Artemether Targets Raptor-Induced Actin Polymerization and Suppresses Migration of Fibroblast-like Synoviocytes in Rheumatoid Arthritis

Background: Fibroblast-like synoviocytes (FLS) in rheumatoid arthritis (RA) may cause articular damage as a result of its aggressive features including direct adhesion and invasion of surface cartilage in joints. Artemether (ART), one of the artemisinin derivatives with antimalarial properties, showed inhibitory effect on inammation and destruction of joints in collagen-induced arthritis rats, which might be applied in RA treatment. However, whether ART has effects on the aggressive properties of human RA-FLS remains unexplored. Methods: Synovium was obtained from patients with active RA (n=18) and FLS were isolated in vitro. RA-FLS were subjected to cell migration, invasion assays, live-cell imaging analysis and Rho GTPase activation assay after ART treatment. To identify the therapeutic target of ART, key signaling molecules of PI3K/Akt, AMPK, MAPK, NF-κB and mTOR pathways from RA-FLS were examined by Western Blot after ART treatment. Raptor was knockdown or overexpressed by siRNA or lentivirus transfection to reveal its role on regulating the aggressive properties of RA-FLS. Results: ART treatment signicantly suppressed the transwell migration and invasion of synovial FLS from RA patients. Time-lapse microscopy revealed that ART treatment reduced random migration velocity of RA-FLS, as well as the directional persistence. ART also impaired the formation of lopodia and lamellipodia in RA-FLS. Further mechanism investigation showed that ART reduced the protein level of Raptor, a critical component of the mTOR pathway, and its downstream target 4E-BP1. It also inhibited the activation of Rho GTPases and the expression of actin binding proteins, including Prolin 1 and p-Colin. Raptor overexpression could reverse the anti-migration and anti-invasion effects of ART on RA-FLS as well as the suppression of Rho GTPases activation and the expression of actin binding proteins. Conclusion: ART can inhibit migration of RA-FLS by blocking Raptor-induced actin polymerization. ART might be a potential agent targeting FLS in RA treatment.


Introduction
Rheumatoid arthritis (RA) is an aggressive immune-mediated joint disease with persistent synovitis and bone erosion, leading to disability and increased risk of mortality [1]. Systemic immune dysregulation and autoimmunity are considered as driving forces in the evolution of RA. However, though increasing number of drugs are available including the biologics and small molecules which target immune cells, cytokines or in ammatory signaling pathway, less than half of RA patients achieve ACR70 e cacy in clinical trials and there are still 10%-15% patients with refractory RA in real world practice [2]. On the other hand, current therapies for RA targeting the immune system are accompanied by potential risk of infection and malignancy. Therefore, new treatment options need to be developed especially effective and safe drugs with different targets for RA.
The main clinical manifestation of RA are synovitis and joint destruction which highlights the roles of stromal elements especially broblast-like synoviocytes (FLS) [3]. More and more evidences showed that RA-FLS not only strongly respond to an in ammatory environment, but also exhibit unique aggressive behaviors, actively and autonomously promote in ammation and directly adhere and invade to adjacent cartilage [4]. Bene ting from the new technologies, the characterization of speci c cell types of RA synovium was analyzed in single-cell level, resulting in the identi cation of broblast subset with CD90+ which exhibited elevated migratory and invasive behavior [5,6]. In vivo study showed that injection of PDPN+FAPα+CD90+ broblast into the ankle of serum transfer induced arthritis mice resulted in more severe and sustained joint swelling with increased leukocyte in ltration [7]. The invasive behavior of RA-FLS is also positively associated with the rate of radiographic destruction of RA patients [8]. Recurrent or refractory RA might be due to the tumor-like phenotype of RA-FLS, and targeting FLS to restore articular homeostasis might be a potential option for RA therapy.
Methotrexate (MTX) is generally an anchor drug for RA treatment for its multiple mechanisms contributing to anti-in ammatory actions. But it also causes a variety of adverse events ranging from hair loss to fatal myelosuppression [9]. Antimalarials drugs are also suggested in RA management recommendation especially hydroxychloroquine (HCQ) which is mainly reserved for patients with mild RA because of its weak clinical e cacy and no joint protection [10]. Recent evidences showed that artemisinin and its synthetic derivatives had the potential therapeutic value for several autoimmune diseases including RA [11]. Artemether (ART), a peroxide sesquiterpenoid lipid-soluble derivative of artemisinin, is widely used in the treatment of Plasmodium falciparum and cerebral malaria [12]. It also exhibits a variety of potential anti-cancer, anti-in ammatory and anti-oxidant abilities [11,13,14]. More recently, a study showed that ART could inhibited RANKL-induced osteoclastogenesis in vitro via inhibition of MAPKs activation in bone marrow-derived macrophages isolated from mice [15]. These results implied that ART might be a new option for RA therapy. However, whether ART has effects on the invasive features of RA-FLS remains unexplored. Here, we aimed to explore whether ART had effects on the migration and invasion of RA-FLS as well as its regulation mechanism. FLS were isolated in vitro and identi ed by the light microscope and ow cytometry according to our previous described methods [18,19]. FLS was identi ed under the light microscope for morphology and further determined by the surface markers of CD90 and CD68 using ow cytometry. Primary FLS from passage 3~5 were used in further functional and molecular biological experiments.

Patients and FLS Preparation
Pharmacological Intervention on RA-FLS Primary RA-FLS were serum starvation for 6 hours before treated with corresponding concentrations of ART (Sigma-Aldrich, St. Louis, MO, USA) for indicated time, which was dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich, St. Louis, MO, USA, less than 0.1% v/v) and then diluted to working concentration with Dulbecco's Modi ed Eagle Medium (DMEM, Thermo Fisher Scienti c, Waltham, MA, USA). MTX IC 5 -10 nM and HCQ IC 5 -20 μM were used as control according to our previous study [19]. DMSO diluted with DMEM in equal volume was taken as untreated.

ELISA Assay
Primary RA-FLS were cultured in serum-free medium for 24 hours before incubated with 100 pg/mL tumor necrosis factor α (TNF-α, PeproTech, Rocky Hill, NJ, USA) and corresponding concentrations of ART, MTX or HCQ. ELISA assays were performed to detect IL-1β, IL-6 and IL-8 in the cell culture supernatants with a detecting absorbance of 450 nm. Each plate test was repeated for three times.

Cell Migration and Invasion Assays and Live-cell Imaging Analysis
Primary RA-FLS were seeded at a density of 1 × 10 5 cells/well in six-well plates. After reaching a monolayer con uence over 90%, RA-FLS were treated with Interventions of ART, MTX or HCQ at indicated concentrations were added to RA-FLS for 24 hours. Horizontal migration was detected by wound healing assay, while vertical migration and invasion were detected using transwell assays [19].
Primary RA-FLS were seeded at a density of 2 × 10 3 cells/well in 12-well plates and treated with indicated concentrations of ART, MTX or HCQ for 24 hours. Then live-cell imaging was performed on an inverted microscope Zeiss Axio Observer Z1 (Carl Zeiss, Berlin, Germany) using phase contrast illumination with a 10× (0.30 Ph1) objective at 37℃ under 5% CO 2 . For random migration, images were captured at 15-min intervals and cells tracked using the Manual Tracking plugin for ImageJ software (NIH, Bethesda, Maryland, USA). In order to avoid errors caused by cells that divided or contacted with other cells during the experiment by the software analysis, we use manual methods for data analysis. Velocity was calculated as the total track distance divided by the total time (720 mins), and displacement rate (D/T) was calculated as the linear distance (D) divided by the total track distance (T). Thirty cells from at least three independent experiments were analyzed. Quantitative Real-time PCR Analysis (qPCR) and Western Blot Analysis Total RNA of primary FLS was extracted with Trizol (Takara, Kusatsu, Japan) and reverse-transcription was performed using PrimeScript TM RT reagent Kit (Takara, Kusatsu, Japan) [20]. The primer sequences of CDC42, Rac1, RhoA, Pro lin 1, Co lin, mTOR, Raptor, Rictor, GβL, 4E-BP1, p70S6K and GADPH were showed in Supplementary information table 1.

Cells Transfection for Knockdown or Overexpression of Raptor
Primary RA-FLS were transfected with Raptor-siRNA and negative control (Genepharma Co., Ltd, Shanghai, China). One of three independent designed siRNAs for Raptor knockdown was selected according to the transfection e ciency: 5'-GCUAAACAACAUGUCGCCATT-3', 5'-UGGCGACAUGUUGUUUAGCTT-3. Transfections were performed using the Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Rho GTPase Activation Assay GTP-bound RhoA, Rac1, and Cdc42 were measured using corresponding Activation Assay Kits (Cytoskeleton, Denver, CO, USA). The PAK-PBD a nity beads were incubated together with glutathioneagarose beads and then the complex (PAK-PBD/GTP) were recovered from the beads and analyzed using SDS-PAGE experiment as indicated by the manufacturer's protocol. Protein concentrations and activity forms of RhoA, Rac1, and Cdc42 in supernatants were assessed according to the manufacturer's instructions.

Immuno uorescence Staining
Primary RA-FLS were treated with different interventions for 24 hours and incubated with monoclonal rabbit anti-human Pro lin antibody and mouse anti-human p-co lin antibody (Cell Signaling Technology, Danvers, MA, USA) overnight at 4°C was followed by incubation with phalloidin-conjugated rhodamine (Beyotime, Bejing, China), Alexa Fluor® 488-labeled goat anti-rabbit IgG and Alexa Fluor® 647-labeled goat anti-mouse IgG secondary antibody for 60 mins in a dark wet box at 37°C. Nucleus was stained with DAPI for 3 mins and coverslips were mounted with ProLong Gold Antifade Reagent (Invitrogen, Carlsbad, CA, USA). Images were analyzed using a Zeiss LSM 710 Confocal Imaging System.

Statistical Analysis
All statistical analyses were conducted using SPSS 21.0 statistical software and GraphPad 8.0 Software. Comparison between two groups was detected using Student's t-test. The data were presented as mean ± SD or median and range for continuous variables and presented as frequencies and percentages for categorical variables and P < 0.05 was considered signi cant [22].

Results
ART Inhibited the Production of Cytokines from RA-FLS, but not the Cell Viability, Proliferation or Apoptosis Among 18 RA patients recruited in this study, their mean age was 49 ± 8 years with the median disease duration 64 months (range from 3 to 240 months), and the median CDAI 32 (range from 21 to 44).
Among these patients, 78% of them had positive rheumatoid factor, 61% had positive anticyclic citrullinated peptide antibody. There were 22% patients had not received any glucocorticoids or disease modifying anti-rheumatic drugs (DMARDs) therapy for three months before recruitment.
Primary RA-FLS were cultured in vitro and identi ed with unique spindle morphology and high level of CD90 expression (94.62%±5.37%) but low level of CD68 expression (1.40%±0.75%) using ow cytometry ( Figure S1). CCK-8 assay showed the IC 50 of ART under 24 hours treatment on primary RA-FLS were up to 761 μM ( Figure S2A). RA-FLS exhibits tumor-like features including excessive proliferation and resistance of apoptosis. CCK-8, EdU and ow cytometry assays showed that ART IC 5 -20 μM as well as the controls of MTX IC 5 -10 nM or HCQ IC 5 -20 μM had no effect on cell viability, proliferation, cell-cycle or apoptosis of RA-FLS (all P > 0.05, Figure S2B-D).
RA-FLS have also emerged as important immune modulator as they can secrete an abundance of proin ammatory cytokines, such as IL-1β, IL-6 and IL-8. Mimicking the in ammatory microenvironment of RA articular cavity by TNF-α 100 pg/mL stimulation for 24 hours, ART IC 5 -20 μM, MTX IC 5 -10 nM or HCQ IC 5 -20 μM signi cantly inhibited the level of IL-1β, IL-6 and IL-8 by 18.2% ~ 63.1% reduction (all P 0.05, Figure S2E). These results showed that ART can inhibit the proin ammatory cytokines production of RA-FLS at safety concentrations.
ART Exhibited Anti-migration and Anti-invasion Effects on RA-FLS RA-FLS have the migratory potential by forming lopodia and lamellipodia, which leads to invasion of the pannus-cartilage interface of joints. To explore the effects of ART on RA-FLS migration and invasion, primary RA-FLS were pretreated with ART, MTX or HCQ at indicated concentrations for 24 hours. Compared with untreated RA-FLS, ART at 20 μM signi cantly inhibited their migration distance in wound healing assay (1063±84 μm vs. 632±43 μm) and 38% less migrated cells in transwell migration assay at 12 hours (364±27 cells/well vs. 226±21 cells/well), as well as their invasion with 43% less invaded cells in transwell invasion assay at 24 hours (312±50 cells/well vs. 178±21 cells/well, Figure 1).
Tracing of migration tracks of sparsely seeded cells of RA-FLS by time-lapse microscopy, the results showed that ART at 20 μM signi cantly inhibited random migration velocity (0.294±0.099 μm/min vs. 0.386±0.104 μm/min, P<0.01), and the directional persistence with 36% decrease in displacement rate. Phalloidin staining images showed that ART at 20 μM could impair the formation of lopodia and lamellipodia of RA-FLS with 45% reduction (41.14%±8.69% vs. 22

ART Inhibited the mTOR Pathway Activation in RA-FLS
There are several signaling pathways usually involved in RA-FLS which might be the upstream regulators of the Rho GTPases, including MAPK, NF-κB and mTOR pathways which are associated with in ammation, and integrin/galectin triggered PI3K/Akt pathway, modulator of energy balance AMPK pathway [23][24][25]. In order to identify the therapeutic target of ART, we examined the key signaling molecules of PI3K/Akt (PI3K, p-PI3K, Akt and p-Akt), AMPK (AMPK-α and p-AMPK-α), MAPK (p38-MAPK and p-p38 MAPK), NF-κB (NF-κB p65 and NF-κB p-p65) and mTOR (mTOR, p-mTOR, Raptor, Rictor and GβL) pathways after treatment with ART, MTX or HCQ at indicated concentrations on primary RA-FLS for 24 hours. ART at 20 µM signi cantly inhibited Raptor expression, whereas it did not alter the expression of signaling molecules from other pathways. MTX at 10 nM signi cantly inhibited Raptor and NF-κB p-p65, whereas HCQ at 20 μM had no effect on these pathways ( Figure 3A, B and Figure S3).
We further investigated whether ART had effect on the activation of the mTOR pathway components. Raptor and Rictor are the key proteins of mTORC1 and mTORC2, respectively. We found that ART at 20 µM signi cantly inhibited the expression of Raptor but not Rictor ( Figure 3C). 4E-BP1 and p70S6K1 are the major substrates of Raptor. After treatment with ART, MTX or HCQ at indicated concentrations on primary RA-FLS for 24 hours, western blot showed that ART at 20 µM signi cantly inhibited the expression and the phosphorylation of 4E-BP1, but not p70S6K, which implied that ART might inhibit the activation of mTOR pathway through suppressing Raptor and 4E-BP1 expression. MTX at 10 nM signi cantly inhibited Raptor and its downstream regulators of 4E-BP1, p-4E-BP1 and p70S6K, whereas HCQ at 20 μM had no effect on them ( Figure 3D).

Elevated Raptor in RA-FLS Promoted Migration and Invasion with Actin Cytoskeleton Regulation
Little was known about Raptor expression in RA-FLS as well as its function. The results of qPCR and western blot showed that Raptor prominently up-regulated over 2.2-fold in RA-FLS in mRNA and protein levels when compared with that in OA-FLS or Orth.A-FLS, respectively (both P<0.05, Figure 4A). Raptor knockdown in RA-FLS by siRNA transfection signi cantly suppressed their migration and invasion, as well as the Rho GTPase activity and the expression of Pro lin 1 and p-Co lin (Ser3). Conversely, Raptor overexpression by lentivirus infection had opposite effects on RA-FLS ( Figure 4B~D).
Immuno uorescence staining of Pro lin 1, p-Co lin (Ser3) and F-actin in RA-FLS showed that Raptor knockdown in RA-FLS led to a reduction in lamellipodia, lopodia and actin stress formation, whereas Raptor overexpression promoted their formation ( Figure 5). These results implied that elevated Raptor in RA-FLS can promote the aggressive properties, activate Rho GTPase and regulate actin cytoskeleton.

Discussion
In the present study, we rst report that ART has anti-migration and anti-invasion as well as antiin ammatory effects on RA-FLS, which are similar to MTX but greater than HCQ. We further nd a novel role of Raptor, a critical regulator among mTOR pathway, on promoting migration and invasion of primary RA-FLS via activating Rho GTPase and regulating actin cytoskeleton, while ART can block the Raptor/4E-BP1/Pro lin 1 pathway and regulate actin cytoskeleton to exert its inhibitory effects ( Figure 6). Our study provides experimental evidence of the potential therapeutic value of ART on targeting FLS in RA treatment and reveals its therapeutic target of Raptor on regulating the aggressive properties of RA-FLS.
The antimalarial effects of artemisinin and its derivatives base on the peroxide bridges which will be broken into the nucleophilic radical metabolites by the heme or free iron especially in parasite-infected red blood cells. Then the free radicals acting as alkylating agents, leading to parasitic death [26]. Previous studies showed artesunate, one of the artemisinin derivatives, suppressed IL-1β, IL-6, IL-8 and VEGF expression by inhibition of the key in ammatory signaling such as NF-κB, PI3K/Akt or MAPK pathways [27][28][29]. More recently we reported that artesunate could inhibit the migration and invasion of RA-FLS, implying the potential therapeutic value of artemisinin derivatives for RA. 19 ART is a lipid-based derivative and has a longer half-life of 1.6 hours than that of water-soluble artesunate as 0.3 hours according to the labels from FDA [30]. Here, we nd that ART has anti-migration and anti-invasion effects on RA-FLS, suggesting that ART might be another e cient agent targeting FLS in RA.
We further investigated the mechanism of ART on inhibiting RA-FLS migration and invasion. RhoA, Rac1, and Cdc42 are the best characterized members of Rho GTPases. RhoA initiates the formation of contractile actin, as well as stress bers, while Rac1 and Cdc42 can promote the lamellipodia and lopodia formation, respectively [31]. Rho GTPases are reported to play critical roles on the aggressive properties of RA-FLS which majorly regulate their migration [32,33]. The actin binding proteins including Pro lin 1 and Co lin are downstream targets of Rho GTPases which regulate actin polymerization and enable cell motility. Co lin mediates the actin laments disassembly but the co lin (Ser3) phosphorylation inactivates its function, while Pro lin 1 plays a critical role on actin polymerization [34]. In our study, we nd that ART signi cantly inhibits the total mRNA and protein expression of Rho GTPases and their activation, as well as the actin binding proteins of Pro lin 1 and pco lin (Ser3), which clarify the role of ART on suppressing RA-FLS migration and the following invasion by blocking the actin polymerization.
The mTOR is a serine/threonine kinase regulating various cell functions such as cell proliferation and survival under the stimulation of cytokines and hormones [35]. It is composed of two multi-protein complexes, which are termed mTORC1 and mTORC2. The mTORC1 is involved in mRNA translation, protein turnover and metabolism, while mTORC2 regulates proliferation and survival [36]. Raptor promotes the recruitment of mTOR substrates and is necessary for mTORC1 to achieve subcellular localization [37]. 4E-BP1 is one of the signature substrates phosphorylated by Raptor and regulates protein synthesis of RhoA, Rac1, and Cdc42 [38]. Inhibition of mTOR reduced synovitis as well as articular cartilage and bone destruction in human TNF-transgenic mice [39]. Rapamycin inhibited lamellipodia formation and RA-FLS invasion by suppressing the phosphorylation of mTOR and its substrates including p70S6K1 and 4E-BP1 [40]. Recent in vitro study showed that IL-17-induced L-type Amino Acid Transporter 1 overexpression might enhance the migration of RA-FLS via the mTOR/4E-BP1 pathway [41]. In this study, we nd that Raptor is the therapeutic target of ART and veri ed that Raptor is the critical component of mTORC1 on promoting migration and invasion and regulating actin cytoskeleton.
We also compared the e cacy of ART with MTX or HCQ at the same safety concentrations. MTX has anti-in ammatory effect on RA-FLS by stimulating adenosine and its receptor, leading to inhibition of NF-κB activity [9]. However, MTX might also cause toxic effects including stomatitis, nausea, and bone marrow depression because of its antifolate property, so it is used in a low dose in clinical practice. Previous study on collagen-induced arthritis rats showed that compared with intraperitoneally injection of MTX, intramuscularly administration of ART (3 mg/kg/day) similarly reduced in ammation and bone erosion in the paws with greater tolerability [42]. In another study, intraperitoneal injection of ART (10 mg/kg/day) showed anti-brosis effect in hepatic brosis mice model [43]. HCQ has antiin ammatory effect on lymphocytes by inhibiting lysosomal and autophagosome functions [44].
Intervention of HCQ on RA-FLS is rarely reported. Our previous study showed that HCQ could inhibit IL-1β, IL-6 and IL-8 production, but not the migration and invasion of RA-FLS [19]. In the present research, we nd that ART has anti-migration and anti-invasion as well as anti-in ammatory effects on RA-FLS which is similarly to MTX and superior to HCQ. The effective dose of ART in our study was 20 µM (approximately equivalent to 5.96 mg/kg), which was higher than the dose of previous animal study but was in the safety range of no more than 40 mg/kg [30,42]. Though the working dose of ART is obviously higher than MTX in vitro study, the dose of ART (3 mg/kg/day) reduces to about 23 times of that of MTX (0.3 mg/kg for 3 times weekly) in the treatment of collagen-induced arthritis rats [42]. It is applicable to obtain the effective dose in our study in vivo through systemic or localized administration and ART might be a safety and effective medicine for RA therapy.
In conclusion, our study rst reports the novel anti-migration and anti-invasion effects of ART on RA-FLS, and illuminates its mechanism of blocking Raptor-induced actin polymerization and therefore suppressing lamellipodia, lopodia and actin stress formation. ART might be a potential agent targeting FLS in RA. Representative images were shown (original magni cation, ×400 above and analyzed by qPCR and western blot. (B) Activity of RhoA, Rac1 and Cdc42 were measured by RhoA/Rac1/Cdc42 activation assay. (C) Total mRNA and protein expression of Pro lin 1, Co lin and pco lin (Ser3) were analyzed by qPCR and western blot. Relative protein expression was normalized by GAPDH. Data were represented as means ± SD from 6 RA patients. *P<0.05, **P<0.01, ***P<0.001, compared with untreated (RA-FLS treated with DMSO). ART: artemether, MTX: methotrexate, HCQ: hydroxychloroquine.
Relative protein expression was normalized by GAPDH. Data were represented as mean ± SD from 6 RA patients. **P<0.01, ***P<0.001, compared with untreated (RA-FLS treated with DMSO). ART: artemether, MTX: methotrexate, HCQ: hydroxychloroquine.  and F-actin were performed in primary RA-FLS. Representative images were shown (original magni cation, ×200 above and ×1000 below). The yellow box indicates lamellipodia and lopodia formation. (B) Statistics for the mean uorescence intensity of Pro lin 1 or p-co lin and the percentage of RA-FLS with pseudopodia. Data were represented as means ± SD from 6 RA patients. *P<0.05, **P<0.01.