View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Elsevier - Publisher Connector Immunity, Vol. 22, 679–691, June 7, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.immuni.2005.03.015 Serine Protease Cathepsin G Regulates Adhesion-Dependent Neutrophil Effector Functions by Modulating Integrin Clustering Sofia Z. Raptis,1,2 Steven D. Shapiro,3 mediators (Ravetch, 1994). It has been known that IC Pamela M. Simmons,1 Alec M. Cheng,1,2,4 activation of PMNs requires the cooperation of FcγR 1,2, and Christine T.N. Pham * and β2 integrins, particularly Mac-1 (CD11b/CD18) on 1Division of Rheumatology PMNs (Jones et al., 1998; Zhou and Brown, 1994). In Washington University School of Medicine addition, Mac-1-deficient PMNs have blunted re- Saint Louis, Missouri 63110 sponses to IC stimulation in vitro and decreased PMN 2 Department of Pathology and Immunology recruitment in a model of IC-mediated glomerulone- Washington University School of Medicine phritis in vivo (Tang et al., 1997). Thus FcγR and inte- Saint Louis, Missouri 63110 grins play an important role in the development of IC- 3 Department of Medicine mediated inflammation. Pulmonary and Critical Care Cathepsin G (CG), neutrophil elastase (NE), and pro- Brigham and Women’s Hospital teinase 3 (PR3) are a family of highly related neutral Harvard School of Medicine serine proteases expressed in the primary granules of Boston, Massachusetts 02115 PMNs (Owen and Campbell, 1999). The importance of NE and CG during infections has been clearly demon- strated in serine protease-deficient mice. These mice Summary are more susceptible to various pathogens, including gram-positive and -negative bacteria and certain fungal The polymorphonuclear leukocyte (PMN)-derived ser- organisms (Belaaouaj et al., 1998; Reeves et al., 2002; ine proteases play a key role in immune complex (IC)- Tkalcevic et al., 2000). Recent studies have also shown mediated inflammation. However, the mechanisms by that mice lacking in CG and NE are resistant to IC- which these proteases regulate inflammatory re- mediated diseases, further suggesting an active role for sponse remain largely undefined. Here, we show that these proteases in immune responses other than their IC-activated cathepsin G- and neutrophil elastase- antibacterial functions. Consistent with this emerging deficient (CG/NE) PMNs adhered normally to IC- view of protease functions, studies have demonstrated coated surfaces but did not undergo CD11b cluster- that NE-deficient mice are resistant to an experimental ing and failed to initiate cytoskeletal reorganization mouse model of bullous pemphigoid (Liu et al., 2000), and cell spreading. As a result, CG/NE-deficient PMNs and CG/NE mice were resistant to an IC-mediated, neu- exhibited severe defects in MIP-2 secretion and trophil-dependent arthritis model (Adkison et al., 2002). reactive oxygen intermediates production. Exoge- In the later study, there was no PMN accumulation in the joints of CG/NE mice that received arthrogenic anti- nously added CG, but not proteolytically inactive CG, bodies. Currently, the role of serine proteases in PMN was sufficient to restore these defects. These find- transmigration and recruitment remains controversial. ings identify an important role for CG in integrin- Although some studies suggest that serine proteases dependent PMN effector functions that are separate might be required for the migration of PMNs to inflam- from and downstream of integrin-dependent ad- matory sites (Carden and Korthuis, 1996; Carney et al., hesion. 1998; Woodman et al., 1993), others reported that CG and NE are not essential for normal chemotaxis to vari- Introduction ous inflammatory stimuli (Hirche et al., 2004; Huber and Weiss, 1989; Mackarel et al., 1999). Thus, the fact that PMNs are an essential component of the host innate PMNs deficient in CG and NE exhibit normal chemo- response against invading pathogens. However, exces- taxis but do not accumulate in vivo in response to IC- sive PMN responses are thought to contribute to the induced inflammation points to an undefined role for development and perpetuation of autoimmune dis- these proteases in regulating leukocyte recruitment. eases. Although many aspects of PMN activation lead- According to the current paradigm, CG and NE, ing to transmigration across the endothelial barrier secreted upon PMN activation, are thought to degrade have been elucidated, the mechanisms that regulate specific target molecules of the ECM, thus contributing PMN activation and effector functions in the extracellu- to tissue damage and the propagation of inflammation. lar matrix (ECM) are still undefined. Recent reports suggest, however, that serine proteases One of the major mechanisms by which PMNs accu- remain bound to the cell surface membrane upon re- mulate and become activated at sites of inflammation lease from primary granules, where they become resis- in noninfectious diseases is through the interaction with tant to circulating endogenous protease inhibitors pres- IC. Binding of PMNs to IC through their Fc gamma re- ent in the serum (Owen et al., 1995a, 1995b). Hence, ceptors (FcγR) leads to phagocytosis, generation of re- the mode of action of membrane bound serine prote- active oxygen intermediates (ROI), release of proteo- ases may be more complex than previously thought. lytic enzymes, and secretion of soluble inflammatory These studies and the results from the experimental ar- thritis model aforementioned led us to consider the un- *Correspondence: [email protected] explored notion that the membrane bound form of ser- 4 Present address: Genentech, Inc., South San Francisco, CA ine proteases may directly regulate PMN activation and 94108. effector functions. In this report, we test this hypothesis Immunity 680 1F). In this case, the number of leukocytes recruited at four hr after injection of zymosan was equivalent in both wild-type (wt) and mutant mice (Figure 1D). Zymo- san has been shown to be a strong activator of the complement system, resulting in the formation of C5a, a potent PMN chemoattractant (Fearon and Austen, 1977). Thus, the C5a gradient generated by zymosan is likely responsible for the equivalent numbers of PMNs initially recruited into the air pouch, as CG/NE PMNs chemotax normally in response to C5a (see Figure S1 in the Supplemental Data available with this article on- line). As C5a has a short half-life, other chemoattrac- tants are likely required to sustain the recruitment of leukocytes over time. In support of this notion, we pre- viously detected a significant reduction in the number of PMNs recruited to the air pouch of CG/NE mice at 12 hr post zymosan injection (Adkison et al., 2002). PMNs themselves have been shown to synthesize and release CXC chemokines when activated (Scapini et al., 2000). PMNs are among the earliest cell types recruited to an inflammatory site, and their ability to secrete chemo- kines may constitute a mechanism by which PMNs can Figure 1. Defective Chemokine Production in CG/NE Mice influence the early trafficking of inflammatory cells. Be- Air pouches were formed as described in Experimental Procedures. cause CG and NE are PMN-specific proteases, we pos- For IC formation (A–C), mice were injected with 20 mg/kg chicken tulate that these enzymes may directly regulate PMN egg albumin (OVA) intravenously followed by injection of 800 ␮g effector functions that in turn shape the inflammatory of rabbit anti-OVA Abs in the air pouch (n = 4–6 mice per group). responses. Alternatively, mice were injected with 300 ␮g of zymosan in 1 ml of saline in the air pouch (D–F; n = 4 mice per group). Leukocyte counts in air pouch were enumerated after 4 hr; MIP-2 and KC in CG and NE Are Required for Optimal Adhesion- the lavage fluid were measured by ELISA. Asterisks denote signifi- Dependent Effector Functions cant difference between wt and CG/NE mice. *p < 0.05, **p < 0.01, Sustained PMN effector functions in response to IC ***p < 0.001. All values represent mean ± SD. stimulation require adhesion and have been shown to depend on the ligation of both FcγR and β2 integrins (Jones et al., 1998; Zhou and Brown, 1994). To recapitu- by examining the response of purified PMNs to IC stim- late these requirements in an in vitro system, wt and ulation in an in vitro cell culture system in which PMN CG/NE PMNs were stimulated on immobilized IC in the activation can be evaluated independent of other cell presence of Ca2+ and Mg2+ and TNF-α priming. IC-acti- types. Our findings suggest that neutrophil-derived ser- vated wt PMNs in adherent culture released significant ine proteases regulate PMN effector functions by mod- levels of the chemokine MIP-2, but not KC, IL-1β,or ulating integrin clustering. TNF-α (data not shown). In contrast, MIP-2 release by CG/NE and CG-deficient PMNs was markedly reduced Results compared to wt PMNs (Figure 2A). PMNs lacking NE alone were minimally compromised in MIP-2 release, Defective Chemokine Release in the Subcutaneous suggesting that CG has a critical role in regulating this Air Pouch of CG/NE Mice effector function (Figure 2A). We also examined NADPH The migration of PMNs to inflammatory sites is regu- oxidase activation in response to IC stimulation. Wt lated by the activity of CXC chemokines such as MIP-2 PMNs generated a robust respiratory burst, whereas and KC (Olson and Ley, 2002). We reasoned that the the amount of ROI produced by CG/NE PMNs was re- absence of PMNs in CG/NE joints in the arthritis model duced by >50%. PMNs lacking CG exhibited a more was due to a defect in the local generation of a chemo- severe defect than those lacking NE, suggesting that tactic gradient. To test this idea, we turned to the sub- CG has a predominant role in IC-mediated ROI pro- cutaneous air pouch model that offered us a direct way duction (Figure 2B).