Cellular Distribution, Post-translational Modification, and Tumorigenic Potential of Human Group III Secreted Phospholipase A 2 *

Human group III secreted phospholipase A 2 (sPLA 2 -III) consists of a central group III sPLA 2 domain flanked by unique N- and C-terminal domains. We found that the sPLA 2 domain alone was sufficient for its catalytic activ- ity and for its prostaglandin E 2 (PGE 2 )-generating functions in various cell types. In several if not all cell types, the N- and C-terminal domains of sPLA 2 -III were proteo- lytically removed, leading to the production of the form containing only the sPLA 2 domain, which could be fur- ther N -glycosylated at two consensus sites. Immunohistochemistry demonstrated that sPLA 2 -III was preferen- tially expressed in the microvascular endothelium in human tissues with inflammation, ischemic injury, and cancer. In support of this, sPLA 2 -III was induced in cul- tured microvascular endothelial cells after stimulation with proinflammatory cytokines. Expression of sPLA 2 -III was also associated with various tumor cells, and colorectal cancer

From the ‡Department of Health Chemistry, School of Pharmaceutical Sciences, Showa University, 1-5-8 Hatanodai, Shinagawa-ku, Tokyo 142-8555, the §Tokyo Metropolitan Institute of Medical Science, 3-18-22 Honkomagome, Bunkyo-ku, Tokyo 113-8613, and the ʈDepartment of Pathology, Toho University, School of Medicine, 5-21-16 Omori-Nishi, Ohta-ku, Tokyo 143-8540, Japan Human group III secreted phospholipase A 2 (sPLA 2 -III) consists of a central group III sPLA 2 domain flanked by unique N-and C-terminal domains. We found that the sPLA 2 domain alone was sufficient for its catalytic activity and for its prostaglandin E 2 (PGE 2 )-generating functions in various cell types. In several if not all cell types, the N-and C-terminal domains of sPLA 2 -III were proteolytically removed, leading to the production of the form containing only the sPLA 2 domain, which could be further N-glycosylated at two consensus sites. Immunohistochemistry demonstrated that sPLA 2 -III was preferentially expressed in the microvascular endothelium in human tissues with inflammation, ischemic injury, and cancer. In support of this, sPLA 2 -III was induced in cultured microvascular endothelial cells after stimulation with proinflammatory cytokines. Expression of sPLA 2 -III was also associated with various tumor cells, and colorectal cancer cells transfected with sPLA 2 -III exhibited enhanced PGE 2 production and cell proliferation, which required sPLA 2 -III catalytic activity. When implanted into nude mice, the sPLA 2 -III-transfected cells formed larger solid tumors with increased angiogenesis compared with control cells. Moreover, small interfering RNA for sPLA 2 -III significantly reduced PGE 2 production and proliferation of colorectal cancer cells. Taken together, these results reveal unique cell type-specific processing and N-glycosylation of sPLA 2 -III and the potential role of this enzyme in cancer development by stimulating tumor cell growth and angiogenesis.
In the phospholipase A 2 (PLA 2 ) 1 family, secreted PLA 2 (sPLA 2 ) enzymes represent a group of structurally related, disulfide-rich, low molecular weight (typically 14 -18 kDa) enzymes with strict Ca 2ϩ dependence and a His-Asp catalytic dyad (1,2). To date, 11 sPLA 2 enzymes have been identified in mammals (IB, IIA, IIC, IID, IIE, IIF, III, V, X, XIIA, and XIIB). On a structural basis, these enzymes are further subdivided into three branches, namely groups I/II/V/X, group III, and group XII (1,2). Individual sPLA 2 s exhibit unique tissue and cellular localizations and enzymatic properties, suggesting their distinct, tissue-specific roles in various pathophysiological events.
Human group III sPLA 2 (sPLA 2 -III) is particularly unique among mammalian sPLA 2 s in that it consists of a central sPLA 2 (S) domain and unique N-terminal (N) and C-terminal (C) domains, that the molecular mass of its full-length protein is estimated to be 55 kDa, and that the sPLA 2 domain is homologous to bee venom group III sPLA 2 rather than to other mammalian sPLA 2 s (3). Recombinant expression of this enzyme by bacterial and mammalian cell expression systems reveals that the central S domain alone is sufficient for its enzymatic functions (3,4). On the basis of overexpression studies using HEK293 cells, the cellular arachidonate (AA)releasing function of sPLA 2 -III is similar to that of sPLA 2 -X, which acts primarily on the phosphatidylcholine (PC)-rich outer leaflet of the plasma membrane (4 -6). Because bee venom sPLA 2 does not contain N and C domains corresponding to those found in human sPLA 2 -III, it was anticipated that these domains in the human enzyme would be removed proteolytically, resulting in the production of the form containing only the S domain, although no experimental data that directly support this hypothesis has yet been provided. Also, there is no information as to which cell types express sPLA 2 -III in human tissues and which functions this enzyme exerts in particular tissue microenvironments.
In an effort to address these questions, we herein examined the cellular localization and processing of sPLA 2 -III by immunohistochemistry and immunoblotting with an antibody raised against the S domain of sPLA 2 -III. Our results indicate that sPLA 2 -III is processed to the S domain-only form in several cell types that intrinsically express this enzyme. Furthermore, we show that sPLA 2 -III is expressed in the microvascular endothelium in various human pathologic tissues as well as in tumor cells. These observations, together with cell biological studies, reveal the unexplored tumorigenic potential of this unique sPLA 2 enzyme.

EXPERIMENTAL PROCEDURES
Materials-Normal human pulmonary fibroblasts (NHPF), human pulmonary vascular smooth muscle cells (NHPVSMC), human dermal microvascular endothelial cells (HMVEC), and their culture media and supplements were purchased from BioWhittaker . The human colorectal  cancer cell line HCA-7, the human bronchial epithelial cell line BEAS-2B, the human prostate cancer cell line PC3, and the mouse Leydig cell  line I-10 were cultured in RPMI 1640 medium (Nissui Pharmaceutical Co.) containing 10% (v/v) fetal calf serum, as described previously (7,8). Rabbit antisera specific for individual human sPLA 2 s and pure recombinant human sPLA 2 s expressed by a bacterial expression system were generous gifts from Dr. M. H. Gelb (University of Washington) (9). Goat anti-human cyclooxygenase (COX)-2 antibody was purchased from Santa Cruz Biotechnology. Rabbit antibodies against human CD31 (an endothelial cell marker) and HAM56 (an alveolar macrophage marker) were purchased from Dako. Human sPLA 2 -III cDNA was kindly provided by Dr. G. Lambeau (CNRS-UPR 411, France) (3). cDNAs for truncated mutants (III-N ϩ S (N domain plus S domain), III-S ϩ C (S domain plus C domain), and III-S (S domain alone)) for sPLA 2 -III (4), c-myc, and rhoA (10) were described previously. Human interleukin (IL)-1␤ and tumor necrosis factor ␣ (TNF␣) were purchased from Genzyme. Tunicamycin, endoglycosidase H, and peptide N-glycanase F were obtained from Sigma. MTT cell counting kit was from Dojindo. Fluorescein isothiocyanate-conjugated and horseradish peroxidase-conjugated anti-IgG antibodies were purchased from Zymed Laboratories Inc.. Primers for reverse transcriptase (RT)-PCR were from FASMAC. Oligonucleotides for small interfering RNA (siRNA) were from Sigma.
Preparation of an Antibody against Human sPLA 2 -III-Rabbit antiserum against the synthetic peptide PRTFYNASWSSRATS, which corresponds to amino acid residues 276 -290 of human sPLA 2 -III protein, was prepared by BioLogica. The antiserum was subjected to enzyme immunosorbent assay to determine the antibody titer against the peptide and was used for subsequent analyses. The specificity of the antibody was evaluated by immunoblotting, as noted below.
Assays for AA Release and Prostanoid Generation-Cells grown to near confluence in 24-well plates (Iwaki Glass) were incubated with [ 3 H]AA (Amersham Biosciences) (0.1 Ci/ml) overnight. After three washes with fresh medium, 250 l of culture medium with or without 2 ng/ml IL-1␤ was added to each well, and the radioactivities released into the supernatants after incubation for appropriate periods were measured. The percentage release was calculated using the formula (S/(S ϩ P)) ϫ 100, where S and P are the radioactivities measured in the supernatant and cell pellet, respectively, as described previously (4 -6). Aliquots of the supernatants from replicate cells (without preincubation with radiolabeled fatty acids) were subjected to prostaglandin (PG) E 2 or 6-keto-PGF 1␣ enzyme immunoassay (Cayman Chemical).
Northern Blotting-Equal amounts (ϳ10 g) of total RNA obtained from cells by use of TRIzol reagent (Invitrogen) were applied to separate lanes of 1.2% (w/v) formaldehyde-agarose gels, electrophoresed, and transferred to Immobilon-N membranes (Millipore). The resulting blots were then probed with appropriate cDNA probes that had been labeled with [ 32 P]dCTP (Amersham Biosciences) by random priming (Takara Biomedicals). Hybridization and subsequent membrane washing were carried out as described previously (4 -6).
Expression of sPLA 2 s by the Adenovirus System-Adenoviruses bearing sPLA 2 cDNAs were prepared with the ViraPower adenovirus expression system (Invitrogen), as described previously (7,8). Briefly, the cDNAs were subcloned into the pENTER/D-TOPO vector using the pENTER Directional TOPO cloning kit (Invitrogen). After purification of the plasmids from the transformed Top10 competent cells (Invitrogen), the cDNA inserts were transferred to the pAd/CMV/V5-DEST vector (Invitrogen) by means of the Gateway system using LR clonase (Invitrogen). The plasmids were purified and digested with PacI (New England Biolabs). The linearized plasmids (1-2 g) were then mixed with 4 l of Lipofectamine 2000 (Invitrogen) in 200 l of Opti-MEM medium (Invitrogen) and transfected into subconfluent 293A cells (Invitrogen) in 1 ml of Opti-MEM in 6-well plates (Iwaki Glass). Then 293A cells were cultured for 1-2 weeks in RPMI1640 medium containing 10% fetal calf serum, with replacement of the medium every 2 days. When most cells became detached from the plates, the cells and culture medium were harvested together, freeze-thawed twice, and centrifuged to obtain the adenovirus-enriched supernatants. Then aliquots of the supernatants were added to fresh 293A cells and cultured for 2-3 days to amplify adenoviruses. After 2-4 times of amplification, the resulting adenovirus-containing media were used as virus stocks. Viral titers were determined by the plaque-forming assay with 293A cells. As a control, the pAd/CMV/V5-GW/lacZ vector (Invitrogen) was digested with PacI and transfected into 293A cells to produce lacZ-bearing adenovirus. Aliquots of the adenovirus-containing medium were added to the cells for subsequent analyses.
Expression of sPLA 2 s by the Lentivirus System-sPLA 2 cDNAs were stably transfected into HCA-7 cells with a ViraPower lentiviral expression system (Invitrogen), as described previously (11). Briefly, the cDNAs were subcloned into the pLenti6/V4-D-TOPO vector (Invitrogen). The resulting plasmid was transfected into 293FT cells (Invitrogen) with Lipofectamine 2000 in Opti-MEM medium, and aliquots of the supernatants harvested 3 days after transfection were then added to HCA-7 cells. The cells were cultured in the presence of 30 g/ml blasticidin, and the surviving cells that expressed appropriate levels of sPLA 2 s were used in subsequent studies.
Construction of cDNAs for sPLA 2 -III Point Mutants-cDNAs for sPLA 2 -III point mutants were constructed by the mismatched PCR method, as described previously (4 -6). The primers used were as follows: III-N1 sense (N1-S) primer (5Ј-TCTGCTGGGAGCTCCTCGGAG-3Ј) and III-N1 antisense (N1-AS) primer (5Ј-CTCCGAGGAGCTC-CCAGCAGA-3Ј); III-N2 sense (N2-S) primer (5Ј-ACCTTCTACAGT-GCCTCCTGG-3Ј) and III-N2 antisense (N2-AS) primer (5Ј-CCA-GGAGGCACTGTAGAAGGT-3Ј); III-HQ sense (III-HQ-S) primer (5Ј-TGCCGGGAACAAGACCGCTG-3Ј) and III-HQ antisense (III-HQ-AS) primer (5Ј-CAGCGGTCTTGTTCCCGGCA-3Ј) as well as III-S sense (III-S-S) primer, to which the signal sequence for human sPLA 2 -IIA was attached at the 5Ј-end, and antisense (III-S-AS) primer (4). To construct the III-S-N1 mutant, PCR was conducted with a set of III-S-S and III-N1-AS primers (for forward strand) and with a set of III-N1-S sense and III-S-AS primers (for back strand) using Pyrobest polymerase (Takara Biomedicals) and sPLA 2 -III cDNA in pCR3.1 (Invitrogen) (4) as a template with 25 cycles of 95°C for 30 s, 57°C for 30 s, and 72°C for 30s. The III-S-N2 and III-S-N1N2 mutants were constructed with a set of III-S-S and III-N2-AS primers (for forward strand) and III-N2-S and III-S-AS primers (for back strand) using cDNAs for sPLA 2 -III and III-S-N1 mutant as templates, respectively. To construct the III-HQ mutant, PCR was carried out with a set of III-S-S and III-HQ-AS primers (for forward strand) and with a set of III-S-HQ-S and III-S-AS primers (for back strand) using sPLA 2 -III cDNA as a template. In each case, the resulting forward and back strands were mixed, heated, annealed, and subjected to second PCR with III-S-S and III-S-AS primers under the same thermal conditions. The PCR products were subcloned into the pENTER/D-TOPO vector for adenoviral expression, as described above. The plasmids were sequenced using a Taq cycle sequencing kit (Takara Biomedicals) and an autofluorometric DNA sequencer 310 Genetic Analyzer (Applied Biosystems) to confirm the sequences. Preparation of the truncated sPLA 2 -III mutants was described previously (4).
RT-PCR-Synthesis of cDNA was performed using 0.5 g of total RNA from cells and tissues and avian myeloblastosis virus-reverse transcriptase, according to the manufacturer's instructions supplied with the RNA PCR kit (Takara Biomedicals). For amplification of sPLA 2 -III, primers N1-S and N2-AS were used. The PCR condition was 94°C for 30 s and then 30 cycles of amplification at 94°C for 5 s and 68°C for 4 min, using the Advantage cDNA polymerase mix (Clontech). The PCR condition for glyceraldehyde-3-phosphate dehydrogenase was described previously (11). The PCR products were analyzed by 1% agarose gel electrophoresis with ethidium bromide.
Immunohistochemistry-Immunohistochemistry of human lung was performed on autopsy specimens obtained within 3 h postmortem from three cases with normal lung structure (all male, 61-76 years old). Synovial tissue sections were obtained (with approval from the ethical committee of the University) from three patients with rheumatoid arthritis (RA) (all female, 68 -74 years old) having surgery at Toho University Ohmori Hospital. All patients were RA factor-seropositive according to the RA criteria (12) and had been treated intermittently with steroid and nonsteroidal anti-inflammatory drugs for similar pe-riods (Ͼ10 years) and in similar ways. RA states were evaluated on the basis of the morphology of the tissue sections, in which outgrowth of synovial cells, formation of lymph follicles, and infiltration of lymphocytes and plasma cells were obvious. Heart tissues with infarction were obtained from three patients (all male, 65-70 years old) within 3 h postmortem. Colorectal, uterus, and breast cancer tissues were obtained from patients having surgery at Toho University Ohmori Hospital (three patients for each tumor type).
The paraffin-embedded tissue sections (4 m thickness) were incubated with Target Retrieval Solution (Dako) as required, incubated for 10 min with 3% (v/v) H 2 O 2 , washed three times with PBS for 5 min each, incubated with 5% (v/v) skim milk for 30 min, washed three times with PBS/Tween for 5 min each, and incubated for 2 h with anti-human sPLA 2 antibodies (1:200 -500 dilutions) in PBS. Then the sections were treated with a catalyzed signal-amplified system staining kit (Dako) with diaminobenzidine substrate. The cell type was identified by conventional hematoxylin and eosin staining of serial sections adjacent to the specimen used for immunohistochemistry as described (13) or staining of serial sections with anti-CD31 and -HAM56 antibodies.
Measurement of sPLA 2 Activity-sPLA 2 activities were assayed by measuring the amounts of radiolabeled linoleic acid released from the substrate 1-palmitoyl-2-[ 14 C]linoleoyl-phosphatidylethanolamine (Amersham Biosciences). The substrate in ethanol was dried under an N 2 stream and was dispersed in water by sonication. Each reaction mixture (total volume 250 l) consisted of the appropriate amounts of the required samples, 100 mM Tris-HCl (pH 7.4), 4 mM CaCl 2 , and 2 M substrate. After incubation for 30 min at 37°C, [ 14 C]linoleic acid was extracted by the Dole method, and the radioactivity was quantified by liquid scintillation counter, as described previously (4 -6).
Confocal Laser Microscopy-Cells grown to subconfluence on glass bottom dishes (Matsunami) pre-coated with 5 g/ml fibronectin (Sigma) were fixed with 3% paraformaldehyde for 1 h in TBS. After three washes with TBS, the fixed cells were treated with 1% (w/v) bovine serum albumin and 0.5% (w/v) saponin in TBS (blocking solution) for 1 h, with anti-sPLA 2 -III antibody at 1:200 dilution in blocking solution for 2 h, and then with fluorescein isothiocyanate-conjugated anti-rabbit IgG antibody at a 1:200 dilution in blocking solution for 2 h, with three washes in each interval. After six washes with TBS, specific immunofluorescent signals were visualized with a laser-scanning confocal microscope (IX70; Olympus).
Experiments with sPLA 2 -III siRNA-A set of synthetic hairpin-forming oligonucleotides directed to human sPLA 2 -III was prepared: 5Ј-CA-CCGCTATGGCATCCGAAACTACTTCAAGAGAGTAGTTTCGGATG-CCATAG-3Ј (sense) and 5Ј-AAAACTATGGCATCCGAAACTACTCTCT-TGAAGTAGTTTCGGATGCCATAGC-3Ј (antisense). After annealing, the oligonucleotides were subcloned into pENTER TM /U6 vector (Invitrogen) using BLOCK-iT TM U6 RNA interference entry vector kit (Invitrogen). After transformation into Top10 competent cells and plasmid purification, the insert was transferred to the pAd/BLOCK-iT TM -DEST vector (Invitrogen) by means of the Gateway system using LR clonase. After transformation into Top10 competent cells and plasmid purification, the plasmid was linearized with PacI and transfected into 293A cells with Lipofectamine 2000 to produce adenovirus bearing sPLA 2 -III siRNA. After 2-4 times amplification, the resulting adenovirus-containing media were used for subsequent analyses.
Experiments with Nude Mice-HCA-7 cells (10 6 cells) were suspended in 100 l of PBS and injected subcutaneously into BALB/cnu/nu mice (6-week-old males) (Crea, Japan). After 3 weeks, the solid tumors were removed surgically and fixed in 10% (v/v) formalin. After embedding in paraffin, thin sections (4 -6 m thickness) of the tumor tissues were prepared on glass slides for histochemical analyses.

sPLA 2 -III Is Capable of Augmenting PGE 2 Production in Various Cell
Types-Although our previous study (4) demonstrated the AA-releasing and PGE 2 -synthetic actions of human sPLA 2 -III in HEK293 transfectants, it remained unclear whether this observation is a cell type-specific event or would be applicable to other cell types. To address this issue, we prepared adenoviruses bearing cDNAs for the full-length (FL) human sPLA 2 -III and for its truncated mutants (N ϩ S, S ϩ C, and S), and we infected them into various mammalian cell types (including primary and transformed cells) to investigate their effects on PGE 2 production. Examples of these analyses are shown in Fig. 1, in which adenoviral overexpression of sPLA 2 -III-FL or -S resulted in marked increases in IL-1␤stimulated PGE 2 production in normal human pulmonary fibroblasts (NHPF) and vascular smooth muscle cells (NHPVSMC), human colorectal (HCA-7) and prostate (PC3) cancer cells, and mouse Leydig cells (I-10). Appropriate expression of the enzyme in each cell type was verified by RNA blotting (Fig. 1, top panels). As exemplified in I-10 cells, inducible expression of sPLA 2 -III after sPLA 2 -III-adenovirus infection was correlated with ongoing PGE 2 synthesis (Fig. 1).
N-Glycosylation of sPLA 2 -III-To detect sPLA 2 -III protein expressed in cells, we prepared an antibody against a synthetic peptide corresponding to the hydrophilic region near the Cterminal portion of the S domain of human sPLA 2 -III ( Fig. 2A). Note that the production of PGE 2 was negligible in the absence of IL-1␤ stimulation in NHPF, NHPVSMC, PC3, and I-10 cells, because COX-2 was not expressed without this cytokine treatment (7,8). HCA-7 cells, which expressed COX-2 constitutively, produced PGE 2 with or without IL-1␤ (10).
To verify the specificity of the antibody, we initially used NHPF, which do not express sPLA 2 -III endogenously (see below) and are therefore useful to evaluate the expression and dynamics of transfected sPLA 2 -III protein. Thus, lysates of NHPF that had been infected with adenovirus for III-S or control (LacZ) were subjected to immunoblotting with the anti-sPLA 2 -III antibody. Although the predicted size of III-S protein is ϳ16 kDa (3), our immunoblot revealed an intense 28-kDa band as well as a minor 16-kDa band in cells infected with III-S, but not with mock, adenovirus (Fig. 2B). Because there are two predicted N-glycosylation sites that are conserved in both human and mouse sPLA 2 -III proteins (Asn-167 and Asn-280 in human sPLA 2 -III and Asn-163 and Asn-276 in mouse sPLA 2 -III; Fig. 2A), we anticipated that the 28-kDa species would represent an N-glycosylated form of III-S. Indeed, when the III-S adenovirus-infected cells were cultured in the presence of tunicamycin, an N-glycosylation inhibitor, there was a molecular mass shift of the main band from 28 to 16 kDa (Fig.  2B). Furthermore, when the lysates of III-S-transfected cells were incubated with endoglycosidase H (Fig. 2C) or peptide N-glycosidase F (Fig. 2D), the 28-kDa band was again shifted to the 16-kDa band. These results indicate that III-S is expressed mainly as an N-glycosylated protein in NHPF. An additional 25-kDa band detected in the lysate of III-S-transfected cells after treatment with intermediate doses of peptide N-glycosidase F (Fig. 2D) may be a partially glycosylated form of III-S.
To determine which consensus N-glycosylation sites in III-S were glycosylated, adenoviruses containing N-glycosylation site mutants of III-S, in which either or both Asn residues were replaced with Ser residues (designated as III-S-N1 (N167S), -N2 (N280S), and -N1N2 (N167S/N280S) mutants), were constructed and infected into NHPF. Under the condition where the expression levels of III-S-WT, -N1, -N2, and -N1N2 mRNAs were comparable (Fig. 3A), III-S-N1 and -N2 mutant proteins provided 22-and 25-kDa bands, respectively, relative to 28-kDa III-S-WT protein (Fig. 3B). This suggests that both the N1 and N2 sites are glycosylated and that a longer sugar chain is attached to the N1 site than to the N2 site. Most interestingly, the expression level of the III-S-N1 mutant protein was far lower than those of the III-S-WT and -N2 proteins (Fig. 3B). Moreover, the protein harboring mutations at both sites (N1N2) was barely detectable (Fig. 3B), even when 10 times more III-S-N1N2 adenovirus was added to the cells (data not shown). These results suggest that mutation at the N-glycosylation sites, particularly at the N1 site, influences the synthesis or stability of III-S protein in NHPF.
To assess if the N-glycosylation event would affect the enzymatic activity of III-S, lysates of NHPF that had been infected with adenovirus for III-S-WT, -N1, or -N2 (10 times more III-S-N1 adenovirus was added to the cells so that III-S-WT, -N1, and -N2 proteins were expressed almost equally, as assessed by immunoblotting (Fig. 3C)) were taken for PLA 2 enzyme assay. As shown in Fig. 3D, PLA 2 activities in the lysates of cells expressing III-S-WT, -N1, and -N2 were similar, indicating that the N-glycosylation does not influence the catalytic activity of III-S. However, we found that the PLA 2 activity secreted into the culture supernatant was significantly lower for the mutants than for III-S-WT; the activities of III-S-N1 and -N2 were about only 35 and 75%, respectively, of the activity of III-S-WT (Fig. 3E). This suggests that N-glycosylation, particularly at the N1 site, is important for proper secretion of III-S protein from the cells. PGE 2 synthesis in NHPF adenovirally transfected with III-S-WT, -N1, or -N2 is shown in Fig. 3F. Most interestingly, the PGE 2 biosynthetic abilities of III-S-WT, -N1, and -N2 (Fig. 3F) were correlated with their enzymatic activities in the culture supernatants (Fig. 3E) rather than with those remaining in the cells (Fig. 3D). These results suggest that, in this setting, III-S acts on cellular membranes to release AA mainly after secretion.
Collectively, our present results indicate that III-S can be N-glycosylated at two consensus sites and that this post-translational event affects the synthesis/stability and secretion of III-S protein in NHPF. However, subsequent studies (see below) revealed that sPLA 2 -III can exist as a nonglycosylated form in particular cell types that intrinsically express this enzyme. Therefore, the physiological and functional relevance of the N-glycosylation of sPLA 2 -III observed in NHPF (Figs. 2 and 3) and in several other cell types (data not shown) needs further elucidation. In addition, although there are other potential N-glycosylation sites in the C domain of human sPLA 2 -III (3), we did not focus on these sites in this study because these sites are not conserved between human and mouse enzymes.
Proteolytic Processing of sPLA 2 -III-When NHPF were infected with adenovirus for III-N ϩ S and their lysates were then taken for immunoblot analysis with anti-sPLA 2 -III antibody, III-N ϩ S protein (of which predicted size is 34 kDa (4)) appeared as multiple bands with molecular masses of ϳ34 and 45 kDa as well as 40 kDa at a higher adenovirus dose (Fig. 4A). Given that the S domain undergoes N-glycosylation (as noted above), the 40-and 45-kDa species may represent partially and fully N-glycosylated forms of III-N ϩ S, respectively. In cells adenovirally transfected with III-S ϩ C (of which the predicted size is 42 kDa (4)), a 55-kDa band was apparent at a higher adenovirus dose (Fig. 4A), suggesting that III-S ϩ C protein is also N-glycosylated. Indeed, treatment of the cells with tunicamycin resulted in the shift of the bands for III-N ϩ S and -S ϩ C to their predicted molecular sizes (data not shown). Of interest, in cells expressing III-S ϩ C, and to a much lesser extent III-N ϩ S, an additional 28-kDa band, which co-migrated with the N-glycosylated form of III-S, was evident (Fig. 4A). In particular, III-S ϩ C was almost entirely converted to the 28-kDa form when the cells were infected with a lower dose of III-S ϩ C adenovirus. These results suggest that both the N and C domains can be cleaved leading to the production of III-S (or a closely related form) and that in NHPF the peptide bond between the S and C domains is more susceptible to this processing than is that between the N and S domains.
When immunoblotting with anti-sPLA 2 -III antibody was performed on parental (data not shown) or mock adenovirusinfected (Fig. 4B) BEAS-2B cells, a human bronchial epithelial cell line that expressed endogenous sPLA 2 -III mRNA (4), a 16-kDa band (likely corresponding to endogenous III-S protein) was faintly detected. The expression level of this band was unaltered after stimulation of the cells with IL-1␤ (data not shown). Following the infection of BEAS-2B cells with III-S adenovirus, the expression of the 16-kDa protein (exactly the same size as that detected in control cells) was markedly increased, with only a minor fraction expressed as a 28-kDa N-glycosylated form (Fig. 4B). In cells infected with adenovirus for III-N ϩ S or -S ϩ C, a major 16-kDa band and a faint 28-kDa band were again evident, whereas no immunoreactive bands with molecular masses of Ͼ35 kDa were detected. Neither the 16-nor the 28-kDa band was detected by control antibody. These results suggested that both the N and C domains were entirely removed by certain processing proteases leading to the formation of III-S, the majority of which exists as a form with no or few sugar moieties, in BEAS-2B cells.
Immunohistochemistry-To determine which cell types intrinsically express sPLA 2 -III in human tissues, we performed immunohistochemistry with anti-sPLA 2 -III antibody on several tissues. In normal human lungs, sPLA 2 -III immunoreactivity was localized in bronchial epithelial cells (Fig. 5A, panels  a and b), whereas it was scarcely detected in the alveolar epithelium, arterial walls, and interstitial fibroblasts. This distribution is consistent with the expression of sPLA 2 -III in bronchial epithelial cells (BEAS-2B) but not in pulmonary fibroblasts (NHPF) in culture (see above). Staining of sPLA 2 -III was also associated with alveolar macrophages (Fig. 5A, panel c), which were also positively stained in serial sections with an antibody against HAM56, a human alveolar macrophage marker (Fig. 5A, panel d).
Staining of sPLA 2 -III in endothelial cells, but not other cell types, was also observed in joints of patients with osteoarthritis (Fig. 5B, panel g).
In human myocardial tissues with acute infarction, vascular endothelial cells adjacent to cardiomyocytes (Fig. 5C, panels a and c) and those in lesions with granulation (Fig. 5C, panel b) showed sPLA 2 -III staining, whereas staining in cardiomyocytes was scarce. Distribution of sPLA 2 -III in endothelial cells (Fig. 5C, panel c) was verified by staining of serial sections with anti-CD31 antibody (Fig. 5C, panel d). In normal human hearts, staining of sPLA 2 -III in endothelial cells was poor, although a weak signal was found in some vessels (Fig. 5C, panel e). In comparison, sPLA 2 -IIA (Fig. 5C, panel f) and sPLA 2 -V (Fig. 5C, panel g) showed intense staining in cardiomyocytes that were devoid of nuclei, implying that these two enzymes are associated with severely damaged cardiomyocytes. Vascular smooth muscle cells in the coronary arteries were positively stained for sPLA 2 -IIA (Fig. 5C, panel d) but not for sPLA 2 -V (Fig. 5C, panel e). There was no appreciable staining of sPLA 2 -IIA and -V in endothelial cells. Control antibody did not provide significant signals in the tissues tested above (data not shown).
We next examined the localization of sPLA 2 -III in the human uterus (Fig. 6A), breast (Fig. 6B), and colon (Fig. 6C) cancer FIG. 5. Immunohistochemistry of sPLA 2 -III and other sPLA 2 s in human lung, RA joints, and ischemic hearts. A, localization of sPLA 2 -III in human lungs. Intense sPLA 2 -III immunoreactivity was located in the bronchial epithelium (dark arrowheads) (panels a and b). Alveolar macrophages (red arrowheads) were also stained with anti-sPLA 2 -III antibody (panel c) as well as with anti-HAM56 antibody (panel d). B, localization of sPLA 2 -III in human RA joints. Immunoreactivity of sPLA 2 -III was strictly confined to the microvascular endothelium (red arrows) in the interstitium (panel a). tissues. Although no obvious staining of sPLA 2 -III was found in the normal uterus (Fig. 6A, panel a), it was clearly detected in uterus cancer cells and the adjacent microvascular endothelium (Fig. 6A, panels b and c). Similar distribution of sPLA 2 -III was observed in breast (Fig. 6B, panels a and b) and colon (Fig.  6C, panels b and c) cancer tissues, in which the enzyme was located in tumor cells and neighboring microvascular endothelial cells but not in normal glandular tissues (Fig. 6C, panel a). Control antibody did not provide positive staining in the uterus (Fig. 6A, panel e), breast (Fig. 6B, panel f), and colon (Fig. 6C,  panel d) cancer tissues.
The localizations of other sPLA 2 s in these cancer tissues were also examined. In uterus cancer, staining of sPLA 2 -V in interstitial fibroblasts, but not in tumor cells, was obvious (Fig.  6A, panel d). In breast cancer, the expression of sPLA 2 -IIA (Fig.  6B, panel c) and -IID (Fig. 6B, panel d), but not other sPLA 2 s (as exemplified by sPLA 2 -IIE; Fig. 6, panel e), was associated with tumor cells. sPLA 2 -IIA was expressed widely in normal colorectal glandular cells (Fig. 6D, panel a), although it was poorly stained in colorectal adenocarcinoma cells (Fig. 6D,  panel b). This is in line with a report that sPLA 2 -IIA expression shows an inverse relationship with the states of human gastric cancer (14). sPLA 2 -IID was detected in normal glandular cells with preferential distribution in the surface mucosa (Fig. 6D, panel c) and was also intensely expressed in colorectal adenocarcinoma cells (Fig. 6D, panel d). This is consistent with our previous observation that the expression of this enzyme is detected in various human colorectal cancer cell lines (15). sPLA 2 -V was not detected in normal colorectal glands (Fig. 6D, panel e), and its scattered staining was found in some populations of cancer cells (Fig. 6D, panel f). sPLA 2 -X was distributed in the lamina propria beneath the normal glandular cells (Fig.  6D, panel g). Although one report has shown the association of sPLA 2 -X immunoreactivity with colorectal cancer cells (16), in our analysis it was preferentially localized in the interstitium adjacent to cancer cells, and the staining of adenocarcinoma cells was scarce (Fig. 6D, panel h). Taken together, individual sPLA 2 s exhibited unique distributions in each tissue, and the localization of sPLA 2 -III in the microvascular endothelium in inflamed, injured, and cancer tissues was unique among the sPLA 2 s.
Expression of sPLA 2 -III in Microvascular Endothelial Cells-The results from the immunohistochemical analyses (Figs. 5 and 6) prompted us to examine the expression of sPLA 2 -III in cultured primary HMVEC. RT-PCR revealed that the expression of sPLA 2 -III transcript was markedly elevated in HMVEC after stimulation with IL-1␤ or TNF␣ over 3-24 h (Fig. 7A). This induction of sPLA 2 -III mRNA was accompanied by increased expression of its protein, as assessed by immunocytostaining with anti-sPLA 2 -III antibody (Fig. 7B). Although the sPLA 2 -III signal was faint in unstimulated cells, it was markedly induced in the endoplasmic reticulum and perinuclear Golgi (as judged by the reticular staining pattern and the secreted property of sPLA 2 -III) in cells stimulated for 6 -24 h with IL-1␤. Control antibody did not provide fluorescent signals in IL-1␤-stimulated cells (Fig. 7B).
Immunoblotting of IL-1␤-stimulated HMVEC with anti-sPLA 2 -III, but not control, antibody revealed a single 53-kDa immunoreactive band (Fig. 7C). The size of this band was similar to that of III-N ϩ S adenovirally overexpressed in HMVEC. Adenoviral transfection of III-S ϩ C into these cells provided doublet bands around 60 kDa, and transfection of III-S resulted in the expression of a 28-kDa protein (Fig. 7C), implying that they undergo N-glycosylation. Although these observations suggest that endogenous sPLA 2 -III may be expressed as III-N ϩ S (or a closely related form) with N-glycosylation, rather than III-S, in HMVEC, its precise entity needs further investigation. Infection of III-N ϩ S adenovirus into HMVEC in the presence of IL-1␤ led to increased PGI 2 synthesis (Fig. 7D), which was correlated kinetically with increased expression of III-N ϩ S and IL-1␤-dependent induction of endogenous COX-2 (Fig. 7D, top panel). sPLA 2 -III Promotes Cancer Cell Growth-Because sPLA 2 -III expression is associated with tumor cells in various cancer tissues (Fig. 6), we next examined the expression and possible functions of this enzyme in the human colorectal cancer cell line HCA-7. Immunoblotting with anti-sPLA 2 -III antibody showed the expression of endogenous sPLA 2 -III protein, which appeared mainly as a 16-kDa nonglycosylated III-S form, in HCA-7 cells (Fig. 8A). The expression of this protein was unaltered when the cells were stimulated with IL-1␤ (data not shown). When III-FL cDNA was stably transfected into HCA-7 cells by lentivirus-mediated gene transfer, there was a marked increase in the 16-kDa III-S protein band, as well as in minor N-glycosylated species with molecular masses of ϳ28 -34 kDa, FIG. 7. Expression and function of sPLA 2 -III in primary human microvascular endothelial cells. A, HMVEC were stimulated for the indicated periods with 1 ng/ml IL-1␤ or TNF␣, and the expression of transcripts for sPLA 2 -III and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was assessed by RT-PCR. B, HMVEC grown on collagen-coated dishes were stimulated for the indicated periods with IL-1␤. Then the cells were fixed, permeabilized, and immunostained with anti-sPLA 2 -III or control antibody. C, HMVEC were infected with adenovirus for III-S, III-N ϩ S, III-S ϩ C, or LacZ for 36 h and then stimulated for 24 h with IL-1␤. The cell lysates were subjected to immunoblotting with anti-sPLA 2 -III or control antibody (Ab). Positions of III-S, III-N ϩ S, and III-S ϩ C, which are N-glycosylated (glyco), as well as that of endogenous sPLA 2 -III are indicated by arrows. D, HMVEC were infected for the indicated periods with III-N ϩ S or LacZ adenovirus (multiplicity of infection ϭ 10) in the presence of IL-1␤. Then the supernatants were taken for enzyme immunoassay for 6-keto-PGF 1␣ (a stable end product of PGI 2 ) (bottom panel), and the cells were subjected to RNA blotting with sPLA 2 -III cDNA as a probe and to immunoblotting with anti-COX-2 antibody (top panel). No induction of COX-2 and therefore no production of 6-keto-PGF 1␣ were observed in these cells in the absence of IL-1␤. Representative results of three independent experiments are shown.
whereas no larger immunoreactive bands corresponding to the III-FL, -N ϩ S, and -S ϩ C forms were detectable (Fig. 8B).
These results indicate that III-FL is entirely processed to III-S in HCA-7 cells.
When these cells were prelabeled with [ 3 H]AA for 24 h and then cultured for an additional 4 h, more [ 3 H]AA was released from sPLA 2 -III-transfected cells than from control cells in an FCS concentration-dependent manner (Fig. 8C). Accordingly, PGE 2 generation was also markedly increased in sPLA 2 -IIItransfected cells as compared with that in replicate control cells (Fig. 8C). It is known that PGE 2 facilitates the proliferation of several colorectal cancer cells, including HCA-7 cells (10). In fact, the proliferation of sPLA 2 -III-transfected HCA-7 cells was apparently faster than that of control cells, as revealed by their photographs in culture (Fig. 8E) and by the MTT cell growth assay (Fig. 8F). Consistent with this, the expression of c-myc and rhoA, which is up-regulated in microsomal PGE 2 synthase-transfected HCA-7 cells (10), was markedly increased in sPLA 2 -III-transfected cells relative to that in control cells (Fig. 8G).
To assess whether these PGE 2 -biosynthetic and growth-promoting actions of sPLA 2 -III depend on its catalytic activity, the catalytically inactive III-S mutant III-HQ, in which the catalytic center His was replaced with Gln, was lentivirally transfected into HCA-7 cells. Appropriate expression of III-HQ as a 16-kDa protein was verified by immunoblotting (Fig. 8H, inset). No increases in PGE 2 synthesis (data not shown) and cell proliferation (Fig. 8H) were observed in III-HQ-transfected cells as compared with those in control cells, implying an absolute requirement for catalytic activity for these events.
When control and sPLA 2 -III-transfected HCA-7 cells were subcutaneously injected into athymic nude mice, the latter formed larger solid tumors over 3 weeks than did the former (Fig. 9, A and B). In contrast, growth of III-HQ-expressing HCA-7 cells implanted into two nude mice was similar to that of control cells (data not shown). Histopathologic examination of a fraction of whole tumor tissues from the xenografts revealed that the tumor derived from sPLA 2 -III-transfected cells appeared as a nodular mass and was well demarcated (Fig. 9C, right panel). The tumor cells were proliferating with scanty interstitium that contained a number of capillary vessels and had scarce cytoplasm and hyperchromatic nuclei of various sizes with sporadic mitosis. Some tumor cells were proliferating with tubular formation, indicating adenocarcinoma differentiation. The capillaries consisted of swollen endothelial cells and were likely to be newly formed vessels. As compared with the sPLA 2 -III-transfected HCA-7 xenografts (Fig. 9C, right  panel), in the control HCA-7 xenografts the capillaries were scarce and were of smaller size (Fig. 9C, left panel). In addition, there were lesions with features of necrosis in the control xenografts (Fig. 9C, left panel), probably because of poor angiogenesis. Finally, we asked if endogenous sPLA 2 -III contributes to PGE 2 production and thereby cell proliferation in HCA-7 cells. To this end, adenovirus harboring sPLA 2 -III-directed siRNA was infected into HCA-7 cells, and accumulation of PGE 2 in the culture medium and cell proliferation were investigated. As shown in Fig. 10A, the expression of endogenous sPLA 2 -III protein was considerably decreased in cells infected with sPLA 2 -III siRNA adenovirus as compared with those infected with control (LacZ or random oligo) adenovirus. In this setting, there were partial but significant reductions in PGE 2 produc- tion and cell proliferation in sPLA 2 -III siRNA-transfected cells relative to those in control cells (Fig. 10B), suggesting the partial involvement of endogenous sPLA 2 -III in these events. DISCUSSION In the present study, we reported the following novel aspects of sPLA 2 -III, an sPLA 2 isozyme that has unique structural properties among the mammalian sPLA 2 s. (i) sPLA 2 -III is capable of augmenting the PG-biosynthetic response in various cell types, and the S domain alone is sufficient for this function.
(ii) sPLA 2 -III is often N-glycosylated at two consensus sites in the S domain, and mutations in these sites decrease the synthesis/stability and secretion of its protein. (iii) sPLA 2 -III is proteolytically processed to the S domain-only form according to cell type. (iv) In human pathologic tissues, sPLA 2 -III is preferentially expressed in microvascular endothelial cells as well as in tumor cells. (v) sPLA 2 -III has the ability to facilitate the growth of human colorectal cancer cells both in vitro and in vivo, the latter being associated with increased angiogenesis.
PG Biosynthetic Action-We have shown that sPLA 2 -III is capable of augmenting PG production in various primary and transformed cells (Figs. 1 and 7D). When the PG biosynthetic capacity of sPLA 2 -III was compared with that of other sPLA 2 s, increased PG synthesis was observed, in general, in the order sPLA 2 -X Ͼ -V Ͼ -IIF Ն -III Ͼ Ͼ IIA in many cell types tested (data not shown). This order appears to be correlated with the ability of these sPLA 2 s to interact with the PC-rich outer plasma membrane (4 -6, 17-21). Bee venom group III sPLA 2 contains a membrane-binding surface composed mainly of hydrophobic residues and two basic residues that come into close contact with the membrane, and interfacial binding of this enzyme to PC-containing membrane vesicles occurs predominantly by a nonelectrostatic mechanism (22). Because of the structural similarity between human sPLA 2 -III (S domain) and bee venom sPLA 2 , the human enzyme may also interact with the PC-rich plasma membrane surface in a similar way. Although a recent report (23) has proposed that sPLA 2 -IIA and -X release AA prior to secretion in HEK293 cells, this does not seem to be the case for human sPLA 2 -III (at least in NHPF) because the PGE 2 synthetic capacity of sPLA 2 -III mutants showed good correlation with their enzymatic activities released from the cells rather than those remaining in the cells (Fig. 3).
Post-translational Modification-We have provided evidence that human sPLA 2 -III is processed to the S domain-only form in a cell type-related fashion. Thus, the linkage between the S and C domains of sPLA 2 -III is more susceptible to cleavage than that between the N and S domains in NHPF (Fig. 4A), whereas the enzyme is converted almost completely to the S domain-only form in BEAS-2B (Fig. 4B) and HCA-7 (Fig. 8, A  and B). This suggests that bronchial epithelial cells and colorectal cancer cells, which intrinsically express sPLA 2 -III, may contain certain processing proteases that can sufficiently cleave the N-S and S-C domain linkages. In any case, it will be essential to determine the precise cleavage sites on sPLA 2 -III in order to fully understand the mechanism for the unique proteolytic processing of this enzyme. The presence of a basic doublet KR at the end of the N domain of sPLA 2 -III suggests that this part can be cleaved by subtilisin-like protein convertase in the Golgi (3). Unlike sPLA 2 -IB and -X, which are activated only after proteolytic removal of the N-terminal propeptide (17,24), the N and C domains do not influence the catalytic activity of sPLA 2 -III (4). Therefore, by assuming that sPLA 2 -III is indeed expressed as a III-N ϩ S or closely related form in endothelial cells (Fig. 7C), the N domain may be associated with some endothelial cell-specific functions of this enzyme. For instance, the N domain might be important for spatiotemporal sorting of the enzyme to its proper target membrane compartments, as has been proposed in HEK293 cells (4).
We have also found that the S domain of sPLA 2 -III transfected into NHPF (Fig. 2) and several other cell types (data not shown) undergo N-glycosylation at two consensus sites, and this post-translational modification is required for the optimal synthesis or stability and secretion, but not the catalytic activity, of the sPLA 2 -III protein (Figs. 2 and 3). In relation to this, we have recently shown that the secretion of human sPLA 2 -X can be affected by its N-glycosylation in certain if not all cell types (25). Paradoxically, however, the majority of sPLA 2 -III exists as a nonglycosylated (or very poorly glycosylated) III-S form in BEAS-2B (Fig. 4B) and HCA-7 (Fig. 8, A and B). Although the reason for this discrepancy is presently unclear, a likely explanation is that there may be a certain interacting factor(s) that stabilizes the nonglycosylated III-S protein in these cells or that a particular machinery that leads to destabilization of the nonglycosylated III-S is lacking from these cells.
Expression in Microvascular Endothelial Cells-sPLA 2 -III is preferentially distributed in microvascular endothelial cells in inflamed joints, ischemic hearts, and various types of cancer (Figs. 5 and 6). The expression of sPLA 2 -III in cultured HMVEC is induced after stimulation with the proinflammatory cytokines (Fig. 7, A and B), whereas it was unaffected by these cytokines in BEAS-2B and HCA-7 cells (4), suggesting that cytokine induction of sPLA 2 -III is a cell type-specific event. The fact that microvascular endothelial cells adjacent to tumor cells, but not those in normal tissues, showed intense sPLA 2 -III immunoreactivity (Fig. 6) raises the intriguing possibility that the expression of this enzyme in endothelial cells might be related to cancer-associated angiogenesis. As the vasoactive prostanoids have angiogenic potential (26 -31), it can be anticipated that the increased production of these lipid mediators by sPLA 2 -III would contribute to endothelial cell migration and proliferation, thereby leading to increased angiogenesis in inflamed and cancer tissues.
Expression and Potential Function in Tumor Cells-Our results argue that sPLA 2 -III aberrantly expressed in tumor cells may have the potential to facilitate cancer development in a manner dependent on its catalytic activity (Figs. 8 -10). The role of PGE 2 in the development of cancer by promoting tumor cell growth and angiogenesis has been demonstrated by studies employing mice deficient in cytosolic PLA 2 ␣ (32), COX-2 (33), and several PGE receptors (34,35). Expression of COX-2 and microsomal PGE 2 synthase is elevated in many, if not all, types of cancer cells (10), and overexpression of these enzymes promotes cancer development (36,37). In addition, lysophosphatidylcholine and lysophosphatidic acid, as well as AA, can promote the migration of endothelial cells, and exposure of these cells to exogenous sPLA 2 s, such as bee venom sPLA 2 , elicits the same response in a manner partially dependent upon their catalytic activity (38). In this context, sPLA 2 -III expressed in tumor cells and adjacent microvascular endothelial cells can also be involved in this pathway leading to cancer development through augmenting the release of AA/PGE 2 and lysophospholipids. Thus, our observations provide the basis for future studies that will evaluate whether substances that specifically inhibit the activity of sPLA 2 -III are applicable to prophylaxis and/or treatment of cancers.
Future Prospects-sPLA 2 -III is also expressed in bronchial epithelial cells and macrophages in the human respiratory tract (Fig. 5A). Thus, sPLA 2 -III, as well as sPLA 2 -V and -X that are expressed in these cells (7), may participate in lung pathology by enhancing eicosanoid synthesis or by facilitating other events such as surfactant hydrolysis (39,40) and antimicrobial defense (41). A recent study (42) has shown that both bee venom sPLA 2 and human sPLA 2 -III induce dendritic cell maturation in culture. Given that the sPLA 2 -III transcript is also detected in human brain (3), sPLA 2 -III might also contribute to certain neuronal processes. Indeed, bee venom sPLA 2 can elicit neuronal actions through catalytic activity-dependent (43) and -independent (receptor-mediated) (44 -46) fashions. Studies using sPLA 2 -III gene-manipulated mice will help to clarify the physiological relevance of the present observations and speculations.