Role for Extracellular Signal-Responsive Kinase-1 and -2 in Retinal Angiogenesis

Abstract

purpose. Vascular endothelial growth factor (VEGF) is a potent mitogen for micro- and macrovascular endothelial cells (ECs). Evidence points to a possible role for two mitogen-activated protein (MAP) kinases, the extracellular-signal responsive kinases (ERK)-1 and -2, in VEGF signaling in ECs. This study was undertaken to begin to define the precise role of MAP kinases in VEGF signal transduction related to angiogenesis.

methods. Bovine retinal microvascular endothelial cells (BRMECs) and a well-established rat model of retinopathy of prematurity (ROP) were used to investigate the role of ERK-1/2 in EC proliferation and tube formation and in retinal angiogenesis in vivo.

results. Administration of VEGF to BRMEC cultures increased ERK-1/2 phosphorylation, cell proliferation, and tube formation in a dose-dependent manner. Phosphorylation of retinal ERK-1/2 also was increased in the ROP model. An inhibitor of ERK, AG126, and an inhibitor of ERK kinase (MEK), PD98059, exhibited a dose-dependent reduction of ERK phosphorylation and EC proliferation, but not tube formation, in VEGF-stimulated BRMECs. In the ROP model, intravitreous injection of 10 μM AG126 or PD98059 reduced the retinal neovascular area by 71% and 48%, respectfully. No effect was seen on intraretinal blood vessel growth.

conclusions. These experiments point to a critical role for ERK and MEK in proliferation of ECs, but not in tube formation. Furthermore, inhibition of either of these two signal intermediates can significantly retard retinal neovascularization. This suggests that the MAPK pathway may provide rational targets for therapeutic intervention in ocular and other diseases with an angiogenic component.

Vascular endothelial growth factor (VEGF) is a potent mitogen for micro- and macrovascular endothelial cells (ECs), with an ED₅₀ of 2 to 10 pM. ¹ VEGF stimulates proliferation of ECs in a variety of in vitro models, ^{2 3 4} and it induces ECs to undergo differentiation and remodeling into capillary-like structures in a three-dimensional gel matrix. ^{5 6} In vivo, VEGF displays a proangiogenic influence in the chick chorioallantoic membrane assay, ^{7 8} in the corneal neovascularization model, ^{9 10 11} and in the subcutaneous implant model. ¹² Its proangiogenic influence also has been linked to several conditions of clinical importance, including rheumatoid arthritis ^{13 14 15} and inflammatory skin diseases, such as psoriasis, erythema, and dermatitis. ^{16 17} Increased VEGF levels have been found in tumors of the breast, ^{18 19} brain and meninges, ^{20 21} lung, ^{22 23} gastrointestinal tract, ^{24 25 26} and bladder and kidney. ^{27 28 29 30} Relevant to this work, elevated VEGF has been measured in vitreous and aqueous humor samples of patients with a variety of potentially blinding neovascular ocular conditions. ^{31 32 33}  

VEGF exerts its effects by binding with high affinity to its two known EC-specific receptors, flt-1 and KDR, the latter homologous to murine fetal liver kinase (flk-1). ^{2 34 35 36} Flt-1 ³⁷ and KDR ³⁸ each demonstrate VEGF-induced dimerization and tyrosine autophosphorylation. ³⁹ Phosphorylated VEGF receptor, in turn, promotes tyrosine phosphorylation of several known signal-transduction proteins, including phospholipase C-(PLC)-γ, phosphatidylinositol 3-(PI 3) kinase, Ras guanosine triphosphatase (GTPase) activating protein (GAP), and the oncogenic adapter protein NcK in bovine aortic ECs. ³⁹ Increased phosphorylation of PLC-γ and the GAP complex, as well as increased p42/p44 mitogen-activated protein (MAP) kinase activity, have been found in sinusoidal ECs after VEGF stimulation. ⁴⁰ Increased phosphorylation of MAP kinase has also been reported in vivo in retinal ischemia–reperfusion injury models. ^{41 42} MAP and extracellular-signal responsive kinase (ERK) kinases (MEKs) are proteins in the Ras-Raf-MAP kinase cascade. This is one of the first protein kinase networks to be fully elucidated and is the central effector of cellular differentiation in development. Moreover, its inappropriate and continuous activation provides a potent promitogenic force and is a very common occurrence in human cancers. The cascade is propagated by phosphorylation of serine/threonine residues on substrates with intrinsic kinase activity. ⁴³ The focus of the present work is on ERK-1 (p44^MAPK) and ERK-2 (p42^MAPK), and the principle kinases responsible for activating ERKs: MEK-1 and -2. ⁴⁴  

This study was designed to determine the importance of MAP kinases in VEGF signal transduction in vitro and in vivo. VEGF-induced MEK and ERK activities were targeted in proliferation and differentiation assays using cultures of bovine retinal microvascular endothelial cells (BRMECs). These kinases were also targeted in a well-established rat model of retinopathy of prematurity (ROP). 

Methods

In Vitro Experiments

Cell Culture.

Primary cultures of BRMECs (VEC Technologies, Rensselaer, NY) from the 6th to the 10th passages were used in all in vitro experiments. Initially, cells were grown in tissue culture flasks coated with a 50:50 solution of 100 μg/mL fibronectin-100 μg/mL hyaluronic acid (HA) in calcium-/magnesium-free (cmf)-PBS (GibcoBRL/Life Technologies, Inc., Rockville, MD). Cells were incubated at 37°C with 5% CO₂ in MCDB-131 complete medium (MCDB-131, 10% fetal bovine serum [FBS], 10 ng/mL epidermal growth factor [EGF], 1 μg/mL hydrocortisone, 100 mg/500 mL endothelial cell medium additive [Endo Gro; VEC Technologies], 45 mg/500 mL heparin, and 5 mL/500 mL 100× antibiotic-antimycotic solution, hereafter referred to as growth medium). When experimental conditions required a serum and growth factor–free medium, the antibiotic–antimycotic solution was added at the above concentration to the MCDB-131 medium (hereafter referred to as serum-free [SF]medium). 

BRMECs were plated in fibronectin-HA–coated wells of six-well culture plates (Corning, Acton, MA). The cells were incubated in growth medium overnight, then in SF medium for 24 hours, and finally, in experimental medium. A VEGF dose–response effect was defined by using SF medium plus 0, 1, 5, 10, 25, or 50 ng/mL recombinant human (rh)VEGF₁₆₅ (R&D Systems, Minneapolis, MN). Optimal VEGF exposure duration was determined by exposing the cells to VEGF for 0, 5, 10, 30, 60, or 120 minutes at optimum concentration. 

Immunoblot Analysis.

After removing medium and washing in PBS, samples were collected in 300 μL of cold lysis buffer (150 mM NaCl, 1.0% Triton X-100, 0.1% SDS, 50 mM Tris-HCl, 100 μg/mL phenylmethylsulfonyl fluoride [PMSF], 1 mM orthovanadate, 0.3 μg/mL EDTA, 0.5% deoxycholate acid, 50 μM NaF, 0.5 μg/mL leupeptin, 0.7 μg/mL pepstatin A, and 100 μL of 1.5 mg/mL aprotinin), lysed for 20 minutes and centrifuged at 5000 rpm for 15 minutes at 4°C. After protein concentrations were determined with the bicinchoninic acid protein assay (BCA) kit (Pierce, Rockford, IL), the volumes of these samples were adjusted to obtain equal protein concentrations. Samples were heated to 95°C for 5 minutes and resolved by SDS-polyacrylamide gel electrophoresis (Bio-Rad, Hercules, CA). Proteins were then transferred to 0.2-μm nitrocellulose membranes (Bio-Rad) which were probed according to the antibody manufacturer’s instructions and visualized with enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech, Piscataway, NJ). Antibodies against phosphorylated and nonphosphorylated ERK-1/2 and MEK were purchased from New England Biolabs (Beverly, MA). 

In addition to analyzing ERK and MEK phosphorylation in the VEGF concentration- and time-dependent assays, Western blot analyses were also conducted on BRMECs preincubated with SF medium containing one of two specific signal-transduction inhibitors for 2 hours, followed by the addition of VEGF. The two compounds used for these experiments target two distinct intermediates in the same MAP kinase signal-transduction pathway. AG126 (Calbiochem, La Jolla, CA) is a synthetic tyrphostin that blocks tyrosine phosphorylation of p42^MAPK. ⁴⁵ PD98059 (New England Biolabs) binds to the inactive form of MEK-1 and prevents its activation by Raf. ⁴⁶ The vehicle for both AG126 and PD98059 was 0.5% dimethyl sulfoxide (DMSO). 

MAP Kinase Activity Assay.

The effects of VEGF and transduction inhibitors on MAP kinase activity were studied in BRMECs by a nonradioactive method (New England Biolabs). This method includes immunoprecipitation of sample protein with anti-phospho-ERK, followed by incubation with Elk-1 fusion protein. Phosphorylation of Elk-1 was then analyzed by Western immunoblot with anti-phospho-Elk-1. The manufacturer’s kit provided all necessary reagents. 

Proliferation Assay.

VEGF-induced proliferation of ECs was measured with a modified MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay. ⁴⁷ BRMECs were plated at 3 × 10³ onto a fibronectin-HA matrix in 96-well plates (Corning). Growth medium was added for 2 days, followed by SF overnight, and then test medium containing 0, 5, 10, 25, or 50 ng/mL VEGF in 100 μL of SF medium. After incubation for 24 hours at 37°C in 5% CO₂, 25 μL MTT was added to each well and incubated for 4 hours. One hundred microliters of lysis buffer (20% SDS in 50:50 n,n-dimethylformamide (DMF) and H₂O with 2.0% acetic acid and 0.05% HCl) was then added to each well, and the plates were incubated overnight at 37°C and read (Spectramax 190; Molecular Devices, Sunnyvale, CA) at 570 nm. Absorbance values were translated to cell number using standard curves consisting of six cell densities assayed in quadruplicate. For experiments with inhibitors of signal transduction proteins, 25 ng/mL VEGF was combined with the inhibitors at various concentrations. All treatments were performed in quadruplicate in each of three separate experiments. The percentage of dead cells was determined in concentrations of AG126 and PD98059 ranging from 1 to 100 μM, by using a live–dead assay kit (Molecular Probes, Eugene, OR). Under culture conditions identical with those used in the proliferation studies, concentrations of 10 μM and lower resulted in percentages of dead cells that were not significantly different from the percentage in untreated cultures. Both agents at concentrations ranging from 20 to 100 μM resulted in percentages of dead cells that were significantly higher than those in untreated cultures (P < 0.05). 

Gel Tube-Formation Assay.

VEGF-induced BRMEC differentiation was assessed with a modified tube-formation assay. ⁴⁸ A mixture of 8 vol of collagen matrix (Vitrogen 100; Cohesion, Palo Alto, CA), 1 vol of 0.2 N NaOH, and 1 vol of 10× RPMI-1640 medium containing 5 μg/mL fibronectin and 5 μg/mL laminin was prepared, and 400 μL was added to each well of a 24-well plate. After the plates were incubated for 3 hours at 37°C to solidify the gel, 1 × 10⁴ BRMECs were added to each well and incubated in growth medium for 3 days. The medium was carefully aspirated, and 200 μL of the gel solution was layered on top of the cells and incubated at 37°C for 1 hour. After addition of growth medium for 24 hours, 2 mL of test medium containing SF medium plus VEGF or SF medium plus VEGF and inhibitors was added to each well. The gels were observed after 24 hours in the experimental medium. After a brief dose–response experiment, the broad-spectrum tyrphostin herbimycin A was used at 0.3 μM as a positive control for the two kinase-specific inhibitors. 

The plates containing BRMEC tubes were placed on an inverted microscope (IMT-2; Olympus, Melville, NY) with a digitizing camera (DMC; Polaroid, Cambridge, MA) coupled to a computer (G4 Macintosh; Apple Computer, Inc., Cupertino, CA). The wells were visualized with the 10× objective, and the image was digitized. Eight fields per treatment group were randomly chosen for quantitative analysis. The digitized images were imported into NIH Image (available by ftp from zippy.nimh.nih.gov/or from http://rsb.info.nih.gov/nih-image; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD). Capillary-like structures of more than two cell lengths were assessed. The measurement function was used to determine the lengths of the tubes, and the data were expressed in micrometers as the total length per field of view. 

In Vivo Experiments

Rat Model of Retinopathy of Prematurity.

The rat model used for in vivo experiments is described in detail elsewhere. ⁴⁹ Briefly, within 4 hours after birth, complete litters of Sprague-Dawley rats were placed with their mothers into infant incubators (Isolette, Hill-Rom Services, Inc., Batesville, IN) under controlled oxygen environments. The animals were exposed to alternating concentrations of oxygen (50% and 10%) every 24 hours for 14 days. On day 14, the litters were removed to room air and treated as described in the following section. After 6 days in room air (postnatal day 20), the animals were killed for assessment of normal and abnormal vascular development. 

Intravitreous Injections.

In vivo experiments were conducted to test the effect of the two inhibitors using a well-established intravitreous injection procedure. ⁵⁰ Immediately after rats were removed to room air, eyes were assigned to one of three treatment groups: noninjected, vehicle (0.5% DMSO)-injected, or inhibitor (AG126 or PD98059 in 0.5% DMSO)-injected. Animals were anesthetized by methoxyflurane (Pitman-Moore, Mundelein, IL) inhalation. A single drop of 0.5% proparacaine hydrochloride (Allergan, Hormigueros, PR) was topically applied to the cornea, and 5-μL intravitreous injections were administered with a 30-gauge syringe (Hamilton, Reno, NV). After the injection, topical antibiotics were applied to the eye. Dose–response experiments were conducted to determine the best concentrations of both inhibitors for use in subsequent in vivo experiments. Appropriate control experiments were conducted to examine the complication of bilateral or systemic effects. Sample sizes for each treatment can be found in Figures 8 and 9 . All experiments conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

VEGF ELISA.

Intravitreous injections of vehicle or inhibitors were made immediately after rats were removed from the oxygen-exposure chamber. Rats received injections of 0.5% DMSO in the right eye and AG126 or PD98059 in 0.5% DMSO in the left eye. Rats were killed 24 hours later, and the retinas were dissected and frozen at −80°C until assayed. The samples were sonicated in Tris-buffered saline (TBS) with protease inhibitors (100 μg/mL PMSF, 0.3 μg/mL EDTA, 0.5 μg/mL leupeptin, 0.7 μg/mL pepstatin A, and 100 μL of 1.5 mg/mL aprotinin), and centrifuged at 5000 rpm for 15 minutes at 4°C. The supernatants were collected and subjected to colorimetric sandwich VEGF ELISA assay (R&D Systems), according to the manufacturer’s instructions. The sensitivity of the assay was 3 pg/mL. Total protein concentration in each sample was measured by BCA protein assay (Pierce). The final mass of retinal VEGF was standardized to total retinal protein. Sample sizes for each treatment can be found in Figure 7 . 

Tissue Histochemistry.

Animals were killed by decapitation 6 days after removal to room air (day 20). The eyes were enucleated, and the neural retinas were dissected and stained by using a histochemical method for detecting adenosine diphosphatase (ADPase) activity, according to a previously described technique ⁵¹ adapted for use in the current study. ⁵² The stained retinas were then flatmounted onto slides. 

Vascular Assessment.

All vascular assessment was conducted on day 6 after oxygen exposure. After histochemical staining, images of the retinas were captured using the digitizing camera (DMC; Polaroid) fixed to a dissecting microscope (SZ40; Olympus). For manipulation and quantitative analysis, the digitized images were imported into an image-analysis program (Photoshop; Adobe Systems, San Jose, CA). With this software, normal intraretinal vascular growth was determined by tracing the vascularized area of a retina and converting the units from pixels to square millimeters. The extent of pathologic neovascular development was semiquantitatively determined by the clock-hour method, as previously described. ⁴⁹  

Protein Analysis of Retinal Tissue.

Either immediately after removal from the exposure chamber or 6 days later, rats were killed, and their retinas were dissected and prepared for analysis of ERK and MEK phosphorylation. In addition, rats that had received PD98059 injections were killed 10 minutes, 30 minutes, 4 hours, or 24 hours later, and retinal tissue was harvested and processed for analysis of phosphorylation of ERK. In all cases, two retinas treated similarly were pooled in 100 μL of lysis buffer and homogenized, and the samples were centrifuged at 5000 rpm for 15 minutes at 4°C. Protein concentrations were normalized by adjusting volume, and samples were then analyzed by SDS-PAGE and immunoblot, as described for BRMECs. 

Statistical Analysis.

Data were analyzed on computer (StatView software; Abacus Concepts, Berkeley, CA). Parametric statistics were used to analyze proliferation and tube-formation assay results. The data were considered to be statistically significant if P < 0.05, by analysis of variance. In vitro data are expressed as the mean result ± SD. Western blot analysis were semiquantitatively analyzed by using raw optical density measurements. Data are expressed as the mean ± SD of results derived from triplicate experiments. Data from the ROP model using the ordinal clock-hour measurement were analyzed with the nonparametric Mann-Whitney test and are expressed as the mean ± SD. Again, significance was determined by P < 0.05. 

Results

Effect of RhVEGF₁₆₅ on Phosphorylation of ERK-1/2 and Kinase Activity of ERK-1/2

The phosphorylation levels of p42 and p44 MAP kinases were semiquantified with a mouse monoclonal antibody that recognizes only ERK-1/2 phosphorylated on both threonine and tyrosine amino acids in the -T-E-Y- motif. The control for these blots was an anti-ERK-1/2 antibody that does not discriminate between phosphorylated and nonphosphorylated states. Peak phosphorylation of ERK-1/2 was observed after a 10-minute incubation in 25 ng/mL VEGF (Figs. 1A 1B) . Highest MAP kinase activity, as measured by increased phosphorylation of the exogenous substrate Elk-1 fusion protein, was associated with MAP kinase immunoisolates of cells exposed to 25 ng/mL VEGF for 10 minutes (Fig. 1C) . 

Effect of PD98059 and AG126 on RhVEGF₁₆₅-Stimulated ERK Phosphorylation and Elk Phosphorylation

In BRMECs incubated with 25 ng/mL VEGF for 10 minutes, PD98059 caused a dose-dependent reduction in ERK phosphorylation, with a nearly complete elimination of ERK-1 phosphorylation evident at 10 μM (Fig. 2A) . Cells that were preincubated in medium containing 10 μM AG126 demonstrated reduced MAP kinase activity, as measured by a 69% decrease in Elk-1 phosphorylation (Fig. 2B) . 

Effect of AG126 and PD98059 on RhVEGF₁₆₅-Induced BRMEC Proliferation

VEGF-induced proliferation of BRMECs was studied to provide a baseline for inhibitor studies. Under the conditions used, maximum proliferation was stimulated by 25 ng/mL VEGF (Fig. 3A) . All subsequent inhibitor experiments were performed in the presence of 25 ng/mL VEGF. Both AG126 and PD98059 at 10 μM (targeting ERK and MEK, respectively) significantly inhibited (P < 0.05) VEGF-induced proliferation of BRMECs, indicating that the MAP kinase pathway is involved in this response (Figs. 3B 3C) . Concentrations of 100 μM were toxic. 

Effect of AG126 and PD98059 on RhVEGF₁₆₅-Induced Tube Formation

Baseline studies were performed to quantify the effects of VEGF on the formation of tubelike structures in a collagen-based matrix. VEGF concentrations of 0, 1, 10, 25, and 50 ng/mL were applied to the cells sandwiched between the two layers of the collagen matrix (Vitrogen 100; Cohesion). BRMECs exposed to SF medium generally demonstrated very little organization, whereas incubation in 25 ng/mL VEGF for 24 hours elicited widespread formation of tubelike structures (Fig. 4A) . Preincubation with 10 μM PD98059 or 10 μM AG126, concentrations sufficient to inhibit proliferation of BRMECs, did not prevent the formation of organized tubelike structures when cultures were stimulated by VEGF (Fig. 4B) . Conversely, 0.3 μM herbimycin A completely inhibited VEGF-induced tube formation (difference from 0 ng/mL VEGF, NS). 

Effect of PD98059 on Phosphorylation of ERK and MEK in Retinas of Oxygen-Treated Rats

Both ERK isoforms were more phosphorylated in retinas of oxygen-treated rats than they were in the retinas of room air control rats at 20 days of age (P < 0.05; Figs. 5A 5B ). The same was true of the samples probed with antibodies against phospho-MEK (P < 0.05). Significant differences were not noted at 14 days, the onset of the post–oxygen-exposure room air period. Intravitreous injection of 10 μM PD98059, which is specific for MEK-1, eliminated phosphorylation of ERK-1 within 10 minutes and phosphorylation of ERK-2 after 4 hours (Fig. 5C) . 

Effect of Intravitreous Injection of AG126 or PD98059 on Retinal VEGF

To investigate a possible role for MEK and ERK upstream of VEGF, ELISA was performed to measure retinal VEGF levels in rats treated with the inhibitors. Retinal VEGF was increased four- to fivefold in oxygen-exposed rats at the time of removal from the chamber compared with levels in age-matched room air–exposed rats (P < 0.002; Fig. 6 ). Rats injected with AG126 or vehicle at this time showed no difference in retinal VEGF when levels were compared 1 day later (P > 0.2). Nor did PD98059 injection result in a significant reduction in retinal VEGF relative to contralateral, vehicle-injected eyes (P > 0.2). However, there is evidence of a systemic effect of intravitreous injection of PD98059. Even the contralateral, vehicle-injected eyes showed significantly reduced (P < 0.01) retinal VEGF relative to contralateral eyes from AG126-treated rats that received precisely the same vehicle (0.5% DMSO). 

Effect of Intravitreous Injection of AG126 or PD98059 on Angiogenesis in the Variable-Oxygen Model of ROP

The importance of the MAP kinase pathway in the rat model of ROP is evidenced by the effect of the two specific inhibitors. Dose–response experiments identified 10 μM as the optimum concentration for in vivo studies (Fig. 7) . At this concentration, both compounds inhibited pathologic retinal angiogenesis (Fig. 8) . Retinas treated with 10 μM AG126 averaged 1.6 ± 0.3 clock hours of neovascular area compared with 5.6 ± 0.9 clock hours in DMSO vehicle-injected control retinas (P < 0.005). Retinas treated with 10 μM PD98059 retinas had an average area of 3.3 ± 0.6 clock hours of retinal disease, whereas the DMSO vehicle-injected control animals had an average area of 6.3 ± 1.0 clock hours of pathologic vessels (P < 0.05). Examples of AG126, PD98059, and vehicle-injected retinas are illustrated in Figure 9 . 

Notably, these compounds successfully inhibited the formation of neovascular tufts without altering the growth of the previously attenuated intraretinal vasculature. Retinal vascular areas were 30.4 ± 2.5 μm² in AG126-treated eyes, 32.2 ± 2.7 μm² in the age-matched control (P > 0.2), and 29.8 ± 1.9 and 31.7 ± 2.6 μm² with PD98059 and its vehicle control, respectively (P > 0.1). 

Discussion

Several antiangiogenic therapeutic strategies have targeted VEGF, including soluble VEGF receptor-Ig chimeras, ⁵³ antibodies targeting the VEGF-binding region of the receptors, ⁵⁴ and altered VEGF isoforms. ⁵⁵ The partial success of these studies serves to reinforce the potential value of VEGF-directed therapies for the angiogenic aspect of vision-threatening ocular conditions. 

Our initial in vitro studies demonstrated that BRMECs respond to exposure to VEGF with increased phosphorylation of ERK-1 and -2, in agreement with previous studies in other systems. ^{40 41 42 56 57} This increase in phosphorylation is paralleled by an increase in kinase activity, as estimated by Elk-1 phosphorylation. We have also observed that blocking the activity of MEK and ERK kinases prevents VEGF-induced proliferation in vitro. Hence, we provide further evidence that the MAP kinases ERK-1 and -2 are important in the VEGF signal transduction cascade of ECs. 

Angiogenesis, however, is a complex and multifaceted process involving more than simple proliferation of ECs. It also is dependent on extracellular matrix proteolysis, migration of ECs, and remodeling. These events can be approximated in vitro with the tube-formation assay, where the cells aggregate into clusters, and extend processes that undergo formation of lumens and develop highly organized capillary structures. The ability of VEGF to stimulate the formation of tubelike structures in vitro has been well documented. ^{6 48 58 59} In contrast to the clear role of ERK-1/2 MAP kinases in VEGF-induced proliferation of BRMECs, our data suggest that they are not essential for differentiation of BRMECs under the conditions tested. At concentrations sufficient to significantly retard VEGF-induced proliferation, AG126 and PD98059 had no effect on VEGF-induced tube formation. This indication that the ERK-1/2 MAP kinases contribute to signal transduction pathways leading to VEGF-induced EC proliferation, but not tube formation, has not been reported previously. The role of ERK in tube formation of EC stimulated with fibroblast growth factor (FGF), however, has been investigated recently. Tanaka et al. ⁶⁰ found that ERK-1/2 is necessary for FGF-stimulated tube formation in a collagen matrix, using MSS31 murine ECs. They also found that FGF-induced tube formation could be prevented if the cells were incubated in the presence of the MEK inhibitor PD98059. By contrast, Klint et al. ⁶¹ found that PD98059 had no effect on FGF-initiated tube formation in capillary ECs derived from brain of transgenic “Imortomice,” although protein analysis indicated that kinase activation was completely blocked for the duration of the assay. Thus, the MAP kinase signal-transduction pathway appears critical to VEGF-initiated proliferation of BRMECs, whereas its role in differentiation is less clear. The discrepancies between our findings and those of Tanaka et al. ⁶⁰ may be due to differences in cell type, genetic background, growth factor or some as yet unidentified feature of the experiments. Figure 10 provides a hypothetical schematic of BRMEC VEGF transduction, based on the results of these experiments and others not included in this report. 

Using the information gained from the in vitro experiments, we turned our attention to our in vivo model of retinopathy of prematurity. Phosphorylation of both ERK-1/2 and MEK was increased in oxygen-exposed rats compared with room air–exposed control rats. Phosphorylation of ERK-1 and -2 was increased 19% and 46%, respectively, in samples obtained at the time experimental animals were taken from variable oxygen. The increases were 34% and 64%, respectively, 6 days later when angiogenesis peaks in this model. Similarly, MEK phosphorylation was increased 42% at the time of removal and 54% 6 days later. This suggests that extended stimulation by retinal VEGF occurred during the 6 days after exposure to oxygen, an implication that is supported by extensive work in rodent ROP models. ^{63 64 65} Increased phosphorylation of ERK-1/2 has been observed in other models of retinal ischemia. These studies examined rats and cats subjected to hypoxia created by retinal vein occlusion or rats subjected to ischemia–repurfusion injury after ligation of the optic nerve for 2 hours. Although the vein occlusion experiments were followed up for up to 6 weeks, no subsequent angiogenic component was demonstrated. ^{41 42 66}  

Intravitreous injections of PD98059 and AG126 were each found to be effective at inhibiting ERK phosphorylation and retinal neovascularization, indicating that the MAP kinase pathway plays a role in the rat ROP model. The in vitro data outlined in this study and information published elsewhere strongly suggest that blocking the MAP kinase pathway inhibits VEGF-induced BRMEC proliferation. Yet, neither AG126 nor PD98059, as determined by the tube-formation assay, affected differentiation of BRMECs. This may explain why these agents inhibited extraretinal vessel growth, but did not affect intraretinal growth of primary vessels. The former occurs by an angiogenic mechanism, which contains a critical cell proliferation component. The latter occurs primarily by a vasculogenic mechanism, involving differentiation of pluripotent precursors. This discrimination capacity increases the therapeutic potential of MAP kinase inhibitors to address ROP, a condition that arises when both processes are ongoing simultaneously in one location. 

Extrapolation of in vitro findings to the in vivo rat model of ROP is complicated by the likelihood that the targeted signal transduction proteins are also found upstream of local VEGF release. VEGF mRNA is upregulated in hamster fibroblasts in an ERK-1/2-MAP kinase–dependent manner. ⁶⁷ PD98059 blocks IL-1–induced production of a variety of cytokines, including VEGF, in human synoviocytes and synovial sarcoma cells. ⁶⁸ More relevant, adenosine agonists cause an ERK-dependent increase in VEGF mRNA and protein expression in human retinal endothelial cells (HRECs) that is blocked by PD98059. ⁶⁹ This inhibitor also exhibited the capacity to reduce VEGF in the retinas of our rats, although the route of its influence was apparently systemic, rather than local. Paradoxically, AG126 demonstrated no effect on retinal VEGF levels, but performed better in the efficacy trial. 

This report describes the first time that PD98059 or AG126 has been injected into the vitreous of an animal. The possibility that these agents can affect EC functions unrelated to VEGF or can affect other retinal cell types should be considered. For example, the influence of adenosine analogues, such as 5′-N-ethylcarboxamido-adenosine (NECA), on HRECs is not limited to upregulation of VEGF. NECA independently induces ERK-dependent proliferation, migration, and tube formation in these cells, and PD98059 inhibits both the ERK phosphorylation and the cells’ angiogenic behavior. ⁶⁹ Like VEGF, adenosine is increased during retinal ischemia, and evidence points to a role for adenosine in both vasculogenesis and angiogenesis. ⁷⁰ Thus, although our in vitro experiments focused exclusively on VEGF, inhibition of adenosine signal transduction could have played a role in our rat experiments. We conclude that MAP kinase inhibitors can provide angiostatic influence both upstream and downstream of VEGF and through mechanisms unrelated to VEGF, amplifying their therapeutic potential. However, because MAP kinases are relatively ubiquitous signal intermediates, nonendothelial cells are likely to be affected by ERK inhibitors. Intravitreous injection of several neurotrophic factors, including brain-derived neurotrophic factor (BDNF), ciliary neurotrophic factor (CNTF), and fibroblast growth factor (FGF), causes increased phospho-ERK immunoreactivity in Müller cells and in neurons of the inner retina. ^{71 72} Careful monitoring of retinal toxicity is a prerequisite for development of these or similar compounds for clinical application. 

 Supported in part by National Eye Institute Grant EY07533 (JSP), a Wasserman Merit Award to JSP, and a Challenge Grant to Vanderbilt Ophthalmology from Research to Prevent Blindness, Inc.

 Submitted for publication November 29, 2001; revised October 15, 2002; accepted November 12, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 8.

 

Both 10 μM AG126 and 10 μM PD98059 were effective at preventing neovascular disease from developing in retinas. Each compound significantly inhibited the formation of angiogenic vessels compared with the vehicle-injected controls (P < 0.005 for AG126 and P < 0.05 for PD98059). There was no apparent effect on growth of intraretinal vessels, as determined by measurement of retinal vascular area. Sample sizes for each treatment appear in parentheses.

Figure 9.

 

Comparison of ADPase-stained, flatmounted retinas from three treatment groups. (A) DMSO vehicle (0.05%), (B) 10 μM AG126, and (C) 10 μM PD98059. (A, arrows) Locations of neovascular tufts in the vehicle-treated tissue.

Figure 7.

 

AG126 (A) and PD98059 (B) inhibited neovascularization in a dose-dependent manner. A 10-μM concentration of each inhibitor was selected for further studies. Sample sizes for each treatment appear in parentheses.

Figure 1.

 

(A) VEGF dose–response profile with a 10-minute incubation. The blot was probed with anti-phospho-ERK1/2. (□) Mean integrated density of phospho-ERK-1 bands on this and all other anti–ERK-probed Western blots; (▪) phospho-ERK-2. (B) VEGF time course profile. Cells were exposed to 25 ng/mL VEGF, and the blot was probed with anti-phospho-ERK1/2 and stripped and reprobed with anti-ERK1/2. (C) VEGF dose–response profile using 10-minute incubation. After immunoprecipitation with anti–phospho-ERK, samples were incubated with Elk-1 fusion protein. The resultant mixture was analyzed by electrophoresis and anti-Elk-1 immunoblot analysis.

Figure 2.

 

(A) BRMECs were exposed to 25 ng/mL VEGF for 10 minutes. ERK-1/2 phosphorylation was inhibited by PD98059 in a dose-dependent manner. (B) Elk-1 phosphorylation by immunoisolated ERK-1/2 from VEGF-stimulated BRMEC was reduced by 10 μM, but not 0.1 μM, AG126.

Figure 3.

 

(A) After a 24-hour incubation in SF media, BRMECs were incubated in VEGF for 24 hours, where 25 ng/mL stimulated maximum proliferation. (B) PD98059 and (C) AG126 each inhibited the VEGF-induced BRMECs proliferation in a dose-dependent manner. Concentrations of 10 μM of each compound significantly inhibited (P < 0.05) the VEGF-induced response without significant toxicity (43% and 38% inhibition, respectively). Treatment with 100-μM concentrations of either inhibitor was significantly toxic to the cells.

Figure 4.

 

(A) After 24 hours under the conditions tested, maximum BRMEC tube formation was stimulated by 25 ng/mL VEGF. (B) The addition of 10 μM AG126 or 10 μM PD98059 did not inhibit the formation of VEGF-induced tubelike structures, but 0.3 μM herbimycin A did (P < 0.002).

Figure 5.

 

Retinas harvested from oxygen-treated and room air–exposed animals at 14 and 20 days of age showed a significant increase in phosphorylated ERK-1/2 in samples from oxygen-treated rats at 20 days (A; P < 0.05). The same samples were incubated with antibodies against phospho-MEK, and then stripped and reprobed with anti-MEK. Phosphorylated retinal MEK also was increased significantly in oxygen-treated rats (OX) compared with room air–exposed control animals (RA) at 20 days (B; P < 0.05). No significant differences in total MEK were observed in the samples. Intravitreous injection of PD98059 at 14 days inhibited phosphorylation of ERK-1/2 in retinas of oxygen-treated rats (C). The top blot was probed with anti-phospho-ERK-1/2. The bottom blot was probed with anti-ERK-1/2. Significant differences (P < 0.05) were detected in ERK-1 phosphorylation at 10 minutes and in ERK-2 phosphorylation at 4 hours. (A, C) (□) ERK-1; (▪) ERK-2.

Figure 6.

 

ELISA revealed a fivefold increase in retinal VEGF levels of oxygen-treated rats compared with room air–treated rats at the time of removal from the exposure chamber (P < 0.002). AG126 treatment at this time resulted in no effect on retinal VEGF measured 1 day later (P > 0.2). PD98059 treatment resulted in no difference between vehicle- and drug-injected eyes, but showed a 37% reduction relative to noninjected eyes (P < 0.01); the contralateral, vehicle-injected eyes showed a 47% reduction (P < 0.01). Sample sizes for each treatment appear in parentheses.

Figure 10.

 

Schematic representation of VEGF signal transduction in BRMECs. Each of the indicated inhibitors was used in the current study to define critical stages of the cascade. The focus of this report is on the roles of MEK and ERK and the effect of inhibiting their participation in VEGF transduction with PD98059 and AG126. Based on work in this laboratory and others, ⁶² but beyond the scope of this report, we conclude that the branch point that discriminates the tube formation signal from the proliferation signal occurs at the level of Raf serine/threonine kinase.