Complement activation on neutrophils initiates endothelial adhesion and extravasation

Molecular Immunology 114 (2019) 629–642

Contents lists available at ScienceDirect

Molecular Immunology journal homepage: www.elsevier.com/locate/molimm

Complement activation on neutrophils initiates endothelial adhesion and extravasation

T

Antonina Akka, Luke E. Springera, Lihua Yangb, Samantha Hamilton-Burdessb, John D. Lambrisc, ⁎⁎ Huimin Yana, Ying Hua, Xiaobo Wua, Dennis E. Hourcadea, Mark J. Millerb, , ⁎ Christine T.N. Phama,d, a

Division of Rheumatology, Department of Medicine, Washington University School of Medicine, Saint Louis, MO, USA Division of Infectious Diseases, Department of Medicine, Washington University School of Medicine, Saint Louis, MO, USA c Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA, USA d John Cochran VA Medical Center, Saint Louis, MO, USA b

ARTICLE INFO

ABSTRACT

Keywords: Complement Neutrophils Fc gamma receptor C5a receptor Inflammation Two-photon microscopy

Neutrophils are essential to the pathogenesis of many inflammatory diseases. In the autoantibody-mediated K/ BxN model of inflammatory arthritis, the alternative pathway (AP) of complement and Fc gamma receptors (FcγRs) are required for disease development while the classical pathway is dispensable. The reason for this differential requirement is unknown. We show that within minutes of K/BxN serum injection complement activation (CA) is detected on circulating neutrophils, as evidenced by cell surface C3 fragment deposition. CA requires the AP factor B and FcγRs but not C4, implying that engagement of FcγRs by autoantibody or immune complexes directly triggers AP C3 convertase assembly. The absence of C5 does not prevent CA on neutrophils but diminishes the upregulation of adhesion molecules. In vivo two-photon microscopy reveals that CA on neutrophils is critical for neutrophil extravasation and generation of C5a at the site of inflammation. C5a stimulates the release of neutrophil proteases, which contribute to the degradation of VE-cadherin, an adherens junction protein that regulates endothelial barrier integrity. C5a receptor antagonism blocks the extracellular release of neutrophil proteases, suppressing VE-cadherin degradation and neutrophil transendothelial migration in vivo. These results elucidate the AP-dependent intravascular neutrophil-endothelial interactions that initiate the inflammatory cascade in this disease model but may be generalizable to neutrophil extravasation in other inflammatory processes.

1. Introduction Neutrophils are innate immune cells that form the first line of defense against invading pathogens. Neutrophil recruitment to specific tissue sites also marks the initiation of a variety of non-infectious inflammatory processes, including arthritis. The transfer of serum from arthritic K/BxN mice into naïve mice leads to a well-characterized

inflammatory arthritis, which is independent of B and T cells but requires contribution from several innate immune mediators. The absolute requirement for neutrophils in the initiation of K/BxN arthritis is well described (Wipke and Allen, 2001; Monach et al., 2010; Sadik et al., 2012). Likewise complement, specifically the alternative pathway (AP) plays an essential role (Ji et al., 2002). Indeed, mice deficient in AP complement proteins such as factor B (FB) are protected against K/

Abbreviations: AAA, abdominal aortic aneurysm; AP, alternative pathway; anti-GPI, anti-glucose-6-phosphate isomerase; ANCA, anti-neutrophil cytoplasmic antibody; Ab, autoantibody; C5aR, C5a receptor; C5aRA, C5a receptor antagonist; CG, cathepsin G; CP, classical pathway; CA, complement activation; FB, factor B; FcγRs, Fc gamma receptors; HI, heat inactivated; HUVEC, human umbilical vein endothelial cell; IC, immune complex; LTB4, leukotriene B4; LP, lectin pathway; MASP, mannose-binding lectin-associated serine protease; MBL, mannose-binding lectin; MMP9, metalloprotease 9; mPAP, mouse peroxidase anti-peroxidase; MPO, myeloperoxidase; NE, neutrophil elastase; P, properdin; PR3, proteinase 3; PRM, pattern recognition molecule; PSGL-1, P-selectin glycoprotein ligand-1; RA, rheumatoid arthritis; 2P, two-photon ⁎ Corresponding author at: Division of Rheumatology, Department of Medicine, Washington University School of Medicine, 660 South Euclid Avenue, Box 8045, Saint Louis, MO, 63110 USA. ⁎⁎ Corresponding author at: Division of Infectious Diseases, Department of Medicine, Washington University School of Medicine, 660 South Euclid Avenue, Box 8051, St. Louis, MO, 63110, USA. E-mail addresses: [email protected] (M.J. Miller), [email protected] (C.T.N. Pham). https://doi.org/10.1016/j.molimm.2019.09.011 Received 29 March 2019; Received in revised form 8 September 2019; Accepted 8 September 2019 Available online 19 September 2019 0161-5890/ © 2019 Elsevier Ltd. All rights reserved.

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

BxN arthritis (Ji et al., 2002). It is generally accepted in this model that anti-glucose-6-phosphate isomerase (anti-GPI) autoantibody (Ab) found in arthrogenic K/BxN serum forms immune complexes that deposit in the joint, leading to intra-articular AP activation, releasing C5a that recruits neutrophils to the joint space (Grant et al., 2002). Why the AP, but not the classical pathway (CP), is required to initiate this effector response is not well understood. Yet others have shown that anti-GPI autoantibody fails to localize to the joints in the absence of neutrophils, suggesting that neutrophils trigger a sequence of intravascular events, which permit subsequent autoantibody entry into the joint space (Wipke et al., 2004). We posit that AP complement-dependent intravascular neutrophil activation promotes endothelial adhesion and neutrophil extravasation, thus initiating the inflammatory cascade.

0.1% BSA in PBS, pelleted and analyzed by FACS analysis using the same antibody panel as in vivo analysis (see above). Flow cytometry was performed on the BD FACSCalibur™. Data analysis was performed using BD CellQuest™ Pro software. 2.4. C5a ELISA Fresh plasma was prepared from collected blood for C5a ELISAbased quantification. Briefly, plates were coated overnight at 4 °C with rat anti-mouse C5a (5 μg/ml) monoclonal antibody (cat#558027, BD Pharmingen, San Jose, CA). After blocking with reagent diluent (1% BSA in PBS) for 1 h at RT, the plates were washed 3x with ELISA wash buffer [0.05% (v/v) Tween 20 in PBS] and incubated with samples (100 μL of same day, freshly obtained plasma diluted 1:20 in reagent diluent) for 1 h at RT. The plates were washed 3x, followed by incubation with biotinylated anti-mouse C5a (500 ng/mL) monoclonal antibody (cat#558028, BD Pharmingen) for 1 h at RT. After washing, the plates were incubated with streptavidin-HRP, 1:200 dilution in reagent diluent (cat# 890803, R&D Systems, Minneapolis, MN) for 30 min, washed and 100 μl of peroxidase substrate (cat# DY999, R&D Systems, Minneapolis, MN) was added. The reaction was stopped with 1 M H2SO4 and OD of samples measured at 450 nm (Molecular devices SpectraMax Plus 384, Molecular Devices, LLC, Sunnyvale, CA).

2. Materials and methods 2.1. Animals All animal experiments were performed in compliance with federal laws and in strict accordance with the guidelines established by the Institutional Animal Care and Use Committee at Washington University. The animal protocol was subjected to annual review and approval by The Animal Studies Committee of Washington University. WT C57BL/6 (cat#00664) and C5-deficient (B10.D2-Hc0H2dH2-T18c/ oSnJ (cat#00461) mice were obtained from the Jackson Laboratory (Bar Harbor, Maine). FB−/− mice in C57BL/6 strain were previously described (Matsumoto et al., 1997). Heterozygous embryos carrying loss-of-function mutation in properdin (Cfptm1Cmst, fully backcrossed to C57BL/6 background) were obtained from the Transgenic Unit of the Division of Biomedical Services at University of Leicester, UK (Stover et al., 2008) and transferred to pseudopregnant Swiss Webster mice. Pups were screened for the presence of the targeted properdin gene. Intercrossing resulted in properdin-deficient hemizygous males (PNull) due to X chromosome linkage of the properdin gene. FcγR-/- mice were a generous gift from Dr. Tanya Mayadas (Brigham and Women’s Hospital) (Ji et al., 2002). All mice were kept under pathogen free conditions at Washington University Specialized Research Facility.

2.5. Two-photon (2 P) microscopy Imaging of mouse paws and neutrophil trafficking was performed as previously described (Wang et al., 2012). Briefly, neutrophils were labeled in vivo with an i.v. injection of 4 μg of PE-conjugated anti-Gr-1 antibody (cat#108408, Biolegend) or anti-Ly6G antibody (cat# 127654, Biolegend, San Diego, CA). Administration of 1–40 μg of PEconjugated anti-Gr-1 antibody or anti-Ly6G antibody does not deplete or interfere with neutrophil trafficking (Yipp and Kubes, 2013; Bucher et al., 2015). Blood vessels were visualized by injecting 50 μl of 10 kD FITC-Dextran (10,000 MW, cat#D1821, Molecular Probes, OR). Mice were anesthetized, placed on a warming pad and their rear paw was secured to the imaging chamber using VetBond (cat #1469SB, 3 M). Time-lapse imaging was performed using a custom-built dual-laser video-rate 2 P microscope. PE-labeled neutrophils and FTIC-dextran were excited by a Chameleon Vision II Ti:Sapphire laser (Coherent) tuned to 910 nm. Fluorescence emission was detected as red (> 560 nm) and green (495–560 nm). Each plane represents an image of 200 μm (x) by 220 μm (y) at 2 pixels/μm. Z-stacks were acquired by taking 21 sequential steps at 2.5 μm spacing. Image reconstruction and multidimensional rendering was performed with Imaris (Bitplane). Enumeration of rolling, crawling, and extravasating neutrophils could not be conducted in blinded manner due to the very different behavior of WT vs. FB−/− neutrophils. The C5aRA, JPE1375 [Hoo-Phe-Orn-Prohle-Pff-Phe-NH2] (1 mg/kg in 0.2 ml PBS/mouse/injection) (Schnatbaum et al., 2006) and control peptide (AcF[OpdChaAdr]) (Strey et al., 2003) were administered i.v. 10 min prior to s.c. injection of K/BxN serum into the hindfoot pad. JPE1375 was used for in vivo C5aR blockade as it appears to be more specific for mouse C5aR (Schnatbaum et al., 2006). PMX53 (Ac-Phe-[Orn-Pro-cha-Trp-Arg]) was used to block in vitro NET formation by human neutrophils.

2.2. In vivo administration of K/BxN serum Male and female mice 6–10 week old were injected i.v. with 400 μl of K/BxN serum at T = 0 and blood was obtained at 10 or 20 min for analysis by flow cytometry with the following antibodies: anti-C3 (cat# 0855500, MP Biomedical Life Sciences), anti-Gr1 (cat# 108425; BioLegend); anti-Ly6G (cat# 127654, Biolegend, San Diego, CA); antiCD11b (cat# 101212; BioLegend), anti-CD 11a (cat#101008, Biolegend, San Diego, CA); anti-CD162 (PSGL-1, cat# 555306; BD Pharmingen). In general, 106 cells were blocked with the anti-FcR mAb 2.4G2, stained with the indicated Abs for 20 min at 4 °C and then washed and resuspended for FACS analysis. Flow cytometry was performed on the BD FACSCalibur™. Data analysis was performed using BD CellQuest™ Pro software. For two-photon (2 P) microscopy mice were injected with 200 μl of K/BxN serum at T = 0 and 24 h later they were boosted with a s.c. injection of 40 μl of K/BxN serum into one hind foot as previously described (Wang et al., 2012). The animals were then subjected to 2 P microscopy between 15–60 min after s.c. injection.

2.6. Immunofluorescence of mouse paws with acute arthritis

2.3. Ex vivo whole blood stimulation with K/BxN serum

The frozen sections (9 μm) of harvested paws were fixed in 4% paraformaldehyde for 20 min at room temperature, blocked with 8% BSA for 1 h at RT, incubated with biotinylated anti-Gr-1 antibody overnight at 4 °C (1:100 dilution, cat# 108404, Biolegend, San Diego, CA) followed by rhodamine red-conjugated streptavidin for 1 h at RT (1:360 dilution, cat# 016-290-084, Jackson ImmunoResearch Inc., West Baltimore, PA). For F-actin visualization slides were permeabilized with 0.1% Triton-X-100 and incubated with 5 mU/ml AlexaFluor

Blood was collected from the inferior vena cava into collection tubes containing PPACK (cat# 520222, EMD Millipore, Billerica MA) as anticoagulant at final concentration of 93 μM. In some experiments peripheral blood was stimulated with heat inactivated (56 °C for 30 min) K/BxN serum (20% v/v) or soluble mouse peroxidase anti-peroxidase (mPAP) immune complexes (Sigma) at different concentrations ex vivo for 15 min at RT. Red blood cells were lysed, cells were washed with 630

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

488 phalloidin (cat# A12379, Molecular Probes, Eugene, OR) for 30 min, washed 3x with PBS, and mounted with Vectorshield, cat# H1000 (Vector laboratories, Burlingame, CA) with 300 nM of DAPI, cat# D3571 (Molecular Probes, Eugene OR). Images were visualized by epifluorescence on a Leica DM 2000 LED microscope and acquired with a Leica DMC 4500 camera using Leica Application Suite or LAS X software. F-actin disruption per visualized medium blood vessel was scored on a scale of 1–4, 1 ≤ 25% disrupted, 2 = 25–50% disruption, 3 = 50–75% disruption, 4 ≥ 75% disruption. Density of microvascular F-actin fluorescent peaks was calculated using ImageJ software (http:// rsb.info.nih.gov/ij). At least 10 blood vessels were counted per paws, n = 3 animals per treatment group. Quantitative scoring was performed in a blinded fashion by two independent observers. For in vivo neutrophil degranulation and VE-cadherin degradation, paw sections were fixed and blocked as above then incubated with biotinylated anti-mouse MPO (1:100 dilution, cat# HM1051BT, Hycult Biotech, Plymouth Meeting, PA) plus either anti-neutrophil elastase (M18) (1:100 dilution, cat# sc-9521, Santa Cruz Biotechnology, Dallas, TX) or anti-VE-cadherin (1:100 dilution, cat# 555289, BD Biosciences, San Jose, CA) antibody overnight at 4 °C followed by Streptavidin-FITC (1:100 dilution, cat# 7100-02, Southern Biotech, Birmingham, AL) and the corresponding secondary antibody (1:100 dilution, cat# 705-025147, cat# 712-295-153, Jackson ImmunoResearch Inc., West Baltimore, PA) for 1 h at RT, washed 3x with PBS and mounted with Vectorshield (cat# H-1000, Vector laboratories, Burlingame, CA) with 300 nM of DAPI (cat# D3571, Molecular Probes, Eugene OR). Epifluorescent images were visualized on a Leica DM 2000 LED microscope and acquired with a Leica DMC 4500 camera using Leica Application Suite or LAS X software. Confocal microscopy was performed on a Zeiss LSM 880 Confocal Laser Scanning Microscope and images acquired/analyzed with software ZEN.

2.8. Neutrophil isolation Neutrophils were isolated from the peripheral blood of healthy volunteers (n = 3) using Ficoll- Paque PLUS (GE Healthcare Biosciences, Piscataway, NJ) centrifugation. The lower phase was treated with erythrocyte lysis buffer solution (0.01 M HEPES, 2.5 mM CaCl2, 0.04% BSA in HBSS), washed and enumerated. Purity of neutrophils was greater than 95%, as assessed by flow cytometry. Mouse neutrophils were isolated from bone marrow by discontinuous Percoll gradient as previously described (Adkison et al., 2002). 2.9. Isolation of C5a-induced degranulated neutrophil contents Human neutrophils were primed with 250 ng/ml human GM-CSF and stimulated with 250 ng/ml human C5a and cultured for 4 h in 6 well plates (5 × 106 cells per well). Wells were washed twice and 10 U/ ml of micrococcal nuclease (Thermo Fisher Scientific, Waltham, MA) was added for 20 min to the cells to release cell-bound degranulated contents. Supernatants containing degranulated neutrophil contents were harvested and added to endothelial cell culture. 2.10. Endothelial cell culture Human umbilical vein endothelial cells (HUVECs) were cultured in flasks coated with Quick Coating Solution (cat# cAP-01, AngioProteomie, Boston, MA) supplemented with Endothelial Cell Growth Medium (cat# 50306188, Thermo Fisher Scientific, Waltham, MA). HUVEC passages 6–9 were used for all in vitro experiments. 2.11. Effects of C5a-induced degranulated neutrophil contents on HUVEC monolayer HUVECs were seeded at low density on coated with Quick Coating Solution (cat# cAP-01 Angio-Proteomie, Boston, MA) in Chamber Slide Systems (cat# 154917 Thermo Fisher Scientific, Waltham, MA) to confluence. Monolayers were treated with C5a-induced degranulated neutrophil contents (30 ng) for 5, 15 or 30 min at 37 °C in a humidified atmosphere of 5% CO2. Cells were fixed with 2% paraformaldehyde and washed 3 times with PBS. For immunofluorescence the chamber slides were blocked with 0.3% Triton X-100 (Thermo Fisher Scientific, Waltham, MA) with the addition of 10% goat serum (Sigma, Saint Louis MO) for 1 h at RT followed by incubation with anti-human VE-Cadherin antibody (1:100 dilution, cat# MAB9381, Bio-Techne Corporation, Minneapolis, MN) in a dilution buffer (PBS with 1% BSA, 1% goat serum and 0.3% Triton X-100) for 48–72 h at 4 °C. The slides were washed extensively and incubated with rhodamine red goat anti-mouse IgG (1:100 dilution, cat#115-2905-164, Jackson ImmunoResearch Inc., West Baltimore, PA) for 1 h at RT. After washing AlexaFluor 488 phalloidin (5 mU/ml, cat# A12379, Molecular Probes, Eugene, OR) in PBS was added to the chamber slides for 30 min at RT, washed and mounted with Vectorshield (cat# H-1000, Vector laboratories, Burlingame, CA) with 300 nM of DAPI (cat# D3571 Molecular Probes, Eugene OR). All images were visualized on a Leica DM 2000 LED microscope and acquired with a Leica DMC 4500 camera using Leica Application Suite or LAS X software. HUVEC monolayer integrity was evaluated using ImageJ. Gaps in F-actin staining due to VE-cadherin disassembly and cell retraction were assessed using the region of interest (ROI) manager and measurements for number of pixels. To calculate percent of gap area the number of pixels covering the gaps was divided by the total number of pixels in a selected area of HUVEC monolayer.

2.7. Immunofluorescence of peripheral blood cells For anti-GPI staining, peripheral blood was obtained from the inferior vena cava; red blood cells were lysed; cells were washed and cytospin samples prepared. Samples were fixed with 4% PFA for 10 min at RT, blocked with anti-FcR mAb 2.4G2 (1:100) and 5% normal mouse serum in 5% BSA in PBS for 1 h at RT. Anti GPI (1:100 dilution, cat#P06744, Biorbyt), anti-Gr1 (1:100, cat#553127, BD Pharmingen) were added to cells overnight at 4 °C, followed by addition of antirabbit IgG (1:200, cat#711-025-152, Jackson ImmunoReseach), antiF4/80 (1:100 dilution, cat#123106, Biolegend), anti-CD19 (1:100 dilution, cat#09654D, BD Pharmingen), or anti-CD3 (1:100 dilution, BD Pharmingen) for 1 h at RT. Vectorshield with DAPI (cat#H-1200, Vector Laboratories) was used to stain nuclei. All images were acquired with a Leica DMC 4500 camera using Leica Application Suite or LAS X software. In separate experiments, blood was collected into tubes with PPACK (cat# 520222, EMD Millipore,) as anticoagulant at final concentration of 93 μM. Peripheral blood was stimulated ex vivo with HI K/BxN or normal mouse serums (20% v/v) for 10 min at RT. After stimulation cells were fixed with methanol for 30 min at −20 °C and red blood cells were lysed. Cytospin samples were blocked with the anti-FcR mAb (1:100 dilution, cat#101302, BioLegend) in 5% BSA in PBS buffer for 1 h at RT and probed with anti-mouse IgG (1:100 dilution, cat#115295-164, Jackson ImmunoReseach) and biotinylated anti-mouse myeloperoxidase (MPO, 1:100 dilution, cat#S22223, Hycult Biotech) and for 2 h at RT followed by streptavidin Alexa Fluor 488. Vectorshield with DAPI (cat#H-1200, Vector Laboratories) was used to stain nuclei. Note: anti-MPO was used since we were unable to detect neutrophils with anti-Gr-1 antibody after K/BxN serum activation ex vivo and cell fixation.

2.12. In vitro effects of C5a-induced degranulated neutrophil contents on VE-cadherin HUVECs were grown to confluence in 24 well plates and incubated 631

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

with C5a-induced degranulated neutrophil contents (30 ng) for 5, 15 and 30 min or C5a-induced degranulated neutrophil contents pretreated with 10 μM cathepsin-G inhibitor (cat# J65957, Alfa Aesar, Ward Hill, MA), 12.5 μM metalloproteinase-9 inhibitor (cat# SB-3CT, Selleckchem, Houston, TX), 10 μM elastase IV inhibitor (cat# 14922, Cayman Chemical, Ann Arbor, MI), 20 μM aprotinin, a PR3 inhibitor (cat#191158, MP Biomedicals, Solon, OH) or a combination of all 4 inhibitors for 10 min at RT. After 30 min, supernatants were discarded and cells were lysed directly in SDS sample buffer transferred to tubes, boiled for 10 min and stored at −20 °C. For Western blot analysis, equivalent amounts of boiled samples in SDS sample buffer were fractionated on an 8% SDS-PAGE under reducing conditions, transferred to PVDF membrane (Millipore, Burlington, MA) and blocked with 5% non-fat milk. The membrane was probed with anti-human VE-Cadherin (1 μg/ml, cat#MAB9381 R&D Systems, MN). Actin (1:3,000, Cat # SC-1615, Santa Cruz) served as control for quantity and quality of protein. Appropriate HRP-conjugated secondary antibodies were used from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA).

Detection of CA on neutrophils in vivo was transient, as neutrophils decorated with C3 fragment were no longer detected in the circulation after 20 min (Figs. 1A and S1). Activated neutrophils opsonized with complement fragments likely leave the blood vessels to translocate to the site of inflammation. In contrast, the level of C3 fragment deposition on neutrophils activated in vitro remained high, in the absence of clearance or transmigration (Fig. S2). CA led to activation of neutrophils, with enhanced CD11a and CD11b expression (Fig. 1A) and a small but detectable decrease in the expression of P-selectin glycoprotein ligand-1 (PSGL-1, CD162), the counter receptor for P-selectin expressed on activated endothelial cells (Fig. 1A). Loss of PSGL-1 from the surface of neutrophils is thought to be mediated by neutrophil serine proteases released from primary granules, providing additional evidence of neutrophil activation and degranulation (Gardiner et al., 2001). Upregulation of CD11a/CD11b persisted at 20 min (Fig. 1A) but was no longer apparent at 24 h (Fig. S3A). C3 fragment deposition on neutrophils required the AP factor B while C4 was dispensable (Fig. 1B), suggesting that neutrophils can directly trigger the AP in the absence of the CP. CA on the surface of neutrophils requires FcγRs, as the absence of the Fc receptor common gamma chain (γc) abrogated C3 fragment deposition (Fig. 1B). This requirement is consistent with previous studies showing that mice lacking γc were protected against K/BxN serum transfer arthritis (Ji et al., 2002; Corr and Crain, 2002). Lastly, we found that the absence of C5 did not prevent CA on neutrophils but diminished the upregulation of adhesion molecules (Fig. 1B and S3B). C3 fragment deposition on C5-deficient neutrophils was higher than WT, suggesting a delay in peripheral clearance of these opsonized cells (Fig. 1B).

2.13. Statistical analysis Comparisons between groups were performed by two-tailed, unpaired t test without correction. Grading data was analyzed with χ2 test. Comparisons between multiple groups (≥3) were performed by oneway ANOVA. Equality of variance assumption was tested and Bonferroni’s correction for multiple comparisons was performed. P < 0.05 between experimental groups was considered significant. 2.14. Study approvals

3.2. Complement activation on neutrophils leads to C5a generation

All animal experiments were performed in compliance with federal laws and in strict accordance with the guidelines established by the Institutional Animal Care and Use Committee at Washington University. The animal protocol is subjected to annual review and approval. De-identified human neutrophils were obtained from healthy donors who consented to donate blood to the Division of Rheumatology through a protocol approved by the Washington University Institutional Review Board (IRB ID#201805088).

C5a is a soluble complement activation product known to promote neutrophil activation (Blatt et al., 2016) through its engagement with the neutrophil C5a receptor (C5aR) (Monk et al., 2007). We first examined whether CA by K/BxN serum administration leads to the generation of C5a in vivo but were unable to detect significant levels of C5a systemically (< 3 pg/ml). We reasoned that C5a was likely generated at high levels locally but was rapidly removed (within 2 min) from the circulation (Webster et al., 1982). We therefore turned to an ex vivo assay that allowed the accumulation of C5a, using whole blood incubated with heat inactivated (HI) K/BxN serum, to remove the intrinsic complement activity of the serum, especially that of FB. Ex vivo whole blood stimulation with HI K/BxN serum led to CA, as evidenced by C3 fragment deposition (Fig. 2A) and generated a significant amount of C5a from WT and C4−/− whole blood, but not FB−/− or FcγR−/− (Fig. 2B). Properdin (P) (Blatt et al., 2016), a component of the AP convertase that lends the complex stability (Fearon and Austen, 1975), has been shown to play an important role in K/BxN serum-induced arthritis (Kimura et al., 2010). Prompted by a recent study that suggested C5a production in human whole blood is mediated by properdin (Blatt et al., 2016; Saggu et al., 2013) we asked whether properdin was also required for K/BxN serum-mediated C5a production. P-deficient whole blood was incubated with HI K/BxN serum and level of C3 fragment deposition on neutrophils and C5a production were examined. Absence of properdin did not prevent C3 fragment deposition on neutrophils (Fig. 2A) but largely abrogated C5a generation (Fig. 2B). These results suggest that C3 fragment deposition is robust and proceeds quickly without or with the residual properdin present in HI K/ BxN serum; however, a higher level of P is required for C5a generation, likely due to the slow turnover of C5 by the AP C5 convertase (Pangburn and Rawal, 2002). Indeed, the addition of WT serum (33% v/v) to P-deficient whole blood led to significantly higher production of C5a following incubation with HI K/BxN serum (Fig. S4). K/BxN serummediated C5a generation in whole blood also required FcγR (Fig. 2B).

2.15. Data availability Data generated in this current study are available from the corresponding authors on reasonable request. 3. Results 3.1. K/BxN serum transfer triggers complement activation on the surface of neutrophils K/BxN serum transfer arthritis is a well-established rodent model that recapitulates key features of the effector phase of rheumatoid arthritis (RA) observed in humans (Kyburz and Corr, 2003). While arthritis development in this model is independent of adaptive immunity, innate immune effectors (including neutrophils and complement) are critically involved. Neutrophil depletion (Wipke and Allen, 2001) or absence of complement AP FB completely abrogates disease manifestation while the complement classical pathway (CP) is dispensable (Ji et al., 2002). How the AP and neutrophils orchestrate the inflammatory cascade following K/BxN serum transfer is unclear. We posit that the pathogenic anti-GPI autoantibody (Ab) in K/BxN serum is capable of directly activating neutrophils, triggering “in situ” CA on the cell surface. To test this possibility, we administered K/BxN serum i.v. to naïve WT mice and showed that within minutes of K/BxN serum injection CA was detected on circulating (Gr1high or Ly-6Ghigh) neutrophils, as evidenced by cell surface deposition of C3 fragment (Figs. 1A and S1). 632

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 1. In vivo K/BxN serum administration triggers intravascular complement activation. A) Wild type (WT) C57BL/6 mice were injected i.v. with K/BxN serum. After 10 or 20 min, blood was collected and white blood cells stained for C3 fragment deposition and CD11b, CD11a, PSGL-1 expression on Gr-1high cells. Quantitative assessment of C3, CD11b, CD11a, and PSGL-1 expression level on Gr-1high/Ly-6Ghigh cells is presented in Figure S1. B) In other experiments, C4−/−, FB−/−, FcγR−/−, and C5null mice were injected with K/BxN serum i.v. for 10 min, blood was collected and stained for C3 fragment deposition on Gr-1high cells. Histograms are representative of n = 4–9 animals per genotype.

3.3. Complement activation by soluble ICs is CP-dependent

dose-dependent manner (Fig. S5). mPAP also led to CD11b (but not CD11a) upregulation (Fig. S5). Unexpectedly, we found that mPAP failed to activate complement in C4−/− as well as FB−/− whole blood (Fig. S5). Moreover, mPAP-mediated C5a generation was blunted in both C4−/− and FB−/− whole blood (Fig. S5). These results suggest that CA by irrelevant ICs is CP- and dose-dependent. In contrast, the requirement for the AP in anti-GPI Ab-mediated CA is reminiscent of the mechanism by which anti-neutrophil cytoplasmic antibody (ANCA) causes AP-dependent vasculitis (Xiao et al., 2007). We hypothesized

Previous studies suggest that anti-GPI Ab encounters antigen in the circulation, forming immune complexes (ICs) that engage FcγRs on neutrophils (Wipke et al., 2004), triggering CA. If this were the case then we posit that irrelevant, pre-formed ICs would also trigger CA on neutrophils. To this end, we incubated soluble preformed mPAP (Wipke et al., 2004) with whole blood ex vivo and examined neutrophils for surface CA. mPAP triggered C3 fragment deposition on neutrophils in a 633

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 2. K/BxN serum activates the alternative pathway of complement leading to C5a production ex vivo. A) Whole blood from different genotypes (WT, C4−/−, FB−/−, Pnull, and FcγR−/−) was obtained and stimulated ex vivo with heat inactivated (HI) K/BxN serum for 15 min and examined for C3 fragment deposition on Gr-1high neutrophils. Histograms are representative of at least 3 independent experiments. B) Plasma from stimulated whole blood was assayed for C5a. Values represent mean ± SEM, n = 3 mice per genotype. Statistical analysis was performed using oneway ANOVA.

Fig. 3. Anti-GPI Ab recognizes an antigen on neutrophil cell surface. A) Peripheral white blood cells were stained for GPI (red) and Gr-1 (green, neutrophils), F4/80 (green, monocytes), CD19 (green, B cells), or CD3 (green, T cells). Colocalization appears yellow (see also Figure S6). Scale bar = 10 μm. B) Whole blood was stimulated with normal serum or HI K/BxN serum. Red blood cells were lysed and white blood cells were stained for MPO (green) and mouse IgG (red). DAPI stains nuclei blue. Scale bar = 5 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

that anti-GPI Ab may directly recognize an antigen expressed on the surface of neutrophils. Indeed, we found that neutrophils abundantly express an antigen that is recognized by a polyclonal rabbit anti-GPI Ab (Figs. 3A and S6). Moreover, IgG in K/BxN serum also binds to an

antigen present on neutrophil surface (Fig. 3B). These results suggest that anti-GPI Ab may initiate CA as soluble ICs that engage FcγRs or, alternatively, by directly binding to surface neutrophils to trigger CA, or a combination of both processes. 634

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

3.4. AP activation promotes neutrophil adhesion and transmigration

use a transcellular route (Choi et al., 2009). For the paracellular pathway, endothelial adherens junctions represent a major barrier. Inhibition of VE-cadherin leads to actin cytoskeleton reorganization (Hordijk et al., 1999), increases vascular permeability and accelerates the rate of neutrophil extravasation in vivo, suggesting that adherens junctions formed by VE-cadherin regulate barrier integrity and act as gatekeepers for neutrophil transendothelial migration (Gotsch et al., 1997). Moreover, in vitro studies suggest that neutrophils transmigrate through VE-cadherin gaps formed during diapedesis (Shaw et al., 2001). How neutrophil adherence to the endothelium leads to VEcadherin loss and gap formation is unclear. We hypothesize that activated neutrophils release extracellular proteases that contribute to adherens junction disruption and increased endothelial permeability. To test this possibility, we stimulated human neutrophils with recombinant C5a and exposed human umbilical vein endothelial cells (HUVECs) to C5a-induced degranulated neutrophil contents. Exposure of HUVECs to degranulated neutrophil contents led to the time-dependent appearance of intercellular gaps (Figs. 7A, B and S7A) and loss of VE-cadherin (Figs. 7A and S7A, B). Treatment of HUVECs with neutrophil degranulated contents also led to VE-cadherin degradation (Figs. 7C and S7B). Heat inactivation of degranulated contents blocked VE-cadherin degradation (Fig. S7C), suggesting that proteases released from neutrophil degranulation were responsible for the degradation process. To identify which neutrophil-derived proteases (neutrophil elastase [NE], cathepsin G [CG], proteinase 3 [PR3], or metalloprotease 9 [MMP9]) were responsible for VE-cadherin degradation, we tested specific inhibitors to these enzymes. NE inhibitor greatly attenuated VE-cadherin degradation while a mixture of all 4 inhibitors completely blocked proteolytic cleavage (Fig. 7D). These results support the hypothesis that proteases released from C5a-activated neutrophils degrade VE-cadherin, inducing gaps and increased vascular leakage. To determine the role of C5a in endothelial barrier function in vivo, we returned to the accelerated model of K/BxN arthritis. In the paws of animals receiving s.c. K/BxN serum and treated with the control peptide, an abundance of neutrophils adherent to the endothelial surface displayed extracellular myeloperoxidase (MPO) and NE (Figs. 8 and S8), as evidence of degranulation. We also found that in vivo degranulated neutrophils (displaying prominent extracellular MPO staining) were associated with areas of VE-cadherin loss (Fig. 8). In contrast, treatment with C5aRA limited degranulation and preserved the integrity of endothelial VE-cadherin (Fig. 8). These findings corroborate the in vitro data that extracellular granule proteases released from activated neutrophils interacting with the endothelium affect the integrity of cell-cell adherens junction. To further quantify the effects of adherent neutrophils at the endothelial level, we sectioned mouse paws and performed F-actin staining to examine for gap formation in the endothelial layer, as evidence of adherens junction disruption. Intravascular neutrophils (stained with Gr-1) were observed in both control peptide- and C5aRAtreated animals (Fig. 9A), consistent with the in vivo 2 P microscopy observations (Fig. 6). In animals administered the control peptide, abundant neutrophils were observed adherent to the endothelium, in various stages of diapedesis, and in the perivascular space and tissue parenchyma (Fig. 9A). In contrast, the majority of neutrophils in the C5aRA treated group were located in the vessel lumen (Fig. 9A). In addition, we observed many breaks and gaps in the endothelial F-actin cytoskeleton, especially in areas of neutrophil extravasation (Fig. 9A and B). In contrast, the integrity of endothelial cytoskeleton was significantly preserved in the C5aRA-treated group (Fig. 9A and B). Disruption of cytoskeletal spatial ordering was observed as a decrease in the density of microvascular F-actin fluorescent peaks (Fig. 9C and D). Taken together, these findings suggest a high degree of endothelial barrier disruption.

We next evaluated whether AP activation on neutrophils promoted their recruitment by performing in vivo two-photon (2 P) microscopy and analyzing single-cell neutrophil dynamics during the first 1 h after K/BxN serum transfer. We used a modified accelerated model of K/BxN arthritis in which mice were administered serum i.v. followed 24 h later by subcutaneous (s.c.) injection of a smaller volume of serum into one hind footpad immediately before imaging (Wang et al., 2012). This s.c. boost acutely activates complement and upregulates adhesion molecules on circulating neutrophils, producing a synchronized and robust recruitment within minutes of injection and thus facilitating the quantitative analysis of in vivo neutrophil trafficking behavior. Neutrophil rolling and intravascular crawling were measured in medium size vessels in the plantar tissue of the digits in the injected paw. Neutrophil rolling involves the interaction of PSGL-1 with its counterreceptor on endothelial cells (Muller, 2013). The binding of activated neutrophils to vascular cell adhesion molecules (i.e. intercellular adhesion molecule 1, ICAM-1) via activated integrins such as CD11a leads to neutrophil arrest and firm adhesion on the endothelium while CD11b mediates intravascular crawling prior to neutrophil transmigration (Ley et al., 2007). Within 15 min of s.c. K/BxN serum injection we observed a significant number of FB−/− neutrophils rolling along the surface of the endothelium (Fig. 4A and B, Video 1). This result was expected in view of the normal surface expression of PSGL-1 on FB−/− neutrophils (Fig. S1). While WT neutrophils also showed rolling behavior at 15 min, a significant number of these had arrested on the endothelium and were observed crawling along the intraluminal surface (Fig. 4A and B, Video 2). These results are consistent with the upregulation of CD11a/CD11b (Fig. 1A). In addition, arrested WT neutrophils displayed a "flattened" polarized morphology as compared to the spherical shape of the rolling FB−/− neutrophils (Fig. 4A and B). Within the first hour of s.c. K/BxN serum injection, a significant number of WT neutrophils were seen extravasating into the peri- and extra-vascular space (Fig. 5A and B). In contrast the number of extravascular FB−/− neutrophils was markedly reduced (Fig. 5A and B). These results suggest that AP activation is required for efficient neutrophil arrest and intraluminal crawling followed by subsequent transmigration across the endothelium. 3.5. Blockade of C5a-C5aR interaction mitigates neutrophil transmigration In addition to promoting neutrophil activation, C5a has also been implicated in neutrophil transmigration (DiScipio et al., 1999; DiScipio et al., 2006) but the precise mechanism by which C5a binding to its receptor promotes leukocyte transmigration is poorly understood. To further elucidate the contribution of C5a-C5aR in neutrophil transmigration in vivo, we administered a C5a receptor antagonist (C5aRA) (Schnatbaum et al., 2006) or a control peptide 10 min prior to s.c. K/ BxN serum injection into the footpad of WT mice. Neutrophil transmigration was assessed 60 min after serum injection by 2 P microscopy. Although, neutrophil trafficking to vessels at the inflamed site was observed in both control peptide and C5aRA treated groups, C5aRA administration significantly attenuated neutrophil extravasation in WT mice as compared to control peptide (Fig. 6A and B). 3.6. C5a promotes the release of extracellular proteases that degrade adherens junctions Neutrophil recruitment is a multistep process that begins with rolling interactions along the endothelium, followed by arrest/firm adhesion, intravascular crawling and finally transmigration across the endothelium. Although the steps leading to neutrophil arrest/firm adhesion and the molecules involved are well described, the exact mechanisms that underlie neutrophil transendothelial migration are still unclear. Many studies suggest that neutrophils exit the blood vessels predominantly through a paracellular route, while 5% of neutrophils 635

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 4. Alternative pathway complement activation modulates neutrophil trafficking behavior at site of inflammation. A) Two-photon (2 P) microscopy in the accelerated model of K/BxN arthritis. Within 15 min of s.c. K/BxN serum injection, a number of WT neutrophils were seen rolling, crawling then transmigrating (see supplemental video). Firmly adherent WT neutrophils exhibited a flattened morphology (seen at higher magnification in the far right, upper panel). L = length; W = width. In contrast, most FB−/− neutrophils retained a rounded morphology (far right, lower panel and supplemental video). The yellow arrow in the upper panels (WT) follows the same neutrophil after it transmigrated across the endothelial layer and meandered through the extracellular matrix. Red = neutrophils; green = FITC dextran; blue = collagen rich extracellular matrix. Note the disappearance of FITC dextran over time in WT animal indicating increased vascular permeability and enhanced clearance of the dextran. In contrast, FITC dextran outlining the blood vessel persisted in FB−/− animal. Scale bar = 20 μm. B) Enumeration of rolling (> 20 μm/min), crawling (< 10 μm/min), transmigrated neutrophils and neutrophil morphology (length:width ratio) in WT and FB−/− mice. At least 20 neutrophils were counted in 5 medium size blood vessels (20–50 μm in diameter) per animals and results presented as mean ± SEM (n = 3 animals per genotype). Statistical significance between phenotypes was determined by student’s t-test.

4. Discussion

dependent intravascular neutrophil-endothelial interactions that initiate the inflammatory cascade in K/BxN arthritis. GPI is a ubiquitous cytoplasmic enzyme that elicits joint-specific disease (Korganow et al., 1999). Extracellular GPI deposits have been found lining the surface of cartilage in normal mouse joint and human arthritic joints (Matsumoto et al., 2002), leading to the hypothesis that GPI:anti-GPI complexes on articular surface initiate the inflammatory cascade. However, GPI expression on circulating peripheral cells has not been described. Our finding that GPI or GPI-like antigen is expressed on neutrophil surface is intriguing. How GPI or GPI-like antigen is presented on the surface of neutrophils remains to be determined. The C3 and C5 convertases are multicomponent serine proteases that cleave C3 and C5, respectively. While the CP and lectin pathway (LP) C3 convertase requires C2 and C4 (C4bC2b), the AP convertase is generated through cleavage of C3 and FB (Ricklin et al., 2010). The

Neutrophils and the complement AP have been implicated in experimental inflammatory arthritis; however, the requirement for the AP remains unexplained. In other disease models, such as the anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis, autoantibodies have been shown to directly activate neutrophils (Xiao et al., 2007). By analogy and from our current findings we propose that in the K/BxN arthritis model, anti-GPI autoantibodies in the serum may engage GPI or a GPI-like antigen, directly triggering the AP on neutrophils in an Fcγ-dependent fashion. AP activation leads to local C5a production and stimulates the release of neutrophil-associated proteases, which contribute to the degradation of VE-cadherin, an adherens junction protein, promoting endothelial barrier dysfunction and enhancing vascular permeability. These results elucidate the AP636

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 5. Alternative pathway complement activation promotes neutrophil transmigration. A) Acute inflammation was induced with s.c. injection of K/BxN serum. One hour after serum injection, 2 P microscopy was performed on the inflamed paws to enumerate intravascular (white arrow), perivascular (green arrow) and extravascular (yellow arrow) neutrophils. Scale bar = 20 μm. Lower panels represent higher magnification. Red = neutrophils; green = FITC dextran; blue = collagen rich extracellular matrix. B) Enumerated number of intravascular, perivascular and extravascular neutrophils in WT versus FB−/− mice. Neutrophils were visualized and counted in at least 10 medium size blood vessels (20–50 μm in diameter) per animals and results presented as mean ± SEM (n = 3 animals per genotype). Statistical analysis was performed by one-way ANOVA.

same holds true for C5 convertase. Properdin is a component of both the C3 and C5 AP convertases that serves to stabilize these otherwise shortlived complexes (Fearon and Austen, 1975) and in some circumstances can provide a platform for de novo convertase assembly (Kemper and Hourcade, 2008). Stimulated neutrophils have been shown to secrete properdin upon stimulation (Wirthmueller et al., 1997) although in the K/BxN serum transfer model, plasma properdin alone appears sufficient to sustain pathogenesis (Kimura et al., 2010). It has been shown that C3b, which can be spontaneously generated in the circulation by AP activation through a “tick-over” process (Pangburn et al., 1981), forms a C3b-IgG complex that exhibits increased affinity for FcγRs (Malbran et al., 1987) and serves as a powerful precursor of the AP C3 convertase, especially in the presence of properdin (Jelezarova et al., 2000). We posit that in a microenvironment containing ICs (i.e. GPI:anti-GPI), this nascent AP C3 convertase is protected from surface complement regulatory proteins and serves as nidus to focus AP activation on neutrophils, independent of the classical pathway C4. Once activated, the AP can further maintain and amplify neutrophil inflammatory

responses via release of C5a. Herein, we show that properdin is indeed required for C5a production following K/BxN serum activation of neutrophils. Interestingly, properdin was not necessary for C3 fragment deposition, possibly because CA on activated neutrophils is robust enough to support C3 fragment deposition even with the short-lived Pdeficient C3 convertases. Alternatively, there was enough residual functional properdin in the K/BxN serum to support C3 fragment deposition at near WT levels. In general, the assembly / function of C5 convertase, formed by the addition of a C3b subunit to a C3 convertase, is regarded as relatively inefficient compared with that of C3 convertase (Pangburn and Rawal, 2002), which could make the stability conferred by properdin to the C5 convertase particularly critical for C5a generation. Although the role of FcγRs in the K/BxN serum transfer arthritis model is well established and mice lacking the common γc are protected against disease development (Ji et al., 2002; Corr and Crain, 2002), the exact receptor(s) engaged on neutrophils are still undetermined. While FcγR−/− mice are completely resistant to K/BxN 637

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 6. C5a receptor antagonist (C5aRA) blocks neutrophil transmigration. A) WT mice were administered C5aRA or control (Ctrl) peptide 10 min prior to s.c. K/BxN serum injection. Neutrophil trafficking to the inflamed paw was assessed by 2 P microscopy 60 min after K/BxN injection. White arrow = intravascular neutrophils; green arrow = perivascular neutrophils; yellow arrow = extravascular neutrophils. Red = neutrophils; green = FITC dextran; blue = collagen rich extracellular matrix. Scale bar = 20 μm. B) The number of intravascular, perivascular and extravascular neutrophils in control peptide or C5aRA treated animals was enumerated. Neutrophils were visualized and counted in at least 10 medium size blood vessels (20–50 μm in diameter) per animal; results are presented as mean ± SEM (n = 3 animals per treatment condition). Statistical analysis was performed using one-way ANOVA.

serum transfer arthritis, FcγRI and FcγRIII are more or less dispensable (Monach et al., 2010; Ji et al., 2002). Others found that FcγRIV is sufficient for K/BxN serum transfer arthritis (Mancardi et al., 2011). FcγR ligation can enhance or attenuate C5aR-mediated functions, depending on whether ICs engage the activating or inhibitory FcγRs (Karsten and Kohl, 2012). Case in point, the expression of activating human FcγRIIa in a γc-deficient mouse was sufficient to restore susceptibility to K/BxN serum transfer arthritis in a C5aR-dependent manner (Tsuboi et al., 2011). Moreover, C5a-C5aR signaling may regulate the FcγR inhibitory/activating ratio, setting the threshold for FcγR-dependent function (Shushakova et al., 2002). Thus, the mechanism by which anti-GPI Ab activates neutrophils through Fc receptors is complex and likely involves interactions with multiple FcγRs, both activating and inhibitory. The role of C5aR on hematopoietic cells has long been appreciated: C5a binding to neutrophils and signaling through C5aR leads to

phagocytosis, oxidative burst, release of granule proteases, upregulation of adhesion molecules and chemotaxis (Lee et al., 2008). The essential role of C5aR in experimental arthritis is also well established (Grant et al., 2002; Lee et al., 2006). Blockade or absence of C5aR protects against arthritis by preventing the recruitment of neutrophils to the joint. Our results elucidate the initial sequence of intravascular events involved in the antibody-mediated neutrophil activation. We show that AP-dependent C5a generation leads to neutrophil activation and upregulation of adhesion molecules that initiate neutrophil-endothelial interaction. Our findings do not preclude the participation of additional AP-dependent interactions: Neutrophil transmigration in K/ BxN arthritis is also dependent on the LTB4 receptor, which is known to be released upon C5aR activation (Miyabe et al., 2017). VE-cadherin is a type I transmembrane protein exclusively expressed in endothelial cells and concentrated at adherens junctions. VEcadherin is linked to the cytoskeleton through its interaction with 638

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 7. In vitro C5a-induced degranulated neutrophil contents degrade VE-cadherin, generating intercellular gaps. A) C5a-induced degranulated neutrophil contents (30 ng of supernatant) were incubated with HUVEC monolayers. Loss of VE-cadherin (asterisks, left lower panel) and appearance of intercellular gaps (right lower panel) at time 0 (Media) and 30 min. VE-cadherin (red), F-actin (green), DAPI (blue). Scale bar = 30 μm. B) Quantification of intercellular gap formation over time, as detailed in Methods. C) Incubation of C5a-induced degranulated neutrophil supernatant with HUVECs led to VE-cadherin degradation. D) Neutrophil protease inhibitors, specifically NE inhibitor (NEi) or the combination of CGi, NEi, PR3i and MMPi (x4i), blocked VE-cadherin degradation. Actin served as control for protein loading. Statistical analysis was performed using one-way ANOVA.

β−catenin (Vestweber et al., 2009). During transendothelial migration, VE-cadherin can either be cleared from the site of paracellular transmigration through internalization into clathrin-coated vesicles (Chiasson et al., 2009) or pushed aside along the plane of the junction (Shaw et al., 2001). VE-cadherin proteolytic cleavage also occurs (Hermant et al., 2003), but its role in neutrophil transmigration remains a matter of debate. Previous studies revealed a non-redundant role for NE in neutrophil transmigration in vivo (Young et al., 2004). Our results corroborate these findings and strongly support the hypothesis that AP activation and C5a generation stimulate neutrophils to release their granule proteases, including serine proteases and metalloproteases capable of disassembling VE-cadherin and remodeling the F-actin cytoskeleton, generating gaps that facilitate neutrophil transmigration. These findings reveal a mechanism by which the crosstalk between adherent neutrophils decorated with extracellular proteases and the endothelium may contribute to the endothelial barrier dysfunction and increased permeability during the inflammatory response. These findings should be contrasted with the mechanism by which K/BxN serum causes immediate vascular leak that requires FcγR, mast cells, and neutrophils but is independent of complement and not completely essential for arthrogenesis (Binstadt et al., 2006). In summary, we have demonstrated that neutrophil responses to K/ BxN serum stimulation are amplified by complement AP activation at the cell surface and C5a generation locally, leading to increased neutrophil adhesion to the endothelium. And although CP is dispensable,

the possible participation of the LP has not been fully explored. Mannose-binding lectin A (MBL-A), a major LP pattern recognition molecule (PRM) is not required for K/BxN arthritis (Ji et al., 2002), but other LP PRMs, including MBL-C, ficolin-1, -2, and -3, collectin-10 and -11, may play critical roles (Garred et al., 2016). In addition, the C2 bypass mechanism for LP-dependent AP-activation (Selander et al., 2006) may also participate. Of note, in our previous investigation of a mouse model of abdominal aortic aneurysm (AAA) we observed that natural IgG antibody deposited in the aortic wall activated the LP, which in turn activated the AP (Zhou et al., 2013). In the case of AAA, complement activation is CP-independent but MBL-A/C-dependent. We also show that C5a contributes to endothelial cytoskeleton remodeling and vascular leakage by stimulating neutrophils to release proteases that promote VE-cadherin degradation. While we did not directly assess arthritis, but rather the local effects of immune complexes on neutrophil activation / extravasation, the accelerated model has previously been shown to induce a similar but more transient and localized arthritis compared to systemic K/BxN serum transfer (Wang et al., 2012). Thus, our findings are relatable to inflammatory arthritis. However, the clinical use of C5a antagonists in RA, has so far been met with disappointing results (Mojcik et al., 2004; Vergunst et al., 2007). On the other hand, complement activation on neutrophils following K/BxN serum administration is reminiscent of ANCA-associated vasculitis, another disease where the AP, autoantibodies and neutrophils play critical roles is a vasculitic process and where blocking C5aR is as 639

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 8. C5aRA blocks protease release and VE-cadherin degradation. A) Mice treated with control peptide or C5aRA were sacrificed 60 min after s.c. K/BxN serum injection. Their paw sections were stained for MPO (green), VE-cadherin (red) and DNA (DAPI). In mice treated with Ctrl peptide Z-stacks confocal images of blood vessels in inflamed paws revealed that in vivo neutrophils adherent to the endothelium displayed abundant extracellular MPO and were associated with areas of degraded VE-cadherin. C5aRA blocked degranulation and preserved adherens junctions. Scale bar = 25 μm. B) Higher magnification of inset areas in (A). Scale bar = 10 μm.

effective as high dose corticosteroids in patients with renal involvement (Jayne et al., 2017; Tesar and Hruskova, 2018). This suggests that targeting C5aR may have therapeutic implication in RA and a wide range of other neutrophil-dependent autoimmune diseases. JL is the inventor of patents and/or patent applications that describe the use of complement inhibitors for therapeutic purposes; the founder of Amyndas Pharmaceuticals, which is developing complement inhibitors (i.e., next generation compstatins) for clinical applications; and the inventor of the compstatin technology licensed to Apellis Pharmaceuticals (i.e., 4(1 MeW)7W/POT-4/APL-1 and its PEGylated derivatives such as APL-2).The rest of the authors have declared that no conflict of interest exists.

the use of complement inhibitors for therapeutic purposes; the founder of Amyndas Pharmaceuticals, which is developing complement inhibitors (i.e., next generation compstatins) for clinical applications; and the inventor of the compstatin technology licensed to Apellis Pharmaceuticals (i.e., 4(1 MeW)7W/POT-4/APL-1 and its PEGylated derivatives such as APL-2).The rest of the authors have declared that no conflict of interest exists. Acknowledgements The authors would like to thank Dr. Tanya Mayadas (Brigham and Women’s Hospital) for the generous gift of common gamma chain knockout mice. This work was supported in part by the National Institutes of Health (grants AR067491, AR073752, AI115139, AI051436, AI077600, U01 AI095550, AI068730, P30 AR073752) and VA Merit Review grant I01BX002714.

Declaration of Competing Interest JL is the inventor of patents and/or patent applications that describe 640

Molecular Immunology 114 (2019) 629–642

A. Akk, et al.

Fig. 9. C5a-C5aR interaction induces endothelial F-actin cytoskeleton remodeling. A) Mice treated with control (Ctrl) peptide or C5aRA were sacrificed 60 min after s.c. K/BxN serum injection. Their paws were sectioned and stained for neutrophils (Gr-1, red), F-actin (green) and nuclei (DAPI). Scale bar = 100 μm. B) Integrity of the endothelial cytoskeleton was scored on a scale of 0–4 (see Methods). C) Density of microvascular F-actin fluorescent peaks was calculated using ImageJ software. D) The number of F-actin peaks per μm of endothelium was obtained from at least 10 medium size blood vessels per animals; results presented as mean ± SEM (n = 3 animals per treatment condition). Statistical significance between treatment group was determined by student’s t-test.

Appendix A. Supplementary data

Gardiner, E.E., et al., 2001. Regulation of P-selectin binding to the neutrophil P-selectin counter-receptor P-selectin glycoprotein ligand-1 by neutrophil elastase and cathepsin G. Blood 98 (5), 1440–1447. Garred, P., et al., 2016. A journey through the lectin pathway of complement-MBL and beyond. Immunol. Rev. 274 (1), 74–97. Gotsch, U., et al., 1997. VE-cadherin antibody accelerates neutrophil recruitment in vivo. J. Cell. Sci. 110 (Pt 5), 583–588. Grant, E.P., et al., 2002. Essential role for the C5a receptor in regulating the effector phase of synovial infiltration and joint destruction in experimental arthritis. J. Exp. Med. 196 (11), 1461–1471. Hermant, B., et al., 2003. Identification of proteases involved in the proteolysis of vascular endothelium cadherin during neutrophil transmigration. J. Biol. Chem. 278 (16), 14002–14012. Hordijk, P.L., et al., 1999. Vascular-endothelial-cadherin modulates endothelial monolayer permeability. J. Cell. Sci. 112 (Pt 12), 1915–1923. Jayne, D.R.W., et al., 2017. Randomized trial of C5a receptor inhibitor avacopan in ANCA-Associated vasculitis. J. Am. Soc. Nephrol. 28 (9), 2756–2767. Jelezarova, E., Vogt, A., Lutz, H.U., 2000. Interaction of C3b(2)–IgG complexes with complement proteins properdin, factor B and factor H: implications for amplification. Biochem. J. 349 (Pt 1), 217–223. Ji, H., et al., 2002. Arthritis critically dependent on innate immune system players. Immunity 16 (2), 157–168. Karsten, C.M., Kohl, J., 2012. The immunoglobulin, IgG Fc receptor and complement triangle in autoimmune diseases. Immunobiology 217 (11), 1067–1079. Kemper, C., Hourcade, D.E., 2008. Properdin: new roles in pattern recognition and target clearance. Mol. Immunol. 45 (16), 4048–4056. Kimura, Y., Zhou, L., Miwa, T., Song, W.C., 2010. Genetic and therapeutic targeting of properdin in mice prevents complement-mediated tissue injury. J. Clin. Invest. 120 (10), 3545–3554. Korganow, A.S., et al., 1999. From systemic T cell self-reactivity to organ-specific autoimmune disease via immunoglobulins. Immunity 10 (4), 451–461. Kyburz, D., Corr, M., 2003. The KRN mouse model of inflammatory arthritis. Springer Semin. Immunopathol. 25 (1), 79–90. Lee, H., Whitfeld, P.L., Mackay, C.R., 2008. Receptors for complement C5a. The importance of C5aR and the enigmatic role of C5L2. Immunol. Cell Biol. 86 (2), 153–160.

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.molimm.2019.09.011. References Adkison, A.M., Raptis, S.Z., Kelley, D.G., Pham, C.T., 2002. Dipeptidyl peptidase I activates neutrophil-derived serine proteases and regulates the development of acute experimental arthritis. J. Clin. Invest. 109 (3), 363–371. Binstadt, B.A., et al., 2006. Particularities of the vasculature can promote the organ specificity of autoimmune attack. Nat. Immunol. 7 (3), 284–292. Blatt, A.Z., et al., 2016. Properdin-mediated C5a production enhances stable binding of platelets to granulocytes in human whole blood. J. Immunol. 196 (11), 4671–4680. Bucher, K., et al., 2015. Fluorescent Ly6G antibodies determine macrophage phagocytosis of neutrophils and alter the retrieval of neutrophils in mice. J. Leukoc. Biol. 98 (3), 365–372. Chiasson, C.M., Wittich, K.B., Vincent, P.A., Faundez, V., Kowalczyk, A.P., 2009. p120catenin inhibits VE-cadherin internalization through a Rho-independent mechanism. Mol. Biol. Cell 20 (7), 1970–1980. Choi, E.Y., Santoso, S., Chavakis, T., 2009. Mechanisms of neutrophil transendothelial migration. Front Biosci 14, 1596–1605 (Landmark Ed). Corr, M., Crain, B., 2002. The role of FcgammaR signaling in the K/B x N serum transfer model of arthritis. J. Immunol. 169 (11), 6604–6609. DiScipio, R.G., Daffern, P.J., Jagels, M.A., Broide, D.H., Sriramarao, P., 1999. A comparison of C3a and C5a-mediated stable adhesion of rolling eosinophils in postcapillary venules and transendothelial migration in vitro and in vivo. J. Immunol. 162 (2), 1127–1136. DiScipio, R.G., Schraufstatter, I.U., Sikora, L., Zuraw, B.L., Sriramarao, P., 2006. C5a mediates secretion and activation of matrix metalloproteinase 9 from human eosinophils and neutrophils. Int. Immunopharmacol. 6 (7), 1109–1118. Fearon, D.T., Austen, K.F., 1975. Properdin: binding to C3b and stabilization of the C3bdependent C3 convertase. J. Exp. Med. 142 (4), 856–863.

641

Molecular Immunology 114 (2019) 629–642

A. Akk, et al. Lee, H., et al., 2006. Human C5aR knock-in mice facilitate the production and assessment of anti-inflammatory monoclonal antibodies. Nat. Biotechnol. 24 (10), 1279–1284. Ley, K., Laudanna, C., Cybulsky, M.I., Nourshargh, S., 2007. Getting to the site of inflammation: the leukocyte adhesion cascade updated. Nat. Rev. Immunol. 7 (9), 678–689. Malbran, A., Frank, M.M., Fries, L.F., 1987. Interactions of monomeric IgG bearing covalently bound C3b with polymorphonuclear leucocytes. Immunology 61 (1), 15–20. Mancardi, D.A., et al., 2011. Cutting Edge: the murine high-affinity IgG receptor FcgammaRIV is sufficient for autoantibody-induced arthritis. J. Immunol. 186 (4), 1899–1903. Matsumoto, M., et al., 1997. Abrogation of the alternative complement pathway by targeted deletion of murine factor B. Proc Natl Acad Sci U S A 94 (16), 8720–8725. Matsumoto, I., et al., 2002. How antibodies to a ubiquitous cytoplasmic enzyme may provoke joint-specific autoimmune disease. Nat. Immunol. 3 (4), 360–365. Miyabe, Y., et al., 2017. Complement C5a receptor is the key initiator of neutrophil adhesion igniting immune complex-induced arthritis. Sci. Immunol. 2 (7). Mojcik, C.F., et al., 2004. Results of a phase 2b study of the humanized anti-C5 antibody eculizumab in patients with rheumatoid arthritis. Ann. Rheum. Dis. 63, 301. Monach, P.A., et al., 2010. Neutrophils in a mouse model of autoantibody-mediated arthritis: critical producers of Fc receptor gamma, the receptor for C5a, and lymphocyte function-associated antigen 1. Arthritis Rheum. 62 (3), 753–764. Monk, P.N., Scola, A.M., Madala, P., Fairlie, D.P., 2007. Function, structure and therapeutic potential of complement C5a receptors. Br. J. Pharmacol. 152 (4), 429–448. Muller, W.A., 2013. Getting leukocytes to the site of inflammation. Vet. Pathol. 50 (1), 7–22. Pangburn, M.K., Rawal, N., 2002. ) Structure and function of complement C5 convertase enzymes. Biochem. Soc. Trans. 30 (Pt 6), 1006–1010. Pangburn, M.K., Schreiber, R.D., Muller-Eberhard, H.J., 1981. Formation of the initial C3 convertase of the alternative complement pathway. Acquisition of C3b-like activities by spontaneous hydrolysis of the putative thioester in native C3. J. Exp. Med. 154 (3), 856–867. Ricklin, D., Hajishengallis, G., Yang, K., Lambris, J.D., 2010. Complement: a key system for immune surveillance and homeostasis. Nat. Immunol. 11 (9), 785–797. Sadik, C.D., Kim, N.D., Iwakura, Y., Luster, A.D., 2012. Neutrophils orchestrate their own recruitment in murine arthritis through C5aR and FcgammaR signaling. Proc Natl Acad Sci U S A 109 (46), E3177–3185. Saggu, G., et al., 2013. Identification of a novel mode of complement activation on stimulated platelets mediated by properdin and C3(H2O). J. Immunol. 190 (12), 6457–6467. Schnatbaum, K., et al., 2006. Peptidomimetic C5a receptor antagonists with hydrophobic substitutions at the C-terminus: increased receptor specificity and in vivo activity. Bioorg. Med. Chem. Lett. 16 (19), 5088–5092. Selander, B., et al., 2006. Mannan-binding lectin activates C3 and the alternative complement pathway without involvement of C2. J. Clin. Invest. 116 (5), 1425–1434.

Shaw, S.K., Bamba, P.S., Perkins, B.N., Luscinskas, F.W., 2001. Real-time imaging of vascular endothelial-cadherin during leukocyte transmigration across endothelium. J. Immunol. 167 (4), 2323–2330. Shushakova, N., et al., 2002. C5a anaphylatoxin is a major regulator of activating versus inhibitory FcgammaRs in immune complex-induced lung disease. J. Clin. Invest. 110 (12), 1823–1830. Stover, C.M., et al., 2008. Properdin plays a protective role in polymicrobial septic peritonitis. J. Immunol. 180 (5), 3313–3318. Strey, C.W., et al., 2003. The proinflammatory mediators C3a and C5a are essential for liver regeneration. J. Exp. Med. 198 (6), 913–923. Tesar, V., Hruskova, Z., 2018. Avacopan in the treatment of ANCA-associated vasculitis. Expert Opin. Investig. Drugs 27 (5), 491–496. Tsuboi, N., et al., 2011. Regulation of human neutrophil Fcgamma receptor IIa by C5a receptor promotes inflammatory arthritis in mice. Arthritis Rheum. 63 (2), 467–478. Vergunst, C.E., et al., 2007. Blocking the receptor for C5a in patients with rheumatoid arthritis does not reduce synovial inflammation. Rheumatology Oxford (Oxford) 46 (12), 1773–1778. Vestweber, D., Winderlich, M., Cagna, G., Nottebaum, A.F., 2009. Cell adhesion dynamics at endothelial junctions: VE-cadherin as a major player. Trends Cell Biol. 19 (1), 8–15. Wang, B., et al., 2012. In vivo imaging implicates CCR2(+) monocytes as regulators of neutrophil recruitment during arthritis. Cell. Immunol. 278 (1–2), 103–112. Webster, R.O., Larsen, G.L., Henson, P.M., 1982. In vivo clearance and tissue distribution of C5a and C5a des arginine complement fragments in rabbits. J. Clin. Invest. 70 (6), 1177–1183. Wipke, B.T., Allen, P.M., 2001. Essential role of neutrophils in the initiation and progression of a murine model of rheumatoid arthritis. J. Immunol. 167 (3), 1601–1608. Wipke, B.T., Wang, Z., Nagengast, W., Reichert, D.E., Allen, P.M., 2004. Staging the initiation of autoantibody-induced arthritis: a critical role for immune complexes. J. Immunol. 172 (12), 7694–7702. Wirthmueller, U., et al., 1997. Properdin, a positive regulator of complement activation, is released from secondary granules of stimulated peripheral blood neutrophils. J. Immunol. 158 (9), 4444–4451. Xiao, H., Schreiber, A., Heeringa, P., Falk, R.J., Jennette, J.C., 2007. Alternative complement pathway in the pathogenesis of disease mediated by anti-neutrophil cytoplasmic autoantibodies. Am. J. Pathol. 170 (1), 52–64. Yipp, B.G., Kubes, P., 2013. Antibodies against neutrophil LY6G do not inhibit leukocyte recruitment in mice in vivo. Blood 121 (1), 241–242. Young, R.E., et al., 2004. Neutrophil elastase (NE)-deficient mice demonstrate a nonredundant role for NE in neutrophil migration, generation of proinflammatory mediators, and phagocytosis in response to zymosan particles in vivo. J. Immunol. 172 (7), 4493–4502. Zhou, H.F., et al., 2013. Fibrinogen-specific antibody induces abdominal aortic aneurysm in mice through complement lectin pathway activation. Proc Natl Acad Sci U S A 110 (46), E4335–4344.

642