Learn more: PMC Disclaimer | PMC Copyright Notice
From the Cover
Angiopoietin-2 displays VEGF-dependent modulation of capillary structure and endothelial cell survival in vivo
Abstract
Modulation of Tie2 receptor activity by its angiopoietin ligands is crucial for angiogenesis, blood vessel maturation, and vascular endothelium integrity. It has been proposed that angiopoietins 1 (Ang1) and 2 (Ang2) are pro- and anti-angiogenic owing to their respective agonist and antagonist signaling action through the Tie2 receptor. The function of Ang2 has remained controversial, however, with recent reports suggesting that in some circumstances, it may be pro-angiogenic. We have examined this issue using the transient ocular microvessel network called the pupillary membrane as a unique in vivo model for studying the effects of vascular regulators. We show that in vivo, in the presence of endogenous vascular endothelial growth factor (VEGF)-A, Ang2 promotes a rapid increase in capillary diameter, remodeling of the basal lamina, proliferation and migration of endothelial cells, and stimulates sprouting of new blood vessels. By contrast, Ang2 promotes endothelial cell death and vessel regression if the activity of endogenous VEGF is inhibited. These observations support a model for regulation of vascularity where VEGF can convert the consequence of Ang2 stimulation from anti- to pro-angiogenic.
Angiogenesis is the process of formation of new capillaries from preexisting blood vessels (1) and is an essential component of embryogenesis, normal physiological growth, repair, and tumor expansion. Although a variety of factors can modulate endothelial cell (EC) responses in vitro and blood vessel growth in vivo, only vascular endothelial growth factor (VEGF) family members and the angiopoietins are believed to act almost exclusively on vascular ECs (2, 3). Multiple members of the VEGF family can bind to the VEGF receptors Flt-1, Flk-1/KDR, and Flt-4 and regulate the endothelial cell proliferation and migration that are essential elements of the angiogenic process (2, 4). EC-specific signaling via the Tie2 receptor by the angiopoietins is also crucial for angiogenesis, but in contrast, angiopoietins are believed to modulate blood vessel maturation and maintenance of endothelium integrity (2, 3).
Two ligands for Tie2, angiopoietins 1 and 2 (Ang1 and Ang2) were originally identified in tissue culture experiments as agonist and antagonist, respectively (5–7). Ang1 was shown to support EC survival and to promote endothelium integrity (5, 7), whereas Ang2 had the opposite effect (6) and promoted blood vessel destabilization and regression in the absence of the survival factors VEGF-A or basic fibroblast growth factor (8). Ang1 and Ang2 were also shown to only weakly influence EC proliferation (9). The characterization of Ang1 and Ang2 as agonist and antagonist was based on the ability of Ang2 to bind the Tie2 receptor and to inhibit the pro-angiogenic action of Ang1. However, a number of studies of Ang2 function have suggested a more complex situation.
Corneal pocket assays have shown that both Ang1 and Ang2 had similar effects acting synergistically with VEGF-A to promote growth of new blood vessels (10). This tended to suggest that in this context, Ang2 was pro-angiogenic. The possibility that there was a dose-dependent endothelial response was raised by the observation that in vitro at high concentration, Ang2 can also be pro-angiogenic (11). Furthermore, if ECs were cultivated on fibrin gel (a substrate that stimulates EC differentiation), activation of Tie2 with Ang2 was also observed (12), perhaps suggesting that the action of Ang2 could depend on EC differentiation state. In microvascular EC cultured in a three-dimensional collagen gel, Ang2 can also induce Tie2 activation and promote formation of capillary-like structures (13). At the same time, in vivo, Ang2 is expressed during development at sites where blood vessel remodeling is occurring (6), as well as in highly vascularized tumors (14, 15). These two observations might suggest that Ang2 is involved in promoting blood vessel plasticity.
To examine this question further, we have developed the pupillary membrane (PM) as a unique in vivo model system. This structure is a temporary vascular network that surrounds the anterior part of the lens in the developing eye (16). In humans, the PM is present only during embryogenesis as it regresses during the third trimester (17). In rodents, regression occurs in the second week after birth. Being situated in the anterior chamber of the eye, the PM can be visualized vitally (18) and is accessible for manipulation in vivo via transcorneal injection (19). Because the PM is composed of a two-dimensional array of capillaries that can be rapidly dissected from the eye, this structure is uniquely suited to test the immediate response of microvessels to angiogenic modulators in vivo. In this study, we compared the responses of PM capillaries to injected Ang2 and VEGF-A as well as to their antagonists. In the presence of endogenous VEGF, Ang2 had a complex effect and efficiently induced increased blood vessel diameter, remodeling of the basal lamina, EC proliferation, migration, and sprouting. If endogenous VEGF activity was inhibited, Ang2 effectively promoted capillary regression. This study demonstrates that in vivo, Ang2 can stimulate angiogenesis or capillary regression depending on the presence of VEGF and also establishes the PM as a useful in vivo model for assessing vascular modulators.
Materials and Methods
Reagents, Antibodies, and Animals.
Reagents were purchased as follows. Biotinylated human recombinant Ang2 was purchased from Calbiochem, and mouse recombinant VEGF (the 164-aa isoform), recombinant human Flt-1/Fc chimera, and Tie2 blocking antibody from R & D Systems (Minneapolis). Anti-VE-cadherin antibody was purchased from Santa Cruz Biotechnology. Anti-BrdUrd and anti-smooth muscle actin antibodies were obtained from Sigma. HUI77 monoclonal antibody against a collagen cryptic domain has been described (20). Secondary antibodies labeled with Alexa Fluor 488 or 568 were from Molecular Probes. Timed pregnant Sprague–Dawley rats were obtained from Taconic Farms and housed in accordance with institutional guidelines.
Transcorneal Injections, Indirect Immunofluorescent Staining, and Imaging.
Using techniques previously described (19), transcorneal injection of growth factors was performed at day 5 after birth [A5; 2 days before the normal onset of pupillary membrane regression (21)] or at day 8 when the rate of apoptosis was high (21). Growth factors or soluble growth factor receptor/Fc chimera was dissolved in PBS with 0.1% BSA. For mock injections, 0.1% BSA solution in PBS was used. The volume of the anterior chamber was estimated to be approximately 10 μl. The injected volume of growth factor solution was 0.5–0.05 μl, suggesting an immediate 10–200× dilution. Twenty-four hours after growth factor injection, animals were killed, perfused with 4% paraformaldehyde in PBS, and the eyeballs were enucleated. Dissection of the PM was performed according to previously established techniques (22). PMs mounted on glass microscopy slides were permeabilized with 0.05% Triton/PBS, washed with PBS, and immunostained with primary antibodies at 1:100 dilution (anti-VE-cadherin antibody) or 1:500 dilution (anti-smooth muscle actin and HU177 antibodies) followed by fluorescently labeled secondary antibodies at a 1:500 dilution. Membranes were counterstained with Hoechst 33258. Images were taken using a Zeiss Axioplan microscope and a Sony DKC5000 digital camera. Figures were assembled using CANVAS and ADOBE PHOTOSHOP software.
Labeling of Apoptotic and Proliferating Cells.
Apoptotic cells were recognized according to the characteristic nuclear fragmentation after staining with Hoechst 33258. Mitotic cells were counted based on their characteristic chromatin morphology after they were stained with Hoechst 33258. Alternatively, to quantitate the proliferative response of capillary cells, BrdUrd was injected i.p. at either 1 or 13 h after Ang2 injection. Twenty-four hours after Ang2 injection, PMs were dissected and stained with anti-BrdUrd antibody. Adjacent BrdUrd-positive cells with the same intensity of labeling were considered recently divided daughter cells. In contrast, isolated labeled cells or those that were adjacent but of distinct labeling intensities have presumably not progressed through mitosis (for more details see Fig. Fig.11 B and C). BrdUrd labeling was performed as described (21). For quantitation of BrdUrd labeling, the total number of labeled cells per membrane was determined.
Ang2 promotes cell division in the pupillary membrane. (A) Graph showing the number of mitotic capillary cells in the pupillary membrane after injection of BSA (blue bar), VEGF-A (green bar), or Ang2 (red bar). (B) Experimental timeline and graph describing the increased BrdUrd labeling of pupillary membrane cells 24 h after injection of BSA (blue bars) or Ang2 (red bars). Error bars in A and B are standard errors. Individual and paired BrdUrd-positive cells were counted according to the labeling patterns observed in micrographs (C and D) in which PM cells were exposed to BSA (C) or Ang2 (D). The white arrows indicate pairs of daughter cells and the green arrowheads individual labeled cells. Other labeled nuclei belong to macrophages. Images at ×200.
Results
Ang2 Promotes Proliferation of Capillary ECs.
To examine the influence of Ang2 on microvessels in vivo, we injected recombinant human Ang2 into the anterior chamber of rat pups at 5 days after birth (A5). For comparison, we injected recombinant human VEGF-A, and mock injections were performed as a negative control. In cell culture assays, the optimal concentration of VEGF-A to produce proliferation or survival responses is about 10 ng/ml, whereas the optimal concentration of Ang1 or Ang2 that invokes a response is more than 100 ng/ml (9). For this reason, we injected a 10-fold higher amount of Ang2 compared with VEGF. In initial experiments, we injected 1 ng of Ang2 and 0.1 ng of VEGF-A and observed modest responses (data not shown). In a second series of experiments, we injected 10 ng of Ang2 or 1 ng of VEGF-A (the 164-aa disulfide-linked homodimeric isoform) and saw more dramatic changes. Injections were performed at 3 p.m. on day 1 of the experiment, with fixation and dissection performed 24 h later.
Despite the expectation that Ang2 alone might mediate capillary regression, we did not observe increased numbers of capillary cells with apoptotic morphology (data not shown). On the contrary, when the injections were performed at 5 days after birth (A5) (before the normal onset of PM regression), 10 ng of Ang2 stimulated proliferation of ECs. In quantifying this response, we counted mitotic figures in control, Ang2 treated, and VEGF-treated PMs. Injections of Ang2 resulted in a significant increase in the number of mitotic cells in blood vessels of the PM (Fig. (Fig.11A). At the levels of factors used, VEGF-A (1 ng) produced an 8-fold and Ang2 (10 ng) an 18-fold increase.
To further investigate the proliferation response, BrdUrd labeling experiments were performed according to the timeline in Fig. Fig.11B. In this experiment, PMs were harvested 24 h after Ang2 injection, with BrdUrd labeling at 13 h. The timing of Ang2 and BrdUrd injection was based on previous characterization of the cell cycle duration in PM cells (21). Cells incorporating BrdUrd were detected by immunofluorescence labeling. Because the PM is a whole-mount preparation, capillary cells that have divided after BrdUrd incorporation are apparent as pairs of nuclei with matching labeling intensity. Matching labeling intensity of adjacent nuclei are two important criteria for identifying cells that have divided after BrdUrd incorporation, as some labeled cells will arise in close proximity purely by chance. Thus, labeled nuclei are scored as “singles” if there is no adjacent labeled nucleus of the same intensity (Fig. (Fig.11 B–D, green arrowheads) and as “pairs” if this criterion is met (Fig. (Fig.11 B–D, white arrowheads).
We have previously used this scoring technique to assess the cycling behavior of PM cells (21), and this has revealed that capillary cells are a cycling population in the absence of exogenous growth factor stimulation. With the described protocol, the control BSA injection resulted in approximately equal numbers of single and paired BrdUrd-labeled cells. Untreated PMs show an identical response (data not shown). By contrast, Ang2-treated PMs showed a 3-fold increase in the number of paired BrdUrd-labeled cells and a 2-fold decrease in the number of single labeled nuclei (Fig. (Fig.11B). The total number of BrdUrd-labeled cells was increased 1.9-fold in Ang2-treated PMs. This suggests that Ang2 stimulates cells of the PM to enter the cell cycle and to proceed through mitosis.
Ang2 Stimulates a Rapid Change in EC Shape and Increased Capillary Diameter.
Interestingly, transcorneal injection of 10 ng of Ang2 also induced a rapid increase in the diameter of capillaries. This was obvious when comparing control and Ang2-treated PMs 24 h after injection (Fig. (Fig.22 A and B) where the vessel diameter had dramatically increased. Measurement indicated that the width of the flattened tubes present in the whole-mount preparations had increased 2.1-fold in Ang2-treated examples (Fig. (Fig.22C). The increased diameter of PM capillaries apparent when comparing control (Fig. (Fig.22D) and Ang2-treated PMs (Fig. (Fig.22E) was eliminated if an anti-Tie2 blocking antibody (50 ng) was co-injected (Fig. (Fig.22F). This indicated that the response to Ang2 was mediated through the conventional signaling pathway. Injection of anti-Tie2 blocking antibody alone had no effect on capillaries. When 0.1 and 1 ng of Ang2 was injected, a less dramatic response was observed, indicating that capillary diameter increase was dose-dependent (data not shown).
Ang2 induces rapid capillary diameter increase and EC shape change. Micrographs of whole-mount control (A) and Ang2-treated (B) PMs showing capillary diameter increase at 24 h (×100). (C) Graph showing the measured increase in capillary diameter of Ang2 (red bar) vs. BSA-treated (blue bar) pupillary membrane capillaries. Error bars are standard errors. (D–G) Higher magnification (×400) paired brightfield and Hoechst 33258-labeled fluorescence micrographs showing PMs exposed to BSA (D), Ang2 (E), Ang2 + anti-Tie2 antibodies (F), and VEGF-A (G). (E) Increased vessel diameter and cell number is apparent; (F) no change is observed; and (G) increased cell number is apparent in the absence of vessel diameter increase. (H–K) Fluorescence micrographs showing anti-VE-cadherin antibody staining (green) for control (H) and Ang2-treated PMs at 4 (I), 12 (J), and 24 h (K) after Ang2 injection. The progressive change in cell shape is apparent from 4 h onward. This is emphasized by the typical cell outlines that are produced in red both on the cell image and projected onto a black area of the micrograph. (H–K) Images at ×630.
Ang2 also invoked a response distinct from that of VEGF. This was most obvious when comparing the density of nuclei present in control (Fig. (Fig.22D), Ang2 (Fig. (Fig.22E), or VEGF-A (Fig. (Fig.22G) treated PMs. For this comparison, we increased the level of VEGF-A injected to 10 ng, but even at this level, VEGF-A did not induce an increase in vessel diameter. As a consequence of the proliferation induced, the density of nuclei appeared higher in VEGF-treated capillaries. This response could be quantified by measuring capillary area and counting the number of nuclei within a given area. This showed that, in arbitrary units (expressed as the value ± the standard error), VEGF treatment produced the greatest increase at 23.9 ± 2.6 nuclei per unit area, whereas control and Ang2 injection gave similar values at 14.1 ± 0.9 and 12.4 ± 1.3 nuclei per unit area, respectively.
In seeking an explanation for the observed increase in vessel diameter upon Ang2 treatment, we determined whether this might be the result of a rapidly induced change in cell shape. Thus, we performed a second series of Ang2 injection experiments but examined PMs at 4, 12, and 24 h and labeled for the junctional protein VE-cadherin using indirect immunofluorescence. The labeling of EC junctions gives a good indication of cell shape.
This analysis showed that the change in capillary diameter was rapid and could first be observed as early as 4 h after Ang2 injection (compare Fig. Fig.22H, control, with 2I, Ang2 treated). The difference in capillary diameter became more apparent 12 h after Ang2 injection (Fig. (Fig.22J) and was maintained at 24 h (Fig. (Fig.22K). Over the 24-h time course, ECs exposed to Ang2 lost their conventional elongated shape (Fig. (Fig.22H) and became broader (Fig. (Fig.22 I–K). At 24 h, VE-cadherin junctional labeling was weaker and more convoluted. The appearance of capillaries of increased diameter within 4 h of Ang2 injection suggested that the cause was unlikely to be cell division, a proposal consistent with the changes in EC shape apparent from junctional labeling. It is likely that at later time points, both the changes in cell shape and increased proliferation induced by Ang2 contribute to increased capillary diameter. This is evidence that Ang2 induces rapid increases in capillary diameter.
We also addressed the question of whether pericytes might modulate the effects of Ang2 on capillary ECs (23). In the PM, thicker capillaries located at the membrane periphery contain many pericytes that can be labeled with antibodies against smooth muscle actin. However, thinner capillaries closer to the center of membrane lack pericytes. In both types of capillaries, Ang2 induced similar changes in EC shape, and thus the vessel diameter increased (Fig. (Fig.3).3). This result suggests that Ang2 acts directly on ECs.
Ang2 induces pericyte-independent EC shape changes and remodeling of the basal lamina. Control (A) and Ang2-treated (B) PM preparations were labeled with Hoechst 33258 for nuclei (blue) and stained with anti-VE-cadherin (green) and antismooth muscle α-actin antibodies (red). This shows that Ang2-induced shape changes occur in capillaries with and without pericytes. Images at ×630. Control (C) and Ang2-treated (D) PM preparations were labeled with Hoechst 33258 for nuclei (blue) and with the HUI77 antibody against a cryptic collagen epitope (red). The higher level of signal present in the dilated capillaries shown in D indicates that the basal lamina is remodeled in response to Ang2. Images at ×400.
Ang2 Induces Remodeling of Capillary Basal Lamina.
Changes in capillary diameter after Ang2 injection suggested remodeling of the basal lamina. To determine whether this was the case, we used the HUI77 monoclonal antibody that recognizes a collagen cryptic domain that becomes accessible only in denatured or proteolytically cleaved collagen (20). HUI77 antibody recognizes different collagen types including the interstitial matrix collagen type I and the blood vessel basal lamina collagen type IV (20). Virtually no staining occurred in the control membranes (Fig. (Fig.33C), whereas in Ang2-treated membranes there was staining at the edges of blood vessels (Fig. (Fig.33D). The lack of significant staining in the control and intense labeling of the Ang2-treated sample indicate that dramatic collagen remodeling is taking place.
Ang2 Stimulates EC Sprouting and Migration.
A significant response to Ang2 injection was the migration of ECs from blood vessels. Migrating cells could be detected if Ang2-treated PMs were labeled fluorescently for VE-cadherin (Fig. (Fig.4).4). Macrophages were distinguished by their characteristic morphology and lack of VE-cadherin junctional labeling (Fig. (Fig.4,4, red dots). Ang2-treated membranes show multiple ECs extending away from the capillaries and onto the anterior surface of the lens capsule (Fig. (Fig.44A, red dashed lines, and B–F). In these examples, VE-cadherin has characteristic junctional labeling that appears either as a single intense line in mature capillaries (Fig. (Fig.44 A–F) or with a linear, punctate pattern where junctions are immature (Fig. (Fig.44 A–D, white arrowheads). Cytoplasmic VE-cadherin labeling also revealed that many of the migrating ECs had very fine cytoplasmic extensions (Fig. (Fig.44 C–E, orange arrowheads) that appeared to follow the extracellular matrix fibers that are part of the PM and lie on the surface of the lens capsule (18). Migrating EC also formed intercapillary monolayers on the lens capsule surface (Fig. (Fig.44B). In some cases, ECs migrating from existing capillaries in response to Ang2 had the morphology of angiogenic sprouts (Fig. (Fig.44F), although this was relatively infrequent. Sprouting ECs were elongated with multiple pseudopodia and had VE-cadherin labeling both at the pseudopodial tips and at junctions with ECs remaining within the capillary (Fig. (Fig.44D).
Ang2 stimulates EC migration and sprouting. Micrographs show Ang2-treated PM preparations labeled fluorescently with Hoechst 33258 for nuclei (blue) and anti-VE-cadherin antibody (green). In all panels, cells that can be identified morphologically as macrophages are labeled with a red dot. ECs are identified through junctional immunoreactivity with anti-VE-cadherin antibodies. These panels show typical examples of membranes where ECs have migrated from the confines of the original capillaries. (A) The outline of capillaries is indicated by the white dashed line and groups of cells that have undergone migration by the red dashed lines. These lines are omitted from other panels to afford an unobstructed view. Cells that have migrated form either small groups (A, dashed red lines, and C–E) or complete monolayers between capillaries (B and D). ECs that have migrated often show junctional labeling that is punctate rather than the continuous line found in capillaries (A, B, and D, white arrowheads for examples). Responsive ECs feature cell extensions that label at a low level with anti-VE-cadherin antibodies (C–E, orange arrowheads). In some membranes (F), EC sprouting can be identified where a VE-cadherin junction-positive cell (red arrow) has extended pseudopodia (white arrowheads). (A–E) Images at ×630. (F) Image at 400×.
VEGF Is Required for Survival of EC in Capillaries of the PM.
The PM normally starts to regress between A5 and A7 in the rat (19). At A5 and A6 when our injections and dissections were performed, there are only a few apoptotic cells in capillaries of the PM. In contrast, by A9 the number of apoptotic cells in the PM is dramatically increasing up to approximately 40 per membrane (19). At this stage, many of the capillaries show the characteristic synchronous pattern of apoptosis (22). VEGF-A and Ang2 injections performed at A9 followed by dissections after 24 h revealed that both VEGF-A and Ang2 had a pro-survival effect, reducing the number of the regressing apoptotic capillaries (Fig. (Fig.55A).
Capillary regression is induced by VEGF inhibition and promoted by Ang2. (A) Histogram showing the number of apoptotic capillary cells during the normal regression phase at A9 after BSA injection (blue bar), VEGF injection (green bar), or Ang2 injection (red bar). (B) Histogram showing the number of apoptotic capillary cells during the normal regression phase at A5 after injection of 0.5 μl of the PBS/0.1% BSA vehicle (blue bar), 1 ng of VEGF (green bar), combined 1 ng of VEGF + 20 ng of Flt-Fc (green/yellow bar), 10 ng of Ang2 (red bar), combined 10 ng of Ang2 + 20 ng of Flt-Fc (red/yellow bar), and 20 ng of Flt-Fc (yellow bar). Error bars in A and B are standard errors. (C and D) Paired differential interference contrast (Upper) and Hoechst 33258 fluorescent staining (Lower, false colored green) micrographs showing PMs exposed to BSA (C, control) and Ang2 + Flt-Fc (D). Cells with apoptotic nuclear morphology are indicated (Lower) by the red arrowheads and the pattern of apoptotic cell distribution transferred to the differential interference contrast image. (Insets) A higher magnification of the indicated areas to emphasize apoptotic morphology. Within the Insets, arrows point to apoptotic bodies. Main images at ×400.
To determine whether an endogenous VEGF activity modulated the Ang2-induced capillary changes, we tested the effect of a soluble Flt receptor (Flt-Fc chimera) that can bind VEGF and inhibit its activity. Interestingly, when Flt-Fc was injected alone at A5, before the beginning of normal regression, it induced a dramatic response; 24 h after the injection, the number of apoptotic capillary cells was increased approximately 25 times in comparison with the controls (Fig. (Fig.55B). This indicates that VEGF is crucial for the survival of EC in this structure. When Flt-Fc was co-injected with VEGF-A, no increase in apoptosis occurred, suggesting that the injected VEGF-A could effectively counter the effects of Flt-Fc. When Flt-Fc was co-injected with Ang2, the number of apoptotic cells was increased approximately 48-fold compared with BSA-treated controls (Fig. (Fig.55B). Thus, unlike the pro-survival effect of Ang2 observed during naturally occurring regression, under circumstances of low or absent VEGF activity, Ang2 had a pro-regression activity. Compared with BSA-treated PMs (Fig. (Fig.55C), Ang2, Flt-Fc treated examples showed distinct morphological changes at 24 h that included dramatic capillary thinning and many ECs with apoptotic morphology (Fig. (Fig.55D).
Discussion
An early model of angiogenesis regulation describing the combined effects of VEGF, Ang1, and Ang2 has proposed that Ang1 is important for blood vessel maturation and stability and acts synergistically with VEGF-A to promote angiogenesis (2). By contrast, it was proposed, based on tissue culture experiments, that Ang2 was an antagonist of Ang1 and could promote vessel destabilization and regression in the absence of VEGF-A (6, 8). The action of Ang2 has become controversial, however, with recent experiments showing that in some circumstances including the corneal pocket assay (10) and in vitro culture (11), this factor can promote angiogenesis.
The current analysis provides support for a model in which the primary function of Ang2 is to destabilize blood vessels and to promote plasticity. In the PM system, where we can examine the effects of a factor on capillaries, Ang2 initially induces a rapid change in endothelial cell shape and increased capillary diameter. This response for Ang2 has been observed in transgenic mice overexpressing Ang1 (24). Ang2 also induces remodeling of the basal lamina as evidenced by the appearance of immunoreactivity for a collagen cryptic epitope (20) throughout the vessels of the PM. Subsequently, endothelial cells extend processes and migrate from existing capillaries. These events closely reproduce the initial steps of the angiogenic process.
It must be appreciated, however, that the PM is normally stimulated by endogenous VEGF. The observation that injected Flt-Fc can induce vessel regression indicates this and shows that VEGF activity is essential for cell survival. This is consistent with the general conclusion that VEGF-A has survival stimulating activity (25–27), that VEGF-A is present in rabbit aqueous (28), and that VEGF-A can suppress normal regression of the PM (29). Thus, the observed activity of injected Ang2 in supporting survival and proliferation of EC and promoting angiogenesis is consistent with existing models (2), as it is likely that endogenous VEGF-A modulates the effects of the angiopoietin.
Of great interest is the observation that the level of EC death increases when Ang2 is injected along with the VEGF inhibitor Flt-Fc. This suggests that when levels of VEGF-A are low, Ang2 has the effect of promoting cell death. This observation is also consistent with existing models suggesting that the presence or absence of VEGF-A is the switch that determines the outcome of Ang2 stimulation. We speculate that when capillary structure is disrupted by Ang2, ECs will be deprived of cell–cell and matrix contacts and, as a consequence, may be more dependent than usual on the survival stimulating activity of VEGF. The molecular basis of this response may be that integrity of VE-cadherin containing adherens junctions is required for signal transduction by VEGF-A (27). Because we observe that Ang2 and endogenous VEGF-A stimulation results in the destabilization of existing VE-cadherin junctions, particularly in cells that have migrated, it may be that these cells are more susceptible to withdrawal of the VEGF-A survival stimulus. This observation also suggests that the most effective way to impose microvascular regression in other settings would be to simultaneously deliver Ang2 while inhibiting VEGF.
Coinjection of the anti-Tie2 antibody and Ang2 into the anterior chamber effectively eliminated the capillary destabilization response. This is an important observation as it confirms the specificity of both the recombinant Ang2 and the anti-Tie2 antibody; it is very unlikely that the Ang2 preparation would contain a cryptic activity that could be neatly inhibited by a complementary cryptic activity in the anti-Tie2 receptor antibody preparation. This observation also confirms that the Ang2 response is mediated by the conventional Tie2 signaling pathway. Interestingly, when injected alone, the antibodies did not have a discernible effect on the PM. This suggests that the constituent cells are not under constant angiopoietin stimulation. In turn, this suggests that Ang2 does not act simply as an antagonist of Ang1, as Ang1 is apparently not present and therefore not regulating PM responses. All these data suggest that Ang2 can induce a highly specific destabilization response in microvascular ECs independent of Ang1.
This study also establishes the PM as a valuable in vivo system for assessing the activities of vascular modulators. With the lack of enveloping tissue, the responses of endothelial cells in the PM can be examined in the kind of detail that previously has been reserved for tissue culture studies. Although requiring technical expertise, the injection and dissection techniques described for the assay can be completed rapidly. The assay can also be combined with vital imaging if the need arises (18), and this extends the utility of the system to examination of very short term responses, for example, in studies of inflammatory modulators. In summary, the PM system represents an important in vivo assay option for vascular modulators whether they are, as in this study, based on gene products, or alternatively, small molecules with empirically determined activities.
Acknowledgments
We thank Tom Sato for reading the manuscript. The Brooks lab is supported by RO1 CA91645 and by a grant from the International Retinal Research Foundation. The Lang lab is supported by National Institutes of Health Grants RO1s EY10559, EY11234, EY12370, and EY14102 and by funds from the Abrahamson Pediatric Eye Institute Endowment at Children's Hospital Medical Center of Cincinnati.
Abbreviations
- VEGF, vascular endothelial cell growth factor
- PM, pupillary membrane
- EC, endothelial cell
Notes
This paper was submitted directly (Track II) to the PNAS office.