Targeted Delivery of Anti–Angiogenic Agent TNP-470 Using Water-Soluble Polymer in the Treatment of Choroidal Neovascularization

Abstract

purpose. The conjugation of drugs with water-soluble polymers such as poly(vinyl alcohol) (PVA) tends to prolong the half-life of drugs and facilitate the accumulation of drugs in tissues involving neovascularization. The purpose of this study was to evaluate the effect of TNP-470–PVA conjugate on the proliferation of endothelial cells in vitro and on experimental choroidal neovascularization (CNV) in vivo.

methods. TNP-470 was conjugated in PVA by a dimethylaminopyridine-catalyzed reaction. The effects of TNP-470–PVA and free TNP-470 on the proliferation of human umbilical vein endothelial cells (HUVECs) and bovine retinal pigment epithelial cells (BRPECs) were evaluated by the tetrazolium-based colorimetric assay (XTT assay). Experimental CNV was induced by subretinal injection of gelatin microspheres containing basic fibroblast growth factor, into rabbits. Thirty rabbits were intravenously treated either with TNP-470–PVA (n = 8), free TNP-470 (n = 5), free PVA (n = 5), or saline (n = 12) daily for 3 days, 2 weeks after implantation of gelatin microspheres. Fluorescein angiography was performed to detect the area with CNV, and the evaluation was made by computerized measurement of digital images. These eyes were also examined histologically. To observe the accumulation of conjugate, 3 rabbits with CNV received rhodamine B isothiocyanate–binding PVA (RITC–PVA), and the lesion was studied 24 hours later by fluorescein microscopy.

results. The TNP-470–PVA inhibited the growth of HUVECs, similar to that of free TNP-470. The BRPECs were less sensitive to TNP-470–PVA than were the HUVECs. TNP-470–PVA significantly inhibited the progression of CNV in rabbits (P = 0.001). Histologic studies at 4 weeks after treatment demonstrated that the degree of vascular formation and the number of vascular endothelial cells in the subretinal membrane of the eyes treated with TNP-470–PVA were less than those of the control eyes. RITC–PVA remained in the area with CNV 24 hours after administration.

conclusions. These results suggest that TNP-470–PVA inhibited the proliferation of HUVECs more sensitively than that of BRPECs, and the targeted delivery of TNP-470–PVA may have potential as a treatment modality for CNV.

Age-related macular degeneration (AMD) is a major cause of visual impairment in patients over 50 years of age. Cases of severe visual loss are often caused by choroidal neovascularization (CNV). Laser photocoagulation, vitreous surgery, and radiation have been used for the treatment of AMD. ^{1 2 3 4 5 6 7 8 9 10 11 12} However, no satisfactory therapy has been established clinically to conserve visual function of patients with subfoveal CNV. ^{1 2 3 4 5 6 7 8 9 10 11 12} These treatments carry a potential risk of accelerating the activity of angiogenesis or of injuring normal retinal tissue unwillingly ¹³ and, therefore, cannot be applied in patients with AMD with good visual acuity in an early stage without hesitation. It is important to identify potent anti-angiogenic agents to arrest AMD in the incipient stage. 

TNP-470, a synthetic analogue of fumagillin, is a most promising anti-angiogenic agent, and it was demonstrated to inhibit the growth and the capillary-like tube formation of endothelial cells more sensitively than other types of cells. ^{14 15} Recently, TNP-470 has been tested as a potential new anticancer agent, because it is recognized that tumor angiogenesis is an essential phenomenon to sustain tumor growth over a few millimeters. ¹⁶  

Although TNP-470 showed a potent inhibitory effect on angiogenesis, side effects such as granulocytopenia and general fatigue also have been reported after general administration of TNP-470. ¹⁵ To apply this drug clinically to treat ocular angiogenesis such as AMD, drug targeting to CNV is necessary to increase site specificity and reduce side effects. Conjugation of the drug with water-soluble polymers such as poly(vinyl alcohol) (PVA), poly(ethylene glycol), and dextran has shown to prolong a circulating life of the drug and increase the accumulation of the drug in tissue with new vessels such as tumor mass. ^{17 18 19 20 21 22 23} We have developed a conjugate of TNP-470 with PVA (TNP-470–PVA) to apply in the treatment of CNV. 

In the present study, we investigated in vitro activity of TNP-470–PVA to inhibit the proliferation of endothelial cells and its in vivo effect on experimental CNV in rabbits. 

Materials and Methods

Chemicals

TNP-470 was kindly supplied by Takeda Chemical Industries (Osaka, Japan). Poly(vinyl alcohol) with an average molecular weight of approximately 220,000 and 4-dimethylaminopyridine were purchased from Wako Pure Chemical Industries (Osaka, Japan). Basic fibroblast growth factor (bFGF) was kindly supplied by Kaken Pharmaceutical (Osaka, Japan). All other chemicals were reagent-grade products obtained commercially. 

Synthesis of TNP-470–PVA

TNP-470 was introduced into the hydroxyl group in PVA. Briefly, 400 mg of PVA was dissolved in 40 ml of dimethyl sulfoxide. Then, 4-dimethylaminopyridine (1.11 g) and TNP-470 (0.345 g) were added. The mixture was kept at approximately 80°C for 3 hours, with occasional stirring. The product, TNP-470–PVA, was dialyzed against water, freeze-dried, and stored at −20°C. 

The conductivity of the solution was measured to calculate the binding molar ratio of the conjugate, using Conductivity Meter (model DS–12; HORIBA, Kyoto, Japan). The molar binding ratio of TNP-470 to PVA was estimated at approximately 60. 

Synthesis of Rhodamine B Isothiocyanate–Binding PVA

Rhodamine B isothiocyanate (RITC) was conjugated to PVA (RITC–PVA) with amino groups synthesized by being reacted with 6-bromohexanoic acid and ethylenediamine. In brief, 400 mg of PVA was dissolved in 10 ml of distilled water, and 6-bromohexanoic acid (3.52 g) in an 8 N NaOH solution (4.8 ml) was added stepwise. The mixture was kept at 80°C for 3 hours with occasional stirring. The product, PVA with carboxyl groups, was dialyzed against water. Next, 1.26 g of 1-ethyl-3-[3-(dimethylamino) propyl] carbodiimide (EDC) was added to the solution of spacer-introduced PVA, and ethylenediamine (61 μl) was added. The reaction was allowed to proceed for 12 hours at room temperature. The pH of the solution was maintained between 5.0 and 5.5 with 0.1 N HCl throughout the procedure. The resulting solution, including PVA with amino groups, was dialyzed against 0.5 M sodium carbonate–bicarbonate buffer (pH 9.5). The PVA with amino groups was incubated with 9.6 mg of RITC at room temperature for 3 hours. Finally, the product was dialyzed against water, freeze-dried, and stored at− 20°C. 

Cell Culture

Human umbilical vein endothelial cells (HUVECs) were purchased from Kurabo (Okayama, Japan). HUVECs were grown as monolayer cultures in HuMedia EG–2 (Kurabo) containing 1% fetal bovine serum. 

Bovine retinal pigment epithelial cells (BRPECs) were obtained from bovine eyes. Bovine eyes were washed in calcium- and magnesium-free phosphate-buffered saline solution containing penicillin G potassium (1000 IU/ml), streptomycin (1 mg/ml), and amphotericin B (2.5 mg/l). The BRPECs were removed from the choroid gently with a pipette after 10 minutes of treatment with 0.1% trypsin and 0.02% EDTA. Freed BRPECs were recovered by centrifugation at 1000 rpm for 5 minutes and then resuspended in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, penicillin G (100 IU/ml), streptomycin (0.1 mg/ml), and amphotericin B (0.25 mg/l). The cells of passages 3 to 5 were used for the experiments. 

Growth Inhibition Assay

HUVECs and BRPECs were maintained in 10-cm cell culture dishes. For the cell inhibition assay, a tetrazolium-based colorimetric assay (XTT assay) method was used to determine cell numbers. ²⁴ HUVECs and BRPECs were plated onto 96-well cell culture plates (Iwaki Glass, Tokyo, Japan), and TNP-470–PVA (1 ng/ml to 600 μg/ml), free TNP-470 (1.0 × 10⁻³ pg/ml to 30 μg/ml), or free PVA (60 μg/ml) with different concentrations was added to the cultures on the next day. At the end of culture, 50 μl of XTT solution was added to the culture. After the additional 4 hours’ incubation, the absorbance at 450 nm was determined by spectrophotometry (model DU–64; Beckman Instruments, Tokyo, Japan). Each experiment was done in quadruplicate and repeated three times. 

Animals and Anesthesia

Pigmented rabbits, weighing 1.8 to 2.6 kg each, were used in this study. The animals were treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The right eye of rabbits was used. The rabbits were anesthetized with intramuscular ketamine (5 mg/kg) and xylazine (2 mg/kg). Topical 1% tropicamide and 2.5% phenylephrine hydrochloride were instilled for mydriasis during surgery and fluorescein angiography and to observe the fundus. The rabbits were killed with an overdose of intravenous sodium pentobarbital. 

Induction of Choroidal Neovascularization in Rabbits

Experimental choroidal neovascularization (CNV) in rabbits was induced as previously reported by us. ²⁵ Briefly, 5 mg of cross-linked gelatin microspheres, 90 μm or less in diameter in wet form, was added to 100 μl of phosphate-buffered saline (PBS; pH 7.5) containing 100 μg of bFGF. The mixture was allowed to keep at room temperature for 1 hour. The resultant product, bFGF-loaded gelatin microsphere suspension, was diluted with an additional 400 μl of PBS. 

A 1-mm sclerotomy was made 2 mm from the limbus on the right eye of anesthetized rabbits. The gelatin microsphere suspension (50 μl) was injected into the subretinal space via the neurosensory retina adjacent to the disc between the medullary wings using a micropipette, with a 100-μm internal diameter, inserted through the sclerotomy site. 

A total of 36 (71%) of 51 eyes that received the microspheres developed CNV lesion with mild or moderate fluorescein leakage 2 weeks after induction. Because the fluorescein leakage in such eyes persists for the next 2 to 4 weeks, ²⁵ we used these eyes in the following in vivo studies. 

Accumulation of RITC–PVA in CNV Lesions

To investigate the accumulation of TNP-470–PVA in the CNV lesions, we administered PVA binding RITC instead of TNP-470 into rabbits with CNV. RITC–PVA (n = 3) or free RITC counteracted with glycine (n = 3) was injected intravenously. Twenty-four hours later, the eyes were enucleated after the rabbits were killed. The eyecups were observed and photographed with a fluorescent microscope (model BH2-RFK; Olympus, Tokyo, Japan). 

Efficacy of TNP-470–PVA on Experimental CNV

To determine the anti-angiogenic efficacy of TNP-470–PVA, rabbits were treated with saline (n = 12), free PVA (n = 5), free TNP-470 (n = 5), and TNP-470–PVA (n = 8). Free PVA, free TNP-470, and TNP-470–PVA were administered intravenously at doses of 30, 3, and 30 mg/kg, respectively, for 3 consecutive days, 2 weeks after implantation of gelatin microspheres. At 3, 4, 5, and 6 weeks after the induction of CNV, fluorescein angiography was repeated. To evaluate the effects of treatment, all angiographic negatives taken 7 to 9 minutes after injection to document the degree of late leakage of fluorescein were converted to digital images, and the area of fluorescein leakage was quantified in a masked fashion using computer software, NIH image (Research Service Branch). To prevent the introduction of variability during image processing, the density of choroidal background on the images from the same eye was harmonized without changing the contrast. The change of the area with fluorescein leakage reflecting the effect of used compounds was calculated by the following formula: (area of fluorescein leakage at 4 weeks after induction of CNV)/(area of fluorescein leakage at week 2, the initiation of the treatment)×100 (%). 

Histologic Studies

The eyes treated with saline (n = 2) and TNP-470–PVA (n = 2) were subjected to histologic examination. These rabbits were killed 6 weeks after implantation of the microspheres. The eye was enucleated and immediately placed in 2.5% glutaraldehyde and 2% formaldehyde in 0.1 M phosphate-buffered saline (pH 7.4) for 15 minutes. The cornea, lens, and vitreous were carefully removed from the eye and placed in the fixative for an additional 24 hours at 4°C. From each eye, the area where the hydrogel had been implanted was resected under the dissecting microscope. Each specimen was embedded in paraffin. Sections of 2 to 3μ m were made and stained with hematoxylin-eosin for light microscopy. 

Statistical Analysis

All values are mean ± SEM. Student’s t-test was used to analyze the in vitro data. ANOVA was used to compare the increased rates of fluorescein leakage in the different groups that received TNP-470–PVA or others, with post hoc comparisons tested using Scheffé’s procedure. Differences were considered statistically significant when the probability values were less than 0.05. 

Results

In Vitro Effect of TNP-470–PVA

As shown in Figure 1 , TNP-470–PVA inhibited the growth of HUVECs in a biphasic manner, similar to that of free TNP-470 (Fig. 2) . In the first phase, inhibition of cell growth was not associated with reduction in the cell number below the initial plating number. The inhibition of HUVEC growth by TNP-470–PVA in the first phase occurred in a wide range of concentrations (complete inhibition of 12 to 60μ g/ml with EC₅₀ of 0.6 μg/ml), showing a plateau in the dose–response curve. In the second phase, growth inhibition was observed at concentrations higher than 60 μg/ml. The inhibition in the second phase resulted in a reduction in the cell number below the initial plating number. TNP-470–PVA exhibited cytotoxic effects on HUVECs at the concentrations in the second phase but not in the first phase. 

On the other hand, the BRPECs were less sensitive to TNP-470–PVA and free TNP-470 than were the HUVECs. Although TNP-470 inhibited the growth of BRPECs in a biphasic manner, the effect of TNP-470 was not enough to exhibit complete inhibition at relatively low doses at which the growth of HUVECs was inhibited completely. TNP-470–PVA inhibited the growth of BRPECs partially (EC₅₀ of 60μ g/ml) at the concentrations tested in this study. 

Free PVA never inhibited the growth of HUVECs or BRPECs at the concentration of 60 μg/ml. 

Accumulation of RITC–PVA in CNV Lesions

RITC–PVA was still in the CNV lesions 24 hours after intravenous administration, whereas most of the free RITC was washed out from these tissues (Fig. 3) . 

Efficacy of TNP-470–PVA on Experimental CNV

The area of fluorescein leakage from the CNV lesion was decreased remarkably in the eyes treated with TNP-470–PVA, whereas leakage from the CNV was unchanged or increased in area in the control groups (Fig. 4) . The CNV lesions that showed fluorescein leakage in the eyes treated with TNP-470–PVA were gradually covered with hyperpigmented tissues, resulting in decreased fluorescein leakage. 

The change of the area with fluorescein leakage was 33.1% ± 8.5% (mean ± SEM), 88.6% ± 6.4%, 103.3% ± 7.1%, and 102.5% ± 3.5%, respectively, in the eyes treated with TNP-470–PVA, free TNP-470, free PVA, and saline (Fig. 5) . TNP-470–PVA showed a statistically significant decrease of fluorescein leakage (P = 0.001). The area with fluorescein leakage at 6 weeks was similar to that at 4 weeks. 

Histologic Studies

Six weeks after implantation of the microspheres, fibrovascular membranes (which were mainly composed of RPE-like cells, an extracellular matrix, and new vessels) were observed beneath the retina in the control eyes (Fig. 6) . Numerous new vessels extended from the choriocapillaris. A few microspheres were still present beneath the retina. In contrast, the subretinal membrane of the eyes treated with TNP-470–PVA consisted mainly of a large number of RPE-like cells that showed pseudoacinar structures (Fig. 7) . A few new vessels, some of which were surrounded by RPE-like cells, were observed in the membrane. Moderate outer retinal degeneration and mild choroidal atrophy were seen only in the areas where microspheres were implanted, not only in the treated eyes but also in the control eyes. 

Discussion

AMD often causes loss of vision. Such cases exhibit the growth of new blood vessels as a common pathologic feature. Therapies such as photocoagulation, vitreous surgery, and radiation are not always effective to repair visual function. Therefore, we have aimed to develop an anti-angiogenic substance that would be available for the treatment of CNV in the early stage. 

It was first reported in 1990 that the anti-angiogenic action of fumagillin, a natural product of Aspergillus fumigatus, and its potent analogue TNP-470 inhibited tumor growth in vivo. ²⁶ Since then other studies have demonstrated its effect on the growth of several kinds of cells and the mechanism of its anti-angiogenic and anti-tumor actions. ^{14 15 27 28 29 30 31 32 33 34 35} TNP-470 was found to relatively selectively inhibit the capillary-like tube formation and the proliferation of endothelial cells in vitro and tumor growth and tumor metastasis in vivo. ^{26 27}  

Kusaka et al. ¹⁴ have demonstrated that TNP-470 exerted its specific anti-angiogenic action by inhibiting cytostatically the growth of endothelial cells in a relatively specific manner. Furthermore, the cytostatic inhibition by TNP-470 is durable after washing out TNP-470 in culture. ¹⁴ For example, if endothelial cells were cultured for 2 hours with 100 ng/ml of TNP-470, the inhibition of endothelial cell growth was sustained for at least 6 days. ¹⁴ These characteristics of TNP-470 are beneficial in clinical use, which suggests that continuous exposure of TNP-470 at a low dose could inhibit cell growth with little toxicity. In fact, TNP-470 is effective against tumor growth and metastasis with daily or intermittent administration in vivo. ¹⁵ On the basis of the promising results obtained with TNP-470 in the in vitro and in vivo studies, this anti-angiogenic agent has already entered clinical trials for a variety of solid tumors. However, side effects of TNP-470, such as granulocytopenia and a modest fatigue in patients treated with high doses of TNP-470, also have been reported. ¹⁵ These side effects may restrict the clinical use of this agent in diseases such as AMD, which itself is not fatal, especially when it is used to arrest the development of a disease in the incipient stage. 

Drug targeting may be one possible way to overcome this limitation. Chemical conjugation of drugs with water-soluble polymers can modify the pattern of drug distribution in the body, resulting in not only an increase in therapeutic efficacy but also a decrease in side effects. ^{17 18 19 20 21 22 23} Yamaoka et al. ¹⁸ demonstrated that PVA had few significant interactions with cell components, such as macrophages and blood cells, and that the half-life of intravenously injected PVA in the blood was mainly determined by the permeation characteristics of the kidney. Therefore, we chose PVA as a water-soluble polymer in the present study. The conjugate of TNP-470 with PVA has several advantages over free TNP-470. First, TNP-470–PVA is supposed to prolong the half-life of TNP-470 in the blood. In this study, the conjugates remained in the subretinal space 24 hours after intravenous injection. It was reported that PVA, with an average molecular weight of approximately 220,000, has a half-life of approximately 18 hours. ¹⁸ Second, large substances tend to accumulate and prolong their retention in tissues with hyperpermeability of vasculature and immaturity of lymphatic system, such as a tumor tissue, to a greater extent than normal tissue. ³⁶ On the basis of this anatomic feature, it has been demonstrated that passive targeting of anti-tumor drugs to a tumor site can be achieved by increasing their apparent molecular size. ²³ We believe that choroidal neovascular membranes have anatomic characteristics similar to tumor tissue because the retinal tissue has only prelymph system ³⁷ and that they are revealed as the staining of fluorescein in a late phase of fluorescein angiography. Third, TNP-470–PVA acts as a reservoir of TNP-470, supplying active TNP-470 to the body. These properties make the effect of passive targeting more likely, possibly allowing the clinical use of this agent to be successful. 

TNP-470–PVA inhibited the growth of HUVECs in a biphasic manner similar to that of TNP-470. On the other hand, the BRPECs exhibited less sensitivity to TNP-470–PVA than did the HUVECs. These findings suggest that TNP-470–PVA preserves the original bioactivity of TNP-470 and that, if this relationship between the two types of cells corresponds to that between choroidal endothelial cells and RPE cells, this conjugate may inhibit the growth of endothelial cells and produce less interference in the proliferation of RPE cells. 

The efficacy of TNP-470–PVA was also evaluated in vivo, where it was shown to inhibit the progression of CNV in rabbits. The dose of free TNP-470 used in this study was not effective in the treatment of CNV. In fact, greater doses of TNP-470 have been used in the treatment of cancer. ¹⁵ Although the dose of TNP-470 included in TNP-470–PVA was equal to or less than that of free TNP-470 used in this study, the conjugates inhibited CNV growth, suggesting that drug targeting made these results successful. In rabbits that underwent intravenous administration of TNP-470–PVA, CNV seemed to be promptly covered with RPE cells. This finding and the in vitro results suggest that TNP-470–PVA does not interfere with RPE cell proliferation, which may inhibit the progression of CNV. However, because the proliferation of human RPE cells may be much slower compared with that in rabbits, the dosage and frequency in the use of TNP-470–PVA will need to be modified to arrest human CNV clinically. 

In conclusion, the conjugate of the angiogenic inhibitor with the water-soluble polymer TNP-470–PVA may enable TNP-470 to be used safely because of prolonged circulation time, passive targeting, and slow release of free TNP-470. Although the influence of PVA on the body remains to be investigated clinically, TNP-470–PVA may be a promising tool in the treatment of diseases that involve angiogenesis such as AMD. 

 Supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan, 1998.

 Submitted for publication December 29, 1998; revised April 30, 1999; accepted June 7, 1999.

 Commercial relationships policy: N.

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1998.

 

Figure 1.

 

Inhibition of HUVEC (circles) and BRPEC (squares) growth by TNP-470–PVA. The arrow indicates averaged initial cell number. Values are mean ± SEM. Note that the HUVECs exhibited more sensitivity to TNP-470–PVA than did the BRPECs. *P < 0.01 compared with the control. §P < 0.01 compared with values of BRPECs.

Figure 2.

 

Inhibition of HUVEC (circles) and BRPEC (squares) growth by free TNP-470. The arrow indicates averaged initial cell number. Values are mean ± SEM. Both kinds of cells exhibited sensitivity to free TNP-470 in a biphasic manner. The HUVECs were more sensitive to this agent than were the BRPECs. *P < 0.01 compared with the control. §P < 0.01 compared with values of BRPECs.

Figure 3.

 

The accumulation of RITC–PVA (B) and free RITC (D) in the subretinal space. RITC was strongly detected by fluorescence microscopy in an eye with RITC–PVA injected (B). The area stained with RITC corresponded with CNV revealed as the staining of fluorescein in a late phase of fluorescein angiography (A). Most of free RITC was not retained in the CNV lesion (D), whereas the fluorescein leakage was intense (C). (Figs. B and D correspond with the fields defined by lines in Figs. A and C, respectively.)

Figure 4.

 

Fluorescein angiograms from a rabbit treated with saline as a control (A, B) and two rabbits treated with TNP-470–PVA (C, D for one eye; E, F for another eye). Late-phase angiograms 2 weeks after implantation of microspheres all show actively leaking fluorescein (A, C, E). Late-phase angiograms 4 weeks after implantation suggest that TNP-470–PVA decreased the area of fluorescein leakage, compared with that area before treatment (arrows; D, F). The area is unchanged or shows increased leakage in a control group (B).

Figure 5.

 

The effect of TNP-470–PVA on the area of fluorescein leakage in the lesion with CNV. Values are mean ± SEM. The conjugates inhibited the growth of CNV significantly. *P < 0.01 compared with control eyes.

Figure 6.

 

Light micrograph of a control eye, 6 weeks after implantation of the microspheres. A massive subretinal neovascular membrane is observed in a subretinal space. This neovascular membrane consists of RPE-like cells, an extracellular matrix, and vascular formation. A few microspheres are still present beneath the retina. Hematoxylin–eosin; original magnification, ×80.

Figure 7.

 

Light micrographs of an eye treated with TNP-470–PVA, 6 weeks after implantation of the microspheres (4 weeks after treatment). The subretinal membrane consists mainly of a large number of RPE-like cells with pseudoacinar structures. Few new vessels can be seen in the membrane. Hematoxylin—eosin; original magnification, ×80.

The authors thank Hisako Okuda, who processed the histologic specimens.