Gelatinase B/MMP-9 and neutrophil collagenase/MMP-8 process the chemokines human GCP-2/CXCL6, ENA-78/CXCL5 and mouse GCP-2/LIX and modulate their physiological activities

Abstract

On chemokine stimulation, leucocytes produce and secrete proteolytic enzymes for innate immune defence mechanisms. Some of these proteases modify the biological activity of the chemokines. For instance, neutrophils secrete gelatinase B (matrix metalloproteinase-9, MMP-9) and neutrophil collagenase (MMP-8) after stimulation with interleukin-8/CXCL8 (IL-8). Gelatinase B cleaves and potentiates IL-8, generating a positive feedback. Here, we extend these findings and compare the processing of the CXC chemokines human and mouse granulocyte chemotactic protein-2/CXCL6 (GCP-2) and the closely related human epithelial-cell derived neutrophil activating peptide-78/CXCL5 (ENA-78) with that of human IL-8. Human GCP-2 and ENA-78 are cleaved by gelatinase B at similar rates to IL-8. In addition, GCP-2 is cleaved by neutrophil collagenase, but at a lower rate. The cleavage of GCP-2 is exclusively N-terminal and does not result in any change in biological activity. In contrast, ENA-78 is cleaved by gelatinase B at eight positions at various rates, finally generating inactive fragments. Physiologically, sequential cleavage of ENA-78 may result in early potentiation and later in inactivation of the chemokine. Remarkably, in the mouse, which lacks IL-8 which is replaced by GCP-2/LIX as the most potent neutrophil activating chemokine, N-terminal clipping and twofold potentiation by gelatinase B was also observed. In addition to the similarities in the potentiation of IL-8 in humans and GCP-2 in mice, the conversion of mouse GCP-2/LIX by mouse gelatinase B is the fastest for any combination of chemokines and MMPs so far reported. This rapid conversion was also performed by crude neutrophil granule secretion under physiological conditions, extending the relevance of this proteolytic cleavage to the in vivo situation.

Abbreviations

APMA

amino-paraphenyl mercuric acetate

CTAP-III

connective tissue activating peptide-III

ENA-78

epithelial cell-derived neutrophil activating peptide-78

GCP-2

granulocyte chemotactic protein-2

IL

interleukin

MCP-3

monocyte chemotactic protein-3

MMP

matrix metalloproteinase

MS

mass spectrometry

NAP-2

neutrophil activating peptide-2

PF-4

platelet factor-4

SDF-1

stromal-derived factor-1

TIMP

tissue inhibitor of metalloproteases

Chemokines and matrix metalloproteases (MMPs), in particular gelatinase B (MMP-9) and neutrophil collagenase (MMP-8), play key roles in the migration of immune cells to sites of inflammation. MMPs degrade basement membranes and extracellular matrix components and are therefore important effector molecules for cell migration. However, MMPs also have an important regulatory role [1], as they can regulate cytokine and chemokine activity by proteolytic processing [2–4]. Chemokines, which form a concentration gradient within tissues to attract leucocytes, can be subdivided into subgroups, depending on the position of the two most N-terminal cysteines in the sequence [5]. CC chemokines, in which the first two cysteines are adjacent, are active on mononuclear cells, basophils and eosinophils. In contrast, the CXC chemokines have one amino acid between the first two cysteines and are active on neutrophils and T-lymphocytes. CXC chemokines, which contain the Glu-Leu-Arg (ELR) motif in front of the CXC sequence, are responsible for the fast chemoattraction of neutrophils to sites of inflammation [6]. Other effects of ELR-positive CXC chemokines include the promotion of angiogenesis [7] and mitogenic activity on various cell types [8,9]. The first discovered chemokine is interleukin-8 (IL-8) [10]. In terms of abundancy, IL-8 is the major ELR-positive CXC chemokine in humans with high chemoattractive potency. In the mouse, the counterpart of IL-8 in humans remains elusive.

Other ELR-containing CXC chemokines in humans are granulocyte chemotactic protein-2 (GCP-2), epithelial-cell-derived neutrophil attractant-78 (ENA-78), GRO-α, GRO-β and GRO-γ and connective tissue-activating peptide-III (CTAP-III), which is an inactive precursor of neutrophil-activating peptide-2 (NAP-2). In the mouse, the only reputed counterpart for the two related chemokines GCP-2 and ENA-78 is named mouse GCP-2/LIX [11–13]. Mouse GCP-2/LIX is believed to have the same roles as IL-8 in the human system. The CXC chemokines without the ELR motif do not stimulate neutrophils, but rather attract lymphocytes [14], and, in contrast with the ELR-positive CXC chemokines, have angiostatic activity [15].

The main receptors for ELR-containing chemokines are CXCR-1 and CXCR-2. IL-8 and GCP-2 bind to both CXCR-1 and CXCR-2, while ENA-78 and NAP-2 bind only to CXCR-2 with high affinity [16–18]. Binding to the receptor activates signal transduction mechanisms, including an increase in intracellular Ca²⁺ concentration, that can produce diverse effects. These include the migration of the neutrophils towards higher chemokine concentrations and the release of the content of their granules containing gelatinase B [19]. In addition, the respiratory burst [20] and the expression of activated adhesion molecules is initiated [21,22].

Degranulation of neutrophils under the influence of chemokines leads to the release of two MMPs, neutrophil collagenase (MMP-8) and gelatinase B (MMP-9). After activation, e.g. by reactive oxygen species produced by the neutrophil [23] or by stromelysin-1 produced by the surrounding tissues [24,25], these two proteases degrade the extracellular matrix and allow the neutrophil to migrate through the tissues. Indeed, gelatinase B has been shown to be an essential enzyme for the migration of various cell types, including metastasizing cancer cells [26], Langerhans cells [27], megakaryocytes [28], and also neutrophils [29]. Because inhibition of the enzyme might diminish inflammation and because excessive gelatinase B activity leads to tissue destruction and pathology, gelatinase B is an attractive target for therapeutic drugs in various diseases [30].

Recently, we have shown that gelatinase B processes chemokines, leading to, for example, the potentiation of IL-8 and the degradation of CTAP-III, GRO-α and PF-4 [3]. This revealed an important positive feedback loop between gelatinase B and IL-8, indicating that gelatinase B is not only an effector but also a regulatory enzyme. Furthermore, another similar positive feedback has been shown between endothelin-1 and gelatinase B [31]. Here we extend these findings by demonstrating the processing of GCP-2, ENA-78 and mouse GCP-2/LIX by gelatinase B and neutrophil collagenase, by comparison of the cleavage efficiencies and by focus on the two major neutrophil MMPs, gelatinase B and neutrophil collagenase. From this, we can report that the cleavage of mouse GCP-2/LIX by gelatinase B is the most efficient of all chemokine–MMP pairs tested so far. Furthermore, this cleavage was also detected with crude neutrophil secretions.

Materials and methods

Chemokines and MMPs

Natural gelatinase B from human neutrophils was purified to homogeneity and activated with 1 : 100 stromelysin-1 as described [3]. Recombinant human neutrophil collagenase and recombinant mouse gelatinase B (R & D, Abingdon, Oxfordshire, UK) were activated during 1 or 2 h, respectively, with 1 mm para-aminophenyl mercuric acetate (APMA) at 37 °C and were subsequently dialyzed against assay buffer (100 mm Tris/HCl, pH 7.5, 100 mm NaCl, 10 mm CaCl₂, 0.01% Tween 20).

Recombinant human ENA-78 was purchased from R & D and further purified by RP-HPLC. Recombinant human GCP-2 and mouse GCP-2(1–79) were produced in the periplasm of Escherichia coli as described for human MCP-2 [32]. Proteins from the periplasm were loaded on a heparin/Sepharose affinity column in 50 mm Tris/HCl, pH 7.4, and eluted in an NaCl gradient (50 mm to 2 m NaCl). GCP-2-containing fractions (determined by ELISA) were dialyzed against 50 mm formic acid pH 4.0, loaded on a 1-mL Mono S cation-exchange column (Amersham Pharmacia Biotech) and eluted with an NaCl gradient (0–1 m). Contaminating proteins were further removed by C-8 RP-HPLC on an Aquapore RP-300 column (4.6 × 220 mm; Perkin–Elmer) and the average relative molecular mass of the proteins was verified by electrospray ion trap MS (Esquire-LC; Bruker Daltonics, Bremen, Germany). As the mouse GCP-2-containing fractions were still contaminated with other proteins, mouse GCP-2 was purified by Mono S cation-exchange chromatography in 50 mm malonic acid, pH 6.4, and eluted with a 0–1 m NaCl gradient. Salts were removed from the cation-exchange fractions by C-8 RP-HPLC on a 2.1 × 220 mm Aquapore RP-300 column.

Digestion of chemokines with gelatinase B or neutrophil collagenase

Human GCP-2(1–77) (4 µm) and human ENA(1–78) (2 µm) were digested under similar conditions to those for IL-8 [3] with activated gelatinase B, purified from human neutrophils (0.4 µm) in assay buffer at 37 °C for the indicated times. Control digestions of these chemokines were performed without gelatinase B but with 0.004 µm stromelysin-1 (used to activate the progelatinase B). Human GCP-2 (4 µm) and human ENA-78 (4 µm) were digested with APMA-activated neutrophil collagenase (0.4 µm) under the same conditions, with only assay buffer added to the control digestions. Mouse GCP-2(1–79) (4 µm) was digested with APMA-activated mouse gelatinase B (20 nm) under the same conditions. Inhibition experiments were performed under identical conditions with the addition of the following inhibitors: 20 mm EDTA, 7 mmo-phenanthroline, 1.2 µm TIMP-1, 2 µg·mL⁻¹ E64, 50 µg·mL⁻¹ leupeptin, 50 mm benzamidine or 2 mm pefabloc. The resulting cleavage products were analyzed by Tris-tricine SDS/PAGE or, after being desalted using a C18 ZIPTIP (Millipore), subjected to MS analysis on an Esquire-LC ion trap apparatus (Bruker). For further identification and sequencing of chemokine fragments, tandem MS/MS was used on quadrupole time-of-flight apparatus (QTOF-II; Micromass, Manchester, UK). Edman degradation was performed on a Procise 491 cLC protein sequencer (Applied Biosystems, Foster City, CA, USA).

Determination of k_cat/K_m

Chemokines were digested with natural human gelatinase B (0.4 µm) or recombinant mouse gelatinase B (10 nm) in assay buffer without Tween 20 at four different chemokine concentrations varying from 1 to 6 µm. Samples were collected at various time intervals, desalted with the use of C18 ZIPTIPs, and analyzed by ion trap MS. Formation of the products was evaluated by comparison of the relative intensity of the product peaks with the substrate peaks after charge deconvolution of the mass spectrum. The velocity of each reaction was determined using at least four different time points before 25% of the substrate was consumed. k_cat/K_m could be determined by linear plotting of the velocity compared with the substrate concentration, and the separate k_cat and K_m constants were determined on a Lineweaver–Burk plot.

Detection of intracellular Ca²⁺ concentrations

The concentration of intracellular Ca²⁺ ([Ca²⁺]_i) was measured as described previously [33,34]. Briefly, purified human granulocytes (10⁷·mL⁻¹) were loaded with the fluorescent indicator fura-2 (2.5 µm fura-2/AM; Molecular Probes Europe BV, Leiden, the Netherlands) for 30 min at 37 °C. After two washes, cells were stored on ice at 10⁶ cells·mL⁻¹ for a maximum of 1.5 h. After excitation at 340 and 380 nm, fura-2 fluorescence was detected at 510 nm at 37 °C in an LS50B luminescence spectrophotometer (Perkin-Elmer) and used to calculate [Ca²⁺]_i.

Conversion of mouse GCP-2(1–79) by neutrophil granule secretion

Neutrophils were isolated from human blood, resuspended in degranulation buffer (20 mm Tris/HCl, pH 7.4, 113 mm NaCl, 10 mm CaCl₂) at 10⁷ cells·mL⁻¹ and stimulated to degranulate with 0.5 µm fMLP at 37 °C for 20 min. Subsequently, the cells were removed by centrifugation. Where indicated, 0.58 µg·mL⁻¹ stromelysin-1 was added to the granule secretagogue and incubated for 3 h. Mouse GCP-2(1–79) was incubated at a concentration of 2 µm with 10-fold diluted granule secretion in assay buffer at 37 °C for 1 h. As a control, mouse GCP-2 was incubated under identical conditions with the corresponding concentration of stromelysin-1 without neutrophil granule secretion. Inhibition experiments were performed under identical conditions with the addition of 20 mm EDTA or 2 mm pefabloc. The resulting products were analyzed by MS after being desalted as described above.

Results

Processing of chemokines by gelatinase B and neutrophil collagenase

Gelatinase B has been found to process the CXC chemokines IL-8, CTAP-III, GRO-α, PF-4 [3] and SDF-1 [35]. To complement and compare the processing of other chemokines by gelatinase B, human GCP-2 was incubated with natural gelatinase B from human neutrophils at an enzyme to substrate ratio of 1 : 10. SDS/PAGE analysis showed that gelatinase B processes GCP-2 (Fig. 1A). The digestion could be inhibited by the metalloproteinase inhibitors EDTA, o-phenanthroline and TIMP-1 but not by thiol or serine protease inhibitors (E64, benzamidine, leupeptin). MS analysis of the cleavage products revealed two alternative cleavage sites, behind residue 4 or 5. Cleavage thus generates GCP-2(5–77) and GCP-2(6–77) (Fig. 2). A trace of GCP-2(7–77) was also detected after digestion with gelatinase B. The relative amounts of the different forms were 78% for GCP-2(6–77), 19% for GCP-2(5–77), and 3% for GCP-2(7–77). These were not modified by prolonged incubation (data not shown), in line with the fact that the gelatinase B used was pure with no exopeptidase activity. GCP-2(6–77) has been isolated previously from a natural source, i.e. cytokine-induced sarcoma cells [11]. Incubation of human GCP-2 with neutrophil collagenase also results in N-terminal cleavage. This cleavage can be inhibited by EDTA, o-phenanthroline and TIMP-1 but not by the thiol or serine protease inhibitors E64 or pefabloc (data not shown). MS analysis indicated that neutrophil collagenase generates GCP-2(6–77) (55%) and GCP-2(7–77) (45%), and that, after 24 h, only half of the substrate is cleaved (Fig. 3). GCP-2(5–77) was not detected after prolonged incubation of intact GCP-2 with neutrophil collagenase.

Figure 1

Processing of GCP-2 by activated gelatinase B. (A) Purified recombinant human GCP-2(1–77) was incubated with stromelysin-1-activated gelatinase B from human neutrophils (+) or with stromelysin-1 alone (–) for 16 h at 37 °C and subsequently analyzed by SDS/PAGE and silver staining. The metalloproteinase inhibitors EDTA, o-phenanthroline (PHEN) and TIMP-1 and the thiol protease inhibitor E64 and serine protease inhibitors benzamidine (Benz) and leupeptin (Leu) were used to control the specificity of the reaction. (B) Purified recombinant mouse GCP-2(1–79) was incubated with APMA-activated mouse gelatinase B (+) or without gelatinase B (–) for 6 h at 37 °C. The indicated protease inhibitors were used to confirm the specificity of the cleavage.

Figure 2

MS analysis of human GCP-2 after cleavage by gelatinase B. Human recombinant GCP-2 was analyzed by electrospray ion trap MS before (A) and after (B) incubation with activated natural gelatinase B from human neutrophils. The unprocessed (m/Z) and charge-deconvoluted (M) spectra are shown. The theoretical masses of GCP-2(1–77), GCP-2(5–77), GCP-2(6–77) and GCP-2(7–77) are 8311.9, 7971.55, 7900.5 and 7801.3 Da, respectively.

Figure 3

MS analysis of human GCP-2 after cleavage by neutrophil collagenase. Human recombinant GCP-2 was analyzed by electrospray ion trap MS before (A) and after (B) incubation with APMA-activated neutrophil collagenase. The unprocessed (m/Z) and charge-deconvoluted (M) spectra are shown. The theoretical masses of GCP-2(1–77), GCP-2(6–77) and GCP-2(7–77) are 8311.9, 7900.5 and 7801.3 Da, respectively.

The closest human relative of human GCP-2 is ENA-78. As shown in Fig. 4A, ENA-78 is also processed by gelatinase B, and this cleavage is also inhibitable by metalloproteinase inhibitors but not by thiol or serine protease inhibitors. As shown by SDS/PAGE analysis of samples taken at various incubation times, digestion by gelatinase B results first in the formation of shorter forms of ENA-78, and thereafter ENA-78 is completely degraded into fragments (Fig. 4B). By MS analysis, the intermediate shorter forms were determined to be ENA(6–78) (relative amount 46%), ENA(7–78) (relative amount 36%) and ENA(8–78) (relative amount 18%) (Fig. 5). The final degradation products were also identified using MS/MS on a quadrupole time-of-flight mass spectrometer (Table 1). ENA-78 and IL-8 are not processed by neutrophil collagenase (data not shown).

Figure 4

Cleavage of ENA-78 by gelatinase B. (A) Recombinant human ENA-78 was incubated with stromelysin-1-activated gelatinase B from human neutrophils (+) or with stromelysin-1 alone (–) during 24 h at 37 °C and subsequently analyzed by SDS/PAGE and silver staining. The metalloproteinase inhibitors EDTA, o-phenanthroline (PHEN) and TIMP-1 and the thiol protease inhibitor E64 and serine protease inhibitor leupeptin (data not shown) were added to control the specificity of the reaction. (B) Recombinant human ENA-78 was incubated at 37 °C with stromelysin-1-activated gelatinase B from human neutrophils (+) or with stromelysin-1 alone (–). Samples were taken at different time intervals (indicated at the top in hours) and analyzed by SDS/PAGE and silver staining.

Figure 5

MS analysis of ENA-78 before and after cleavage by gelatinase B. Human recombinant ENA-78 was analyzed by electrospray ion trap MS before (A) and after (B) incubation with activated natural gelatinase B from human neutrophils for 4 h at 37 °C. The unprocessed (m/Z) and charge-deconvoluted (M) spectra are shown. The theoretical masses of ENA(1–78), ENA(6–78), ENA(7–78) and ENA(8–78) are 8352.9, 7985.5, 7914.4 and 7815.3 Da, respectively.

Table 1. Determination of late cleavage sites of gelatinase B in ENA-78. The sequence of ENA-78 is AGPAA *A *V *LRE°LRCVCLQTTQGVHPKMISN°LQVFAIGPQC°SKVEVVAS°LKNGKEICLDPEAPFLKKVIQKILDGGNKEN, where fast cleavages as shown in Fig. 5 are indicated with *, whereas ° indicates slow cleavages (after 24 h incubation).

Mass (Da)^a	Theoretical mass^a	Fragment^b
1812.99	1812.92	11 LR C V C LQTTQGVHP KM 26
1873.98	1874.00	30 LQVFAIGP QCSKVEVVAS 47
1074.56	1074.55	30 LQVFAIGP QC 39
3450.66	3450.89	48 LKNGKEI CLD…. GGNKEN 78

• a Monoisotopic masses;
• ^b sequence confirmed by tandem MS/MS; amino acids indicated in bold were additionally confirmed by Edman degradation, and the numbering in subscript indicates the location of the first and last residues in the mature protein.

IL-8 does not exist in the mouse, and only one homologue of human GCP-2 and human ENA-78 has been identified and named mouse GCP-2/LIX [36]. Using the same methods as for human GCP-2 and human ENA-78, we found that mouse GCP-2(1–79) is also processed by mouse gelatinase B to GCP-2(5–79) (1, 6). Interestingly, this cleavage was by far the most efficient, occurring at an enzyme to substrate ratio of 1 : 200. In analogy with human gelatinase B cleaving human IL-8 in only one place, mouse GCP-2 was also cut by mouse gelatinase B at a unique site. Human gelatinase B was able to process mouse GCP-2 at the same site and with a similar efficiency. On prolonged incubation with an enzyme to substrate ratio of 1 : 10, the mouse chemokine was further degraded by human gelatinase B into smaller fragments (data not shown).

Figure 6

MS analysis of mouse GCP-2 cleaved by mouse gelatinase B. Mouse GCP-2 was analyzed by electrospray ion trap MS before (A) and after (B) incubation with APMA-activated gelatinase B for 3.5 h at 37 °C. The unprocessed (m/Z) and charge-deconvoluted (M) spectra are shown. The theoretical masses of mouse GCP-2(1–79) and mouse GCP-2(5–79) are 8452.2 and 8109.9 Da, respectively.

Determination of k_cat/K_m

The best way to characterize the velocity of an enzyme-catalyzed reaction is by determining the Michaelis–Menten constants k_cat/K_m. The k_cat/K_m values of the cleavage of human GCP-2 and ENA-78 by activated human gelatinase B and mouse GCP-2 by activated mouse gelatinase B were determined by measurement of the cleavage rate at chemokine concentrations varying between 1 and 6 µm before 25% of the substrate was consumed. For each chemokine concentration, four samples were taken at different time intervals and analyzed by MS. The ratio between the relative signal intensity of each form of the chemokine was used to determine the conversion, and the conversion rate was calculated from a linear plot of product versus time (the correlation coefficient r² was always 0.98 or higher). The k_cat/K_m was calculated from the slope of the plot of conversion rate versus substrate concentration (Fig. 7, Table 2). This plot was linear, indicating that the K_m is significantly higher than the highest substrate concentration used (6 µm), and therefore the k_cat and K_m values could not be determined separately. For comparison, the k_cat/K_m of the previously described cleavage of IL-8 by gelatinase B [3] was determined in a similar way. Clearly, mouse GCP-2 is the most efficiently processed chemokine by gelatinase B, at a cleavage rate slightly higher than that of MCP-3 by gelatinase A [4], whereas the rates of cleavage of IL-8, GCP-2 and ENA-78 by gelatinase B are considerably lower. Nevertheless, cleavage of human IL-8, GCP-2 and ENA-78 is believed to be physiologically relevant, because in biological samples the gelatinase B concentration is often higher than the chemokine concentration.

Figure 7

Determination of k_cat/K_m for the cleavage of IL-8, GCP-2, ENA-78 and mouse GCP-2 by gelatinase B. The chemokines IL-8 (◊), GCP-2 (•), ENA-78 (▴) and mouse GCP-2 (▪) were incubated at the indicated concentrations with activated gelatinase B. At various time intervals, before conversion of 25% of the substrate, samples were taken and analyzed by MS to determine the cleavage rate. Quantification was by determination of the relative abundance of the products versus the substrate on the mass spectra. (A) Comparison of the cleavage of IL-8 by human gelatinase B and of mouse GCP-2 by the mouse enzyme. (B) Comparison of the velocities of the processing of the human chemokines IL-8, GCP-2 and ENA-78. Notice that the scales on the y axes are different.

Table 2. Kinetics of the cleavage of chemokines by gelatinase B and neutrophil collagenase. NI, Not indicated.

Chemokine	Enzyme	Products	Relative product amount ^a	k cat/Km (m^−1·s^−1) ^b	r ^2 ^c
Mouse GCP-2(1–79)	Mouse gelatinase B	mGCP-2(5–79)	100%	11667	0.984
Human GCP-2(1–77)	Human gelatinase B	GCP-2(6–77), GCP-2(5–77),	78%, 19%, 3%
		GCP-2(7–77)		163	0.988
Human GCP-2(1–77)	Human neutrophil collagenase	GCP-2(6–77), GCP-2(7–77)	55%, 45%	< 100	NI
Human ENA(1–78)	Human gelatinase B	ENA(6–78), ENA(7–78), ENA(8–78)	46%, 36%, 18%	350	0.997
		Further cleavage behind residues 10, 26, 29, 39 and 47
Human ENA(1–78)	Human neutrophil collagenase	No cleavage	–	0	–
Human IL-8(1–77)	Human gelatinase B	IL-8(7–77)	100%	233	0.994
Human IL-8(1–77)	Human neutrophil collagenase	No cleavage	–	0	–
Human MCP-3(1–76)^d	Human gelatinase A	MCP-3(5–76)	100%	8000	NI

• ^a Relative amounts of truncated chemokine forms were derived from the relative intensity of the corresponding peaks on the mass spectra; ^b Calculated from the slopes in Fig. 7; ^c Correlation coefficients of the linear regression analysis, according to Fig. 7; ^d For comparison, the cleavage of MCP-3 by gelatinase A was determined by McQuibban et al. [4].

Effect of processing by gelatinase B on the biological activity of human GCP-2 and ENA-78 and mouse GCP-2

Recently, we described the unique 10–30-fold potentiation of IL-8 by N-terminal processing by gelatinase B [3]. The processing of human GCP-2(1–77) into GCP-2(5,6,7–77) by gelatinase B did not influence its biological activity, as analyzed by measurement of the increase in [Ca²⁺]_i (data not shown). This observation confirmed previous results [11].

Different N-terminally truncated forms of ENA-78 have previously been extensively compared. The data indicated that shorter forms are threefold more potent than intact ENA-78 [34,37]. As the processing of ENA-78 by gelatinase B consists first of N-terminal truncation followed by degradation, it is expected to result in a transient increase in activity of the chemokine, followed by inactivation. Under the conditions used, however, the potentiation was mainly masked by the degradation (data not shown).

The removal of four N-terminal residues of mouse GCP-2(1–79) by mouse gelatinase B resulted in a twofold potentiation (P < 0.05, n = 3) (Fig. 8). Our biochemical analysis is in line with previous results with natural isoforms of mouse GCP-2/LIX [38]. In the latter study it was also found that progressive truncation results in increased biological activities.

Figure 8

[Ca²⁺]_i-mobilizing activity of mouse GCP-2(1–79) and mouse GCP-2(5–79). The biological activity of mouse GCP-2(1–79) (white bars) and mouse GCP-2(5–79) (black bars) were compared by measuring the ability to induce increases in [Ca²⁺]_i in human neutrophils. After purification, the neutrophils were loaded with the fluorescent dye Fura-2 and stimulated with various concentrations of mouse GCP-2. The increase in [Ca²⁺]_i was monitored by measuring the fluorescence of free and Ca²⁺-bound Fura-2. Significant differences are indicated with * (P = 0.05, n = 3) or ** (P = 0.02, n = 3). With 2 nm mouse GCP-2, no increase in [Ca²⁺]_i was observed, and the detection limit is indicated with a dotted line.

Processing of mouse GCP-2(1–79) by neutrophil granule secretion

To determine whether the chemokine conversions by neutrophil collagenase and gelatinase B also occur under physiological conditions, mouse GCP-2(1–79) was incubated with neutrophil granule secretion at 37 °C for various times. This did not result in processing of mouse GCP-2(1–79) (data not shown), except for a slow conversion into mouse GCP-2(7–79). The latter could be inhibited with pefabloc, showing that a serine protease is responsible. As gelatinase B and neutrophil collagenase are secreted as proenzymes, it was hypothesized that the MMPs have to be activated before being able to convert chemokines. Under physiological and pathological conditions, e.g. inflammation, considerable amounts of stromelysin-1 may be produced by surrounding cells, and this will efficiently activate gelatinase B [24,25]. Therefore, the neutrophil granule secretion was first incubated with 10 nm stromelysin-1, resulting in activation of gelatinase B, as verified by zymography analysis. Subsequently, mouse GCP-2(1–79) was incubated with the activated granule secretion and analyzed by MS, showing clearly the conversion of mouse GCP-2(1–79) into mouse GCP-2(5–79) (Fig. 9). This rapid conversion was not obtained by incubation with stromelysin-1 alone and was inhibited by EDTA and not by pefabloc (data not shown), confirming that it was due to the activity of the neutrophil MMPs, in particular gelatinase B.

Figure 9

Conversion of mouse GCP-2(1–79) by neutrophil granule secretion. Mouse GCP-2(1–79) was analyzed by electrospray ion trap MS after incubation with neutrophil granule secretion for 1 h at 37 °C. In (A), mouse GCP-2(1–79) was incubated with neutrophil granule secretion containing progelatinase B, and in (B) mouse GCP-2(1–79) was incubated with neutrophil granule secretion in which gelatinase B was first activated by incubation with stromelysin-1. The unprocessed (m/Z) and charge-deconvoluted (M) spectra are shown. The theoretical masses of mouse GCP-2(1–79) and mouse GCP-2(5–79) are 8452.2 and 8109.9 Da, respectively. The small peak at M = 7898.5 corresponds to mouse GCP-2(7–79), which was already present in low amounts in the mouse GCP-2 sample before the incubation (data not shown) and which increased slightly during the incubation.

Discussion

Neutrophils are first-line defence cells of the innate immune system and are equipped with a battery of effector molecules for the destruction of bacteria and other invading micro-organisms. In addition, these cells can respond extremely rapidly (within minutes) to signals such as chemotactic gradients generated by ELR-positive CXC chemokines. The neutrophil MMPs, gelatinase B and neutrophil collagenase, contribute largely to this fast response, as they are prepacked in the granules and help the neutrophil to migrate through basement membranes and connective tissues. We have shown previously that gelatinase B processes the most potent human neutrophil chemokine, IL-8, into a 10–30-fold more active chemokine. This results in an important positive feedback loop, as IL-8 induces the rapid release of gelatinase B from the granules [3]. The CXC chemokines CTAP-III, GRO-α and PF-4 are degraded by gelatinase B [3]. Gelatinase A and other MMPs have been shown to process MCPs and SDF-1 N-terminally to inactive forms [4,35,39].

These findings are further extended and compared here by the discovery of novel chemokine–MMP interactions: the processing of the human CXC chemokines GCP-2 and ENA-78 by human gelatinase B, of human GCP-2 by neutrophil collagenase, and of the single mouse counterpart of these chemokines, named mouse GCP-2/LIX, by mouse gelatinase B. Gelatinase B removes four to six N-terminal residues from human GCP-2, and a slower cleavage by neutrophil collagenase was observed, resulting in the removal of five or six N-terminal residues. The activity of human GCP-2 remains unchanged after these cleavages. In contrast, gelatinase B first processes ENA-78, the closest homologue of GCP-2, by the removal of five to seven N-terminal residues, and prolonged incubation results in complete degradation. Previous studies [34,37] have amply shown that N-terminally processed forms of ENA-78 are 3–8-fold more active than the full length form, confirming that a transient positive feedback loop exists between gelatinase B and ENA-78, before ENA-78 activity is down-regulated by degradation. No processing of ENA-78 by neutrophil collagenase was observed.

In the mouse, no close homologue of IL-8 exists, but its role is thought to be assumed by mouse GCP-2/LIX, which is the closest mouse homologue of both human GCP-2 and human ENA-78. Similar to human IL-8, mouse GCP-2/LIX is the most potent mouse CXC chemokine. It has been shown to activate both IL-8 receptors, CXCR-1 and CXCR-2. The cleavage of mouse GCP-2 by gelatinase B is highly efficient (k_cat/K_m = 11667 m⁻¹·s⁻¹, which is so far the highest value for any chemokine–MMP pair) and also results in potentiation of its biological activity, although to a lesser extent than with human IL-8. However, isolation and comparison of natural isoforms shows that further progressive truncation by other, as yet unknown, proteases takes place and leads to an up to 30-fold potentiation [38], which is similar to the potentiation of IL-8 in man. Here we show that incubation of mouse GCP-2 with neutrophil granule secretion results in the same truncation as with purified gelatinase B, if the gelatinase B in the secretion is activated. This activation is performed by, e.g. stromelysin-1, as has been shown in vitro and in vivo[24,25].

Other proteases have been shown to process chemokines. For instance, CXC chemokines have been shown to be processed by the neutrophil proteases proteinase-3, elastase and cathepsin G [37,40]. However, these proteases are not rapidly released from neutrophils upon stimulation with chemokines, unless synthetic cytochalasin B is present [41]. The need for the cytochalasin stimulus makes the physiological consequences of these cleavages as yet less clear. The serine protease dipeptidyl peptidase IV/CD26 removes two to four N-terminal residues from several chemokines. The CC chemokines RANTES, MDC and eotaxin are inactivated or even converted into chemotaxis inhibitors by CD26, while LD78β is the only chemokine to be potentiated by CD26 [42–45]. The CXC chemokines, without the ELR motif, SDF-1α, IP-10, Mig and I-TAC are also rapidly inactivated by CD26 [46].

In conclusion, gelatinase B is an important protease for the processing of ELR-positive CXC chemokines. It is able to potentiate the most active CXC chemokines in man (IL-8) and mouse (GCP-2/LIX), whereas other CXC chemokines are functionally unaffected by clipping (human GCP-2) or are degraded (e.g. ENA-78) by gelatinase B. Neutrophil collagenase, the other secreted neutrophil MMP, also plays a role in the processing of human GCP-2. Typical examples where these feedback loops may occur in vivo are bacterial pyogenic infections, in which neutrophils are massively attracted and stimulated to degranulate gelatinase B and neutrophil collagenase under the pressure of the ELR-positive chemokines [47,48]. Also, in rheumatoid arthritis, high levels of gelatinase B activity are found in the synovial fluid together with IL-8 and ENA-78 [30,49]. Another process in which both gelatinase B and chemokines have been implicated is angiogenesis, in which gelatinase B seems to trigger an angiogenic switch [50], whereas the ELR-positive chemokines have clear angiogenic activity [51–53]. Tumors expressing ELR-positive chemokines may also gain advantage, not only by promoting angiogenesis, but also by attracting neutrophils, which are then stimulated to degranulate and release gelatinase B. The neutrophil gelatinase B is then used by the tumor cells to promote angiogenesis and also to degrade extracellular matrix components, thereby allowing migration of the tumor cells to the blood vessels [54–57]. In line with this countercurrent model [54], it was recently shown that GCP-2 expression in vivo favors tumor growth by angiogenesis [56].

Acknowledgements

We thank René Conings, Jean-Pierre Lenaerts and Roos Cruysberghs for technical assistance and Dr Annemie Lambeir (University of Antwerp) for helpful discussions. We also thank the F.W.O.-Vlaanderen particularly for funding two mass spectrometers. This work was supported by the Geconcerteerde OnderzoeksActies 2002-06, the Cancer Reseach Fund of Fortis AB, the Belgian Federation against Cancer, and the National Fund for Scientific Research (F.W.O.-Vlaanderen). A.W. and P.P. are postdoctoral fellows of the F.W.O.-Vlaanderen.