The following were purchased: l-glutamine, antibiotic–antimycotic solution (10,000 IU/ml penicillin, 10 mg/ml streptomycin, and 25 µg/ml amphotericin B), Triton X-100, Histopaque^®-1077, bovin serum albumin (BSA), Heparinase I and III Blend from Flavobacterium heparinum, N-formyl-met-leu-phe (fMLF), phenolphthalein β-d-glucuronide sodium salt, detoxified LPS (from E. coli serotype 0111:B4), PMA, brefeldin A, and cycloheximide (Sigma-Aldrich, St. Louis, MO, USA); RPMI and fetal calf serum (FCS, endotoxin level <0.1 EU/ml) (MP Biomedicals Europe, Illkirch, France); P11, TCS 2314 (Tocris Bioscience, UK); anti-human VEGF-A₁₆₅ (monoclonal mouse IgG_2B; Clone 26603) and anti-human VEGF_165b (monoclonal mouse IgG₁; Clone 56-1) (R&D System, Minneapolis, MN, USA). Target-specific primers for VEGFA_165a, VEGFA_165b, VEGFB, VEGFC, VEGFD, Ang1, Ang2, and β-actin were produced and purified by Custom Primers (Life Technologies, Milan, Italy). The recombinant sPLA₂s human group IB, hGIIA, hGIIE, hGIIF, hGV, hGX, and hGXIIA and the inhibitors Me-Indoxam and RO092906A were prepared in the laboratory of Gerard Lambeau and were a generous gift from and Michael H. Gelb (Departments of Chemistry and Biochemistry, University of Washington, Seattle, WA, USA). sPLA₂ preparations were routinely checked for LPS contamination (Limulus amebocyte Test, MP Biomedicals) and discarded if the LPS concentration was above the detection limit of the assay (0.125 EU/ml). All other reagents were from Carlo Erba (Milan, Italy).

Granulocytes were isolated from buffy coats of healthy donors obtained from the Leukapheresis Unit. After dextran sedimentation, PMNs were obtained by centrifugation over Histopaque^®-1077 at 400 × g for 30 min, at 22°C, at a 1:1 ratio. Finally, PMNs were isolated by negatively removing all contaminating cells using the MACSxpress Neutrophil Isolation Kit and MACSxpress Erythrocyte Depletion Kit (Miltenyi Biotec, Bologna, Italy). This procedure yields a population of CD66b⁺ cells with a purity greater than 99% as assessed by flow cytometry. PMNs were suspended (5 × 10⁶ cells/ml) in complete medium (RPMI 1640 containing 5% FCS, 2 mM l-glutamine, and 1% antibiotic–antimycotic solution) and incubated in different plates (Falcon, Becton Dickinson, Franklin Lakes, NJ, USA) at 37°C in a humidified atmosphere of 5% CO₂ and 95% air. After 30 min of rest, the cells were used for the experiments.

PMNs were incubated (37°C, 5 min–3 h) in RPMI 1640 containing 5% FCS, 2 mM l-glutamine, and 1% antibiotic–antimycotic solution and stimulated with various concentrations (0.3–10 µg/ml) of human GIB, GIIA, GIIE, GIIF, GIII, GV, GX, GXIIA, fMLF (50 nM), detoxified LPS (100 ng/ml), and PMA (80 nM). In selected experiments, hGV and hGX (3 µg/ml) were preincubated (37°C, 20 min) with increasing concentrations (0.01–10 µM) of their inhibitors Me-Indoxam or RO092906A. In other experiments, PMNs were preincubated (37°C, 1 h) with heparinase (0.4 U/ml) or (37°C, 30 min) with P11 (100 nM) and/or TCS 2314 (100 nM), brefeldin A (10 µg/ml), cycloheximide (10 µg/ml) and then stimulated (37°C, 30 min) with hGV (3 µg/ml). In selected experiments, PMNs were preincubated (37°C, 15 min) with Me-Indoxam or RO092906A and then stimulated with fMLF (37°C, 10 min). At the end of the experiment, the supernatants were removed, centrifuged (1,000 × g, 4°C, 5 min) and stored at −80°C for subsequent determination of mediator release.

Total cellular RNA was isolated from PMNs using the SV RNA isolation system (Promega, Madison, WI, USA), treated with RNase-free DNase I and resuspended in DEPC water. RNA concentration and quality were assessed by spectrophotometry. Total mRNA was reverse-transcribed (Superscript III Reverse Transcriptase 200 U, Life Technologies) and quantitative PCR (qPCR) was carried out in iCycler-iQ5 real-time PCR detection system (Bio-Rad, Hercules, CA, USA) using SYBR Green Master Mix (Bio-Rad). Target-specific primers for VEGF_165a, VEGFA_165b, VEGFB, VEGFC, VEGFD, Ang1, Ang2, and β-actin suitable for qPCR were produced and purified by Custom Primers (Life Technologies, Milan, Italy) and are reported in Table 1. β-Actin was used as housekeeping gene to normalize cycle threshold (Ct) values using the 2^−ΔCt formula. The data were analyzed with iCycler iQ analysis software (Bio-Rad).

The concentration of VEGF-A, VEGFA_165b, VEGF-B, VEGF-C, VEGF-D, Ang1, Ang2, and CXCL8/IL-8 in the supernatants, lysed or freshly isolated PMNs (lysed in Tryton 0.1%) was measured in duplicate determinations using commercially available ELISA kits (R&D System). The ELISA sensitivity is 31.1–2,000 pg/ml for VEGF-A, 31.1–4,000 pg/ml for VEGF-A_165b, 9.4–300 pg/ml for VEGF-B, 62–4,000 pg/ml for VEGF-C, 31.1–2,000 pg/ml for VEGF-D, 156.25–10,000 pg/ml for Ang1, 31.1–4,000 pg/ml for Ang2, and 31.1–2,000 pg/ml for CXCL8/IL-8. β-Glucuronidase release was measured with a colorimetric assay.

A modified liposomal-based fluorescent assay was used to measure PLA₂ activity in neutrophil supernatants. Briefly, a PLA₂ substrate cocktail consisting of 7-hydroxycoumarinyl-arachidonate (0.3 mM), 7-hydroxycoumarinyl-linolenate (0.3 mM), hydroxycoumarinyl-6 heptenoate (0.3 mM), 10 mM dioleoylphosphatidylcholine (DOPC), and 10 mM dioleoylphosphatidylglycerol (DOPG) was prepared in ethanol. Liposomes were formed by gradually adding 77 µl substrate/lipid cocktail to 10 ml PLA₂ buffer (50 mM Tris–HCl at pH 8.9, 100 mM NaCl, 1 mM CaCl₂) while stirring rapidly over 1 min using a magnetic stirrer (Invitrogen EnzChek^® phospholipase A₂ assay). Neutrophil supernatants (50 µl) was added to 96-well plates, and PLA₂ activity was initiated by adding 50 µl substrate cocktail. Fluorescence (excitation at 360 nm and emission at 460 nm) was measured, and specific activity (relative fluorescent units/μg protein/min) for each sample was calculated.

Non-tumor and tumor lung tissues were obtained from patients affected by lung adenocarcinoma (HCV⁻, HBsAg⁻, and HIV-1⁻) undergoing lung resection. Immunohistochemical staining has been carried out on 4-µm lung cancer serial sections from formalin-fixed, paraffin-embedded tissues, in order to evaluate the expression of hGV and CD66b. Negative control slides without primary antibody were included for each staining (not shown). Paraffin slides were deparaffinized in xylene and rehydrated through graded alcohols. Antigen retrieval was performed with slides heated in 1 mM EDTA buffer (pH 9.0) for hGV and 0.01 M citrate buffer (pH 6.0) for CD66b, in a bath for 20 min at 97°C. After antigen retrieval, the slides were allowed to cool and rinsed with TBS. The endogenous peroxidase activity was inactivated with 3% hydrogen peroxide. Following the protein block (BSA 5% in PBS 1×), the slides were incubated with a polyclonal rabbit antibody against hGV (NBP2-31558, Novus Biologicals, Littleton, CO, USA, dilution 1:100) or a monoclonal mouse anti-human CD66b antibody (555723, BD Pharmingen, San Jose, CA, USA, dilution 1:400 at 4°C overnight). The sections were rinsed in TBS and incubated with biotinylated anti-rabbit or anti-mouse antibodies, respectively, for 1 h at room temperature. Immunoreactivity was visualized using 3,3′-diaminobenzidine (DAB) and avidin–biotin–peroxidase complex. Finally, sections were weakly counterstained with hematoxylin, mounted, and interpreted using light microscope.

The data are expressed as mean values ±SD of the indicated number of experiments. Statistical analysis was performed with Prism 6 (GraphPad Software) by one-way analysis of variance followed by Dunnett’s test (when comparison was made against a control) or Bonferroni’s test (when comparison was made between each pair of groups). Statistically significant differences were accepted when the p value was at least ≤0.05.

To evaluate the impact of sPLA₂ on the production of angiogenic and lymphangiogenic factors by human PMNs, we first assessed the basal expression of VEGFs and Angs. Freshly isolated PMNs constitutively expressed VEGF-A, VEGF-B, and Ang1 at both mRNA and protein levels. By contrast, no expression of VEGF-C and VEGF-D (lymphangiogenic factors) and Ang2 could be detected (Figures 1A,B). Interestingly, PMNs constitutively expressed VEGF-A_165b (Figures 1A,B), the antiangiogenic splice variant of VEGF-A₁₆₅ (42).

We then evaluated the effects of several human recombinant sPLA₂s on the secretion of VEGF-A, VEGF-A_165b, VEGF-B, and Ang1 as well as the proangiogenic chemokine CXCL8/IL-8 from PMNs. Several sPLA₂s induced the release of VEGF-A (Figure 2A) and CXCL8/IL-8 (Figure 2B) and promoted the release of Ang1 (Figure 2C). However, at the concentration used (5 µg/ml), the release of VEGFA and CXCL8 was significant upon stimulation with hGV and hGX, and the release of Ang1 was significant upon stimulation with hGIIF, hGV, and hGX. These results were paralleled by the highest β-glucuronidase release used as a marker of exocytosis (Figure S1A in Supplementary Material) suggesting the release of preformed mediators rather than de novo synthesis. By contrast, no secretion of VEGF-B (Figure S1B in Supplementary Material) could be observed in any of the tested conditions. Interestingly, hGV and, to a lesser extent, hGIIA were the only sPLA₂s to induce the secretion VEGF-A_165b (Figure 2D).

To better understand the mechanisms of sPLA₂ neutrophil stimulation, we performed dose–response and time-dependent experiments with hGV and hGX sPLA₂s, which are more effective in stimulating human PMNs and are the only sPLA₂s expressed by these cells (16). Figure 3 shows that the release of VEGF-A (Figure 3A), CXCL8/IL-8 (Figure 3B), and Ang1 (Figure 3C) in response to hGV and hGX was induced by concentrations as low as 1 µg/ml. By contrast, Figure 3D shows that VEGF-A_165b secretion was observed only with hGV, but not with hGX, at all tested concentrations. Moreover, hGV induced the release of these mediators as early as after 5 min of stimulation (Figures 3E,H).

To verify whether hGV-induced secretion of proangiogenic and antiangiogenic mediators requires de novo protein synthesis, neutrophils were stimulated with hGV in the presence or absence of brefeldin A (an inhibitor of anterograde cellular transport and protein secretion) or cycloheximide (an inhibitor of protein synthesis). Neither brefeldin A nor cycloheximide affected the spontaneous release of VEGF-A, CXCL8/IL-8, Ang1, and VEGF-A_165b (data not shown). However, brefeldin A but not cycloheximide significantly inhibited the release of these mediators induced by hGV (Table S1 in Supplementary Material). Moreover, we measured VEGF-A, Ang1, and VEGF-A_165b protein levels in supernatants and cellular lysates of unstimulated and hGV-stimulated PMNs and found that total protein levels (supernatants plus cellular lysates) of angiogenic mediators were not significantly modulated by hGV stimulation (Table S2 in Supplementary Material).

To corroborate these findings and to exclude a non-specific PMN activation by sPLA₂, we assessed the secretion of VEGF-A, VEGF-A_165b, Ang1, and CXCL8/IL-8 following the stimulation of PMNs with well-known neutrophil stimuli, such as fMLF, PMA, and LPS. Even though all these stimuli induced the release of VEGF-A (Figure 4A) and CXCL8/IL-8 (Figure 4B), the secretion of Ang1 (Figure 4C) was only induced by PMA. By contrast, none of the tested stimuli induced the release of VEGF-A_165b (Figure 4D). Taken together, these results indicate that hGV was the only sPLA₂ able to induce the release of preformed pro- and antiangiogenic molecules, whereas the effect of hGX was limited to proangiogenic factors.

Several evidences demonstrate that PMNs express PLA₂R1 that is involved in sPLA₂-induced neutrophil activation (8, 21–23). To verify whether hGV enzymatic activity and/or PLA₂R1 were involved in the production of pro- and antiangiogenic factors induced by hGV, we stimulated PMNs in the presence of Me-Indoxam. This inhibitor protrudes out of the catalytic groove when bound to sPLA₂s, thereby leading to steric hindrance and hence interfering with the sPLA₂–receptor interaction (52). We have previously shown that Me-Indoxam prevents receptor-mediated activation of HLMs and PMNs stimulated with sPLA₂s (17, 18, 21, 53). hGV also binds HSPGs that mediate its internalization (8). To verify a possible role for HSPGs, PMNs were pre-treated with heparinase to eliminate surface HSPGs before stimulation with hGV. Both Me-indoxam and heparinase pre-treatment markedly reduced the secretion of VEGF-A (Figure 5A), CXCL8/IL-8 (Figure 5B), and Ang1 (Figure 5C). Surprisingly, the release of VEGF-A_165b (Figure 5D) was not inhibited, instead it was significantly increased by Me-Indoxam and to a lesser extent also by heparinase pre-treatment. To explain these findings, we reasoned that Me-Indoxam and heparinase pre-treatment could enhance hGV binding to a different cell surface target on PMNs.

Two independent studies have shown that sPLA₂s can activate human monocytes by binding to α_Vβ₃ and α₄β₁ integrins (3, 7), which are also expressed on PMNs (54, 55). hGIIA is structurally related to hGV (8, 56) and binds to α_Vβ₃ and α₄β₁ integrins through a specific domain, different from the catalytic center or the PLA₂R1-binding site (3, 7). To investigate whether integrins could be involved in hGV-mediated release of VEGF-A_165b, we stimulated PMNs with hGV in the presence of the two integrin antagonists P11 and TCS 2314, which inhibit α_Vβ₃ and α₄β₁, respectively (57, 58). The combination of these inhibitors markedly reduced the release of VEGF-A_165b (Figure 5E). The specificity of this finding is supported by the observation that the secretion of CXCL8/IL-8 was not affected (Figure 5F).

fMLF induces the release of hGV by human PMNs (16, 19). We confirmed these results, showing an increased sPLA₂ activity in supernatants from fMLF-stimulated PMNs which was inhibited by Me-Indoxam but not by the hGX-specific inhibitor RO092906A (Figure S2 in Supplementary Material) (59). Since we found that hGV induces VEGF-A, CXCL8/IL-8, Ang1, and VEGF-A_165b (Figure 3) release, it would be conceivable that also fMLF induced the release of these mediators in an hGV-dependent manner. However, in our model, fMLF only promoted the secretion of VEGF-A and CXCL8/IL-8 (Figure 4). Figure 3 shows that the concentration of hGV required for the release of VEGF-A (Figure 3A) and CXCL8/IL-8 (Figure 3B) was 0.3 µg/ml whereas that for the release of Ang1 (Figure 3C) and VEGF-A_165b (Figure 3D) was at least 1 µg/ml. Therefore, we measured the enzymatic activity of several concentrations of recombinant hGV and found that the sPLA₂ activity in supernatants from fMLF-stimulated PMNs corresponded to a concentration of recombinant hGV ≤0.3 µg/ml (data not shown). Thus, it is likely that the release of hGV in response to fMLF is in our conditions insufficient to induce VEGF-A_165b and Ang1 release. Nevertheless, we asked whether fMLF-induced secretion of VEGF-A and CXCL8/IL-8 was mediated by the release of endogenous hGV. To address this question, PMNs were stimulated with fMLF in the presence or absence of Me-Indoxam or RO092906A. As expected, Me-indoxam reduced VEGF-A (Figure 6A) and CXCL8/IL-8 (Figure 6B) secretion while RO092906A had no effect (Figures 6C,D), suggesting an involvement of hGV but not hGX in fMLF-mediated neutrophil activation.

Our results show that hGV induces the release of proangiogenic and antiangiogenic factors by PMNs. Since angiogenesis is a hallmark of cancer-related inflammation (60) and PMNs infiltrate several human tumors (61), we assessed the expression of hGV and CD66b⁺ neutrophils by immunohistochemistry in neoplastic lung tissue samples. Figure 7 shows that hGV (Figures 7G,H) and CD66b (Figures 7E,F) were expressed in lung cancer samples but not in non-tumor areas (Figures 7A–D, respectively). These data suggest that hGV, expressed in lung adenocarcinoma microenvironment, can modulate tumor angiogenesis through the activation of TAN.