Decorin Is a Novel VEGFR-2-Binding Antagonist for the Human Extravillous Trophoblast

Abstract

Extravillous trophoblasts (EVT) of the human placenta invade the uterine decidua and its arteries to ensure successful placentation. We previously identified two decidua-derived molecules, TGF-β and a TGF-β-binding proteoglycan decorin (DCN), as negative regulators of EVT proliferation, migration, and invasiveness and reported that DCN acts via multiple tyrosine kinase receptors [epidermal growth factor-receptor (EGF-R), IGF receptor-1 (IGFR1), and vascular endothelial growth factor 2 receptor (VEGFR-2)]. Because binding of DCN to VEGFR-2 has never been reported earlier, present study explored this binding, the approximate location of VEGFR-2-binding site in DCN, and its functional role in our human first trimester EVT cell line HTR-8/SVneo. Based on far-Western blotting and coimmunoprecipitation studies, we report that DCN binds both native (EVT expressed) and recombinant VEGFR-2 and that this binding is abrogated with a VEGFR-2 blocking antibody, indicating an overlap between the ligand-binding and the DCN-binding domains of VEGFR-2. We determined that ¹²⁵I-labeled VEGF-E (a VEGFR-2 specific ligand) binds EVT with a dissociation constant (K_d) of 566 pM, and DCN displaced this binding with an inhibition constant (K_i) of 3.93–5.78 nM, indicating a 7- to 10-fold lower affinity of DCN for VEGFR-2. DCN peptide fragments derived from the leucine rich repeat 5 domain that blocked DCN-VEGFR-2 interactions or VEGF-E binding in EVT cells also blocked VEGF-A- and VEGF-E-induced EVT cell proliferation and migration, indicative of functional VEGFR-2-binding sites of DCN. Finally, DCN inhibited VEGF-E-induced EVT migration by interfering with ERK1/2 activation. Our findings reveal a novel role of DCN as an antagonistic ligand for VEGFR-2, having implications for pathophysiology of preeclampsia, a trophoblast hypoinvasive disorder in pregnancy, and explain its antiangiogenic function.

Decorin (DCN), a member of the small leucine-rich proteoglycan family is a ubiquitous component of the extracellular matrix, where it is synthesized by mesenchymal cells and serves multiple functions including regulation of collagen fibrillogenesis, maintenance of tissue integrity, and serving as a reservoir for TGF-β (1–3). It has been implicated in regulating multiple cellular functions because of its ability to bind to a variety of molecules both in the extracellular matrix as well as on the cell surface (4, 5). DCN contains a 40-kDa core protein linked at its N terminus to a single tissue-specific glycosaminoglycan (GAG) chondroitin sulfate or dermatan sulfate chain. The mature protein is highly conserved across species and consists of a central domain harboring multiple leucine-rich repeats (LRR). Most of the biological functions of DCN are mediated by the organization of the LRR in the core protein, which folds into an arch-shaped structure with a concave surface well suited to bind both globular and nonglobular proteins (2). High-affinity binding sites for collagen have been located in LRR4 and 5 (6) and for TGF-β between LRR3 and -5 (2, 7). DCN binding to mature TGF-β inactivates TGF-β in some cases (8) by interfering with TGF-β receptor activation and signaling (2, 9). Although crystal structure analysis of DCN indicated that it is a stable dimer with large interfaces (10), it was shown that biologically active DCN is a monomer in solution and thus is a monovalent ligand for various extracellular matrix proteins, growth factors, and cell surface receptors (11).

DCN can negatively regulate a variety of cellular functions, either by binding to certain ECM molecules or cell surface receptors. For example, DCN interaction with fibronectin (12) and thrombospondin (13) inhibited cellular adhesion. An interaction with fibronectin and collagen-I via the DCN GAG chain impeded migration-promoting effects of these substrates on an osteosarcoma cell line (14). DCN was shown to exert antiproliferative effects on many cell types (15–18). It inhibited angiogenesis in vitro by blocking migratory function of endothelial cells (19, 20) and also tumor-induced angiogenesis in vivo (21) by antagonizing endogenous vascular endothelial growth factor (VEGF). Purified GAG-free DCN and its 26-residue leucine-rich repeat, LRR5, was shown to inhibit VEGF and serum-induced angiogenesis by endothelial cells in vitro (22).

Although the identity of possible receptors in multiple DCN actions remains untested in many studies, a number of tyrosine kinase receptors (TKR) have been implicated in receptor-mediated actions of DCN (5). It was shown to interact with epidermal growth factor receptor (EGF-R) in a squamous cell carcinoma cell line and led to dimerization and autophosphorylation of the receptor, triggering a signal cascade including activation of MAPKs, mobilization of intracellular calcium, eventual up-regulation of p21, and finally growth suppression, associated with protracted internalization of EGF-R by the caveolar pathway and a retardation of EGF-R recycling to the cell surface (23–25). Sustained down-regulation of EGF-R in squamous cell carcinoma cells (26) and ErbB2 in breast carcinoma cells (27) contributed to exogenous DCN-mediated growth suppression of tumor cells in vivo (28, 29). DCN was reported to interact with IGF receptor-1 (IGFR1) in endothelial cells, leading to its phosphorylation, followed by a down-regulation of the receptor, promoting cell survival (30). Interestingly, DCN-mediated activation of this receptor, in cooperation with its interaction with α2/β1 integrin via the GAG chain, promoted endothelial cell adhesion and motility on collagen 1 (31). DCN-mediated synthesis of fibrillin-1 in the kidney also depends on the binding of DCN to IGFR1 on renal fibroblasts (32). Recently, DCN was shown to bind and interact with Met, the receptor for hepatocyte growth factor on HeLa cells, leading to receptor phosphorylation, an enhanced degradation of the endocytosed receptor (33), and a concurrent suppression of β-catenin and Myc within the cells (34).

The human placenta is an invasive structure, in which a highly migratory and invasive cell population known as the extravillous trophoblast (EVT) invades the uterine decidua and its arteries to allow an efficient exchange of oxygen and other key molecules between the maternal and the fetal blood. EVT cell proliferation, migration, and invasiveness are exquisitely regulated in situ by a large number of locally derived molecules in a positive or negative manner to maintain a healthy uteroplacental homeostasis. A breakdown in this homeostasis can occur in preeclampsia, a trophoblast hypoinvasive disorder and choriocarcinoma, a trophoblast hyper invasive disease of the placenta (35). Previously we identified two decidua-derived negative regulators of normal EVT cell proliferation, migration, and invasiveness: TGF-β and DCN, and found that choriocarcinoma cells are resistant to negative regulation by both molecules (18, 36, 37). Recently, we demonstrated that mechanisms responsible for DCN action on EVT cells are mediated through multiple TKR. By following DCN-mediated TKR phosphorylation and the effects of selective TKR inhibitors on DCN actions in EVT cells, we noted that the antimigratory action of DCN on the fibronectin substrate was mediated primarily through IGFR1, and the antiproliferative action was mediated primarily through EGF-R and VEGF receptor (VEGFR)-2 (38).

DCN action through its binding to VEGFR-2 has never been reported. For this reason, the present study was designed with the following objectives: 1) to test whether there is a direct binding of DCN with VEGFR-2 in EVT cells or in a cell-free system; 2) to map the peptide sequence in DCN core protein responsible for VEGFR-2 binding and functionality in EVT cells; 3) to test the binding affinity of DCN and DCN peptide fragments with VEGFR-2 in EVT cells relative to VEGF-E, a VEGFR-2 specific ligand; 4) to test whether the DCN-binding domain of VEGFR-2 overlapped with its ligand-binding domain; 5) to identify DCN antagonism of signals in VEGFR-2 mediated migration of EVT cells. We demonstrate that DCN binds directly to VEGFR-2 in EVT as well as human umbilical vein endothelial cell (HUVEC) extracts and also to recombinant VEGFR-2 Fc chimera. DCN binding to VEGFR-2 in EVT cells had a 7- to 10-fold lower affinity than VEGF-E. Interaction of VEGFR-2 with the DCN core protein was DCN peptide sequence selective both in binding and antagonizing VEGF-A- or VEGF-E-induced proliferation and migration of EVT cells. The DCN-binding domain of VEGFR-2 was found to overlap with the ligand-binding domain. Finally, DCN inhibited VEGFR-2-mediated EVT cell migration by interfering with ERK1/2 activation.

Results

Cell surface human leukocyte antigen (HLA)-G expression by the HTR 8/SV neo EVT cell line

To confirm HLA-G expression levels in the EVT cell line during the current experiments, we conducted a flow cytometric analysis as shown in Fig. 1.

Fig. 1.

HLA-G expression by the HTR-8/SV neo EVT cell line. Relative incidence of labeled EVT cells plotted against fluorescence intensity for control Ig (background, plotted in light line) and anti-HLA-G (dark line)-treated cells. FITC, Fluorescein isothiocyanate.

Binding of DCN to VEGFR-2 detected with far-Western blotting

Our earlier evidence for DCN-VEGFR-2 interactions in EVT cells was indirect (38). To obtain direct evidence for the binding, we first performed far-Western blotting with EVT lysates expressing native VEGFR-2, serving as the target Prey, probed with DCN core protein as the bait, as detailed in Materials and Methods. Recombinant human VEGFR-2/Fc chimera and VEGFR-2-expressing HUVEC lysate were used as positive controls, whereas VEGFR-2 lacking human embryonic kidney (HEK)-293T cell lysate served as the negative control. The binding of the prey with the bait proteins was detected with the specific anti-DCN antibody. In the lane containing the VEGFR-2/Fc chimera, DCN bound to a protein band of 165 kDa, matching the size of this molecule (Fig. 2A). In EVT and HUVEC lysates, DCN bound to multiple protein molecules varying in sizes, the strongest reactivity ranging between 150 to 170 kDa, which may include nonglycosylated and partially glycosylated forms of VEGFR-2 (39). However, a weak DCN-reactive band of approximately 160 kDa was also noted in the lane containing VEGFR-2-negative HEK-293 cell lysate (used as a negative control). After stripping of the membrane and reprobing with the rabbit monoclonal VEGFR-2 antibody, the above band in HEK-293T cell lysate disappeared, showing that it is not VEGFR-2 (Fig. 2B). On the other hand, 150- to 170-kDa bands both in EVT and HUVEC lysates were retained in the VEGFR-2 immunoblot, demonstrating that they included VEGFR-2. Interestingly, a strong band of 230 kDa, representing the mature form of VEGFR-2 (40) was only noted in the HUVEC lysate (Fig. 2B), although such a band was very faint in Fig. 2A, possibly indicative of very weak DCN binding. These results, taken together, revealed specific binding of DCN to VEGFR-2 contained in EVT cell and HUVEC proteins, as well as in the cell-free system.

Fig. 2.

Far-Western blot analysis of DCN binding to VEGFR-2. A, We used recombinant VEGFR-2/Fc chimera (positive cell-free control), EVT lysate protein (experimental sample), HUVEC lysate protein (positive cell lysate control), and VEGFR-2-negative HEK-293T cell lysate protein (negative control). All samples were resolved by SDS-PAGE. VEGFR-2 was used as prey and GAG-free (chondroitin ABC lyase treated) DCN or GST-DCN used as bait protein. The signals were detected using anti-DCN antibody or anti-GST-antibody (data not shown), revealing a 165-kDa band representing the recombinant VEGFR-2/Fc chimera and multiple bands of different sizes in both EVT and HUVEC lysates including a strong band spanning between 150 and 170 kDa (possibly including nonglycosylated and partially glycosylated forms of VEGFR-2) in both EVT cell and HUVEC lysates. B, Same membranes were striped, reprobed with anti-VEGFR-2 antibody revealing that the 150- to 170-kDa band span in both EVT and HUVEC lysates and the 165-kDa band in VEGFR-2/Fc chimera lane were genuine VEGFR-2 that bound to DCN. HUVEC revealed an additional 230-kDa band representing fully glycosylated form of VEGFR-2 that had very faint DCN reactivity in panel A. Results are from a representative experiment from three independent experiments.

DCN binding to VEGFR-2 detected by coimmunoprecipitation

To test further the validity of the results of far-Western blotting, we performed a series of coimmunoprecipitation experiments using EVT lysates as well as recombinant VEGFR-2/Fc chimera as detailed in Materials and Methods. The immunocomplexes were resolved by SDS-PAGE and immunoblotted for DCN core protein or glutathione-S-transferase (GST)-DCN. Figure 3A shows that VEGFR-2 coimmunoprecipitated with the 65-kDa GST-DCN. The immunoblotting for DCN core protein (shown in Fig. 3, B and C) produced identical results showing that VEGFR-2 Fc chimera (Fig. 3A) or VEGFR-2 in EVT cell lysate (Fig. 3B) coimmunoprecipitated with the 40-kDa DCN. In this experiment recombinant VEGFR-2/Fc chimera was used as the positive control (showing identical band) and IgG was used as the negative control (no band). These results further confirm the specific binding of DCN with VEGFR-2 in EVT cells.

Fig. 3.

Coimmunoprecipitation of DCN by VEGFR-2: GST-DCN (panel A) or GAG-free DCN (panels B and C) was incubated with recombinant VEGFR-2 /Fc chimera protein (positive control) or EVT cells lysate (experimental) to pull down the VEGFR-2 using the anti-VEGFR-2 antibody. The Western blot analysis was performed with anti-GST-antibody (shown in panel A) or anti-DCN antibody (panels B and C) The detected signals corresponded to the GST-DCN (65 kDa) or DCN (40 kDa). VEGFR-2/Fc chimera (panel B) or EVT cell lysate protein (panel C) were preincubated with a VEGFR-2 blocking monoclonal antibody, recognizing the ligand-binding terminus of VEGFR-2, before incubation with GAG-free DCN. IgG served as the negative control. Pretreatment with the blocking antibody completely abrogated DCN binding to VEGFR-2 in positive control as well as EVT cell lysate, indicating an overlap between the DCN-binding and ligand-binding domains of VEGFR-2. These experiments were reproduced twice. IB, Immunoblot; IP, immunoprecipitation.

DCN binds to the ligand-binding domain of VEGFR-2

Next, to test whether DCN binds to the ligand-binding domain of VEGFR-2 provided as the VEGFR-2/Fc chimera (Fig. 3B) or EVT-expressed VEGFR-2 (Fig. 3C), we used a VEGFR-2 blocking monoclonal antibody recognizing the ligand-binding domain of VEGFR-2 in coimmunoprecipitation studies using GAG-free DCN core protein (results presented) or GST-DCN (not presented, because results were identical) to pull down VEGFR-2. IgG was used as a negative control. Results revealed that the 40-kDa band (corresponding to DCN core protein), indicative of DCN binding to recombinant VEGFR-2 (Fig. 3B) or endogenous VEGFR-2 in EVT cell lysate (Fig. 3C) fully disappeared after preincubation with the VEGFR-2 blocking antibody, indicating a complete inhibition of the binding in both cases. These results demonstrate that DCN binds to VEGFR-2 in the ligand-binding domain. Thus, there is an overlap between the ligand-binding and DCN-binding domains of VEGFR-2.

LRR5 sequences in DCN core protein interacting with VEGFR-2

It was reported that the LRR5 peptide of the DCN core protein blocked VEGF-induced angiogenesis of endothelial cells in vitro more efficiently than the whole DCN protein (22). This finding prompted us to test whether LRR5 may include VEGFR-2 binding sites. To identify the possible amino acid sequences within the 26-residue LRR5 segment of the DCN core that may mediate the interaction with VEGFR-2, we used different truncated synthetic peptides having overlapping sequences as shown in Fig. 4A. They were tested for their blocking ability on the binding of VEGFR-2 to DCN in coimmunoprecipitation studies. We used recombinant VEGFR-2/Fc chimera as the positive control (Fig. 4B) and EVT lysate (Fig. 4C) as experimental material. Typically, 3–4 μg of recombinant VEGFR-2 and 30–40 μg of EVT lysate proteins were preincubated for 3 h with 1 μg/ml of the synthetic peptides. Peptide 4 (13 amino acid) strongly inhibited, and peptide 3 (12 amino acid) weakly inhibited DCN binding with VEGFR-2 in both cases. A modest inhibition was noted with the peptide 1 (26-amino acid LRR5) applied to EVT lysate only (Fig. 4C), whereas little or no inhibition was noted with peptide 2 (13 amino acid) in either case (Fig. 4, B and C). VEGFR2/Fc chimera used a loading control in Fig. 4B (165 kDa) and mature VEGFR-2 (230 kDa) in cell lysates (Fig. 4C) identified with the rabbit monoclonal antihuman VEGFR-2 antibody in the same experiment confirmed an equal amount of loading. From the amino acid sequences shown in Fig. 4A, it appears that the overlapping amino acid sequences in peptides 1, 3, and 4 (L G T N P L K, Fig. 4A) are required for the interaction of DCN with VEGFR-2. We investigated further the binding affinities of DCN and individual DCN peptides to VEGFR-2 from their ability to compete with VEGF-E binding by EVT cells.

Fig. 4.

Approximate identification of Core DCN protein sequences interacting with VEGFR-2. A, Amino acid sequences of the synthetic DCN LRR-5 peptide and its fragments used in our experiments to show the overlapping sequences in peptides 1–4. B, Coimmunoprecipitation of the purified full-length DCN protein with VEGFR-2/Fc chimera after preincubation with different peptides. C, Coimmunoprecipitation of purified full-length DCN with endogenous VEGFR-2 in the EVT lysate protein after preincubation with the same peptides. No peptide was added in controls in panels B and C. VEGFR-2/Fc chimera and endogenous VEGFR-2 in EVT lysate, detected with the rabbit monoclonal anti-VEGFR-2 antibody, were used as intrinsic loading controls in panels B and C, respectively. IB, Immunoblot; IP, immunoprecipitation.

Binding affinity of DCN and DCN peptides for VEGFR-2 in EVT cells

Figure 5 presents data on the binding kinetics of VEGF-E (Fig. 5A) to EVT cells and its displacement with DCN (Fig. 5, B and C) and DCN peptides (Fig. 5, D and E), presented from two different experiments. Based on the data, the B_max and K_d for VEGF-E binding to EVT cells was calculated as 19.68 pm and 566.5 pm, respectively (Fig. 5A). The respective K_i and IC₅₀ values for the displacement of VEGF-E by DCN were computed as 5.78 nm and 16 nm in experiment 1 (Fig. 5B) and 3.93 nm and 10.88 nm in experiment 2 (Fig. 5C). Thus DCN showed approximately 7- to 10-fold lower affinity than VEGF-E for VEGFR-2 binding in EVT cells. The VEGF-E displacement kinetics with DCN peptide 1 is presented in Fig. 5D. Figure 5E summarizes the IC₅₀ and K_i values for DCN peptides 2, 3, and 4 in this experiment. In summary, all the DCN peptides showed a stronger ability to displace VEGF-E than the DCN core protein in the following order: peptide 3, 4, 2, and 1. These data confirm the earlier biochemical data (Fig. 4C) that DCN peptides 1, 3, and 4 variably blocked DCN binding to VEGFR-2 in EVT cell lysate. However, in the present assay, peptide 2 is also shown to bind to VEGFR-2 with a stronger affinity than DCN. Furthermore, whereas peptide 4 showed the highest blocking activity in the biochemical assay (Fig. 4), peptide 3 showed the highest ability to displace VEGF-E in the binding studies. In further experiments we tested whether different DCN peptides could antagonize VEGF-A or VEGF-E in functional assays.

Fig. 5.

Affinity of VEGF-E and DCN or DCN peptide binding to VEGFR-2 in EVT cells. A, The binding kinetics of VEGF-E in EVT cells using a fixed concentration ¹²⁵I-labeled VEGF-E (1000 pm), in combination with increasing concentrations of cold VEGF-E. B and C, Radioligand binding in EVT cells with increasing concentrations of DCN in two separate experiments. The B_max and K_d for VEGF-E binding were estimated as 19.68 pm and 566.5 pm. The K_i and IC₅₀ for DCN in displacing VEGF-E binding were respectively estimated as 5.78 nm and 16 nm, respectively, in experiment 1 (panel B), and 3.93 nm and 10.88 nm respectively in experiment 2 (panel C). Binding kinetics using DCN peptides are shown for peptide 1 alone (panel D) and summarized for peptide 2, peptide 3, and peptide 4 in panel E. CPM, Counts per minute.

DCN peptides antagonize VEGF-121- and VEGF-E-induced EVT cell proliferation and migration

Dictated by the IC₅₀ and K_i values for DCN and DCN peptides, we decided to employ 50 nm DCN and an equivalent concentration of DCN peptides in functional assays to maximally antagonize VEGF actions produced by 20–50 ng VEGF121 (0.17–0.43 nm) and 30 ng/ml VEGF-E (1.00 nm). Neither DCN nor DCN peptides 1, 2, 3, and 4 exerted any cytotoxic effects on EVT cells at the highest concentration (50 nm) used in functional assays based on trypan blue staining, and cell viability was indistinguishable (96–98%) in untreated and treated cells (data not shown).

In the proliferation assay using VEGF-121 (Fig. 6A) both fetal bovine serum (FBS) and VEGF-121 (at 10 and 50 ng/ml) stimulated EVT cell proliferation of serum-starved cells. Presence of 10 and 50 nm DCN significantly inhibited basal proliferation of serum-starved cells. When DCN or DCN peptides (50 nm) were tested for their ability to antagonize VEGF-121 (50 ng/ml)-induced proliferation, peptide 1 (LRR5) was found to be as efficient as DCN core protein. Small but significant antagonism was also noted with peptides 3 and 4 but not peptide 2. In an independent experiment, using VEGF-E (Fig. 6B), DCN alone significantly inhibited basal proliferation of serum-starved cells at 20 and 50 nm, so also all DCN peptides at 50 nm concentration. DCN at 20 and 50 nm concentration and all DCN peptides (at 50 nm) also antagonized VEGF-E (30 ng/ml)-induced proliferation.

Fig. 6.

Effects of DCN and DCN peptides on VEGF-121- and VEGF-E-induced EVT cell proliferation. EVT cell proliferation was measured as relative levels of BrdU incorporation in serum-starved cells under various treatment conditions (24 h). FBS served as the positive control. Data are presented as mean of triplicates ± sd. *, Significant difference (P < 0.05) resulting from various treatments (with DCN or peptides) of serum-starved cells, as compared with SFM. #, Significant difference (P < 0.05) resulting from DCN or DCN peptide treatments of VEGF-A (50 ng/ml; presented in Fig. 6A) or VEGF-E (30 ng/ml; presented in Fig. 6B)-treated cells, compared with VEGF-A (panel A) or VEGF-E (panel B) treatment alone. Experiments in panel A (presented for VEGF-A) and panel B are from two different experiments, representative of four separate experiments.

In the migration assays conducted with VEGF-121 (Fig. 7B) migration of serum-starved cells was stimulated in a concentration-dependent manner (Fig. 7A), reaching a maximum at 20ng/ml. Again, similar to the proliferation assay, DCN and DCN peptides 1, 3, and 4, but not peptide 2, significantly antagonized VEGF-121 (20 ng/ml)-induced migration of serum-starved EVT cells. In an independent migration assay using VEGF-E (Fig. 7, C and D), both FBS and VEGF-E (30 ng/ml) caused a 3-fold increase in EVT cell migration of serum-starved cells, and DCN (20–50 nm) as well as DCN peptides (50 nm) significantly inhibited the basal migration of serum-starved cells. Significant inhibition of VEGF-E (30 ng/ml)-induced migration was noted at DCN concentrations 10–50 nm, reaching a maximum at 50 nm to below the level of basal migration in serum-free medium (SFM). All DCN peptides strongly abrogated VEGF-E-induced migration, the maximal antagonism noted with peptide 4.

Fig. 7.

Effects of DCN and DCN peptides on VEGF-121- and VEGF-E-induced EVT cell migration. The dose response for migration of EVT cells with respect to VEGF-A and VEGF-E treatment, respectively, was standardized (panels A and C); letters a,b,c,d differing in the bar graph indicate significant differences between them. The 24-h migration data represent the mean (of triplicate wells ± sd) absolute numbers of migrant cells under different treatment conditions using serum-starved cells. Cells were stimulated with either VEGF-A (panel B) or VEGF-E (panel D). Treatments with DCN or DCN peptides alone significantly reduced basal migration with certain treatments. *, Significant difference (P < 0.05) resulting from various treatments (with DCN or peptides) of serum-starved cells, as compared with SFM. #, Significant difference (P < 0.05) resulting from DCN or DCN peptide treatments of VEGF-A (50 ng/ml; presented in panel B) or VEGF-E (30 ng/ml; presented in panel D)-treated cells, compared with VEGF-121 (panel 7A) or VEGF-E (panel C) treatment alone. Experiments in panels B and D are two different experiments, representative of four separate experiments.

DCN inhibits VEGF-E-induced EVT cell migration by interfering with ERK1/2 activation

In a migration assay (Fig. 8A), the MEK1/2-specific inhibitor U0126 inhibited VEGF-E-induced EVT cell migration to that of a serum-free state. Significant inhibition occurred at concentrations as low as 5.0 μm of U0126 (Fig. 8A) that had been tested to have no effect on cell viability. When EVT cell lysates were subjected to immunoblots for phospho ERK1/2 (T202/Y204), total ERK and β-actin, U-0126 also blocked VEGF-E-induced ERK activation (data not shown). Similarly VEGF-E-stimulated ERK activation was blocked with DCN at a concentration of 50 nm (Fig. 8, B and C), without having any effect on cell viability.

Fig. 8.

ERK1/2 (p44/42) is required for VEGF-E-induced cell migration. A, EVT cell migration was measured as the number of migratory cells after 24 h incubation on transwell migration chambers. Cells were pretreated with varying concentrations of the MEK1/2 inhibitor, U0126 (0.5, 5.0, 10.0 μm) for 1 h. VEGF-E (30 ng/ml) was used as a chemotactic agent. Migration data are presented as a grand mean ± sd for two independent experiments with each condition being tested in triplicates. Significant differences (P < 0.05) in means are indicated as different superscripts. B and C, HTR-8/SVneo cells were serum starved for 24 h followed by pretreatment with varying concentrations of DCN (0, 10, 50 nm). Cells were then stimulated with VEGF-E (30 ng/ml) for 15 min. Cell lysates were prepared and analyzed for ERK1/2 activity by phospho-p44/42 (Thr202/Tyr204) Western blot. Inhibition of VEGF-E-induced p44/42 activity occurred at a concentration of 50 nm DCN.

Discussion

Using biochemical and functional approaches, the present study reveals for the first time that DCN is a novel ligand for VEGFR-2, antagonizing VEGF action. DCN binding to VEGFR-2 was demonstrated in human EVT cells that had earlier been shown to express functional VEGFR-2 (38, 41) and also in a cell-free system using recombinant VEGFR-2/Fc chimera as well as in HUVEC used as positive controls. VEGFR-2 lacking HEK-293T cells were used as a negative control for the binding. We also demonstrated that the binding of DCN to VEGFR-2 could be blocked with a VEGFR-2 blocking antibody, indicating an overlap between the ligand (VEGF)-binding and the DCN-binding domains. The precise DCN -binding sites on the N terminus of VEGFR-2 remain to be mapped.

Because our EVT cell line had been shown to express all the three VEGF receptors including VEGFR-1, VEGFR-2 (41, 42), and VEGFR-3 (43), we employed a VEGFR-2-specific ligand VEGF-E (44) in radioligand-binding assays to determine the binding affinity of DCN to VEGFR-2 in EVT cells relative to VEGF-E. This affinity was found to be 7- to 10-fold lower than VEGF-E, a ligand shown to have an affinity similar to VEGF-169, the most common isoform of VEGF-A in mammals (44). The small difference between the K_d value of 566 pm for VEGF-E in EVT shown here and the K_d value of 333 pm reported for a VEGF-R2 overexpressing NIH 3T3 cell line (44) may be due to the differences in the cell type used. The demonstration that DCN exhibits a low affinity for VEGFR-2 was also observed for DCN binding to two other members of the TKR family: EGF-R (24), and IGFR1 (30). Only in the case of Met, the third DCN-binding receptor in the TKR family identified recently, the affinity was found to be similar to its natural ligand hepatocyte growth factor (33).

As the first step in approximating the identity of the VEGFR-2 binding sites in the DCN core protein, we focused on LRR5 and its peptide fragments with overlapping amino acid sequences, because LRR5 had been reported to be more efficient than the DCN core protein in blocking VEGF-induced angiogenesis by endothelial cells in vitro (22). In the binding assay using coimmunoprecipitation, peptide fragment 4 was most efficient in competing for DCN binding, whereas peptides 1 (LRR 5) and 3 were less efficient. In the VEGF-E radioligand displacement assay, peptide 3 was most efficient, followed by peptides 4, 2, and 1. These findings were validated further by their antagonistic actions on VEGF-A- and VEGF-E-induced EVT proliferation and migration. Because VEGF-A can bind both VEGF-R1 and R2, both of which are expressed by EVT cells, small differences in results obtained with the two ligands, such as lack of effectiveness of peptide 2 in antagonizing VEGF-A actions, can be explained by compensatory effects of VEGFR-1.

In our earlier studies (38) we had shown that both EGF-R and VEGFR-2 were responsible for antiproliferative action of DCN on EVT cells, whereas IGFR1 was primarily responsible for the migration-inhibitory actions of DCN on EVT cells plated on a fibronectin substrate. Under the present experimental conditions using no fibronectin substrate, VEGFR-2 was also shown to participate in the migration-inhibitory actions of DCN. Indeed, VEGF-E, the VEGFR-2-specific ligand, was a potent stimulator of EVT migration, allowing us to use the migration assay to identify VEGFR-2 antagonism by DCN and DCN peptides.

Based on three different approaches (blocking DCN binding of VEGFR-2 in immunoprecitipation presented in Fig. 4; displacements of VEGF-E binding to EVT cells presented in Fig. 5, C–E, and finally antagonism of VEGF-E-induced proliferation and migration in Figs. 6B and 7D, which showed minor differences in results, it is evident that all the peptides contribute to DCN binding to VEGFR-2, most avidly by peptides 3 and 4. In Fig. 9 we present a model of DCN binding to VEGFR-2. Considering the overlapping amino acid sequences among peptides 1, 2, 3, and 4 (Fig. 4A), we propose that amino acid sequences L G T N P L K S S G I E represent the minimal VEGFR-2-binding site in LRR5 of the DCN core protein. This sequence should serve as a guide in further validation studies with mutant DCN proteins to be used in binding and functional assays.

Fig. 9.

A model for DCN-VEGFR-2 interactions. Based on our results we suggest that DCN binds to the N terminus of VEGFR-2, overlapping with the VEGF-binding domain, the degree of overlap remaining undetermined. The VEGFR-2 binding site of the DCN core protein includes 12 amino acid (AA) sequence L G T N P L K S S G I E in the LRR5 domain, most avid binding represented by L G T N P L K at the proximal end.

The precise mechanism(s) responsible for the antagonistic actions of DCN by binding to VEGFR-2 in EVT cells remain to be elaborated. In a previous series of elegant studies (5, 24, 26, 30, 33) on DCN-mediated antagonism of several members of TKR, namely EGF-R, IGFR1 and Met, a common theme emerges: DCN binding to the receptor causes transient receptor phosphorylation, accelerates receptor endocytosis and degradation, and retards receptor recycling. The most profound and rapid receptor degradation was noted in the case of Met (33). We believe that DCN action on VEGFR-2 shares some of these features, including a transient receptor phosphorylation, which had been reported by us earlier (38).

Until now, the precise signaling mechanisms responsible for DCN antagonism of VEGF-induced proliferation and migration of EVT cells remained to be identified. VEGFR-2 is considered the major mediator of several physiological and pathological effects of VEGF including cell migration and proliferation (45). VEGF has been reported to stimulate cell migration by activating ERK1/2 in a variety of cell types (46, 47). In this study we found that blocking ERK1/2 using the pharmacological inhibitor, U0126, inhibited EVT migration induced by VEGF-E, the VEGF-R2-specific ligand. Hence we examined whether DCN inhibits VEGF-E-induced EVT migration by interfering with ERK1/2 activation. We show that VEGF-E-induced ERK activation could be blocked by very low concentrations of DCN. These results, taken together, reveal that DCN inhibits VEGF-E-induced EVT migration by interfering with VEGF-E-induced ERK1/2 activation (Fig. 8, B and C), an event that requires ligand binding to VEGFR-2. Finally, an up-regulation of P21 mediating antiproliferative effects of DCN has been reported for many cells, including our earlier findings on the EVT (18). However it remains to be tested whether P21 activation is the final common pathway for the antiproliferative effects of DCN on EVT cells resulting from binding to EGF-R (38) as well as VEGFR-2 (Ref. 37 and present study).

The biological significance of DCN antagonism of VEGFR-2 by a direct binding to this receptor is profound. VEGFR-2 is the most important receptor responsible for angiogenesis under physiological and pathological conditions, including tumor-induced angiogenesis. Antiangiogenic action of DCN in vitro (19, 20, 22) and also tumor-induced angiogenesis in vivo (21) can now be fully explained on the basis of its binding to VEGFR-2. It is likely that the antitumor action of exogenous DCN in a number of experimental tumor models (27–29, 48) can be explained on the basis of DCN antagonism of multiple TRK including VEGFR-2.

Although DCN can antagonize multiple tyrosine kinase receptors expressed by the trophoblast, DCN antagonism of VEGFR-2 has strong implications for placental physiology and pathology because of its dual role on the trophoblast and endothelial cells. Now it is evident that DCN colocalizing with TGF-β in the decidual extracellular matrix (ECM) (49) serves a dual physiological purpose: 1) storage of TGF-β in an inactive form until it is cleaved and activated by the EVT-derived proteases to limit trophoblast invasion; and 2) limiting trophoblast migration and invasiveness independent of its binding to TGF-β (18). Preeclampsia is a serious pregnancy-associated disease of the pregnant mother ascribed to a hypoinvasive placenta resulting from multiple causes. Poor EVT invasion leads to an inadequate remodeling of the endometrial arteries, resulting in poor perfusion of the placenta with maternal blood, which in turn, can cause fetal growth restriction and trigger vascular damage in the mother (35). Increased levels of antiangiogenic molecules such as soluble VEGF-R1 and soluble endoglin (50–52), leading to a reduction of bioavailable placental growth factor and VEGF at the feto-maternal interface, has been proposed in the genesis of preeclampsia. We suggest that DCN overexpression or activity (e.g. release of highly active DCN peptides) in the decidua may also contribute to this disease by its concurrent roles in antagonizing trophoblast invasion and angiogenesis, a hypothesis that is currently being tested.

Materials and Methods

Materials

Human recombinant VEGFR-2/Fc chimera was purchased from R&D Systems (Cedarlane, Hornby, Ontario, Canada). Pure bovine articular cartilage-derived DCN and Chondroitin ABC lyase were from Sigma (Oakville, Ontario, Canada). GST-tagged DCN was from Abnova (Taipei, Taiwan). Mouse monoclonal anti-DCN antibody and protein A/G agarose were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Rabbit monoclonal antihuman VEGFR-2 antibody (no. 2479) used for immunoblots and immunoprecipitation was from Cell Signaling Technology (Danvers, MA). A rat antimouse VEGFR-2 (KDR/Flk-I) monoclonal antibody, Clone RAFL-2 (53), used for blocking VEGFR-2, was a kind gift of Dr Sophia Ran (Southern Illinois University School of Medicine). This antibody was found to be cross-reactive with human VEGFR-2, blocking ligand binding to VEGFR-2. Falcon cell culture inserts (8.0-μm pore size), used for migration assays, and 96-well cell culture plates used for cell proliferation assays were from BD Labware (Franklin lakes, NJ). Polyvinylidene difluoride (PVDF) membranes (ImmunoBlot PVDF) were from Bio-Rad Laboratories, Inc., Hercules, CA). VEGF-E was from GenWay Biotech (San Diego, CA).

Core DCN peptides: source and purification

Crude DCN peptides were purchased from GenScript Corp. (Piscataway, NJ) and purified by reversed-phase HPLC, freeze-dried, and stored at −20 C until used. Peptides used in this study are: peptide 1 (QMIVIELGTNPLKSSGIENGAFQGMK, the 26-residue LRR5); and its fragments: peptide 2 (SSGIENGAFQGMK, 13 residues); peptide 3 (LGTNPLKSSGIE, 12 residues); and peptide 4 (QMIVIELGTNPLK, 13 residues) (22).

Deglycosylation of DCN

Bovine cartilage-derived DCN (100 μg) (Sigma, St. Louis, MO) was digested with 1 U of Chondroitin ABC lyase in 40 mm Tris-HCl, 40 mm sodium acetate, 10 mm EDTA, 0.1 mm phenylmethylsulfonyl fluoride, 0.36 mm pepstatin buffer (pH 8.0) for 3 h at 37 C. The resultant GAG-free, 40-kDa DCN core protein was used in most experiments unless specified otherwise.

Cell lines

We used a human EVT cell line, HTR-8/SVneo (henceforth referred to as EVTs), derived in our laboratory by simian virus (SV) 40 tag immortalization of a short-lived first-trimester primary EVT cell line HTR-8 that was produced by propagation of EVT cells migrating out of a first-trimester chorionic villus explant (54). HTR-8/SVneo cells, like the primary parental line, express all the markers of EVT in situ, including cytokeratin 7, 8, and 18, placental-type alkaline phosphatase, high-affinity uPA receptor, human leukocyte antigen (HLA) framework antigen w6/32, IGF-II mRNA and protein, and a selective integrin repertoire α1, α3, α5, β1, and the vitronectin receptor α _vβ₃/β₅ (55). These cells also express HLA-G mRNA and protein (56, 57) and exhibit phenotypic behavior of freshly isolated cytotrophoblast cells during Matrigel invasion, including the expression of HLA-G (58). Cells at 92–105 passages (100% positive for the EVT cell marker cytokeratin 7) were grown in RPMI 1640 complete medium, including 10% FBS, 50 U/ml penicillin, unless otherwise specified. HUVEC and HEK-293 cells (obtained from the American Type Culture Collection, Manassas, VA), used as respective positive and negative controls for VEGFR-2 expression (59), were grown in the endothelial cell growth medium and RPMI 1640 complete medium with 10% FBS, 50 U/ml penicillin, respectively.

Flow cytometry

For flow cytometric analysis of HLA-G expression, HTR-8/SVneo cells grown up to 80% confluence were harvested, centrifuged, washed, and resuspended in 2% FBS/Dulbecco's PBS flow buffer at a concentration of 10⁴ cells/ml for labeling. Cells were incubated at 4 C for 1 h with fluorescein isothiocyanate-labeled mouse monoclonal anti-human the HLA-G1 antibody (MEM-G/9, AbCam, Cambridge, MA) or an equivalent concentration of fluorescein isothiocyanate-control mouse IgG, and washed twice in the same buffer. The cells were then analyzed with a Coulter EPICS V analytical flow cytometer (Coulter Electronics, Hialeah, FL), recording 10,000 events.

Coimmunoprecipitation and Western blotting

Coimmunoprecipitation of DCN and VEGFR-2 and their detection by Western blotting was done as described previously (60). Briefly, EVT cell monolayers were scraped off in ice-cold PBS containing 5 mm NaF and 1 mm Na₃VO₄ and then collected by centrifugation at 4 C. Cells were lysed in radioimmune precipitation assay buffer [50 mm Tris-HCl (pH 7.4), 1% Nonidet P-40, 0.25% sodium deoxycholate, 150 mm NaCl, 1 mm EDTA, 1 mm NaF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mm phenylmethylsulfonylfluoride]. Equal concentrations of proteins were incubated with 0.5 μg GST-tagged DCN or GAG-free DCN for 4 h in a shaker at 4 C followed by incubation with 10 μl of rabbit monoclonal anti-VEGFR-2 antibody (no. 2479, Cell Signaling Technology; 1:1000 dilution) overnight. After 2 h of incubation with 25 μl of protein A/G sepharose beads (Santa Cruz) at 4 C, the beads were washed twice with PBS and then boiled in SDS-PAGE sample buffer (1:1) for 5 min to elute proteins for subsequent electrophoresis. Western blotting was done as follows: the eluted protein was resolved by 10% SDS-PAGE, transferred to PVDF membranes, and subjected to immunoblotting using the mouse monoclonal anti-DCN primary antibody (1:500). After incubation with the horseradish peroxidase-linked secondary antibodies, proteins were visualized by the enhanced chemiluminescence detection system (Amersham Pharmacia Biotech, Piscataway, NJ). For competition or blocking experiments, the cell lysate or recombinant VEGFR-2-Fc chimera proteins were preincubated with 100 μg of DCN core synthetic peptides (P1, P2, P3, and P4) or 5 μg VEGFR-2 blocking monoclonal antibody (RAFL-2) for 2 h at 4 C, before adding the DCN core protein. For signaling experiments, cells were serum starved for 24 h followed by pretreatment with varying concentrations of DCN (0, 10, 50 nm) for 1 h. Cells were then stimulated with VEGF-E (30 ng/ml) for 15 min. Cell lysates were prepared, resolved by 10% SDS-PAGE, and transferred to a PVDF membrane. Membranes were probed for phospho-p44/42 (Thr202/Tyr204) and total p44/42 (Cell Signaling Technology).

Far-Western blotting

Far-western blotting of DCN and VEGFR-2 was conducted as reported elsewhere (61). Briefly, 45–50 μg of protein from EVT lysates were boiled 5 min in Laemmli sample buffer (1:1), resolved by SDS-PAGE on 8% gel, and transferred onto PVDF membranes. After 1 h incubation in 1% casein in Tris-buffered saline-0.1% Tween, membranes were incubated with 2 μg of DCN protein for 3 h at 4 C. The membranes were then washed with Tris-buffered saline-0.1% Tween and incubated with anti-DCN antibody (1:500) overnight at 4 C. The membranes were washed again and incubated with an horseradish peroxidase-linked antimouse IgG secondary antibody (1:10,000) (Santa Cruz Biotechnology) for 1 h at room temperature. Finally, the resulting protein bands were detected with ECL plus Western Blotting Detection System (Amersham).

Radioligand binding assay

This assay evaluates the binding affinity of a ligand to its receptor and the ability of specific antagonists to displace the binding (62). VEGF-E is an Orf virus-derived VEGF that binds to VEGFR-2 with high affinity but not to VEGF-R1 (44), allowing us to use this ligand in a radioligand binding assay to measure the ability of DCN or certain DCN peptides (LRR5 and its fragments) to displace ¹²⁵I-labeled VEGF-E binding to EVT, relative to cold VEGF-E. VEGF-E was radioiodinated with the chloramine-T method to produce ¹²⁵I-labeled VEGF-E of high specific activity (225 μCi /μg). For labeling, EVT harvested from semiconfluent cultures were seeded at 10⁵ cells/500 μl medium per well in 24-well plates, allowed to attach overnight, and then preincubated with the binding medium (RPMI-1640; 10 mm HEPES, pH 7; containing 0.1% BSA) for 30 min at 4 C. The cells were then subjected to labeling (44) by incubating them with 10–2000 pm ¹²⁵I-labeled VEGF-E in the binding medium for 90 min at 4 C. In competition assays, a 10-fold concentration cold VEGF-E was mixed with radioactive VEGF-E. In separate competition experiments, 1000 pm of radioactive VEGF-E was mixed with increasing concentrations of cold VEGF-E or cold DCN or cold DCN peptides. The cells were washed twice with PBS and then lysed with radioimmune precipitation assay cell lysis buffer for radioactivity counts in a γ-counter (LKB Wallac Clinigamma 1272, Wallac Oy, Finland). The binding kinetics (B_max and K_d for VEGF-E, and K_i and IC ₅₀ for DCN or DCN peptides) were measured with the GraphPad program (GraphPad Prism version 4, Graph Pad Software, Inc., San Diego, CA).

Proliferation assay

The effects of treatments with VEGF-A or VEGF-E on proliferation of HTR-8/SVneo cells in the presence or absence of DCN or various DCN peptides (P1, P2, P3, and P4) were assessed with a colorimetric immunoassay based on the measurement of bromodeoxyuridine (BrdU) (pyrimidine analog) incorporation during DNA synthesis in proliferating cells using the cell proliferation ELISA, BrdU colorimetric kit from Roche Diagnostics (Mannheium, Germany), according to the manufacturer's instructions. Briefly, EVT were cultured in RPMI 1640 complete medium in the presence of 10% FBS and 50 U/ml penicillin until they reached 50% confluency. Confluent EVT were serum starved overnight, trypsinized, and seeded at 5 × 10³/well on 96-well culture plate (Corning, Inc., Corning, NY) in RPMI 1640 serum-free medium containing 1% BSA (SFM). Treatment conditions included: SFM, FBS (10%, used as a positive control), DCN (10–50 nm), DCN peptides P1, P2, P3, and P4 (50 nm), VEGF-121 (10–50ng/ml), VEGF-E (30 ng/ml), and a combination of DCN or DCN peptides with VEGF-121 or VEGF-E at concentrations specified in Results. The cells were allowed to adhere to the plate and after 24 h, the degree of cell proliferation was measured by colorimetry using an ELISA plate reader (Tecan infinite M200 microplate reader, Grödig, Austria). The data were presented as the relative values for absorbance. All assays were conducted in triplicate, and experiments were repeated for reproducibility at least three times.

Migration assay

Migration of EVT cells was determined using Costar transwells (6.5-mm diameter; 8-μm pore size) as previously described (55). Briefly, HTR-8/SV-neo cells were serum starved by culturing in SFM overnight, trypsinized, and suspended in SFM. Cells were pretreated with varying concentrations of DCN (0, 5, 10, 20, 50, 100 nm), or 50 nm DCN peptide P1, P2, P3, and P4 or varying concentrations of the MEK1/2 inhibitor, U0126 (0.5, 5, 10 μm) for 1 h. Cells (7.5 × 10⁴) were placed on polycarbonate membranes of Transwell chambers. Lower chambers were filled with 0.6 ml of SFM or SFM containing 20 ng/ml VEGF121, or 30 ng/ml VEGF-E. In selected experiments, 10% FBS was employed as a positive control. Cells were allowed to migrate in a humidified incubator with 5% CO₂ at 37 C for 24 h, when migration had been shown to reach a plateau in control pilot experiments. The upper surfaces of the membranes were wiped gently with cotton swabs to remove nonmigratory cells. The membranes were then fixed and stained with 0.1% crystal violet in methanol, and the absolute number of migrant cells was scored visually using a light microscope at ×400 magnification. Migration data were presented as the mean of absolute values or the percentage of control values, each condition was tested in triplicate, and experiments were repeated at least three times.

Statistical analysis

Data on cell proliferation and migration were expressed as the mean ± sd. They were analyzed with ANOVA and two-tailed t test. Differences between two treatment groups were accepted as significant at P < 0.05.