveness, and leading to stemness of various cancers 18,19,30 , while achieving robust ECM to restrict metastasis of cancer cells 9,10 .
Earlier interactions between HAPLN1 and other molecules beyond hyaluronic acid and proteoglycans were investigated as well. For example, TNF-activated mitogen-activated kinase (MEK) in chondrocytes regulates the expression of HAPLN1, and controls the catabolism and anabolism of the extracellular matrix of chondrocytes 31 . In multiple myeloma cells, HAPLN1 can activate the NF-B pathway to acquire resistance to bortezomib 20 . In granulosa cells, HAPLN1 was proposed to be promoted through PKA-RUNX1/RUNX2 pathway 32 , which was earlier con rmed in metformin treated RA-FLSs 17 . In this study, relative mRNA expression of AMPK-and HAPLN1 in untreated RA-FLSs showed a positive correlation.
However, both si-HAPLN1 and HAPLN1 OE treated RA-FLSs had down-regulated AMPK-expression at mRNA but not protein level, which suggests a complex feedback circle between AMPK-and HAPLN1.
Furthermore, on silencing HAPLN1, pro-in ammatory factors such as TNF-, MMPs and IL-6 as well as structure related molecules like TGF-β, Fibronectin and ACAN were down-regulated. Conversely, HAPLN1 OE treated RA-FLSs had up-regulated TNF-, MMPs, IL-6 and ACAN expression. In untreated RA-FLSs, relative mRNA expression of HAPLN1 was positively associated with TGF-β and IL-6. The Ki-67 has been widely used as a proliferation marker for most human tumor cells for decades 33 , which was decreased after si-HAPLN1 transfection but increased by HAPLN1 OE treatment. In addition, cyclin D1 has an important part in regulating cell proliferation during G1 phase of the cell cycle. Given the frequent overexpression of cyclin D1 in cancer cells, its expression appears to be closely linked with carcinogenesis 34 .
The oncological ndings suggest that cyclin D1 has a central role in mediating invasion and metastasis of cancer cells by controlling Rho/ROCK signaling and matrix deposition of thrombospondin-1 35 . In this study, the mRNA expression levels of cyclin D1 in RA-FLSs was signi cantly decreased by HAPLN1 OE treatment, which might possibly explain its inhibitory ability on FLSs migration.
Proteomic and mRNA-seq results showed HAPLN1 function in RA-FLSs from a holistic view. Proteomic analysis suggested si-HAPLN1 transfected RA-FLSs were enriched in pro-in ammatory pathways with down-regulated DEGs. It is not strange that mRNA and protein levels seem to have a low correlation, as the multi-step process of gene expression involves transcription, translocation, turnover of mRNAs and proteins 36 . Although only 2 DEGs overlapped with proteomic and transcriptional studies with rHAPLN1 treated RA-FLSs, the omics study re ected activation of in ammation, proliferation, an increased cell adhesion and strengthening of ECM function.

Conclusion
In conclusion, HAPLN1 accelerates proliferation of RA-FLSs to form a pathological pannus, mimicking cancer cells. Based on physiological development and oncology studies, HAPLN1 seems to be an oncogene, however, with an opposing function on cell adherence and inhibition of migration ability. These ndings were also con rmed by molecular network consisting of MMPs, IL-6, Ki-67 and TGF-β. Thus, HAPLN1 can be considered as a pathogenic factor; however, more questions are inevitable: Why HAPLN1 expression is higher in RA patients with shorter disease course? HAPLN1 seems closely related with metabolism, and positively correlated with AMPK, but which one is an upstream regulator? Availability of supporting data The datasets used and/or analysed during the current study are available from the corresponding author or rst author on reasonable request.

Competing interests
The authors declare no competing interests.
HAPLN1 is up-regulated in RA patients. (A) Volcano plot showing higher expression of HAPLN1 gene in FLSs from RA than OA patients. (B) Increased HAPLN1 expression in RA than OA synovial membranes (C) Plasma HAPLN1 levels were signi cantly enhanced in RA patients than OA patients and healthy people. Error bars indicate ± SD. *p < 0.05, ***p < 0.001.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.