Expression of Pigment Epithelium–Derived Factor in Experimental Choroidal Neovascularization

Abstract

purpose. To investigate the expression of pigment epithelium–derived factor (PEDF) in the rat laser-injury model of choroidal neovascularization (CNV).

methods. Retinas were immunostained for PEDF at different times (1, 2, and 3 weeks) after laser injury. Levels of PEDF protein in the vitreous at 1, 3, 7, 14, and 28 days after laser injury were also assayed by Western blot.

results. Protein levels of PEDF in the vitreous were increased during the first 7 days after CNV induction. Immunostaining for PEDF was observed throughout normal nonlasered control retinas, sham-lasered retinas, and areas remote to laser lesions, which were generally more intense in the outer nuclear layer (ONL) and less intense in the internal nuclear layer (INL). Decreased expression of PEDF was observed in flanking areas adjacent to the injury site and was confined mainly to the ONL. In the injury sites, immunostaining within the ONL was either absent or decreased for up to 3 weeks after laser injury (the duration of the study). Preadsorption of the anti-PEDF antibody with the immunizing peptide blocked specific labeling in the retina.

conclusions. These results demonstrate an inverse correlation of expression of PEDF and formation of CNV in the experimental model and suggest that decreased expression of PEDF plays a permissive role in the formation of CNV. PEDF analogues may be a reasonable treatment strategy for CNV.

Age-related macular degeneration (AMD) is the leading cause of severe vision loss in people aged 65 and older in Western countries. ^{1 2 3} Dry AMD is the more common form of the disease, characterized by drusen and pigmentary and atrophic changes in the macula, with slowly progressive loss of central vision. Wet or neovascular AMD is characterized by subretinal hemorrhage, fibrosis, and fluid secondary to choroidal neovascularization (CNV), with more rapid and pronounced loss of central vision. ⁴ Although less common than dry AMD, neovascular AMD accounts for 80% of the severe vision loss due to AMD. ^{5 6} Current treatments for CNV are designed to destroy or remove the abnormal blood vessels and do not address the underlying stimuli responsible for neovascularization. ^{7 8} Development of pharmacologic therapy for CNV would be a major advance. One strategy could include the development of analogues to endogenous inhibitors of angiogenesis. 

Angiogenesis is controlled by the local balance between factors that either stimulate or inhibit vessel growth. In most normal tissues, inhibitory influences predominate, and vessels remain quiescent. ^{9 10} In contrast, in a variety of pathologic states, such as tumor growth and neovascular AMD, neovascularization occurs because of decreased production of inhibitors and/or increased production of angiogenic stimulators. ^{11 12}  

Although efforts have focused so far on identifying new angiogenic stimulators and investigating the role of these factors in ocular neovascularization, ¹³ little attention has been given to the identification of angiogenic inhibitor factors involved in ocular neovascularization. 

Pigment epithelium–derived factor (PEDF) is a member of the serine protease superfamily secreted by the RPE cells in the developing and adult retina. ¹⁴ It localizes to the interphotoreceptor matrix (IPM), the functional complex wherein the RPE interacts with the photoreceptors. PEDF is present in bovine IPM as a soluble extracellular monomeric glycoprotein that by itself confers neurotrophic activity to the IPM. ¹⁵ In vitro, it induces neuronal differentiation and promotes survival of the cerebellar granule neurons. ¹⁴ Gene expression of PEDF was found in human ciliary epithelium, and PEDF protein was found in the aqueous humor. ¹⁶ Furthermore, proteolytic activity directed toward PEDF was found in vitreous of bovine eyes, indicating that the vitreous has a serine-proteolytic activity that cleaves PEDF and may play a role in modulating PEDF in vivo. 

Expression of PEDF appears to be growth-state dependent: senescent cells do not express PEDF in vitro. ¹⁷ In addition, the PEDF gene is closely linked to an autosomal dominant locus for retinitis pigmentosa, ¹⁸ suggesting that PEDF could be a survival factor for photoreceptors. A role for PEDF has also been suggested in central areolar choroidal dystrophy (CACD), ¹⁹ autosomal dominant progressive cone dystrophy (CORDS), ²⁰ and Leber congenital amaurosis. ²¹ Intravitreal injections of PEDF delay the death of photoreceptors in mouse models of inherited retinal degeneration ¹⁴ and the RCS rat. Moreover, the vitreous humor is generally antiangiogenic and devoid of vessels, and it also normally contains a high concentration of PEDF. ²² PEDF has been shown recently to be a potent inhibitor of ischemia-driven retinal neovascularization. ^{23 24}  

Disruption of the Bruch membrane in the rat by laser treatment results in formation of CNV. ²⁵ This effect is usually attributed to changes in the level of angiogenic growth factors such as vascular endothelial growth factor (VEGF), which is upregulated during formation of CNV. ^{26 27} Capillaries originate within the choroid and extend through the disrupted Bruch membrane into the outer nuclear layer (ONL). Multiple layers of pigment-laden cells are found interspersed with choroidal capillaries within the neovascular membrane. Angiographic fluorescein leakage is observed in approximately half of laser-injury sites. Fluorescein leakage peaks within 2 to 3 weeks after induction and shows regression thereafter, although CNV may still be observed histologically. ²⁵ The purpose of this study was to determine the expression of PEDF during the course of CNV’s development in the rat model of laser-induced CNV. 

Materials and Methods

Experimental CNV Model

The laser-injury rat model of CNV was modified from earlier reports. ^{25 28} Forty adult male pigmented brown Norway rats (Jackson Laboratories, Bar Harbor, ME) were used in the study. The study design is summarized in Table 1 . All procedures were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the guidelines of the Massachusetts Eye and Ear Infirmary’s Animal Care Committee. The rats were anesthetized with an intraperitoneal injection of 0.2 mL of a 50:50 mixture of ketamine hydrochloride (20 mg/mL; Bayer Corp., Kansas City, MO) and xylazine hydrochloride (100 mg/mL; Abbott Laboratories, Abbott Park, IL). Animals were killed with an overdose of the same anesthetic mixture, followed by neck dislocation. 

The pupils were dilated with 1% tropicamide, and four to six photocoagulation lesions, using an argon laser (100-um spot size, 0.1-second duration, 120–160 mW; model 920; Coherent, Palo Alto, CA) were delivered between the retinal vessels in a peripapillary distribution in each fundus, using a slit lamp delivery system and a cover glass as a contact lens. Production of a bubble at the time of laser treatment confirmed the rupture of the Bruch membrane. Baseline fundus photographs were taken before laser treatment, and fluorescein angiograms were performed in each animal on the assigned date of death, with a fundus camera (TRC-50VT; Topcon, Paramus, NJ) with images captured on computer (Imagenet for Windows; Topcon) after an injection of 1 mL 1:10 diluted 10% fluorescein sodium (Alcon, Fort Worth, TX). Sham laser treatment involved the placement of three to four laser burns of 1000 μm at 120 to 160 mW, yielding an irradiance 1000 times less than the actual laser-injury lesions. No breaks in the bubble formation in the Bruch membrane or hemorrhages were noted in the sham-lasered eyes. 

Protein Electrophoresis and Western Blot Analysis

The eyes were enucleated at 1, 3, 7, 14, 21, and 28 days after laser photocoagulation (four rats at each time point). Control animals were four nonlasered rats. To remove the vitreous, eyes were bisected posterior to the iris, the lens was lifted out, and the vitreous removed and placed in tubes (Eppendorf; Brinkman Instruments, Westbury, NY) on ice after careful examination and dissection to remove any contaminating material. Rat vitreous from both eyes was pooled, and volume was measured by pipette aspiration. Next, vitreous was quickly homogenized with ice-cold lysis buffer (pH 7.5) containing 10 mM Tris, 130 mM NaCl, 1% Triton X-100, 10 mM NaF, 10 mM NaPi, 10 mM NaPPi, 16 μg/mL benzamidine, 10 μg/mL phenanthrolene, 10 μg/mL aprotinin, 10 μg/mL leupeptin, 10 μg/mL pepstatin, and 4 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF). An aliquot was removed for protein quantitation using a bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL), and the rest of the supernatant was stored at −70°C for future analysis of PEDF levels. 

Electrophoresis of proteins was performed with 12% SDS-polyacrylamide gels. All samples were boiled in denaturing sample buffer, and equal amounts of proteins were loaded on each lane. Proteins were separated at room temperature under reducing conditions at 120 V for 1 hour. Western blot transfer of separated proteins was performed at room temperature, using polyvinylidene difluoride membranes at 50 mA for 1 hour. To verify equal protein loading, blots were stained with 0.1% ponceau red (Sigma, St. Louis, MO) diluted in 5% acetic acid. Afterward, blots were blocked for 1 hour in TBS (10 mM Tris-HCl [pH 7.5], 150 mM NaCl) containing 5% nonfat dried milk. Next, the membranes were probed with a 1:250 dilution of primary antibody in TBS containing 2.5% nonfat dried milk for 1.5 hours. Rabbit polyclonal antibody against PEDF (anti-EPC-1) was a generous donation of Vincent J. Cristafalo and Mary K. Francis (Lankenau Medical Research Center, Wynnewood, PA). After incubation with primary antibody, the blots were washed for 30 minutes with frequent changes of TBS and blocked in 1% nonfat dried milk in TBS for 30 minutes, followed by incubation in a peroxidase-coupled secondary antibody for 1 hour in TBS containing 1% nonfat dried milk. The blots were washed for 1 hour with frequent changes of TBST (TBS with 0.1% Tween). Immunoblot analysis was performed using enhanced chemiluminescence with Western blot detection reagents (Amersham, Piscataway, NJ) followed by exposure to autoradiograph film (ML; Eastman Kodak, Rochester, NY). Blots were analyzed by densitometry on computer (ImageQuant software; Molecular Dynamics, Inc., Sunnyvale, CA). Amount of PEDF expressed in arbitrary units (AU) was normalized to the total amount of protein per sample. Data were plotted as AU PEDF per microliter vitreous versus time point. 

Histopathology

Twelve animals were killed after fluorescein angiography at 1 to 3 weeks after CNV induction. The control consisted of nonlasered eyes and sham-lasered eyes at 1 and 2 weeks. Four eyes were processed for each time interval. Eyes were enucleated, and the lens and anterior segment were removed. Remaining eyecups were fixed in formalin for 1 hour at room temperature and processed for paraffin-embedded sections. All sections were mounted on coated slides (Superfrost Plus; Fisher Scientific, Fairlawn, NJ). Serial sections were cut at 8-μm thickness. Before immunohistochemistry, some sections were stained with hematoxylin and eosin (H&E) and observed using a light microscope to help localize the CNV. 

Immunohistochemistry

Immunohistochemical staining with rabbit polyclonal anti-PEDF raised to 327 to 343 amino acid residues (generous donation of Noel Bouck, Northwestern University, Chicago, IL) was performed using the antigen-retrieval method. All incubations were performed in a moist chamber at room temperature. Briefly, sections were deparaffinized and then placed in antigen-retrieval solution (Dako, Carpinteria, CA) for 20 minutes at 95°, after which sections were blocked with 5% fetal bovine serum (FBS) in 0.1 M PBS (pH 7.4) for 1 hour. The sections were incubated with anti-PEDF (1:50 dilution in 1% FBS-PBS) overnight at room temperature. The slides were washed in PBS for 10 minutes, followed by incubation in a biotinylated-coupled secondary antibody (Jackson ImmunoResearch, West Grove, PA) for 1 hour in PBS containing 1% FBS. The slides were washed in PBS for 10 minutes and then visualized with the avidin-biotin complex (ABC) method (Vectastain Elite; Vector Laboratories, Burlingame, CA). Slides were incubated with diaminobenzidine to give a brown reaction product and were lightly counterstained with Mayer hematoxylin before mounting. Photographs were then obtained (Eclipse E600; Nikon, Melville, NY, and a Spot charge-coupled device [CCD] camera; Diagnostic Instruments, Sterling Heights, MI). Normal nonlasered and sham-lasered retinas were used as the positive control. Control sections were treated in the same way with omission of primary antibody or preincubation of the primary antibody overnight at 4°C with PEDF protein at a concentration of 100 μg/mL. 

The degree and pattern of immunostaining, both within and between specimens, was assessed by a masked histologist, using standard light microscopy. The intensity of labeling was graded qualitatively as grade 1, slightly stained; grade 2, moderately stained; grade 3, strongly stained; and grade 4, maximum staining as in the nonlasered control. To compare the immunostaining levels in a quantitative fashion, sections were imaged with a digital camera using identical microscope settings. Labeling in the ONL was then quantified by computer (Image; Scion Corp., Frederick, MD). 

Statistical Analysis

Data were analyzed using a one-way analysis of variance (ANOVA). Statistical significance was set at P ≤ 0.05. 

Results

PEDF Levels in Vitreous during the Course of CNV’s Development

To investigate changes in the vitreous levels of PEDF protein during the course of CNV’s development, CNV lesions were induced in rats, as previously described. Four animals were used per time point (Fig. 1) . At days 1, 3, 7, 14, 21, and 28 after induction, animals were killed, and vitreous was assayed for PEDF levels by Western blot analysis. Control experiments were performed with vitreous of eyes that had not been lasered. Densitometric analysis of Western blot was normalized to the volume loaded. Figure 2 represents the average of relative densities of vitreous PEDF over the time course of development of CNV. There seemed to be an increase in PEDF vitreous protein levels during the first week after induction of CNV. However, no difference was recorded afterward, and levels were comparable with control levels at week 1 and up to at least 28 days after CNV. 

PEDF Immunostaining in Laser-Induced CNV

Immunohistochemical expression of PEDF during the course of CNV was examined. Immunostaining was analyzed in 11 CNV lesions in 7 of 11 rats with laser injury at week 1 (two animals, two lesions), week 2 (three animals, seven lesions) and week 3 (two animals, two lesions). Control eyes were those of normal nonlasered rats and sham-lasered ones. All immunostaining was accompanied by appropriate control staining, and none of the negative control samples generated detectable immunostaining. The generation of laser burns required the use of pigmented animals. As a result of the density of melanin pigmentation, it was difficult to detect immunopositivity in cells in the choroid and RPE layer. Results of the immunohistochemical analysis by a masked histologist are summarized in Tables 2 and 3 . In addition, analysis of immunostaining by computer (Image; Scion) allowed quantitative analysis of intensity of staining and confirmation of findings (Fig. 3) . 

Nonlasered Control Eyes.

Immunolabeling for PEDF was observed throughout normal nonlasered control retinas of brown Norway rats. Intense immunostaining was observed in the ONL and moderate immunoreactivity in the INL. PEDF immunoreactivity in nonlasered eyes was most intense and was scored as grade 4. 

Sham-Lasered Eyes.

PEDF immunoreactivity in sham-lasered retinas was found to be diminished compared with that in nonlasered retinas. Immunolabeling in sham eyes was typically grade 3. 

Laser-Treated Eyes.

Marked focal downregulation of PEDF’s expression (weak immunoreactivity) was observed throughout the CNV regions (Fig. 4C 4D) and adjacent flanking areas (Fig. 4B) at 1, 2, and 3 weeks after laser injury. Immunoreactivity in the area of CNV was judged to be predominantly grade 1 (81.8% grade 1 and 18.25% grade 2), whereas that of flanking regions was predominantly grade 2 (27.3% grade 1 and 72.7% grade 2). Downregulation of PEDF signal persisted through the 3 weeks of follow-up. 

In the nonlasered regions, including areas remote to the laser injury (Fig. 4A) , there was little variation in the immunolabeling pattern for PEDF at the times after laser treatment, compared with sham-lasered retinas (Fig. 4E) . Areas remote to CNV were found to be predominantly grade 3 (12.5% grade 2, 62.5% grade 3, and 25% grade 4) and sham-lasered retinas were all grade 3. 

The intensity of PEDF staining using image software in sham-lasered, normal, areas remote to the CNV was almost two times that overlaying CNV and flanking areas (Fig. 3) . 

PEDF Immunostaining in CNV at 1, 2, and 3 Weeks after Laser Injury.

PEDF immunostaining in CNV at 1 and 2 weeks after laser injury was mostly grade 1, whereas in CNV 3 weeks after laser treatment, injury was grade 2. Staining in flanking areas at 1 and 3 weeks after laser injury was grade 2 and was distributed almost equally between grade 1 and grade 2 at 2 weeks. Staining in areas remote to the CNV was mostly grade 3 at 1 week, grade 2 at 2 weeks, and grade 4 at 3 weeks after laser injury (Table 3 , Fig. 5 ). 

Discussion

The purpose of this study was to investigate changes in expression of the antiangiogenic factor PEDF in the retina after laser injury over the course of CNV’s development and to try to correlate these findings with intravitreal levels of PEDF. First, our results demonstrate that PEDF immunostaining in the normal rat retinal layers localized mainly to the ONL—the most avascular layer of the retina—and, to a lesser extent, to the INL. The rat vitreous was found to contain abundant amounts of PEDF proteins. This confirms the findings of others ^{15 23} who have studied murine and bovine retinas and vitreous. 

We also found that there was a significant increase in vitreous protein levels of PEDF during the first 7 days after CNV induction; however, PEDF levels returned to baseline control levels thereafter. The increase of PEDF in the vitreous after laser injury may be the result of PEDF’s release from damaged retinal cells, in particular from the RPE cells that produce it and the IPM where it localizes. These findings must be interpreted with caution, because changes in the level of an abundant protein may be difficult to detect after focal injury. 

In the laser-injury model of CNV, expression of PEDF during CNV’s development followed a pattern that might be expected of an angiogenic inhibitor, in that the amount of inhibitory PEDF was downregulated in the areas of laser injury and CNV’s formation, suggesting that its loss plays a permissive role in formation of CNV. It is worth noting that mild laser injury (sham laser) alone without subsequent development of CNV appeared to result in mild focal downregulation of PEDF’s signal compared with nonlasered retinas, but to a much lesser extent than the downregulation witnessed in the areas that sustained significant injury and in which CNV developed. Moreover, during the laser induction of CNV, some RPE cells are invariably injured, and this may contribute to the decreased immunostaining for RPE-produced PEDF. However, the number of RPE cells that were damaged was small, and flanking areas with undamaged RPE also showed decreased expression of PEDF. In addition, RPE-produced VEGF has been shown to be increased after laser-induced CNV. 

The most pronounced downregulation of PEDF’s expression occurred in 2-week-old CNV lesions, which corresponds to the peak temporal incidence of angiographically leaking CNV, as described by other groups. ^{25 28} The PEDF downregulation pattern during CNV’s development seems to parallel somewhat the upregulation of expression of VEGF, ²⁹ the latter being strongest at 1 week after photocoagulation in lasered lesions and decreased by 4 weeks. 

The mechanism of PEDF’s action is still being elucidated. However, a recent report by Stellmach et al. ²⁴ shows that PEDF causes apoptosis in activated endothelial cells, and this may be the mechanism of inhibition of angiogenesis. They found that PEDF induces apoptosis in cultured endothelial cells and leads to an eightfold increase in the number of apoptotic endothelial cells as detected in situ, when the ischemic retinas of PEDF-treated animals were compared with vehicle-treated control retinas. Recent evidence points to the possible existence of a PEDF receptor. ³⁰ Purification, isolation, and characterization of such receptors may help unravel the mechanism of action and regulation of PEDF. PEDF agonists that bind PEDF receptors may be a reasonable therapeutic approach to CNV. Because PEDF synthesis and secretion are diminished in senescent cells, ¹⁷ it can be speculated that senescent RPE in AMD may produce less PEDF and thereby facilitate the development of CNV. The Tie family of receptors and their ligands the angiopoietins (Ang) also appear to be involved in the control of vascular proliferation or regression. A recent report by Hangai et al. ³¹ demonstrated that Ang1 and Ang2 colocalize with VEGF in CNV stromal cells and that VEGF induces the upregulation of Ang1’s expression in cultured RPE cells. A study of the temporal expression of angiopoietins during the course of development of CNV would be of great interest and help elucidate the interrelationship of PEDF, VEGF, and angiopoietins. 

In summary, expression of PEDF has been shown to be decreased at sites of laser injury and development of CNV, suggesting that its release or activation may be part of that development. Because focal downregulation of PEDF may play a role in the pathogenesis of CNV, these results suggest that PEDF analogues may be a useful treatment for CNV in AMD and other diseases. 

 Supported by the Massachusetts Lions Eye Research Fund and the S. Elizabeth O’Brien Trust.

 Submitted for publication March 23, 2001; revised December 26, 2001; accepted January 14, 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.

 

Table 1.

 

Summary of Animals (n = 40)

Eyes (n)	Group	Time Point at Examination*	Assay
8	Control, not lasered	None	Immunoblot of vitreous PEDF protein level
8	Laser injury	1 day post-injury	Immunoblot
8	Laser injury	3 days post-injury	Immunoblot
8	Laser injury	7 days post-injury	Immunoblot
8	Laser injury	14 days post-injury	Immunoblot
8	Laser injury	21 days post-injury	Immunoblot
8	Laser injury	28 days post-injury	Immunoblot
2	Control, no laser injury	None	Immunostaining for PEDF
4	Sham laser	1 week post-injury	Immunostaining
4	Sham laser	2 weeks post-injury	Immunostaining
4	Laser injury	1 week post-injury	Immunostaining
6	Laser injury	2 weeks post-injury	Immunostaining
4	Laser injury	3 weeks post-injury	Immunostaining

 Time after laser-induced injury.

Figure 1.

 

Sample fluorescein angiogram 3 weeks after laser injury. Late fluorescein angiogram showing hyperfluorescence and leakage of fluorescein in the photocoagulated lesions, demonstrating the formation of choroidal neovascularization.

Figure 2.

 

Quantitative densitometric analysis of PEDF vitreous levels after CNV induction over time. The average density per microliter vitreous is plotted (n = 4, ± SD). Results show an increase in PEDF vitreous level during the first 7 days after CNV induction (P ≤ 0.05).

Table 2.

 

Grading of PEDF Immunostaining in Rat CNV

	Grade 1	Grade 2	Grade 3	Grade 4
CNV (n = 11)	81.8	18.2	0	0
Flanking CNV (n = 11)	27.3	72.7	0	0
Remote from CNV (n = 8)	0	12.5	62.5	25
Sham (n = 2)	0	0	100	0
Nonlasered (n = 2)	0	0	0	100

 Data are percentage of total group achieving each grade.

Table 3.

 

Grading of PEDF in 1-, 2-, and 3-Week-Old CNV

Time after Induction	CNV	Flanking CNV	Remote from CNV
1 Week (n = 2)	Grade 1 (100%)	Grade 2 (100%)	Grade 3 (100%)
2 Weeks (n = 7)	Grade 1 (85.7%)	Grade 1 (42.9%)	Grade 1 (20%)
	Grade 2 (14.3%)	Grade 2 (57.1%)	Grade 2 (80%)
3 Weeks (n = 2)	Grade 2 (100%)	Grade 2 (100%)	Grade 4 (100%)

 Data are immunostaing grades, with percentages of total group at 2 weeks in parentheses.

Figure 3.

 

Quantitative analysis of PEDF immunostaining intensity in the ONL after experimental choroidal neovascularization induction. The average pixel intensity is plotted ± SD.

Figure 4.

 

Focal downregulation of PEDF protein expression over laser-induced CNV in adult rat. Retinas were harvested at 1, 2, and 3 weeks after induction from brown Norway rats and stained for PEDF. Note accumulation of PEDF (E, arrow), stained reddish brown. Control sections directly adjacent to lesion (B), remote to lesion (A) or sham (E) stained with primary antibody, (F) with preadsorbed primary antibody, or (G) without primary antibody. Note downregulation of PEDF (D; arrowhead) in the area of CNV (C, D; arrow). Retina layers indicated in (B) include retinal pigment epithelium (RPE), outer nuclear layer (ONL), inner nuclear layer (INL), and ganglion cell layer (GCL). Rat eyes were fixed in formalin within 1 to 5 minutes of harvest. For immunostaining, paraffin-embedded sections were incubated with anti-PEDF and visualized with the avidin biotin complex method. Scale bar, 50 μm.

Figure 5.

 

PEDF immunostaining grade in 1,2 and 3 week-old CNV.