Analysis of Individual Azurocidin N-glycosylation Sites in Regard to Its Secretion by Insect Cells, Susceptibility to Proteolysis and Antibacterial Activity

Azurocidin is an inactive serine protease homolog with primary sequence similarity to neutrophil elastase, cathepsin G, and proteinase 3. The aim of this study was to investigate possible consequences of differential glycosylation of azurocidin in regard to its secretion, protein stability as measured by susceptibility to proteolysis, and antibacterial activity. Site-directed mutagenesis was employed to generate mutant azurocidin variants lacking individual N-glycosylation sites. Our results show that N-linked glycans may play a role in proper azurocidin folding and subsequent secretion by insect cells. We also demonstrate that N-linked glycosylation contributes to azurocidin stability by protecting it from proteolysis. The lack of N-glycosylation at individual sites does not significantly influence the azurocidin antibacterial activity. or cytokines. During evolution, azurocidin adopted new functions in host defense against invading organisms by mechanisms not involving proteolytic activity. Azurocidin is a potent chemoattractant for monocytes (Pereira et al., 1990) and microglia (Pereira et al., 2003), stimulates monocyte survival and differentiation and thrombospondin secretion (Ostergaard & Flodgaard, 1992), enhances the LPS-stimulated release of prostaglandin E2 (Heinzel-mann et al.din also stimulates reversible contraction of fibrob-lasts and endothelial cells (Ostergaard & Flodgaard, 1992) and activates endothelial cell protein kinase C (Pereira et al., 1996b). Several studies have shown that azurocidin has a potent bactericidal activity against a variety of bacteria (Shafer et al. INTRODUCTION Azurocidin, also known as heparin-binding protein (HBP) or 37-kDa cationic antimicrobial protein (CAP37), is an inactive serine protease ho-molog with the highest primary sequence similarity to neutrophil elastase and proteinase 3 (for references see Watorek, 2003). Human azurocidin lacks proteolytic activity due to substitutions (H41S and S175G) in the catalytic triad. The protein is consti-tutively expressed during neutrophil differentiation and targeted to azurophilic granules. Recently , the presence of azurocidin has been shown in the endothelium of Alzheimer's brain microvessels (Pereira et al., 1996a), in the endothelium associated with atherosclerotic plaques (Lee et al., 2002), in pancreatic juice (Gronborg et al., 2004), and several reports have shown that its expression is induced in endothelial (Lee et al., 2002) and corneal epithe-lial cells (Ruan et al., 2002) in response to inflam-matory mediators such as lipopolysaccharide (LPS) It has been widely shown that N-glycosylation significantly affects intracellular proteolytic processing , secretion efficiency, and posttranslational stability of secreted proteins. In human azurocidin, three N-linked glycosylation sites have been identified The Asn114 glycosylation site is part of a putative endothelial cell protein ki-nase C activation site, thus it …


INTRODUCTION
Azurocidin, also known as heparin-binding protein (HBP) or 37-kDa cationic antimicrobial protein (CAP37), is an inactive serine protease homolog with the highest primary sequence similarity to neutrophil elastase and proteinase 3 (for references see Watorek, 2003).Human azurocidin lacks proteolytic activity due to substitutions (H41S and S175G) in the catalytic triad.The protein is constitutively expressed during neutrophil differentiation and targeted to azurophilic granules.Recently, the presence of azurocidin has been shown in the endothelium of Alzheimer's brain microvessels (Pereira et al., 1996a), in the endothelium associated with atherosclerotic plaques (Lee et al., 2002), in pancreatic juice (Gronborg et al., 2004), and several reports have shown that its expression is induced in endothelial (Lee et al., 2002) and corneal epithelial cells (Ruan et al., 2002) in response to inflammatory mediators such as lipopolysaccharide (LPS)

2007
K. Indyk and others It has been widely shown that N-glycosylation significantly affects intracellular proteolytic processing, secretion efficiency, and posttranslational stability of secreted proteins.In human azurocidin, three N-linked glycosylation sites have been identified (Asn100, Asn114, and Asn145) (Iversen et al., 1997;1999, Karlsen et al., 1998)).The Asn114 glycosylation site is part of a putative endothelial cell protein kinase C activation site, thus it can be important for the cellular activation mechanism.The Asn100 and Asn114 glycosylation sites are located in a long surface loop which serves as a linker connecting two similar β-barrel domains, each composed of six antiparallel β-strands.These glycosylation sites may be involved in the folding of the two domains into the correct position with respect to each other.The Asn145 glycosylation site is situated in a β-strand in the second β-barrel (Iversen et al., 1997;Karlsen et al., 1998).The native protein isolated from neutrophils has a high variety of N-linked glycan structures (Olczak & Watorek, 2002), and the dominant glycans are GlcNAc 2 Man 2 (Fuc), GlcNAc 2 Man 3 GlcNAc GalNeu(Fuc), and GlcNAc 2 Man(Fuc).The dominant glycoform found in recombinant protein produced in Sf9 insect cells contains two GlcNAc 2 Man 3 (Fuc) and one GlcNAc 2 Man 3 unit (Rasmussen et al., 1996).Both native azurocidin purified from neutrophils and recombinant one produced in insect cells show high glycosylation microheterogeneity (Pereira, 1995;Rasmussen et al., 1996;Karlsen et al., 1998).The aim of this study was to investigate possible consequences of differential glycosylation of azurocidin in regard to its secretion, stability as measured by susceptibility to proteolysis, and antibacterial activity.

MATERIALS AND METHODS
Cell culture and transfection.Sf9 TriEx insect cells (Novagen) were propagated in TriEx insect cell medium at 28°C as recommended by the manufacturer.Transient transfection and protein analyses were performed in 12-or 6-well plates as described previously (Olczak & Olczak, 2006).Transfection with different amounts of wild-type DNA (pAzu1 plasmid; Table 1) was used in preliminary experiments to evaluate the influence of the amount of transfected DNA on azurocidin secretion.No significant dependence was found on the amount of plasmid DNA used (not shown), therefore for all experiments presented in this study 2 µg of plasmid DNA was used.Cells and media were collected after 96 h.Before harvesting, the integrity of the cells was checked using Trypan Blue staining.The cells were confluent and viable (viability at least 95%; data not shown).Cells were separated by centrifugation of the suspension at 800 × g for 10 min at 4°C.Non-concentrated medium was subjected to all analyses.
Purification of azurocidin.Recombinant azurocidin was purified from culture medium using cobalt-affinity chromatography as described previously (Olczak & Olczak, 2006).For this purpose, Sf9 cells were grown in 10 ml of TriEx medium at 28°C with shaking at 145 r.p.m. for 90 h.Native human azurocidin was purified from neutrophils (obtained from "buffy coats" provided by the Wroclaw District Blood Bank) according to Watorek et al. (1996) with the following modifications.Briefly, frozen granules were suspended in 50 mM acetate buffer, pH 4.0, containing 1 M NaCl and homogenized using a glass homogenizer.The granule extract was centrifuged at 30 000 × g for 30 min at 4°C and the pellet was re-homogenized five times.Granule extract was adjusted to pH 8.0 using solid Tris and applied to a Sepharose-Trasylol column equilibrated in 50 mM Tris/HCl buffer, pH 8.0, containing 1 M NaCl.After elution of non-adsorbed proteins with the equilibration buffer, human neutrophil elastase and cathepsin G were eluted with 50 mM citrate buffer, pH 4.0, containing 1 M NaCl.Then, elution of azurocidin was accomplished by eluting with 100 mM glycine/ HCl buffer, pH 3.0.Fractions containing azurocidin were concentrated using an Amicon ultrafiltration unit (YM1).
Tryptic digestion.Culture media containing azurocidin glycosylation variants were adjusted with the medium of cells transfected with an empty vector to the same concentration of recombinant protein.Samples were incubated with trypsin (the molar trypsin : protein ratio was 1 : 500; Sigma) at 37°C in 100 mM Tris/HCl, pH 8.0, containing 20 mM CaCl 2 .At various time points aliquots were removed in which trypsin activity was stopped by adding SDS/PAGE sample buffer and immediate boiling.Controls without trypsin were included in the protocol.
Antibacterial assay.Radial diffusion assays were performed according to Lehrer et al. (1991) and McCabe et al. (2002) with the following modifications.Escherichia coli K12 Row cells (PCM 1862 strain) were grown overnight in 3% Trypticase soy broth (TSB; Beckton Dickinson).To obtain mid-logarithmic growth phase (OD 600 = 0.55-0.65),150 µl of overnight culture was inoculated into 50 ml of fresh TSB and incubated for additional 3-3.5 h at 37°C.Bacteria were harvested by centrifugation at 2500 × g for 7 min at 4°C and washed twice with cold 10 mM sodium phosphate buffer, pH 7.5.Suspension containing 10 8 cfu (OD 620 = 0.2 corresponds to 5 × 10 7 cfu/ml) was added to 12 ml of warm 0.8% low electroendosmosis agarose (Sigma) in 10 mM sodium phosphate buffer, pH 7.5, containing 0.02% Triton X-100 (Roth) and 0.02% bovine serum albumin (BSA) (Roth).After dispersing of the bacteria, the suspension was poured into 90-mm diameter Petri dishes.Several dilutions of neutrophil-derived and recombinant azurocidin were added to wells punched in the agarose layer (7.5 µl per well).The plates were incubated for 3 h at 37° C to allow protein diffusion and then overlaid with 10 ml of 1% agar (BIOCORP) containing 6% TSB.Following an 18-h incubation at 37°C the plates were stained for 24 h according to Lehrer et al. (1991).The areas of the clearing zones surrounding the wells were measured using a Stemi SV 11 stereomicroscope (Carl Zeiss).All assays were performed at least three times.In all experiments lysozyme (Serva) was used as a positive control.Buffer used for azurocidin dilution served as a negative control.
Bacterial membrane permeabilization assay.The assay was performed according to Mak et al. (2007) using E. coli JM83 strain harboring pCH110 plasmid (Pharmacia-Amersham) encoding β-galactosidase and ampicillin resistance, with the following modifications.Aliquots of 4 µl of bacterial suspension (2.5 × 10 8 cfu/ml) were mixed in wells of 96well round-bottom polypropylene microtiter plates (Greiner) with 60 µl of azurocidin samples in a buffer at the pH range 4.5-7.5.In addition, several NaCl concentrations (0-0.35M) were tested.The final amount of azurocidin was up to 2.6 µg per well.The plates were incubated with shaking for 15 min at 37°C.Then, 150 µl of 3.3 mM β-galactosidase colorimetric substrate (o-nitrophenyl-β-d-galactopyranoside, ONPG; Sigma) in 0.2 M Tris/HCl, pH 8.0, was added to each well and the plates were again incubated for 30 min at 37°C.The reaction was stopped by adding 45 µl of 1.7 M Na 2 CO 3 .The mixtures (200 µl) were individually transferred to flat-bottom clear polyethylene 96-well plates (Nunc) and the absorbance at 420 nm was measured using a Spectra Max 340 microplate spectrophotometer (Molecular Devices).As control samples for 0% perforation served bacteria incubated without azurocidin, and as control samples for 100% perforation, bacteria disintegrated by sonication prior to the assay.

Secretion of azurocidin glycosylation variants by insect cells
N-glycosylation is acquired in the endoplasmic reticulum and Golgi apparatus, both of which form part of the secretory pathway.Therefore, understanding the importance of individual N-linked sites with respect to azurocidin secretion would be advantageous.It has been shown that recombinant azurocidin expressed in baculovirus-infected insect cells possesses similar oligosaccharide structures as the native neutrophil-derived protein (Almeida et al., 1996;Rasmussen et al., 1996;Karlsen et al., 1998).Recently, we have reported on a directly transfected insect cell system suitable for analysis of multiple azurocidin variants, including glycosylation mutants (Olczak & Olczak, 2006).We showed that recombinant azurocidin containing its native signal peptide and a fusion protein at the C-terminus was effectively expressed in transiently transfected insect cells and secreted into the culture medium at high levels (Olczak & Olczak, 2006).In this study we used site-directed mutagenesis to remove, either singly or in combination, the N-glycosylation sites by replacing the Asn residue from the consensus glycosylation sequon (Asn-X-Thr/ Ser) by a Ser residue (Table 1).To determine whether azurocidin affected in the glycosylation sites can be effectively secreted, the respective protein mutants were expressed in transiently transfected insect cells and analyzed by SDS/PAGE and Western blotting.A clear decrease in the molecular mass of the mutated variants of azurocidin was observed after detection with Coomassie Brilliant Blue G-250 or anti-His antibodies (Fig. 1).Each successive mutation reduced the apparent molecular mass of the recombinant mutant protein, suggesting that the mutations prevented the addition of oligosaccharides.Our results showed that secretion of single glycosylation mutants was comparable with that observed for the wild-type re-combinant protein, secretion of double glycosylation mutants was slightly decreased, whereas in the case of the triple glycosylation mutant the decrease in secretion was pronounced (Fig. 1).This effect could be explained by the function of Asn-linked oligosaccharides in the overall protein stability and/or folding.Iversen et al. (1999) showed that the lack of glycosylation did not affect the folding of azurocidin, as the secondary and tertiary structures of non-glycosylated recombinant azurocidin and wild-type recombinant azurocidin were nearly identical.The surface loops to which some glycans are attached were also unaffected by the removal of the glycosylation.In contrast, Almeida et al. (1996) demonstrated that recombinant azurocidin produced in tunicamycin-treated insect cells was found exclusively in a cell-associated form, suggesting that glycosylation may be important for azurocidin stability during secretion process.In addition, all attempts to produce azurocidin in E. coli cells have failed (Almeida et al., 1996).Based on those data and our results presented in this study one may suggest that N-linked glycans might be important for proper azurocidin folding and subsequent secretion.

Susceptibility of azurocidin glycosylation variants to tryptic digestion
It is generally accepted that glycosylation is required to stabilize the protein structure.However, results concerning the susceptibility of glycoproteins to proteolysis are not convincing (van Berkel et al., 1995;van Heen et al., 2004;Valmu et al., 2005).In this study we were interested whether N-linked glycosylation is important for azurocidin stability as measured by susceptibility to proteolysis.Recombinant azurocidin glycosylation variants were examined in culture medium without purification.Proteins were digested with trypsin, separated by SDS/ PAGE, and visualized by anti-His antibodies.The digestion was repeated three times under the same conditions giving similar patterns.As shown in Fig. 2, the decrease of glycosylation of recombinant azurocidin was correlated with an enhanced susceptibility to degradation.These results indicated that N-linked glycosylation does contribute to azurocidin stability by protecting it from proteolysis.

Antibacterial activity of azurocidin glycosylation variants
It has been reported that the N-terminal part of azurocidin may be important for its antimicrobial activity.A synthetic peptide corresponding to az- The presence of oligosaccharide chains (•) was checked experimentally by determination of electrophoretic mobility after SDS/PAGE of secreted azurocidin.b Recombinant protein secretion was examined by Western blotting.
urocidin amino-acid residues 20 to 44 mimicked the bactericidal activity of the protein against a range of bacteria (Pereira et al., 1993;Polikandritou et al., 1997).This peptide also exhibited a capacity to neutralize endotoxin activity in vitro and in vivo (Brackett et al., 1997).A recent report showed that binding of heparin, which occurred at a highly cationic region located within loops 3 and 4 of azurocidin, inhibited the azurocidin antimicrobial activity (McCabe et al., 2002).However, removal of the four positively charged amino-acid residues from the 20-44 azurocidin sequence did not affect the activity against bacteria and C. albicans (McCabe et al., 2002).Additional mutations generated within this region also did not lead to significant changes of the bactericidal activity of azurocidin (Kastrup et al., 2001).
Based on these data one may suggest that other regions or components, such as N-linked glycans, may contribute to the antibacterial function of azurocidin.To date, no reports have been published on the importance of individual N-linked glycans of human azurocidin for its antibacterial activity.
To answer this question, the antibacterial activity of native neutrophil-derived and recombinant wildtype azurocidin against E. coli, as determined by radial diffusion assay, was first compared.The dose dependence of the activity of recombinant azurocidin was similar compared with that observed for the native neutrophil-derived protein (not shown), which is in agreement with other studies (Almeida et al., 1996).As shown in Fig. 3, neutrophil-derived azurocidin showed a higher antibacterial activity compared with wild-type recombinant azurocidin.This may be explained by the presence of the Cterminal fusion protein attached to the recombinant protein or by different purification methods used.When the results of a radial diffusion experiment performed with wild-type recombinant azurocidin were compared with those for single glycosylation mutants it was clear that only the N145S azurocidin mutant exhibited a slightly decreased antibacterial activity (Fig. 3).It has been shown previously that a complete lack of glycosylation strongly reduced the biological activity of azurocidin, since the nonglycosylated protein mediated a limited stimulation of LPS-induced cytokine release from monocytes (Iversen et al., 1999).Although the lack of N-glycosylation at individual sites did not significantly influence the antibacterial activity, all these data support the hypothesis that the oligosaccharide moiety of human azurocidin may play a role in its biological function.
Several investigators have suggested that the bacterial killing mechanism is initiated by the binding of azurocidin to bacterial LPS/lipid A (Pereira et al., 1993;Iversen et al., 1997;Brackett et al., 1997;Polikandritou et al., 1997).Upon binding to the lipid A component of LPS, azurocidin may induce leakage of the bacterial membrane.However our results showed that azurocidin, neutrophil-derived or recombinant, did not show membrane permeabilizing effect on E. coli cells under conditions used (not shown).Antibacterial activity against E. coli was determined using radial diffusion assay.The area (mm 2 ) of the clearing zone surrounding the wells was measured and results are shown as means ± S.D. for three independent experiments performed in duplicates, using 2.5 µg of azurocidin per well.nAzu WT, native neutrophil-derived azurocidin; rAzu WT, recombinant wild-type azurocidin; rAzu N100S, rAzu N114S, and rAzu N145S, recombinant azurocidin glycosylation mutants.

CONCLUDING REMARKS
In this study, site-directed mutagenesis was employed to generate various azurocidin mutant proteins lacking individual N-glycosylation sites.Our results showed that although the lack of single N-linked glycans did not influence azurocidin secretion, the absence of two or three N-linked glycans might play a role in azurocidin folding and subsequent secretion by insect cells.Azurocidin lacking N-linked glycans, either singly or in combination, appeared to be more susceptible to trypsin digestion than its fully glycosylated form.The lack of occupancy of single N-glycosylation sites did not significantly influence the azurocidin antibacterial activity.We also found that azurocidin did not exhibit a membrane permeabilizing effect on E. coli cells.Based on the literature data and the results presented in this study it seems that further characterization of this important inflammatory mediator is likely to be critical to understanding the pathogenesis of various diseases and the design of effective therapeutics.

Figure 1 .
Figure 1.Secretion of recombinant azurocidin glycosylation variants by insect cells.Protein secretion was examined by SDS/PAGE and Western blotting.Secreted azurocidin (shown by arrows) was visualized in culture media by Coomassie Brilliant G-250 (A) or anti-His antibodies (B) staining.The lack of individual N-glycans was evidenced by the decrease in the molecular mass of the mutant proteins.At least three clones were examined, for each mutation, giving similar secretion levels; the results are shown for representative clones.

Figure 2 .
Figure 2. Susceptibility of recombinant azurocidin glycosylation variants to proteolysis.Culture media containing azurocidin were adjusted to contain the same concentrations of the recombinant protein.The molar trypsin : protein ratio was 1 : 500.Samples were removed at the indicated time points and analyzed by SDS/PAGE and Western blotting and visualized with anti-His antibodies.Three independent experiments were performed; the results are shown for one representative.

Figure 3 .
Figure 3. Antimicrobial activity of recombinant azurocidin glycosylation variants.Antibacterial activity against E. coli was determined using radial diffusion assay.The area (mm 2 ) of the clearing zone surrounding the wells was measured and results are shown as means ± S.D. for three independent experiments performed in duplicates, using 2.5 µg of azurocidin per well.nAzu WT, native neutrophil-derived azurocidin; rAzu WT, recombinant wild-type azurocidin; rAzu N100S, rAzu N114S, and rAzu N145S, recombinant azurocidin glycosylation mutants.