Overproduced bone marrow neutrophils in collagen‐induced arthritis are primed for NETosis: An ignored pathological cell involving inflammatory arthritis

Abstract Objectives Bone marrow edema is a universal manifestation of rheumatoid arthritis (RA), and its pathological essence is a bone marrow lesion (BML) formed by various bone marrow (BM) immune cells. Neutrophils play an important role in inflammatory arthritis, but the role and mechanism of neutrophils in BML are not clear. Materials and methods Granulocyte colony‐stimulating factor (G‐CSF) −/− mice and wild type (WT) C57BL/6 mice were immunized for collagen‐induced arthritis (CIA). Histological scores of arthritis were evaluated. Immunohistochemistry staining with anti‐Ly6G was conducted. Neutrophil extracellular traps (NETs) in joint sections were determined by immunofluorescence staining. BM neutrophils were isolated for flow cytometry and NETosis induction in vitro. Results Histological study showed significant neutrophil infiltrations in BML of CIA mice. Inhibition of BM neutrophil production by G‐CSF knock out can obstruct the induction of BML and CIA. In addition to abundant infiltrated NETs intra‐articular, remarkable NETosis primed BM neutrophils were infiltrated in BML of CIA mice, which was positively related to bone erosion. Neutrophils derived from G‐CSF−/− mice have diminished ability of NETs formation in vitro, while G‐CSF induction can enhance its capacity of NETs formation. Conclusions We propose for the first time that the overproduced BM neutrophils in CIA mice are primed for NETosis in a G‐CSF dependent manner, and these pathogenic cells may have an important role in inflammatory arthritis. Blocking this pathological process could be a potential strategy for the treatment of RA.


| INTRODUC TI ON
Rheumatoid arthritis (RA) is a complex autoimmune disease involving immune cells, synovial cells, osteoclasts, and other cells. T, B lymphocytes, dendritic cells, macrophages, and other immune cells play an important role in the occurrence and development of RA, but the exact pathogenesis of RA is still unclear. 1 The majority of researchers have focused on lesions in the synovium, cartilage, and bone cortex, but in fact, immune cells in the subchondral bone marrow (BM) also play a role in RA. 2 Bone marrow edema (BME) is a non-specific imaging manifestation of the high signal lesions in the BM region on the fat-suppressed image of MRI. 3 BME is common in inflammatory joints of RA and can appear in the early stage of disease. 4 Studies have shown that BME score is closely related to cartilage and bone erosion, and is a sensitive indicator that reflects the risk and progress of cartilage and bone erosion in RA. 5 Besides, BME was correlated with anti-citrullinated protein antibodies (ACPA) levels in RA patients. 6 Therefore, controlling BME may have potential benefits for the treatment and prognosis of RA.
From a pathological point of view, the essence of BME is that normal adipose tissue is replaced by inflammatory cells mainly composed of T, B lymphocytes. 7,8 This pathological manifestation is called bone marrow lesions (BMLs). 9 McQueen et al proposed the pathogenesis hypothesis of BML in RA, suggesting that pathological cells migrate from subchondral bone to synovium to play a role. 10 However, there are few studies in this field, and the exact role and mechanism of BML in RA is still unclear.
Neutrophils are differentiated from precursor cells in the BM and released into the peripheral blood. In the inflammatory state, BM neutrophil production increases, which can migrate and infiltrate into the joint cavity of RA patients under chemotaxis, and further participate in arthritis by phagocytosis, degranulation, and production of reactive oxygen species (ROS). 11 In recent years, the discovery of neutrophil extracellular traps (NETs) has further elucidated the role of neutrophils in the pathogenesis of RA. 12 Whether neutrophils are involved in BML of RA has not been reported yet. The purpose of this study was to determine the role of neutrophils in BML of inflammatory arthritis and its possible pathways and mechanisms by a murine model of RA.

| Collagen-induced arthritis
Collagen-induced arthritis (CIA) which is the major murine model of RA was induced as described previously. 13 Briefly, mice were immunized intradermally in the proximal tail with 100 μg of chicken type II collagen (2 mg/mL, 20012; Chondrex, Inc) emulsified in 250 μg of Freund's complete adjuvant (5 mg/mL, 7023; Chondrex, Inc). A booster injection was given 21days after the first immunization. The incidence and severity of arthritis were assessed from the second immunization up to the termination of the experiment. A clinical arthritis score from 0 to 4 was given for each limb with a maximal score of 16 for each mouse (0: no erythema and swelling; 1: slight edema and erythema limited to the tarsals or ankle; 2: slight edema and erythema spreading from the ankle to the tarsal bone; 3: moderate edema and erythema from the ankle to the metatarsal joints; 4: severe edema and erythema from the ankle to the entire paw). 14 The mice were sacrificed at day 51 and the ankle joints were collected.

| Histological examination
The ankle joins were fixed in 4% paraformaldehyde overnight, decalcified in ethylenediaminetetraacetic acid (EDTA) for 4 weeks, processed and embedded in paraffin. Joint sections were stained with hematoxylin and eosin. Histological score of arthritis was evaluated by two investigators separately who were blinded to grouping information, according to a previously described scoring system on the extent of synovitis, pannus formation, bone and/or cartilageerosion, as following criteria: 0: no signs of inflammation; 1: mild inflammation with hyperplasia of the synovial lining layer, minimal without cartilageerosion; 2 to 4: increasing degrees of inflammatory cell infiltrate, or cartilage and bone erosion. 15

| Immunohistochemistry
For immunohistochemistry, paraffin-embedded slices on slides were deparaffinized with xylene and rehydrated. Antigens were unmasked by boiling in EDTA antigen retrieval buffer pH 9.0.
Endogenous peroxidases were blocked with 3% hydrogen peroxide solution and then slides were blocked with 10% normal goat serum.

| Immunofluorescence staining for joint sections
After sectioning to 4 µm, the joint section was blocked with 10%

| Isolation of mouse BM neutrophils
Bone marrow cells were flushed from the tibias and femurs of WT and G-CSF−/− mice into Roswell Park Memorial Institute (RPMI) media-1640.
Neutrophils were purified from BM as described previously. 16 10 7 -10 8 neutrophils were obtained from each mouse and the cell survival rate was more than 95% counted by trypan blue stain. The purity of isolated neutrophils was more than 90% assessed by Wright-Giemsa stain.

| Flow Cytometry
Single-cell suspensions of BM cells were re-suspended in RPMI-1640 for staining. Cell surface staining was performed to evaluate frequencies of mature BM neutrophils using antibodies against CD45, CD11b and Ly6G directly labeled with APC, PE-Cy7 or FITC (BD pharmingen).
Appropriate isotypes were used as antibody controls. Stained cells were analyzed using a FACSCalibur flow cytometer (BD Biosciences).

| NETosis induction in vitro
Neutrophils obtained from WT and G-CSF−/− mice were seeded at a density of 10 5 cells/well in a 24-well cell culture plate with polylysine-coated coverslips for 30min. Then, the cells were exposed to PBS ormice serum or phorbol 12-myristate 13-acetate (25 nmol/L, PMA; Sigma)with or without G-CSF (100 ng/mL, Recombinant Human G-CSF; Proteintech) for 3 hours at 37°C with 5% CO 2 .
For G-CSF induction, BM neutrophils were primed with G-CSF at 100 ng/mL for 1 hour before PMA stimulation.

| Statistical analysis
All statistical analyses were performed using GraphPad Prism (version 7.0). Between-group difference was analyzed by Student's t test or Mann-Whitney U test. The correlation between groups was analyzed using the Pearson correlation coefficient. A value of P < .05 was considered statistically significant. F I G U R E 2 Suppression of bone marrow (BM) neutrophil production by granulocyte colony-stimulating factor (G-CSF) knockout can significantly obstruct the induction of collagen-induced arthritis (CIA). A, The cumulative incidence of arthritis and (B) clinical scores (mean ± SEM) in immunized WT and G-CSF−/− mice during days 21-51 of CIA induction. Results shown are from 8 individual mice per group and two independent experiments. C, Representative histopathological sections of ankle joints from immunized WT and G-CSF−/− mice by hematoxylin and eosin staining. D, The semi-quantitative histological score was evaluated and compared, as determined by two-tailed, unequal variance Student's t test (mean ± SEM). Results shown are from 8 individual mice per group and two independent experiments. E, Representative immunohistochemical staining of Ly6G in subchondral BM. Brown staining indicates Ly6G positive cells. F, The percentage and number of Ly6G positive cells per field in subchondral BM from normal control or immunized WT and G-CSF−/− mice were enumerated for at least five different fields of each section by Image-pro plus 6.0, as determined by two-tailed, unpaired Student's t test (mean ± SEM). Results shown are from 8 individual mice per group and two independent experiments | 7 of 14 XU et al.

CIA mice
Previous studies have reported that the main histological change of RA subchondral BML is the infiltration of a large number of lymphocytes, 7,8 but whether there is infiltration of neutrophils has not been reported in the literature. Therefore, we induced CIA model and determined it through pathological methods. HE staining showed that CIA mice had marked BML in which normal adipose tissues were replaced by infiltrated immune cells, while the BM of the normal control mice were mainly adipose tissue ( Figure 1A). Immunohistochemical staining using the murine neutrophil surface marker Ly6G showed that Ly6G positive neutrophils infiltrated into the articular cavity, synovium and articular cartilage of CIA mice and also in subchondral BML significantly, in contrast, no neutrophils were found in the joints and BM of control mice ( Figure 1B,C). These results show that there were significant neutrophil infiltrations in BMLs of CIA mice.
We speculate that the intensive neutrophil infiltration in BML may be due to the excessive production of BM neutrophils under the stimulation of inflammation.

| Inhibition of BM neutrophil production by G-CSF knock out obstruct the induction of CIA
The differentiation and maturation of BM neutrophils are mainly regulated by G-CSF. G-CSF is not only a hematopoietic factorbut also a proinflammatory factor by promoting BM granulopoiesis under inflammatory stimulation. 17  Therefore, excessive production of BM neutrophils is necessary for BML formation and it plays an important role in the pathogenesis of CIA, this process is G-CSF dependent.

| BM neutrophils involved in inflammatory arthritis through NETosis
Previous studies have shown that BME is associated with ACPA antibody, 6 meanwhile,the citrullinated histones generated by neutrophils in the process of NETs formation are the real antigens of ACPA. 18 Therefore, we assume that BM neutrophils are involved in the onset of arthritis through NETosis. The role of NETs in a variety of diseases, including RA, has been confirmed, but current studies are limited to NETosis in peripheral blood or affected organs, and it is unclear whether BM neutrophils are also involved in NETosis.
Therefore, we further investigated the NETosis phenomenon in the joints and subchondral BM of CIA mice. In order to analyze the formation of NETs in CIA, we used anti-citrullinated histone H3 antibody, a NETs-specific marker, to stain the joint tissues by immunofluorescence. There was a large amount of NETs infiltration in the joints of CIA mice, distributed mainly in the articular cavity, articular cartilage, synovial tissue, etc, which was consistent with the location of neutrophil infiltration ( Figure 3A

| BM neutrophils are primed for NETosis
Histone citrullination is a key step in NETosis, but it does not represent the formation of NETs. To further clarify whether BM neutrophils have initiated NETosis or only primed for NETosis, we

| G-CSF affects the ability of BM neutrophils to form NETs in vitro
To further investigate the effect of G-CSF on the formation of NETs in BM neutrophils, we extracted and isolated BM neutrophils from WT and G-CSF−/− mice to induce NETs in vitro. The lower proportion of CD11b + Ly6G + mature neutrophils in G-CSF−/− BM was confirmed by flow cytometry (Figure 6A,B).
Then, we observed the formation of NETs stimulated by PMA with or without G-CSF induction in neutrophils from WT and G-CSF−/− mice. We found that a lower proportion of G-CSF−/− neutrophils released NETs compare to WT neutrophils after induction of PMA ( Figure 7A,B). With G-CSF induction, BM neutrophils from both WT and G-CSF−/− mice showed enhanced NETosis ( Figure 7A,B). These results show that neutrophils derived from BM lacking G-CSF have diminished ability of NETs formation in vitro, and G-CSF induction can enhance its capacity of NETs formation.Therefore, G-CSF has a role in NETosis of BM neutrophils.

| D ISCUSS I ON
Bone marrow edema is an early imaging manifestation of RA. 19 BME score is closely related to the erosion of cartilage and bone in RA and is a sensitive indicator of the risk and progression of bone erosion. 5 From the perspective of pathology, BME in inflammatory joints of RA indicates abnormal aggregation of T or B lymphocytes in this area, replacing normal adipose tissue, but its exact role and mechanism are still unclear. 9 Neutrophils are differentiated and mature in the BM. In the state of inflammation, neutrophils can infiltrate in large numbers in the joint cavity of RA after chemotaxis and participate in joint inflammation. 20 However, whether neutrophils participate in BML of RA has not been reported.
In this study, we used a mouse model of CIA to confirm the presence of significant neutrophil infiltration in subchondral BML of inflammatory joints. Further CIA induction experiments on G-CSF−/− mice showed that G-CSF−/− mice had no BML formation and neutrophil infiltration, and arthritis induction was completely blocked. Therefore, G-CSF-dependent neutrophil overproduction in CIA mice may lead to numerous neutrophil infiltration in BML.
Based on the discovery that the citrullinated histone produced by neutrophils during NETosis is the authentic antigen of ACPA 18,21 and the existed correlation between ACPA and BME, 6,22 we speculated that the BM neutrophils were involved in inflammatory arthritis via NETosis. Further immunofluorescence staining results confirmed our speculation that not only the infiltration of migrate from subchondral bone to synovium to play a role. 10 Our study reveals that the overproduced BM neutrophils in CIA mice are primed for NETosis, blocking the excessive production of BM neutrophils can inhibit the occurrence of inflammatory arthritis through suppressing the process of BML and NETosis. Therefore, we suppose that these pathological cells within BML play a key role F I G U R E 5 Bone marrow (BM) neutrophils are primed for NETosis. A, Representative confocal images of BM neutrophils isolated from collagen-induced arthritis (CIA) mice and normal control mice before and after induction of serum. B, Percentage of BM neutrophils primed for NETosis per field in neutrophils from CIA mice as compared to neutrophils from normal control mice before induction, (C) Percentage of BM neutrophils forming neutrophil extracellular traps (NETs) per field from normal control and CIA mice before induction as compared to that of after induction with their corresponding serum, as determined by two-tailed, unequal variance Student's t test (mean ± SEM). Results shown are representative of at least eight fields in different regions for each condition, and representative of two independent experiments. D, Magnified representative confocal images of BM neutrophils from CIA mice forming NETs in vitro after induction. E, Magnified representative confocal images of NETosis primed neutrophils in BM and NETs formation in cartilage and joint cavity in the development of arthritis, providing new clues for elucidating the role of BME in the pathogenesis of RA.
Neutrophils are terminally differentiated cells and unable to further proliferate. In the absence of stimuli, neutrophils have a very short life span in the circulation (8-20 hours) before undergoing apoptosis. 25 In RA patients, it can quickly migrate and infiltrate into the inflamed joint. 11 The number of neutrophils in the joint cavity is extremely high. 26 The neutrophils in the synovial fluid of RA patients NETosis caused by different stimuli may have different molecular mechanisms and upstream pathways, including NOX-dependent and NOX-independent mechanisms. 39,40 In addition to citrullinated histones, ROS is also an important marker of NETosis. ROS、Ca 2+ and PAD4 are key participants in NETosis and they are also the focus of in-depth study for the exact mechanism of BM neutrophils primed for NETosis in CIA.
In conclusion, overproduced BM neutrophils are primed for NETosis in BM of CIA mice in a G-CSF dependent manner, and these pathogenic cells may have an important role in initiating and promoting the development of inflammatory arthritis, especially in bone erosion (Figure 8). In the future, more in-depth and detailed researches are needed to clarify the role and exact mechanism of these cells within the development of RA and further clarify the regulatory mechanism of these pathological BM neutrophils. The crosstalk between BM neutrophils and inflammatory arthritis is an interesting topic worthy of further study. Targeting this pathological process may be a potential strategy for the treatment of RA.

ACK N OWLED G M ENTS
This work was supported by grants from the Medical and Health Research Program of Zhejiang Province (2018KY067 and F I G U R E 8 Under inflammation, bone marrow (BM) neutrophils are produced in large quantities under the action of granulocyte colonystimulating factor (G-CSF), meanwhile primed for NETosis. Once released into the periphery and subjected to chemotaxis of inflammation, these pathological cells can infiltrate into inflammatory joints and form neutrophil extracellular traps (NETs), promoting joint inflammation and autoantibody production, eventually leading to bone erosion 2020KY116). We would like to thank prof. Jiong Chen, Laboratory of Biochemistry and Molecular Biology, School of Marine Sciences, Ningbo University, Ningbo, China for kindly supporting. We also thank Dr Weiqian Chen, Dr Jiong Yu, Dr Ye Yu and Dr Heng Cao for discussion on experimental methods.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.