VEGF Increases Retinal Vascular ICAM-1 Expression In Vivo

Abstract

purpose. Intraocular injections of vascular endothelial growth factor (VEGF), a peptide implicated in the pathogenesis of diabetic retinopathy, can induce retinal ischemia. Diabetic retinal ischemia may be caused, in part, by the adhesion of leukocytes to the retinal vasculature. In this study, the ability of VEGF to increase the expression of intercellular adhesion molecule-1 (ICAM-1) and other adhesion molecules in capillary endothelium and the retinal vasculature was examined.

methods. The expression of ICAM-1, vascular cell adhesion molecule-1 (VCAM-1), E-selectin, and P-selectin on human brain capillary endothelial cell monolayers exposed to VEGF was quantitated by immunoassay. The effect of VEGF on retinal vascular ICAM-1 expression was determined in ICAM-1 immunofluorescence studies of retinal flat-mounts and in RNase protection assays.

results. VEGF increased capillary endothelial cell ICAM-1 levels in a dose- and time-dependent manner (6–24 hours, plateau after 6 hours; EC₅₀, 25 ng/ml). VEGF failed to alter E-selectin, P-selectin, or VCAM-1 levels under the conditions tested. Intravitreal injections of pathophysiologically relevant concentrations of VEGF increased ICAM-1 protein and mRNA levels in the retinal vasculature.

conclusions. VEGF increases retinal vascular ICAM-1 expression. VEGF-induced increases in ICAM-1 may promote retinal leukostasis in diabetic eyes.

Vascular endothelial growth factor (VEGF) is a major mediator of retinal ischemia–associated ocular neovascularization. Intraocular VEGF levels have been temporally and spatially correlated with neovascularization, ¹ and the specific inhibition of VEGF in animal models has been shown to suppress the neovascularization of the iris ² and the retina. ³ More recently, exogenous VEGF has also shown to induce many of the angiographic and histopathologic features of diabetic retinopathy in nonhuman primates, including retinal ischemia. ⁴ The cellular and molecular mechanisms underlying this phenomenon have not been defined. 

Fluorescein angiographic studies in humans have demonstrated that some diabetic retinal ischemia is reversible. ⁵ However, in most eyes a large proportion of the diabetic retinal ischemia that develops is irreversible and results from the formation of acellular capillaries. ⁶ Diabetic retinal leukostasis ⁷ may be operative in the development of both types of retinal ischemia. Increased numbers of monocytes and granulocytes have been demonstrated to occupy the lumens of retinal capillaries and postcapillary venules in diabetic rats, with some leukocytes lying adjacent to dying endothelial cells. ⁸ Increased numbers of granulocytes have also been demonstrated in the retinas of diabetic humans. ⁹ One potential mechanism mediating this phenomenon is the increased adherence of leukocytes to the vascular endothelium. Intercellular adhesion molecule-1 (ICAM-1) mediates the adhesion of neutrophils and monocytes to vascular endothelium, and increased ICAM-1 levels have been found in the retinal vasculature of human diabetics. ⁹ In addition, human diabetic neutrophils have been shown to be more prone to upregulate ICAM-1 ligand CD18. ¹⁰  

Because of these observations, studies were undertaken to characterize the effect of VEGF on the expression of endothelial cell adhesion molecules in vitro and in retina in vivo. 

Methods

Cell Culture

Human brain capillary endothelial cells (HBCEC) were obtained from Cell Systems Corporation (Kirkland, WA) and maintained in medium 199 (GIBCO BRL, Gaithersburg, MD), containing HEPES (25 mM), heparin (1%), endothelial cell growth factor (50 μg/ml), L-glutamine (2 mM), 100 U/ml penicillin, 100 mg/ml streptomycin, and 15% fetal bovine serum (HyClone Laboratories, Logan, UT), according to the manufacturer’s instructions. For enzyme immunoassay (EIA), HBCECs were grown on tissue culture–treated plastic microtiter 96-well plates, coated with 0.1% (wt/vol) gelatin (Difco, Detroit, MI), and allowed to reach confluence. 

Adhesion Molecule Quantitation via EIA

Endothelial cell adhesion molecule levels were measured by EIA in HBCECs. HBCECs were grown to confluence in 96-well plates and treated with serum-free medium with or without various concentrations of recombinant human VEGF₁₆₅ (a gift of Napoleone Ferrara, Genentech, South San Francisco, CA; endotoxin < 0.03 EU/ml) or recombinant human tumor necrosis factor-α (TNF-α; R&D Systems, Minneapolis MN; positive control) for 24 hours. Cell surface EIAs were completed using mouse anti-human monoclonal antibodies against ICAM-1 (Ab HU 5/3), VCAM-1–1 (Ab E1/6), and E-selectin (Ab H18/7). ¹¹ The EIAs were carried out by incubating monolayers first with saturating concentrations of specific monoclonal antibodies against the target molecule, followed by biotinylated goat anti-mouse IgG, and finally with streptavidin-alkaline phosphatase. The surface expression of each protein was quantified spectrophotometrically, reading the optical density of the wells (410 nm) 15 to 60 minutes after the addition of a chromogen (p-nitrophenylphosphate), as described previously. ¹¹ The data were normalized against protein concentrations and expressed as mean ± SEM. 

ICAM-1 Immunofluorescence

Animals were cared for in accordance with the ARVO Resolution on the Use of Animals in Ophthalmic and Vision Research.. The experiments were approved by the Schepens Eye Research Institute Animal Care and Use Committee. Normal C57B/6 mice (male, 6 to 8 weeks old) were examined for retinal ICAM-1expression after intravitreal injections of recombinant murine VEGF₁₆₄ (R&D Systems, Minneapolis, MN) or vehicle alone (n = 6; endotoxin < 0.1 EU/mg). After anesthesia with a mixture of ketamine (Ketalar; Parke–Davis, Morris Plains, NJ; [150 mg/kg]) and xylazine (Rompun; Harver–Lockhart, Morris Plains, NJ;[ 60 mg/kg]), 2 μl concentrated VEGF or phosphate-buffered saline (PBS) solvent was injected slowly into the vitreous cavity to attain a final concentration of 100 ng/ml VEGF. After 24 hours, the animals were killed by cervical dislocation, and the eyes were enucleated and immersion-fixed in 0.5% (wt/vol) paraformaldehyde in 0.1 M PBS (pH 7.4) for 30 minutes. The intact retinas were dissected out under a surgical microscope as previously described ¹² and further fixed in 70% ethanol at room temperature for an additional 30 minutes. The retinas were washed in 0.1 M PBS (pH 7.4) with 0.1% (vol/vol) Triton X-100 for 30 minutes and then incubated with biotinylated anti-mouse ICAM-1 (1:50 diluted in 0.1 M PBS containing 1% [wt/vol] bovine serum albumin, BSA) overnight at 4^oC. After washing at room temperature with PBS containing 1% BSA, the retinas were further incubated with streptavidin–fluorescein isothiocyanate (1:100 diluted in PBS containing 1% BSA; Amersham, Arlington Heights, IL) for 2 hours at room temperature. An isotype-matched IgG was used as a negative control. The retinas were co-stained with 2 ng/ml ethidium bromide for 10 seconds and mounted. Endothelial cell nuclei were counted per unit length of vessel. 

RNase Protection Assay for ICAM-1

The experiments were approved by Children’s Hospital Animal Care and Use Committee. Male Sprague–Dawley rats weighing 200 to 250 g were anesthetized with 0.1 mg/kg sodium amobarbital. Pupils were dilated with 0.5% tropicamide and 2.5% phenylephrine hydrochloride (Alcon, Humancao, Puerto Rico). Murine VEGF₁₆₅ or human TNF-α, in a total volume of 5 μl PBS, was injected into the right vitreous at a site 1 mm posterior to the limbus using a Hamilton syringe with a 30-gauge needle. The final concentrations of VEGF and TNF-α were calculated to be 100 and 10 ng/ml, respectively. The contralateral control eyes received 5-μl PBS injections. The eyes were enucleated 2.5 hours later. The rats were euthanatized with an anesthetic overdose followed by CO₂ incubation. All injections were done under direct observation using a surgical microscope. Any eyes that had damage to the lens or retina were not used for analyses. The retinas were gently dissected free and cut at the disc. The tissue was placed in an Eppendorf tube, snap-frozen in liquid nitrogen, and stored at −80^oC. The retinas were homogenized in 1 ml RNAzol (Biotecx Laboratories, Houston, TX) at 4^oC and prepared for RNase protection assays. 

The ICAM-1 riboprobe was produced by subcloning the coding sequence of the rat ICAM-1 cDNA into the EcoRI–BamHI site of the pBluescript II KS vector. Transcription by T7 RNA polymerase after linearization by EcoRI resulted in a probe of 225 nucleotides (nt). This probe protects a 166-nt fragment of ICAM-1. All samples were simultaneously hybridized with an 18S riboprobe (Ambion, Austin, TX) to normalize for variations in loading and recovery of RNA. Full-length protection of this probe results in an 80-nt fragment. The assay was performed as previously described. ¹³ Ten micrograms of total cellular RNA was hybridized with ³²P-labeled antisense ICAM-1 and 18S riboprobes (200,000 cpm of each) overnight at 42^oC in 30μ l hybridization buffer. Hybridized RNA was digested with nuclease P1 (20 μg/ml) and RNase T1 (2 μg/ml) for 1 hour at 25^oC in 300 μl digestion buffer. Digestions were terminated by the addition of 20 μl of 10% sodium dodecyl sulfate and 50 μg proteinase K for 15 minutes at 37^oC. After phenol/chloroform extraction and ethanol precipitation, the protected fragments were resolved on 6% polyacrylamide, 7 M urea gels and quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). 

Statistics

Significance was calculated using the paired Student’s t-test. P < 0.05 was deemed significant. 

Results

VEGF increased the expression of ICAM-1 on HBCEC in a time- and dose-dependent manner (6–24 hours, plateau after 6 hours; EC₅₀, 25 ng/ml) (Figs. 1 A, 1B). Expression levels of surface VCAM-1 (Fig. 2 A), E-selectin (Fig. 2B) , and P-selectin (Fig. 2C) were unaffected by incubation with VEGF₁₆₅ for 6 hours (data not shown) or 24 hours. Incubation of HBCEC monolayers with TNF-α (positive control) produced significant increases in VCAM-1 (2.4 ± 0.1–fold, n = 8, P < 0.05 versus control), E-selectin (17.5 ± 1.3–fold, n = 8, P < 0.05 versus control), P-selectin (10.7 ± 1.2–fold, n = 8, P < 0.01, data not shown), and ICAM-1 (2.5 ± 0.2–fold, n = 8, P < 0.05 versus control, data not shown). 

Compared to the PBS-injected eyes, the VEGF-injected eyes had a marked homogeneous upregulation of ICAM-1 immunoreactivity, particularly in the mid-peripheral retina and the postcapillary venules. There was also a moderate upregulation in the arterioles and capillary beds (Fig. 3) . Because VEGF is an endothelial cell mitogen, we examined whether VEGF increased endothelial cell numbers, possibly confounding the increases in ICAM-1 expression. Twenty-four hours of VEGF treatment did not significantly increase the number of endothelial cell nuclei per fixed segment length in the retinal vasculature (data not shown). 

To examine whether VEGF increases ICAM-1 expression at the transcript level, RNase protection assays of rat retinal RNA were performed after intravitreal injections of VEGF or PBS buffer. Compared to the PBS-injected eyes, the VEGF-injected eyes had a 2.1 ± 0.4–fold (n = 4, P < 0.05) upregulation of ICAM-1 mRNA (Fig. 4) . As a positive control, TNF-α induced a 3.1 ± 0.8–fold increase of ICAM-1 mRNA (n = 5, P < 0.05). 

Discussion

These data demonstrate that VEGF can increase ICAM-1 expression on capillary endothelial cells in vitro and in the retinal vasculature in vivo. The concentration of VEGF required to increase ICAM-1 in vitro (EC₅₀, 25 ng/ml) is comparable to that measured in the vitreous of eyes with retinal ischemia and neovascularization. ¹ The VEGF-induced effects appear to be specific for ICAM-1 under the conditions tested. These data are consistent with those of Melder and associates ¹⁴ showing an increase in bioactive ICAM-1 in angiogenic tumor endothelium in response to VEGF. However, our data differ regarding the VEGF-induced upregulation of VCAM-1, because it was not observed in our system. One reason for these divergent results may be the different types and strains of endothelial cells used in our studies. 

In addition to VEGF, other pathophysiologically relevant stimuli may increase ICAM-1 in the diabetic retinal vasculature. Factors relevant to diabetes, such as the cytokine TNF-α, ¹⁵ have been shown to increase ICAM-1 in other systems. TNF-α levels are increased in the serum of diabetics, ¹⁶ and as confirmed by our data, TNF-α is a potent inducer of endothelial cell ICAM-1 expression. As in other tissues, the induction of ICAM-1 may be causal for the leukostasis observed in the diabetic retina. In the brain, ICAM-1 mediates leukocyte adhesion to postcapillary venules and is associated with capillary occlusion and vascular endothelial cell damage. ¹⁷  

Retinal VEGF levels appear to be elevated early in diabetes, before there is any histopathologic evidence of retinal ischemia. ¹⁸ This makes it more likely that endogenous VEGF serves to increase retinal vascular ICAM-1 in vivo. We hypothesize that VEGF, in whole or in part, increases retinal ICAM-1 expression and that ICAM-1 mediates the binding of leukocytes to the vasculature, accounting for the early reversible phase of retinal ischemia. With time, the retinal leukostasis becomes chronic and widespread, hastening the death of endothelial cells and pericytes. With the development of more widespread ischemia, retinal VEGF levels are further upregulated, triggering greater ischemia and ultimately neovascularization. 

This hypothetical sequence of events remains to be proven. Proof requires the specific inhibition of VEGF and ICAM-1 in the diabetic retina and the direct monitoring of its functional and anatomic consequences. These studies are under way. 

 Reprint requests: Anthony P. Adamis, Massachusetts Eye and Ear Infirmary, 243 Charles Street, Boston, MA 02114.

⁵  These authors contributed equally to the work presented here and should therefore be regarded as equivalent senior authors.

 Supported by the National Eye Institute (APA), the Massachusetts Lions, and the Roberta W. Siegel Fund (APA).

 Presented in part at the 1997 and 1998 annual meetings of the Association for Research in Vision and Ophthalmology, Ft. Lauderdale, Florida, May 1997 and May 1998.

 Submitted for publication June 30, 1998; revised January 8, 1999; accepted February 16, 1999.

 Proprietary interest category: N.

 

Figure 1.

 

Time (A) and dose (B) responses of surface ICAM-1 levels in HBCECs after exposure to serum-free medium with and without VEGF₁₆₅. Time course experiment was performed using 50 ng/ml VEGF₁₆₅. Dose response of ICAM-1 (B) was performed after a 24-hour exposure to serum-free medium with and without VEGF₁₆₅. *P < 0.05 vs control, n = 8. Data are expressed as percentage over control (no VEGF treatment) optical density, normalized against protein concentration in each well.

Figure 2.

 

Expression of VCAM-1 (A), E-selectin (B), and P-selectin (C) in HBCECs after 24-hour exposure to serum-free medium with and without VEGF₁₆₅ or TNF-α.* P < 0.05 vs control, n = 8.

Figure 3.

 

ICAM-1 immunofluorescence study of C57B/6 mouse retina 24 hours after exposure to 100 ng/ml murine VEGF₁₆₄. Ethidium bromide staining identifies cell nuclei. Compared to the PBS-injected eyes (A), the VEGF-injected eyes (B) had a marked upregulation of ICAM-1 (1) in the retinal postcapillary venules and veins and a moderate upregulation in the arterioles and capillary bed. Twenty-four hours of VEGF treatment did not significantly increase the number of endothelial cell nuclei in the retinal vasculature (2). ICAM-1 (1) and ethidium bromide (2) staining are shown separately and superimposed (3). (Magnification, ×25.)

Figure 4.

 

Rat retina ICAM-1 RNase protection assays 2.5 hours after intravitreal injection of VEGF or TNF-α. The contralateral eyes were injected with PBS. These are representative data from 1 of 4 rats injected with VEGF and 1 of 5 rats injected with TNF-α. Graph shows 18S normalized expression levels.