INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

ANGIOGENESIS IS A well-recognized process of tissue remodeling that is of vital importance for the execution of developmental programs and physiological adaptations. The meander of newly formed vessels is guided by a complex interplay of angiogenic and angiostatic stimuli (4, 5, 25). Directed cell migration is a key component of the angiogenic process. An array of growth and motility factors induces endothelial cell sprouting and migration. Among those factors, vascular endothelial growth factor (VEGF) gained prominence due to its specificity for the endothelium. VEGF serves as a potent motogen for cultured endothelial cells and displays angiogenic activity in vivo (7, 14, 15). It is secreted by mesenchymal and some other cells in response to stimuli as diverse as hypoxia, various hormones, cytokines, and growth factors (2, 15, 27) and acts on cognate receptors expressed on endothelial cells (19, 21). However, under pathological conditions such as atherosclerosis, heart failure, or some forms of hypertension, among others, the angiogenic response is deficient. Interestingly, these same pathological conditions are also characterized by the diminished production of the endothelium-derived relaxing factor, nitric oxide (NO) (3, 12, 16). We have recently established the role of NO as a requisite component of epithelial cell migration and wound healing (23). In epithelial cells, locomotion stimulated by various agonists (hepatocyte growth factor, epidermal growth factor, fibroblast growth factor 1, or insulin-like growth factor I) was decelerated by inhibitors of NO synthase. Furthermore, in rat microvascular endothelial cells and human umbilical vein endothelial cells (HUVECs), we observed that inhibition of endothelial NO synthase by L-arginine derivatives or by antisense oligonucleotides attenuates endothelin-induced cell migration (22). Collectively, these observations prompted current investigation of a possible link between directed migration stimulated by the endothelium-specific angiogenic factor, VEGF, and NO production in endothelial cells. In two independent experimental systems (transmigration in a Boyden chemotactic apparatus and wound healing), functional NO synthase was obligatory for VEGF-stimulated endothelial cell migration. Analysis of endothelial cell impedance showed that NO mediates spontaneous micromotion (which we refer to as "podokinesis") even in stationary cells. This scalar cell movement is transformed to vectorial locomotion when VEGF guidance is present. The above model emphasizes the need for the dual stimulatory input: 1) NO-induced switch from a stationary to locomoting endothelial cell phenotype, which results in scalar motion or podokinesis, and 2) VEGF-provided guidance cues that impart a vectorial component to the movement of endothelial cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell culture. Microvascular endothelial cells were previously established and characterized by our laboratory; these simian virus 40 (SV40)-immortalized cells established from explant cultures of microdissected rat renal resistance arteries express receptors for acetylated low-density lipoprotein and immunodetectable von Willebrand antigen and are capable of capillary tube formation (31). Cells were grown in gelatin-coated dishes in M199 culture medium (Mediatech, Washington, DC) supplemented with 5% fetal bovine serum (HyClone, Logan, UT), 100 U/ml penicillin, and 100 µg/ml streptomycin (GIBCO BRL, Gaithersburg, MD). HUVECs were obtained from Clonetics and maintained in ECM-2 medium supplemented with 2% fetal calf serum.

Cell migration assay. The migration assay was performed according to previously described technique (24), with minor modifications, in a Boyden chemotaxis apparatus (Neuroprobe, Cabin John, MD). Endothelial cells were lifted with 0.05% trypsin-0.53 mM EDTA (GIBCO BRL) and washed, and 10⁶ cells/ml suspended in 25 µl of M199 with 0.1% bovine serum albumin were added to the lower chamber of the Boyden apparatus. Polycarbonate filters with 8-µm pores were coated with 10 µg/ml ProNectin F (Protein Polymer Technologies, San Diego, CA), washed twice with phosphate-buffered saline, and positioned above the wells of the lower chemotactic chamber, which contained cells. The top half of the chamber was reattached, and the chamber was incubated in an inverted position at 37°C in 95% air-5% CO₂ for 2 h to allow a uniform cell attachment to the filter. In experiments in which L-arginine was added, an L-arginine-free basal medium Eagle (BME; GIBCO) with 0.1% bovine serum albumin was used. The test agents, L-arginine (0.1-1.0 mM), N^G-nitro-L-arginine methyl ester (L-NAME; 400 µM to 2 mM), S-nitroso-N-acetyl-DL-penicillamine (SNAP, 100-500 µM; obtained from Molecular Probes, Eugene, OR), recombinant VEGF165 (1-10 ng/ml, obtained from Collaborative Biomedical Products, Bedford, MA), or a vehicle suspended in 50 µl of M199 or BME with 0.1% bovine serum albumin, were added to upper chambers. The chambers were wrapped with Parafilm and incubated for an additional 5 h in an upright position. After incubation, the filter was removed from the apparatus and cells were fixed with methanol and counterstained. The number of migrated cells on the upper surface of the filter was counted in six randomly chosen fields under ×400 magnification and averaged. All experiments were performed in quadruplicate, and each experiment was repeated at least three times.

Antisense oligodeoxynucleotide treatments. The antisense phosphorothioate derivative of oligodeoxynucleotides (S-ODNs) to the human endothelial constitutive NO synthase (ecNOS) (5'-AGT TGC CCA TGT TAC TGT GCG TCC GTC-3'), nucleotides 56-30 of human ecNOS cDNA (20), as well as the sense 5'-GAC GGA CGC ACA GTA ACA TGG GCA ACT-3' and scrambled 5'-CTG GGA CCT GTT CGT ACA GGT CTC TTC-3' phosphorothioate sequences were synthesized using an automated solid-phase DNA synthesizer (Applied Biosystems, Foster City, CA). These sequences included the 5'-untranslated region of ecNOS cDNA and initiation codon. Thus designed sequences showed no homology with other known mammalian sequences deposited in the GenBank database. All the S-ODNs were purified with oligonucleotide purification cartridges, dried down, resuspended in tris(hydroxymethyl)aminomethane (Tris)-EDTA [10 mM Tris (pH 7.4)-1 mM EDTA (pH 8.0)], and quantified spectrophotometrically.

Monitoring NO release. Cells grown in gelatin-coated 35-mm dishes were preincubated in Krebs-Ringer buffer of the following composition (in mM): 136 NaCl, 5.4 KCl, 1.8 CaCl₂, 0.8 MgSO₄, 1.0 NaH₂PO₄, and 10 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid and 1 g/l glucose, adjusted to pH 7.4, which was supplemented with 7.5 U/ml superoxide dismutase (Sigma, St. Louis, MO). NO release was monitored with an NO-selective microprobe (Inter Medical, Nagoya, Japan) as previously detailed (22, 23). The working electrode made of Pt/Ir alloy was coated with a film containing KCl, NO-selective nitrocellulose resin (pyroxyline lacquer), and a gas-permeable silicon membrane (8). The counter electrode was made of carbon fiber. The redox current was detected by a current-voltage converter circuit and continuously recorded. Tip diameter of the probe (25 µm) permitted the use of a micromanipulator (Zeiss-Eppendorff) attached to the stage of an inverted microscope (Nikon Diaphot) and enclosed in a Faraday's chamber to position the sensor 3-5 µm above the cell surface. Calibration of the electrochemical sensor was performed using different concentrations of a nitrosothiol donor (SNAP) as previously detailed (8).

Cell adhesion, wound healing, and micromotion assays. Electrode fabrication and the design of electric cell-substrate impedance sensor (ECIS) have been reported previously (6, 17). Electrodes were precoated with 10 µg/ml fibronectin. Endothelial cells at a density 2 × 10⁵ were seeded on electrodes in the presence of different agents, as detailed in RESULTS, and cell adhesion was recorded as an increase in the electrical resistance. To study cell detachment from the electrodes, confluent endothelial monolayers grown on electrodes were exposed to different agonists. To generate "wounds" on the surface of microelectrodes, confluent endothelial monolayers were electropermeabilized with direct current from a 6-V battery generating a local electrical current of ~0.12 mA. This procedure resulted in a loss of cells from the surface of the microelectrode, while all surrounding cells remained intact. The rate of repopulation of the microelectrode with migrating cells was monitored for 24 h as changes in resistance and capacitance.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of VEGF on NO production and cell migration. Application of VEGF caused a dose-dependent increase in NO release from SV40-immortalized renal microvascular endothelial cells (Fig. 1A). The 50% effective concentration (EC₅₀) was ~8 ng/ml. VEGF exhibited a similar potency in accelerating endothelial cell migration, with an EC₅₀ of ~10 ng/ml (Fig. 1B). In transwell migration experiments, the above effect of VEGF was blunted in the presence of the NO synthase inhibitor, L-NAME (Fig. 1C).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Effects of vascular endothelial growth factor (VEGF) on nitric oxide (NO) production and endothelial cell migration. A: representative recordings (n = 4) of VEGF-induced release of NO from cultured endothelial cells. Experiments were performed with an NO-selective microelectrode, as detailed in MATERIALS AND METHODS. Application of VEGF resulted in a transient release of NO. The 50% effective concentration for VEGF-induced NO production was 8 ng/ml. NO release could be blocked by pretreatment with 2 mM NG-nitro-L-arginine methyl ester (L-NAME) (not shown). Inset: calibrated NO concentration and time scale. B: dose-response curve of VEGF-induced endothelial cell migration in a Boyden apparatus. * P < 0.05 vs. control (n = 4). C: VEGF-induced endothelial cell migration requires functional NO synthase. VEGF was used at concentration of 10 ng/ml, resulting in a more than 2-fold increase in the number of migrated endothelial cells; L-NAME (2 mM) abrogated the effect of VEGF. These experiments were performed in medium depleted of growth factors but containing L-arginine. C, control.

Effect of antisense oligonucleotides on VEGF-guided endothelial cell migration. To secure the highest possible selectivity of eNOS inhibition, presently not achieved with the available L-arginine derivatives, we next utilized the antisense strategy to selectively knock down the enzyme. Because the antisense construct was designed to target the initiation codon of the human eNOS cDNA, HUVECs were utilized in these experiments. After a 12-h incubation with 10 µM antisense, sense, or scrambled S-ODNs in serum-free medium, cells were lifted and seeded into wells of a Boyden chamber in the continuous presence of S-ODNs, as detailed in MATERIALS AND METHODS. Two hours later, when cells effectively adhered and spread, culture medium containing 10 ng/ml VEGF was added to the opposite side of the membrane. Transmigration assessed 7 h later showed that VEGF-induced stimulation of HUVEC migration was virtually abrogated in cells pretreated with the antisense construct, whereas effects of VEGF in the presence of sense and scrambled S-ODNs were not significantly different from nonpretreated control cells (Fig. 2). The efficacy of antisense treatment has been validated previously in the same experimental setting (22). Furthermore, the toxicity of S-ODNs at the concentrations used was undetectable, based on experiments with ethidium homodimer staining of nuclei of damaged cells (not shown). These findings, therefore, reconfirm the above data, which demonstrate the requisite role of functional eNOS for VEGF-guided endothelial cell migration.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Effect of antisense oligonucleotides to endothelial NO synthase (eNOS) on transmigration of human umbilical vein endothelial cells. Experiments were performed in a Boyden apparatus, as detailed in MATERIALS AND METHODS. Endothelial cells were pretreated with antisense (AS), sense (S), or scrambled (SCR) phosphorothioated oligonucleotides [all at 10 µg/ml, previously shown to be nontoxic and sufficient for eNOS knockdown (22)].

Wound healing experiments. Endothelial cells were damaged by the application of a direct current in the ECIS chamber. This resulted in a localized injury to the endothelial cells occupying the electrode surface of 50,000 µm² and denudation of the electrode, as shown in Fig. 3A. Consequently, the resistance of cell monolayers precipitously dropped to 24 ± 3% of the basal level detected at confluence. The resistance of the monolayers increased to 57 ± 4% of basal level by 24 h after wounding, due to the migration of cells from wound edges (confirmed microscopically at the end of experiments, as exemplified in Fig. 3A). The addition of VEGF accelerated the process of wound healing (Fig. 3B), resulting in 69 ± 3% (P < 0.05, n = 5) restoration of the resistance (resistance at confluence was taken as 100%). This process was dependent on the functional NO synthase because pretreatment of endothelial cells with L-NAME virtually abrogated wound healing [29 ± 5% of basal level (Fig. 3B), whereas D-NAME was ineffective (not shown)].

View larger version (44K): [in this window] [in a new window]   Fig. 3.   Endothelial wound healing is enhanced by VEGF in NO-dependent fashion. A: micrographs of the fibronectin-coated platinum electrode surface immediately after an intact confluent cell monolayer was destroyed on the electrode surface by a direct current pulse (note that cells are absent on the electrode, whereas they are preserved around it) and 24 h later, when the electrode surface has been repopulated by migrated endothelial cells. Diameter of an electrode = 250 µm. B: tracings of the electrical impedance of endothelial cells. Initially, all monolayers had resistances of >9,000 /cm2 (normalized resistance is shown) that fell to ~20-30% of initial resistance after the application of direct current pulse. In the absence of growth factors, endothelial cells slowly repopulated the electrode surface, as judged by the increasing resistance. VEGF accelerated the process of endothelial wound healing, whereas addition of L-NAME, alone or in combination with VEGF, abrogated endothelial cell migration. These data are representative of 7 separate experiments performed with different batches of endothelial cells.

Role of NO in cell matrix adhesion. To elucidate the cellular basis for the observed permissive role of NO in VEGF-guided migration, endothelial cell monolayers were subjected to electrical impedance analysis. Resting cells displayed spontaneous fluctuations in electrical impedance that promptly disappeared after addition of a fixative (Fig. 4A). These data are consistent with the previously described micromotion in ostensibly stationary cells (6). Furthermore, confluent endothelial cell monolayers subjected to the NO donor, SNAP, displayed spontaneous higher-amplitude fluctuations of impedance (at 100 µM) and a decrease in the resistance of monolayers (at 500 µM), as shown in Fig. 4, B and C, respectively. To test the possibility that NO interferes with cell matrix adhesion, the attachment of endothelial cells to fibronectin-coated microelectrodes was examined in the presence and absence of SNAP. When endothelial cells were seeded on the surface of fibronectin-coated electrodes, cell attachment and spreading (confirmed microscopically at 4 h) resulted in a progressive increase of the resistance (Fig. 4D). The addition of SNAP resulted in a dose-dependent inhibition of cell attachment [100 µM SNAP resulted in 92.7 ± 0.87% of control (mean ± SE), n = 4, P > 0.05; 500 µM SNAP decreased adhesion to 65.5 ± 3.57% of control, n = 4, P < 0.005 vs. both control and 100 µM SNAP], thus supporting the notion that NO affects endothelial cell matrix adhesion and implying observed spontaneous fluctuations in cell resistance could be attributed to changes in adhesive interactions.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Real-time recordings of the electrical impedance of endothelial cells. A: spontaneous fluctuations in the resistance of confluent endothelial cell monolayers disappear after fixation with 4% paraformaldehyde (PFA). B and C: NO donor S-nitroso-N-acetyl-DL-penicillamine (SNAP; arrow) increases the amplitude of spontaneous fluctuations in the electrical resistance of confluent endothelial cells (at 100 µM in B) and, at a higher concentration (500 µM in C), decreases the resistance. Note that the ordinate scale in B and C is different (to demonstrate absolute and relative value of these miniature spontaneous fluctuations detectable with high-resolution amplification). D: real-time recordings of the electrical impedance of freshly seeded endothelial cells, as an indicator of their adhesion and spreading. Rate of endothelial cell adhesion and spreading on fibronectin-coated microelectrode surface is decreased in a dose-dependent fashion by the addition of the NO donor SNAP. All tracings are representative of at least 4-5 experiments performed on separate occasions using different passages of endothelial cells.

View larger version (27K): [in this window] [in a new window]   View larger version (27K): [in this window] [in a new window]   Fig. 5.   Effect of modulating NO availability on endothelial cell matrix adhesion. Endothelial cells pretreated with 200 µM sodium nitroprusside (SNP) or 2 mM L-NAME were plated into wells coated with the indicated matrix proteins. After 60-min adhesion period, reaction was stopped, cells were stained, and cell density was determined by measuring the absorbance (optical density units). LN, laminin; Col I, collagen I; Col IV, collagen IV; FN, fibronectin.

View larger version (24K): [in this window] [in a new window]   Fig. 6.   Analysis of changes in resistivity of endothelial cell monolayers after application of 500 µM SNAP (A) or pretreatment with 1 mM L-NAME (B). Top: scheme of endothelial cell monolayer considered as a series of resistors, with the electrical current passing either paracellularly or subcellularly underneath the ventral cell surface (h space). Parameter  equals the product of cell radius (r) and square root of a ratio between a constant  (resistance of the culture medium) and subendothelial space (h). A and B: changes in the  before and after specified treatments, based on results of individual experiments. Calculations were performed using an assumption that cell radius did not change after the treatments. * Significant difference (P < 0.05) between pre- and posttreatment data.

View larger version (56K): [in this window] [in a new window]   Fig. 7.   Serial images of focal adhesions in endothelial cells (reflective confocal microscopy). Endothelial cells were seeded on fibronectin-coated glass coverslips and allowed to attach for 30 min. Unattached cells were removed by changing the medium, and images were recorded at 5-min interval. During the recorded 30-min period of cell attachment (images obtained at 0-30 min), focal contacts, visualized as dark streaks using this technique, gained in prominence and eventually stabilized (images obtained at 10-30 min). Addition of 100 µM SNP, a short-acting NO donor, resulted in the transient disappearance of some focal contacts (images obtained at 35-45 min). Quantitative analysis of these changes was performed using Universal Imaging software program. Histograms of light intensity (right) correspond to the flanking images and graphically illustrate the shift in intensity occurring after the addition of SNP (detected as the right shift toward brighter pixels in images obtained at 35 and 40 min).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In the present study, the absolute requirement for functional NO synthase in executing the VEGF-guided program of endothelial cell migration and wound healing was demonstrated in two independent experimental settings. First, VEGF, shown to stimulate NO release from endothelial cells, induced endothelial cell migration (75% stimulation), whereas L-NAME or antisense oligonucleotides targeting eNOS completely suppressed VEGF-stimulated migration in a Boyden apparatus. Second, in a series of experiments on endothelial cell wound healing, the rate of VEGF-stimulated cell migration was significantly blunted by the inhibition of NO synthesis. The above findings led us to conclude that VEGF-induced endothelial cell locomotion is mediated in part via NO production.

To gain insight into the possible mode of NO action, impedance analysis of endothelial cell monolayers subjected to NO was performed. Using this impedance bioprobe, we have previously demonstrated micromotion, i.e., spontaneous fluctuations in the resistance, in quiescent cells (6). In the present study, we show that the micromotion in ostensibly stationary endothelial cells is elicited in part by NO. We proposed to term this process podokinesis to emphasize a scalar form of cell locomotion occurring in nonmigrating endothelial cells, which, in the presence of guidance cues, e.g., VEGF, may be transformed to a vectorial movement. Thus the execution of the program for a directional endothelial cell migration requires two simultaneous messages: scalar NO-induced podokinesis and guidance cues provided by motogens like VEGF (the present study), epidermal growth factor (23), or endothelin-1 (22).

The above data on the role of NO in mediating motogenic effects of VEGF are in concert with the findings of substance P-stimulated cell migration. Recently, it has been shown that the angiogenic effect of substance P is mediated by NO (33). NO synthase inhibitors abolished substance P-induced migration of microvascular endothelial cells, whereas SNP potentiated cell migration. Leibovich et al. (13) have observed angiogenic activity produced by stimulated human monocytes in the presence of L-arginine, but not in its absence, and suppressed angiogenesis in the presence of inhibitors of NO synthase. Because these manipulations of the NO pathway did not affect the release of several cytokines (tumor necrosis factor- and interleukin-8), it is conceivable that the angiogenic activity in question was NO per se. Similarly, we have recently demonstrated that epithelial cell migration and wound healing are NO dependent: NO serves as a switch from a stationary to locomoting phenotype in BS-C-1 cells (23). Furthermore, exogenous NO has been shown to elicit chemotaxis of neutrophils in vitro (1). Most recently, we have presented experimental evidence that NO plays a permissive role in endothelin-1-elicited locomotion of rat microvascular and of HUVECs (22). Inhibition of endothelial NO synthase with either L-NAME or antisense S-ODNs prevented endothelin-induced locomotion. A similar effect was observed in Chinese hamster ovary cells expressing endothelin B receptor and NO synthase (22). When the present study was in preparation, Morbidelli et al. (20) reported that VEGF-induced proliferation of coronary venular endothelial cells is mediated by NO and NO synthase inhibition attenuates this effect. Indeed, previous studies in canine coronary arteries have strongly suggested that VEGF-mediated vasorelaxation is dependent on the production of endothelium-derived relaxing factor, thus suggesting that this growth factor stimulates generation of NO (10). In the present study, through the use of an NO-selective microelectrode, the VEGF-stimulated release of NO was documented directly (Fig. 1A). Furthermore, the role of cell migration, with lesser impact of cell proliferation, is emphasized in experiments presented herein, since the time course of experiments and incubation of cells in low-serum medium were designed so as to minimize the mitogenic effects of VEGF.

It should be realized that NO effects on locomotion may be cell specific and dose dependent. For instance, after balloon injury of rat carotid artery endothelium, in vivo gene transfection of the endothelial NO synthase into vascular smooth muscle cells abolished neointimal formation (32). Similar results have been reported by Marks et al. (18), demonstrating that a protein adduct of NO prevented neointimal formation. Although the observed prevention by NO of neointimal formation could be secondary to the inhibition of smooth muscle cell proliferation, there was a biphasic effect of protein adduct of NO on smooth muscle cell migration in a Boyden apparatus: low concentrations of the protein enhanced it, whereas high concentrations suppressed it (18).

Sporadic fluctuations in endothelial resistance and their enhancement by NO, as resolved due to unique amplification of miniature changes in impedance of cells grown on the surface of gold microelectrodes in ECIS (Fig. 5), are due at least in part to changes in cell matrix adhesion. Teleologically, they may represent the state of cellular "vigilance" (preparedness to migrate when the integrity of endothelial layer becomes compromised or motogenic guidance cues are applied). The observed spontaneous changes in cell matrix adhesion, referred to as podokinesis, may represent an essential step in transforming the scalar motion of nonmigrating endothelial cells to the vectorial movement on application of VEGF or endothelin. Experiments with SNAP-induced stimulation of podokinesis implicate NO in its generation. Furthermore, NO attenuated endothelial cell adhesion to several matrix proteins, including fibronectin, and, at a higher concentration, mediated cell detachment, indicating that NO affects cell matrix adhesion. In this context, observations by Oliver et al. (Ref. 24 and reviewed in Ref. 28) that the rate of locomotion has a biphasic dependence on the tightness of cell-substratum adhesion may explain the general phenomenon of NO-mediated motility. Somewhat similar antiadhesive effects of NO have been described in leukocytes. Adhesion of leukocytes to the venular endothelium in a superfused cat mesenteric preparation is inhibited by L-arginine and enhanced by inhibitors of NO synthase (11). Analogous inhibition of leukocyte adhesion to type I collagen was observed with SNAP or 8-bromoguanosine 3',5'-cyclic monophosphate (29). Thus it appears that NO affects cell adhesion and migration in a broad range of cells. Experiments conducted in collaboration with Zachary and Abedi in HUVECs further confirm the effect of NO on formation and stability of focal adhesions with reference to different matrix proteins (M. S. Goligorsky, H. Abedi, and I. Zachary, unpublished observations). The proposed hypothesis on NO-driven changes in cell matrix adhesion, resulting in podokinesis, may provide the explanation for these diverse observations. Our model predicts the requirement for two coexisting inputs: first, NO-dependent podokinetic scalar movements and, second, presentation of guidance cues, e.g., VEGF or endothelin, to execute the transition from a stationary phenotype to vectorial locomotion resulting in angiogenesis (as schematically depicted in Fig. 8).

View larger version (28K): [in this window] [in a new window]   Fig. 8.   Hypothetical scheme of the dual-signal requirement (NO-driven podokinesis and guidance cues) for directed migration of endothelial cells in vitro and angiogenesis in vivo with the emphasis on the role of functional NO synthase in these processes.

The implications of the above findings are twofold. First, in situations in which angiogenesis is undesirable, i.e., tumor angiogenesis, a targeted inhibition of NO production by endothelial cells may provide a tool to halt neovascularization. In this vein, some tumors express enhanced activity of NO synthases (9), in addition to the production of various stimulators of angiogenesis. This may explain in part the enhanced angiogenic meander toward such foci (5). Second, the above findings may shed light on the reasons for inadequate angiogenesis, despite the presence of appropriate gradients of angiogenic stimuli, in the states of impaired endothelial NO production, i.e., atherosclerosis (3, 12, 16). Under these circumstances, supplying NO to the endothelium at ischemic sites may improve angiogenesis. Application of the above model predicting the requirement for two messengers, NO and VEGF, for execution of a program for the directional locomotion should provide a theoretical basis for the design of future angiogenic and angiostatic therapies.