Neutrophil Elastase, Proteinase 3, and Cathepsin G as Therapeutic Targets in Human Diseases

A. History

Proteases are proteolytic enzymes that catalyze the splitting of proteins into smaller peptides or amino acids and are important in many ways to human health and biotechnology. Four major classes of proteases are distinguished on the basis of key residues involved in catalysis: serine proteases, cysteine proteases, aspartic acid proteases, and metalloproteases. A new type of catalysis that involves a threonine in the catalytic site has been identified more recently in studies of the proteasome. There are at least 500 to 600 different proteases in humans; most of them are serine, cysteine, and metalloproteases (López-Otín and Bond, 2008). The serine endopeptidases (EC 3.4.21.−) are characterized by the presence of a conserved serine residue as part of the catalytic center. In trypsin/chymotrypsin-type serine proteases, for example, this catalytic triad consists of the three residues Asp, His, and Ser (Polgár, 2005). The major clans found in humans include chymotrypsin-like, α/β-hydrolase, and signal peptidase. Serine proteases represent approximately one third of total proteases currently identified and are present in eukaryotes, prokaryotes, bacteria, and archaea (Hedstrom, 2002). They display diverse biological functions and include digestive enzymes of exocrine glands, clotting factors, and leukocyte granule-associated proteases, such as NE, PR3, and CG.

Although the activity of leukocyte proteases had been described early in the 20th century, HNE was identified only in 1968 by Janoff and Scherer (1968). The term “elastase” is used to describe an enzyme capable of releasing soluble peptides from insoluble elastin, which was the first substrate employed for characterizing its activities, but does not necessarily imply that its activities are always physiologically related to digestion of this target. Subsequently, another protease retaining enzymatic activity on synthetic substrates for chymotrypsin was described (Dewald et al., 1975; Rindler-Ludwig and Braunsteiner, 1975) and designated “cathepsin G” (Starkey and Barrett, 1976). A third protease, “myeloblastin,” was first identified in 1978 and more extensively characterized 10 years later. This name was originally used by Bories et al. (1989) to describe a serine protease in promyelocytic leukemia cells HL-60. It was subsequently shown that myeloblastin was identical to a protease known independently as “proteinase 3” that had been studied for its role in emphysema (Baggiolini et al., 1978). The same molecule had been also described as a protein with antimicrobial activity (Campanelli et al., 1990a) and called azurophilic granule protein 7, and it was identified as the p29 autoantigen or ANCA auto-antigen in Wegener granulomatosis before it finally became referred to uniformly as “proteinase 3.”

B. Biosynthesis, Processing, and Localization in Neutrophil

HNE, PR3, and CG are three homologous proteases that belong to the chymotrypsin superfamily of serine proteases. They evolved from a common ancestor through gene duplication. The genes encoding HNE, PR3, and CG consist of five exons and four introns (Zimmer et al., 1992; Caughey et al., 1993). The gene for HNE, formerly called ELA2 and newly officially dubbed ELANE, is located in the same cluster as that of PR3 (PRTN3) and azurocidin in the terminal region of the short arm of chromosome 19p13.3 (Zimmer et al., 1992; Caughey et al., 1993). In contrast to its related family members, azurocidin has lost its catalytic serine and is devoid of proteolytic activity (Watorek, 2003). It primarily exhibits antimicrobial function. The CG gene (CTSG) is located on chromosome 14q11.2 in a cluster containing other genes for the related serine proteases chymase, granzyme H (Fellows et al., 2007), and granzyme B (Caughey et al., 1993). HNE, PR3, and CG genes are expressed in granulocytes, monocytes, and mast cells. High-level transcription of these HNE, PR3, and CG genes is limited to the promyelocytic stage of neutrophil maturation (Garwicz et al., 2005). PR3 is also found in basophiles (Braun et al., 1991). The main characteristics of HNE, PR3, and CG are presented in Table 1.

The synthesis of HNE, PR3, and CG is regulated first at the transcriptional level during granulocyte development and second at the post-translational level before they are stored in their proteolytically active mature form within neutrophil azurophilic granules. They are synthesized as an inactive prepro-protein containing a signal peptide and an amino-terminal prodipeptide (Rao et al., 1996; Garwicz et al., 1997, 1998; Gullberg et al., 1997). They also contain a C-terminal propeptide, the removal of which is not necessary for NSP activity. The processing and granule targeting of NSPs has been studied in transfected cells producing active proteases in a fashion similar to that of human neutrophils. Activation of NSPs occurs through cleavage of the amino-terminal peptide by signal peptidase, followed by the removal of an aminoterminal dipeptide by DPPI, a tetrameric cysteine protease also known as cathepsin C (EC 3.4.14.1). DPPI is a highly conserved lysosomal cysteine exopeptidase expressed mostly in the lung, spleen, kidney, and myeloid cells and belongs to the papain family (McGuire et al., 1993). It is composed of an oligomeric structure derived from four identical subunits and represents the only member of this family that is functional as a tetramer (Turk et al., 2001). DPPI is initially synthesized as a 55-kDa monomeric pro-enzyme rapidly processed to generate a mature form, but the regulation of its processing is still unknown. The proteolytically active enzyme adopts a papain like-structure made of a 23-kDa heavy chain (Leu207–Arg370) and a 7-kDa light chain (Asp371–Leu439) linked by disulfide bonds and an “exclusion” propeptide domain of 16 kDa (Asp1–Gly119) that remains bound to the mature active enzyme (Fig. 2A). This so-called exclusion domain transforms the framework of a papain-like endopeptidase into a dipeptidyl exopeptidase by blocking the active site beyond the S2 pocket (Turk et al., 2001) (Fig. 2B). In addition to its key role as an activator of hematopoietic granule serine proteases, DPPI also participates in lysosomal peptide degradation via its dipeptidyl peptidase activity. The enzyme cleaves two-residue units from the N termini of proteins until it reaches a stop sequence, typically an arginine or lysine in P2, a proline residue in P1 or P1′, or an isoleucine residue in P1. The C-terminal propeptide of HNE, PR3, and CG is normally excised by an as-yet-unknown protease(s), although this does not seem to be required for either granule targeting or proteolytic activity (Gullberg et al., 1995; Capizzi et al., 2003).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Structure of human pro-dipeptidyl peptidase I. A, amino acid sequence of human pro-dipeptidyl peptidase I. Primary sequences corresponding to exclusion domain (Asp1–Gly119), activation peptide (Thr120–His206), heavy chain (Leu207–Arg370), and light chain (Asp371–Leu439) are colored in red, gray, light, and dark blue, respectively. B, ribbon representation and solvent-accessible surfaces of the DPPI functional monomer in complex with the inhibitor Gly-Phe-CHN₂ [Protein Data Bank 1K3B (Turk et al., 2001), 2DJF, and 2DJG (Mølgaard et al., 2007)]. Color codes in the ribbon plot are the same as in A except that Asp1 in the hairpin loop is colored yellow (left). Solvent-accessible surfaces with positive or negative electrostatic potential are colored dark blue and red, respectively (right). The side chain of the catalytic Cys234 is showed with a cyan stick. The figures have been created with Yasara (http://www.yasara.org).

Fully processed mature HNE, PR3, and CG isolated from azurophilic granules contain, respectively, 218 (Bode et al., 1986; Sinha et al., 1987), 222 (Campanelli et al., 1990b), and 235 (Salvesen et al., 1987; Hof et al., 1996) residues. They are present in several isoforms depending on their carbohydrate content, with apparent mass of 29 to 33 kDa upon SDS-polyacrylamide gel electrophoresis (Twumasi and Liener, 1977; Watorek et al., 1993). HNE and PR3 display two sites of N-glycosylation, whereas CG possesses only one. NSPs are stored mainly in neutrophil azurophilic granules, but HNE is also localized in the nuclear envelope, as revealed by immunostaining and electron microscopy (Clark et al., 1980; Benson et al., 2003), whereas PR3 is also found in secretory vesicles (Witko-Sarsat et al., 1999a). Upon neutrophil activation, granular HNE, PR3, and CG are secreted extracellularly, although some molecules nevertheless remain at the cell surface (Owen and Campbell, 1999; Owen, 2008a). The mechanism through which NSPs are sorted from the trans-Golgi network to the granules has not been completely defined, even though an intracellular proteoglycan, serglycin, has been identified as playing a role in elastase sorting and packaging into azurophilic granules (Niemann et al., 2007). Unlike HNE and CG, PR3 is constitutively expressed on the membranes of freshly isolated neutrophils (Csernok et al., 1990; Halbwachs-Mecarelli et al., 1995). Stimulation of neutrophils at inflammatory sites triggers intracytoplasmic granules to translocate to the phagosomes and plasma membrane, thereby liberating their contents. The first step of the translocation to the target membrane depends on cytoskeleton remodeling and microtubule assembly (Burgoyne and Morgan, 2003). This is followed by a second step of granule tethering and docking, which are dependent on the sequential intervention of SNARE proteins (Jog et al., 2007).

C. Structural Characteristics and Proteolytic Specificity

As found for other members of this family (i.e., the chymotrypsin/trypsin-type family), HNE, PR3, and CG display a three dimensional structure consisting of two homologous β-barrels and a C-terminal α-helix (Bode et al., 1986; Fujinaga et al., 1996; Hof et al., 1996) (Fig. 3A). Each barrel contains six antiparallel β-sheets connected through a linker segment. Residues of the catalytic triad [Ser195, Asp102, and His57 (chymotrypsin numbering)] are located at the junction of the two β-barrels, whereas the active site cleft runs perpendicular to this junction. This arrangement of amino acids in the active site presumably allows nucleophil attack by Ser195 on the carbonyl carbon (C = O) of the substrate scissile bond, thus setting off the catalysis process.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Three-dimensional architecture of the trypsin/chymotrypsin family and structural differences between neutrophil elastase, proteinase 3, and cathepsin G. A, ribbon plot of neutrophil elastase showing the characteristics of the trypsin/chymotrypsin family, the two asymmetric β-barrels and the C-terminal α-helix, front view on the left, back view after a rotation of 180° around a vertical y-axis on the right. N- and C-terminal β-barrels are presented in red and yellow, respectively. The three flexible loops of the C-terminal β-barrels forming the activation domain are colored in pink and indicated by arrows. Disulfide bounds are depicted in green. B, the solvent accessible surface based on the atom coordinates of elastase [Protein Data Bank 1PPF (Bode et al., 1986)], proteinase 3 [Protein Data Bank 1FUJ (Fujinaga et al., 1996)], and cathepsin G [1CGH (Hof et al., 1996)] is colored to show its positive (blue) and negative (red) electrostatic potential. Surface-accessible residues forming a hydrophobic patch on proteinase 3 are depicted in orange. The loop of the inhibitor turkey ovomucoid third domain (P4–P3′) complexed to elastase and the irreversible phosphonate inhibitor Suc-Val-Pro-Phe^P-(OPh)₂ complexed to cathepsin G are given as a cyan stick model. The serine of the catalytic triad is yellow. Ribbon plot of neutrophil elastase and the molecular surfaces are generated with Yasara (http://www.yasara.org).

Upon processing of a dipeptide at the amino terminus of NSPs by DPPI, the free ammonium group of the first N-terminal residue (Ile16) forms an internal salt bridge with the side-chain carboxylate of Asp194 that renders the active site S1 pocket [Schechter and Berger (1967) nomenclature; Fig. 4A] accessible to substrate. Ile16 together with the three flexible surface loops (217–225 loop, 180 loop, and autolysis loop) of the C-terminal β-barrel form the activation domain of NSPs (Fig. 3A), the conformation of which differs after processing by DPPI. The main differences between the topologies of pro- and mature forms of related serine proteases have been previously documented for thrombin, prochymase, and progranzyme K (Hink-Schauer et al., 2002; Reiling et al., 2003; Jenne and Kuhl, 2006). HNE, PR3, and CG are highly cationic proteases as a result of their content of positively charged residues. HNE and PR3 display a cluster of positively charged residues located in the loops of the activation domain (Korkmaz et al., 2007). This positive cluster is disrupted by negatively charged residues in PR3. Four of the five hydrophobic residues forming the surface hydrophobic patch in PR3 are located on loop 217–225 (Fig. 3B). This most probably explains the specific binding of mature PR3 but not of its proform to the neutrophil membrane (Korkmaz et al., 2008b).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Synthetic chromogenic and fluorogenic substrates of NSP. A, schematic diagram illustrating the Schechter and Berger (1967) nomenclature. This nomenclature defines a linear topology of the interactions between a protease and a substrate: S subsites on the protease accommodate P residues of the substrate upstream of the scissile bond, whereas S′ subsites accommodate P′ residues from the C-terminal part of the substrate, downstream of the cleavage site. Numbering starts with the first residue next to the cleavage site, P1 and P1′, and continues toward the N terminus (P2, P3, P4, etc.) and the C terminus (P2′, P3′, P4′, etc.). B, chromogenic substrates commonly used to measure the activity of NSPs, peptidyl-paranitroanilides (top), peptidyl-thiobenzylesters with chromogenic groups in gray. Paranitroanilide may be detected by direct measurement at 410 nm, but a coupled assay with a thiodisulfide reagent such as DTNP is required for measuring the absorbance of the released thiobenzyl group. Fluorogenic aminomethylcoumarin substrates contain a fluorescent group that is released by proteolytic cleavage. FRET substrates contain a fluorescent and a quenching group as a donor/acceptor pair [here a fluorescent ortho-aminobenzoyl (Abz) group and a N-(2,4-dinitrophenyl ethylenediamine (EDDnp) quenching group] at their N- and C-terminal ends, respectively. Fluorescence is released after cleavage of any peptide bond within the amino acid sequence (shown in gray). FRET substrates are convenient tools to investigate the protease specificity around the cleavage site. DTNB, 5,5′-dithiobis(2-nitrobenzoic acid).

The three dimensional structure of the chymotrypsin family of serine proteases in complex with inhibitors provides useful information regarding the operation of the active site. Structures of HNE in complex with turkey ovomucoid third domain (Bode et al., 1986), secretory leukocyte protease inhibitor (SLPI) domain 2 (Koizumi et al., 2008), and synthetic chloromethyl ketone inhibitors (Wei et al., 1988; Navia et al., 1989) have been determined as that of CG in complex with Suc-Val-Pro-PheP-(OPh)2 (Hof et al., 1996) and β-ketophosphonate 1 (de Garavilla et al., 2005). The crystal structure of PR3 has been solved by molecular replacement using the crystallized HNE structure (Fujinaga et al., 1996). HNE, PR3, and CG display a common fold, but their distinct properties with respect to particular substrate cleavage specificity arises from the substitutions in the S1 pocket and also from the surface loops surrounding the active site (99 loop, 60 loop, 37 loop, and autolysis loop) defining the environment of the active site. Solvent-accessible surfaces in HNE, PR3, and CG show that their charge distributions differ significantly in the vicinity of the substrate binding region (Korkmaz et al., 2008c).

The S1 pocket in individual chymotrypsin-like serine proteases plays an important role in defining their substrate specificity. The HNE S1 pocket is hemispherical and hydrophobic because of the presence of Val190, Phe192 Ala213, Val216, Phe228, and the disulfide bridge Cys191–Cys220 (Bode et al., 1986). The S1 pocket of PR3 is also hemispheric but seems to be smaller than that of HNE because of the Val/Ile substitution at position 190 (Fujinaga et al., 1996). The S1 subsite of CG, which is larger than that of HNE and PR3, is divided into two compartments with a negatively charged Glu226 at the bottom (Hof et al., 1996). The specificity of NSPs has been investigated using chromogenic (p-nitroanilide, Anb^5,2-NH₂, thiobenzylester) and fluorogenic [aminomethylcoumarin, fluorescence resonance energy transfer (FRET)] substrates (Castillo et al., 1979; Nakajima et al., 1979; McRae et al., 1980; Tanaka et al., 1985; Brubaker et al., 1992; Kam et al., 1992; Réhault et al., 1999; Wysocka et al., 2007; Korkmaz et al., 2008a; Carmona et al., 2009) (Fig. 4B). In agreement with structural data, HNE and PR3 preferentially accommodate small hydrophobic residues (Val, Cys, Ala, Met, Ile) (Brubaker et al., 1992; Koehl et al., 2003; Korkmaz et al., 2007; Wysocka et al., 2007) in the S1 pocket, whereas CG has both a chymotrypsin-like and a trypsin-like specificity, permitting both large, hydrophobic P1 residues (Phe, Leu, Met) and positive Lys or Arg residues (Tanaka et al., 1985). The S2 subsites of HNE and CG are bordered by Phe215, Leu99, and the flat side of the imidazole ring of the catalytic triad His57, which is quite hydrophobic. In contrast, the S2 subsite of PR3 is a deep polar pocket of increased polarity as a result of the Leu-to-Lys substitution at position 99 (Fujinaga et al., 1996). The size and hydrophobic properties of the S2 subsite of HNE and CG account for the fact that medium-sized hydrophobic side chains, such as that of Pro, are preferred at the P2 position (Tanaka et al., 1985; Réhault et al., 1999). Indeed, most commercially available chromogenic and/or fluorogenic substrates used to measure HNE and CG display a proline at the P2 position. Because of the presence of Lys99, PR3 preferentially accommodates a negatively charged residue at P2 (Hajjar et al., 2006; Korkmaz et al., 2007). The S3 subsite of CG is formed by a Lys at position 192, which favors interaction with an acidic P3 residue (Tanaka et al., 1985). The residue at position 99 in all three proteases borders both the S2 and S4 subsites, which makes it smaller and more polar in PR3 than in HNE or CG because of the Leu-to-Lys substitution. Regarding the S′ subsites, the S1′ and S2′ subsites in HNE are relatively hydrophobic and lined with Cys42–Cys58 and Phe41, Leu143, respectively. The S2′ subsite in PR3 lies in the vicinity of the charged residues Asp61 and Arg143, neither of which is conserved in HNE (Fujinaga et al., 1996). The 60 loop in PR3 containing Asp61 is significantly displaced to bring the negatively charged side chain to the S1′ and S3′ sites, thereby making them smaller and more polar. These peculiarities have been exploited to develop the first specific synthetic substrates of human PR3 (Korkmaz et al., 2004, 2007). It is noteworthy that Asp61 and Arg143 are conserved in mouse and human PR3, whereas the Lys99–Arg60 doublet in human is replaced by Asn99–Gln60 in the mouse (Fig. 5A), which explains why mouse PR3 does not cleave human PR3 substrates bearing a negatively charged residue at P2 (Kalupov et al., 2009) (Fig. 5B). The P′-S′ specificity of CG has not yet been fully elucidated. X-ray structural analysis showed that CG displays two positively charged residues (Arg41 and Arg143) that reside in the immediate vicinity of the active S2′ site and that could contribute to the specificity of the protease (Hof et al., 1996; Réhault et al., 1999).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

A, interspecies comparisons of the active site region of proteinase 3 and neutrophil elastase from men and mice. Values of electrostatic potential were mapped onto the solvent accessible surface of each protease with positively and negatively charged residues colored in blue and red, respectively. Positions of the subsites S4 to S2′ are shown in yellow. Key residues that lie in close vicinity to the substrate binding pockets are labeled by their three-letter codes. The hydrophobic patch in human PR3 is missing in the mouse homolog as the Trp218 residue is substituted by an Arg. B, sequences derived from natural human and mouse substrates that are cleaved by human and murine NE and PR3. Arrows indicate identified and potential cleavage sites. IL-8: (Padrines et al., 1994) (1); IL-18: putative cleavage sites [(2) and (7)]; SDF-1α: (Valenzuela-Fernandez et al., 2002) (3), putative cleavage site (8); NFκB: (Preston et al., 2002) (4), NFκB (Kalupov et al., 2009) (9); p21 (Dublet et al., 2005) (5), T. Kalupov et al., unpublished data (10); and MIP-2 (T. Dau and D. E. Jenne, unpublished data) (6). The structural models of murine proteases were derived from X-ray structures of human homologs by homology modeling (http://www.expasy.ch).

D. Plasma Membrane Association

Exposure of neutrophils to cytokines (TNF-α), chemoattractants (platelet-activating factor, formyl-Met-Leu-Phe, or IL-8), or bacterial lipopolysaccharide leads to rapid granule translocation to the cell surface with secretion of HNE, PR3, and CG into the extracellular medium (Owen and Campbell, 1999). A fraction of secreted HNE, PR3, and CG is detected at the surface of activated neutrophils (Owen et al., 1995a, 1997; Campbell et al., 2000). Resting purified neutrophils from peripheral blood express variable amounts of PR3 on their surface. A bimodal, apparently genetically determined, distribution has been observed with two populations of quiescent neutrophils that express or do not express the protease at their surface (Halbwachs-Mecarelli et al., 1995; Schreiber et al., 2003). The percentage of mPR3-positive neutrophils ranges from 0 to 100% of the total neutrophil population within individuals. Furthermore, the percentage of mPR3-positive neutrophils remains stable over time and is not affected by neutrophil activation (Halbwachs-Mecarelli et al., 1995).

The mechanism through which HNE and CG are associated with the outer surface of the plasma membrane of neutrophils mainly involves electrostatic interactions with the sulfate groups of chondroitin sulfate- and heparan sulfate-containing proteoglycans (Campbell and Owen, 2007). These two proteases are released from neutrophil cell surfaces by high concentrations of salt (Owen et al., 1995b, 1997; Korkmaz et al., 2005a) and after treatment with chondroitinase ABC and heparinase (Campbell and Owen, 2007). Membrane PR3 is not solubilized by high salt concentrations, which means that its membrane association is not charge dependant (Witko-Sarsat et al., 1999a; Korkmaz et al., 2009). Unlike HNE and CG, PR3 bears at its surface a hydrophobic patch formed by residues Phe166, Ile217, Trp218, Leu223, and Phe224 that is involved in membrane binding (Goldmann et al., 1999; Hajjar et al., 2008) (Fig. 3B). Several membrane partners of PR3 have been identified, including CD16/FcγRIIIb (David et al., 2005; Fridlich et al., 2006), phospholipid scramblase-1, a myristoylated membrane protein with translocase activity present in lipid rafts (Kantari et al., 2007), CD11b/CD18 (David et al., 2003), and human neutrophil antigen NB1/CD177 (von Vietinghoff et al., 2007; Hu et al., 2009), a 58- to 64-kDa glycosyl-phosphatidylinositol anchored surface receptor belonging to the urokinase plasminogen activator receptor superfamily (Stroncek, 2007). NB1 shows a bimodal distribution that superimposes with that of PR3 on purified blood neutrophils (Bauer et al., 2007). Active, mature forms of PR3 but not pro-PR3 can bind to the surface of NB1-transfected human embryonic kidney 293 cells (von Vietinghoff et al., 2008) and Chinese hamster ovary cells (Korkmaz et al., 2008b). Interaction involves the hydrophobic patch of PR3 because specific amino acid substitutions disrupting this patch in the closely related gibbon PR3 prevent binding to NB1-transfected cells (Korkmaz et al., 2008b). Decreased interaction of pro-PR3 with NB1-transfected cells is explained by the topological changes affecting the activation domain containing the hydrophobic patch residues. Together, these results support the hydrophobic nature of PR3-membrane interaction.

Because a significant fraction of HNE, PR3, and CG is detected at the surface of activated neutrophils, measuring and quantifying membrane-bound activities is of major interest for understanding their role during inflammation. The availability of sensitive and specific FRET substrates for each NSP has allowed the quantification of active proteases on the surface of stimulated neutrophils (Owen et al., 1995a, 1997; Korkmaz et al., 2005a, 2009). With the use of PR3-specific FRET substrate, it has been shown that constitutively expressed PR3 on the surface of quiescent blood neutrophils is either enzymatically inactive or is unable to interact with substrates (Korkmaz et al., 2009).

A. Antimicrobial Roles in Infections

HNE, PR3, and CG participate in direct intracellular killing of phagocytosed bacteria in phagolysosomes in combination with myeloperoxidase and reactive oxygen species generated by the NADPH oxidase complex (Kobayashi et al., 2005). In addition to intracellular killing, extracellular killing can occur through trapping of bacteria in NET composed of filamentous DNA structures decorated with cationic proteases, including NSP secreted by activated neutrophils (Brinkmann et al., 2004). HNE exerts its antimicrobial activity on Gram-negative bacteria by cleaving the outer membrane protein A of Escherichia coli (Belaaouaj et al., 2000) and other virulence factors of Salmonella enterica, Yersinia enterocolitica, and Shigella flexneri (Weinrauch et al., 2002). HNE also prevents the escape of S. flexneri from phagolysosomes in neutrophils (Weinrauch et al., 2002). Extracellular HNE and CG cleave the proinflammatory bacterial virulence factor flagellin (López-Boado et al., 2004) and degrade leukotoxin, a membrane pore-forming virulence factor of the Gram-negative bacteria Actinobacillus actinomycetemcomitans, which can lyse leukocytes and inhibit neutrophil functions (Tsai et al., 1979; Johansson et al., 2000). This pathogen has attracted attention because of its implication in severe destructive periodontal disease. However, the proteolytic activity of NSPs is not necessarily crucial for their antimicrobial activity. Their positive surface charges mediate strong binding to bacterial membranes. This binding alone may inhibit bacterial protein synthesis and induces membrane depolarization and disruption (Zasloff, 2002). Several peptides derived from the CG structure possess antimicrobial properties in vitro (Shafer et al., 1996, 2002). The antimicrobial properties of PR3 are also independent of its protease activity. PR3 efficiently kills both Gram-negative bacteria such as E. coli and Gram-positive bacteria such as Streptococcus faecalis, as well as fungi such as Candida albicans. PR3 also processes human cathelicidin (hCAP18) to its active LL-37 form in the extracellular environment after neutrophil activation (Sørensen et al., 2001). Human cathelicidin is an 18-kDa cationic anti-bacterial protein produced by epithelial cells of the gastrointestinal and respiratory tract and by sebocytes of the skin but also by blood cells, including neutrophils (Doss et al., 2010). Antimicrobial activity requires a proteolytic cleavage of hCAP-18 to liberate the active C-terminal 37-residue polypeptide that exerts broad antimicrobial activity against Gram-negative and -positive bacteria (Dürr et al., 2006).

Mice deficient in NE and CG have been employed as models for exploring their antibacterial functions in microbial infectious diseases. Previous studies showed that mice deficient in NE and CG are more susceptible to infection by Gram-negative and -positive bacteria (Reeves et al., 2002). In neutrophils from NE-deficient [NE(−/−)] mice infected by S. flexneri, 12% of bacteria escaped from the phagolysosome (Weinrauch et al., 2002).

B. Roles in Inflammatory Process Regulation

NSPs are abundantly secreted into the extracellular environment upon neutrophil activation at inflammatory sites. A fraction of the released proteases remain bound in an active form on the external surface of the plasma membrane so that both soluble and membrane-bound NSPs are able to proteolytically regulate the activities of a variety of chemokines, cytokines, growth factors, and cell surface receptors. Secreted proteases also activate lymphocytes and cleave apoptotic and adhesion molecules (Bank and Ansorge, 2001; Pham, 2006; Meyer-Hoffert, 2009). Thus, they retain pro- and anti-inflammatory activities, resulting in a modulation of the immune response at sites of inflammation.

A. Serine Protease Inhibitors

Serpins are the largest and most diverse family of protease inhibitors; more than 1000 members have been identified in human, plant, fungi, bacteria, archaea, and certain viruses (Silverman et al., 2001; Mangan et al., 2008). They share a similar highly conserved tertiary structure and similar molecular weight of approximately 50 kDa. Human serpins belong to the first nine clades (A–I) of the 16 that have been described based on phylogenic relationships (Irving et al., 2000; Silverman et al., 2001; Mangan et al., 2008). For historical reasons, α1-protease inhibitor (α1-PI) was assigned to the first clade. Clade B, also known as the ov-serpin clan because of the similarity of its members to ovalbumin (a protein that belongs to the serpin family but lacks inhibitory activity), is the second largest clan in humans, with 15 members identified so far. Ov-serpin clan members are generally located in the cytoplasm and, to a lesser extent, on the cell surface and nucleus (Remold-O'Donnell, 1993).

Serpins play important regulatory functions in intracellular and extracellular proteolytic events, including blood coagulation, complement activation, fibrinolysis, cell migration, angiogenesis, and apoptosis (Potempa et al., 1994). Serpin dysfunction is known to contribute to diseases such as emphysema, thrombosis, angioedema, and cancer (Carrell and Lomas, 1997; Lomas and Carrell, 2002). Most inhibitory serpins target trypsin-/chymotrypsin-like serine proteases, but some, termed “cross-class inhibitors,” have been shown to target cysteine proteases (Annand et al., 1999). The crystal structure of the prototype plasma inhibitor α1-PI revealed the archetype native serpin fold (Loebermann et al., 1984). All serpins typically have three β-sheets (termed A, B, and C) and eight or nine α-helices (hA–hI) arranged in a stressed configuration. The so-called reactive center loop (RCL) of inhibitory molecules determines specificity and forms the initial encounter complex with the target protease (Potempa et al., 1994; Silverman et al., 2001). Serpins inhibit proteases by a suicide substrate inhibition mechanism. The protease initially recognizes the serpin as a potential substrate using residues of the reactive center loop and cleaves it between P1 and P1′ This cleavage allows insertion of the cleaved RCL into the β-sheet A of the serpin, dragging the protease with it and moving it over 71 Å to the distal end of the serpin to form a 1:1 stoichiometric covalent inhibitory complex (Huntington et al., 2000). Such cleavage generates a ∼4-kDa C-terminal fragment that remains noncovalently bound to the cleaved serpin. Displacement of the covalently attached active site serine residue from its catalytic partner histidine explains the loss of catalytic function in the covalent complex. The distortion of the catalytic site structure prevents the release of the protease from the complex, and the structural disorder induces its proteolytic inactivation (Huntington et al., 2000). Covalent complex formation between serpin and serine proteases triggers a number of conformational changes, particularly in the activation domain loops of the bound protease (Dementiev et al., 2006).

B. Canonical Inhibitors

Canonical inhibitors include a still increasing number of families with different folds but sharing a canonical conformation of their inhibitory protease binding loop in the P3-P3′ region (Bode and Huber, 1992). This loop is a convex, extended, and solvent-exposed structure that is highly complementary to the concave active site of the enzyme. A huge number of canonical inhibitors have been isolated from various cells, tissues, and organisms, including plant seeds, avian eggs, and various body fluids (Krowarsch et al., 2003). The chelonianin family of canonical inhibitors includes the physiological endogenous inhibitors of NSPs, SLPI, elafin, and its precursor trappin-2 (Moreau et al., 2008).

C. α-Macroglobulins

Human α2-macroglobulin is a polyvalent homotetrameric inhibitor of 725 kDa present at a concentration of 2 mg/ml in serum (Petersen, 1993). It inhibits all classes of proteases, including NSPs (Travis, 1988). Because its high molecular mass impairs diffusion to inflammatory sites during neutrophil extravasation, its major role is probably restricted to controlling protease activity within the circulation. Its mechanism of interaction with proteases has been described in detail (Sottrup-Jensen, 1989). Its covalent association with proteases sterically shields their active sites from large molecular substrates, hence permitting enzymatic hydrolysis of only small synthetic substrates (Doan and Gettins, 2007).

A. Chronic Inflammatory Lung Diseases

In many instances, the initiation and propagation of lung damage is a consequence of an exaggerated inappropriate inflammatory response, which includes the release of proteases and leukocyte-derived cytotoxic products (Owen, 2008b; Roghanian and Sallenave, 2008). Inflammation is a physiological protective response to injury or infection consisting of endothelial activation, leukocyte recruitment and activation, vasodilation, and increased vascular permeability. Although designed to curtail tissue injury and facilitate repair, the inflammatory response sometimes results in further injury and organ dysfunction. Inflammatory chronic lung diseases, chronic obstructive pulmonary disease, acute lung injury, acute respiratory distress syndrome, and cystic fibrosis are syndromes of severe pulmonary dysfunction resulting from a massive inflammatory response and affecting millions of people worldwide. The histological hallmark of these chronic inflammatory lung diseases is the accumulation of neutrophils in the microvasculature of the lung. Neutrophils are crucial to the innate immune response, and their activation leads to the release of multiple cytotoxic products, including reactive oxygen species and proteases (serine, cysteine, and metalloproteases). The physiological balance between proteases and antiproteases is required for the maintenance of the lung's connective tissue, and an imbalance in favor of proteases results in lung injury (Umeki et al., 1988; Tetley, 1993). A number of studies in animal and cell culture models have demonstrated a contribution of HNE and related NSPs to the development of chronic inflammatory lung diseases. Available preclinical and clinical data suggest that inhibition of NSP in lung diseases suppresses or attenuates the contribution of NSP to pathogenesis (Chughtai and O'Riordan, 2004; Voynow et al., 2008; Quinn et al., 2010). HNE could also participate in fibrotic lung remodeling by playing a focused role in the conversion of latent transforming growth factor-β into its biologically active form (Chua and Laurent, 2006; Lungarella et al., 2008).

B. Anti-Neutrophil Cytoplasmic Autoantibody-Associated Vasculitides

ANCA-associated vasculitides encompasses a variety of diseases characterized by inflammation of blood vessels and by the presence of autoantibodies directed against neutrophil constituents. These autoantibodies are known as ANCAs (Kallenberg et al., 2006). In Wegener granulomatosis (WG), antibodies are mostly directed against PR3. WG is a relatively uncommon chronic inflammatory disorder first described in 1931 by Heinz Karl Ernst Klinger as a variant of polyarteritis nodosa (Klinger, 1931). In 1936, the German pathologist Friedrich Wegener described the disease as a distinct pathological entity (Wegener, 1936, 1939). WG is characterized by necrotizing granulomatous inflammation and vasculitis of small vessels and can affect any organ (Fauci and Wolff, 1973; Sarraf and Sneller, 2005). The most common sites of involvement are the upper and lower respiratory tract and the kidneys. WG affects approximately 1 in 20,000 people; it can occur in persons of any age but most often affects those aged 40 to 60 years (Walton, 1958; Cotch et al., 1996). Approximately 90% of patients have cold or sinusitis symptoms that fail to respond to the usual therapeutic measures and that last considerably longer than the usual upper respiratory tract infection. Lung involvement occurs in approximately 85% of the patients. Other symptoms include nasal membrane ulcerations and crusting, saddle-nose deformity, inflammation of the ear with hearing problems, inflammation of the eye with sight problems, and cough (with or without hemoptysis).

The cause of WG has not yet been elucidated, but infection with S. aureus could well contribute to its pathogenicity (Kallenberg, 2008). Deficiency in α1-PI also is a risk factor of WG. Studies of gene polymorphisms showed that the frequency of the α1-PI deficiency allele PI*Z is increased in WG and is related to its outcome (Esnault et al., 1993; Callea et al., 1997; Needham and Stockley, 2004). WG is characteristically associated with the presence of ANCA, directed against proteins of neutrophilic granulocytes and monocytes stored in cytoplasmic granules (Reumaux et al., 2004; Kallenberg et al., 2006; Kallenberg, 2008). The occurrence of this cytoplasmic staining pattern was first reported in 1982 in patients with necrotizing glomerulonephritis (Davies et al., 1982). Jenne et al. (1990) identified PR3 as the target antigen for ANCA. Though most patients with WG are positive for ANCAs, antibody titer in the serum of patients does not always correlate with the severity of the disease, which raises the question of the pathogenicity of ANCAs (Merkel et al., 2009). In addition to cytoplasmic ANCA (or cANCA, which usually recognizes PR3 in WG), perinuclear ANCA (pANCA) have been demonstrated by enzyme-linked immunosorbent assay and indirect immunofluorescence on ethanol-fixed neutrophils (Hoffman and Specks, 1998). pANCA are predominantly directed against myeloperoxidase but some are also directed against HNE and CG (Savige et al., 1998) and can additionally be found in a variety of inflammatory diseases, including sometimes Crohn disease, rheumatoid arthritis, and drug-induced vasculitis (Hoffman and Specks, 1998; Radice and Sinico, 2005). The precise subcellular localization resulting in the apparent perinuclear distribution of pANCA remains uncertain (Yang, 1997; Hoffman and Specks, 1998).

Mouse monoclonal anti-PR3 antibodies and cANCA from WG patients recognize conformational epitopes on PR3 (Bini et al., 1992) that may vary during the course of the disease (Rarok et al., 2003). Four groups of mouse monoclonal anti-PR3 antibodies have been identified. Antibodies in the same group recognize similar epitopes on PR3 (Van Der Geld et al., 1999). Autoantibodies to HNE are responsible for a syndrome of midline destructive vasculitis, which is associated with use of cocaine via nasal insufflations (Wiesner et al., 2004).

Resting neutrophils from purified peripheral blood of healthy persons express PR3 on their membranes in a genetically determined manner, and this proportion of PR3 is increased in a stepwise fashion by neutrophil priming and subsequent activation (Schreiber et al., 2003). Circulating blood neutrophils from patients with active WG express PR3 on their cell membranes in larger quantities than healthy persons as a result of an ongoing inflammatory process (Rarok et al., 2002). Likewise, neutrophils from healthy volunteers can be primed with proinflammatory cytokines (e.g., TNF-α to increase the fraction of surface-exposed PR3 directly accessible to circulating cANCA) (Csernok et al., 1994). Neutrophils become fully activated upon cANCA binding to their membranes. They produce reactive oxygen species and release lysosomal proteases from their azurophilic granules to the pericellular environment. In recent work, Kessenbrock et al. (2009) showed that TNF-α-primed neutrophils stimulated by cANCA produce NET containing PR3 and myeloperoxidase. Deposition of NET in inflamed kidneys and the presence of circulating myeloperoxidase-DNA complexes suggest that NET formation contributes to focal necrotizing glomerulonephritis and enhances the autoimmune response against neutrophil components in individuals with small-vessel vasculitis.

C. Hereditary Neutropenias

Neutropenia is a hematological disorder characterized by an abnormally low number of neutrophils (Horwitz et al., 2007). The normal neutrophil count fluctuates across human populations and within individual patients in response to infection but typically lies in the range of 1.5 to 5 × 10⁹ cells/liter. Neutropenia is categorized as severe when the cell count falls below 0.5 × 10⁹ cells/liter. Hence, patients with neutropenia are more susceptible to bacterial infections and, without prompt medical attention, the condition may become life-threatening. Common causes of neutropenia include cancer chemotherapy, drug reactions, autoimmune diseases, and hereditary disorders (Berliner et al., 2004; Schwartzberg, 2006).

Cyclic neutropenia and severe congenital neutropenia are the two main forms of hereditary neutropenia (Ancliff, 2003; Horwitz et al., 2007). Their prevalence has been estimated at 1 in 200,000 in newborn. Cyclic neutropenia (MIM 162800) is an autosomal disease in which blood-cell production from the bone marrow oscillates with a 21-day periodicity. Circulating neutrophils vary between almost normal numbers and zero. Monocytes also cycle but do so in a phase that is opposite that of neutrophils (Berliner et al., 2004). During neutropenic periods, affected persons are at risk for opportunistic infections (Reimann and DeBerardinis, 1949). The periodic fluctuation between high and low neutrophil counts could be explained by a feedback mechanism through which mature neutrophils inhibit maturation of progenitor cells (Horwitz et al., 2003; Salipante et al., 2009). The first case of human cyclic neutropenia was reported by Leale (1910), but only in 1999 did Horwitz et al. (1999) reported that mutations of the elastase gene ELA2, since then renamed ELANE, were responsible for human cyclic neutropenia.

Severe congenital neutropenia (MIM 202700), also called Kostmann disease, refers to a noncyclic neutropenia present at birth (absolute neutrophil count below 0.5 × 10⁹/liter). It was first reported as infantile agranulocytosis in a large Swedish family of consanguineous parents (Kostmann, 1956; Carlsson et al., 2006). A characteristic of severe congenital neutropenia is that the bone marrow displays a defect in maturation of developing myeloid cells at the promyelocyte stage. Mutations in ELANE are present in approximately 60% of severe congenital neutropenia cases, and approximately 50 different mutations scattered throughout the coding regions have been identified. Many of the mutations do not lead to loss of enzymatic activity (Horwitz et al., 2007). Bioinformatic analysis of different pathogenic missense mutations in ELANE suggests diverse deleterious effects on structural and physicochemical properties of HNE (Thusberg and Vihinen, 2006). Additional rare cases of severe congenital neutropenia arise from mutations in the genes encoding zinc finger transcription factor GFI1 (Person et al., 2003), glucose-6-phosphatase (Boztug et al., 2009), and HS1-associated protein X-1 (Klein et al., 2007). Patients with severe congenital neutropenia present symptoms of recurrent bacterial infections mostly in the skin, the respiratory system, and the oral cavity, whereas some patients may also present with severe generalized infections. Neutropenic patients often respond to granulocyte colony-stimulating factor, which stimulates neutrophil production in bone marrow. Bone marrow transplantation is warranted in patients who are unresponsive to granulocyte colony-stimulating factor or who develop leukemia.

The heterotetrameric adaptor protein complex 3 (AP-3) residing on the cytoplasmic surface of intracellular vesicles has been shown to function in the sorting of proteins to the endosomal/lysosomal system (Badolato and Parolini, 2007). Mutation of the gene encoding the AP-3 β subunit, which disrupts the complex, have been found to cause Hermansky Pudlak syndrome type 2 (MIM 203300) (Dell'Angelica et al., 1999; Huizing et al., 2002), a disorder characterized by oculocutaneous albinism, platelet granular dysfunction, and neutropenic immunodeficiency (Huizing et al., 2008), and also to cause a canine form of cyclic neutropenia associated with pigmentation dilution known as “gray collie syndrome” (Lothrop et al., 1987). Benson et al. (2003) have argued that mutations in either the HNE or AP-3 gene perturbs intracellular trafficking of HNE, implying that mislocalization of HNE may be involved in the pathogenesis of the congenital neutropenias associated with ELANE mutations. Alternatively, mutations in ELANE may induce the unfolded protein response as a consequence of incorrect folding and targeting of the mutant proteins, with associated programmed cell death (Köllner et al., 2006; Grenda et al., 2007). Expression of mutant HNE, but not wild-type HNE, strongly induced BiP/GRP78 mRNA expression and XBP1 mRNA splicing, two classic markers of the unfolded protein response (Grenda et al., 2007).

D. Papillon-Lefèvre Syndrome

Papillon-Lefèvre syndrome (PLS; MIM 245000) is a hereditary skin disease characterized by a diffuse palmoplantar keratosis and severe periodontitis, leading to premature loss of both the deciduous and permanent teeth (Dhanrajani, 2009). Approximately 15 to 20% of patients are predisposed to recurrent infections. The combination of oral retinoids and antibiotics complemented with professional tooth cleaning are currently used in the treatment of severe periodontitis occurring with this disorder (Dhanrajani, 2009). PLS is an extremely rare inherited autosomal recessive disease with an estimated prevalence of 1 to 4 cases per million people, affecting children between the ages of 1 and 4 (Gorlin et al., 1964). It was first described by two French physicians, Papillon and Lefèvre (1924), as a form of “Mal de Meleda” and was only later classified as a distinct entity (Dekker and Jansen, 1956). More than 200 cases worldwide have been reported in the literature. Consanguinity of the parents was observed in approximately one third of the cases described. Both sexes are equally affected; no racial predominance seems to exist (Haneke 1979; Hattab et al., 1995). Loss-of-function mutations in the DPPI gene (CSTC) on chromosome 11q14-q21 are strongly correlated with PLS and the related Haim-Munk syndrome (Hart et al., 1999, 2000; Toomes et al., 1999).

The exact cause of periodontal disease in PLS has not been elucidated, but it has been attributed to a neutrophil defect resulting in decreased phagocytosis and chemotaxis (Brown et al., 1993; Bullon et al., 1993; Firatli et al., 1996). Association of neutrophil or myeloid abnormalities with severe periodontal destruction and the development of rapid periodontal infections in experimentally induced neutropenia in animals suggest that neutrophils are protective against periodontal destruction. The 116-kDa pore-forming leukotoxin is the prime virulence factor of the periodontopathogen A. actinomycetemcomitans and promotes its colonization while inducing granule exocytosis and lysis of primed neutrophils. NSPs degrade this toxin extracellularly (Johansson et al., 2000, 2001). Furthermore, they convert the neutrophil-derived hCAP-18 into LL-37, an antimicrobial peptide with activity against A. actinomycetemcomitans (de Haar et al., 2006). In PLS, the loss of DPPI activity within neutrophils results in a severe reduction in the activity and stability of NSPs (Pham et al., 2004; de Haar et al., 2004), leading to the generation of lower levels of LL-37 by neutrophils (de Haar et al., 2006) and reduced efficiency of degradation of leukotoxin produced by A. actinomycetemcomitans (de Haar et al., 2006).

Pham et al. (2004) have shown that neutrophils from most patients with PLS display normal microbicidal activity against E. coli and S. aureus despite their severe deficiency in NSPs, but neutrophils from certain patients were defective in microbial killing. This clinical observation suggests that NSPs are not essential antimicrobial molecules and may explain why only a few patients with PLS demonstrate severe recurrent systemic infections.

A. Therapeutic Inhibitors

CF, ALI, and ARDS are characterized by the presence of pathogens in the lung. Migration of neutrophils to sites of infection is beneficial because of their capacity to kill pathogens intracellularly after phagocytosis and extracellularly upon release of NET. However, the presence of neutrophils at infection sites may also be deleterious when an excessive, uncontrolled amount of NSP is released into the extracellular medium and leads to lung matrix degradation. Supplementation with exogenous, natural, synthetic, or recombinant NSP inhibitors administered by the vascular route or by aerosols has been tried as a means for restoring the protease/inhibitor balance but without fully convincing results. This is not surprising, however, because 1) exogenous inhibitors do not target all three NSPs released from activated neutrophils, (Owen and Campbell, 1999; Owen, 2008b) and 2) proteolysis in the lung, for example, may occur even in the presence of active endogenous inhibitors as neutrophils adhere to extracellular matrix components and to the vascular endothelium and deposit NSPs in direct proximity to their protein substrates (Owen and Campbell, 1999; Owen, 2008b). Extracellular proteases may also escape inhibition through their association with cell membranes or with negatively charged macromolecular components such as DNA or proteoglycans. Thus, an ideal NSP inhibitor that would control proteolytic activities within given limits should be able to inhibit HNE, PR3, and CG with similar efficiency; would resist rapid oxidation and proteolysis; and would be of small size, thereby affording better access to NSPs when they are bound to extracellular matrices and to molecular or cellular components. It should also resist aerosolization and in vivo degradation. None of the inhibitors developed to fight chronic inflammatory diseases so far address all of these criteria. Nevertheless, replacement therapy by intravenous administration of α1-PI has long been shown to be efficient for α1-PI deficiency associated with emphysema (Wewers et al., 1987; Hubbard et al., 1988), and aerosolized administration of α1-PI or SLPI to patients with cystic fibrosis suppressed or significantly reduced HNE in the respiratory epithelial lining fluid and restored the anti-HNE capacity (McElvaney et al., 1991, 1993). α1-PI augmentation therapy also abrogates HNE-induced up-regulation of cathepsin B and MMP-2 expression, further demonstrating the central role of this protease as a target in many inflammatory diseases (Geraghty et al., 2007, 2008). A remaining problem, however, is that serpins may have impaired access to the microenvironment created by the contact between neutrophils and proteins of the extracellular matrix because of their large molecular weight and their potential sensitivity to oxidation and proteolysis. On the other hand, therapeutically applied inhibitors should not completely abrogate the putative physiological roles of proteases in the pericellular milieu of the neutrophil. This might happen with inhibitors that instantaneously and irreversibly inactivate the target proteases. Hence, competitive inhibitors that can dissociate from proteases, as after withdrawing inhibitor administration, seem to be more appropriate and better suited to curtail excessive proteolytic activity. Irreversible inhibitors such as serpins, however, may also preserve a basal low level of proteolytic activity provided the delay time for full inhibition is long enough to allow substantial substrate cleavage by the protease before its complete inactivation (Bieth, 1986). The delay time to reach a complete irreversible inhibition depends on the local concentration of the two partners, the association rate constant, and the specificity constant toward the biological substrate. In this way, partial oxidative inactivation of α1-PI in inflammatory biological fluids contributes to the maintenance of some local HNE activity and its pericellular functions (Luisetti and Travis, 1996). The chemical inhibitor sivelestat and the recombinant low-molecular-weight inhibitor depelstat are resistant to oxidative and proteolytic inactivation but are targeting only HNE (Kawabata et al., 1991; Attucci et al., 2006). Most synthetic and chimeric elastase inhibitors developed so far have not proceeded beyond phase II of development because of side effects in addition to poor efficacy (Tremblay et al., 2003). Polyvalent recombinant protein inhibitors that target all three NSPs have recently been developed (Zani et al., 2009). One is a recombinant chimera resulting from the fusion of the C-terminal domain of SLPI and elafin; the other is a trappin-2 variant in which the P1 residue Ala is replaced by Leu. These recombinant low-M_r proteins are tight-binding inhibitors of all three NSPs, demonstrating nanomolar or subnanomolar K_i values. They inhibit membrane-bound NSPs but do not resist oxidative inactivation because of the presence of a critical methionine residue at the P1′ position (Zani et al., 2009). Further investigation is needed to identify critical residues involved in the interaction of each protease with physiological, recombinant, and synthetic inhibitors. Specific and highly sensitive FRET substrates designed and developed to measure the activities of HNE, PR3, and CG at the surface of triggered neutrophils and in whole biological fluids are important tools for quantifying active HNE, PR3, and CG in lung diseases (Korkmaz et al., 2008a). These substrates, allowing measurement of proteolytic activities in both solid and soluble fraction of BAL fluid or sputum from patients with chronic inflammatory diseases, will no doubt aid the investigation of spatiotemporal efficacy of administered therapeutic inhibitors.

cANCA are directed against PR3 and are a diagnostic marker of WG, although cANCAs are not found in approximately 10% of patients with Wegener granulomatosis (Radice and Sinico, 2005). Their simultaneous interaction with mPR3 and Fc receptors is believed to trigger full activation and degranulation of neutrophils around small vessels in patients with active disease (Kallenberg, 2008). Unlike HNE and CG, PR3 is constitutively expressed on the surface of circulating neutrophils but is proteolytically inactive (Korkmaz et al., 2009). This prevents constitutive PR3 from interacting with circulating protease inhibitors but not with cANCAs, which, in turn, may activate neutrophils and favor expression of active NSPs at the cell surface. Designing a specific PR3 inhibitor that removes active PR3 from the surface of activated neutrophils would probably help reduce ANCA binding to the cell surface and therefore small vessel inflammation. Unfortunately, such an inhibitor is not yet available. The main circulating PR3 inhibitor is α1-PI, which forms soluble protease-inhibitor complexes but inhibits PR3 at a rate approximately 100 times lower than that of HNE. This may result in a transient increase of active mPR3 on primed neutrophils with cANCA binding and subsequent neutrophil activation. Although it was found that heterozygotes for the PiZ variant of the α1-PI gene are at greater risk than the general population for developing WG (Elzouki et al., 1994), there is currently no evidence that AAT augmentation therapy is effective in the treatment of WG associated with AAT deficiency (Silverman and Sandhaus, 2009). In keeping with this, we recently found that pathogenic cANCAs react with membrane-bound PR3 faster than does α1-PI (Kuhl et al., 2010), which supports the idea that the inhibitor cannot significantly impair neutrophil triggering by cANCA.

Designing specific inhibitors capable of targeting a particular protease such as PR3 (for the treatment of WG) or polyvalent inhibitors (to improve control of multiple proteases in infectious and inflammatory disorders) requires a deeper investigation of the interaction between protease active sites and inhibitors at the molecular level. Determination of the three-dimensional structures of protease-inhibitor complexes and characterization of protease specificities through kinetic analysis using synthetic substrates has provided information indicating the residues and subsites that are involved in such interactions. A strategy for engineering α1-PI to become a specific and sensitive inhibitor of PR3 could consist of substituting the exposed reactive center loop of α1-PI, which could then interact with charged residues present in the active site of PR3. This approach might also help to elucidate the mechanism of degradation by PR3 of the α1-PI RCL (Duranton and Bieth, 2003). Such a hypothetical recombinant serpin-like inhibitor might be capable of rapidly removing active PR3 from the activated neutrophil membranes and could be helpful in modulating cANCA-neutrophil interactions in WG.

This approach could be extended by development of monoclonal humanized mini antibodies capable of masking the essential cANCA epitope on mPR3, which will impair neutrophil activation. It is noteworthy that monoclonal antibodies represent a promising class of biotherapeutics for the treatment of autoimmune diseases and cancers, and more than 20 mAbs have been approved for therapy or are in clinical trials (Lynch et al., 2009). Rituximab, a chimeric monoclonal antibody directed against CD20, which is present primarily on B cells and is currently used in the treatment of non-Hodgkin's lymphoma and rheumatoid arthritis (Plosker and Figgitt, 2003; Fleischmann, 2009), significantly decreases the level of anti PR3 autoantibodies, resulting in patient clinical remission (Ferraro et al., 2008) that is possibly related to a lower neutrophil activation.

B. Reducing Intracellular Levels of Active Elastase, Proteinase 3, and Cathepsin G

Impairment of access of therapeutic inhibitors to extracellularly active NSPs represents a possible limitation of an inhibitor-based approach to the treatment of infectious and inflammatory pulmonary diseases. This may occur both because of the composition of the extracellular medium (mucus, NET, proteoglycans, and other macromolecules) and because of the spatiotemporal localization of NSPs. Reducing the intracellular levels of active NSPs in mature neutrophils, without perturbing their functions and accumulation at infectious sites, in theory, represents an alternative approach for limiting uncontrolled proteolysis (Fig. 6). Potentially, this could be achieved through disruption of the intracellular trafficking of mature NSPs. A decrease in the immune reactive level of NSPs is observed when intracellular proteins involved in protein sorting and vesicle transport are mutated in human diseases. Mutations of the AP-3 cargo protein and of the CHS protein in Hermansky Pudlak syndrome-2 and Chediak-Higashi syndrome, respectively, result in decreased intracellular levels of HNE and CG (Ganz et al., 1988; Fontana et al., 2006). AP-3 cargo protein and the CHS protein are believed to be involved in vesicle formation and transport. A low HNE level is also observed in neutrophils from neutropenia patients with HS1-associated protein X-1 mutations (B. Korkmaz and M. S. Horwitz, unpublished results). The proteoglycan serglycin has recently been shown to play a role in the sorting of NE (Niemann et al., 2007). In serglycin knockout mice, NE is absent from mature neutrophils but serglycin inactivation has no effect on correct granular targeting and on intracellular quantities of CG and PR3. Furthermore, the lack of serglycin has no effect on neutrophil count and ultrastructure (Niemann et al., 2007), demonstrating that inactivation of a protein involved in the sorting of NE does not necessarily induce neutropenia, whereas dysfunctional AP3 and CHS proteins impair granule formation and protein targeting in general. Thus, there is cumulative evidence that the intracellular level of HNE and possibly of other related NSPs depends on the functionality of molecules involved in its intracellular trafficking.

Identifying and inactivating other molecular targets that deregulate protease trafficking without inducing neutropenia or altering neutrophil function or chemotaxis could possibly be exploited for a therapeutic strategy aiming at controlling the production and the activity of NSPs in chronic inflammatory lung diseases. It has been shown, however, that NSPs liberated at the inflammatory sites regulate local production of cytokines, leading to neutrophil infiltration and accumulation. In fact, neutrophil infiltration at sites of inflammation is severely diminished in PR3(−/−)/NE(−/−) mice unless only NE is deficient (Kessenbrock et al., 2008). In that case, neutrophil infiltration and accumulation is only marginally affected. Because both PR3 and NE in mice share similar active sites (Fig. 5) and are able to activate proinflammatory cytokines IL-8 and IL-1β, it can be concluded that only one is necessary to promote neutrophil recruitment at inflammatory sites. In humans, however, IL-8 and IL-1β are specifically activated by human PR3. This means that altered intracellular trafficking of HNE is not expected to modify neutrophil migration toward inflammatory sites. Using an animal (primate) model in which the NSP specificities are closer to those in humans should permit the testing of this hypothesis.

C. Reducing Neutrophil Accumulation at Inflammatory Sites

In noninfectious lung diseases and other inflammatory human diseases characterized by an excessive neutrophil accumulation, such as rheumatoid arthritis and inflammatory bowel diseases, containment of NSP activities could occur before proteases are released from neutrophils at the inflammatory site (Fig. 6). Strategies for impairing inflammatory cell recruitment to the lung have already been proposed and explored in vitro and in vivo. For example, inhibitors of phosphodiesterase E4 (PDE4) prevent hydrolysis of cAMP in inflammatory cells, thus reducing their chemoattractant and protease-secreting function (Fan Chung, 2006). By suppressing the release of inflammatory signals, PDE4 inhibitors display a high therapeutic potential in COPD and asthma (Schmidt et al., 1999; Giembycz, 2002). Intraperitoneal administration of the PDE4 inhibitor Rolipram in mice receiving an intranasal challenge of LPS significantly reduces neutrophil accumulation in the lungs (H. Korideck and J. D. Peterson, unpublished data). This observation holds true for mice undergoing long-term exposure to cigarette smoke and treated orally with the long-acting PDE4 inhibitor roflumilast (Martorana et al., 2005). Anti-oxidant supplementation, to block the transcription factor NF-κB (Szulakowski et al., 2006) and thus the production of several pro-inflammatory molecules causing lung inflammation, also reduces the accumulation of inflammatory cells, and the same result is obtained using inhibitors of chemokine receptors (Donnelly and Barnes, 2006).

DPPI is the major protease involved in the activation of NSPs; it also influences the total intracellular levels of these proteases, as observed in neutrophils from patients with PLS (Pham et al., 2004). In addition, mice deficient in DPPI demonstrate impaired neutrophil accumulation in their joints in response to noninfectious inflammatory stimuli but are protected against cartilage destruction in acute arthritis experimentally induced by immune complexes (Adkison et al., 2002). The decrease in neutrophil accumulation in DPPI(−/−) mice is correlated with cytokine/chemokine dysregulation at inflammatory sites resulting from the absence of proteolytically active NSPs (Adkison et al., 2002; Akk et al., 2008). In patients with PLS, DPPI deficiency leads to the accumulation of neutrophils in the blood as they are hampered to enter the extravascular tissue. Taken together, these observations support the idea that DPPI inhibitors could be useful for the treatment of a number of important inflammatory diseases, including lung diseases. Using such inhibitors in infectious diseases, on the other hand, seems to be risky because efficient elimination of pathogens requires both efficient neutrophil recruitment to the site of microbial invasion and rapid destruction of microbes in phagolysosomes. As a significant number of PLS patients display a normal anti-microbial response to common pathogens, NSPs are presumably not essential to overcome invading pathogens (Pham, 2006). Hence, redundancy in antimicrobial defense systems provides a favorable window for the application of NSP inhibitors. In a therapeutic inhibitor-based approach, DPPI may be specifically targeted and inactivated in the bone marrow. DPPI inhibition can be achieved with general inhibitors of cysteine proteases such as diazomethyl ketones, dipeptide nitriles, vinylsulfones, and (2S,3S)-3-[[[(1S)-1-[[[4-[(aminoiminomethyl)amino]butyl]amino]carbonyl]-3-methylbutyl]amino]carbonyl]-2-oxiranecarboxylic acid (E-64) (Barrett et al., 1982; McGuire et al., 1992; Méthot et al., 2007). Among the existing inhibitors, Gly-Phe-diazomethylketone is one of the most effective compounds (Kam et al., 2004; Méthot et al., 2007). Dipeptide nitriles and vinyl sulfone inhibitors are nontoxic to cells and effective toward intracellular DPPI and are therefore potentially suitable for functional studies. Long-term administration of a dipeptide nitrile DPPI inhibitor transported by a poly(ethylene glycol)-based vehicle results in the nearly complete inhibition of DPPI and causes marked reduction of neutrophil accumulation at inflammatory site (Méthot et al., 2008). Specific, nontoxic inhibitors of DPPI may be delivered to the bone marrow by liposomal formulations originally developed for the treatment of leukemias (Gregoriadis et al., 1974; Kohlschütter et al., 2008). As a consequence of therapeutic DPPI inhibition, pro-NSPs would not be converted into active enzymes and that in turn would modulate the levels of inflammatory mediators, such that neutrophils will not infiltrate and accumulate at inflammatory sites. In vivo inhibition of NSP activation by targeting DPPI requires high fractional inhibition of this protease in the bone marrow (Méthot et al., 2008). Because mature neutrophils are released from the bone marrow into the peripheral circulation after approximately 11 to 14 days of development (Walker and Willemze, 1980), a continuous administration of DPPI inhibitor over this period would probably be required. Possible side effects of such a prolonged administration cannot be predicted yet, because severe periodontitis and hyperkeratosis, as found in DPPI-deficient patients with PLS, are not observed in mice successfully treated with DPPI inhibitor to prevent NSP activation (Méthot et al., 2008).

Loss-of-function mutations in DPPI have provided important insight into its role in inflammation, but little information is available on intracellular NSP activation during neutrophil development. Hematopoietic cell lines that express DPPI and are able to activate and target NSPs in a fashion similar to that of human neutrophils represent good models for analyzing the processing of proenzymes by DPPI. The HL-60 (human promyelocytic leukemia cells) cell line contains DPPI and expresses HNE, PR3, and CG (Meagher and Cotter, 1988; Bories et al., 1989; Yoshimura and Crystal, 1992). But other mammalian hematopoietic cell lines [rat basophilic leukemia cells (RBL-1), murine 32D cells and the mast cell line-1] produce DPPI and can be transfected to produce active NSPs that will be stored in granules in a similar fashion as in human neutrophils (Gullberg et al., 1994, 1995; Specks et al., 1996; Garwicz et al., 1997; Li and Horwitz, 2001). Generally, processing and targeting is more rapid and efficient in RBL-1 cells than in other cell lines. Thus, RBL-1 cells have been successfully used to investigate NSP activation by DPPI and to study the influence of HNE mutations on the intracellular trafficking of the protease. Identification of the location of rat DPPI and its colocalization with active HNE by fluorescence microscopy may lead to a better understanding of where pro-NSPs are activated, a question remaining unanswered until now. To this end, chemical fluorescent tracers that irreversibly inactivate human NSPs could be used.

Recently, we showed that pharmacological inactivation of DPPI in RBL cells transfected with HNE impairs HNE targeting to the nuclear envelope and the same result was seen using neutrophils purified from patients with DPPI mutations (B. Korkmaz and M. S. Horwitz, unpublished results). PR3 transfected into RBL-1 cells is present in lysosomes but also on cellular surfaces. One might expect that inactivation of intracellular DPPI by cell permeable inhibitors could well reduce the interaction of PR3 with cellular membranes, because purified pro-PR3 displays less affinity to cellular surfaces than mature PR3 (Korkmaz et al., 2008b). This may be of major importance in WG, because it would allow for a reduction in membrane binding of pathogenic autoantibodies, which results in neutrophil priming and activation.