INTRODUCTION

  • I. NEUTROPHIL MOLECULES AND FUNCTIONS

    • I.A. ADHESION AND MIGRATION

      • I.A.1. Traffic and margination

      • I.A.2. Adhesion to the Endothelial Wall

        • Rolling and Tethering

        • Neutrophil Priming During Rolling

        • Firm Adhesion and Spreading

      • I.A.3 Extravasation and Diapedesis Toward Inflammatory Stimuli

        • Transendothelial Migration

        • Migration Within Interstitial Tissues

        • Signaling by Chemoattractants

        • Transepithelial Migration

    • I.B. PHAGOCYTOSIS, DEGRANULATION AND BACTERIA KILLING

      • I.B.1. Phagocytosis

      • I.B.2. Degranulation

        • Granule Biogenesis

        • Mechanisms of Degranulation

      • I.B.3. Microbicidal Molecules

        • NADPH-Derived Oxidants

        • The H2O2-Myeloperoxidase System

        • Nitric Oxide-Synthase-Derived Reactive Nitrogen Intermediates

        • Granule Proteins

          • Antimicrobial Proteins

          • Proteases

    • I.C. CYTOKINE SYNTHESIS

      • I.C.1. TNF-α as a Proinflammatory Cytokine

      • I.C.2. IL-1 and IL-1 Receptor Antagonist (IL-1-Ra)

      • I.C.3. IL-8 as a Prototype of Chemokines

      • I.C.4. Modulation of Cytokine Expression by Neutrophils

        • IFN-γ

        • IL-10

        • IL-4 and IL-13

    • I.C.5. Molecular Regulation of Cytokine Production

  • I.D. APOPTOSIS AND RESOLUTION OF ACUTE INFLAMMATION

    • I.D.1. Progressive Decrease of Neutrophil Recruitment

    • I.D.2. Apoptosis in Resolution of Inflammation

  • II. NEUTROPHILS IN PATHOLOGY

    • II.A. Bacterial Infection

    • II.B. Tissue Injury-Induced Inflammation: Ischemia-Reperfusion Injury

    • II.C. Crystal-Induced Inflammation

    • II.D. Complement-Induced Inflammation and Oxidative Stress: Hemodialysis

    • II.E. Immune Complex-Induced Inflammation: Antibody-Mediated Glomerunephritis

    • II.F. Cytokine-Induced Inflammation: Rheumatoid Arthritis

    • II.G. Antineutrophil Cytoplasmic Antibodies and Vasculitis: Autoimmunity Against Neutrophil Components

    • II.H. Genetic Disorders of Neutrophil Regulations: Hereditary Periodic Fever Syndromes

    • II.I. Cystic Fibrosis: The Paradox of an Exacerbation of Neutrophil-Mediated Tissue Damage and a Concomitant Persistence of Infection

  • CONCLUSION

The notion that inflammation is the net resultant of pro and contra inflammatory pathways (Fauve, 1980) is well illustrated by the dual role of neutrophils which combine an anti-infectious and a proinflammatory role (Klebanoff, 1992; Ward, 1999; Weiss, 1989). The aim of this report is to review the main physiological and pathogenic activities of neutrophils—ie, adherence and migration, degranulation and release of inflammatory mediators, phagocytosis and apoptosis—in the light of the most recent molecular data on extracellular effectors and regulators, membrane receptors, and intracellular signaling pathways involved in these functions.

Leukocyte adhesion processes have been studied extensively during the last decade and most membrane molecules responsible for leukocyte interactions with other cells or with the extracellular matrix have been identified. New data are constantly reported on sophisticated intracellular pathways that allow neutrophils to integrate signals transmitted by adhesion partners with those of chemoattractants and cytokines. What remains puzzling is that leukocytes mostly use the same adhesion molecules to adhere to inflamed endothelium, as do lymphocytes to constantly recirculate from the blood to lymphoid tissues. Still, naive lymphocytes are the only leukocytes to cross endothelia of lymphoid high endothelial venules, while neutrophils are the first leukocytes, hours before monocytes or lymphocytes, to migrate specifically across the endothelium adjacent to the inflammation site. The selectivity and specific timing of such a highly redundant system is just becoming comprehensible, in particular with the description of the chemokine and serpentine families.

Another aspect reviewed here is the wide variety of effector molecules required to achieve the usual microbicidal role of neutrophils, including radical oxygen species (ROS), proteinases, bactericidal proteins and cytokines, which either alone or in concert may interact in up- or down-regulating the major inflammatory processes. We emphasize new directions of investigation regarding these neutrophil-derived effector molecules, as exemplified by myeloperoxidase-derived oxidants whose implications go far beyond inflammatory diseases. The potential clinical use of neutrophil-derived antibiotic proteins is illustrated by the Bactericidal Permeability Increasing protein (BPI), now undergoing clinical trials, and the design of novel antimicrobial peptides based on studies on defensins and cathelicidins.

Finally, we illustrate the various aspects of neutrophil biology by classifying, according to their predominant neutrophil-activating mechanism, diseases in which neutrophils play a pivotal role. Comprehension of the activation pathways will allow us to analyze, and possibly prevent, chronic inflammation processes where dysregulated neutrophil recruitment and activation results in severe damage of adjacent normal tissues.

I. Neutrophil Molecules and Functions

I.A. Adhesion and Migration

I.A.1. Traffic and Margination

Neutrophils are partitioned in the blood between a circulating pool, present in large blood vessels and in the axial stream of small vessels, and a marginating pool. In the absence of inflammation, the marginating pool, better called “physiological regional granulocyte pool” (Peters, 1998), comprises granulocytes transiently arrested in narrow, mainly pulmonary, capillaries. This physiological retention of neutrophils in capillaries appears to be a mechanical process due to the stiffness of neutrophils—as compared with the high deformability of erythrocytes (Downey et al, 1990)—and does not involve cell adhesion (Doyle et al, 1997; Mizgerd et al, 1996; Yamaguchi et al, 1997; Yoder et al, 1990).

Conversely, in inflamed organs, neutrophil traffic involves a selectin- and integrin-dependent sequestration in capillaries and post-capillary venules (Adams and Shaw, 1994; Springer, 1994). Intravascular injection of inflammatory mediators first results in rapid sequestration that involves a decrease in neutrophil deformability, followed by prolonged accumulation of neutrophils in the lung and liver parenchyma, by a process involving CD11b/CD18 integrins and L-selectin (Doerschuk, 1992; Doyle et al, 1997; Erzurum et al, 1992; Hogg and Doerschuk, 1995; Jaeschke and Smith, 1997; Tedder et al, 1995). Mechanisms involving adhesion also occur when neutrophil emigration follows instillation of stimuli in airways, peritoneum, or skin. Defective neutrophil recruitment to inflamed sites in leukocyte adhesion deficient (LAD) patients and in adhesion molecules-knock-out mice shows that these emigration processes require selectins (Borges et al, 1997; Bullard et al, 1996; Doyle et al, 1997; McEver and Cummings, 1997; Tedder et al, 1995) and the interactions of leucocyte CD18 integrins with endothelial ICAM-1 (Mizgerd et al, 1997; Sligh et al, 1993). However, animal models using intratracheal instillation of Streptococcus pneumonia suggest that neutrophil pulmonary traffic, at least in mice and rabbits, may differ from what happens in the systemic circulation and in some cases may involve selectin- and integrin-independent emigration from systemic venules (Mizgerd et al, 1996).

I.A.2. Adhesion to the Endothelial Wall

The dual neutrophil functions of immune surveillance and in situ elimination of microorganisms or cellular debris require a rapid transition between a circulating non-adherent state to an adherent state, allowing them to migrate into tissues where necessary. The initial event is the appearance, on the endothelium adjacent to the inflamed site, of new adhesion molecules, induced by inflammation mediators released by damaged tissues, which result in local extravasation of leukocytes. In postcapillary venules or in pulmonary capillaries, the slow flow rate, further reduced by vessel dilatation at sites of inflammation, allows a loose and somewhat transient adhesion, referred to as “tethering,” and resulting in the rolling of leukocytes along the endothelium. During this tethering step, neutrophils respond to ligands—mainly chemokines—dispatched on the endothelium surface by a signaling event that activates integrin-mediated sustained, stationary adhesion and spreading (Springer, 1994).

Rolling and Tethering

The rolling step is mediated by neutrophil L-selectin and by E- and P-selectins newly expressed on inflamed endothelial cells. Rare deficits in neutrophil selectin ligand expression, due to a metabolic defect in a synthetic pathway common to all selectin ligands, lead to faulty neutrophil trafficking in humans suffering from the LAD type 2 syndrome (Phillips et al, 1995). P-selectin, readily mobilized in a few minutes to the endothelial cell surface following stimulation by thrombin, histamine, or oxygen radicals, interacts primarily with a mucin-like ligand PSGL-1 (P-selectin glycoprotein ligand-1), located at the tip of leucocyte microvilli ( McEver and Cummings, 1997; Moore et al, 1995).

Rolling subsequently involves E-selectin, which appears on endothelial cells one to two hours after cell stimulation by IL-1, TNFα, or LPS (Lawrence and Springer, 1993; Patel et al, 1995). E-selectin counter-receptors include PSGL-1 and ESL1 (E-selectin-ligand 1), a molecule highly homologous to the cystein-rich FGF receptor (CFR) and located on neutrophil microvilli ( Steegmaier et al, 1997).

The kinetic of neutrophil recruitment in selectin-deficient mice suggests that P- and L-selectin contribute sequentially to leucocyte rolling and shows that L-selectin is involved in the prolonged neutrophil sequestration in inflamed microvasculature (Doyle et al, 1997; Ley et al, 1995; Steeber et al, 1998). Unlike P- and E- selectins, L-selectin is constitutively present on leukocytes. Its binding capacity is however rapidly and transiently increased after leukocyte activation, possibly via receptor oligomerization (Li et al, 1998). So far, only one inducible L-selectin counter-receptor, specifically expressed on inflamed endothelium, has been described, which is bearing the cutaneous lymphocyte antigen (CLA) (Tu et al, 1999). In addition to its binding to endothelial ligands, leucocyte PSGL-1 is a counter-receptor for leukocyte L-selectin and there is evidence that neutrophils roll, via L-selectin, on previously adherent neutrophils (Alon et al, 1996; Bargatze et al, 1994). This secondary tethering would synergistically enhance leukocyte accumulation on inflamed endothelium.

Neutrophil Priming During Rolling

The endothelium of inflamed microvessels produces chemoattractants such as platelet-activating factor (PAF), leukotriene B4, and various chemokines, immobilized via a “presentation molecule” (proteoglycan) on the luminal surface of endothelial cells. Among these chemokines, interleukin 8 (IL-8) specifically attracts neutrophils, while having no effect on monocytes (Premack and Schall, 1996; Rollins, 1997) and being unable to promote lymphocyte transmigration through endothelium (Roth et al, 1995). IL-8 is a major neutrophil chemoattractant, as shown by the complete inhibition of neutrophil recruitment in inflammation sites by anti-IL8 monoclonal antibodies in animal models (Folkesson et al, 1995; Matsumoto et al, 1997; Sekido et al, 1993). There is evidence that microvascular endothelial cells not only synthesize IL-8 in response to IL-1 or LPS, but also store IL-8 in Weibel-Palade bodies and release it upon stimulation by histamine or thrombin (Utgaard et al, 1998; Wolff et al, 1998). Moreover, tissue-derived IL-8 is internalized by endothelial cells of postcapillary venules and small veins, transcytosed in the abluminal-to-luminal direction via plasmalemmal vesicles (caveolae), and presented at the tips of microvilli of the endothelial cell luminal surface (Middleton et al, 1997).

Neutrophils bear several receptors for chemoattractants, which belong to the super-family of seven-transmembrane receptors associated with intracellular GTP-binding heteroproteins. As described below, the engagement of these G protein-coupled receptors triggers a variety of signal transduction cascades that lead to firm leukocyte adhesion and activation of direction-specific movement, but also to a wide range of functions such as degranulation or respiratory burst. These latter effects do not normally occur during the initial adhesion to endothelial cells, but are delayed until leukocytes have reached the inflammatory focus. Ceramide, resulting from TNF-activated sphingomyelinase, has been proposed as one of the regulating mediators responsible for this delay (Fuortes et al, 1996).

Firm Adhesion and Spreading

Firm neutrophil adhesion to endothelial cells appears to involve exclusively the interaction of leucocyte integrins of the β2 subfamily (CD11a, CD11b, CD11c/CD18) with ICAM-1, as shown by the defects observed in CD18-deficient LAD patients and in ICAM-1 knock-out mice (Anderson et al, 1984; Fischer et al, 1983; Sligh et al, 1993). This differs from monocytes and lymphocytes, which also react via α4β1 integrin with endothelial VCAM-1.

β2 integrins are unable to interact with their physiological ligands in unstimulated neutrophils, a safety mechanism that controls acute and chronic inflammatory responses. The ligand binding capacity is acquired upon activation signals (“inside-out signaling”) that lead to integrins clustering and to a transition of a β2-integrin subpopulation to a high affinity state (Rieu and Arnaout, 1996; Stewart and Hogg, 1996). Various agonists trigger CD11b/CD18 activation in neutrophils, including chemoattractants (PAF, IL8, fMLP, C5a), cytokines and growth factors (TNFα or GMCSF), and bacterial products (formylated peptides and LPS). During the initial rolling on endothelial cells, integrin “activation” signals are given by chemoattractants displayed on the endothelial membrane and presumably also by the engagement of selectins and their counter-receptors. Indeed, ligation of L-selectin by antibodies or carbohydrates that mimic natural L-selectin ligands (Simon et al, 1995)—or PSGL-1 interaction with P-selectin (Yago et al, 1999)—signal neutrophil adhesive functions via CD11b/CD18 integrins (Brenner et al, 1996; Steeber et al, 1997). Inside-out signaling pathways that lead to integrin switch to an active conformation differ with the stimulating agonist and are still incompletely characterized (Blouin et al, 1999; Capodici et al, 1998; Jones et al, 1998).

The regulation of β2-integrin avidity (clustering) involves interactions of both α and β chain cytoplasmic tails with the cytoskeleton (Van Kooyk et al, 1999) and the membrane association of cytohesin-1, a guanine nucleotide exchange protein that binds to the cytoplasmic portion of CD18 and up-regulates β2-integrin avidity (Kolanus et al, 1996; Nagel et al, 1998). Integrins transmit signals triggered by their clustering and multiple engagements with adhesion substrates (“outside-in signaling”). Neutrophils integrate these signals of integrin engagement and those delivered simultaneously by inflammatory cytokines or chemoattractants to activate a cascade of intracellular events resulting in cell spreading, locomotion, degranulation, and oxidative burst. These outside-in transduction pathways include the activation of various tyrosine kinases (Berton, 1999a; Fuortes et al, 1999; Lowell and Berton, 1999).

Finally, CD11b/CD18 integrin interacts in cis with GPI-anchored membrane proteins, such as FcγRIIIb (CD16b), the LPS receptor CD14 or the urokinase receptor uPAR (CD87). Integrins behave as promiscuous transducers mediating signals triggered by these GPI-linked receptors (Petty and Todd, 1996). FcγRIIIb interaction with CD11bCD18 promotes antibody-dependent phagocytosis (Todd and Petty, 1997), while CD14 interaction with CD11bCD18 only occurs in the presence of LPS and LPS-binding protein and may play a role in the generation of proinflammatory mediators (Zarewych et al 1996, Todd and Petty, 1997).

I.A.3. Extravasation and Diapedesis Toward Inflammatory Stimuli

Transendothelial Migration

Neutrophil transmigration occurs prominently at the borders of endothelial cells, where discontinuities of tight junctions are observed. P-selectin has been shown to be concentrated along endothelial borders and may target there neutrophil adhesion (Burns et al, 1999). Extravasation requires, however, modifications of endothelial cell-to-cell adherent junctions. Indeed, disorganization of the junctional components VE-cadherin, β-catenin, and plakoglobin has been observed in the vicinity of regions of firm adhesion between neutrophils and endothelial cells (Del Maschio et al, 1996).

Two cell adhesion molecules of the Ig-superfamily (CAMs) have been shown to be involved in leukocyte transmigration, the platelet endothelial cell adhesion molecule-1 (PECAM-1 or CD31) and, more recently, the junctional adhesion molecule (JAM) (Martin-Padura et al, 1998; Muller et al, 1993; Vaporciyan et al, 1993). PECAM-1 is expressed both on the neutrophil surface and at the endothelial cell junction and mediates neutrophil extravasation via PECAM-1/PECAM-1 homophilic interactions. A “zipper” model has been proposed to account for a transmigration of leukocytes that maintains the permeability barrier of the endothelial cell monolayer (Muller et al, 1993). The JAM is selectively concentrated at inter-endothelial tight junctions but is not present on neutrophils. Antibodies to JAM inhibit in vitro leukocyte transmigration but the leukocyte counter-receptor of JAM is not known (Dejana et al, 1999; Martin-Padura et al, 1998). Finally, PECAM-1 is able to transduce signals into the cell, and its dimerization, by antibody cross-linking, increases CD11b/CD18 binding capacity via an inside-out signal transduction that involves PI3-kinase (Berman and Muller, 1995; Pellegatta et al, 1998).

Migration Within Interstitial Tissues

Neutrophils migrate in tissues by haptotaxis, ie, up a gradient of immobilized, rather than soluble (chemotaxis) chemoattractants. These chemoattractants, produced by bacteria, by dying cells, or by various stromal and epithelial cells of inflamed tissues, are bound to extracellular matrix components because of their negative charge. In vitro models show that leukocytes navigate through complex chemoattractant fields by migrating in a multistep process in response to one agonist source after the other (Fig. 1). Cross-talks between chemoattractant receptors and their signaling pathways may result in desensitization to one chemoattractant by another. In particular, signals delivered by “end target-derived” chemoattractants—such as formyl peptides, released by bacteria or by mitochondria from dying cells, or complement C5a, produced in their immediate surrounding—are dominant and override “regulatory cell-derived” attractants, such as bioactive peptides (LTB4) or chemokines (IL8) (Foxman et al, 1999; Kitayama et al, 1997). This will allow, for example, leukocytes recruited by endothelial-derived chemoattractants to migrate away from the endothelial agonist source towards their final target within a tissue.

Figure 1
figure 1

Schematic view of the chemotactic migration of leukocytes towards an inflammation site. Neutrophils move through the endothelium and within tissues by responding to successive combinations of chemoattractant gradients. Chemoattractants are released by endothelial cells, by activated stromal cells (macrophages, epithelial cells…), and by the inflammatory targets, ie, bacteria or dying cells. The direction of neutrophil movement is first guided by the steepest local chemoattractant gradient and is then regulated by successive receptor desensitization and attraction by secondary distant agonists. Finally, end-target attractants are dominant over regulatory cell-derived agonists (adapted from Foxman et al, 1999)

Neutrophil migration through the extracellular matrix is mediated by β2 integrins, in concert with β1 and β3 integrins: the laminin-, fibronectin-, and vitronectin receptors α6β1, α5β1 and αvβ3 are mostly stored in neutrophil granules and rapidly expressed on the plasma membrane upon stimulation by chemoattractants and during transendothelial migration (Bohnsack et al, 1995; Hendey et al, 1996; Loike et al, 1999; Roussel and Gingras, 1997). Leukocyte locomotion requires the continuous formation of new adhesive contacts at the cell front, while the cell rear detaches from the adhesive substrate (Lauffenburger and Horwitz, 1996). During neutrophil locomotion on vitronectin, αvβ3 integrins, which are mainly expressed at the leading edge, have been shown to be endocytosed when reaching the cell rear and finally recycled at the front (Lawson and Maxfield, 1995). This αvβ3 polarized distribution involves the activation of the (Ca2+)i-dependent phosphatase calcineurin, possibly resulting in integrin de-activation at the cell rear (Hendey et al, 1996). Similarly, a tyrosine-containing motif of the β2 integrin cytoplasmic domain, involved in spontaneous receptor recycling, is required for cell migration (Fabbri et al, 1999). Detachment from the adhesion substrate could also be favored by the clustering in the cell rear of anti-adhesive membrane molecules, such as leukosialin (CD43) (Seveau et al, 2000). Dynamic 3-D imaging of neutrophils migrating through the matrix has revealed that cells crawl along matrix fibers, sometimes pulling elastic portions of these fibers to move through openings, but with no proteolytic degradation of the matrix (Mandeville et al, 1997). Such proteolytic degradation appears, however, to be required for the crossing of basement membranes (Huber and Weiss, 1989; Mandeville et al, 1997).

Signaling by Chemoattractants

Neutrophils display multiple structurally related receptors for chemoattractants that can trigger adhesion, direct cell migration, and promote degranulation and oxidative responses. These G-protein-coupled seven-transmembrane glycoproteins, also called “serpentines,” include receptors for complement C5a, for formylpeptides, for the PAF, for leukotriene B4 (Yokomizo et al, 1997), and receptors for C-X-C or α-chemokines (mainly CXCR1 and CXCR2 receptors for IL-8) (Premack and Schall, 1996). Ligation of chemoattractants to such receptors activates phospholipases, via heterodimeric G proteins, resulting in intracellular Ca2+ release, Ca2+ channel opening, and activation of conventional proteine kinase C isoforms (Bokoch, 1995; Premack and Schall, 1996). Tyrosine kinases (mainly Lyn of the Srk-family) (Berton, 1999a; Ptasznik et al, 1996; Welch and Maridonneau-Parini, 1997) and the GTP-binding protein Ras (Worthen et al, 1994) are also activated. Ras activation triggers the MAPK/ERK cascade, which appears to be involved in various chemoattractant-induced neutrophil functions (Krump et al, 1997; Nick et al, 1997; Pillinger et al, 1996). Activation of small GTP-binding proteins of the Ras, Rac, and Rho families regulate actin-dependent processes such as membrane ruffling, formation of filopodia and stress fibers, mediating cell adhesion and motility (Benard et al, 1999; Cox et al, 1997; Nobes and Hall, 1999). Moreover, Rho family members relay signals from chemokine receptors to the outside-in activation of integrins. The Ca++- and DAG-independent protein kinase C-ζ has recently been proposed as a downstream effector of Rho signaling in this process (Laudanna et al, 1996, 1998). Finally, chemoattractant receptors, via their coupled G-protein heterodimers, activate PI3-Kinase, which is involved in the pathways leading to degranulation and NADPH-oxidase activation (Klippel et al, 1996; Okada et al, 1994; Thelen and Didichenko, 1997). The role of PI3-Kinase in neutrophil adhesion promoted by G-protein-coupled receptors is not clearly defined (Akasaki et al, 1999; Shimizu and Huntiii, 1996). Specific PI3-Kinase inhibitors block chemoattractant-induced neutrophil locomotion or homotypic aggregation, but have no effect on integrin CD11b/CD18 expression and activation triggered by these agonists (Capodici et al, 1998; Jones et al, 1998; Niggli and Keller, 1997).

Transepithelial Migration

In many inflammatory diseases (gastrointestinal, respiratory, urinary) neutrophils finally transmigrate across a polarized epithelium to accumulate within a lumen (Parkos, 1997). Transepithelial migration involves a disruption of intercellular tight junctions, which modifies the epithelial barrier and allows the entry of noxious lumenal contents and microorganisms. Complex signaling events lead to cortical restructuration of epithelial F-actin (Hofman et al, 1996) and the reversible disruption of tight junctions, followed by their resealing after passage of neutrophils (Nash et al, 1987; Parsons et al, 1987). The basolateral-to-apical crossing of an epithelium layer, which is often more than two times higher than the size of a neutrophil, involves a complex series of adhesive and de-adhesive events driven by mechanical forces—as shown by neutrophils squeezing through tight junctions and triggered by potent chemoattractants. Specifically, N-formyl peptides are released by bacteria in the lumen and transported across epithelial cells (Chadwick et al, 1988; Merlin et al, 1998), while chemokines such as IL-8 are secreted by infected epithelia on their basolateral side (Eckmann et al, 1993; Kunkel et al, 1991; Richman-Eisenstat et al, 1993). Finally, recent studies with Salmonella typhi have shown that pathogen interactions with the lumenal epithelial surface result in the release of an additional, currently undefined, “transcellular” chemotactic factor(s), which could drive neutrophil migration across epithelium (McCormick et al, 1998). Neutrophil transepithelial migration is mediated by β2 integrin CD11b/CD18 interaction with unknown epithelial ligand(s) distinct from ICAM-1, which might include members of the proteoglycan family (Parkos, 1997). Integrin-associated protein CD47 appears to be involved in neutrophil transepithelial migration as shown by the delayed neutrophil recruitment to infectious sites in the presence of anti-CD47 blocking antibodies or in CD47 knock-out mice (Lindberg et al, 1996; Parkos, 1997). Although this could suggest involvement of αvβ3 integrin, whose functions are regulated by CD47, no evidence has been reported for a role of β3 integrins in transepithelial migration.

Finally, neutrophils may modify the epithelial electrolyte secretion and thus hydration of lumenal surfaces. Indeed, upon activation by bacterial products in the lumen, neutrophils release 5′AMP, which, in the intestine, has been shown to be rapidly converted to adenosine and to interact with an A2b type epithelial receptor, resulting in vectorial secretion of chloride ions into the lumenal compartment (Parkos, 1997; Strohmeier et al, 1995).

I.B. Phagocytosis, Degranulation, and Bacteria Killing

I.B.1. Phagocytosis

Neutrophil phagocytosis involves two different receptor classes, FcγReceptors—FcγRIIA (CD32), and FcγRIIIB (CD16)—and complement receptors CR1 (CD35) and CR3 (or CD11b/CD18 integrin). Among these, the functional phagocytic receptors are FcγRII and CR3, while CR1 and FcγRIIIB appear mostly as co-receptors facilitating the function of the former receptors. Signaling pathways triggered by these two classes of receptors are different, as are the phagocytic processes themselves.

The ingestion of IgG-coated targets is promoted by the aggregation of FcγRII receptors and the phosphorylation of their cytoplasmic ITAMS (immunoreceptor tyrosine-based activation motifs) via the activation of Src-tyrosine kinases. Src kinase (Hck, Fgr, Lyn)-deficient mice exhibit poor Syk activation upon FcγR engagement, which results in a delay in phagocytosis (Crowley et al, 1997). Phosphorylated ITAMS indeed serve as docking sites for SH2 domains of Syk tyrosine kinase, which triggers various pathways involving the activation of PI3-kinase and of Rho proteins. As a result of Rho protein activation, membrane protrusions extend over the surface of the opsonized particle to form a “phagocytic cup” which engulfs the particle (Greenberg et al, 1996; Massol et al, 1998; Swanson and Baer, 1995). Phagocytes from Syk-deficient mice, or treated with a PI-3 kinase inhibitor, form actin-rich phagocytic cups that fail to proceed to particle engulfment (Crowley et al, 1997; Kiefer et al, 1998). Analysis of transfectants expressing GTPases defective in binding guanine nucleotides allowed to specify the involvement of Rho proteins: RhoA appears to be involved in the early F-actin recruitment and phagocytic cup formation, but may not be absolutely required for FcR-mediated phagocytosis (Caron and Hall, 1998; Hackam et al, 1997); CDC42 would regulate the extension of membrane over the particle edges, and Rac1, together with PI3-kinase, would allow membrane fusion and the final closure of the phagocytic cup (Cox et al, 1997; Massol et al, 1998). PI3-kinase is indeed involved in the myosin-induced “purse-string-like” contraction of pseudopods that closes phagosomes (Swanson et al, 1999).

The exact role of the FcγRIII receptor, anchored via a C-terminus-linked GPI moiety in the neutrophil membrane, is not yet clear. It has been recently proposed that, upon cross-linking by immunoglobulin ligands, FcγRIII receptors would recruit FcγRII receptors in “signaling raft-like” membrane domains and allow the clustering of ITAMS (Chuang et al, 2000).

Phagocytosis of C3bi-opsonized targets by complement receptor 3 (CR3) involves a different process: complement-opsonized targets sink into the cell, which produces little protrusions. CR3-mediated phagocytosis has recently been shown, in macrophages, to involve Rho but neither Rac nor Cdc42 (Caron and Hall, 1998). Unlike FcγR-mediated phagocytosis, the ingestion of C3bi-opsonized particles occurs independently of a rise of cytosolic-free Ca++ and of increased inositol phosphate production (Fällman et al, 1989). While FcR-mediated phagocytosis is accompanied by the activation of the respiratory burst and by the production of arachidonic metabolites and cytokines, this does not occur during C3bi-dependent uptake (Wright and Silverstein, 1983; Yamamoto and Johnston, 1984).

Adhesion of neutrophil CR1 and CR3 to particles exclusively coated with C3b/iC3b is not sufficient to promote phagocytosis, unless neutrophils are activated by PMA or by formyl-peptides and a contact with fibronectin or laminin (Brown, 1986; Wright and Meyer, 1986). These stimuli result in the phosphorylation of CR1 and trigger the “inside-out” signaling that activates CR3 binding capacity. Cooperativity between Fcγ- and complement-receptors occurs when C3b/iC3b-bearing targets are also opsonized by antibodies or display glycosylated CR3 ligands (Ehlenberger and Nussenzweig, 1977). Cross-talks between phagocytic receptors are suggested by the observation that neutrophils from CR3-deficient (CD18-deficient LAD) patients display an impaired antibody-dependent phagocytosis (Dana et al, 1984) and that FcγRIIIB interacts in cis with CR3, via a lectin-carbohydrate interaction (Todd and Petty, 1997). Complex signaling pathways promoted by the engulfment of opsonized targets lead to the fusion of protease-rich granules with the phagosome and the triggering of the oxydative burst, as described in detail below.

I.B.2. Degranulation

Granule Biogenesis

Neutrophil-derived microbicidal molecules are packed in granules that are released upon cell activation (Elsbach, 1998; Lehrer and Ganz, 1999; Spitznagel, 1990) (Fig. 2). Granule biogenesis follows the granulocyte differentiation pathway (Borregaard and Cowland, 1997). The azurophilic granules first emerge at the stage of promyelocytes and contain myeloperoxidase, serine proteases, and antibiotic proteins (Fouret et al, 1989). Azurophil granules are thus considered as the true microbicidal compartment mobilized upon phagocytosis. Still, morphological heterogeneity has been described within azurophil granules (Egesten et al, 1994). According to the observation that azurophil granules do not contain lysosome-associated membrane proteins (LAMP), they cannot be classified as lysosome, but rather appear to have the functional characteristics of a regulated secretory granule (Cieutat et al, 1998). Later in differentiation, at the metamyelocyte stage, specific granules containing lactoferrin and collagenase emerge, followed by the tertiary granule population containing gelatinase. A fourth type of granules, called the secretory vesicles, appears at the stage of mature neutrophil. Their origin might be endocytic, because they contain plasma proteins such as albumin. However, recent findings have shown that this strict compartmentalization is not a dogma. For instance, proteinase 3, a serine protease described in azurophil granules is also localized in the membrane of secretory vesicles, the most mobilizable compartment of neutrophils (Witko-Sarsat et al, 1999a). The mechanisms underlying the secretion of the four morphologically distinct populations of granules may be under separate control. The order of exocytosis observed after ionophore-induced progressive elevation of cytosolic calcium was secretory vesicles, gelatinase granules, specific granules, and lastly azurophilic granules (Sengelov et al, 1993).

Figure 2
figure 2

Electron microscopy showing the various intracytoplasmic granules of a resting neutrophil. Resting neutrophils were first fixed in 1.25% glutaraldehyde in 0.1 m phosphate buffer followed by an incubation in diaminobenzidine to label peroxidase-positive granules. Neutrophils were then post-fixed with OsO4. Peroxidase-positive granules are azurophil granules (or primary granules, pg), which appear as large dark granules. Specific granules (or secondary granules, sg) are smaller in size. Nucleus (N); centriole (ce); mitochondri (m). (Courtesy of Dr Elizabeth Cramer, INSERM U474, Cochin Hospital, Paris.)

Two inherited defects affect neutrophil granule structure, as reviewed in Malech and Nauseef (1997). The first is the specific granule deficiency, which is a rare congenital disorder marked by frequent and severe bacterial infections. Neutrophils are characterized by a lack of specific granules and defensins, abnormalities in neutrophil migration, and impaired bactericidal activity. The molecular basis of this defect has been recently characterized in one patient and involves a deletion in the CCAAT/enhancer binding protein ([C/EBP]ε) gene, encoding for a member of the leucine zipper family of transcription factor primarily expressed in myeloid cells (Lekstrom-Himes et al, 1999). The second inherited granule deficiency is the Chediak-Higashi syndrome (CHS), which is a rare autosomal recessive disorder associated with an immune deficiency leading to increase susceptibility to infection and a life-threatening lymphoma-like syndrome. A lack of natural killer cell function and a neutropenia may be found, with a prominent defect in formation of neutrophil granules. The gene for CHS has been cloned, based on its homology to the murine gene responsible for the Beige phenotype that corresponds to the mouse CHS (Nagle et al, 1996). The CHS protein has structural features homologous to a yeast vacuolar sorting protein thought to be associated with vesicle transport. In fact, the CHS abnormalities are not restricted to neutrophils and all cell types show some oversized lysosomes, the disease affecting thus several organ systems.

Mechanisms of Degranulation

Intracellular transport of proteins, their delivery to various compartments, and their eventual secretion in the extracellular milieu represent crucial mechanisms in the activity of neutrophils. Degranulation of vesicles into the phagolysosome or in the extracellular space are key events for microbicidal activity (Berton, 1999b). With the exception of secretory vesicles, which are of endocytic origin, fusion of neutrophil granules with the plasma membrane represents a heterotypic fusion event. This fusion involves protein-protein interactions that dock a vesicle to its final destination and proteins that favor the interaction between the phospholipid bilayer of the vesicle and its target membrane.

I.B.3. Microbicidal Molecules

The antimicrobial efficiency of human neutrophils depends on two concurrent events occurring in the nascent phagolysosome of stimulated neutrophils: the generation of ROS by assembly and activation of the NADPH-dependent oxidase and the release of enzymatic or antimicrobial protein content in the granules. These responses are triggered by numerous agonists promoting adhesion or by phagocytic targets (Fig. 3).

Figure 3
figure 3

Neutrophil effector mechanisms involved in the defense against pathogens and in the inflammatory process. Neutrophil effector systems are mobilized following phagocytosis of a pathogen. Complement opsonins C3b and C4b are recognized by CR1 and CR3. IgG opsonins are recognized via the immunoglobulin receptors (FcγR). The first microbicidal pathway is the oxidative reponse, which consists of the production of radical oxygen species following NADPH-oxidase complex activation, including superoxide anion (O2), hydrogen peroxide (H2O2), and, via myeloperoxidase, hypochlorous acid (HOCl) and chloramines. The second microbicidal pathway is non-oxygen–dependent and consists of the release in the phagolysosome or in the extracellular medium of preformed proteins stored in granules. Serine proteases, antibiotic proteins, as well as myeloperoxidase are contained in azurophilic granules. Metalloproteinases (collagenase and gelatinase) and antimicrobial proteins (lactoferrin and cathelicidin) are contained in specific granules. Gelatinase is also contained in tertiary granules, also called gelatinase granules. (Adapted from Witko-Sarsat and Descamps-Latscha, 1994.)

NADPH-Derived Oxidants

The activation of the oxidative metabolism, known as the respiratory burst, first involves NADPH oxidase, an enzymatic complex composed of cytosolic (p40phox, p47phox, and p67phox) and membrane proteins (p22phox and gp91phox), which constitute a heterodimeric flavohemoprotein known as cytochrome b558, as reviewed in Babior (1999), Clark (1999), and Deleo and Quinn (1996). Two low-molecular weight guanine nucleotide-binding proteins are involved: Rac2, which is located in the cytoplasm in a dimeric complex with RhoGDI (Guanine nucleotide Dissociation Inhibitor), and Rap1A, which is located in membranes. Upon activation of neutrophils, p47phox becomes phosphorylated and cytosolic components migrate to the plasma membrane where they associate with cytochrome b558 to assemble the active oxidase. This enzymatic complex is thus able to generate superoxide anion (O2), which can dismutate into H2O2 (Babior, 1984; Nathan, 1987). There are three intermediates in the reduction of O2 to H2O2, namely O2, H2O2, and the hydroxyl radical (OH°), which are formed by successive one electron additions. Despite numerous studies, the formation of OH° in phagocytes is still controversial (Britigan et al, 1986; Rosen and Klebanoff, 1979; Tauber and Babior, 1977; Ward et al, 1983). The formation of singlet oxygen appears to be an important event in the microbicidal potential of neutrophils (Allen et al, 1972; Harrison et al, 1978).

Much of what is known about the NADPH oxidase has come from studies of patients deficient in the system, who have chronic granulomatous disease (CGD). Due to a genetic defect in any of the four phox subunit genes, phagocytes of CGD patients fail to mount a respiratory burst (Gallin et al, 1991; Segal and Abo, 1993). With regard to molecular defect, 60% to 80% of cases are due to the X-linked gp91phox deficiency, 30% of cases are due to the autosomal recessive p47phox, and 2% to 3% are due to the autosomal recessive p22phox or p67phox deficiency (Roos et al, 1996). Although their phagocytic capacity is normal, CGD phagocytes are incapable of producing ROS and subsequently of killing ingested targets. Patients with CGD experience recurrent and often life-threatening bacterial and fungal infections, as well as a granulomatous response in affected tissues. One exception to this rule is the normal killing of microorganisms that themselves produce significant quantities of H2O2 (eg, pneumococci), thereby supplying a missing ingredient used by the CGD neutrophils to reconstitute the activity of the myeloperoxidase- H2O2-halide antimicrobial system that we describe below. The severe clinical picture observed in CGD clearly demonstrates the vital importance of NADPH oxidase in host defense against infection. Two distinct murine animal models of CGD-p47phox gene knockout (Jackson et al, 1995) and gp91phox gene knockout (Pollock et al, 1995) have been developed and show a phenotype resembling human CGD. Circulating neutrophils in these mice lack NADPH activity and CGD mice die prematurely from bacterial or fungal infections. These animal models are potentially interesting for the investigation of gene therapy protocols for CGD (Bjorgvinsdottir et al, 1997; Dinauer et al, 1999; Mardiney et al, 1997). The human gene therapy for the correction of the p47phox-deficient form of CGD is currently under study (Malech, 1999). The cell targets for p47phox gene transfer are hematopoietic progenitor/stem cells (CD34+) using retroviral vectors (Malech et al, 1997). In a phase I clinical trial of ex vivo gene therapy of p47phox-deficient CGD using retroviral vectors, prolonged production (2–6 months) of a low number (1:5000) of oxidase-normal neutrophils was achieved. Although the correction of the defect was transient, this therapy might provide beneficial effects in increasing host defense potential.

The H2O2-Myeloperoxidase System

The generation of superoxide anion via the activation of NADPH oxidase is the starting material for the production of a vast assortment of reactive oxidants, including halogenated oxidants generated through the myeloperoxidase (MPO) pathway (Klebanoff, 1969, 1999). MPO is a heme protein present in azurophil granules of neutrophils and monocytes, which is released upon cell activation into the phagolysosome or into the extracellular space. MPO amplifies the toxic potential of H2O2 by producing reactive intermediates. At plasma concentrations of chloride ion, the major product of MPO is hypochlorous acid (HOCl). This potent oxidant chlorinates electron-rich substrates and oxidatively bleaches heme proteins and nucleotides. MPO has a wide range of substrates leading to a wide variety of byproducts. Amino acids, especially taurine, can be chlorinated to yield chloramines, the so-called long-lived oxidants (Rotrosen, 1992; Winterbourn, 1990). Several studies have demonstrated that the MPO-H2O2 system results in the formation of tyrosyl radical and chlorination products, the generation of tyrosine peroxide, reactive aldehydes (Hazen et al, 1998), and the oxidation of serum proteins and lipoproteins (Heinecke, 1999; Leeuwenburgh et al, 1997). Interestingly, MPO can utilize nitrite and hydrogen peroxide as substrate to catalyze tyrosine nitration in proteins (Sampson et al, 1998) and can react with peroxinitrite (Floris et al, 1993; Podrez et al, 1999), thus providing a link with the nitric oxide-synthase system, as described below.

The importance of MPO in microbicidal activity has historically been the focus of studies. Given the important role of this system in the antimicrobial activity of circulating neutrophils, it seemed predictable that inherited deficiency of MPO would be severe and likely to be uncommon in the general population. Several inherited MPO deficiencies were described with severe infections with Candida (Nauseef, 1998). Likewise, knock-out MPO mice, which have recently been described (Koyama et al, 1999), show increased susceptibility to Candida infection. However, the use of flow cytochemistry to quantitate leukocytes in clinical samples revealed a large population of asymptomatic subjects with MPO deficiency (Nauseef et al, 1996; Petrides, 1998). MPO-derived oxidants have also been implicated in other processes unrelated to host defense, including carcinogenesis ( London et al, 1997; Pero et al, 1996), atherosclerosis (Heinecke, 1999), and chronic renal failure (Witko-Sarsat et al, 1998). In this latter condition, high levels of advanced oxidized protein products, derived from MPO activity, circulate in the plasma of uremic patients and can mediate inflammatory functions.

One should mention that an allelic polymorphism in the SP1 binding site in MPO promoter (Sp/Sp genotype leading to higher MPO mRNA expression) has been recently associated with an increased frequency of acute promyelocytic leukemia (Reynolds et al, 1997), as well as with gender-specific risk for Alzheimer’s disease (Reynolds et al, 1999). In a study pointing out the influence of host defense molecule polymorphisms in CGD complications, it has clearly been demonstrated that CGD patients having the Sp/Sp genotype had significantly increased frequency of gastrointestinal complications (Foster et al, 1998). It thus appears that MPO-derived oxidants have important regulatory functions in various pathophysiological conditions, not restricted to the field of inflammation. Although neutrophils remain the main source of MPO, it is also present in monocytes that could also be involved in MPO-mediated biological activity.

Nitric Oxide-Synthase-Derived Reactive Nitrogen Intermediates

NO-synthases (NO) are unique among eukaryotic enzymes in being dimeric, calmodulin-dependent or calmodulin-containing cytochrome P450-like hemoproteins that combine reductase and oxygenase catalytic domains in one monomer. They bear both FAD and FMN, and carry out a 5-electron oxidation of a non-aromatic amino acid (L-arginine) with the aid of tetrahydrobiopterin (Marletta, 1993; Nathan and Xie, 1994). Reactive nitrogen intermediates include nitric oxide (NO), which can react with oxygen to form much stronger oxidants such as nitrogen dioxide (NO2). The direct toxicity of NO is modest, but is greatly enhanced by reacting with superoxide to form peroxynitrite (ONOO-) (Beckman and Koppenol, 1996).

The large amounts of NO produced by murine macrophages may contribute to their microbicidal activity. Because human phagocytes and especially neutrophils, appear to generate so little nitric oxide (Klebanoff and Nathan, 1993; Padgett and Pruett, 1995), the microbicidal role of NO in the human remains controversial (Albina, 1995). However, in urinary tract infection, neutrophils isolated from urine showed a dramatic increase in nitric oxide synthase activity, the major isoform being the inducible NO-synthase (Wheeler et al, 1997). NOS knockout mice have demonstrated the enzyme’s essential contribution to host defense only against a restricted set of pathogens, including Mycobacterium tuberculosis and Leishmania major. Mice doubly deficient in both NO and NADPH-oxidase (gp91phox-knockout) formed massive abscesses containing commensal organisms, mostly enteric bacteria, even when reared under specific pathogen-free conditions with antibiotics, whereas neither parental strain showed such infections. No data on neutrophil functions in these mice are yet available (Shiloh et al, 1999).

Granule Proteins

Antimicrobial Proteins Neutrophils use an array of antimicrobial peptides and proteins to destroy invading microorganisms. This has been reviewed in (Lehrer and Ganz, 1999; Levy, 1996). The azurophilic granules contain the majority of the antimicrobial proteins that should be released in the phagolysosome. One of the most active participants in host defense against Gram-negative bacterial infections is bactericidal/permeability increasing protein (BPI), a 50- kDa protein stored in azurophil granules but also expressed at the plasma membrane of neutrophils ( Elsbach, 1998; Weersink et al, 1993). The selective toxicity of BPI for Gram-negative bacteria relies on the binding capacity of its 21–25 kDa aminoterminal fragment to LPS (Ooi et al, 1987). This property of BPI has prompted preclinical and subsequent clinical testing of recombinant amino-terminal fragments of BPI. Phase I trials in healthy human volunteers and multiple clinical trials have now well proven that BPI is neither toxic nor immunogenic in normal individuals or in seriously ill patients (Von Der Mohlen et al, 1995). Trials of BPI administration have been or are being performed in various pathologic conditions including severe pediatric meningococcemia, hemorrhagic trauma, peritoneal infections, and cystic fibrosis (Elsbach, 1998).

Another important group of antimicrobial peptides is the group of beta-sheet defensins that comprises four members: HNP1 to NHP4 (Ganz and Lehrer, 1995). Defensins are small cationic, antibiotic peptides that contain six cysteines in disulfide linkage. They are active against Gram-positive and Gram-negative bacteria and act by inducing microbial membrane permeabilization. Besides their bactericidal role, defensins appear to have the ability to regulate the inflammatory process through binding to protease inhibitors such as alpha-1-antitrypsin and alpha-1-antichymotrypsin. At high concentration, defensins can ablate the inhibitory effect of normal human serum on cathepsin G and human neutrophil elastase (Panyutich et al, 1995). Defensins are mitogenic for fibroblasts, thus suggesting a role in wound healing (Murphy et al, 1993).

Specific granules also contain antimicrobial molecules destined predominantly for extracellular release. Among them, hCAP-18 is a cathelicidin. This family of antimicrobial peptides has a conserved N-terminal precursor segment named “cathelin” and a highly variable carboxy-terminal microbicidal peptide. In neutrophils, they are stored in specific granules in an inactive form (Cowland et al, 1995; Sorensen et al, 1997). LL-37, a 37-residue peptide at the carboxy-terminal domain of hCAP-18, is released after processing by elastase (Zanetti et al, 1997), thus pointing to a cooperation with azurophil granules. Synthetic LL-37 was active against Gram-negative and Gram-positive microorganisms (Turner et al, 1998). Interestingly, LL-37 may act synergistically with lactoferrin, which indeed possesses its own antimicrobial activity (Zanetti et al, 1997). Lactoferrin is a well studied example of an antimicrobial peptide generated by limited proteolysis of a longer protein. Lactoferrin is an 80 kDa iron-binding protein whose antimicrobial amino-terminal domain (lactoferricin) is liberated by pepsin cleavage (Hwang et al, 1998). Other proteins stored in specific granules, such as phospholipase A2 and lysozyme, may account for bactericidal activities, as demonstrated in biological fluids (Harwig et al, 1995; Weinrauch et al, 1996).

Secretory leukoproteinase inhibitor (SLPI) is a 12 kD nonglycosylated protein which is present in neutrophils (Sallenave et al, 1997) and produced by cells of mucosal surfaces. This two-domain polypeptide has a carboxy-terminal domain expressing antiproteinase activity, whereas its aminoterminal domain has broad-spectrum antimicrobial properties (Tomee et al, 1997). LPS and lipoteichoic acid from Gram-positive bacteria cell walls have also been shown to induce SLPI synthesis in murine macrophages (Jin et al, 1998).

Studies of antimicrobial peptides are providing new insights into the complex interactions between microbes and their hosts. Characterization of structural antimicrobial motifs of peptides isolated from neutrophils or from other sources (plants, insects) might offer novel templates for pharmaceutical compounds that could be effective against increasingly resistant microbes. For instance, protegrins are stored in granules from pig neutrophils as cathelin-containing precursors and are processed by elastase in a 16–18 amino-acid active peptide (Shi and Ganz, 1998). Protegrins display a strong toxic activity against bacteria, fungi, and enveloped viruses (Cho et al, 1998; Lehrer and Ganz, 1999). These small peptides are particularly well suited for chemical peptide synthesis. A clinical trial on prevention of oral mucositis caused by cancer therapy has been undertaken (Ganz and Lehrer, 1999).

Proteases Neutrophil-derived proteases have the ability to degrade the majority of extracellular matrix components and, as a result, play fundamental roles in physiological processes, reviewed in Owen and Campbell (1999). According to the biochemistry of the active site, four distinct classes of proteinases can be identified, namely, serine-proteases, metalloproteases, thiol-proteases, and aspartate proteases. The two first classes, most active at neutral pH, play a major role in degradation of extracellular proteins, whereas the two others, most active at acidic pH, are involved in intracellular protein digestion.

Serine Proteinases Neutrophil-derived proteases are packed in azurophil granules. Serine proteases are a large family of enzymes characterized by their active site, the so-called “catalytic triad” composed of histidine, aspartic acid, and serine. The group of the neutral serine protease homologs stored in the azurophilic granules of the neutrophil includes cathepsin G (Salvesen et al, 1987), elastase (Takahashi et al, 1988), proteinase 3 (Bories et al, 1989; Campanelli et al, 1990b), and the enzymatically inactive azurocidin or CAP-37 (Almeida et al, 1991; Campanelli et al, 1990a; Morgan et al, 1991), which are cationic glycoproteins of similar size (25–29 kD) which have been cloned. Neutrophil serine proteinases exhibit sequence homologies between each other and with T cell proteases, human lymphocyte proteases, granzyme B, and rat mast cell proteases (Hudig et al, 1993). Genomic cloning has revealed that neutrophil elastase, proteinase 3, and azurocidin genes form a cluster of genes, located in the terminal region of the short arm of chromosome 19 and coordinately regulated in the promonocytic cell line U937 during induced terminal differentiation (Sturrock et al, 1992; Zimmer et al, 1992). The gene of cathepsin G belongs to another cluster of genes encoding hematopoietic serine proteases along with granzyme H and granzyme B genes on chromosome 14q11.2 (Haddad et al, 1991; Hanson et al, 1990).

Elastase and PR3 display very similar patterns of proteolytic activities (Kam et al, 1992). They are both capable of cleaving insoluble elastin and a variety of matrix proteins, including fibronectin, laminin, vitronectin, and collagen type IV They show minimal activity against interstitial collagens, type I and III (Rao et al, 1991). The main physiologic defense against elastase and proteinase 3 is plasma α1-antitrypsin (α1-AT) and α2-macroglobulin (Mason et al, 1991; Travis and Salvesen, 1983). Several intracellular elastase inhibitors have been characterized, including the human monocyte/neutrophil elastase inhibitor, a 42 kD glycoprotein, and member of the serpin family (Remold-O’Donnell et al, 1992). Cathepsin G displays a different pattern of inhibition, inasmuch as the serpin which accounts for the greatest inhibition is α1-antichymotrypsin. Elastase and PR3 have been implicated in pulmonary pathology including emphysema (Janoff et al, 1977; Kao et al, 1988), chronic bronchitis, and cystic fibrosis (last section of this review). Current concepts on the pathogenesis of emphysema emphasize the role of unrestrained proteolytic activity in the lung extracellular matrix. Because α1-AT provides almost all the protective screen of the lower respiratory tract against neutrophil elastase, emphysema might result from inactive α1-PI unable to inhibit neutrophil elastase in the lung (Janoff, 1985; Stockley, 1987). Of particular interest is α1-AT deficiency, an autosomal hereditary disorder characterized by reduced levels of α1-AT in plasma and lung fluids, thereby leading to unopposed proteinase activity and culminating in pulmonary emphysema (Crystal et al, 1989). In this condition, quantum proteolytic events in neutrophils are large and prolonged, leading directly to an increased risk of tissue injury in the immediate vicinity of neutrophils (Campbell et al, 1999). SLPI may be important for local anti-elastase protection in the lung, but it has antiproteinase activity against neutrophil elastase and cathepsin G, but not against proteinase 3 (Rao et al, 1993).

Neutrophil serine proteinases, especially cathepsin G, can mediate platelet aggregation (Renesto and Chignard, 1993). Serine proteinase homologs have been shown to exert immunomudulatory effects. Azurocidin (CAP-37) has a chemotactic activity for monocytes and stimulates PKC in endothelial cells (Pereira et al, 1990, 1996). Elastase can cleave monocyte CD14, thus inhibiting lipopolysaccharide-mediated cell activation (Le-Barillec et al, 1999). Both elastase and proteinase 3 are able to induce the synthesis of IL-8 in endothelial cells (Berger et al, 1996) and to process IL8 into more potent N-terminal truncated forms of IL8 (Padrines et al, 1994). Similarly, PR3 can process the membrane-bound TNF-α precursor into its mature form (Robache-Gallea et al, 1995). Elastase and PR3 could play a significant role in pulmonary inflammation through their secretagogue activity on goblet cells. They are so far the most potent agonist for goblet cell secretory activity that is dependent on their catalytic activity (Sommerhoff et al, 1990; Witko-Sarsat et al, 1999b).

Several groups have investigated the conditions required for plasma membrane expression of serine proteases that could be of great relevance in the pathophysiology of inflammatory disease. Both elastase and cathepsin G can be expressed at the plasma membrane of activated neutrophils, where they appear to be bound via an ionic interaction (Owen et al, 1995). In contrast, proteinase 3, also expressed at the plasma membrane, is bound via a covalent interaction (Witko-Sarsat et al, 1999a). It has been described that in the case of elastase and cathepsin G, this tight association of serine-proteinases with the plasma membrane confers resistance to physiological inhibitors such as α1-AT via a steric mechanism. The use of inhibitors seems to be the only means to regulate the activity of neutrophil serine proteinases, inasmuch as these proteases are probably functional in their packaged forms, and specific serine protease inhibitors could be of potential interest as anti-inflammatory drugs.

Among purified neutrophil-derived proteases, those that so far appear to have significant in vitro antimicrobial potential independent of their enzymatic action are cathepsin G (Bangalore et al, 1990; Shafer et al, 1991), PR3, and azurocidin (Campanelli et al, 1990b). Their antimicrobial activity is widely distributed because they are active against Gram-positive, Gram-negative, yeast, and fungi. Although in vitro studies did not reveal antimicrobial activity of elastase, knock-out elastase mice show an impaired host defense against Gram-negative bacterial sepsis (Belaaouaj et al, 1998). The discrepancy between in vitro and in vivo studies suggests that the antimicrobial potential of elastase could be via an indirect mechanism involving the proteolytic activation of antimicrobial peptides synthesized as proforms, such as defensin (Daher et al, 1988) and cathelicidin (Zanetti et al, 1997).

Metalloproteinases Matrix metalloproteinases (MMPs) constitute a family of closely related enzymes that play important roles in a variety of physiological and pathological processes of matrix degradation (Owen and Campbell, 1999; Shapiro, 1998; Weiss and Peppin, 1986). Neutrophils contain metalloproteinases such as collagenase (MMP-8), which are stored in the specific granules and specifically cleave type I collagen, whereas the 92 kDa-gelatinase (MMP-9), which is stored in separate secretory vesicles, degrades native type V collagen. In addition, neutrophil metalloproteinases have been involved in the modulation of neutrophil functions such as the shedding of Fcγ receptor (Middelhoven et al, 1997).

Comparison of the primary structures of MMP shows that they are structurally homologous with defined functional domains (Sanchez-Lopez et al, 1988; Wilhelm et al, 1989). All of these enzymes contain an essential catalytic zinc-binding domain, an NH2-terminal domain that preserves the latent state of the enzyme and a COOH-terminal domain that plays a major role in substrate specificity.

Metalloproteinases are stored in latent form within granules. When neutrophils are stimulated to release them, these latent enzymes must be activated before they can attack their substrate, by mechanisms that are still not completely understood. In vitro experiments have shown the involvement of serine proteinases, chlorinated oxidants, chemicals such as organomercurials, and SH-modifying agents. Latent collagenase can be directly activated by HOCl, whereas progelatinase seems to require both oxidant- and serine protease-dependent pathways (Knauper et al, 1996; Murphy et al, 1992; Peppin and Weiss, 1986). Secretion of metalloproteinase inhibitors in an inactive precursor form is an important feature that regulates their activity in the extracellular milieu.

Further regulation of the activity of metalloproteinases in the extracellular milieu is achieved by specific inhibitors interacting with the activated enzymes. Three tissue inhibitors of metalloproteases (TIMP) have been characterized and cloned (Murphy and Docherty, 1992). TIMP-1 and TIMP-2 are soluble inhibitors that are present in many tissues and biological fluids. They are secreted by several mammalian cell types, including fibroblasts, endothelial cells, smooth muscle cells, and chondrocytes (Cawston et al, 1981; Declerck et al, 1991; Stricklin and Welgus, 1983). TIMP-3 is an insoluble inhibitor that is bound to extracellular matrix (Moses, 1997). The activity of TIMP can be oxidatively regulated and is blocked by various serine proteinases, including human neutrophil elastase, trypsin, and α-chymotrypsin (Okada et al, 1988). In addition, metalloproteinases can be inhibited by α2-macroglobulin and by complexing agents such as EDTA. Targeted disruption of metalloprotease genes has pointed out specific roles for metalloproteases in physiological processes (Shapiro, 1998). For instance, gelatinase B knock-out mice show an impaired primary angiogenesis in bone growth plates and a resistance to bullous pemphigoid. No investigation of neutrophil functions has been performed in this model (Vu et al, 1998).

I.C. Cytokine Synthesis

A great body of evidence has accumulated that the human neutrophil is both a target and a source of various proinflammatory cytokines, chemokines, and growth factors, and therefore often exerts its proinflammatory functions through an autoregulatory pathway. Neutrophils are exquisite targets of proinflammatory cytokines, eg, IL-1 and TNF-α, of chemokines such as IL-8, and growth factors such as granulocyte/monocyte colony stimulating factor (G-CSF and GM-CSF). Indeed, these cytokines have been shown to amplify several functions of neutrophils, including their capacity of adhering to endothelial cells and to produce ROS, as described above; likewise, chemokines act as potent attractants and favor their orientated migration toward the inflammatory site. An important issue is that both cytokines and chemokines may also act as priming agents of neutrophils.

The concept that neutrophils can be a source of cytokines has only recently emerged and has recently been reviewed (Cassatella, 1999). Indeed, neutrophils were long considered to be devoid of transcriptional activity and capable of performing no or little protein synthesis. However, convincing molecular evidence has now been afforded that neutrophils either constitutively or in an inducible manner can synthesize and release a wide range of proinflammatory cytokines, antiinflammatory cytokines, and other cytokines and growth factors (Table 1). The production of cytokines by activated neutrophils is striking in its diversity. However, it remains much lower in its degree than that produced by the mononuclear phagocytes, namely the monocytes (Cassatella, 1995). This important discrepancy between the two cell types leads to the use of extremely purified neutrophil preparations when studying their cytokine production. However, this appears less evident in vivo if one considers that (i) the number of circulating neutrophils is almost 20 times higher than that of monocytes, and (ii) at the site of inflammation, neutrophils are the first to be recruited and largely predominate over monocytes. The production of cytokines is also largely influenced by the stimulating agents and among these, cytokines and bacterial endotoxins (LPS) are the most potent inducers (Table 2). The pattern of cytokines produced by neutrophils also greatly differs depending on the agonist, and for some cytokines co-stimulation by at least two of agonists is required, eg, IFN-γ + LPS in the case of IL-12 (Cassatella et al, 1995). Another important feature is that cytokine production is preceded by a consistent accumulation of the corresponding mRNA cytokines. Finally, the production of cytokines by neutrophils can easily be modulated by immunomodulatory cytokines such as IFN-γ, IL-4, IL-10, and IL-13, suggesting that T helper-1 (Th-1) or Th-2 cells may influence neutrophil cytokine production (Romagnani, 1994).

Table 1 Cytokine Expression by Neutrophils In Vitro
Table 2 Agents Able to Trigger Cytokine Production by Neutrophils

Given the limited space, our description will be restricted to TNFα, IL-1β, and IL-1Ra as representatives of pro- or anti-inflammatory cytokines, and IL-8 as representative of chemokines. Detailed description of the other cytokines produced by neutrophils is given in Cassatella (1999).

I.C.1. TNF-α as a Proinflammatory Cytokine

TNF-α, a homotrimer of 17Kda subunits, was originally described as a product of activated monocytes and macrophages displaying tumoricidal activity. It is a highly pleiotropic cytokine belonging to the superfamily of membrane-anchored and soluble cytokines that are notably involved in T cell-mediated immunity. Although it inhibits the growth of tumor cells, it has an enhancing effect on the proliferation of certain normal cells and has a great variety of nontumoral target cells, for example, monocytes, macrophages, lymphocytes, eosinophils, and neutrophils. TNF-α is involved in septic shock, cachexia, autoimmunity, and inflammatory diseases. Its potent proinflammatory effects mainly result from its capacity to increase expression of endothelial cell adhesion molecules and subsequently promote neutrophil adherence to vascular endothelium (see above). Finally, TNF-α is also a priming agent for neutrophils that notably increases their phagocytosis, degranulation, and oxidative responses.

However, activated neutrophils have been shown to have the capacity to express TNF-α mRNA (Lindemann et al, 1989). Using GM-CSF as a stimulus, no secretion of the related TNF protein was detected. Soon after, the dual observation of expression of TNF-α mRNA and protein secretion was reported with LPS as a stimulating agent. This observation was substantiated by other reports showing that Candida albicans also induces a potent extracellular release of TNF-α (Djeu et al, 1990; Mandi et al, 1991). It is also now generally accepted that cytokines, for example, TNF-α itself, IL-1β, GM-CSF, and IL-2, are also potent inducers of TNF-α mRNA expression and secretion by neutrophils.

As already mentioned, neutrophils are exquisite targets of TNF-α that under certain conditions enhances their expression of adhesion molecules, induces their degranulation and subsequent release of lysosomal enzymes, and primes them for oxidative activity, leading to the production of highly reactive oxygen species. These effects are mediated via TNF-α receptors that have been cloned and are expressed in two types: the type A of 75 kDa (also referred to as p75) and the type B of 55 kDa (p55). Both types are expressed on a wide variety of cells, but it is generally accepted that, whereas the p75 is mainly expressed on cells of myeloid origin, the p55 predominates on epithelioid cells. Even though both receptors display similar architecture, most of the sequence homology is displayed in the extracellular domain and almost no homology is observed in the cytoplasmic domain, suggesting that the two receptors activate distinct intracellular signaling pathways. Soluble forms of p55 and p75 that bind to TNF-α and have similar neutralizing effects have been described (Brockhaus et al, 1990; Tartaglia and Goeddel, 1992). The underlying mechanisms of shedding of TNF receptors are still unclear, but neutrophil elastase has been shown to participate in the shedding of p75 (Porteu et al, 1991). Both soluble receptors are present in blood and urine of normal individuals (Aderka et al, 1992). Elevated circulating levels of these molecules have already been reported in patients with inflammatory diseases such as rheumatoid arthritis (Dayer and Fenner, 1992; Lopez et al, 1995) and in chronic renal failure patients (Descamps-Latscha et al, 1995; Pereira et al, 1994). The biologic functions of these soluble forms of TNF receptors present in plasma are still speculative. One might expect that they play a role in neutralizing circulating TNF-α resulting in a decrease of an inflammatory reaction. However, a role of these receptors as possible “reservoirs” of biologically active TNF-α has also been suggested. In vivo, an imbalance between TNF soluble receptors and TNF-α has been observed in severe meningococcemia (Girardin et al, 1992; Villard et al, 1993).

I.C.2. IL-1 and IL-1 Receptor Antagonist (IL-1-Ra)

The issue of whether or not neutrophils synthesize and secrete IL-1 has been a matter of debate for several years, but molecular studies have now clearly demonstrated that neutrophils indeed express the mRNA of both IL-1α and IL-1β and release the related IL-1 proteins. However, the amount of IL-1 β is up to 10 times more than that of IL-1α. The stimuli listed in Table 2 are in general all capable of inducing IL-1 production by neutrophils, although, among the cytokines, it appears that only IL-1β itself and TNF-α induce IL-1 production by neutrophils. Interestingly, anti-neutrophil cytoplasm autoantibodies (ANCA) have been shown to induce IL-1β mRNA expression in neutrophils. IL-1 is usually released in concert with TNF-α and exerts similar effects on neutrophils (see above).

Two classes of IL-1 receptors (IL-1-RI and IL-1-RII) expressed on a wide variety of cells have been described. IL-1Ra is a 23- to 25-kDa protein made by the same cells as those that produce IL-1. It exerts its inhibitory action on IL-1 by binding to IL-1 receptors without triggering any signal transduction or biological activity. The demonstration of a constitutive secretion of IL-1Ra by neutrophils using GM-CSF combined with TNF-α as agonists (McColl et al, 1992) has been followed by the demonstration that neutrophil-derived products have an inhibitory activity on monocyte derived IL-1 production (Tiku et al, 1986). The expression of IL-1Ra mRNA in LPS-treated neutrophils appears to be greater than that of LPS-treated monocytes and an almost 100-fold excess of IL-1Ra over IL-1 is usually produced by activated neutrophils. More recent studies have shown that IL-4, IL-13, and TGF-β are potent inducers of IL-1Ra mRNA. It has thus become evident that neutrophil-derived IL-1Ra could contribute to modulate the IL-1 induced inflammatory and immune responses.

I.C.3. IL-8 as a Prototype of Chemokines

Chemokines are usually classified as C-X-C or C-C chemokines on the basis of the position of the first two cysteine residues, and IL-8 is a prototype of the C-X-C family (Baggiolini et al, 1994). It was first described as a potent neutrophil chemoattractant and activator (Baggiolini and Clark-Lewis, 1992). It is expressed in response to LPS, mitogens such as PHA, cytokines, eg, TNF-α, and IL-1β or aggregated immune complexes (Deforge et al, 1992). It is secreted by a variety of cells including T lymphocytes, epithelial cells, keratinocytes, fibroblasts, endothelial cells, and neutrophils. Interestingly, IL-8 is the most abundantly secreted cytokine by neutrophils, and on the other hand neutrophils are the primary cellular target of IL-8 (Gainet et al, 1998). The presence of IL-8 mRNA in freshly isolated neutrophils has been widely reported. However, whether it is constitutive or induced by isolation procedure is still debated.

The list of agents capable of triggering IL-8 production by neutrophils never ceases to increase and comprises all the products listed in Table 2. Among these, opsonized zymosan, which involves CR1 and CR3 complement receptors, appears as the most potent agonist of IL-8 production. The potent chemotactic effect of IL-8 on neutrophils is also abundantly documented. Two sulfhydryl groups participate in the binding of the ligand to the receptor and consequently regulate receptor-mediated cell functions (Samanta et al, 1993).

I.C.4. Modulation of Cytokine Expression by Neutrophils

As for other cytokine producing cells, cytokine expression by neutrophils can easily be modulated by the T-cell–derived regulatory cytokines, ie, positively by Th1 type cytokines such as IFNγ, and negatively by Th2 type cytokines such as IL-10, IL-4, and IL-13, reviewed in Cassatella (1999). Given the pathophysiological importance of such regulatory pathways, we will briefly consider each of these cytokines.

IFN-γ

IFN-γ exerts a strong enhancing effect on the production of cytokines, regardless of the agonist used. This priming effect can be further enhanced in the presence of GM-CSF. IFN-γ also up-regulates both the production of IL-1Ra and the accumulation of IL-1Ra mRNA. This positive effect of IFN-γ on neutrophil cytokine production might have important consequences in vivo and may also represent one of the important mechanisms contributing to improve host defense to pathogens.

IL-10

Contrary to IFN-γ, IL-10 exerts an inhibitory effect on neutrophil cytokine production. This negative effect of IL-10 was initially reported on TNF-α, IL-1-β, and IL-8 production (Cassatella et al, 1993), and is now expanded to almost all cytokines and chemokines produced by neutrophils, including IL-8, MIP-1α, MIP-1β, and GRO-α. Interestingly, IL-10 totally abrogated LPS-induced production of the p40 chain of the heterodimeric IL-12. Given the central role of this latter cytokine on Th-1 cells, this effect of IL-10 could contribute to the prevention of Th1 cell mobilization and attraction during infections. In contrast, IL-10 potentiates the expression of IL-1Ra induced by LPS, but not by opsonized zymosan. Taken together, these findings contribute to reinforcing the role of IL-10 as a biological antiinflammatory compound.

IL-4 and IL-13

These two immunomodulatory T cell-derived cytokines have also recently been shown to decrease LPS-induced neutrophil production of cytokines, in particular IL-8. As mentioned above for IL-10, IL-4 also induced an up-regulation of IL-1 Ra synthesis in neutrophils, but this effect was not shared with IL-13. Finally, GM-CSF, which is known to exert a priming effect on neutrophil production of TNF-α and IL-8 induced by LPS, may also potentiate the neutrophil synthesis of IL-1α and IL-1β, but without affecting that of IL-1Ra.

I.C.5. Molecular Regulation of Cytokine Production

The very low transcriptional activity of neutrophils renders difficult the investigation of the regulation of their cytokine gene transcription. Nevertheless, an active transcription of the IL-1β gene in response to IL-1β, TNF-α LPS, and GM-CSF has been demonstrated (Cassatella, 1999). Moreover, recent studies have shown that a wide variety of agonists induce the nuclear accumulation of NF-κB/Rel proteins and their inhibitor I-κB (McDonald et al, 1997). In contrast, numerous studies have shown that cytokine gene expression can be regulated at the level of mRNA stability. This mechanism explains the up-regulation of IL-1β mRNA, whereas that of IL-1Ra has been shown to occur at the post-transcriptional level. The above-mentioned inhibitory effects of IL-10 on neutrophil IL-8 production also result from both an inhibition of IL-8 gene transcription and an enhanced IL-8 mRNA degradation. Finally, a control at the translational level has also been suggested for IL-1Ra under specific experimental conditions.

I.D. Apoptosis and Resolution of Acute Inflammation

Most acute inflammatory responses resolve spontaneously thanks to endogenous “stop programs” that switch off inflammation and limit destruction of host tissues. These include the elimination of infectious agents by phagocytosis, the progressive decrease of leucocyte recruitment promoted by endogenous “braking signals,” and finally, the apoptosis and clearance of leukocytes (Liles and Klebanoff, 1995).

I.D.1 Progressive Decrease of Neutrophil Recruitment

Termination of neutrophil emigration from blood vessels presumably results from (i) changes in the pattern of cytokine/anticytokine (IL-1/IL-1Ra, TNF/soluble TNF-R) and inflammatory/anti-inflammatory cytokines (IL-10, IL-4, IL-13), secreted by tissue cells and infiltrated leukocytes; (ii) the progressive return of endothelial cells to their resting state in terms of membrane adhesion molecules—that are shed or internalized—and in terms of displayed chemokines; and (iii) the inactivation of chemoattractants by specific enzymes or via receptor-mediated endocytosis (Ayesh et al, 1995; Cao et al, 1998; Hofman et al, 1998).

As mentioned above, IL-10, IL-4, and IL-13, synthesized by macrophages and T cells at the inflammation site, down-regulate chemokine synthesis by neutrophils (Cassatella et al, 1993; Wang et al, 1994), and IL-10 enhances the synthesis of IL-Ra (Cassatella et al, 1994; Marie et al, 1996). In contrast, the effects of IL-10 on endothelial cells, such as the up-regulation of E-selectin and ICAM-1 (Sironi et al, 1993; Vora et al, 1996), would favor leucocyte extravasation. The net effect of these cytokines, however, appears to slow down neutrophil recruitment, because anti-IL-13 antibodies enhance the recruitment of neutrophils in the inflamed lung (Lentsch et al, 1999), and IL-10-deficient mice develop chronic enterocolitis characterized by a massive influx of neutrophils (Kuhn et al, 1993).

Other mediators that may slow down neutrophil recruitment are lipoxins, ie, lipoxygenase products generated by transcellular metabolism during host defense and inflammation. PMN-platelet transcellular pathways are a major route to lipoxin formation during PMN/platelet adhesion, PMN donating leukotriene A4 and platelet providing the lipoxin synthase to produce lipoxin LXA4. Lipoxins inhibit neutrophil chemotaxis, adhesion to endothelial cells, and migration across endothelium and epithelium. They were shown, in experimental models of glomerulonephritis, to act as “stop signals” for neutrophil-mediated tissue injury (Diamond et al, 1999; O’Meara and Brady, 1997).

I.D.2. Apoptosis in Resolution of Inflammation

Neutrophil apoptosis and subsequent ingestion by macrophages is the major mechanism for clearing neutrophils that have been recruited to the inflamed sites and thus for promoting resolution of the inflammation (Cox et al, 1995; Savill, 1997).

The constitutive apoptosis of senescent neutrophils involves proteolytic cascades—caspases, calpains, and the proteasome—that activate kinases, eg, caspase 3-mediated activation of protein kinase C-δ (Pongracz et al, 1999), dissociate actin-binding proteins from filamentous actin (Knepper-Nicolai et al, 1998), and participate in cell surface as well as nuclear morphological transformations. Inflammatory mediators, such as LPS or GM-CSF, delay the apoptosis of neutrophils by increasing mitochondrial stability and reducing caspase 3 activity (Watson et al, 1999), and by down-regulating the gene expression of Bax, a pro-apoptotic member of the Bcl-2 family (Dibbert et al, 1999). In contrast, anti-inflammatory cytokines such as IL-10 accelerate the apoptosis of LPS-activated neutrophils (Cox, 1996).

Extravasation and apoptosis of inflammatory neutrophils are normal in Fas ligand- and Fas-deficient mice, showing that the FasL/Fas-mediated apoptosis is not essential in regulating the lifespan of neutrophils during an acute inflammatory response (Fecho and Cohen, 1998). Still, macrophages can trigger neutrophil apoptosis by expressing cell surface Fas ligand (FasL) and releasing soluble FasL, that reacts with the Fas “death receptor” on neutrophils. Ingestion of opsonized particles or of apoptotic neutrophils indeed promotes the release of soluble FasL by macrophages and the killing of bystander neutrophils (Brown and Savill, 1999). This may represent a negative feedback loop accelerating the resolution of inflammation by eliminating recruited leukocytes by apoptosis.

Phagocytosis of apoptotic neutrophils by human macrophages involves the αvβ3 integrin-CD36 complex, on macrophages, which binds thrombospondin, which itself binds an undefined ligand on apoptotic neutrophils (Savill et al, 1992). Macrophages stimulated with digestible particulate glucans lose their ability to use this recognition system, but acquire the ability to recognize exposed phosphatidylserine (PS) on the surface of apoptotic cells. CD36 appears to act as a necessary cofactor either for the αvβ3 system or for PS recognition (Fadok et al, 1998a). CD14 participates in the phagocytosis of apoptotic lymphocyte, but not that of apoptotic neutrophils.

Finally, phagocytosis of apoptotic neutrophils actively inhibits the production of IL-1beta, IL-8, IL-10, GM-CSF, TNFα, leukotriene C4, and thromboxane B2 by human macrophages (Fadok et al, 1998b). This active suppression of inflammatory mediator production is presumably an important step in the resolution of inflammation.

II. Neutrophils in Pathology

The major role of neutrophils in host defense is a rapid response to invading microorganisms. However, neutrophils do not differentiate efficiently between foreign and host antigens without the help of soluble components of the immune system (eg, antibodies, complement, and cytokines) to select their targets. The nonspecific response and the powerful weapons of neutrophils are the two major mechanisms by which they could injure normal tissue. The host-damaging potential of the neutrophils is limited by elimination of the primary event that initiates inflammatory sequences and by mechanisms that inactivate neutrophils such as tachyphylaxis to proinflammatory mediators and apoptosis. Shut off the neutrophil influx involves inactivation of mediators and temporal change in the pattern of chemokines produced. Apoptosis mediates safe clearance of dying neutrophils from the inflammatory site. When these regulatory mechanisms are impaired or when the acute insult cannot be resolved, neutrophils become the predominant contributor to tissue injury (Fig. 4).

Figure 4
figure 4

Neutrophil-induced lesions. A, Leucocytoclastic vasculitis of the gut associated with cryoglobulinemia in a patient with rheumatoid arthritis. B, Necrotizing and crescentic glomerulonephritis associated with anti-proteinase 3 ANCA in a patient with Wegener’s granulomatosis. C, Acute tubular necrosis secondary to renal ischemia after aortic surgery (courtesy of Drs Dominique Droz and Laure-Hélène Noël, INSERM U507, Necker Hospital). D, Bronchiolar infiltration by neutrophils in a patient with cystic fibrosis. (courtesy of Dr Claire Danel, Laennec Hospital, Paris).

The role of neutrophils in pathology is exemplified in this review by different diseases, which we classified according to the stimulus or type of mechanism that activates neutrophils. However, this classification is simplistic, because neutrophils may be activated via several concomitant mechanisms when involved in inflammatory reactions.

II.A. Bacterial Infection

Tissue damage after acute bacterial infection may partly result from excessive neutrophil infiltration and activation in the infected tissue. During pyelonephritis, the large inocula of bacteria in the kidney parenchyma triggers a burst of neutrophil extravascular migration. Microscopic examination of infected kidneys shows intensive neutrophil infiltration, degranulation of neutrophils containing phagocytozed bacteria, severe tubular destruction, and occlusion of small capillaries by leukocyte plugs (Ivanyi and Thoenes, 1987; Ivanyi et al, 1988). Studies using CT imaging indicate that patients with severe infection may develop renal scars (Meyrier, 1989). Experiments in animal models have shown that renal scarring after acute bacterial pyelonephritis results from parenchymal damage by neutrophils. Renal scarring may be prevented by neutrophil depletion or inhibition of leukocyte chemotaxis (Bille and Glauser, 1982; Tardif et al, 1994). Indirect evidence suggests that uropathogenic bacterial strains, which induce excessive neutrophil activation, contribute to tissue damage and renal scarring: strains associated with renal damage in vivo induce higher extracellular release of elastase and reactive oxygen products in vitro than bacterial strains causing pyelonephritis without renal scars (Topley et al, 1989; Monga and Roberts, 1995; Mundi et al, 1991). Some Escherichia coli strains that resist neutrophil engulfment could stimulate an extracellular oxidative metabolic burst, while those that are phagocytozed usually lead to intraphagosomal production of free radicals, minimizing their tissue-damaging effects (Iwahi and Imada, 1988; Lock et al, 1990). The bacterial load at the initiation of the inflammatory response and the phenotypic expression of bacteria, determining their ability to provoke neutrophil activation and to resist neutrophil phagocytosis, are important factors contributing to neutrophil-induced tissue damage during bacterial infection.

II.B. Tissue Injury-Induced Inflammation: Ischemia-Reperfusion Injury

Tissue injury after ischemia appears to be a consequence not only of tissue hypoxia but also of the process of reperfusion that leads to an inflammatory response. After the initial ischemic event, an array of cytokines, complement components, and cell contents are released, activating the endothelium and inducing neutrophil recruitment (Fig. 4C). Infiltrating neutrophils are a potential source of reactive oxygen species, proteolytic enzymes, and cytokines, which during reperfusion may play a detrimental role. In addition, capillaries may become obstructed by aggregated neutrophils, impairing reperfusion of the microcirculation (Bagge et al, 1980). Several studies have shown a pathogenetic role for neutrophils in ischemic insult to the myocardium, intestine, skeletal muscle, liver, and kidney, as reviewed in De Greef et al (1998). Ischemia causes neutrophil activation, retention, and worsening of renal injury in isolated kidneys (Linas et al, 1995). Neutrophil depletion attenuates ischemic renal reperfusion injury in the rat and mouse (Kelly et al, 1996). Blocking antibodies to ICAM-1 or β2-integrin prevent the ischemia-induced renal infiltration of granulocytes and protect the kidney against reperfusion injury in the rat (Kelly et al, 1994; Rabb et al, 1994, 1995). The same results were obtained using antisense oligonucleotides and gene “knock-out” to block ICAM-1 (Kelly et al, 1996). In renal transplantation, neutrophil infiltration into the glomeruli approximately 30 minutes after reperfusion of the cadaveric kidney is significantly associated with long cold-ischemia times and delayed graft function, suggesting that graft function may be influenced by early neutrophil-mediated damage after reperfusion (Koo et al, 1998). Inhibition of leukocyte adhesion with anti-ICAM-1 or anti-LFA1 mAb in clinical renal transplantation trials seems to accelerate recovery of graft function (Haug et al, 1993; Hourmant et al, 1996). Such therapeutic approaches may herald the development of anti-adhesion strategies in other clinical forms of ischemia-reperfusion injury, such as resuscitation following systemic hypotension, or myocardial necrosis.

II.C. Crystal-Induced Inflammation

Acute gouty inflammation is the consequence of the deposition of monosodium urate crystals in joints. Neutrophils appear to be the major effector of acute gout. They accumulate in the joint fluid where they actively ingest urate crystals, aggregate and degranulate. Acute gouty inflammation may be prevented by neutrophil depletion or leukocyte chemotaxis inhibition (Phelps and McCarty, 1966). The considerable reactivity of the urate crystal surface allows it to bind soluble and membrane proteins (Terkeltaub et al, 1983). Precipitated urate crystals activate humoral mediator cascades such as complement and contact activation systems. The generation of C5a, kallikrein, bradykinin, and plasmin induces neutrophil chemotaxis and vascular permeability. Urate crystals also induce the production of inflammatory mediators by synovial cells (TNFα and IL8) and by neutrophils (LTB4 and IL8) (Matsukawa et al, 1998; Rae et al, 1982). TNFα and IL8 induce subsequent production by neutrophils of IL1β and IL8, which amplify the leukocyte influx. This autocrine mechanism appears to regulate the initiation and propagation of the inflammatory reaction.

Interestingly, urate crystals inhibit the production of C-C chemokines by neutrophils in response to TNF-α, preventing the recruitment of mononuclear cells (Hachicha et al, 1995). Binding of urate crystals to neutrophils is followed by the release of lysosomal proteases and superoxide. In addition, phagocytosis of these highly membranolytic crystals may cause perforation of the phagolysosomal membrane leading to liberation of the lysosomal contents and to cell necrosis. However, the tissue injury induced by neutrophils during gout attack is acute and self-limited. The desensitization of neutrophils to chemotactic factors, proteolytic inactivation of soluble mediators, production of cytokine antagonists such as IL-1Ra (Matsukawa et al, 1998), and release of anti-inflammatory molecules such as TGF-β (Brandes et al, 1991) may suppress leukocyte activation and neutrophil influx. Change in the protein coating of the crystal surface during the evolution of acute gouty inflammation may also decrease the ability of urate crystals to activate leukocytes (Terkeltaub et al, 1991) and may partly explain why the presence of intra-articular crystals do not necessary produce arthritis between acute gouty attacks.

II.D. Complement-Induced Inflammation and Oxidative Stress: Hemodialysis

Activation of the complement cascade received much attention in the early literature on the immune system in maintenance dialysis (Craddock et al, 1977). It occurs in the early phase of each dialysis session, mainly through the alternative pathway, and closely reflects dialysis membrane biocompatibility, which is observed in dialysis with cellulose membranes such as cuprophan, but not with synthetic membranes, such as polyacrilonitrile (Cheung, 1990). Activated complement-split products, C5a and C3a, were originally ascribed to neutrophil lung sequestration (Arnaout et al, 1985; Hakim et al, 1984), and are now endowed for several indices of dialysis-induced neutrophil and/or monocyte activation, including triggering of protease (Hörl et al, 1985), reactive oxygen species (ROS) production by neutrophils (Descamps-Latscha et al, 1991), and transcription (but not secretion) of the pro-inflammatory cytokines, interleukin-1 (IL-1) and TNF-α by monocytes (Schindler et al, 1990). The conjunction of massive generation of ROI in the face of a chronic deficiency of antioxidant systems (Céballos-Picot et al, 1996) and of a profound impairment in the balance between pro-inflammatory cytokines and their specific inhibitors (Descamps-Latscha et al, 1995) largely contribute to dialysis-related complications such as an increased rate of atherosclerosis and β2-microglobulin amyloidosis arthropathy (Descamps-Latscha, 1993). In this setting chlorinated oxidants derived from MPO seem to play a critical role (Witko-Sarsat et al, 1998).

In acute ischemic renal failure in rats, chronic activation of the complement system by daily exposure of the blood to cuprophane dialysis membranes is associated histologically with increased neutrophil infiltration in the renal parenchyma, and functionally with a delay in the resolution of the acute renal failure (Schulman et al, 1991). Among patients with acute renal failure requiring hemodialysis, the use of the cuprophane membrane, as compared with the synthetic membrane, leads to a slower resolution of renal failure and a lower survival rate (Hakim et al, 1994; Schiffl et al, 1994). These findings suggest that preactivation of neutrophils by hemodialysis exacerbates the detrimental role of these cells in ischemia-reperfusion injury and in endotoxemia.

II.E. Immune Complex-Induced Inflammation: Antibody-Mediated Glomerunephritis

Immune complex deposition in tissues triggers an inflammatory reaction and is a key pathogenic factor in numerous clinical conditions such as glomerulonephritis, immune vasculitis, arthritis, and systemic lupus (Fig. 4A). Antibodies may deposit in the glomerulus either because they circulate in the form of immune complexes and are passively entrapped, or because they bind to targets fixed in the glomerulus. Among the experimental models developed to study the pathogenesis of antibody-mediated glomerulonephritis, the most extensively studied is the model of antiglomerular basement membrane (anti-GBM)-nephritis, which is induced by injection of heterologous anti-GBM antibodies. In this experimental model, the injection of sufficient heterologous antiserum against the GBM leads to acute (heterologous) and delayed (autologous) phases of injury. The autologous phases result from the host’s immune response to heterologous Ig anti-GBM. Most studies have concentrated on the heterologous phase of anti-GBM-nephritis. This phase is characterized by immediate deposition of nephrotoxic mAb along the GBM (within minutes), transient neutrophil infiltration (maximal at 2 hours) in the glomerular capillaries, a swelling of endothelial cells that may detach from the underlying basement membrane, and the presence of thrombotic lesions. At Day 1 to 4, many glomeruli show segmental and global thrombosis and necrosis. Development of proteinuria starts 2 to 4 hours after injection and is maximal at 8 to 12 hours. Neutrophils play a critical role in glomerular injury in the heterologous phase because neutrophil-depleted mice or beige mice (that are deficient in leukocytic neutral proteinase elastase and cathepsin G) do not develop proteinuria (Schrijver et al, 1990). ICAM-1 or P-selectin deficiency does not impair neutrophil influx in these models (Tang et al, 1997). CD11b/CD18 (CR3) deficiency reduces but does not abolish neutrophil infiltration. The recruitment of neutrophils is Fcγ-dependent and the use of F(ab′)2 fragments of the anti-GBM antibody markedly reduces the neutrophil influx. Immune complex deposits under the fenestrated endothelium are easily accessible to circulating cells. It has therefore been suggested that the initial accumulation of neutrophils is driven by neutrophil Fc receptor engagement with immobilized immune complexes in the glomerular capillary walls. The putative role of complement in pathogenesis of the heterologous phase of the anti-GBM disease is suggested by the deposition of complement components in a distribution matching that of the antibody. C5a may have a role in the chemoattraction and activation of neutrophils; the deposition of iC3b on the GBM may enhance neutrophil adhesion while the deposition of sublytic concentration of C5b-9 on the endothelial surface may promote the surface expression of P-selectin. However, the contribution of complement is dependent on the dose of antibody used (Sheerin et al, 1997). Under a certain threshold level of antibody, complement acts synergistically with anti-GBM antibodies to induce neutrophil infiltration and albuminuria. At higher doses, neutrophil accumulation is complement-independent, but proteinuria remains partially complement-dependent, as shown in an experimental model using C3, C4, and C5-deficient mice (Sheerin et al, 1997; Schrijver et al, 1988). Complement dependent proteinuria may be related to the interaction of CR3 (CD11b/CD18) on neutrophils with complement fragment iC3b, which probably stabilizes neutrophil interaction with immune complexes and favors the release of azurophilic granules (Tang et al, 1997). Indeed, proteinuria is absent in CR3-deficient mice, whereas initial neutrophil infiltration is only partly reduced. Neutrophils, once attracted to the glomeruli, make close contacts of varying extent with the inner side of the GBM, pushing aside the endothelial cells. In the zones of contact of neutrophils with their adhesion partners, neutrophil proteinases are known to be functional, in spite of the presence of powerful circulating inhibitors (Campbell and Campbell, 1988). Neutrophil enzymes can thus fragment GBM and degrade heparan sulfate proteoglycans in vitro and ex vivo (Heeringa et al, 1996). Heparan sulfate constitutes the majority of anionic sites of the capillary permeability barrier and also contributes to the anticoagulant properties of GBM exposed through the glomerular endothelium fenestrae. Experiments performed in Beige mice, which have a congenital defect in neutrophil granules, have demonstrated the involvement of neutrophil proteinases in the induction of albuminuria in the heterologous phase of anti-GBM nephritis (Schrijver et al, 1989). While the importance of neutrophils in mediating glomerular injury has been well demonstrated in this experimental model, the pathogenetic role of these cells in human immune complex mediated glomerunephritis remains to be elucidated.

II.F. Cytokine-Induced Inflammation: Rheumatoid Arthritis

Rheumatoid arthritis (RA) is known to be a predominantly T cell/macrophage driven process in the early stage of the disease, but the cell with the greatest capacity to inflict damage within joints is the neutrophil (Edwards and Hallett, 1997). Neutrophils are numerous in the synovial fluid and joint tissues during the early stages of rheumatoid arthritis and during acute exacerbation of the disease (Mohr et al, 1984). Neutrophils concentrate at the pannus-cartilage junction, which is the site of early cartilage erosion. They are thought to contribute directly to the cartilage damage, through their serine- and metalloproteases (Chatham et al, 1993; Larbre et al, 1995), as well as through the production of ROS and chlorinated oxidants (Edwards and Hallett, 1997). One should emphasize the importance of neutrophil- and macrophage-derived cytokines, found in large quantities in synovial fluids [eg IL-1β, IL-6, TNF-α, TGF-β, and IL-8]. In vitro, neutrophil-mediated cartilage injury is modulated by cytokines such as TNF-α (Kowanko et al, 1990). In vivo, the major role of TNF-α in the pathogenesis of RA has been demonstrated by a successful clinical trial using anti-TNF antibodies (Elliott et al, 1994). TNF-α is therefore likely to be an important stimulator of neutrophils within the joints of RA patients, resulting in amplification of the inflammatory response and contributing to the tissue damage caused by neutrophils.

Another cytokine playing a crucial role in the attraction of neutrophils is IL-8. Abundantly released by chondrocytes, it diffuses toward the joint surface, causing a chemotactic gradient. Although IL-8 alone does not exert a direct effect on cartilage, it induces a rapid cartilage destruction when cocultured with neutrophils (Pillinger and Abramson, 1995) probably due to neutrophil release of oxidants and proteases. Human chondrocytes are also a major source of NO. They express the inducible isoform of NO synthase (iNOS) and NO, in conjunction with ROS, might theoretically contribute to promote cartilage degradation. However, this was not verified in a recent report showing that the inhibition of chondrocyte production enhances neutrophil-induced cartilage breakdown (Clancy et al, 1993).

The question of whether neutrophils in RA differ either functionally or in their molecular expression from non-rheumatoid neutrophils is still a matter of debate. Some studies have reported that the circulating rheumatoid neutrophil exhibits evidence of prior activation, with enhanced expression of receptors including FcγRI, CR1, CR3, and CR4, increased capacities of migration, degranulation, and superoxide anion production (Pillinger and Abramson, 1995). However, other studies have shown that the only neutrophils that show activation characteristics are within the joints (Lopez et al, 1995). Because there is consensus for considering RA as a major T-cell–driven disease, it is likely that the differences between rheumatoid and non-rheumatoid circulating neutrophils are a subsequent rather than a primary event in RA.

With regard to therapeutic strategies developed in RA, it is of interest to note that (i) most of the commonly used antiinflammatory drugs eg, corticosteroids and non-steroidal antiinflammatory drugs, gold salts, sulfasalazine, or D-penicillamine, exert profound effects on neutrophil effector functions; and (ii) among the drugs under trial in RA, several are specifically aimed at targeting neutrophil molecules eg, anti-adhesion molecules and/or neutrophil activation, eg, anticytokine antibodies (Arend and Dayer, 1995). Taken together, these observations derived from RA underscore the importance of considering the neutrophil as a pleiad of actors, each one a potential candidate for targeting future therapeutic strategies.

II.G. Antineutrophil Cytoplasmic Antibodies and Vasculitis: Autoimmunity Against Neutrophil Components

Neutrophil granule proteins are now recognized as target antigens for antineutrophil cytoplasmic antibodies (ANCA) found in sera of patients with systemic necrotizing vasculitides. The pathogenesis of ANCA-associated vasculitis is unknown. However, there is no immunologic evidence for vascular immune complex localization (Csernok et al, 1999; Falk and Jennette, 1997).

The diagnosis of ANCA is based on the pattern of fluorescence observed by indirect immunofluorescence microscopy on alcohol-fixed PMN. ANCA can be divided into a group displaying a cytoplasmic staining pattern (C-ANCA) and a second group displaying perinuclear staining (P-ANCA) (Hoffman and Specks, 1998; Wieslander, 1991). Investigations of antigen specificity have been aimed at identifying the proteins recognized by these ANCA (Lesavre, 1991). It appears that the majority of C-ANCA react with PR3, although in a few cases C-ANCA could be directed against BPI, elastase (Nassberger et al, 1989), or cathepsin G (Halbwachs-Mecarelli et al, 1992). Despite difficulties in classifying vasculitic syndromes, the correlation between clinical expression of Wegener’s granulomatosis and ANCA reactivity has now established that PR3 is the target autoantigen (Ewert et al, 1991; Jenne et al, 1990). Wegener’s granulomatosis is a multisystem disease characterized by granuloma of the respiratory tract and systemic necrotizing vasculitis (Hagen et al, 1993) (Fig. 4B). The major target antigen of P-ANCA is myeloperoxidase (Falk and Jennette, 1988). Anti-MPO was originally described in patients with microscopic polyangiitis and idiopathic necrotizing crescentic glomerulonephritis, which can be considered as a renal limited form of systemic vasculitis. Also presenting anti-MPO is the Churg-Strauss syndrome, which has a history of asthma and hypereosinophilia in conjunction with small vessel vasculitis (Falk and Jennette, 1997; Kallenberg and Tervaert, 1999). Whether ANCA are serologic epiphenomena or play a pathogenic role in the course of the disease is still a matter of debate (Salant, 1999). The correlation of ANCA titers with disease activity suggests that ANCA may be directly involved in the clinical course of the disease.

Many studies have focused on in vitro ANCA-induced activation of neutrophils and consequent damage to endothelial cells. Binding of ANCA to their target results in activation of an oxidative burst (Falk et al, 1990; Keogan et al, 1992), degranulation, and cytokine secretion in neutrophils and in monocytes (Ralston et al, 1997). The binding of ANCA involves the engagement of Fcγ receptors (Kocher et al, 1998). However, hypotheses that postulate a pathogenic role for ANCA must explain how ANCA are able to interact in vivo with target antigens that are sequestered within the cytoplasm of neutrophils. Several mechanisms leading to translocation of intracellular antigens have been proposed, including cytokine priming (Csernok et al, 1994) and apoptosis of neutrophils (Gilligan et al, 1996). Interestingly, PR3 is expressed at the plasma membrane of stimulated neutrophils (Muller Kobold et al, 1998), but also in a constant subset of unstimulated neutrophils. A large subset of neutrophils expressing membrane PR3 is a risk factor for vasculitis (Witko-Sarsat et al, 1999c).

Despite several studies using synthetic peptides or recombinant proteins aiming at mapping ANCA epitopes, no precise information is available. Anti-PR3 recognize conformational epitopes (Bini et al, 1992). Of note, both anti-MPO (Short et al, 1995) and anti-PR3 (Sun et al, 1998) seem to recognize a pro-form that is not completely processed. The relevance of such findings in the pathophysiological mechanisms of vasculitis are still unknown. Moreover, the question of whether ANCA can be a direct target for treatment has still to be elucidated (Kallenberg and Tervaert, 1999).

II.H. Genetic Disorders of Neutrophil Regulations: Hereditary Periodic Fever Syndromes

The hereditary periodic fever syndromes are a group of disorders characterized by self-limited episodes of fever accompanied by localized inflammation. Of these syndromes, Familial Mediterranean Fever (FMF) has been the longest recognized (Sohar et al, 1967). It is an autosomal recessive disease occurring primarily among populations originating in the Mediterranean basin (Arabs, Armenians, Turks, and Sephardic Jews).The inflammatory episodes of FMF affect mainly the serosal or synovial membranes. They are characterized by massive influx of neutrophils into the affected tissues. These recurrent and reversible attacks are brief (1 to 3 days). Secondary amyloidosis, a consequence of long-standing inflammation, is the most severe complication of the disease. The acute attacks and the development of amyloidosis can be prevented by daily oral colchicine (Zemer et al, 1986). The FMF gene has recently been identified by positional cloning (Anonymous, 1997a, 1997b). It encodes a transcript expressed predominantly in neutrophils. The predicted product is a 781 amino acid protein (called pyrin or marenostrin) with homology to several transcription factors. Restriction of marenostrin/pyrin message expression to neutrophils has confirmed that FMF is a genetic disorder of these cells. The function of this protein is still unknown, and the pathogenic basis of the inflammatory attack is not yet understood. It is tempting to suggest that marenostrin/pyrin is an inhibitory regulator of the inflammatory response, controlling for example a chemotaxin-inactivating enzyme (Babior, 1998; Matzner et al, 1984). Another point that is not yet settled is why FMF attacks involve serosal and synovial membranes. One may argue that serosal and synovial surfaces suffer minor injuries that are too mild to cause symptoms but are serious enough to locally activate neutrophils. Physical traumas seems to initiate inflammatory episodes, because many patients report the appearance of a short synovial attack involving the lower extremities after prolonged standing or walking. However, other tissues such as mucous membranes that are more exposed to trivial injury are not affected by the inflammatory episodes of FMF, showing that the regulatory dysfunction of neutrophils from these patients is not implicated in all pathways leading to neutrophil stimulation.

Familial Hibernian Fever (FHF), also termed dominant periodic fever, familial periodic fever, or TRAPS (TNF receptor-associated periodic syndromes), is an autosomal dominant disease occurring primarily among the populations originating in Northern Europe (McDermott et al, 1997). The clinical picture of FHF is recurrent attacks of fever with abdominal pain and severe localized myalgia. Presence of erythematous skin lesions, conjunctivitis, unilateral periorbital edema, and scrotal pain are noted with high frequency. Pleuretic and nonpleuretic chest pain may occur. True arthritis is rare but oligoathralgia is a frequent symptom. Lymphadenopathy is common. Attacks tend to last longer (2 to 3 weeks) than the inflammatory episodes of FMF. Patients tend to respond to corticosteroids rather than colchicine. Amyloidosis has also been documented but with a widely divergent incidence among families. Mutations that cause this disease are point substitutions within cystein-rich motifs of the extracellular domains of the 55 kDa TNF receptor (TNFR1, p55, or CD120a) (McDermott et al, 1999). These mutations impair metalloprotease-mediated shedding of membrane TNFR1. Cleavage of TNF receptors from the membrane of activated monocytes and neutrophils is thought to attenuate the inflammatory response by desensitization of phagocytes to further “juxtacrine” TNF stimulation and by producing a pool of soluble receptors that competes with membrane-bound receptors and inhibits TNF systemic effects. It is therefore likely that impaired post-stimulatory TNFR1 clearance and subsequent reduced levels of shed soluble receptors are responsible for prolonged systemic inflammatory episodes triggered by mild injury in FHF.

II.I. Cystic Fibrosis: The Paradox of Exacerbation of Neutrophil-Mediated Tissue Damage and Concomitant Persistence of Infection

Cystic fibrosis is a hereditary disorder caused by mutations of the cystic fibrosis transmembrane conductance regulator (CFTR), the product of which is a membrane protein thought to function as a chloride channel (Stern, 1997). The lethal clinical manifestations are clearly related to the thick, infected mucous and chronic neutrophil-dominated airway inflammation (Jennings and Crystal, 1992; McIntosh and Cutting, 1992; Welsh and Fick, 1987). It has been suggested that CFTR itself may be a major receptor for binding and internalization of Pseudomonas aeruginosa and that CFTR-mediated ingestion of Pseudomonas aeruginosa is critical for early and effective clearance from the lung (Pier et al, 1997). This pathogen can never be permanently eradicated despite intensive antibiotic treatment and leads invariably to respiratory failure, which is the cause of death in most patients with cystic fibrosis. Although the genetic basis of cystic fibrosis, as well as the molecular structure of CFTR, have been extensively studied, a clear relationship between the genetic defect and the pulmonary pathophysiology, especially chronic infections and neutrophil-dominated airway inflammation, has not been established. A current hypothesis stresses the possibility of impairment in the innate immune system (Bals et al, 1999).

Neutrophils are considered responsible for the early onset and the promotion of the inflammatory process in CF, which starts within the first year of a CF patient’s life (Khan et al, 1995). Several studies investigating the complex relationships between infection and inflammation in CF support the concept that the host inflammatory response is not necessarily proportional to the burden of pathogens in the respiratory tract, although these pathogens may provide the primary stimulus for such responses (Regelmann et al, 1995). Numerous neutrophils are present in the airways (Fig. 4D) and high concentrations of neutrophil-derived mediators have been found, eg, long-lived oxidants (Witko-Sarsat et al, 1995); inflammatory cytokines, such as IL-8 (Richman-Eisenstat et al, 1993), which correlates with disease activity, and TNF-α (Bonfield et al, 1995); metalloproteases such as gelatinase; serine proteases including elastase, cathepsin G (Goldstein and Döring, 1986; Suter et al, 1984), and recently, proteinase 3 (Witko-Sarsat et al, 1999b) with a clear imbalance between proteinases and antiproteinases (Birrer et al, 1994); and antibiotic peptides, defensins. It has been shown that the antibiotic activity of beta-defensins (hBD-1 and hBD-2) might be decreased in the CF lung because of the modification in the ionic concentrations of the bronchial secretions (Bals et al, 1998; Goldman et al, 1997).

With regard to neutrophil functions, myeloperoxidase-dependent oxygenation activities appear to be significantly higher not only in CF homozygotes, but also in heterozygote parents of CF patients (Prince, 1998), providing some evidence for a genetic component in the altered neutrophil function in CF (Witko-Sarsat et al, 1996). Likewise, it has been described as a decrease in the shedding of L-selectin in stimulated CF neutrophils, which was not observed in either stable or acutely infected non-CF bronchiectasis patients, thus suggesting a disturbed control of the migration process in CF neutrophils (Russell et al, 1998).

As CF involves chronic active inflammation and recurrent infections ultimately resulting in inflammation, clinical management of CF presents a paradox combining antibiotherapy and antiinflammatory drugs such as the non-steroidal antiinflammatory agent ibuprofen (Konstan et al, 1995). Antiproteinase therapy has also been proposed and aerosolization of α1-antitrypsin or secretory leukoproteinase inhibitor (SLPI) has proven to exert a beneficial effect on pulmonary function (McElvaney et al, 1991, 1992). New antiinflammatory therapeutic approaches in CF rely on better knowledge of the link between the CFTR mutation and the mechanisms of the neutrophil-dominated airway inflammation function that constitutes a very recent and exciting area of research. For instance, CFTR belongs to the family of ATP-binding cassette (ABC) proteins. The idea is that other ABC proteins—Multi-Drug Resistance (MDR) protein or the Multidrug Resistance-associated Protein (MRP)—might complement CFTR (Lallemand et al, 1997). Interestingly, colchicine, which is both an ABC protein inducer and an antiinflammatory agent with direct effects on neutrophils, has thus been proposed as a good antiinflammatory drug candidate (Sermet-Gaudelus et al, 1999).

Conclusion

Until approximately two decades ago, neutrophils were viewed as short-lived cells having a destructive hardware but no software. This notion has changed since it became clear that neutrophils were able to integrate complex arrays of adhesion-, chemoattractant-, and cytokine-driven signals and were able to synthesize cytokine in response to inflammatory stimuli.

The neutrophil itself appears to be well equipped with an unusual combination of contrasting molecules: ROS, small, ubiquitous and short-lived molecules, and proteases, macromolecules with specific long-lasting activity. The neutrophil safely keeps them apart through separate intracellular compartmentalization under basal conditions. These mediator activities regulate pathophysiological processes not exclusively aimed at the destruction of invading microorganisms. Finally, neutrophils are activated by a great variety of stimuli and are involved in diseases that may be classified, as shown in Table 3, according to the major neutrophil-activating event.

Table 3 Clinical Conditions Associated with Neutrophil-Induced Injury Classified According to their Main Mechanism