RGD-Dependent Vacuolation and Lumen Formation Observed during Endothelial Cell Morphogenesis in Three-Dimensional Fibrin Matrices Involves the α_vβ₃ and α₅β₁ Integrins

Three-Dimensional Fibrin System

Human umbilical vein endothelial cells were grown to confluence in M199 (Gibco-BRL, Grand Island, NY) supplemented with 20% fetal calf serum (Gibco-BRL) and bovine brain extract as described. 41 Before experiments, cells were rinsed in phosphate-buffered saline, trypsinized, and resuspended in Dulbecco’s minimum essential medium (DMEM) (Gibco-BRL) at a density of 5 × 10 6 cells/ml. Fibrinogen (Calbiochem, La Jolla, CA) was suspended at 20 mg/ml in DMEM. Final concentrations in each 25-μl gel included 10 mg/ml fibrinogen and 1 × 10 6 cells/ml in DMEM solidified with 1 μg thrombin (Amersham Pharmacia Biotech, Alameda, CA). Cultures were equilibrated at 37°C with 5% CO₂ before the addition of 100 μl media. DMEM was supplemented with 20% fetal calf serum, 40 ng/ml recombinant vascular endothelial growth factor (VEGF), and basic fibroblast growth factor (bFGF) (Upstate Biotechnologies, Lake Placid, NY); 200 units of aprotinin (American Diagnostica, Piscataway, NJ), 50 μg/ml ascorbic acid (Sigma); and 50 ng/ml 12-O-tetradecanoyl phorbol 13-acetate (TPA). Cultures were maintained for the times indicated in each experiment before being photographed or fixed, stained, and quantitated.

Time-Course Experiments

EC cultures were resuspended three-dimensionally in fibrin gels as described above, using DMEM without phenol red (Gibco-BRL). Random fields were selected, and sequential photographs of individual wells were taken at 0.5, 5, 7, 12, and 24-hour time points. In this experiment, we were able to observe the development of lumenal and vacuolar structures in the same culture at an identical focal plane.

Blockade of Endothelial Cell Vacuolation and Lumen Formation Using Integrin Antibodies

To determine the integrins involved in EC vacuolation, various antibodies directed toward human integrin subunits and heterodimers were added (20 μg/ml). The antibodies included anti-β₁ (MAb13; Becton-Dickinson, Bedford, MA), 42 anti-α₅ (MAb16; Becton-Dickinson), 42 anti-α₂ (A2-IIE10; Upstate Biotechnologies, Lake Placid, NY), 43 anti-α_vβ₃ (LM609; Chemicon, Temecula, CA), 44 and anti-α_vβ₅ (P1F6; Chemicon). 45 Furthermore, in some experiments, a cyclic Arg-Gly-Asp (cRGD) peptide (GPenGRGDSPCA; Gibco-BRL) and its control, Gly-Arg-Gly-Glu-Ser-Pro (GRGESP; Gibco-BRL), was added to both the gel and medium at 500 μg/ml.

Induction of Regression: Collapse of Vacuolation and Lumen Formation

To determine whether integrins expressed by ECs were involved in the maintenance of vacuolar and lumenal structures, cultures were allowed to develop to either 12- or 24-hour time points before we attempted to collapse the structures. Medium was removed and replaced with fresh medium containing either the cRGD or RGE peptides alone or combined with anti-α₅ or anti-α₂ integrin antibodies. Cultures were allowed to regress for 3 hours, at which time medium was removed and wells were fixed with 3% glutaraldehyde in phosphate-buffered saline (PBS) (Sigma, St. Louis, MO) overnight.

Quantitation of Endothelial Vacuolation and Lumen Formation

Cultures were fixed at various time points by removing the medium and replacing it with 3% glutaraldehyde in PBS overnight at 4°C. Wells were washed with water and stained with 1% toluidine blue (Sigma) containing 2% sodium borate (Sigma) for 5 minutes at room temperature. Cultures were destained and photographed and/or quantitated.

Percentage vacuolation was determined by analyzing 200 cells for each condition. A cell was considered to be vacuolating if >30% of the cell’s area contained a lumen or vacuole. As cultures progressed to later stages, we observed that the cells coalesced. If a network of cells was encountered during quantitation, the number of nuclei in the network determined the number of vacuolating cells or lumen-forming networks, as toluidine blue prominently stains nuclei.

Embedding and Sectioning of Fibrin Gels

Fibrin gels at various time points were fixed overnight in 3% glutaraldehyde in PBS. After sequential ethanol washes, gels were stained with 0.1% toluidine blue in ethanol. Gels were embedded and prepared for sectioning with a Polybed 812-BDMA Embedding Kit (Polysciences, Warrington, PA) according to the Glauert method, 46 as instructed by the manufacturer. Sections were stained with 1% toluidine blue and photographed.

Endothelial Cell Vacuolation in Three-Dimensional Fibrin Matrices Regulates Lumen Formation

Human umbilical vein endothelial cells were used to investigate endothelial morphogenesis in a three-dimensional fibrin matrix. ECs were suspended in a solution of fibrinogen, and a fibrin matrix was formed after the addition of thrombin. Cultures were allowed to undergo morphogenesis over a 24-hour period. Photographs of the same culture and focal plane taken at various times (Figure 1) ▶ revealed that suspended cells began forming vacuoles at approximately 3–5 hours of culture. These vacuoles grew in size over time and merged to form lumenal structures. Cultures fixed at various times during morphogenesis were embedded in a plastic matrix. Cross-sectional analysis revealed the presence of intracellular vacuoles and lumens (Figure 2) ▶ . The development of intracellular vacuoles and the transition of these into larger lumenal structures are evident.

Endothelial Cell Vacuolation and Lumen Formation in Three-Dimensional Fibrin Matrices Is RGD-Dependent and Involves Both the α_vβ₃ and α₅β₁ Integrins

The influence of cRGD and control peptides either added alone or in combination with blocking antibodies directed to the α₅ or α₂ integrin subunits were assessed in time-course experiments. Combining cRGD peptide with anti-α₅ integrin blocking antibodies resulted in complete blockade of vacuolation and subsequent lumen formation at all time points. As shown in Figure 3 ▶ , 85–90% of the ECs within the fibrin matrix developed vacuoles and lumens. These morphogenic events became apparent at approximately 3 hours of culture and reached maximum levels at 12 hours. Adding cRGD peptide alone blocked approximately 50% of vacuolation and lumen formation. Combining cRGD with anti-α₂ integrin blocking antibodies did not further increase the effects of cRGD. The addition of RGE peptide either alone or combined with anti-α₂ antibodies had no effect when compared with control, while anti-α₅ blocking antibodies had a slight effect at later time points. Toluidine blue-stained cultures of control versus cRGD/anti-α₅-treated cultures were photographed at various time points (Figure 4) ▶ . Extensive intracellular EC vacuolation and lumen formation was observed with time in the control cultures, whereas complete blockade of these morphogenic changes occurred in the presence of cRGD and α₅ antibodies.

Cyclic RGD peptides are known to inhibit the function of many α_v integrins while showing minimal effects on other RGD-binding integrins such as α₅β₁. 47,48 To assess the extent to which different α_v integrins expressed by ECs were involved in vacuolation and lumen formation, various integrin-blocking antibodies directed toward integrin heterodimers and subunits were added (Figure 5) ▶ . Although none of the various anti-integrin antibodies alone were able to block lumen formation, the combination of anti-α₅ and anti-α_vβ₃ integrin antibodies completely inhibited vacuolation and subsequent lumen formation. The addition of anti-α_vβ₅ with anti-α₅ and anti-α_vβ₃ did not further inhibit this effect. The addition of anti-α_vβ₅ with anti-α₅ had little effect compared to control. Combining anti-α₂ antibodies with anti-α_vβ₃ did not further the effects of anti-α_vβ₃ alone, ruling out the involvement of α₂β₁ in the fibrin system. These data confirm that together the α_vβ₃ and α₅β₁ integrins regulate EC vacuole and lumen development in three-dimensional fibrin matrices. Moreover, blockade of vacuole formation completely inhibited lumenal development. This evidence indicates that lumen formation is preceded by vacuolation in this system and that intracellular vacuolation is a key step in lumen formation.

Regression of Pre-Existing Endothelial Cell Vacuoles and Lumens in Fibrin Matrices Occurs after the Addition of α_vβ₃ and α₅β₁ Antagonists

To determine whether blocking integrin function or ligation affected preformed vacuolar or lumenal structures, various monoclonal anti-integrin blocking peptides and antibodies were added to developed cultures (Figure 6) ▶ . At both 12- and 24-hour time points, we observed that combinations of cRGD peptide and anti-α₅ integrin antibodies induced nearly complete collapse of existing structures, resulting in minimal vacuolation and lumen formation. The addition of cRGD peptides alone or in combination with anti-α₂ integrin antibodies reduced vacuolation to approximately 50%. The RGE peptide combined with anti-α₅ integrin antibodies induced a small amount of regression, reducing vacuolation to approximately 70% at both time points. Addition of the control peptide, RGE, alone or combined with anti-α₂ integrin antibodies did not induce regression and had no effect on vacuolation or lumen formation. Photographs were taken of stained cultures at the 24-hour time point (Figure 7) ▶ . The addition of cRGD peptides and α₅ antibodies induced complete regression and collapse of existing lumenal structures as compared to control. The RGE peptide and α₂ antibodies had no effect. These data indicate that α_vβ₃ and α₅β₁ not only regulate formation of vacuoles and lumens, but are necessary for the maintenance of these structures.

A Case for Dual Integrin Control of Angiogenesis

ECs express the α₁β₁, α₂β₁, α₃β₁, α₅β₁, α₆β₁, α_vβ₁, α_vβ₃, and α_vβ₅ integrins. 49,50 The α₂β₁, α₁β₁, α_vβ₃, and α_vβ₅ integrins have been reported to be involved in angiogenesis both in vivo and in vitro. 15,50-57 Antibody and peptide blocking data in all studies revealed that blocking integrin function resulted in the blockade of angiogenesis. Recent evidence has revealed that the α_vβ₃ integrin is important for angiogenesis, based on the ability of LM609, a monoclonal antibody directed to the α_vβ₃ heterodimer or cyclic RGD peptides, to block angiogenesis. 51-55 However, concern about α_vβ₃ being primarily responsible for the development of angiogenesis has been raised from recent analyses of α_v integrin knockout mice. Although removal of the α_v integrin subunit was lethal, considerable vascular development occurred. 58 One interpretation of this may be that the α_vβ₃ integrin functions to modulate the angiogenic response but is not solely responsible for it, and the addition of α_vβ₃ antagonists provides inhibitory signals that limit the response. It is clear from in vitro models that multiple integrins can regulate EC morphogenesis, and that it appears to depend on the matrix environment. 15,50 For example, α₂β₁ regulates morphogenesis in a collagen type I matrix, 15 α₆β₁ regulates morphogenesis in a laminin-rich matrix, 50 and α_vβ₃ and α₅β₁ regulate morphogenesis in a fibrin matrix as reported here. Thus it appears that multiple integrins, presumably through common signaling pathways in ECs, can regulate the same critical steps in EC morphogenesis, such as migration, sprouting, invasion, vacuole formation, and lumen development. In support of these conclusions are in vivo studies defining the involvement of multiple integrins (eg, α_vβ₃, α_vβ₅, α₁β₁, and α₂β₁) during angiogenesis. 51-57 The matrix environment through which EC morphogenesis proceeds in vivo is likely to be variable, depending on whether the process is occurring during embryonic development, during physiological events such as the female reproductive cycle, or following tissue injury (eg, wound repair and tumorigenesis). During these latter events, a prominent provisional matrix consisting of fibrin and fibronectin is observed. 20 Thus, depending on the circumstances, varying matrix environments may alter the relative contributions of particular integrins in the control of angiogenesis or vasculogenesis.

The finding that multiple integrins regulate EC morphogenesis raises the concept that cooperative interactions and signaling through two or more integrins may be important in these events. Cooperation between integrins has previously been reported for α₄β₁ and α₅β₁ in leukocytes, 59 as well as with α_vβ₃ and α₅β₁ in endothelial cells, 60 a transfected Chinese hamster ovary cell line, 61 and K562 cells. 62 In one study on AIDS-related Kaposi sarcoma, HIV-1 Tat protein binding to α_vβ₃ and α₅β₁ integrins stimulated EC migration, basement membrane degradation, and morphogenesis in vitro. 60 A second study reported that cooperativity between the α₅β₁ and α_vβ₃ integrins was involved in controlling cell migration. 61 Another study using K562 cells transfected with α_vβ₃ determined that antibody blockade of the α_vβ₃ integrin also completely blocked α₅β₁-mediated phagocytosis. 62 Pharmacological data from the same study suggested that ligation of α_vβ₃ blocked α₅β₁ signaling through a protein kinase C pathway. Taken together, these studies support that α_vβ₃-α₅β₁ integrin cooperation and signaling may play a role in the regulation of many cellular activities, including morphogenesis.

Integrin Control of Endothelial Cell Vacuolation and Lumen Formation

The cytoskeletal alterations associated with EC morphogenesis have yet to be elucidated. Recently, however, much progress has been made in this area. 63 What is clear is that outside-in signaling through integrins, which tether the cell’s actin cytoskeleton to the ECM, is one mechanism of controlling cell shape changes. 64 Recent work in our laboratory involving EC morphogenesis in three-dimensional collagen matrices demonstrated α₂β₁-dependent vacuole, lumen, and network formation, 15 which was associated with up-regulation of gelsolin, vasodilator-stimulated phosphoprotein (VASP), and profilin, three proteins that regulate actin cytoskeletal reorganization and signaling. 65 In the former study, anti-α₂β₁ integrin antibodies completely blocked vacuole formation and completely interfered with subsequent lumen and capillary tube formation. Here our data support the same conclusion that when vacuole formation is blocked, subsequent lumenal development is prevented. In the previous study using collagen matrices, EC intracellular vacuole formation was shown to be a pinocytic process that required both actin microfilaments and microtubules. In addition, a number of proteins associated with vesicular trafficking in cells including caveolin and annexin II were found to be associated with the vacuole compartment. 15 These data, combined with the work in this report, indicate that intracellular vacuoles represent a novel intracellular compartment regulating the lumen formation step in EC morphogenesis.

Previous data 15 and the data reported here show that integrins are required for the formation of intracellular vacuoles. Interestingly, the integrin-blocking reagents show no overlapping inhibitory influence in that anti-α_vβ₃ and -α₅β₁ antibodies block EC morphogenesis in a fibrin matrix, but not in collagen, 15 while the opposite is true for anti-α₂β₁ antibodies. We have shown here that α_vβ₃ and α₅β₁ integrins together control EC vacuolation and lumen formation in fibrin. Thus this work provides new evidence for a previously undescribed role for the α_vβ₃ and α₅β₁ integrins in EC vacuolation and lumen formation.

It is not entirely clear which ECM ligands are involved in α_vβ₃- and α₅β₁-mediated EC morphogenesis in fibrin matrices. The fibrin matrix environment in the current study contains fibrin as well as serum-derived adhesive molecules such as fibronectin and vitronectin. Which of these molecules is most relevant to the EC morphogenic events remains to be determined. The α₅β₁ integrin is known to bind fibronectin, 66 but also has recently been reported to bind fibrinogen. 67 In addition, α_vβ₃ is known to bind fibrinogen, fibronectin, and vitronectin, 64 all of which are present in our assays. The important point to be made is that the three-dimensional fibrin matrix system described in this report mimics the fibrin-fibronectin provisional matrices known to be prominent with the wound microenvironment where angiogenesis occurs. In any case, our study indicates that α_vβ₃ and α₅β₁, both receptors for this provisional matrix, together control EC morphogenesis in this matrix environment.

Involvement of Integrins in Vascular Regression

A final point of this discussion deals with the relevance of this work as it relates to mechanisms of vascular regression in either tumor- or wound-derived provisional matrices. We present here evidence that not only are α_vβ₃ and α₅β₁ involved in EC morphogenesis (ie, the formation of vacuoles and lumens), but ligation of these integrins is also necessary for maintenance of the cell’s architecture and three-dimensional structure. Both α_vβ₃ and α₅β₁ have been discussed as targets for anti-angiogenic therapy. 68,69 Some of the regression effects that we describe may be relevant to the known anti-angiogenic influence of α_vβ₃ integrin antagonists in vivo. 51-53 These antagonists are known to induce regression of angiogenic blood vessels through an apoptotic mechanism, and they may act in part to induce regression of angiogenic vessels present in a fibrin-rich provisional matrix environment. The α₅β₁ integrin has been reported as a ligand for fibronectin 66 and fibrinogen 67 and regulates cell-mediated retraction of fibronectin-fibrin matrices. 70 This information, combined with our data, supports the possibility that α_vβ₃ antagonists in combination with α₅β₁ antagonists may be more effective at inducing vascular regression than α_vβ₃ antagonists alone. Further work is needed to investigate this issue.

It is clear that EC morphogenesis and angiogenesis are complicated processes. The involvement of particular integrins in these events is dictated by multiple signals, including EC integrin expression levels, ECM content, and the source of the angiogenic stimulus. The system described here is an experimental method of investigating the pathways transmitted through α_vβ₃ and α₅β₁ integrins that result in EC vacuole formation and morphogenesis in three-dimensional fibrin matrices. Elucidating the molecular events responsible for EC morphogenesis in fibrin matrices will provide important clues to a better understanding of how blood vessel formation and regression are regulated.