Hypoxic tumors in vivo and many cell lines in vitro secrete bFGF and VEGF. Both bFGF and VEGF are components of the tumor microenvironment capable of activating ECs. To evaluate the potential role of LPP3 in angiogenesis, we investigated the effects of treatment of HUVECs with VEGF and bFGF. We stimulated monolayer HUVECs with either VEGF¹⁶⁵ or bFGF for various time periods between 0 and 18 h, and subjected lysates to Western blot analyses using an affinity purified anti-LPP3-cyto antibody (Fig. 1A). The expression of LPP3 protein levels was increased by >3-fold in response to VEGF¹⁶⁵ treatment (100 ng/ml) for 6 or 12 h (relative to control levels), whereas bFGF (20 ng/ml) had a significantly less robust effect on LPP3 levels during the same treatment duration (Fig. 1A). The concentrations of VEGF¹⁶⁵ and bFGF used in this experiment were optimal as evidenced by the observation that both factors activated extracellular-signal-regulated kinase (Erkl/2) in these cells in an independent experiment (data not shown). We observed that anti-LPP3-c-cyto antibody detects three major polypeptides, two of which (~52 and ~46 kDa) are slow and one (~36 kDa) that exhibits high mobility (Fig. 1A). Since, LPP3 contains a single consensus N-glycosylation site (170N) on the proposed 2^nd extracellular loop of the LPP3 protein, the high mobility anti-LPP3-c-cyto immunoreactive species is likely to be a post-translationally modified form. This idea is actually supported by our data with N-glycanase described elsewhere. As a positive control for the cytokines used, the membrane was stripped and re-blotted with an anti-proliferating cell nuclear antigen (PCNA) antibody (Fig. 1B). As expected, both cytokines increased PCNA expression to optimal levels. The level of Focal adhesion kinase (Fak) protein was measured to confirm equivalent protein loading (Fig. 1C).

Next, to determine the ability of anti-LPP3-c-cyto antibody to immunoprecipitate LLP3 antigen, we used ECs and human erythroleukemia (K562) cells (Fig. 1D). K562 cells that do not express LPP3 protein was included as a negative control. K562 cells were left unstimulated (Fig. 1D, lane 1), while ECs were stimulated with VEGF (Fig. 1D, lane 2) and bFGF (Fig. 1D, lane 3), and subsequently subjected to cell surface biotinylation and immunoprecipitation with an anti-LPP3-c-cyto antibody and analyzed by ligand blotting with Streptavidin-HRP (Fig. 1D). We found that the anti-LPP3-c-cyto did not immunoprecipitate 36–52 kDa polypeptides from K562 cells (negative control cell line); in contrast, ligand blotting with streptavidin-HRP detected three major polypeptides (36, 42 and 52 kDa, indicated by arrows) (Fig. 1D). These data suggest that the LPP3 antigen is exposed on the extracellular surface of ECs and can be immunoprecipitated by an anti-LPP3-c-cyto antibody.

LPP3 contains a single N-glycosylation site (170N, accession number O14495) 23. Several pilot experiments suggested that, depending upon how cells are cultured, the variation in the extent of LPP3 glycosylation can be complex and dramatic. To examine this possibility, we prepared cell extracts from ECs and subjected lysates to immunoprecipitation as indicated (Fig. 1E). After several washes with cell lysis buffer, immunoprecipitates were incubated with N-glycanase F (PNGaseF), an enzyme that cleaves the carbohydrate moiety. In doing so, we observed that most of the slow mobility (smeared) polypeptides disappeared, leaving behind a ~36 kDa unprocessed polypeptide (Fig. 1E). Consistent with previous reports, we observed that LPP3 is N-glycosylated 3031, and the extent of N-glycosylation appears to be cell type- and culture condition-dependent.

Anti-LPP3-RGD antibody was raised by injecting rabbit with a synthetic peptide modeled after the proposed 2^nd extracellular loop of LPP3 protein (EGYIQNYRCRGDDSKVQEAR) 23. Previously we relied on the specificity of the anti-LPP3-RGD antibody to analyze tumor sections and inhibit LPP3-mediated cell-cell interactions 23. In the current study, we determined the ability of anti-LPP3-RGD to detect native LPP3 antigen of intact ECs. Towards this end, ECs were either left unstimulated or were stimulated with VEGF, incubated with anti-LPP3-RGD and subjected to fluorescence activated cell sorting (FACS). In contrast to unstimulated ECs that do not express LPP3 protein significantly, the addition of VEGF induced cell surface expression of LPP3 protein of monolayer ECs (Fig. 2A). This result indicates that VEGF stimulates expression of LPP3 antigen on the surface of ECs, which can be detected by anti-LPP3-RGD antibody. This data also suggests that anti-LPP3-RGD antibody detects intact and native LPP3 antigen neoepitope expressed by activated ECs. A schematic diagram showing topological and structural organization of LPP3 is shown (Fig. 2B).

It is apparently clear that the tumor microenvironment contains bFGF and VEGF. Because bFGF and VEGF induce LPP3 expression in cultured ECs, we hypothesized that LPP3 protein might be similarly expressed by tumor-endothelium. To examine this hypothesis, serial angioma and hemangioma sections were subjected to immunostaining with anti-Flk-1, anti-PECAM-1 (CD31), and von Willebrand factor (vWF) to establish the presence of ECs and blood vessels as previously described 23. We have previously shown that quiescent skin blood vessels are negative for anti-LPP3-RGD immunoreactivity [please see reference 23 for online supplemental data]. Hypoxic tissues as well as inflamed tissues are known to express VEGF 14. Consistent with these reports, our data indicate that VEGF (green) is diffusedly distributed throughout angioma and hemangioma tissues, with significantly higher expression in the blood vessels (Fig. 3B,3E,3H). As shown in Fig. 3A,3D and 3G, LPP3 (red) protein appears to co-localize with VEGF in the angioma, including in the endothelium where the yellow ring like structure indicates colocalization; however, their molecular proximity remains unknown. Similarly, the expression of LPP3 protein was coincident with VEGF expression in the hemangioma section examined (Fig. 3G,3H,3I). These data demonstrate that both LPP3 and VEGF are diffusedly distributed within tumor vasculature, and LPP3 expression may not be exclusively restricted to ECs. Incubation of the antibodies with peptides that had been used to generate the primary antibody blocked immunoreactivity, confirming the specificity of antibodies used, as previously described 23. This data is consistent with our earlier observation that the LPP3 protein is highly expressed in tumor endothelium 23.

A considerable number of studies indicate that bFGF- and VEGF-mediated signaling, as well as cell adhesion events mediated by α₅β₁ and α_vβ₃ integrins, critically determine the outcome of angiogenesis. Because the LPP3-RGD protein acts as a cell-associated integrin ligand for α₅β₁ and α_vβ₃ integrins, we hypothesized that an anti-LPP3-RGD antibody could inhibit cell-cell interactions that may impede the ability of ECs to undergo capillary morphogenesis (also called tubulogenesis). To evaluate the capacity of anti-LPP3-RGD to block capillary morphogenesis of ECs, we employed early passage (between 3–4 total) ECs. Typically, we embed ECs between two layers of type I collagen matrices, and treat them with bFGF and VEGF in the presence of 20% serum. As previously described, the process of capillary formation by ECs in a three-dimensional type I collagen take place over a period of 24 to 72 h, and requires the addition of bFGF and VEGF 2324.

For the purpose of this study, capillaries formed after 24 h of culture in 3D collagen were considered "pre-formed vessels", whereas capillaries formed after more than 24 h of culture were considered "new capillaries". We used affinity purified rabbit anti-IgG (control) polyclonal antibodies (pAbs) and mouse anti-MHC class II (W6/32) monoclonal Abs (mAbs), and mouse anti-β₁ (4B4) and anti-α_vβ₃ (LM609) mAbs, as negative and positive controls, respectively.

In the absence of antibodies, we observed an increased number of capillaries formed from 36 to 72 h in culture (Fig. 4, filled black bar). The addition of an anti-MHC mAb (empty bar) and rabbit IgG (filled blue bar) resulted in minimal inhibition, and for the purpose of our study we consider these minimal inhibitions as baseline (Fig. 4). In contrast, the addition of the anti-β₁ (filled red bar) and anti-α_vβ₃ (filled green) mAbs caused regression of the pre-formed capillaries (Fig. 4). Anti-α_vβ₃ mAbs reduced the number of pre-formed capillaries by more than 50%, suggesting that other cell surface proteins, such as fibronectin-binding integrin α₅β₁ and collagen/laminin binding integrins may also mediate capillary morphogenesis. Indeed, anti-β₁ integrin subunit mAbs inhibited pre-formed interconnections and reduced the number of capillaries by ~60–70% (Fig. 4). No two mAbs were added simultaneously since it has been reported that α₅β₁ and α_vβ₃ integrin antibodies together induce complete collapse and regression of tubules in vitro 3738. Unexpectedly, the addition of the anti-LPP3-RGD antibody (filled yellow bar) had a dramatic effect on bFGF and VEGF-induced capillary formation (Fig. 4). The effect of anti-LPP3-RGD was comparable to anti-α_vβ₃ mAbs (Fig. 4, solid yellow bar). Representative cross sections of 3D gel are shown (Fig. 5). Arrows indicate lumen like structures. These data suggest that antibodies affecting cell adhesion, migration and cell-cell interaction events inhibit capillary morphogenesis by ECs in vitro. These data demonstrate that an affinity-purified rabbit anti-LPP3-RGD polyclonal antibody can inhibit bFGF- and VEGF-induced capillary formation in vitro.

Human umbilical vein endothelial cells (HUVECs) were purchased from Cambrex Bio Science Inc. (Walkersville, MD). Media 199, antibiotic solution, L-glutamine and all other cell culture reagents were purchased from InVitrogen Corp. (Carlsbad, CA). Recombinant human vascular endothelial growth factor (VEGF¹⁶⁵), basic fibroblast growth factor (bFGF), and anti-VEGF (MAB293) were purchased from R&D Systems Inc. (Minneapolis, MN); adult human serum-AB from Gemini Bioproducts (Woodland, CA); bovine skin-derived type I collagen from Cohesion Technologies, Inc. (Palo Alto, CA). Affinity-purified anti-α_vβ₃ integrin (LM609), anti-PCNA and anti-VE-cadherin (MAB1989) monoclonal antibodies (mAbs) were purchased from Chemicon International Inc. (Temecula, CA) and anti-KDR/Flk-1 (sc-6251) mAb from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Mouse anti-Fak monoclonal antibody was purchased from Upstate Biotechnology Inc. (Lake Placid, NY); anti-human β₁ integrin subunit (clone 4B4) mAbs from Beckman Coulter Inc. (Fullerton, CA); and anti-MHC class II (W6/32) from Sigma Chemical Com., (St. Louis, MO). The preparation of rabbit anti-LPP3-RGD (previously called anti-VCIP-RGD or anti-PAP2b-RGD) and anti-LPP3-c-cyto (previously called anti-VCIP-c-cyto or anti-PAP2b-c-cyto) polyclonal antibodies (pAbs) has been previously described 2324.

Monolayer ECs (2 × 10⁷) at passage 4 were stimulated with VEGF¹⁶⁵ and subjected to cell surface biotinylation as previously described 2324. Biotinylated ECs were solubilized in RIPA cell extraction buffer [50 mM HEPES, pH 7.5, 150 mM NaCl, 1.0% Triton X-100 (non-ionic), 0.25% SDS (anionic), 0.25% sodium deoxycholate (anionic), and 2 mM EDTA, to which appropriate concentrations of proteases were added prior to use]. Extracts were clarified by centrifugation at 21,000 × g for 45 min at 4°C. Lysates were pre-absorbed at 4°C for 2 h by incubating with sepharose beads conjugated to rabbit IgGs. For each immunoprecipitation, approximately 1.5 mg total protein was used. Immunoprecipitation was carried out for 3 hr at 4°C. Immunocomplexes were washed five times with RIPA cell extraction buffer. For deglycosylation, immunoprecipitates were denatured in 1.0% SDS for 20 min at 90°C and washed twice with deglycosylation reaction buffer (New England Biolab., Beverly, MA). The deglycosylation (PNGaseF) reaction was initiated in a reaction volume of 100 μl containing 0.5% NP40 detergent and 50 units of PNGaseF enzyme at 37°C for 3 h. Samples were boiled in Lammeli reducing sample buffer and resolved by 10% SDS-PAGE and transferred to a nitrocellulose (NC) membrane. The NC membrane was blocked with 5% milk and 1% BSA in 1× TBS, 0.1% Tween and analyzed by incubating with streptavidin conjugated to horse-radish peroxidase (HRP) at a 1:10000 dilution. For FACS, monolayer ECs were starved for 6 h, thereafter, either left unstimulated or stimulated with VEGF (100 ng/ml) for 6 h. Cells were then non-enzymatically detached and subjected to FACS analyses as previously described 36.

Monolayer EC culture was performed as previously described 17182324. The preparation of 3D collagen matrix has also been previously described 2324. Briefly, a viscous gel-like solution was prepared by mixing 7 ml of 3.0 mg/ml type I collagen solution with 1 ml of 10× Ml99 medium at 4°C, adjusting the pH to 7.5 with 0.1 N sodium hydroxide, adding 0.1 ml of 100× ITS, and adjusting to a final volume of 10 ml with sterile distilled water. The matrix was allowed to polymerize (solidify) for 30 min at 37°C. Next, unstarved proliferating ECs were gently resuspended (at 6 × 10⁵ cells/ml in complete media), seeded onto solidified gels, and the dishes (24 well Coster cell culture dishes) were returned to a CO₂ incubator for 2–3 h. At the end of 3 h, unattached cells were removed by gentle aspiration. Onto monolayer cells, a second layer of collagen gel was added and returned to a 37°C humidified CO₂ incubator. Following solidification (~3 h), the matrix was layered with Ml99 medium containing 20% adult human serum-AB, 4 mM L-glutamine, 1× ITS, bFGF (20 ng/ml) and 100 ng/ml VEGF¹⁶⁵. The old growth medium was removed and fresh medium was added every 24 h. Capillary formation was examined under a phase contrast microscope every 12 h.

ECs were embedded in 3D gels as described above, using media M199. At least 10 random 100× fields were examined to assess capillary formation (also called tubulogenesis). The formation of lumen-like structures was visible as early as 24 h. Antibodies were dialyzed in sterile dialysis buffer (25 mM Tris, 175 mM sodium chloride, 2 mM potassium chloride pH 7.4) overnight and passed through a 0.22 μM filter prior to use. The integrity of antibodies was determined by SDS-PAGE. To determine the effects of specific antibodies on pre-formed capillaries, the method of Bayless et al. was used 3738. Briefly, at the end of 24 h, mAbs (50 μg/ml) and pAbs (75 μg/ml) were added. Fresh aliquots of antibodies were added every 12 h for a total of 60 h. To quantify the degree of capillary formation, 3D matrices were fixed at 24, 36, 48, 60, and 72 h by aspirating the medium, washing with PBS, and then fixing with 4% glutaraldehyde in PBS, pH 7.4, overnight at 4°C. Matrices were then washed with distilled water and embedded in paraffin according to the manufacturer's instructions (Richard Allen Scientific, Kalamazoo, MI). Serial sections (4 μm) were prepared, dehydrated, stained with acidified eosin, and destained with distilled water. Capillaries were counted and photographed using a Zeiss Axiovert 25C microscope at 100× magnification. Each capillary tubule was surrounded by least 2 to 5 ECs. Capillary formation was defined as the induction of a minimum of 3 separate capillary events within a single field. At least 10 random fields were counted for each sample. Experiments were performed in duplicate, using triplicate wells in each case. Results were expressed as mean ± SEM.

Student's t tests and ANOVA were used to detect significant comparisons as previously described 23.