The Regulation of Macrophage Polarization by Hypoxia-PADI4 Coordination in Rheumatoid Arthritis

Background: Hypoxia, a common feature of rheumatoid arthritis (RA), induces the overexpression of peptidyl arginine deiminase 4 (PADI4) in broblast-like synoviocytes (FLSs) and macrophages. However, the roles of PADI4 and its inducer hypoxia in the regulation of macrophage polarization remain unclear. This study aimed to investigate the role of hypoxia-PADI4 for macrophage polarization in RA patients. Methods: synovial tissue(ST) and synovial uid(SF) were collected from 3 OA patients and 6 RA patients. The distribution of M1 and M2 in ST and cytokines in SF were examined by immunohistochemical analysis and BioPlex immunoassays. THP-1 macrophages and BMDM polarization were determined under normoxic (21% oxygen) or hypoxic (3% oxygen) conditions. The effects of PADI4 on macrophages were determined by transfection of adenovirus vector-coated PADI4 (AdPADI4) and the use of PADI4 inhibitor. To mimic the environment of RA joints, THP-1 macrophages polarization was examined after coculturing with RA-FLSs. Finally, the roles of PADI4 in joint synovial lesions on macrophage polarization were investigated in collagen-induced arthritis(CIA) rats. Results: We found increased macrophage polarization of M1 and M2 in the RA ST, compared with OA ST. The ratio of M1/M2 for RA and OA was 1.633 ± 0.1443 and 2.544 ± 0.4429, respectively. The concentration of M1- and M2-type cytokines was higher in RA than that in OA patients. Hypoxia contributed to the increase of the gene and protein expression of M1 and M2 markers. M1- but not M2-type gene expression showed a positive relationship with PADI4 expression while the level of expression of M2-type genes showed no signicant difference. RA-FLSs could promote the copolarization of M1 and M2, and PADI4 inhibitor reversed M1 activation. The degree of


Background
Rheumatoid arthritis (RA) is a common chronic autoimmune disease characterized by abnormal activation of the immune system, resulting in synovitis and destruction of joint function (1). Synovitis is characterized by an overabundance of RA broblast-like synoviocytes (RA-FLSs), monocytes, macrophages, and other immune cells within the RA synovial tissue (2). Among these cells, it has been well recognized that macrophages play an essential role in the initiation and perpetuation of RA by secreting multiple in ammatory and anti-in ammatory factors (3). Mature macrophages have two polarization states: classical activation (M1) and the alternative activation (M2) phenotype (4). M1 macrophages, induced by Th-1 cytokine interferon-γ (IFN-γ) or lipopolysaccharide (LPS), contribute to the development of in ammation and joint destruction by producing numerous pro-in ammatory factors, such as IL-12 and TNF-α, and speci cally expressing NOS-2 and CCR7 as biomarkers. The Th-2 cytokines IL-4, IL-10, and IL-13, however, cause macrophages to polarize into the M2 phenotype. They can then secrete anti-in ammatory cytokine IL-10, increasing the speci c expression of the mannose receptor (CD206) (5). The high concentration of TNF-α in the RA joint synovium can promote M1 polarization (6), and the broblast factor secreted by RA-FLSs can promote M2 polarization (7). However, the role and molecular mechanism of M1/M2 in RA synovial lesions remain largely unknown.
An early step in the development of RA is the activation of RA-FLSs. This is followed by mass production of in ammatory mediators, including cytokines and chemokines, which mediate the recruitment and interaction with immune cells (8). This process enhances the oxygen consumption caused by the proliferation of cells in RA synovium, which leads to the oxygen tension in RA, indicated by the 2-4% values, which in diseased synovial tissues can be even lower than 1% of O 2 (hypoxia condition); healthy controls show levels around 8% (9). Macrophages, as well as other leukocytes, are sensitive to tissue hypoxia, which can increase in ammatory cytokine production (10), yet how hypoxia affects macrophage polarization is not well understood in vitro.
Peptidyl arginine deiminase IV (PADI4) is involved in the post-translational modi cation of arginine residues (11). A study revealed that a large number of activated macrophages and FLSs in RA synovial tissues are involved in the overexpression of PADI4 (12). The contribution of PADI4 to the apoptosis of RA-FLSs has been revealed (13). Considering the multiple roles of PADI4 in the regulation of gene expression and immunological functions, PADI4 may be a potential target for the therapy of autoimmune diseases. However, the roles of PADI4 and its inducer hypoxia in the regulation of macrophage polarization remain unclear.
This study aimed to investigate the polarization of macrophages in RA synovial tissue and elucidate the role of hypoxia-PADI4 in macrophage polarization. A collagen-induced arthritis (CIA) mouse model (13) was used to explore the PADI4-mediated development of in ammation.

Patients and controls
Clinical samples were the resident synovial tissue collected from patients undergoing knee arthroscopic or routine examination at Shanghai East Hospital, including six patients with RA and three patients with osteoarthritis (OA). All of the patients provided consent and met the diagnostic criteria of the American College of Rheumatology (ACR) criteria for RA (14) and OA (15). The Ethics Committee approved the study protocol of Shanghai East Hospital.

Animals
Twelve SD rats were administered intradermally at the base of the tail with a dose of bovine type II collagen (100 μg) emulsi ed in complete Freund's adjuvant on day 0. Then a booster injection was provided on day 21 with bovine type II collagen (100 μg) emulsi ed with incomplete Freund's adjuvant.
The rats were then randomly divided into two groups: the CIA model group and the PADI4 inhibition group. Normal non-immunized rats were selected as the normal control group. The rats were sacri ced on the nal day; joints were collected from all of the groups.
Immunohistochemical and immuno uorescence analyses Synovial tissues from humans and joints from rats were xed in 10% neutral buffered formalin and then embedded in para n. Para n sections were depara nized and rehydrated. After blocking with 3%H 2 O 2 , the sections were incubated with a goat polyclonal antibody against human CCR7 (1:500, Abcam, Cambridge, MA, US) and rabbit polyclonal antibody against human mannose receptor (CD206, 1:500, Abcam, Cambridge, MA, US) at 4°C overnight. Next, the sections were incubated with secondary antibody for 30 min at room temperature. Immunoreactive signals were visualized using DAB (diaminobenzidine
Primary RA-FLSs were isolated from RA synovial tissues. After isolation from the tissue, RA-FLSs were grown further over four to six passages for our experiments. Then these cells were cocultured with THP-1 macrophages through a Transwell coculture system.

Virus vector infection
Adenovirus vectors expressing PADI4 (Genechem, China) were used in accordance with the manufacturer's instructions to transfect genes into M0. RT-PCR and Western blot were used to assess PADI4 expression.

Cocultures of RA-FLSs and M0
The coculture systems were established through a 24-well plate with an 8.0-µm Pore Polycarbonate Membrane Insert (Corning, NY, US). To study the impact of the coculture system on macrophages, macrophages cocultured with RA-FLSs were treated as coculture groups, and the macrophages cocultured macrophages as control groups. M0 were seeded in 5 × 10 4 cells/well, and 1 × 10 4 cells/well RA-FLSs were seeded in the transwell inserts in another separated well. After 6 h of incubation, cells had become rmly attached to the wall. Then, the transwell inserts with RA-FLSs were moved to the wells containing the M0 macrophage and after that cocultured for 48-72 h. The coculture system was subjected to normoxia (21% O 2 ) and used as controls, and another set was placed under hypoxia conditions (3% O 2 ) in the incubator.

Statistical analysis
Statistical analysis was performed using Prism GraphPad software (GraphPad Software, Inc., La Jolla, CA, US). Data were collected from each experiment and compared between groups using analysis of variance (ANOVA) or t-tests (two-tailed distribution) as appropriate. A P-value of ≤ 0.05 was considered to be signi cant.

M1 and M2 are both abundant in RA synovial tissues
To determine the distribution of M1 and M2 macrophages in synovial tissues from RA and OA, we performed immunostaining and double immuno uorescent staining. We found CCR7 + (M1 marker) and CD206 + (M2 marker) cells in the synovial tissues of both RA and OA (Fig. 1A). In RA, the mean ratio of M1/M2 was 1.633 ± 0.1443, and in OA was 2.544 ± 0.4429, i.e., the predominant form were the M1 macrophages (Fig. 1B) (P < 0.05). The results of double immuno uorescent staining showed the distribution of M1 and M2 in the RA group to be both more abundant than in the OA group, and the average uorescence intensity of M2 was higher in RA than in OA (Fig.1 C&D) (P < 0.05).
BioPlex array was used to assess the expression of cytokines in synovial uid. The Bioplex suspension array system allowed the simultaneous measurement of 10 human cytokines, including interleukins, interferons, and tumor necrosis factors. The results showed that the levels of M1 and M2 cytokines, IL-1, IL-4, IL-6, IL-10, IL-12, IL-13, IL-15, IL-17, IFN-γ, VEGF, and EGF to be signi cantly higher in RA than in OA (Fig. 1E) (P < 0.05), suggesting that the degree of in ammatory cell in ltration (both M1 and M2) in RA was more pronounced than in OA.
The expression of M1-and M2-related genes increases under hypoxia As shown in Fig. 2A, qPCR revealed that gene expression of M1 markers (Ccr7, Ccl2, and Il-1β) and M2 markers (Tgfβ, Il-10, and Ccl18) in PMA-treated THP-1 macrophages were signi cantly increased under 3% O 2 conditions. We then polarized M1 and M2 with IFN-γ, LPS, and IL-4. The results showed an elevation of hypoxia-induced M1 gene expression in M1, and M2 caused the same phenomenon, but the range of change was not as pronounced as with M1. Immuno uorescence assay was used to analyze the differential expression of M1 (CCR7) and M2 (CD206) in the 3% O 2 groups and 21% O 2 control, and the results were consistent with mRNA expression (Fig. 2B). Fig.2C showed that the mRNA levels of Ccl2 and TNF-α increased under 3% O 2 in BM M0 and M1, the change of M2 marker (Arg-1) was consistent with THP1-derived macrophages. These observations suggest that hypoxia can increase the level of expression of genes related to M1 and M2 genes.

Increased PADI4 expression induced by hypoxia affects the polarization of macrophages
To investigate the effect of hypoxia on PADI4 expression, THP-1 macrophages were cultured under conditions of different oxygen concentrations (3% O 2 and 21% O 2 ). After culture for 36-48 h, the gene expression of PADI4 in macrophages was determined using qPCR and Western blotting. As shown in Fig.  3A, mRNA, and protein levels of PADI4 were upregulated in cells cultured under 3% O 2 conditions relative to those cultured with 21% O 2 . We then evaluated the effect of PADI4 on the polarization of macrophages. Cells were incubated in a hypoxic environment in the presence of Cl-amidine, the PADI4 inhibitor, to block PADI4 (Fig. 3A). As shown in Fig. 3B, the mRNA levels of M1 markers decreased, while the mRNA levels of most M2 markers did not change signi cantly in THP1-derived macrophages. In contrast, the expression of M2 markers (IL-10, Arg-1) increased in BMDM. These results may indicate that PADI4 affects the polarization of macrophages, especially on M1 under hypoxic conditions.

PADI4-induced macrophages convert to M1
To determine whether PADI4 could affect the polarization process of macrophages, we used adenovirus vectors to determine PADI4 overexpression macrophages in normoxia. THP-1-derived macrophages were infected with AdPADI4 and AdGFP as controls. The infection effect was assessed using qPCR and Western blot (Fig. 4A). As shown in Fig. 4B, signi cant increases were found in the mRNA expression of M1 genes in the Padi4 overexpression group compared with the control group under normoxic conditions (P < 0.05). There were no readily visible differences in M2 genes between these two groups. BMDM showed an increase of M1 marker expression and a decrease in M2 markers expression. This phenomenon also happened in M1, which was further polarized into using IFN-γ and LPS (Fig. 4D). The results indicate that PADI4 may be associated with the process by which macrophages convert to the M1 phenotype, but it was not found to have any effect on M2 polarization. The increases in the expression of M2-related genes may have been due to hypoxia.

RA-FLSs-induced PADI4 overexpression affects M1 gene expression
The effects of coculturing with RA-FLSs on macrophage polarization are shown in Figure 6. For macrophages, the quantitative RT-PCR results showed that the mRNA levels of M1 and M2 genes were higher in the coculture groups than in the control groups, which had signi cant differences (Fig. 5A) (P < 0.05). The addition of the Cl-amidine into the coculture system resulted in a decrease in the expression of M1 in ammatory genes in macrophages and an increase in M2 in ammatory genes (Fig. 5B) (P < 0.05). These results suggested that reducing Padi4 down-regulated M1 activation and hypoxia and enhanced the M2 polarization in this coculture system.

Macrophages in ltration in the synovial tissue is decreased by Padi4 downregulation in CIA rats
After the booster immunization, we evaluated the arthritic scores every 7 days for 63 days. The symptoms associated with arthritis gradually appeared in all of the CIA rats on the 7th day and continued to develop severe arthritis until they reached a plateau between days 28 and 35. Si-Padi4-MSN was injected on the 35th day. The control group treated with MSN showed severe swelling, erythema, and joint rigidity in the paw, while the administration of si-Padi4-MSN signi cantly lowered the arthritic scores. Immunohistochemical analysis of the synovial tissue revealed that Padi4 downregulation rats exhibited reduced macrophages(CD68 + ) in ltration compared with the control group mainly composed by M1(CCR7 + ). The general photos and X-ray pictures and similar results are shown in Fig.6. Collectively, our results suggested that PADI4 exhibited a certain effect on the CIA.

Discussion
Macrophages, the main resource of TNF-α and other pathogenic in ammatory factors, play a crucial role in the in ammatory response. Previous work has shown that the number of macrophages correlated with the degree of synovial hyperplasia and is positively correlated with the patient's DAS28 score and the degree of joint bone erosion (17). Several studies have shown that M1 dominates the in ammatory environment in the RA joint, in which synovial uid has an M1/M2 ratio of 32.76 ± 11.02 (18). Using patient synovia tissue, we con rmed that both M1 and M2 were facilitated in the RA joint synovia tissue, with an M1/M2 ratio of 2.544 ± 0.4429. The concentration of M1 and M2 relative cytokines were more abundant (TNF-α, IL-6, IL-1, IL-4, IL-8, IL-10, IL-12, IL-15, IL-17, IFN-γ, GM-CSF, and VEGF) compared to OA. A recent study reported that exposure of M2 macrophages to the particular combination of c-IgG and toll-like receptor (TLR) ligands, as occurs in the synovia in most patients with RA, strongly promotes in ammation (19). Previous immunohistochemistry data showed that more than 80% of CD163 + macrophages (M2) were responsible for secreting TNF-α in synovia of patients with RA (20). Our observations provided experimental evidence that the polarization of M1 and M2 were indeed enhanced in RA synovia tissue, resulting the development of in ammation.
In previous work, we found that PADI4 is overexpressed in RA-FLSs and promotes the proliferation of RA-FLSs under hypoxic conditions (13). It has been reported PADI4 expression in CD68 macrophages, and it showed a positive correlation with the expression of several genes associated with macrophage activation and function (21). However, the effect of PADI4 on macrophage polarization requires further investigation. In this study, we con rmed a marked overexpression of PADI4 expression in macrophages under hypoxic conditions. Here, we substantiate the role of PADI4 by showing that the PADI4 knockdown indeed weakened M1 marker expression and its overexpression-enhanced M1 polarization, whereas the M2 marker expression was unaffected (Fig. 4). Based on these ndings, we suggest that hypoxia enhances M1 and M2 polarization, and hypoxia-induced PADI4 mainly regulates M1 activation.
Various proin ammatory factors secreted from FLSs can in uence the interaction of FLS and other immune cells, especially macrophages, including activating M1 polarization and anti-in ammatory effects. Coculture models of RA-FLSs and THP-1 macrophages in vitro resulted in the activation of M1 and M2, and Padi4 expression increased in macrophages. In the presence of the PADI4 inhibitor Clamidine, M1-related genes were inhibited, especially Tnf-α, and the expression of M2-related genes increased (Fig. 5). These results suggested that targeting PADI4 helped to suppress the pro-in ammatory process. Further investigation is needed to determine whether PADI4 contributes to pro-in ammation in the mouse model.
Si-RNA-MSN, a nanoparticle carrier we have used in previous studies that is slowly released and can easily reach an effective concentration in the body, was injected into CIA rats to knockdown Padi4. Our study demonstrated that joint swelling and macrophages in ltration were improved after PADI4 inhibition, especially the number of M1. M2 marker did not perform well due to the difference in the phenotype of human and mice, however, the signi cant decrease in the number of M1 can also prove our conclusion. Citrullination, catalyzed by PADI4, is implicated in the host cell apoptosis and immune cell clearance postinfection (22). It has been reported that citrullination may dampen host defense and promote pathogen survival (23). Although PADI4 is well-known to be involved in RA, especially in cell apoptosis, our research gives evidence that targeting PADI4 is promising to weaken the active immune cell in ltration and alleviate chronic in ammation.

Conclusion
All of the data in our paper demonstrated that The PADI4 inhibitor could alleviate the symptoms by reducing polarization of M1 then affecting the M1/M2 ratio and their in ltration degree. The underlying mechanism may be the regulation of the hypoxia-PADI4 axis and PADI4 expression level is a critical factor for treating RA.   Expression of PAdi4 and polarization of macrophages after the introduction of PADI4 inhibitor in hypoxia.
(A) The expression of Padi4, Tnf-α, Ccr7, Nos-2, Il-10, Ccl18, and cocultured and control THP-1 macrophages was determined using quantitative RT-PCR. (B) Effects of PADI4 inhibitor on the expression of genes related to in ammation by macrophages. Macrophages in the control group and coculture group were treated with CI-amidine and then incubated at 3% O2 for 24 h. The expression of TNF-α, Ccr7,