Soluble perlecan domain i enhances vascular endothelial growth factor-165 activity and receptor phosphorylation in human bone marrow endothelial cells

Background Immobilized recombinant perlecan domain I (PlnDI) binds and modulates the activity of heparin-binding growth factors, in vitro. However, activities for PlnDI, in solution, have not been reported. In this study, we assessed the ability of soluble forms to modulate vascular endothelial growth factor-165 (VEGF165) enhanced capillary tube-like formation, and VEGF receptor-2 phosphorylation of human bone marrow endothelial cells, in vitro. Results In solution, PlnDI binds VEGF165 in a heparan sulfate and pH dependent manner. Capillary tube-like formation is enhanced by exogenous PlnDI; however, PlnDI/VEGF165 mixtures combine to enhance formation beyond that stimulated by either PlnDI or VEGF165 alone. PlnDI also stimulates VEGF receptor-2 phosphorylation, and mixtures of PlnDI/VEGF165 reduce the time required for peak VEGF receptor-2 phosphorylation (Tyr-951), and increase Akt phosphorylation. PlnDI binds both immobilized neuropilin-1 and VEGF receptor-2, but has a greater affinity for neuropilin-1. PlnDI binding to neuropilin-1, but not to VEGF receptor-2 is dependent upon the heparan sulfate chains adorning PlnDI. Interestingly, the presence of VEGF165 but not VEGF121 significantly enhances PlnDI binding to Neuropilin-1 and VEGF receptor-2. Conclusions Our observations suggest soluble forms of PlnDI are biologically active. Moreover, PlnDI heparan sulfate chains alone or together with VEGF165 can enhance VEGFR-2 signaling and angiogenic events, in vitro. We propose PlnDI liberated during basement membrane or extracellular matrix turnover may have similar activities, in vivo.


Background
Perlecan, a heparan sulfate proteoglycan with preferred localization to vascular basement membranes, is comprised of a~480 kDa protein core with five distinct domains (I -V). Domains II-V share structural homologies with other protein modules [1]. In contrast, N-terminal domain I (PlnDI) is structurally unique. As ã 22 kDa protein core, PlnDI contains 172 amino acid residues that give rise to a sperm protein, enterokinase and agrin (SEA) module localized downstream of three Ser-Asp-Gly motifs that serve as glycosaminoglycan (GAG) attachment sites [2,3].
Through the chondroitin and heparan sulfate GAG chains attached to domain I, perlecan functions as a ligand reservoir for storage, release, and protection of heparin-binding growth factors (reviewed by Whitelock et al., 2008). These interactions allow perlecan to modulate a range of biological functions, including angiogenesis (reviewed by Bix and Iozzo, 2008) [4]. Recent studies suggest immobilized forms of perlecan and PlnDI bind VEGF 165 to coordinate developmental angiogenesis by modulating VEGF 165 /VEGFR-2 signaling [5,6]. However, a role for soluble forms of PlnDI and the mechanism(s) by which it modulates VEGF 165 /VEGFR-2 signaling is unclear.
Herein, we tested the hypothesis that soluble forms of recombinant PlnDI bind and increase VEGF 165 /VEGFR-2 interactions on human bone marrow endothelial cells, in vitro. Observations from this investigation suggests soluble forms of recombinant PlnDI are biologically active and capable of interacting with components of the VEGFR-2 signaling complex, enhance activity and downstream signaling related to endothelial cell angiogenic processes.

Purification and biochemical characterization of PlnDI
Recombinant PlnDI was purified from conditioned media of HEK 293 EBNA clones as reported previously [17], and further enriched by passage through a Sepharose CL-6B column. This additional step removed high molecular weight contaminants secreted into the serum free media (i.e., full length perlecan). Aliquots of the eluted product were subsequently analyzed by SDS-PAGE and Western blotting to identify the GAG chain composition and preparation purity.
In Alcian blue stained SDS-PAGE gels, undigested samples displayed a broad band between~45-117 kDa  The presence of PlnDI was confirmed by Western blotting using anti-PlnDI specific antibodies (CSI-0071) and antibodies (3G10) to anti-Δ-heparan sulfate that recognize heparan sulfate neo-epitopes, generated following heparinase cleavage (arrow Figure 1C and 1D). Neither antibody recognized undigested products; however, anti-PlnDI antibodies recognized partially digested products (bracket in Figure 1C, lane 2) and both antibodies recognize a distinct band at 33 kDa (arrow, Figure  1C and 1D). The 33 kDa band reflects the domain I core protein adorned with GAG chain linkage residues following heparinase digestion.

VEGF 165 binds to PlnDI in a heparan sulfate dependent manner
To identify requirement(s) for VEGF 165 binding to PlnDI, both solid and solution phase binding assays were performed. In solid phase binding assays, immobilized PlnDI binds VEGF 165 in a heparan sulfate dependent manner ( Figure 2). Heparinase cocktail treatment of PlnDI, prior to immobilization on nitrocellulose, reduced VEGF 165 binding by~75% ( Figure 2). In contrast, pre-digestion with chondroitinase ABC did not alter VEGF 165 binding. Studies with the PlnDI protein core, prepared following digestion with a mixture of both enzymes, suggest VEGF 165 poorly binds this region. VEGF antibodies do not bind immobilized PlnDI ( Figure   2). In competitive inhibition assays, heparin [0.25 μg/ml] prevented~80% of VEGF 165 binding to PlnDI ( Figure 2).
In solution, requirements for VEGF 165 binding to PlnDI were similar, but the capacity of binding demonstrated pH dependence ( Figure 3A). When the pH of solution was reduced from 8.0 to 7.0 then 6.0, VEGF 165 binding was reduced by 50% and 80%, respectively (Figure 3A). To identify VEGF 165 specific binding, the background binding of VEGF 165 to nitrocellulose was subtracted from total bound to PlnDI [18]. Employing this approach, PlnDI-HS chains account for nearly all VEGF 165 binding, and the presence of CS chains masks VEGF 165 interaction with HS ( Figure 3B). In panel B, neutral pH was chosen to more closely reflect tissue culture conditions of subsequent experiments.

PlnDI modulation of VEGF 165 bio-activity
To identify a role for PlnDI in modulating VEGF 165 activity in vitro, human bone marrow endothelial cells were employed in two independent assays: 1) VEGF 165enhanced capillary tube-like formation; 2) VEGF 165enhanced phosphorylation of VEGFR-2. In capillary tube-like formation assays, the ability of bone marrow endothelial cells to form tube-like structures in the presence of exogenous VEGF 165 +/-PlnDI was quantified. Under serum free conditions, the addition of soluble VEGF 165 (positive control) and PlnDI demonstrated dose dependent increases in lengths of tube-like structures formed ( Figure 4A-B and 1F). Optimal concentrations for VEGF 165 [20 ng/ml] and PlnDI [12.5 μg/ml] increased tube-like formation 35% and 24%, respectively.
Studies employing PlnDI, pre-treated with either chondroitinase ABC and/or a heparinase cocktail suggests the ability of PlnDI to enhance tube-like formation is HS chain dependent ( Figure 4C). Moreover, PlnDI activity is further enhanced when its CS chains are removed. Interestingly, PlnDI/VEGF 165 mixtures combine to enhance tube-like formation 16% relative to VEGF 165 alone ( Figure 4D). The synergy between PlnDI and VEGF 165 is PlnDI-HS chain dependent ( Figure 4D). PlnDI protein core/VEGF 165 mixtures produce tube-like structures indifferent from those by VEGF 165 alone. Unexpectedly, heparin/VEGF 165 mixtures do not synergize in this system ( Figure 4E).
Since the presence of endogenous cell surface HS complicates the studies above, experiments employing bone marrow endothelial cells without cell surface HS were performed. Under these conditions, VEGF 165 and PlnDI enhance tube-like formation ( Figure 5) restored synergy with VEGF 165 in a PlnDI-HS chain dependent manner ( Figure 5). Because the role of HS in heparin-binding growth factor activity may involve interactions between HS, ligand, and cell surface receptors, the ability of PlnDI-HS to modulate VEGF 165 -induced VEGFR-2 tyrosine phosphorylation was investigated by Western blot using VEGFR-2 (Tyr-951) specific antibodies. VEGFR-2 phosphorylation at Tyr-951 results in recruitment of several adapter proteins whose subsequent downstream signaling supports endothelial cell survival and migration [19]. To perform these studies, we employed bone marrow endothelial cells whose cell surface HS were first removed by exposure to heparinases. Under these conditions, the exogenous addition of PlnDI and VEGF 165 (positive control) enhanced VEGFR-2 phosphorylation at Tyr-951 ( Figure 6A-B). The signal intensity of phosphorylation increased over time, peaked after ten minutes, then returned to control levels after 20 minutes ( Figure 6A-B). The addition of PlnDI, adorned with only HS chains, enhances Tyr-951 phosphorylation~3 fold relative to intact PlnDI ( Figure 6C). Studies employing PlnDI preparations pre-treated with mixtures of chondroitinase ABC and heparinase enzymes did not completely attenuate phosphorylation ( Figure 6C). Heparin addition (positive control) also enhanced VEGFR-2 phosphorylation ( Figure 6C).
Relative to either alone, PlnDI/VEGF 165 mixtures stimulate peak phosphorylation after only 2.5 minutes ( Figure 7A vs. 6A-B). To identify the role of PlnDI-HS in modulating VEGF 165 induced VEGFR-2 phosphorylation at Tyr-951, PlnDI preparations adorned with either CS, HS, or without GAGs were pre-mixed with VEGF 165 . The absence of HS chains on PlnDI reduced the signal intensity of phosphorylation 43% ( Figure 7B). In contrast, preparations decorated only with HS chains enhance the signal intensity of phosphorylation~3 fold ( Figure 7B). The absence of CS and HS chains did not completely reduce the intensity of phosphorylation relative to control (VEGF 165 ).
To determine if PlnDI/VEGF 165 enhanced VEGFR-2 phosphorylation also promotes downstream signaling, blots were stripped then re-probed with antibodies specific for total and phosphorylated forms of Akt. PlnDI/ VEGF 165 mixtures enhance the signal intensity of phosphorylated Akt~4 fold, relative to VEGF 165 alone (Figure 7C), and~40% of this activity is PlnDI-HS chain dependent.

Discussion
For the first time, we have characterized the ability of recombinant PlnDI to bind VEGF 165 and modulate its angiogenic activity, in vitro. We have shown that soluble forms of PlnDI are capable of modulating VEGFR-2 phosphorylation, as well as VEGF 165 -induced phosphorylation of VEGFR-2, and that the heparan sulfate glycosaminoglycan chains adorning PlnDI are important for these activities. Together, our observations suggest soluble forms of PlnDI may form and/or stabilize a complex between VEGF 165, NRP-1, and VEGFR-2 to enhance angiogenic events and VEGFR-2 signaling in human bone marrow endothelial cells (summarized in Figure 9). In contrast to our previous reports [17,20], the purity of PlnDI employed in the present investigation was enhanced by passage through a Sepharose CL-6B column. SDS-PAGE, Western blot and monosaccharide analysis suggest the molecular weight and GAG chain composition of PlnDI are similar to species previously characterized [20,21]. Moreover, these observations predict our preparation contains at least two species of PlnDI: one adorning predominately CS and the other predominately HS chains. Interestingly, the CS and HS disaccharide composition of PlnDI reported herein is different from species recently characterized by Whitelock et al. [22], as well as that reported for full length perlecan purified from bovine rib growth plate cartilage, HUAEC and RT101 cell lines [23][24][25]. These differences could be due to: 1) cell culture conditions; 2) approaches for purification; and 3) approaches employed for disaccharide analysis. Regardless, since fewer 4-sulfated CS residues and more 2-sulfated and 6-sulfated HS residues were identified it is reasonable to conclude that the function of PlnDI employed herein is distinct from forms previously reported. Indeed, subtle variations in HS substructure profoundly affect heparin-binding growth factor and receptor interactions, and thus the activity of perlecan [26][27][28].
While the role(s) of HS chains on perlecan have been most widely investigated with regard to regulation of FGF-2 activity [29,30], few studies have reported on perlecan-VEGF 165 interactions [5,6,22]. Moreover, the GAG modifications required specifically for perlecan-VEGF 165 interactions have not been described. Nevertheless, studies with heparin/HS suggest 2-O-and 6-O-sulfation is important for VEGF binding and activity [31][32][33]. Although the abundance of 2-O-and 6-O-sulfation on  PlnDI-HS suggests it harbors the capacity to interact with VEGF 165 , a correlation between VEGF 165 affinity and abundance of a particular disaccharide or the overall level of HS sulfation has not been observed [31]. Thus, growth factor binding is likely determined by HS domain organization (i.e., length of sulfation and transition domains, as well as their placement along the chain). Since HS chains on recombinant PlnDI are likely to be short (8-10 kDa) relative to those on tumorderived perlecan (30-70kDa) [21,34,35], we predict 48 residues comprise a single HS chain on PlnDI (based on the molecular weight of repeating units of glucuronic acid and N-acetylglucosamine). Moreover, since six or seven oligosaccharide residues are sufficient to fully occupy the HS binding site for VEGF 165 [31], we further predict that six VEGF 165 binding sites (maximally) may be available on each HS chain attached to PlnDI.
The HS dependent binding of VEGF 165 to immobilized PlnDI described herein is consistent with recent reports [5,6]. In contrast, a new communication has reported PlnDI does not bind immobilized VEGF 165 [36]. We suspect the concentration and/or the disaccharide composition of PlnDI employed therein may account for the contrasting observations. Our studies with PlnDI in solution suggest VEGF 165 binding to PlnDI in solution is not only HS but pH dependent. The marked reduction in VEGF 165 binding to PlnDI under acidic conditions, a novel observation, is consistent with previous publications describing the attenuation of VEGF 165 binding with low concentrations of heparin under acidic conditions, and its potentiation at neutral pH [14,37].
To identify the ability of soluble, exogenous PlnDI to modulate VEGF 165 activity, in vitro, tube-like formation studies were performed with human bone marrow endothelial cells seeded on growth factor reduced (GFR) Matrigel. We hypothesized that PlnDI/VEGF 165 mixtures would enhance the lengths of tube-like structures formed over VEGF 165 alone. While our observations support this hypothesis, we were surprised that PlnDI addition, alone, also enhanced the length of tube-like structures. Given our experimental approach, the enhancement of tube-like formation by soluble, exogenous, PlnDI may also reflect interactions with other matrix molecules (i.e., fibronectin and laminin) and heparin-binding growth factors present in GFR Matrigel reported to interact with PlnDI [38]. This possibility, however, should not discount the ability of exogenous PlnDI to interact directly with human bone marrow endothelial cells, or the possibility that the presence of heparin-binding molecules and growth factors may even mask the full activity of PlnDI.
Interestingly, under conditions where bone marrow endothelial cells were pre-treated with a heparinase cocktail, the additive effect of PlnDI/VEGF 165 mixtures on tube-like formation was not observed unless the concentration of PlnDI was increased two fold. While these observations suggest PlnDI-HS chains can modulate VEGF 165 activity, in vitro, heparin/VEGF 165 mixtures (positive control [14,32]), did yield similar results. We remain puzzled by this observation since heparin/ VEGF 165 mixtures combine to enhance VEGFR-2 phosphorylation, suggesting heparin is active in our system. At the cellular/receptor level, we analyzed VEGFR-2 auto-phosphorylation to identify requirements for PlnDI modulation of VEGF 165 activity, in vitro. While both VEGFR-1 and VEGFR-2 contribute to VEGF induced signals, VEGFR-2 dominates VEGF induced mitogenic and angiogenic responses in endothelial cells [11,12]. Of the six tyrosine phosphorylation sites identified on the intracellular domain of VEGFR-2, we report on one associated with endothelial cell survival and migration [39]. Together, our observations suggest exogenous soluble PlnDI, alone, can stimulate VEGFR-2 phosphorylation at Tyr-951. Moreover, PlnDI fragments harboring only HS chains further enhance VEGFR-2 phosphorylation, suggesting the presence of CS chains masks activity. These studies importantly extend those recently reported for full length perlecan [6] by demonstrating delivery of PlnDI or co-delivery with VEGF 165 are sufficient to enhance VEGFR-2 phosphorylation, and promote downstream signaling (i.e., increased Akt phosphorylation). Given our approach (i.e., the use of cells in suspension), our observations suggest PlnDI/VEGF 165   mixtures enhance survival signaling (increased Akt phosphorylation) of human bone marrow endothelial cells, in vitro. Consistent with this conclusion, our unpublished observations suggest VEGFR-2 phosphorylation at Tyr-1175 and Tyr 1214, and phosphorylation of p38 MAPK, Erk1/2 (events associated with endothelial cell proliferative and migratory states) [39], are unaltered.
Finally, to determine if PlnDI has the capacity to bind and modulate the activity of VEGFR-2 directly, we performed PlnDI binding studies against immobilized VEGFR-2, and NRP-1. Outcomes from these studies suggest PlnDI-HS chains, similar to heparin/HS, harbor the capacity to interact with VEGFRs and co-receptors [15,32,40], and enhance VEGFR-2 signaling [41]. We suspect PlnDI-HS chain binding to NRP-1 occurs via its heparin binding domain [15]. In contrast, PlnDI binding to VEGFR-2 is less dependent on HS chains. Heparin concentrations up to [100 μg/ml] did not appreciably alter binding (unpublished observations). Interestingly, the presence of VEGF 165 enhances PlnDI binding to VEGFR-2, suggesting the formation of a complex between PlnDI/ VEGF165/VEGFR-2 is possible. Our observations also suggest that modulation of VEGFR-2 signaling by PlnDI may involve complex interactions with more than one ligand.

Conclusion
The findings presented herein demonstrate exogenous, soluble, recombinant PlnDI is sufficient to bind and modulate the activity of the VEGFR-2 signaling complex via HS interactions, in vitro. Moreover, PlnDI may have activities independent of those with heparin-binding growth factors in supporting tube-like formation, in vitro. Figure 9 provides a simplified visual depiction of how PlnDI may impact angiogenic events in the absence or presence of VEGF 165 . PlnDI unbound or bound to VEGF 165 is liberated via cleavage within its SEA module [42] or the single immunoglobulin G-like region of domain II [43,44] during matrix turnover, wound healing, or disease progression. In the absence of VEGF 165, PlnDI-HS may bind to NRP-1, VEGFR-2, or support complex formation with both to signal downstream angiogenic events. When VEGF 165 is present PlnDI interactions with NRP-1 and VEGFR-2 are optimized, leading to enhanced downstream signaling and angiogenesis.

Immunoassays
Solid phase binding assays were performed as described previously [17]. For solution phase binding assays, PlnDI (5 μg) untreated, or pre-digested with a heparinase cocktail and/or chondroitinase ABC was pre-incubated with 20 ng of VEGF 165 in PBS containing 3% (w/v) BSA, or 25 mM HEPES at either pH 8.0, 7.0, or 6.0 [37], or 50 mM Tris-HCl (pH 8.0), PBS (pH 7.0), 50 mM sodium acetate (pH 6.0) for 1 hr at room temperature. Samples were subsequently blotted onto nitrocellulose, and blocked. Bound VEGF 165 was detected with anti-VEGF 165 antibodies (1 μg/ml in 3% (w/v) BSA in PBST). Primary antibodies were detected with anti-mouse IgG secondary antibodies conjugated to HRP and visualized as described for Western blotting. Binding was quantified by densitometry and expressed as mean density values (DV) from triplicate assays. Specific binding was determined by subtracting VEGF 165 background from total bound [18].

Capillary Tube-like Assay
Growth factor reduced (GFR) Matrigel was added to wells of ice-cold 96-well plates (70 μl/well) for 6 seconds. Excess was removed, leaving a thin coating. Plates were incubated for 6 minutes on ice, 20 minutes at room temperature, and finally warmed for 20 minutes at 37°C. Bone marrow endothelial cells were seeded (6,500 cells/well) in serum free RPMI 1640 media containing 1% (w/v) penicillin/streptavidin, 2 mM glutamax without growth supplements. After cell attachment, the media was replaced with media containing one or more supplements [i.e., PlnDI (12.5 μg/ml), untreated or predigested with a heparinase cocktail and/or chondroitinase ABC, heparin (4.0 μg/ml), VEGF 165 (20 ng/ml)]. For assays conducted in the absence of cell surface heparin sulfate, human bone marrow endothelial cells were cultured for 15 minutes under serum free conditions in RPMI 1640 media supplemented with heparinase cocktail [32]. Such treatments temporarily remove more than 95% of cell surface HS. Prior to seeding cells were washed twice with RPMI 1640 media.
To quantify tube-like formation cells were fixed (4% (v/v) paraformaldehyde) after 18 h, stained (SYTO13, Invitrogen, CA), then photographed with a SPOT CCD camera affixed to an inverted microscope equipped for epifluorescence. Nine random fields, representing 80% of each well, were analyzed for three angiogenic parameters: average tube length (defined as three or more cells connected lengthwise, and exceeding 100 μm in length; [47], number of tube-like structures, and the number of branch points, using Image J software (NIH). When several tube-like structures merged together or branched, the total length was calculated as the sum of the individual branches. All tube-like formation studies were conducted in quadruplicate wells, and repeated at least three times. Since the outcomes of each angiogenic parameter were similar only average tube length is reported. Note: All supplement concentrations employed herein are optimal, and were determined empirically over a broad range. As a control for enzyme activity, assays were also conducted with supplements containing heat inactivated chondroitinase ABC and/or heparinase cocktail.

Monosaccharide analysis
As done previously [46], PlnDI (20 μg) was hydrolyzed with 4 M HCl at 100°C for 6 h, then dried in a Speed-Vac. Residues were dissolved in HPLC grade water then analyzed on a CarboPac PA1 high pH anion-exchange column (4 × 250 mm) using Dionex BioLC HPLC coupled to a pulse amperometric detector.

Statistical analysis
All experiments were conducted in triplicate, repeated at least three times, and analyzed by two-tailed paired Student's t-test using GraphPad Prism version 5.0 for Windows (San Diego California USA). Differences were considered significant at P < 0.05. All results are presented as means ± standard error of the mean.