Research ArticleARTHRITIS

Complement C5a receptor is the key initiator of neutrophil adhesion igniting immune complex–induced arthritis

See allHide authors and affiliations

Science Immunology  20 Jan 2017:
Vol. 2, Issue 7, eaaj2195
DOI: 10.1126/sciimmunol.aaj2195

Adhering to arthritis

Autoimmune diseases, such as arthritis, are exacerbated by an inflammatory cascade that involves both immune complex deposition and neutrophil infiltration. However, how exactly immune complexes lead to neutrophil infiltration and other downstream events remains unclear. Now, Miyabe et al. use intravital imaging in joints of arthritic mice to show that complement C5a receptor (C5aR) was critical for neutrophil adhesion on joint endothelium. C5aR induced integrin-dependent neutrophil arrest and spreading, followed by extravasation mediated by leukotriene B4 receptor and crawling mediated by CCR1. CXCR2 was involved in late neutrophil recruitment and survival. These studies provide insight into the complex mechanism of neutrophil infiltration in arthritic joints.

Abstract

The deposition of immune complexes (ICs) in tissues induces a “type III hypersensitivity” that results in tissue damage and underlies the pathogenesis of many autoimmune diseases. The neutrophil is the first immune cell recruited into sites of IC deposition and plays a critical role in shaping the overall tissue response. However, the mechanism by which ICs initiate and propagate neutrophil infiltration into tissue is not known. Using intravital multiphoton joint imaging of IC-induced arthritis in live mice, we found that the complement C5a receptor (C5aR) was the key initiator of neutrophil adhesion on joint endothelium. C5a presented on joint endothelium induced β2 integrin–dependent neutrophil arrest, facilitating neutrophil spreading and transition to crawling, and subsequent leukotriene B4 receptor (BLT1)–mediated extravasation of the first neutrophils. The chemokine receptor CCR1 promoted neutrophil crawling on the joint endothelium, whereas CXCR2 amplified late neutrophil recruitment and survival once in the joint. Thus, imaging arthritis has defined a new paradigm for type III hypersensitivity, where C5a directly initiates neutrophil adhesion on the joint endothelium igniting inflammation.

INTRODUCTION

Deposition of antibody-antigen immune complexes (ICs) in tissues underlies the pathogenesis of a wide variety of autoimmune diseases, including systemic lupus erythematosus, rheumatoid arthritis (RA), vasculitis, serum sickness, post-streptococcal glomerulonephritis, and cryoglobulinemia. In these diseases, collectively referred to as “type III hypersensitivity,” IC deposition leads to a local inflammatory response characterized by immune cell infiltration and activation. The neutrophil is the first immune cell recruited into tissue; however, the mechanism by which deposited ICs initiate and propagate neutrophil infiltration into tissue is not known and is of considerable therapeutic importance (1).

The prevailing paradigm of type III hypersensitivity is that tissue-resident cells sense ICs through Fcγ receptors (FcγRs) and complement receptors and elaborate secondary mediators, such as tumor necrosis factor (TNF) and interleukin-1 (IL-1), which activate endothelial cells (ECs) (1). Adhesion molecules and chemokines up-regulated on the surface of activated ECs are then thought to promote neutrophil recruitment. Although complement and FcγR activation have been variably implicated in the pathogenesis of IC-induced inflammation, their role in directly regulating neutrophil recruitment has not been elucidated.

The pathogenesis of IC-mediated disease has been interrogated in a variety of mouse models, including models of IC-induced inflammation of the skin, cremaster muscle, kidney, peritoneum, lung, and joint (214). In the reverse passive cutaneous Arthus reaction, the inflammatory response in the skin is dependent on mast cell FcγR activation but independent of complement activation (24). In contrast, neutrophil FcγR activation is required for inflammation in models of IC-induced cremaster muscle and joint inflammation (5, 6), whereas tissue-resident macrophage FcγR activation is required in kidney, peritoneal, and lung IC models (710). Inflammation in the cremaster muscle and kidney IC model is similar to the skin model in that they are independent of complement C3 and C5a (5, 8), whereas the peritoneal model is partially dependent on C5a (11, 12). C5a indirectly contributes to IC-mediated lung and peritoneal inflammation by activated FcγRs on resident macrophages (9, 10). In the K/BxN model of IC-induced arthritis, C5a is required for the development of arthritis (6, 14), although the mechanism by which it contributes to neutrophil recruitment and joint inflammation is not clear. Thus, there are important differences in the critical inducers of IC-induced inflammation depending on the organ where they are deposited; however, the mechanisms underlying these differences have not been elucidated to date.

In the K/BxN model, immunoglobulin G (IgG) autoantibodies to the ubiquitous glycolytic enzyme glucose-6-phosphate isomerase (GPI) form ICs with GPI and are preferentially deposited on the cartilage surface of the joint, where they led to the local activation of the alternative complement pathway (AP) (14, 15). This model has similarities to human RA, where joint tissue is frequently covered with ICs, and autoantibodies to ubiquitously expressed citrullinated self protein antigens (ACPAs) are present in most of the patients with RA (16). As in other IgG IC-mediated diseases, neutrophil infiltration into the joint is critical for arthritis in the K/BxN mouse model as it is for joint inflammation in RA (17). Once in the joint, neutrophils become activated and release powerful proteases, cytokines, chemokines, and neutrophil extracellular traps (18). Thus, neutrophils are essential initiators of IC-induced inflammation, including in the K/BxN model of arthritis and in RA. Neutrophils not only induce tissue destruction themselves but also profoundly influence the biology of other immune and structural cells of tissue, including in the joint that leads to RA (18). Therefore, understanding the molecular mechanisms controlling the entry of neutrophils into tissues, such as the joint, could be broadly helpful in understanding the pathogenesis of IC-induced inflammation and aid in the development of new therapies.

The β2 integrin leukocyte function-associated antigen–1 (LFA-1) and four neutrophil chemoattractant receptors (CARs)—C5aR, leukotriene B4 (LTB4) receptor 1 (BLT1), CC chemokine receptor 1 (CCR1), and CXC chemokine receptor 2 (CXCR2)—are all required on neutrophils for the development of arthritis in the K/BxN model of autoantibody IC-induced arthritis (6, 14, 1922). Although we have been able to suggest a temporal order for the requirement of these diverse CARs (23), we have not been able to decipher their respective functional roles in controlling neutrophil entry into the joint using traditional end point analysis. Further, it is not known what induces the very first neutrophils to enter the joint to initiate arthritis in this model. It has been proposed that sentinel tissue–dwelling neutrophils are the key cell that initiates inflammation in this model of IC-induced tissue inflammation (24).

Recent advances in imaging technology have provided unprecedented views into immune cell migration in live animals, which have deepened our understanding of the molecular control of immune cell trafficking into tissue in vivo (25). This process begins with the capture of free-flowing leukocytes on the vessel wall, followed by (i) tethering and rolling on the vessel wall in the direction of flow, (ii) firm arrest on the endothelium, (iii) spreading out and crawling in all directions on the vessel to locate a receptive location for (iv) transendothelial migration (TEM) to extravasate into the tissue (26). Thus, applying multiphoton intravital microscopy (MP-IVM) to study IC-induced inflammation in the joint has the potential to take our understanding of IC-induced inflammation and immune cell entry into the joint to an entirely new level by allowing the real-time visualization of the events that lead to arthritis in live mice. Here, using a new joint imaging technique in live mice, we demonstrate that C5a is specifically presented on joint endothelium, where it directly initiates neutrophil adhesion through its receptor C5aR expressed on neutrophils, igniting joint inflammation in a model of IC-induced arthritis.

RESULTS

C5aR is required for neutrophil adhesion in IC-induced arthritis

Multiple CARs expressed on neutrophils are required for the development of joint inflammation in the K/BxN serum transfer model of arthritis (Fig. 1A). C5aR and BLT1 are absolutely required for arthritis, as is the enzyme 5-lipoxygenase (Alox-5), which is required for the generation of LTB4, the BLT1 ligand. The chemokine receptors CCR1 and CXCR2 also contribute to the development of arthritis and together are absolutely required for arthritis (Fig. 1A). These end point studies, though informative, cannot ascertain the individual respective roles of the chemoattractant (CA) pathways in the control of neutrophil entry into the joint.

Fig. 1 In vivo imaging of inflammatory arthritis in live mice.

(A) Clinical arthritis score of WT and CAR-KO mice on day 7 after AST. (B to I) In vivo imaging of the joint in live mice. Data are representative of three independent experiments. (B) WT-LysM-GFP, no AST. (C to I) WT or CAR-KO-LysM-GFP mice on day 7 after AST. (C) WT, (D) C5ar-KO, (E) Blt1-KO, (F) Alox-5–KO, (G) Ccr1-KO, (H) Cxcr2-KO, and (I) Ccr1-Cxcr2-DKO mice. GFP, neutrophils; blue, connective tissue; white (Qdots), blood vessels. Time in minutes:seconds. Scale bars, 50 μm. Number of newly adherent cells (J) and newly extravasated cells (K) in the joints during 30-min recording. Data are means ± SEM. n = 3 mice per group; P value was calculated using unpaired two-tailed Student’s t test.

We have adapted MP-IVM in live mice to image the individual roles of the CARs implicated in neutrophil recruitment into the joint and the development of arthritis in real time (27). We have used LysM-GFP mice, where expression of the green fluorescent protein (GFP) is driven by the lysozyme M promoter, resulting in GFP-positive neutrophils and monocytes, to visualize neutrophil migratory behavior in the joints in vivo. Imaging joints of control (no arthritis) and arthritic wild-type (WT)–LysM–GFP mice revealed the applicability of this technique (Fig. 1, B and C, and movie S1). Control mice had very few neutrophils visibly interacting with blood vessels of the joint, and no neutrophils were visible outside of blood vessels (Fig. 1B and movie S1). In contrast, on day 7 after arthritogenic serum transfer (AST), neutrophils that adhered to the blood vessels were detected. This included cells that were still round and firmly arrested, as well as cells that had spread out and were crawling on the endothelium in multiple directions (Fig. 1C and movie S1). In addition, many neutrophils were also seen outside of blood vessels and were observed to be actively mobile and moving in multiple directions in inflamed joints (Fig. 1C and movie S1).

To reveal how deficiency in CARs might influence neutrophil migratory behavior in the joint, we imaged Blt1-deficient (KO)-LysM-GFP, C5ar-KO-LysM-GFP, Alox-5–KO–LysM-GFP, Ccr1-KO-LysM-GFP, Cxcr2-KO-LysM-GFP, and Ccr1-Cxcr2-DKO-LysM-GFP mice 7 days after AST. We observed a notable phenotype in C5ar-KO mice: Neutrophils could not adhere to blood vessels of the joint (Fig. 1D and movie S1). The number of newly adherent neutrophils in C5ar-KO-LysM-GFP mice was markedly less than that in other CAR-KO mice during the 30-min observation (Fig. 1J). In contrast, neutrophils could adhere to joint blood vessels in the other CAR-KO and Alox-5–KO mice on day 7 after AST (Fig. 1, E to I, and movie S1). However, the number of newly extravasated neutrophils that entered the tissue during the 30-min observation was markedly less in all CAR-KO and Alox-5–KO mice compared with WT mice (Fig. 1K). These data demonstrate that C5aR plays a role in initiating neutrophil adhesion in the model and that the other CARs participate in the process of neutrophil entry into the joint but at later stages in the process.

Cell-intrinsic role for C5aR in neutrophil adhesion in the inflamed joint

To study the cell-intrinsic role for each CAR in neutrophil migratory behavior in the setting of full-blown arthritis, we imaged joints 2 hours after adoptively transferring purified Actin-RFP-WT and Actin-GFP-CAR-KO neutrophils into WT mice that had received AST 7 days before (fig. S1A). We confirmed that Actin-RFP-WT and Actin-GFP-CAR-KO neutrophils were present in a 1:1 ratio in the blood (fig. S1B). In these experiments, we could analyze neutrophil recruitment according to three defined sequential phases of the leukocyte migration cascade: firm arrest, spreading/crawling, and extravasation (Fig. 2A). As was observed in C5ar-KO mice, markedly fewer C5ar-KO neutrophils were observed, arrested and crawling on joint blood vessels, compared with WT neutrophils when cotransferred into WT mice with arthritis (Fig. 2, B, G, and H). In contrast, other CAR-KO neutrophils did not display any defect in arrest or crawling compared with coadoptively transferred WT neutrophils (Fig. 2, C to H). However, as was observed for gene-deficient mice, adoptively transferred CAR-KO neutrophils studied had impaired extravasation when compared with coadoptively transferred WT neutrophils (Fig. 2I). In addition, there was a marked delay in the time to firm arrest, spreading/crawling, and extravasation for C5ar-KO neutrophils compared with WT neutrophils, whereas there was only a delay in the time to extravasation for the other CAR-KO neutrophils studied (fig. S1C). As might be expected in the setting of full-blown arthritis, the phenotype of adoptively transferred CAR-KO neutrophils was less marked than what was observed for neutrophils in CAR-KO mice, where all neutrophils were CAR-deficient and mice had no evidence of joint inflammation.

Fig. 2 In vivo imaging WT and CAR-KO neutrophils coadoptively transferred into arthritic WT mice.

(A) Schematic of neutrophil migration cascade. (B to F) In vivo imaging of the joint after coadoptive transfer of WT-Actin-RFP and CAR-KO-Actin-GFP purified neutrophils into WT mice that received AST 7 days before. Data are representative of three independent experiments. (B) WT and C5ar-KO, (C) WT and Blt1-KO (D), WT and Ccr1-KO (E), WT and Cxcr2-KO, and (F) WT and Ccr1-Cxcr2-DKO neutrophils. GFP, CAR-KO neutrophils; RFP, WT neutrophil; blue, connective tissue; white (Qdots), blood vessels. Scale bars, 50 μm. Time in hours:minutes. Green arrows, extravasated CAR-KO neutrophils; red arrows, extravasated WT neutrophils. (G to I) Quantitation of the number of new (G) arrested, (H) crawling, and (I) extravasated neutrophils in the joint during the 120-min observation. Data are means ± SEM. n = 3 mice per group; P value was calculated using unpaired two-tailed Student’s t test. N.S., not statistically significant.

To confirm the cell-intrinsic role of C5aR in neutrophil adhesion and the cell-intrinsic role of the other CARs in neutrophil extravasation, we generated mixed bone marrow chimeric (BMC) mice (fig. S2A). In these mice, BM cells from WT-Actin-RFP and CAR-KO-Actin-GFP mice were transferred in a 1:1 ratio into irradiated WT mice to determine the cell-intrinsic effect of CAR deficiency in mice over the entire course of arthritis development (fig. S2B). All the mixed BMC mice developed arthritis after AST, although, as might be expected, mixed BMC mice developed less severe arthritis than did BMC control mice (WT BM into irradiated WT mice) (Fig. 3A). Again, markedly fewer C5ar-KO neutrophils were observed sticking to the joint endothelium compared with WT neutrophils in these mixed BMC that developed arthritis (Fig. 3, B and F). As was seen in the adoptive transfer experiments, the other CAR-KO neutrophils studied did not exhibit defects in adhesion to the joint endothelium compared with WT neutrophils in arthritic BMC mice (Fig. 3, C to F). However, all the CAR-KO neutrophils studied in the mixed BMC mice had impaired extravasation compared with WT neutrophils after AST (Fig. 3G). Again, as might be expected, the phenotype of CAR-KO neutrophils in the mixed BMC arthritic mice was less marked than what was observed for neutrophils in CAR-KO mice, where all neutrophils were CAR-deficient and mice had no evidence of joint inflammation. These data confirm the conclusion that C5aR is uniquely required for initial neutrophil adhesion in this model, whereas other CARs contribute to extravasation into inflamed joints.

Fig. 3 In vivo imaging joints of mixed BMC mice on day 7 after AST.

(A) Clinical arthritis score on day 7 after AST. (B to E) Representative images of the joints of mixed BMC mice on day 7 after AST. Mixed BMC mice were generated by the transfer of WT-Actin-RFP and CAR-KO-Actin-GFP BM (1:1 ratio) into irradiated WT mice. Whereas leukocytes other than neutrophils were fluorescently labeled in these experiments, neutrophils were easily identified by their size, shape, and migratory properties. Data are representative of three independent experiments. (B) C5ar-KO-Actin-GFP and WT-Actin-RFP BM cells transferred into WT mice. (C) Blt1-KO-Actin-GFP and WT-Actin-RFP BM cells transferred into WT mice. (D) Ccr1-KO-Actin-GFP and WT-Actin-RFP BM cells transferred into WT mice. (E) Cxcr2-KO-Actin-GFP and WT-Actin-RFP BM cells transferred into WT mice. GFP, CAR-KO cells; RFP, WT cells; blue, connective tissue; white (Qdots), blood vessels. Time in minutes:seconds. Scale bars, 50 μm. (F and G) Quantitation of sticking cells and extravasated cells in the joint. (A, F, and G) Data are means ± SEM. n = 3 mice per group; P value was calculated using unpaired two-tailed Student’s t test.

C5aR and CCR1 mediate neutrophil-endothelium interactions at different stages of the adhesion cascade

In the coadoptive transfer experiments, we found that within 5 min after arrest on the endothelium in arthritic joints, WT, Ccr1-KO, Cxcr2-KO, Ccr1-Cxcr2-DKO, and Blt1-KO neutrophils transformed their shape from round to amoeboid and initiated crawling (Fig. 4A and movie S2). In contrast, C5ar-KO neutrophils that managed to arrest on the endothelium in arthritic mice maintained their round shape (circularity) as measured by ImageJ software, could not spread, and eventually detached (Fig. 4, A and B, and movie S2). Notably, ~40% of C5ar-KO–arrested neutrophils detached in the adoptive transfer experiments, which is a phenomenon that was not observed for adoptively transferred WT neutrophils (Fig. 4, C and D, and movie S3). In addition, ~20% of adoptively transferred Ccr1-KO and Ccr1-Cxcr2-DKO neutrophils that initiated crawling were observed to detach from the endothelium, a phenomenon that was also not observed for WT adoptively transferred neutrophils that initiated crawling (Fig. 4, E to G, and movie S3). Cxcr2-KO crawling neutrophils did not detach, suggesting that detachment of Ccr1-Cxcr2-DKO crawling neutrophils was the result of a CCR1 deficiency.

Fig. 4 C5aR and CCR1 mediate interactions of neutrophils with the joint endothelium at different stages of the adhesion cascade.

(A) Shape of WT and CAR-KO neutrophils 5 min after arrest. Broken white line outlines the shape of the neutrophils. Scale bar, 10 μm. (B) Circularity of WT and CAR-KO neutrophils 5 min after arrest. Data are means ± SEM; n = 10 cells per genotype. (C to G) In vivo joint imaging after adoptive transfer of WT-Actin-RFP and CAR-KO-Actin-GFP neutrophils into WT mice that received AST 7 days before. Blue, connective tissue; white (Qdots), blood vessels. (C) WT and C5ar-KO neutrophils. Blue arrows, cells that detached. Time in minutes:seconds. Scale bars, 10 μm. Data are representative of three independent experiments. (D) Quantitation of arrested cells that detached from the joint endothelium over the 120-min observation. WT and Ccr1-KO (E) and WT and Ccr1-Cxcr2-DKO (F) neutrophils. Green arrows, crawling cells that detached; green broken lines, track of a mobile cell. Time in minutes:seconds. Scale bars, 10 μm. (G) Quantitation of crawling cells that detached from the joint endothelium over the 120-min observation. (C, E, and F) Data are representative of three independent experiments. (D and G) Data are means ± SEM; n = 3 mice per group; P value was calculated using unpaired two-tailed Student’s t test.

We observed the same detachment phenotypes for C5ar-KO, Ccr1-KO, and Ccr1-Cxcr2-DKO neutrophils in mixed BMC mice after serum transfer (fig. S2, C to F). C5ar-KO neutrophils that could arrest were observed to detach from the endothelium of arthritic joints in mixed C5ar-KO:WT BMC mice. Likewise, Ccr1-KO and Ccr1-Cxcr2-DKO neutrophils that initiated crawling were observed to detach from the endothelium of arthritis joints in Ccr1-KO:WT and Ccr1-Cxcr2-DKO:WT mixed BMC mice, respectively. These detachment phenotypes were not observed for WT neutrophils in these mice (fig. S2, C to F). Cxcr2-KO neutrophils were not observed to detach after initiating crawling on joint endothelium in Cxcr2-KO:WT mixed BMC mice, again suggesting that detachment of Ccr1-Cxcr2-DKO crawling neutrophils was the result of a CCR1 deficiency. Together, these data suggest that C5aR signaling in neutrophils is also important for transition from firm arrest to spreading and crawling, whereas CCR1 signaling is important to maintain cell crawling on the endothelium of the joint in this model as the cell probes for a receptive location to undergo TEM.

Early generation of C5a in the joint and retention on joint endothelium

To determine the kinetics of neutrophil CA generation in the model, we analyzed synovial fluid (SF) from control and arthritic mice on days 1, 3, and 7 after AST for levels of CAs by enzyme-linked immunosorbent assay (ELISA) (Fig. 5A). In the joint space, C5a generation could be detected on day 1 after AST and rose to 76.33 ± 4.33 pM by day 3 and 856.7 ± 72.19 pM by day 7. LTB4 production was detected beginning on day 3 and rose to 146.7 ± 27.28 pM by day 7. Chemokines were first detected on day 7. These data demonstrate that C5a is generated in the joint first followed by LTB4 and then chemokines.

Fig. 5 C5a expression and binding in the inflamed joint.

(A) Levels of CAs in SF on days 0, 1, 3, and 7 after AST as determined by ELISA. n = 3 mice per time point (B to E) Immunofluorescence staining of joint tissue from arthritic mice on day 7 after AST (B and D) or control mice (C and E). Red, C5a; green, VWF+ ECs; blue, 4′,6-diamidino-2-phenylindole (DAPI). Scale bars, 50 μm. Data are representative of three independent experiments. (F) In vitro binding assay. Amount of C5a bound to uncoated 96-well assay plates or plates coated with HSPG (5 μg/ml) in the presence or absence of heparin (80 μg/ml). (G) In vitro adhesion assay. Number of BM neutrophils adherent to murine aortic ECs that were pretreated with C5a (1 nM) in the presence or absence of heparin (80 μg/ml). (F and G) n = 3 independent experiments. Data are means ± SEM; P value was calculated using unpaired two-tailed Student’s t test.

C5a deposition was detected on joint cartilage in mice with arthritis (day 7) but not in the joints of control mice (Fig. 5, B and C). In addition, C5a immunoreactivity was also detected on von Willebrand factor (VWF+) endothelium of joints in arthritic mice but not on the endothelium in the joints of control mice (Fig. 5, D and E). Further, we did not detect C5a decorating VWF+ endothelium of other organs from the same arthritic mice, including the aorta, coronary artery, superior vein cava, kidney, and lung (fig. S3).

Because other basic CAs, like many of the chemokines, bind heparan sulfate proteoglycans (HSPGs) on the surface of ECs, we tested whether C5a also has this property. C5a bound directly to HSPG, and, as would be expected, heparin inhibited this interaction (Fig. 5F). In addition, C5a could bind to ECs and induce neutrophil adhesion in vitro (Fig. 5G). C5a binding to ECs was also inhibited by heparin, suggesting that C5a was binding to HSPG expressed on the surface of ECs. Thus, C5a generated in the joint space is captured by HSPG expressed on joint ECs, and this initiates the neutrophil adhesion cascade in the joint.

C5a is a strong initiator of neutrophil adhesion in the joint

To directly demonstrate that C5a induces neutrophil adhesion in vivo, we developed an in vivo neutrophil adhesion and extravasation assay. Joints of control WT-LysM-GFP mice were surgically exposed, and exogenous CAs were added to the surface of the exposed joint and imaged for 30 min. All CAs tested could induce neutrophil adhesion and extravasation in the joint compared with the phosphate-buffered saline (PBS) control (Fig. 6, A to F, and movie S4). C5a was the most efficacious CA at inducing adhesion and the least efficacious at inducing extravasation (Fig. 6G). In contrast, CXCL1 was the most efficacious at inducing neutrophil extravasation (Fig. 6H). C5a-induced neutrophil adhesion was dose-dependent (0.1 to 100.0 nM), with maximal adhesion seen at 10 nM (fig. S4, A and B). Although we imaged for 120 min, C5a (10 nM) did not induce neutrophil extravasation to nearly the same extent as CXCL1 (1 nM) (fig. S4, C and D). C5a-induced neutrophil adhesion was also observed in the joint of Blt1-KO-LysM- and Ccr1-Cxcr2-DKO-LysM-GFP mice, suggesting that C5a-C5aR signaling directly induces adhesion (fig. S4, E to G). These data suggest that C5a is a potent, direct initiator of neutrophil adhesion in the joint in vivo.

Fig. 6 In vivo joint adhesion assay.

(A to F) In vivo joint imaging of WT-LysM-GFP mice after exposure to CAs (n = 3). (A) Control. (B) CXCL1. (C) CXCL2. (D) CCL3. (E) C5a. (F) LTB4. GFP, neutrophils; blue, connective tissue; white (Qdots), blood vessels. Time in minutes:seconds. Scale bars, 50 μm. Images are representative from three independent experiments. Number of newly adherent cells (G) and newly extravasated cells (H) in the joints over the 30-min observation after the application of the CA (1 nM). (G and H) Data are means ± SEM; n = 3 mice per group; P value was calculated using unpaired two-tailed Student’s t test.

C5a-induced neutrophil adhesion via β2 integrin activation

To determine whether C5a-induced neutrophil adhesion was mediated by β2 integrins, we performed an in vitro adhesion on intercellular adhesion molecule–1 (ICAM-1)–coated plates. C5a induced neutrophil binding to ICAM-1, suggesting that C5a activates β2 integrins on neutrophils (Fig. 7A). Next, to detect which β2 integrin is important for C5a-mediated adhesion, we inhibited Mac-1 and LFA-1 using blocking antibodies. Inhibition of Mac-1 or LFA-1 partially inhibited C5a-induced neutrophil adhesion, and blocking both Mac-1 and LFA-1 was additive (Fig. 7A). These data suggest that C5a activates both LFA-1 and Mac-1 on neutrophils and induces neutrophil adhesion in vitro.

Fig. 7 β2 integrin activation mediates C5a-induced neutrophil adhesion to the joint endothelium.

(A) In vitro adhesion of C5a (1 nM)–activated WT neutrophils to ICAM-1–coated plates in the presence of β2 integrin–blocking monoclonal antibodies (mAbs) or an isotype control mAb. n = 3 independent experiments. Data are means ± SEM; P value was calculated using unpaired two-tailed Student’s t test. (B to F) In vivo imaging in joints of WT-LysM-GFP mice that were pretreated for 4 hours with β2 integrin–blocking mAbs or an isotype control mAb after exposure to C5a (1 nM). (B) Control antibody. (C) LFA-1 antibody. (D) Mac-1 antibody. (E) LFA-1– and Mac-1 antibody–treated mice. Green, neutrophils; blue, connective tissue; white (Qdots), blood vessels. Scale bars, 50 μm. Time in minutes:seconds. Data are representative of three independent experiments. (F) Number of newly adherent cells on the joint endothelium over the 30-min observation. (G to I) Neutrophil detachment in vivo. (G) LFA-1 antibody. (H) LFA-1 antibody– and Mac-1 antibody–treated mice. Green, neutrophils; blue, connective tissue; white (Qdots), blood vessels. Green arrows, crawling cells that detached; blue arrows, arrested cells that detached. Time in minutes:seconds. Scale bars, 10 μm. Data are representative of three independent experiments. (I) Number of newly adherent cells that detached over the 30-min observation. Striped bar indicates arrested cells that detached and closed bar indicates crawling cells that detached of the total number of adherent cells that detached. (F and I) n = 3 mice per group. Data are means ± SEM; P value was calculated using unpaired two-tailed Student’s t test.

We then compared the expression levels of β2 integrin on BM, blood, and SF neutrophils isolated from arthritic mice (day 7). LFA-1 and Mac-1 were highly expressed on the surface of neutrophils isolated from the BM, blood, and SF of arthritic mice (fig. S5A). In addition, we confirmed that deficiency of CARs on neutrophils did not alter LFA-1 and Mac-1 levels on BM neutrophils (fig. S5B). Next, we analyzed whether C5a-mediated neutrophil adhesion in vivo was mediated by β2 integrin activation using our in vivo adhesion assay. Blocking Mac-1 or LFA-1 inhibited C5a-induced neutrophil adhesion in the joint (Fig. 7, B to E, and movie S5). Dual blockade of LFA-1 and Mac-1 was even more effective than single blockade at inhibiting C5a-induced neutrophil adhesion in vivo (Fig. 7, B and F, and movie S5). Thus, these in vivo studies confirm our in vitro findings demonstrating that C5a-C5aR signaling on neutrophils activates β2 integrin, inducing neutrophil adhesion in the joint.

In addition, blockade of β2 integrins resulted in neutrophil detachment after C5a-induced adhesion (Fig. 7, G to I and movie S6). LFA-1 blockade induced crawling neutrophils to detach from the endothelium (Fig. 7G and movie S6), whereas blockade of both Mac-1 and LFA-1 also induced arrested neutrophils to detach (Fig. 7H and movie S6). These data suggest that LFA-1 primarily mediates the strength of the interaction of C5aR-activated neutrophils with the endothelium.

CXCR2 ligands promote neutrophil recruitment and survival in the joint

The spontaneous production of neutrophil-active chemokines from neutrophils isolated from the blood and joint of arthritic and control mice was determined by ELISA. Neutrophils isolated from the BM and blood of control mice did not spontaneously release neutrophil-active chemokines, including CXCL1, CXCL2, and CCL3. In contrast, neutrophils isolated from SF and blood, but not BM, of arthritic mice spontaneously produced CXCL2 (Fig. 8A), suggesting that activated neutrophils spontaneously release CXCL2.

Fig. 8 CXCR2 propagates neutrophil recruitment and survival in the joint.

(A) Quantitation of CXCL2 produced by neutrophils isolated from the BM, blood, and SF of WT control and arthritic mice (day 7 after AST). n = 3 mice per group. (B) Surface protein CAR expression on neutrophils isolated from the BM, blood, and SF of arthritic mice by flow cytometry. Data are representative of three independent experiments. (C to F) Neutrophils were isolated from blood and SF of mixed BMC mice on days 1, 3, and 7 after AST. WT and (C) Ccr1-KO, (D) Cxcr2-KO, (E) C5ar-KO, and (F) Blt1-KO neutrophils in blood and SF were analyzed by flow cytometry. n = 3 independent experiments per time point. (G and H) Neutrophils were isolated from SF of WT mice on day 7 after AST and incubated with CXCL1 (1 nM) and/or CXCL2 (1 nM) for 48 hours. Neutrophil survival was assessed by flow cytometry using annexin+ (apoptosis) and PI+ (necrosis) (G), and annexin+ cells were quantitated (H). n = 3 independent experiments. (A, C to F, and H) Data are means ± SEM; P value was calculated using unpaired two-tailed Student’s t test.

We also analyzed the cell surface expression of the neutrophil-active CARs studied here—C5aR, BLT1, CCR1, and CXCR2—on neutrophils isolated from the BM, blood, and SF of arthritic mice. C5aR, BLT1, and CCR1 were expressed on BM, blood, and SF neutrophils, and the levels of these CARs were similar on neutrophils isolated from these three compartments (Fig. 8B). In contrast, CXCR2 surface expression was low on BM neutrophils and increased markedly on neutrophils as they moved into the blood and then into the SF (Fig. 8B).

We also analyzed the kinetics of neutrophil accumulation in the SF of mixed BMC mice that had received AST 1, 3, and 7 days before. In all mixed BMC mice, the ratio of WT and CAR-KO neutrophils in blood was equal over the entire course of arthritis (Fig. 8, C to F). Initially, at day 1 after AST, the vast majority of neutrophils in the SF were WT (Fig. 8, C to F). However, the % C5ar-KO, Blt1-KO, and Ccr1-KO neutrophils in SF increased over time (Fig. 8, C, E, and F). In contrast, Cxcr2-KO neutrophils did not increase in SF over time (Fig. 8D). These data suggest that CXCR2 promotes recruitment of C5ar-KO neutrophils into the arthritic joint after disease onset. In addition, we tested whether CXCR2 also contributes to neutrophil survival once in the joint, given reports that RA SF can prevent neutrophil apoptosis (28). SF neutrophils were isolated from the joints of arthritic mice (day 7 after AST) and incubated for 48 hours with CXCL1 and/or CXCL2. Apoptotic cells were identified as annexin V+ and necrotic cells as PI+ (propidium iodide–positive). Culture of SF neutrophils with CXCR2 ligands markedly reduced the population of annexin V+ neutrophils, suggesting that apoptosis was markedly inhibited compared with untreated SF neutrophils (Fig. 8H). These data suggest that, in addition to amplifying neutrophil recruitment into the joint, CXCR2 also directly promotes the survival of neutrophils once in the inflamed joint.

DISCUSSION

Although it is known that ICs are important inducers of tissue inflammation, the mechanism by which IC deposition leads to the recruitment of immune cells into tissue is not known. Using a joint imaging technique in live mice and a murine model of IC-induced arthritis, we have identified a new role for C5a as the direct initiator of this process in the joint: C5a presented on joint endothelium activated C5aR on neutrophils, inducing β2 integrin–dependent neutrophil adhesion. C5aR signaling also facilitated the transition of neutrophils from firm arrest to spreading and crawling for subsequent BLT1-dependent extravasation of the very first neutrophils into the joint. CCR1 contributed to the interaction of crawling neutrophils with the joint endothelium, and CXCR2 amplified late neutrophil recruitment and promoted neutrophil survival once in the joint. Thus, we have defined the roles of the CAR system in collaborating to control neutrophil recruitment into tissue in a type III hypersensitivity IC reaction. In so doing, we have also demonstrated a role for C5a in directly initiating neutrophil recruitment and inflammation in the joint after IC deposition.

In the K/BxN model, circulating ICs induce the production of vasoactive amines from radioresistant FcγR-expressing cells that reside in the liver (13). This results in a localized transient vascular leak in the joint, which leads to the delivery of ICs into the joint space and their deposition on the surface of joint cartilage (13, 15). Given that the cartilage surface has a relative paucity of inhibitory complement regulatory proteins, such as decay-accelerating factor, this leads to the local activation of the AP and the generation of C5a as ICs augment the spontaneous autoactivation of C3 in a process termed “tick-over” (15, 29). Consistent with this hypothesis, C3 is also required for arthritis in this model (14). C3 and its receptor C3aR have also been implicated in RA pathogenesis (30), and AP activation has been associated with disease activity in adult and juvenile RA (31, 32).

Our data now demonstrate that C5a specifically decorates the inflamed joint endothelium, and this is critical for activation of β2 integrins on neutrophils and the adhesion of the very first neutrophils, which initiates joint-specific inflammation. We found that C5a was generated in the joint space before the generation of the other CAs. C5aR was expressed on blood neutrophils and on those recovered from the BM and SF, demonstrating that C5aR can respond to endothelially presented C5a to induce initial neutrophil adhesion on joint blood vessels. C5a could bind directly to joint endothelium and to HSPG in vitro, and both of these interactions were inhibited by heparin. Further, C5a superfused on joints in live mice was the most potent CA at inducing neutrophil adhesion and was the least efficacious at inducing extravasation. We therefore postulate that much like chemokines, locally produced C5a is bound by HSPG on EC, and this retains C5a locally, presenting it to rolling neutrophils. It is also possible that similar to what has been proposed for chemokines, C5a generated in the joint is transported to the surface of ECs by an atypical CAR-like receptor, such as C5L2. The atypical chemokine receptor ACKR1 could transport inflammatory chemokines from the ablumen to the lumen of blood vessels (33). It has also been reported that C5aR is expressed at low levels on EC and that C5aR signaling can induce the expression of ICAM-1 on human umbilical vein endothelial cells in vitro as well as disrupt vascular integrity (34, 35), both of which could contribute to joint inflammation. However, in the K/BxN model, C5aR on ECs does not play a major role, because BMC experiments demonstrated that C5aR was only required on hematopoietic cells for the development of arthritis (36).

In addition to mediating initial adhesion, C5aR signaling also played a role in the transition to spreading and crawling. In coadoptive transfer experiments and in mixed BMC experiments, we found that adherent C5ar-KO neutrophils were prone to detach from the endothelium. This phenotype is something that was not observed for WT neutrophils. Thus, the strength and/or duration of arrest likely contribute to the ability of neutrophils to transition from firm arrest to spreading and crawling. Both Mac-1 and LFA-1 contribute to the strength of zymosan-activated neutrophil adhesion to ECs, with LFA-1 playing a more important role (37). The intraluminal crawling that follows arrest and spreading is essential for TEM (38). In TNF-activated and CXCL2-superfused cremaster muscle venules, neutrophil arrest and crawling were mediated by β2 integrins and their ligands ICAM-1 and ICAM-2. Arrest was mainly mediated by LFA-1, whereas luminal crawling was mainly mediated by Mac-1 (38, 39). Our data demonstrated that C5a induced both Mac-1– and LFA-1–mediated neutrophil adhesion in vitro and in vivo and that LFA-1 played a more important role in mediating the strength of the interaction with the endothelium. Thus, C5a-C5aR signaling activates Mac-1 and LFA-1 on neutrophils for adhesion and interaction with the endothelium.

How other CARs collaborate with C5aR to induce adherent neutrophils to enter into the inflamed joint in this model is largely unknown. However, like C5ar-KO mice, Alox-5–KO and Blt1-KO mice are completely resistant to the development of arthritis in the K/BxN AST model (19, 21). Our in vivo imaging experiments demonstrated that neutrophils could not extravasate into the joint in these mice after AST. However, in contrast to neutrophils in the C5ar-KO mice, neutrophils in Alox-5–KO and Blt1-KO mice could arrest and transition to crawling after AST. However, neutrophils could not undergo TEM, suggesting that LTB4-BLT1 signaling initiates the first neutrophil entry into the joint. Given that C5aR signaling is required for LTB4 production in the model (36) and that C5a is generated before LTB4, our data suggest that C5a-C5aR signaling initiates neutrophil adhesion and the production of LTB4, which results in the initial entry of neutrophils into the joint, triggering the initiation of arthritis.

Once neutrophils enter the joint and initiate arthritis, C5aR and BLT1 were not absolutely required for neutrophil recruitment into the joint. Although our neutrophil coadoptive transfer and mixed BMC experiments revealed a cell-intrinsic role for C5aR for adhesion and BLT1 for neutrophil TEM, there was only a 50% inhibition of these processes compared with their complete absence when the mouse was completely deficient in C5aR or BLT1. The reason for these differences likely reflects the fact that, in the coadoptive transfer experiments and mixed BMC experiments, WT neutrophils are present and enter the joint, where they become activated and produce cytokines, such as IL-1, and chemokines, such as CXCL2, within the joint (20). In addition, neutrophil-produced IL-1 induces neutrophil-active chemokine production from fibroblast-like synoviocytes and ECs, which likely then partially relieves the absolute requirement for C5aR and BLT1. Consistent with this, our data demonstrate that production of CXCR2 and CCR1 ligands in the joint occurs later than the generation of C5a and LTB4, consistent with the hypothesis that C5a and LTB4 are required for adhesion and TEM of the very first neutrophils for disease onset, whereas chemokines mainly contribute to the amplification and sustainment of neutrophil recruitment after disease onset.

In this regard, we also found roles for the chemokine receptors CCR1 and CXCR2 for neutrophil recruitment and the pathogenesis of joint inflammation. Ccr1-KO neutrophils specifically detached after initiating crawling, a behavior that was not seen for WT or for any of the other CAR-deficient neutrophils studied, suggesting that CCR1 contributes to the interaction of neutrophils with the endothelium of the joint. CCR1 ligands are well known to bind HSPG on ECs (40). Thus, CCR1 appears to support neutrophil extravasation by facilitating the ability of neutrophils to crawl on the endothelium of inflamed joints to locate receptive locations for TEM. We have also found that CXCR2 is involved with the sustained accumulation of neutrophil in the joint at later time points. Our mixed BMC mice demonstrated that Cxcr2-KO neutrophils were not found in the SF even at later points. This was in contrast to Blt1-KO, C5ar-KO, and Ccr1-KO neutrophils that accumulated in the SF over time. This suggests that CXCR2 plays a role not only in neutrophil recruitment into the joint but also in neutrophil survival once in the SF. Along these lines, we found that CXCR2 ligands prevented apoptosis of SF neutrophils. It is interesting to note that neutrophils are the main cell type found in RA SF, which has been shown to prevent neutrophil apoptosis (28).

SF neutrophils from arthritic mice spontaneously secreted high levels of CXCL2 and produced IL-1, which can induce CXCR2 ligand generation from resident cells in the joints (20). A recent study also demonstrated that IC can induce CXCL2 production from neutrophils and that CXCL2 released from neutrophils can induce CXCL1 production from macrophages (41), suggesting that neutrophils are well equipped to self-regulate their own recruitment. Using an in vivo recruitment assay, we found that CXCL1 was the most efficacious CA at inducing neutrophil extravasation into the joint. We also found that levels of CXCR2 were higher on neutrophils isolated from the SF compared with the blood and BM of arthritic mice. Our data therefore demonstrate that CXCR2 is an important mediator of late neutrophil recruitment and possibly survival in the joint space of arthritic mice.

Thus, imaging IC-induced arthritis has defined a mechanism for a type III hypersensitivity, where C5a directly initiates neutrophil adhesion on the joint endothelium, resulting in neutrophil recruitment into the joint and the development of arthritis. These conclusions have some important limitations. First, the mechanism described here for the joint is likely different from other organs, where tissue-resident cells have been shown to play an important role in initiating inflammation and where C5a indirectly contributes to inflammation by activating resident immune cells (9, 10). The unique environment of the joint with a paucity of resident immune cells may be a determining factor for the critical proximal role of C5a and C5aR in initiating IC-induced arthritis. Further, the prominent role for C5a described here may not be applicable to other models of RA or to human RA, where the role of IC in driving joint inflammation is likely more heterogeneous.

In summary, imaging arthritis has identified a mechanism whereby the C5a-C5aR pathway directly controls neutrophil adhesion on the endothelium and then facilitates their transition to crawling for subsequent BLT1-mediated entry into the joint to ignite inflammation. We have also found roles of CCR1 in mediating neutrophil crawling on the endothelium and CXCR2 in mediating neutrophil survival in the joint space. Thus, we have defined how the CAR system collaborates to control neutrophil entry into the joint after IC deposition (fig. S6) and, in so doing, have established a paradigm for type III hypersensitivity.

MATERIALS AND METHODS

Study design

The aim of this study was to use in vivo joint imaging in live mice to determine the respective roles of C5aR, BLT1, CCR1, and CXCR2 in the recruitment of neutrophils into the joint in an IC-induced model of arthritis. Age (8 to 10 weeks old)– and sex (only male)–matched WT and CAR-KO mice bred and housed in the same facility at Massachusetts General Hospital were used for this study. The sample size was determined to achieve a power of 90%, accepting type I error rate of 0.05. Analysis of movies, as well as in vitro adhesion assays and flow cytometry, was performed by an investigator blinded to the experimental groups. Randomization and blinding were not used for in vivo joint imaging.

Mice

K/BxN mice were obtained by crossing KRN with NOD/LtJ mice (Jackson Laboratory). Ltb4r1−/− (42), C5ar−/− (43), Alox-5−/− (44), Cxcr2−/− (45), Ccr1−/− (46), and LysM-GFP (47) mice were maintained in our laboratory. WT C57BL/6 mice were purchased from Charles River. C57BL/6-Tg(CAG-EGFP)131Osb/LeySopJ (Actin-GFP) and B6.Cg-Tg(CAG-DsRed*MST)1Nagy/J (Actin-RFP) were purchased from the Jackson Laboratory. Ccr1−/−Cxcr2−/− mice were generated by crossing Ccr1−/− and Cxcr2−/− mice, CAR-KO-LysM-GFP and Actin-GFP mice were generated by crossing CAR-KO mice and WT-LysM-GFP or WT-Actin-GFP mice, and genotypes were confirmed by Transnetyx Inc. All experiments were performed according to the protocols approved by the Massachusetts General Hospital Subcommittee on Research Animal Care.

Serum transfer and arthritis scoring

Pooled serum from 8-week-old arthritic K/BxN mice was transferred into recipient mice (100 μl) intraperitoneally on days 0 and 2, as described (27). Clinical arthritis was scored as described (27).

Multiphoton intravital microscopy and image analysis

To perform MP-IVM, the skin above the lateral ankle joint of anesthetized mice was depilated, the ankle joint was microsurgically exposed, and fixed by tissue bond to a glass slide on the microscope stage as described (27). Multiphoton excitation was obtained through two MaiTai Ti:sapphire lasers (Newport/Spectra-Physics) tuned to 820 and 920 nm to excite all fluorescent probes used. Stacks of 11 square optical sections with 4-μm z-spacing were acquired on an Ultima IV multiphoton microscope (Prairie Technologies) every 15 s with optical zoom of ×2 to provide image volumes 40 μm in depth and 307 μm in width. Emitted fluorescence was detected through 460/50, 525/50, 595/50, and 660/40 band-pass filters and non-descanned detectors to generate four-color images. Sequences of image stacks were transformed into volume-rendered, time-lapse movies with Imaris software (Bitplane). (i) Imaging of endogenous neutrophils: Joints of WT or CAR-KO-LysM-GFP mice on day 7 after AST were observed for 30 min. The number of neutrophils that were observed to newly stick, to arrest, to crawl, and to diapedese the endothelium (extravasate), were quantitated. Neutrophils were identified by their size, shape, and migratory behavior. (ii) Imaging adoptively transferred neutrophils: BM neutrophils were isolated by EasySep kit as described (27). A total of 107 BM neutrophils (5 × 106 BM WT Actin-RFP and 5 × 106 BM CAR-KO Actin-GFP neutrophils) were transferred into arthritic WT mice on day 7 after AST, which were matched for the same clinical arthritis score. Joints were imaged for 120 min after BM neutrophil transfers. Population of circulating WT-GFP+ and CAR-KO-RFP+ neutrophils was confirmed by flow cytometry. Neutrophils were scored for the number that arrested, crawled, and extravasated during the observation period. Arrest was defined as a round shape cell that remained in the same position for at least 30 s. Crawling was defined as an amoeboid shape cell that crawled along the inside of blood vessel. Extravasation was defined as a neutrophil that had entered into the tissue (27). (iii) In vivo adhesion assay: CAs [0.1 pmol: C5a (R&D Systems), CXCL1, CXCL2, CCL3 (PeproTech), and LTB4 (Tocris Bioscience)] dissolved with sterile PBS (100 μl) were placed on the joints of WT-LysM-GFP mice and imaged for 30 min. Mac-1 and/or LFA-1 antibodies or isotype control antibody (100 μg per mouse) was intravenously injected 4 hours before imaging (22). The number of newly sticking neutrophils and extravasated neutrophils was quantitated. All data were analyzed by Imaris software. Circularity was calculated using ImageJ software, where 1.0 represented a complete circle.

Generation of mixed BMC mice

Mixed BMC mice were generated according to established protocols (36). Four weeks after reconstitution, mixed BMC mice were used for experiments. Reconstitution was confirmed by flow cytometry of blood neutrophil counts at 4 weeks after irradiation. BMC mice were used if ≥95% of neutrophils were of donor origin, and WT and CAR-KO neutrophils were present in equal numbers.

Flow cytometry

Neutrophils were isolated from BM, blood, and SF according to established methods (27). Cells were blocked with 2.4G2 anti-FcγRIII/II (BD Biosciences) and stained with the following antibodies: allophycocyanin-conjugated anti-murine Ly6G, fluorescein isothiocyanate (FITC)–conjugated anti-murine CXCR2, phycoerythrin (PE)–conjugated anti-murine C5aR, FITC-conjugated anti-murine CD11a, PE-conjugated anti-murine CD11b (BioLegend), PE-conjugated anti-murine CCR1 (R&D Systems), and biotinylated anti-murine BLT1 (3D7, provided by B.H.). Biotinylated mouse IgG1 (R&D Systems) for BLT1, FITC-conjugated rat IgG2a (BioLegend) for CXCR2 and CD11a, and PE-conjugated rat IgG2a (BioLegend) for C5aR, CD11b, and CCR1 were used as isotype control antibodies. After washing, BLT1 and its isotype control were incubated with PE-conjugated streptavidin (eBioscience). Flow cytometry was performed with FACS LSRFortessa (BD Biosciences) and analyzed with FlowJo software. Neutrophils were identified as Ly6G-positive cells in the granulocyte gate of forward and side scatter plots (fig. S7).

Quantitation of chemoattractants

SF was collected (27) and analyzed by ELISA for levels of CXCL1, CXCL2, CCL3, C5a (R&D Systems), and LTB4 (Thermo Fisher Scientific). Immunomagnetically purified neutrophils from BM, blood, and SF were incubated at 2 × 106/ml in RPMI, 10% fetal calf serum (FCS), 10 mM Hepes, penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C, 5% CO2 for 21 hours. Cell-free cell culture supernatants were assayed for CXCL1, CXCL2, and CCL3 (R&D Systems) by ELISA.

Immunofluorescence staining

Paraffin-embedded tissue sections (4 μm thick) from control and arthritic mice were prepared as described (27). Sections were then blocked with protein block (DakoCytomation) for 15 min and stained with rabbit anti-C5a polyclonal antibody (8 μg/ml; Bioss Antibodies), anti-VWF polyclonal antibody (8 μg/ml; LifeSpan Biosciences), or normal rabbit IgG (Bioss Antibodies) as an isotype control for 60 min at 25°C. Alexa Fluor 488– or Alexa Fluor 555–conjugated donkey anti-rabbit IgG antibodies (Abcam) were used as secondary antibodies and incubated for 1 hour at 25°C. The slides were examined using a fluorescence microscope (Carl Zeiss).

C5a binding to HSPG

HSPG (5 μg/ml; Sigma-Aldrich) was coated on 96-well assay plates (Corning) overnight at 25°C. After washing three times with PBS with Tween 20 (PBST), 1% FCS/PBS was added to each well for 60 min at 25°C to block and then washed three times. C5a (1 nM; R&D Systems) with or without heparin (80 μg/ml; Hospira) was loaded onto the well and incubated for 20 min at 25°C. After washing three times with PBST, C5a detection antibody (200 ng/ml per well; R&D Systems) was incubated for 2 hours at 25°C and washed three times with PBST. Streptavidin-HRP (R&D Systems) was incubated for 20 min at 25°C and washed three times with PBST. Substrate solution (Invitrogen) was loaded onto plate for 20 min, and 2N H2SO4 was used as a stop solution. Absorbance at 450 nm was evaluated with a SpectraMax Plus 384 (Molecular Devices).

In vitro neutrophil adhesion assay

(i) Murine aortic ECs were incubated at a concentration of 2 × 104 per well in Complete Mouse Endothelial Cell Medium with Endothelial Cell Medium Supplement Kit (Cell Biologics) at 37°C, 5% CO2 overnight. The following day, after washing softly three times with sterile PBS, BM neutrophils isolated from naïve WT mice (1 × 105 per well) were loaded onto plates with or without C5a (1 nM) and/or heparin (80 μg/ml) and incubated for 20 min. After removing nonadherent neutrophils by washing three times, the number of adherent neutrophils was enumerated using microscopy. (ii) ICAM-1 (2.5 μg/ml) was coated onto 96-well plates overnight at 4°C. The following day, the plate was washed by sterile PBS three times. BM neutrophils were isolated from naïve WT mice in RPMI, 10% FCS, 10 mM Hepes, penicillin (100 U/ml), and streptomycin (100 μg/ml) and incubated with LFA-1– and/or Mac-1–blocking antibodies or isotype control antibody (40 μg/ml) for 20 min at room temperature. BM neutrophils (1 × 105 per well) were loaded onto plates with or without C5a (1 nM) for 20 min at 37°C. After removing nonadherent neutrophils by washing three times, the number of adherent neutrophils was counted using microscopy.

Analysis of apoptosis

Leukocytes were isolated from the SF of arthritic mice, and neutrophils were purified using EasySep as previously described (48). Neutrophils (1 × 105 per well) were loaded into 96-well plates and incubated with or without CXCL1 (1 nM) and/or CXCL2 (1 nM) for 48 hours. Neutrophils were analyzed for the presence of apoptosis using an Annexin V-FITC Apoptosis Detecting kit (Sigma-Aldrich) and flow cytometry with FACS LSRFortessa.

Statistics

Data are expressed as means ± SEM. All statistical analyses were performed in Prism 5 (GraphPad Software). Means between two groups were compared with unpaired two-tailed Student’s t test. P values less than 0.05 were considered significant.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/2/7/eaaj2195/DC1

Fig. S1. In vivo imaging coadoptively transferred WT and CAR-KO neutrophils into WT mice with full-blown arthritis.

Fig. S2. In vivo imaging demonstrated that C5ar-KO and Ccr1-KO neutrophils detach from the endothelium in the joints of mixed BMC mice.

Fig. S3. Analysis of C5a expression in VWF+ endothelium of (A) aorta and coronary artery, (B) superior vein cava, (C) kidney, and (D) lung from WT mice on day 7 after AST.

Fig. S4. In vivo joint adhesion assay after the application of CA to the exposed joint of WT-LysM-GFP mice (A to D) and Blt1-KO-LysM-GFP- and Ccr1-Cxcr2-DKO-LysM-GFP mice (E to G).

Fig. S5. LFA-1 and Mac-1 expression on neutrophils.

Fig. S6. C5a-induced neutrophil adhesion initiates IC-mediated inflammation.

Fig. S7. Representative gating strategy of Ly6G+ neutrophil.

Table S1. Excel file containing tabulated data for Figs. 1 to 8.

Movie S1. In vivo joint imaging of WT-LysM-GFP and CAR-KO-LysM-GFP mice on day 7 after AST.

Movie S2. Dynamic transition of adoptively transferred WT and CAR-KO neutrophils from arrest to crawling on the endothelium of inflamed joints in WT mice on day 7 after AST.

Movie S3. Detachment of adoptively transferred neutrophils from the endothelium of inflamed joints in WT mice on day 7 after AST.

Movie S4. In vivo adhesion assay: Application of CXCL1, CXCL2, CCL3, C5a, LTB4 (1 nM), or PBS (control) to exposed joints of naïve WT-LysM-GFP mice.

Movie S5. In vivo adhesion assay: β2 integrins mediate C5a-induced neutrophil adhesion on joint endothelium.

Movie S6. In vivo adhesion assay: β2 integrin blockade caused C5a-activated neutrophils to detach from the joint endothelium.

REFERENCES AND NOTES

Acknowledgments: We thank D. Mathis and C. Benoist (Harvard Medical School) for providing KRN mice. Funding: This study was supported by the NIH (R01AI050892 to A.D.L., R01AI097053 to T.R.M., and R01HL125780 to F.W.L.), Rheumatology Research Foundation (A.D.L.), Mallinckrodt Pharmaceutical Research Fellowship Award in Rheumatology Research, Pfizer ASPIRE Rheumatology Award, and Japan Rheumatism Foundation Research Grant (Y.M.). Author contributions: Y.M. designed and performed most experiments, analyzed and interpreted data, and contributed to writing the manuscript. C.M. performed and analyzed flow cytometry, ELISA, and immunohistochemical experiments. T.T.M., E.Y.K., and T.R.M. provided technical assistance for the MP-IVM experiments. G.A.N. and F.W.L. provided technical assistance for in vitro and in vivo adhesion assay. B.H. provided BLT1 antibody. N.D.K. provided K/BxN serum and technical assistance in working with the K/BxN serum transfer model of arthritis. A.D.L. provided overall project supervision, contributed to the design of the experiments, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article