The Anti-Angiogenesis Mechanism of Geniposide on Rheumatoid Arthrits is Related to the Regulation of PTEN

Rheumatoid arthritis (RA) is a systemic immune disease characterized by joint inammation and pannus. The nascent pannus contributes to synovial hyperplasia, cartilage and tissue damage in RA. Based on the previous study on the anti-inammatory effect of Geniposide (GE) on arthritis rats, this study aims to explore the therapeutic effect and potential mechanism of GE on RA angiogenesis, involving the participation of phosphate and tension homology deleted on chromosome ten (PTEN) and downstream pathways. Clinical manifestations, synovial pathomorphology, microvessel density and the level of angiogenesis-related factors were used to evaluate the therapeutic effect of GE on adjuvant-induced arthritis (AA) rats. The proliferation, migration, tube formation of human umbilical vein endothelial cells (HUVECs) indicate the degree of angiogenesis in vitro. Lentivirus overexpression of PTEN was employed to elucidate the potential mechanism. The results showed that GE improved the degree of arthritis and angiogenesis in AA rats. PTEN was decreased signicantly in vivo and in vitro, and over-expression of PTEN improved the biological function of HUVECs to inhibit angiogenesis. GE inhibited the proliferation, migration and tubule formation of HUVECs and plays an anti-angiogenesis role in vitro. Mechanism study showed that PTEN expression was increased and p-PI3K and p-Akt expression was decreased with GE treatment. It suggests that GE up-regulated the expression of PTEN and inhibited the activation of PI3K-Akt signal, which plays a role in inhibiting angiogenesis in RA in vivo and in vitro.


Introduction
Rheumatoid arthritis (RA), as a chronic and in ammatory autoimmune disease, is characterized by multiple in ammatory cells in ltration, pannus formation and synovial hyperplasia. Among them, angiogenesis runs through the whole process of RA, which is one of the main reasons of RA with longstanding incurability [1,2] . Neovascularization provides oxygen, nutrients and signal molecules for highly proliferative synovial tissue, maintains the continuous proliferation of broblast-like synoviocytes (FLS), and increases the recruitment of in ammatory cells into the tissue, leading to synovitis and tumor-like proliferation of FLS. The abnormally proliferating FLS aggravates the process of angiogenesis by secreting in ammatory cytokines and/or angiogenic factors, such as growth factors, hypoxia inducible factors, chemokines, matrix metalloproteinases and adhesion molecules [3,4] .
Angiogenesis is the process of forming new blood vessels from the pre-existing vessels, which has been proved to be one of the pathogenesis of RA [5] . Under physiological conditions, blood vessels remain quiescent and rarely form new branches. Whereas in developmental, tissue repair or pathological situations, endothelial cells (ECs) form new blood vessels through a series of complex processes. Firstly, extracellular matrix is degraded under the stimulation of angiogenesis signals, such as vascular endothelial growth factor (VEGF), which is produced by changes in the external environment or surrounding tissues. Activated ECs rapidly proliferate and migrate into tissues to form new lumens [6,7] . Therefore, the inhibition of angiogenesis caused by the proliferation and migration of ECs in RA has become an important goal in the treatment of RA.
In addition to angiogenic factors or extracellular signals, the genetic changes and mutations of tumor suppressor genes also trigger the process of angiogenesis. Phosphate and tension homology deleted on chromosome ten (PTEN) was rst known as a tumor suppressor gene, which is involved in various cellular processes including cell proliferation, survival, angiogenesis, and tumor growth by regulating intracellular signaling [8][9][10] . PTEN, as a negative regulator of PI3K, regulates PI3K-Akt signal and participates in angiogenesis. The direct evidence is that the extensive sprouting of new blood vessels and the enlargement of original blood vessels can be observed by the over-expression of PI3K and Akt by RCAs retroviral vector [11] . In addition to inhibiting PI3K activation, some studies have found that PTEN regulates the expression of VEGF by activating JNK and ERK signals and participates in the process of angiogenesis [9,12,13] . This provides evidence for the important role of PTEN in regulating cell growth, migration and angiogenesis.
Geniposide (GE) is a kind of iridoid glycoside obtained from Gardenia jasminoides Ellis by modern technology of traditional Chinese medicine, which has pharmacological effects of anti-in ammatory and anti-angiogenesis [14] . Previous work of our group focused on the therapeutic effect of GE on experimental arthritis. It was found that GE regulates the VEGF level in FLS and restores the dynamic balance of pro/ anti-angiogenic factors in AA rats [15,16] . The purpose of this study is to explore the role of PTEN in angiogenesis of AA rats and the relationship between downstream PI3K-Akt signal and the antiangiogenesis of GE in vivo and in vitro.

Drugs and reagents
Geniposide is a white powder with 98% purity and purchased from Guangxi Shanyun Biochemical

Animals and cells
Healthy Sprague-Dawley (SD) rats treated with complete Freund's adjuvant (FCA) at 0.1mL by subcutaneous administration of the left posterior paw to induce adjuvant arthritis, an animal experimental model of RA. Rats without paw swelling were given 0.05 mL FCA supplement before the 7th day. The model was evaluated after 17 days, including arthritis index, paw swelling, and mobility score, oral administration was continued for a week. Normal control and model group were given the same dose of normal saline. The speci c experimental design is shown in Fig. 1. Male SD rats of Speci c pathogen Free (SPF) grade were used in this experiment and weighed 160-190 g. The quality testing unit is Shandong Animal Experimental Center, and the ethics report is issued by the experimental animal ethics committee of Anhui University of traditional Chinese medicine. Human umbilical vein endothelial cells (HUVECs) were provided by BeNa Culture Collection (BNCC, Beijing, China) (No 347734)and grown in an epithelioid adherent manner.

Evaluation of AA rat model
Arthritis index, paw swelling, mobility score were used to evaluate the model. The arthritis index was judged according to the degree of joint swelling, and the speci c criteria were as follows: 0 point indicating normal without swelling; 1 point indicating slight swelling and erythema of ankle joint; 2 points indicating slight swelling and erythema from ankle joint to toe joint or metacarpal joint; 3 points indicating moderate swelling and erythema from ankle joint to toe joint or metacarpal joint; 4 points indicating severe swelling and erythema from ankle joint to toe joint. The maximum score of each rat was 12. Paw swelling was measured with a foot volume meter every 3 d after the appearance of in ammation, and the degree of foot swelling was calculated (Δ mL=foot volume after modeling-foot volume before modeling). Mobility score was performed every 3 d by the non-participants personnel. Each rat was placed in a box with a certain movement space and observed for more than 30 s. The speci c scoring criteria are as follows: 0 point for only lying down; 1 point for crawling; 2 points for standing but di culty walking; 3 points for walking but di culty running; 4 points for normal walking.

Histopathological assessment
The synovial tissue was xed in 4% paraformaldehyde tissue solution, dehydrated and embedded in para n. The sections were then stained with Hematoxylin and Eosin (H&E). The pathological characteristics of synovial tissue were observed under microscope (OLYMPUS BX51, Germany).

Immunohistochemistry
The expression level of CD31 on the surface of vascular endothelial cells (VEC) in synovial tissue, a speci c marker of VEC, was detected by speci c anti-CD31 antibody. Synovial tissue was rst xed with formalin and then embedded in para n. Sections were incubated with H 2 O 2 to block endogenous peroxidase. Then, goat anti-CD31 (1:100) and VEGF antibody (1:150) were incubated overnight at 4 o C.
Finally, sections were incubated with secondary antibodies, treated with 3, 3 '-diaminobenzidine and counterstained with hematoxylin. VEC positive for CD31 clustered into brown as a single microvessel. The average optical density of positive staining was quanti ed as the relative expression level of the target protein by Image J software analysis.

ELISA
The level of angiogenic factor in serum of AA rats was measured by VEGF speci c ELISA kit according to the manufacturer's protocol (J&L Biological, China). Absorbance was determined at 450 nm to quantify the samples. Six replicates were set for each sample.

Cell culture
HUVECs were cultured in high-glucose Dulbecco's modi ed Eagle's medium containing 10% fetal bovine serum and 1% penicillin-streptomycin solution (100×). In order to induce in ammatory conditions, HUVECs stimulated by TNF-α was selected to establish an in ammatory model in vitro. In addition, according to the previous protocol, the cell co-culture technology in vitro has also been applied to better simulate the in ammation in vivo and compare with TNF-α stimulation alone. The speci c method was that FLS derived from synovial tissue of AA rats by in vitro culture were treated with TNF-α (10 ng/mL) for 24 hours. The conditioned medium (FLS-CM) was collected and diluted to different concentrations with DMEM (0%, 5%, 25%, 50%, 75% and 95% ) to stimulate HUVECs for 24 hours.The optimal stimulation concentration was selected for experiments.

Infection and validation of lentivirus
PTEN over-expression lentiviral vectors were obtained from Hanbio Biotechnology (Shanghai, China). Lentivirus infection of HUVECs induced over-expression of PTEN, and an empty GFP vector (Lv-GFP) was used as a control. After infection, the results of over-expression were detected by Immuno uorescence and RT-qPCR.

Edu proliferation assay
HUVECs were seeded on the round coverslip in 48-well plates at a density of 10 5 cells/mL. Different concentrations of GE or inhibitors Bpv were administered after incubation in the DMEM medium containing 10% FBS for 24 hours. The following protocols were performed by k uor488 Edu cell proliferation detection kit (KeyGEN BioTECH, Nanjing, China). After successful staining, the coverslip were then coated with nail polish on the slide. Finally, it was imaged and analyzed under uorescence microscope.

Cell cycle analysis
HUVECs were treated with different concentrations of GE or Bpv to analyze the cell cycle when growing to a density of 50%. The cells were digested and xed in 70% ethanol overnight at 4 o C. The cells were resuspended in PBS the next day, and stained with 400 µL propidium iodide (PI) solution (50 µg/mL) (Solarbio, P8080, Beijing, China) and 100 µL RNase A (100 µg/mL) (Solarbio, R1030, Beijing, China) for 30 min in the dark. Finally, it was analyzed by ow cytometry (Beckman Coulter, FC500) and FlowJo software (Version 10.6.0).

Cell counting kit-8 assay
The cells were planted in 96-well plates at the density of 10 4 cells/well and cultured in 37 o C incubator for 24 hours. Six wells in each group were treated with TNF-α (10 ng/mL) or GE (25, 50, 100 µM) and incubated for 24 hours. Then 10 µL of CCK-8 solution was added to each well and continue to culture for 2 hours. The optical density (OD) at 450 nm was measured with a microplate reader (SpectraMax iD3, Molecular Devices, the United States), and the cell proliferation rate was calculated based on OD450 values.

Wound-healing analysis
The wound-healing experiment was used to determine the migration ability of cells. HUVECs were seeded into 6-well plates at a density of 6×10 5 cells/well, and cells were cultured to adherent and grown to a density of 90%. The cells were scratched vertically with 200 µL sterile pipette and ruler, leaving a vertical line. The slipped cells were washed, and photographed under a microscope observation. This time point was set at 0 h. Next, TNF-α (10 ng/mL) or GE (25, 50, 100 µM) were treated and photographed after incubation for 6, 12, 24 hours, respectively. The scratch area at different time points was measured by Image J software, and the rate of wound healing was calculated to represent the migration ability of cells.

Vertical migration analysis
HUVECs (10 5 cells/mL) were seeded in the upper chamber of a polycarbonate transwell chamber (Corning, United States) with 8 µm pore membranes in order to evaluate the vertical migration ability of cells. The upper chamber medium was replaced with the low concentration FBS (5%) medium containing TNF-α (10 ng/mL) or GE (25, 50, 100 µM), and the lower chamber was the medium containing 10% FBS. After 24 h of culture, the chamber was removed and washed with PBS once. The cells that did not pass through the membrane were carefully wiped off with a cotton swab, and the cells under the membrane were xed with 4% paraformaldehyde. The membrane was dried naturally and then stained with crystal violet for 20 min. Finally, the membrane was xed on a glass slide and observed under an inverted microscope (Leica DMI6000B, Germany). The number of migrating cells was analyzed by Image J software.

Tube formation assay
Before the experiment, the 96-well plate and sterile pipettes were pre-cooled at -20 o C in advance, whereas Matrigel (Corning, United States) was thawed at 4 o C. 50 µL of Matrigel was taken in a 96-well plate and incubated for 30 min to solidify in a 37 o C incubator. Cell suspensions made of different groups were seeded into 96-well plate containing Matrigel at a density of 2×10 4 cells/mL. Images were taken by a microscope after 24 hours of incubation (Leica DMI6000B, Germany). The number of tubes and the length of branches were calculated by Image J software to represent the lumen forming ability of cells.

Western blot analysis
HUVECs of different groups grown to 80% were washed three times with ice-cold phosphate buffered saline (PBS). 500 µL RIPA lysate containing 1% PMSF protease inhibitor was lysed and then centrifuged at 12000 rpm at 4 o C for 30 min. The supernatant was collected and the total protein was quanti ed by BCA protein quanti cation kit (Beyotime, China). The equivalent protein samples were separated by 8-10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to PVDF membranes (Biosharp,Shanghai, China) under constant current condition of 200 mA. The membrane was blocked in 5% (w/v) skimmed milk for 2 h to block non-speci c protein binding. The membrane was incubated with primary antibody at 4 o C overnight after washing with Tris-buffered saline containing Tween-20 (TBST). After that, it was incubated with goat anti-rabbit or mouse IgG horseradish peroxidase conjugated secondary antibody for 1 h at room temperature. The levels of each protein were normalized to β-actin. The protein bands were exposed with the alpha view SA system (protein simple, California, United States) and analyzed by image software.

Quantitative real-time polymerase chain reaction (RT-qPCR) assay
Total RNA from different groups of HUVECs was extracted using Trizol reagent (Invitrogen, United States). The concentration of RNA was determined, and samples with purity (A260 / A280) between 1.8-2.1 were selected for the next step. The extracted total RNA was reverse transcribed into cDNA using the EasyScript® All-in-One First-Strand cDNA synthesis kit (Transgen Biotech, China). The expression of the target gene was relatively quanti ed by SYBR RT-PCR in Lightcycle 480 system (Roche, United States).

Statistical analysis
The experimental data are expressed by mean ± standard deviation (SD) and statistically analyzed by SPSS 23.0 software. The signi cance of differences between groups was compared by one-way ANOVA.
In all results, P < 0.05 or 0.01 was considered statistically signi cant.

GE improved the symptoms of joint in ammation in AA rats
SD rats were injected with FCA for 7 d to induce an AA model. AA rats showed swelling and joint stiffness on the left hind limb, while arthritic nodules appeared at the ankle ( Fig. 2A). From the 7 d to the 17 d after modeling, the arthritis index and paw swelling gradually increased, while mobility score decreased, indicating that the modeling was successful. The arthritis index and paw swelling were signi cantly decreased and mobility score was increased in AA rats after treatment with GE and TG, indicating that GE effectively alleviated joint in ammation in AA rats (Fig. 2B-D). Synovial pathomorphology showed that the synovium of normal rats was composed of 1 or 2 layers of synovial cells overlaid on the surface of connective and adipose tissue, and there was no angiogenesis and aggregation of in ammatory cells. On the contrary, synovial tissues of AA rats showed 3 to 8 layers of hyperplastic synoviocytes, with a large number of in ammatory cells in ltrated the synovium accompanied by subintimal angiogenesis. After treatment with GE or TG, the synovium was signi cantly thinner and orderly arranged than that in AA group, and the in ammatory cell in ltration and vascular proliferation were relatively reduced (Fig. 2E). It further illustrated that GE alleviated the joint in ammation of AA rats, including the histopathological changes.

GE inhibited angiogenesis in vivo in AA rats
The resulting MVD by the positive expression ratio of CD31 was used to represent the level of angiogenesis in vivo. As shown in Fig. 3A, the number of CD31 positive microvessels in AA group was increased compared with the control group, but decreased signi cantly after GE and TG treatment. MVD quanti cation was consistent with the above results (Fig. 3B). In addition, immunohistochemical quantitative results showed that GE signi cantly inhibited the expression of VEGF in synovial tissue. It was also found that PTEN expression was decreased in AA group and GE up-regulated PTEN expression in synovial tissue (Fig. 3C). The level of classical angiogenic factor VEGF shows the degree of angiogenesis. As shown in Fig. 3D, the serum VEGF level in the AA group was increased signi cantly compared with the control group, and the VEGF level in the GE group was decreased in varying degrees compared with the AA group, which was consistent with synovial tissue. The changes of VEGF level further revealed the anti-angiogenesis effect of GE in AA rats. These results suggest that GE downregulated angiogenic factors and play an anti-angiogensis role in AA rats.

Establishment of in ammatory HUVECs model in vitro
In order to con rm the role of PTEN in RA in vitro, an in ammation cell model of HUVECs and a co-culture cell model of FLS-HUVECs were also established in addition to an AA rat model. First, the appropriate concentration of TNF-α (0, 1.25, 2.5, 5, 10 and 20 ng/mL) was screened to induce in ammation in HUVECs in ammatory model. The results showed that the proliferation rate of HUVECs was the highest at the concentration of 10 ng/mL (Fig. 4A). The cell co-culture model simulates the real environment between cells very well. HUVECs stimulated by different concentrations of FLS-CM (0, 5%, 25%, 50%, 75% and 95%) were established to determine the optimal co-culture conditions. The proliferation rate of HUVECs stimulated by different concentrations of FLS-CM is shown in Fig. 4B, 50% FLS-CM promoted the proliferation of HUVECs signi cantly compared with the control group. Western blot showed that PTEN expression in 50%-95% FLS-CM was decreased compared with the control group (Fig. 4C). The above results support the selection of 50% FLS-CM as the stimulation condition for the co-culture model.
Next, the two models were compared by VEGF secretion levels and PTEN expression. Compared with the control group, the VEGF secretion levels was increased in both two models (Fig. 4D). As shown in Fig. 4E, F, PTEN protein expression and mRNA levels were decreased compared with the control group in TNF-α and FLS-CM model. The P values between TNF-α and FLS-CM groups were greater than 0.05, indicating that there was no signi cant difference in VEGF secretion and PTEN protein and mRNA expression between the two groups. The above results show that these two in vitro models simulated the environment of HUVECs in in ammatory and pro-angiogenic media, and the decrease of PTEN expression was consistent with that in synovium of AA rats. For the consideration of experimental operability, the TNF-α in ammation model was selected for subsequent experiments.

PTEN over-expression inhibited proliferation, migration, and tube formation of HUVECs
Our study showed that the expression of PTEN was decreased in synovium of AA rats. In order to provide more evidence that PTEN is involved in RA angiogenesis, a PTEN lentiviral vector was used to overexpress PTEN in HUVECs to assess the effects of over-expression of PTEN on HUVECs proliferation, migration and tube formation in vitro. Immuno uorescence and RT-qPCR analysis showed that the expression levels of PTEN protein and mRNA was signi cantly up-regulated after lentiviral vector infection, demonstrating successful infection (Fig. 5A, B).
The abnormal proliferation of HUVECs was the initial event of angiogenesis. Cell cycle results showed that the G1 phase decreased and the G2 and S phases increased after TNF-α stimulation; Lv-PTEN + upregulated the cells number of G1 phase cells and down-regulated the cells number of G2 and S phase. It was proved that HUVECs proliferation was promoted under in ammatory conditions, and PTEN overexpression inhibited proliferation (Fig. 5C). Similarly, Edu staining analysis showed that Lv-PTEN + inhibited the proliferation of HUVECs, which was determined by the reduction of green uorescence representing proliferating cells (Fig. 5D). The migration of HUVECs as the basic behavior of angiogenesis was also examined by wound-healing and transwell assay. The wound healing rate is directly proportional to the ability of cell migration. As shown in Fig. 5E, F, TNF-α stimulation signi cantly increased the healing rate and the number of migrated cells, while decreased signi cantly in the Lv-PTEN + group. As a classic model of angiogenesis in vitro, HUVECs tube forming experiment is also analyzed.
As shown in Fig. 5G, HUVECs in the TNF-α group was interconnected to form a complete lumen structure, whereas the lumen structure was damaged, the cell edge shrank, and the number of branches decreased signi cantly after treatment with Lv-PTEN + . In conclusion, the over-expression of PTEN maybe is an important factor for inhibition of HUVECs angiogenesis.

GE up-regulated PTEN to inhibit angiogenesis of HUVECs in vitro
To further understand the therapeutic mechanism of GE and the potential involvement of PTEN, the in vitro model was used to explore the therapeutic effect of GE on HUVECs angiogenesis. Based on cell cycle analysis, the number of cells in G1 phase decreased and G2 and S phases increased after PTEN inhibitor Bpv treatment. Different concentrations of GE down-regulated the number of G2 and S phase cells, indicating that GE had an obvious inhibitory effect on the proliferation of HUVECs (Fig. 6A). Edu staining analysis also showed that GE inhibited the proliferation of HUVECs (Fig. 6B). As shown in Fig. 6C, D, Bpv induced the migration of HUVECs by inhibiting PTEN. The healing rate and the number of migrating cells were decreased after GE treatment, indicating that GE inhibited HUVECs migration. In addition, the results of tube forming showed that GE also signi cantly inhibited the tube formation of HUVECs, which was manifested by the reduction of the number of tube and branching length (Fig. 6E).
Our results indicate that the GE treatment attenuated the proliferation, migration and tube formation of HUVECs, which may be related to the up-regulation of PTEN expression.

GE regulated PTEN-PI3K-Akt signaling in HUVECs
Our study shows that over-expressed PTEN plays an important role in inhibiting angiogenesis. However, the mechanism of PTEN in RA angiogenesis remains to be determined. PTEN is closely related to the PI3K-Akt signaling pathway in the pathophysiological process. Expression levels of PTEN-PI3K-Akt signal axis proteins, including PTEN, PI3K, p-PI3K, Akt, p-Akt, were determined to evaluate the regulatory effect of GE on the signal axis. As shown in Fig. 7A-E, PTEN protein levels was decreased, while p-PI3K and p-Akt protein levels were increased in TNF-α group compared with the control group. GE up-regulated PTEN protein level compared with TNF-α group, which was opposite to that of PTEN inhibitor Bpv group. In addition, exerted signi cant inhibitory effects on the expression of the proteins involved in the PI3K-AKT signal axis, as in LY294002 group. WB results showed that GE up-regulated PTEN expression and inhibited the activtion of PI3K-Akt signal.
To further con rm the associations among GE, PTEN, and PI3K-Akt signal axis, mRNA levels in HUVECs from each group were evaluated by RT-qPCR. As shown in Fig. 7F, PTEN mRNA level was signi cantly decreased in TNF-α group, whereas PI3K and Akt mRNA levels were increased compared with the control group. GE up-regulated PTEN and induced signi cant suppression of PI3K-Akt signal axis-related mRNAs, which was consistent with the WB results. Our results suggested that GE inhibited the activation of PI3K-Akt signal axis by indirectly up-regulating PTEN.

Discussion
RA is a chronic disease characterized by persistent in ammation, and the main pathological manifestations are synovial in ammatory hyperplasia, pannus formation, and joint destruction. Angiogenesis, as the basic event of pannus formation, has attracted extensive attention whether contributing to synoviocytes tumor-like proliferation, in ammatory cell in ltration, or articular cartilage destruction [17,18] . The nutrients, oxygen and a variety of signal factors produced by the neovasculature fuel the hypoxia and closed synovial microenvironment undoubtedly, which activates the interaction and signaling crosstalk between cells, prompting abnormal proliferation and aggravation of in ammatory phenotypes in various types of cells [19,20] .
Appropriate animal models are the basis of experimental research and new drug development. The AA rat model shares many similar features with RA in clinical manifestations, pathology, and immunological changes, making it a more ideal animal model to study the pathological mechanisms and evaluate drug e cacy of RA [21.22] . In addition, the AA rat model was successfully established, the characteristics of AA model were described in detail from the whole to the joint, and the evaluation indexes of the model were determined, including systemic evaluation, arthritis index, paw swelling level, movement score. In this study, the cell co-culture model and the TNF-α stimulated cell model were compared to select the suitable model for in vitro study. The cell co-culture model is widely used in vitro because of its similarity with the real internal environment [23] . FLS, as the main effector cell of RA joint synovium, contributes to and participates in synovium in ammation. VEC is not only an important part of vascular structure formation and angiogenesis, but also the material basis for the survival of FLS, which is closely related to the occurrence and development of RA [24,25] . FLS extracted from synovial tissue were used to establish a coculture model with HUVECs to simulate synovial environment. TNF-α acts on VEC as a classical proin ammatory factor, which not only increase the secretion of in ammatory cytokines and angiogenic mediators (such as IL-1 and VEGF), but also damage the function of endothelial cells or lead to vascular dysfunction, resulting in vascular injury and local ischemia and hypoxia [26,27] . Therefore, TNF-α is a good stimulator acting on HUVECs to simulate in ammation and vascular injury. Our experiments have demonstrated that FLS and VEC in synovial tissue of AA rats are abnormally activated, including abnormal proliferation and angiogenesis. The comparison of in vitro models showed that there was no signi cant difference between them in the detection of experimental related indexes (including cell proliferation, secretion of angiogenic factors and PTEN expression). Finally, TNF-α was selected for the in vitro experiment, considering that FLS can not be guaranteed to be in the in ammatory state and the operability of the experiment in the in vitro co-culture experiment.
Angiogenesis, the formation of new capillaries from pre-existing vessels, is associated with in ammation and in ammatory diseases and is also a potential target for the treatment of RA. The pathogenicity of angiogenesis in RA, resulting in enhanced endothelial surface and persistent in ammatory cell in ltration, has been described to play an important role in the pathogenesis of this disease [7,28] . The angiogenesis process of RA is a programmed event. Angiogenic mediators produced by various types of cells (including FLS, macrophages, etc) in the synovial environment activate VEC. The endothelial basement membrane and perivascular extracellular matrix are degraded by proteolytic enzymes. After that, VEC proliferate abundantly and migrate into the interstitial tissue due to the differentiation into invasive tip cells, called sprouting. The formation of capillary loops by sprouting VEC (tip cells) to synthesize new basement membranes helps to maintain the structural and functional integrity of nascent blood vessels and ultimately the formation of new capillaries [29][30][31] . Therefore, cell proliferation, migration and tubulogenesis in vitro were comprehensively studied to evaluate the vascularization ability of the cells in this study.
As more and more angiogenic mediators that activate VEC are identi ed, such as growth factors, proteolytic enzymes, integrins, and adhesion molecules, researchers mostly focus on this [32,33] . Previous studies by our group also focused on VEGF and found that VEGF binds to its surface receptors to activate downstream pathways of the cell and plays a role in mediating angiogenesis [15,16] . However, the discovery of novel approaches to target multiple cascades or to select upstream cascades of many proangiogenic factors may provide promising strategies for RA treatment, because the functions of angiogenic mediators are mostly cross regulated. As early as the 1990s, researchers found the important role of genetic factors regulating angiogenic mediators in tumor angiogenesis. For example, Bouck rst reported that the inactivation of the p53 gene by mutation or deletion downregulateed the expression of thrombospondin, which induces angiogenesis [34][35][36] . Thus, one of the main consequences of tumor suppressor gene inactivation is to promote tumor angiogenesis, which makes people re-recognize the importance of tumor suppressor genes (and oncogenes) in the study of angiogenesis. In this study, PTEN was proved to have a negative regulatory effect on HUVEC angiogenesis in an in ammatory environment.
PI3K is an intracellular phosphatidylinositol kinase that phosphorylates PIP2 to produce PIP3 which in turn activates Akt [37] . PTEN reverses this process by removing phosphate groups. PI3K-Akt signaling plays a role in a variety of cellular functions including proliferation, survival, migration, invasion and cell metabolism [38,39] . Upstream components of the PI3K-Akt signaling pathway such as PTEN and Ras are commonly mutated in many human cancers, especially playing an important role in regulating normal vascularization and pathological angiogenesis. Direct evidence for the involvement of PI3K and Akt in regulating angiogenesis in vivo was observed by forced expression of PI3K and Akt by the RCAs retroviral vector system [11] . Over-expression of PI3K or Akt induces angiogenesis, whereas over-expression of PTEN inhibits angiogenesis in chicken embryos, indicating that PI3K-Akt signal is required for normal embryonic angiogenesis. In addition, PI3K is activated by growth factors and angiogenesis inducers, such as vascular endothelial growth factor (VEGF) and angiopoietins, activates Akt or other targets, and induces HIF-1 and VEGF expression to regulate angiogenesis [40,41] . PTEN de cient endothelial cells showing increased excessive angiogenesis have been demonstrated in the vasculogenesis of tumors [42,43] . Akt regulates multiple downstream targets, including nitric oxide synthase (NOS), nuclear factor kappaB (NF-кB), glycogen synthase kinase 3(GSK-3), and Jun N-terminal kinase (JNK). However, the speci c target of Akt induced angiogenesis remains to be determined. Based on the comprehensive research of RA disease and PTEN signal, we proposed for the rst time that PTEN-PI3K-Akt signal is involved in the regulation mechanisms of RA angiogenesis. In the HUVECs model in vitro, we found that the expression levels of PTEN protein and mRNA were signi cantly decreased, and the expression levels of PI3K-Akt signal axis related protein and mRNA were signi cantly increased after TNF-α stimulation. In order to further verify the involvement of signal axis, we established a PTEN over-expression lentiviral vector. The experimental results are consistent with our expectation that the over-expression of PTEN inhibited the proliferation, migration and tubulogenesis of HUVECs in vitro. These ndings support targeting the PTEN-PI3K-Akt signal as an effective strategy for the treatment of RA angiogenesis.
Inhibitors targeting the PI3K-Akt pathway have been developed, and these drugs can reduce VEGF secretion and angiogenesis as predicted. The traditional PI3K-Akt inhibitors LY294002 and wortmannin showed antiangiogenic activity, but these inhibitors are not suitable for human because of their toxicity and crossover inhibition of other lipid and protein kinases [44] . The treatment of traditional Chinese medicine has attracted more and more attention as an effective treatment with the emergence of new technological approaches. GE, as a iridoid glycoside obtained from Gardenia jasminoides Eills by modern extraction and separation technology of traditional Chinese medicine, is often used in the treatment of RA [14] . In recent years, our team has performed a lot of work around the study of the therapeutic effect and mechanism of GE on RA. Previous studies have shown that GE can restore the balance of pro/antiangiogenic factors and inhibit VEGF-SphK1-S1P signal axis to exert anti-angiogenic effects [15] . This study supports the role of GE in up-regulating PTEN expression to inhibit synovial microvascular angiogenesis. Especially in the in vitro model, the cellular biological function of HUVECs was signi cantly inhibited. Furthermore, we found that the anti-angiogenic effect of GE may be regulated by PTEN-PI3K-Akt signal. The results obtained by lentivirus vector and speci c inhibitors are consistent with our expectation that regulating PTEN-PI3K-Akt signal can signi cantly reduce the angiogenesis of HUVECs, which is consistent with the therapeutic effect of GE (Fig. 8).

Conclusion
Our current results support that the low-expression of PTEN in RA synovium is associated with excessive angiogenesis. GE effectively inhibited secondary in ammation and angiogenesis in AA rats and upregulated the expression of PTEN. In addition, GE treatment attenuated the proliferation, migration and tubulogenesis of HUVECs, which was concerned with up-regulating PTEN and inhibiting the activation of PI3K-Akt pathway. The present ndings offering a new option for therapeutic intervention of RA. Further studies are needed to determine the speci c mechanism of GE up-regulating PTEN expression, and this part of the experiment is in progress.   in ammatory cells (I) respectively. Data are represented as mean ± SD (n = 8). # P < 0.05, ## P < 0.01 versus control group; *P < 0.05, **P < 0.01 versus AA group.   Data are represented as mean ± SD. # P < 0.05, ## P < 0.01 versus control group. Effects of GE on the tube formation of HUVECs (n = 6). Data are represented as mean ± SD. # P < 0.05, ## P < 0.01 versus control group; * P < 0.05, ** P < 0.01 versus TNF-α group.GE up-regulated PTEN to inhibit angiogenesis of HUVECs in vitro. A. Effects of GE on the cell cycle of HUVECs (n = 3). B. Effects of GE on the Edu expression of HUVECs (n = 6). C. Effects of GE on the horizontal migration of HUVECs (n = 6). D.