Biological Activity of CXCL8 Forms Generated by Alternative Cleavage of the Signal Peptide or by Aminopeptidase-Mediated Truncation

Background Posttranslational modification of chemokines is one of the mechanisms that regulate leukocyte migration during inflammation. Multiple natural NH2-terminally truncated forms of the major human neutrophil attractant interleukin-8 or CXCL8 have been identified. Although differential activity was reported for some CXCL8 forms, no biological data are available for others. Methodology/Principal Findings Aminopeptidase-cleaved CXCL8(2-77) and CXCL8(3-77), the product of alternative cleavage of the signal peptide CXCL8(-2-77) and the previously studied forms containing 77 and 72 amino acids, CXCL8(1-77) and CXCL8(6-77), were prepared by solid-phase peptide synthesis, purified and folded into active proteins. No differences in binding and calcium signaling potency were detected between CXCL8(1-77), CXCL8(-2-77), CXCL8(2-77) and CXCL8(3-77) on cells transfected with one of the human CXCL8 receptors, i.e. CXCR1 and CXCR2. However, CXCL8(-2-77) was more potent compared to CXCL8(1-77), CXCL8(2-77) and CXCL8(3-77) in signaling and in vitro chemotaxis of peripheral blood-derived human neutrophils. Moreover, CXCL8(-2-77) was less efficiently processed by plasmin into the more potent CXCL8(6-77). The truncated forms CXCL8(2-77) and CXCL8(3-77) had higher affinity for heparin than CXCL8(1-77), a property important for the presentation of CXCL8 on endothelial layers. Upon intraperitoneal injection in mice, elongated, truncated and intact CXCL8 were equally potent to recruit neutrophils to the peritoneal cavity. Conclusions In terms of their ability to induce neutrophil recruitment in vivo, the multiple CXCL8 forms may be divided in three groups. The first group includes CXCL8 proteins consisting of 75 to 79 amino acids, cleaved by aminopeptidases, with intermediate activity on neutrophils. The second group, generated through proteolytic cleavage (e.g. by Ser proteases), contains 69 to 72 amino acid forms which are highly potent neutrophil attractants in vivo. A third category is generated through the modification of the arginine in the NH2-terminal region into citrulline by peptidylarginine deiminases and has weak potency to induce neutrophil extravasation.


Introduction
During infection, neutrophilic granulocytes are often the first immune cells to invade the inflammatory site. In addition to bacterial formyl peptides, complement fragment C5a and the lipid leukotriene B4, a number of chemokines with a conserved Glu-Leu-Arg (ELR) motif in front of the first cysteine (ELR + CXC chemokines) attract neutrophils [1,2,3]. All human ELR + CXC chemokines, i.e. CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, CXCL7 and CXCL8 attract and activate neutrophils through the specific G protein-coupled CXC chemokine receptor CXCR2. In addition, CXCL6 and CXCL8 [or interleukin (IL)-8] also interact with CXCR1 [4]. Upon stimulation with bacterial or viral agents or inflammatory cytokines such as IL-1b or tumor-necrosis factor-a (TNF-a) CXCL8 can be produced by virtually any cell type including leukocytes, fibroblasts and epithelial and endothelial cells [5]. CXCL8 is one of the most abundant, if not the most abundant, ELR + CXC chemokines with the highest specific biological activity on neutrophils. CXCL8 may be transported across endothelial layers by the Duffy antigen receptor for chemokines (DARC) and presented to neutrophils in the blood flow on top of the glycocalyx through its interaction with glycosaminoglycans [1,6].
To eradicate microbial infection, neutrophils use a number of innate immune mechanisms such as phagocytosis of bacteria and production of intracellular reactive oxygen components and hydrolyzing enzymes such as proteases that may be released in the extracellular milieu upon activation [7]. Recently, the formation of neutrophil extracellular traps (NETs), a mechanism that allows neutrophils to retain antimicrobial activity after death, has been identified as an additional mechanism and alternative for death by necrosis or apoptosis [8,9]. A side effect of these efficient antimicrobial mechanisms and release of highly active molecules and hydrolyzing enzymes is partial tissue destruction. To provide the rapid and adequate immune response that is restricted in time to the duration of the infection and to avoid chronic inflammation, precise control of local neutrophil accumulation and activation is essential. One of the mechanisms that regulate neutrophil recruitment and activation is enzyme-induced posttranslational modification of ELR + CXC chemokines [10]. In fact, the ELR + CXC chemokine CXCL7 only becomes activated upon proteolytic removal of a large part of the NH 2 -terminal region [11,12,13]. For other ELR + CXC chemokines the activity has been reported to be up-regulated upon limited truncation of the NH 2terminus by specific enzymes such as plasmin, thrombin, matrix metalloproteases, etc. [10]. However, further truncation within the ELR motif results in almost complete inactivation of all ELR + CXC chemokines. Numerous posttranslationally modified natural forms of the different ELR + CXC chemokines have been identified and partially characterized [10]. Leukocyte-derived conditioned medium contains at least 10 different truncated and also citrullinated forms of the most potent human ELR + CXC chemokine, CXCL8 [14]. Incubation of CXCL8 with the myeloid aminopeptidase CD13 results in the removal of one or two amino acids from the 77 amino acid CXCL8(1-77) [15]. The Arg in position 5 is a crucial amino acid for cleavage of CXCL8 into CXCL8(6-77) by the serine proteases plasmin and thrombin. CXCL8 truncated by five to eight NH 2 -terminal residues, becomes a three-to ten-fold more potent neutrophil attractant and angiogenic molecule in vitro and in vivo [14,16,17]. Citrullination of natural CXCL8 by peptidylarginine deiminase (PAD)-2 or PAD-4 also occurs specifically on Arg in position 5 [14]. Citrullination significantly reduces the capacity of CXCL8 to induce neutrophil extravasation without affecting its angiogenic activity. However, intravenously injected citrullinated CXCL8 is a more potent mobilizer of mature neutrophils to the blood stream [18]. In addition to proteolytic truncation and citrullination, alternative cleavage of the signal peptide results in a natural CXCL8 form with two extra NH 2 -terminal amino acids, i.e. CXCL8 containing 79 amino acids or CXCL8(-2-77) [11,14,19,20,21].
Significant amounts of natural elongated CXCL8(-2-77) and truncated CXCL8(2-77) and CXCL8  have been reported to be produced by lymphocytes, monocytes and fibroblasts [11,14,19,20,21]. Since these forms co-elute on chromatographic columns with other CXCL8 forms the individual forms were not readily available as pure proteins for the evaluation of their biological activity. Here we produced the different CXCL8 forms by Fmoc solid phase peptide synthesis. The biological activity of three CXCL8 forms for which the activity was not investigated before, i.e. CXCL8(-2-77), CXCL8(2-77) and CXCL8 , was compared with that of the previously studied CXCL8(1-77) and the more potent CXCL8(6-77) form in receptor binding, signaling and neutrophil migration assays.

Chemical synthesis of the naturally occurring CXCL8 variants
Separation of the natural CXCL8 variants CXCL8(-2-77), CXCL8(1-77), CXCL8(2-77), CXCL8(3-77) and CXCL8(6-77) via chromatography to study their biological activity was not feasible. These variants co-elute from analytical ion exchange and reversed-phase high performance liquid chromatography (RP-HPLC) columns [14,20]. Therefore, the individual proteins were prepared by solid phase peptide synthesis, deprotected, folded and purified to homogeneity. The accuracy of the relative molecular mass (M r ) of the synthesized proteins was verified by electrosprayion trap mass spectrometry as shown in Table 1. Edman degradation-based NH 2 -terminal amino acid sequencing confirmed the NH 2 -terminal protein sequence of the synthesized chemokines (data not shown).

Susceptibility of CXCL8 forms to cleavage by thrombin and plasmin
The plasma serine proteases plasmin and thrombin cleave CXCL8(1-77) in a very efficient way, thereby potentiating its neutrophil chemotactic activity [22,23]. Plasmin has a preference for the peptide bond between Arg 5 and Ser 6 , although minor cleavage of the Lys 8 -Glu 9 bond also occurs, whereas thrombin specifically cleaves the Arg 5 -Ser 6 bond of CXCL8. To study whether NH 2 -terminal truncation by one or two amino acids or the additional Glu-Gly dipeptide hinder or promote plasmin-or thrombin-mediated cleavage, processing of CXCL8(1-77), CXCL8(2-77), CXCL8(3-77) and CXCL8(-2-77) by these proteases was analyzed by mass spectrometry. All four substrates were converted with comparable efficiency by thrombin (data not shown), showing that limited variation of the length of the CXCL8 NH 2 -terminus did not affect the cleavage rate of thrombin. Similarly, the susceptibility of CXCL8 for cleavage by plasmin was not changed by limited NH 2 -terminal truncation. In contrast, plasmin cleaved CXCL8(-2-77) less efficiently compared to CXCL8(1-77) (Figure 4). Under the experimental conditions described in the Materials & Methods section, approximately 41% of CXCL8(-2-77) versus 79% of CXCL8(1-77) had undergone proteolysis by plasmin after 30 min of incubation. Complete conversion of CXCL8(1-77) was achieved upon 60 min of incubation, whereas at that moment more than 40% of CXCL8(-2-77) was still intact. This difference in susceptibility may influence the biological activity of CXCL8(-2-77) in vivo.

Neutrophil migration in vivo induced by CXCL8 forms
The in vivo neutrophil recruitment potency of the NH 2 -terminal CXCL8 variants was compared with that of CXCL8(1-77). Although no homologous murine CXCL8 equivalent exists, human CXCL8 activates murine neutrophils. Mice were intraperitoneally (i.p.) injected with the CXCL8 forms and after 2 h the peritoneal cavity was washed and the number of leukocytes and the percentage of neutrophils were determined. In this in vivo model, the neutrophil influx induced by CXCL8(1-77), CXCL8(2-77), CXCL8(3-77) and CXCL8(-2-77) was comparable ( Figure 5). Indeed, injection (i.p.) of 100 pmol of either of these isoforms elevated the number of neutrophils in the peritoneal cavity from about 2.10 4 neutrophils/ml (upon vehicle injection) to approximately 30.10 4 neutrophils/ml. As a positive control, CXCL8(6-77) was included in this assay. The higher potency of CXCL8(6-77) compared to CXCL8(1-77) could be confirmed, as the median number of neutrophils present in the intraperitoneal lavage upon injection of 100 pmol of CXCL8(6-77) is significantly higher than upon injection of 100 pmol of CXCL8(1-77) (70.10 4 neutrophils/ ml vs. 30.10 4 neutrophils/ml). Thus, no difference in neutrophil recruitment potency was detected between CXCL8(1-77), CXCL8(2-77), CXCL8(3-77) and CXCL8(-2-77) in this in vivo assay in mice, despite the small, but significant differences observed in calcium mobilization, chemotaxis and heparin binding assays and regarding the sensitivity to proteolytic activation by plasmin (vide supra).

Discussion
IL-8 or CXCL8 is the most potent human chemokine with neutrophil chemotactic properties. Already upon the initial isolation of natural human CXCL8 it was clear that several different NH 2 -terminal isoforms of CXCL8 are produced by leukocytes or fibroblasts under inflammatory conditions [11,19,20,21,22]. Posttranslational modifications can alter the biological activity (receptor affinity and signaling potency), interaction with other molecules (e.g. glycosaminoglycans), clearance of chemokines and hence the in vivo inflammatory properties [10]. The most abundant CXCL8 isoforms CXCL8(1-77) and CXCL8(6-77), produced by endothelial cells [24,25], fibroblasts [19] and monocytes [21,26] respectively, have been studied extensively. These studies revealed that the removal of five NH 2 -terminal amino acids (AVLPR) potentiates the neutrophil chemotactic activity of CXCL8 both in vitro and in vivo [14,22,23]. Additional limited NH 2 -terminal truncations to CXCL8(8-77) and CXCL8(9-77) gradually increase the chemotactic potency in vitro,  suggesting that CXCL8(6-77) is not the most active form of CXCL8 [16,27]. A crucial role in receptor binding and activation has been ascribed to the ELR motif in front of the CXC sequence using both chemically synthesized truncated and mutated analogs [16,28]. Thus far, truncation of the NH 2 -terminus of CXCL8 has been shown to result in potentiation as long as proteases do not cleave in and beyond the ELR motif.
Although not as explicit as the truncated isoforms, CXCL8(-2-77) also bound heparin with higher affinity than CXCL8(1-77). Conversion of CXCL8 into the more potent neutrophil chemoattractant CXCL8(6-77) by the serine proteases plasmin and thrombin was not affected by prior removal of the NH 2 -terminal amino acids Ala 1 or Val 2 . This is in contrast to CXCL8 citrullination which prevented subsequent CXCL8 activation by these serine proteases [14]. On the other hand, CXCL8(-2-77), generated through alternative cleavage of the signal peptide, was less susceptible to cleavage by plasmin. Finally, the neutrophilattracting activity of CXCL8(-2-77), CXCL8(1-77), CXCL8(2-77) and CXCL8(3-77) were evaluated in vivo upon i.p. injection in mice. During in vivo leukocyte migration, many more parameters come into play, including glycosaminoglycan binding and alteration of the expression pattern of adhesion molecules such as selectins and integrins. Despite the slightly increased in vitro chemotactic potency and GAG binding affinity, the in vivo neutrophil-attracting potency of CXCL8(-2-77) upon i.p. injection in mice did not differ from that of CXCL8(1-77). Injection of 100 pmol of CXCL8(1-77) or 100 pmol of CXCL8(-2-77) both elevated the percentage of neutrophils in the peritoneal cavity from about 2% to 18%. Perhaps, the higher variability of in vivo assays prohibits detection of small differences in biological activity. Alternatively, the reduced susceptibility of CXCL8(-2-77) to cleavage and activation by plasmin may counteract the increased chemotactic potency and GAG binding property of CXCL8(-2-77). Although alternative cleavage of the signal peptide of the CXCL8 precursor was reported in various publications, it does not seem to have an effect on the in vivo neutrophil-attracting activity of CXCL8 and thus, does not constitute a major regulatory mechanism. Furthermore, although NH 2 -terminal removal of amino acids one by one from CXCL8(6-77) was reported to improve the activity in vitro [16], aminopeptidase treatment of CXCL8(1-77) resulting in the generation of CXCL8(2-77) and CXCL8(3-77) did not progressively generate more potent isoforms in vivo compared to CXCL8(1-77). Combining these results with the results previously described for CXCL8(6-77) and [Cit 5 ]CXCL8(1-77) [14], we can conclude that, based on their potency to recruit neutrophils, the NH 2 -terminal variants of CXCL8 can be divided into 3 subgroups (Figure 6). A first group contains CXCL8(-2-77), CXCL8(1-77), CXCL8(2-77) and CXCL8(3-77), which display intermediate neutrophil-attracting capacity. Upon removal of 5 to 8 NH 2 -terminal amino acids, CXCL8 more efficiently attracts neutrophils, therefore these isoforms [i.e. CXCL8(6-77), CXCL8(7-77), CXCL8(8-77) and CXCL8(9-77)] form a second group with enhanced biological activity. Citrullinated CXCL8 belongs to a third category as it displays no or low neutrophil-attracting potency. The significant differences in terms of their ability to recruit neutrophils, point towards the importance of differential detection of these individual isoforms in patient samples and towards a potentially important role for CXCL8-modifying enzymes such as plasmin, thrombin and PADs in fine-tuning neutrophil migration under pathological conditions. Quantification of the individual CXCL8 forms in patient samples will help to unveil the potential pathophysiological role of these different CXCL8 isoforms.

Ethics statement
The experimental protocols (involving the use of laboratory animals) were carried out in strict accordance with the Belgian and European guidelines concerning the handling of laboratory animals and were approved by the Animal Ethical Committee of the K.U.Leuven (approved project P07029).
No specific approval from an institutional review board is required for the use of buffy coats for the following reasons: (1) no personal patient information is made available, (2) buffy coats cannot be used for treatment of patients and are waste products for the blood transfusion centre and (3) blood donors sign an agreement that parts of the donation that cannot be used for patient treatment may be used for scientific research.

Enzymatic modification of CXCL8 in vitro
To evaluate the susceptibility of CXCL8 variants to enzymatic modification by plasmin or thrombin, human CXCL8 variants were incubated with thrombin or plasmin at 37uC for the indicated time intervals at enzyme-substrate (E/S) molar ratios of 1/100 and 1/1000, respectively. Cleavage was accomplished in phosphate-buffered saline (PBS) (pH 7.4). CXCL8 processing was stopped by adding 0.1% (v/v) trifluoroacetic acid (TFA). Processed preparations were desalted on C4 ZipTips (Millipore, Billerica, MA, USA) and analyzed by electrospray ion trap mass spectrometry (Esquire LC ion trap mass spectrometer; Bruker, Bremen, Germany).
Chemical synthesis of CXCL8 and the posttranslationally modified isoforms CXCL8(-2-77), CXCL8(1-77), CXCL8(2-77), CXCL5  and CXCL8(6-77) were chemically synthesized based on fluorenyl methoxycarbonyl (Fmoc) chemistry using a 433A solid phase peptide synthesizer (Applied Biosystems, Foster City, CA, USA), as described by Loos et al. [31]. Intact synthetic peptides were separated from incompletely synthesized by-products by RP-HPLC on a Source 5-RPC column (4.66150 mm; GE Healthcare Bio-sciences, Little Chalfont, United Kingdom) applying an acetonitrile gradient in 0.1% (v/v) TFA. Part (0.7%) of the column effluent was split online to an ion trap mass spectrometer (Esquire LC). Fractions containing the protein with the correct M r were folded overnight, subsequently loaded on a C8 RP-HPLC column (Aquapore-RP 300 2.16220 mm; Perki-nElmer, Waltham, MA, USA) and eluted with an acetonitrile gradient in 0.1% TFA. Proteins eluting from the column were detected spectrophotometrically at 214 nm and the M r was measured online by ion trap mass spectrometry. Homogenous CXCL8 fractions were pooled, lyophilized and dissolved in ultrapure water (MilliQ; Millipore). Several assays were applied to determine protein concentrations, i.e. enzyme-linked immunosorbent assay (ELISA), NH 2 -terminal sequencing based on Edman degradation (491 cLC protein sequencer, Applied Biosystems), sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL, USA). Figure 6. Overview of the naturally occurring CXCL8 isoforms. Several NH 2 -terminally modified isoforms of CXCL8 have been purified from the conditioned medium of leukocytes, fibroblasts and endothelial cells. The NH 2 -terminal and COOH-terminal sequences of these isoforms are depicted in the one-letter code (B = citrulline). Based on the criteria formulated by Von Heijne [29], the 99 amino acid precursor molecule may be cleaved by signal peptidases between Cys 20 and Glu 21 or between Gly 22 and Ala 23 , the latter being more probable. Indeed both resulting CXCL8 proteins have been identified and are indicated in this manuscript as CXCL8(-2-77) and CXCL8(1-77), respectively. CXCL8  can be cleaved by a number of proteases, resulting in the listed NH 2terminally truncated forms. Furthermore, CXCL8 can be citrullinated by peptidylarginine deiminase (PAD). These NH 2 -terminal modifications differently affect the ability of CXCL8 to recruit neutrophils. The in vivo neutrophil-attracting activity is defined in the boxes at the right. Isolation of granulocytes from fresh human buffy coats Fresh human buffy coats were obtained from the Blood Transfusion Center of the Red Cross (Leuven, Belgium). The bulk of erythrocytes were removed by sedimentation in erythrocyteaggregating hydroxyethyl starch (Plasmasteril, Fresenius AG, Bad Homburg, Germany) for 30 min. After centrifugation, the leukocyte pellet was dissolved in PBS and layered on Ficoll-sodium diatrizoate medium (Lymphoprep, Axis-Shield PoC AS, Oslo, Norway) in order to separate mononuclear cells from granulocytes by density gradient centrifugation (30 min, 4006g). Residual erythrocytes in the granulocyte pellet were eliminated by hypotonic shock (30 s) in ultrapure water (MilliQ). Purified granulocytes were used for chemotaxis and calcium signaling experiments.

Calcium mobilization through CXCR1 and CXCR2 by CXCL8
The capacity of the CXCL8 isoforms to induce signaling through human CXCR1 or human CXCR2 resulting in an increase of the [Ca 2+ ] i was tested on both freshly purified granulocytes and CXCR1-and CXCR2-transfected HEK293 cells, as described [32].

Neutrophil migration in vitro and in vivo in response to CXCL8
The chemotactic activity of the different CXCL8 isoforms for neutrophilic granulocytes in vitro was compared using a 96-well Boyden chamber-based chemotaxis assay (Multiscreen plates, Millipore) and migrated neutrophils were quantified with a luminescence ATP detection assay (PerkinElmer), as previously described [32].
In vivo blood vessel extravasation of circulating neutrophilic granulocytes into the peritoneal cavity was examined by i.p. injection of endotoxin-free CXCL8 [in 0.9% (w/v) NaCl] or vehicle in female NMRI mice (8-10 weeks, raised pathogen-free; Elevage Janvier, Le Genest Saint Isle, France). Endotoxin concentrations were evaluated with the Limulus amoebocyte lysate test (Cambrex, East Rutherford, NJ, USA). After 2 h, mice were sacrificed and the peritoneal cavity was washed with 5 ml saline enriched with 2% (v/ v) FBS and 20 U/ml heparin. Total leukocyte counts in the peritoneal lavage were determined. 10 5 cells were used for the preparation of two cytospins/mouse. After Hemacolor (Merck, Darmstadt, Germany) staining, percentages of neutrophils in the peritoneal lavage were determined in duplicate by differential microscopic counting of at least 100 leukocytes per cytospin.

Binding of CXCL8 to chemokine receptors CXCR1 and CXCR2
The binding capacity of the CXCL8 isoforms was evaluated through the measurement of competition with 125 I-labeled CXCL8(6-77) (PerkinElmer). CXCR1-or CXCR2-transfected cells (2610 6 ), reconstituted in binding buffer (50 mM HEPES, 1 mM CaCl 2 , 5 mM MgCl 2 and 0.1% (w/v) bovine serum albumin), were incubated for 2 h with radiolabeled CXCL8 and varying concentrations of unlabeled CXCL8. Subsequently, the cells were washed twice with 1 ml of binding buffer. Finally, gamma radiation emitted by the cells was measured with a gamma counter. Maximal 125 I-CXCL8(6-77) binding was determined by incubating the cells with radiolabeled CXCL8 alone and was set at 100%.

Binding of CXCL8 to heparin
The heparin binding properties of the CXCL8 isoforms were compared on heparin binding plates (Iduron, Paterson Institute for Cancer Research, University of Manchester, Manchester, UK), which adsorb heparin without modification to retain the proteinbinding characteristics. Heparin (Low in calcium; Iduron), diluted in standard assay buffer (SAB; 100 mM NaCl, 50 mM sodium acetate, 0.2% v/v Tween-20; pH 7.2) (25 mg/ml) was coated overnight at room temperature on the plasma-polymerized surface of heparin binding plates. After three wash steps with SAB, the plates were blocked with blocking buffer (SAB enriched with 0.2% w/v gelatin) for 1 h at 37uC. Blocking buffer was discarded and dilutions of CXCL8 (in blocking buffer) were added in triplicate and incubated for 2 h at 37uC. Unbound CXCL8 was removed by three washes with SAB and bound CXCL8 was detected with biotinylated polyclonal rabbit anti-human CXCL8 (PeproTech), diluted in blocking buffer. After 1 h of incubation at 37uC, the plates were washed with SAB. Horse radish peroxidase-labeled streptavidin was added for 30 min at 37uC. Subsequently, the plates were washed and the peroxidase activity was quantified by adding a horse radish peroxidase substrate solution and conversion of 3,39,5,59-tetramethylbenzidine, supplemented with 0.004% (v/ v) H 2 O 2 was measured at 450 nm using a spectrophotometer. Equal affinity of the biotinylated anti-human CXCL8 antibody for the different CXCL8 variants was verified. Percentage heparin binding was calculated by subtracting the mean optical density (OD) of the control samples (blocking buffer) from the average OD of the test sample, followed by division by the average OD obtained with 300 nM of CXCL8(1-77) and multiplication by 100.