Monoclonal Antibody-Mediated Drug Targeting to Choroidal Neovascularization in the Rat

Abstract

purpose. Active drug targeting mediated by monoclonal antibodies (mAbs) of vascular endothelial cells in tumors is a new concept in cancer therapy. Integrin αvβ3 has been reported to be strongly expressed in vascular endothelial cells of surgically excised choroidal neovascular membranes and is thought to be a potential antigen for mAb-mediated drug targeting of choroidal neovascularization (CNV). The objective of this study was to evaluate the efficacy of drug targeting mediated by anti-integrin αvβ3 mAbs in a laser-induced CNV rat model.

methods. The mitomycin C (MMC)-dextran (MMCD) conjugate was synthesized with a carbodiimide-catalyzed reaction. The mAb was conjugated with MMCD (MMCD-mAb). To evaluate the feasibility of mAb-mediated drug targeting in vitro, we investigated the effect of the immunoconjugates involving dextran-binding MMC on the proliferation of human umbilical vein endothelial cells (HUVECs). CNV was induced by laser photocoagulation in male Brown Norway rats. Immunolocalization of integrin αvβ3 in CNV lesions was assessed immunohistochemically with the anti-von Willebrand factor antibody as an endothelial cell marker. Intravenous administration of saline (n = 7), 1 mg/day mAb (n= 7), 100 μg/kg per day free MMC (n = 7), MMCD with irrelevant Ab (n = 7), unconjugated MMCD with unconjugated mAb (MMCD+mAb; n = 7), or MMCD with mAb (MMCD-mAb; n = 8) containing an equal amount of free MMC, was performed daily for 3 days from day 14 after CNV induction. CNV was assessed by fluorescein angiography 2 weeks after treatment. Fluorescein leakage was scored on a four-grade scale. The animals were killed 2 weeks after treatment, and the lesions were evaluated histologically.

results. The inhibition of immunoconjugates on the proliferation of HUVECs was enhanced specifically by the mediatory effect of the mAb. Endothelial cells demonstrated strong immunoreactivity of integrin αvβ3 in the CNV. In the vehicle-treated group, fluorescein leakage equal to that before treatment was observed 2 weeks after treatment, with an average score of 2.00 ± 0.17 (mean ± SEM). MMCD-mAb significantly inhibited the development of CNV in rats (P < 0.01). Moreover, the thickness of the lesions was significantly reduced in the MMCD-mAb–treated group (P < 0.01).

conclusions. Immunoconjugates effectively inhibited progression of CNV in this model. The results suggest that mAb-mediated drug targeting may be beneficial in the treatment of CNV.

Angiogenesis plays an important role in pathologic conditions, such as tumor growth, metastasis, and inflammation as well as in physiological development and processes such as wound repair. In the eye, choroidal neovascularization (CNV) is a major complication of age-related macular degeneration (AMD), which causes severe visual loss among the elderly in developed countries. ^{1 2 3 4 5} Although submacular surgery to remove CNV has been advocated, the efficacy of this therapy has not yet been established. ^{6 7 8 9} Laser treatment of CNV is the standard therapy for AMD at present. However, most patients are not eligible for laser photocoagulation, according to the current guidelines, and the outcome of laser photocoagulation is not always satisfactory. The Macular Photocoagulation Study Group reported that a large percentage of patients who have undergone photocoagulation have reduced visual acuity and recurrence of neovascularization. ^{10 11 12 13 14 15} The effects of several antiangiogenic agents such as interferon-α, thalidomide, and TNP-470 have been described. ^{16 17 18 19 20} However, systemic administration of these drugs has been shown to produce no real benefit for patients with AMD, because these drugs do not have organ-specific affinity, and their in vivo half-lives are too short. ²¹ To reduce the serious side effects that the systemic administration of these agents at the dosages needed to achieve their desired therapeutic effect may bring about, a number of drug delivery systems have been developed. Of these approaches, the one using antibodies against the targeted tissues appears to have the greatest theoretical promise for the treatment of CNV. 

Newly formed vessels in tumor tissue have been clearly demonstrated to exhibit high substance permeability compared with that in normal tissues. In addition, lymph systems in tumor tissue are incomplete. These anatomic features enable macromolecules to accumulate and prolong their retention in the perivascular regions of solid tumors to a greater extent than in normal tissues, which is known as the enhanced permeability and retention (EPR) effect. ^{22 23 24 25} CNV membranes have anatomic characteristics similar to tumor tissue because the retinal tissue surrounding the CNV contains only prelymph systems, ²⁶ and CNV membranes have been identified during late-phase fluorescein angiography. In this connection, we have demonstrated that passive targeting of the anti-angiogenic agent TNP-470 against experimental CNV through chemical conjugation with water-soluble polymer may be beneficial in the treatment of CNV. ²⁷  

Although passive targeting has been effective in the treatment of experimental CNV, monoclonal antibodies (mAbs) that recognize vascular endothelial cells may enhance the targetability of drug conjugates, thus making them more effective for clinical use. Integrin αvβ3 is the endothelial receptor for von Willebrand factor, fibrinogen, and fibronectin, and has been found to be selectively expressed on blood vessels in human wound granulation tissue but not in normal skin. ²⁸ More important, antagonists of this integrin block cytokine or tumor-induced angiogenesis on the chick chorioallantoic membrane. ^{28 29 30} Finally, no detectable signal for integrinα vβ3 was observed in normal human retina, but neovascular tissue removed from the surface of the retina of patients with diabetic retinopathy showed intense staining for this integrin ³¹ and it was observed on blood vessels in ocular tissues with active neovascularization from patients with AMD or presumed ocular angiogenesis. ³² These findings suggest that integrinα vβ3 may be a useful therapeutic target for ocular disease characterized by neovascularization such as AMD. 

In this study, we developed an active drug-targeting system against neovascular vessels, using an antibody directed against vascular endothelial cells. We evaluated the efficacy of an anti-integrinα vβ3 mAb as a mediator for drug targeting by using the conjugates of mAb and dextran-binding mitomycin C (MMC) for the treatment of laser-induced experimental CNV. 

Materials and Methods

Chemicals

Dextran (molecular weight, 66,900) was purchased from Sigma Chemical (St. Louis, MO). MMC, dithiothreitol (DDT), and heterobifunctional reagent N-succinimidyl 3-(2-pyridyldithio)-propionate (SPDP) were obtained from Wako Pure Chemicals Industries (Osaka, Japan). All other chemicals were reagent-grade products obtained commercially. 

Cell Cultures

Human umbilical vein endothelial cells (HUVECs) were purchased from Kurabo (Okayama, Japan) and grown as monolayer cultures in media (HuMedia EG-2; Kurabo) containing 1% fetal bovine serum (FBS). The hybridomas were suspended in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS, penicillin G (100 IU/ml), streptomycin (0.1 mg/ml), and amphotericin B (0.25 mg/ml). The cells of passages 3 thought 5 were used for the experiments. 

Antibody

Anti-human integrin αvβ3 mouse mAb was purified from a hybridoma culture that was purchased from American Type Culture Collection (Manassas, VA) by affinity chromatography on immobilized protein G (MAb Trap GII; Pharmacia Biotech, Uppsala, Sweden). Polyclonal rabbit antibody against human von Willebrand factor and mAb UCD/PR 10.11 against types 8 and 18 cytokeratin were purchased from Dako (Amsterdam, The Netherlands). Conjugated secondary antibodies for FITC and tetrarhodamine isothiocyanate (TRITC) were obtained from Dako A/S (Glostrup, Denmark). 

Synthesis of Immunoconjugate Composed of mAb and MMC-Dextran Conjugate

For this study, we selected MMC with intense cytotoxicity to facilitate evaluation of the effect of the antibody as a mediator for drug targeting. MMC-dextran (MMCD), a polymeric prodrug of MMC, was synthesized by a carbodiimide-catalyzed reaction, and the amino groups were then introduced to the MMCD. The coupling to mAb (anti-integrinα vβ3) was performed by using the heterobifunctional reagent SPDP. The coupling procedure consisted of the following steps: (1) MMC was conjugated to dextran with a carboxyl group by a carbodiimide-catalyzed reaction. MMCD with the amino groups was synthesized by reacting ethylenediamine with the remaining free carboxyl group of dextran. Briefly, 1 g dextran was dissolved in a 4-N NaOH solution (10 ml), and 6-bromohexanoic acid (4 g). The mixture was kept at 80°C for 3 hours with occasional stirring. The product, a spacer-introduced dextran, was washed with water and concentrated by ultrafiltration. Fifty milligrams 1-ethyl-3-[3-(dimethylamino) propyl] carbodiimide was added to the solution of the spacer-introduced dextran, and 7.5 mg MMC was dissolved drop by drop into the solution. Thirty minutes later, a 15-M excess of ethylenediamine over dextran (30 μl, 0.35 M) was added stepwise, and the reaction was allowed to proceed for 12 hours at room temperature in the dark. The pH of the solution was maintained at between 5.0 and 5.5 with 0.1 N HCl throughout the procedure. The product was then washed with 0.1 M phosphate-buffered solution (PBS; pH 7.5) and concentrated by ultrafiltration. (2) The MMCD with the amino groups (50 mg) was incubated first with a 10-fold molar excess of SPDP in 3 ml PBS at room temperature for 30 minutes. The reaction was terminated by the addition of 0.5 ml of 0.1 M Tris-HCl buffer (pH 8.0), and the product was washed with PBS and concentrated by ultrafiltration. The 3-(2-pyridyldithio) propionated (PDP)-MMCD was reduced with 10 mM DDT in a 0.1 M acetate buffer (pH 4.5) at room temperature for 20 minutes. The resultant MMCD with thiol groups (HS-MMCD) was washed with 0.1 M acetate buffer (pH 4.5) and concentrated by ultrafiltration. In another process, mAb (anti-integrinα vβ3; 10 mg) was also incubated with a 10-fold molar excess of SPDP in 1 ml PBS at room temperature for 30 minutes. The PDP-mAb was passed immediately through through a Sephadex G-25 (Pharmacia Biotech). (3) Finally, the PDP-mAb was mixed with its mole equivalent of HS-MMCD, the pH was adjusted to 7.5, and the mixture was allowed to stand at room temperature overnight in the dark. The mixture was then applied to an extraction column (Toyopearl HW-65S, 2.4 × 65 cm; Toyo Soda Kogyo, Tokyo, Japan), equilibrated with PBS (pH 7.0), and eluted with the same buffer. The peak fractions were pooled, concentrated, frozen, and stored at −20°C in the dark. The chemical structure of MMCD-mAb is shown in Figure 1 . mAb UCD/PR 10.11 (an irrelevant Ab) was conjugated to MMCD by the same method (MMCD-irrelevant Ab). 

Synthesis of Immunoconjugate Composed of mAb and FITC-Dextran

Dextran with amino groups was synthesized by the same method. One milligram dextran with the amino groups was incubated in a 0.5 M carbonate-bicarbonate buffer solution (pH 9.5) with 8.4 mg FITC at room temperature for 3 hours. The product, FITC-binding dextran (FITCD), was dialyzed against water, and mAb was conjugated to FITCD through the method just described (FITCD-mAb). 

In Vitro Efficacy of Immunoconjugates

HUVECs were maintained in 10-cm cell culture dishes. For the cell inhibition assay, the tetrazolium-based colorimetric assay (XTT assay) was used. ³³ HUVECs (5 × 10³ cells/well) were seeded into 96-well cell culture plates and incubated overnight. The cells in each well were then exposed to various concentrations of immunoconjugates (MMCD-mAb) for 1 hour at 37°C, washed twice with the fresh growth medium, and incubated for 48 hours at 37°C. For control samples, free MMC, MMCD-irrelevant Ab, unconjugated MMCD with unconjugated mAb (MMCD+mAb), or nothing was added to the plates. At the end of the culture, 50 μl XTT solution was added to the culture. After an additional 4 hours of incubation, the absorbance at 450 nm was determined by spectrophotometry (Model DU-54; Beckman Instruments, Fullerton, CA). Each experiment was performed in quadruplicate and repeated three times. 

Animals and Anesthesia

Male Brown Norway (BN) rats, weighing 180 to 200 g, were used for this study. The animals were treated in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The right eye of the rats was used. The rats were anesthetized with intramuscular ketamine (10 mg/kg) and xylazine (4 mg/kg). Topical 1% tropicamide and 2.5% phenylephrine hydrochloride were instilled for mydriasis during laser photocoagulation and fluorescein angiography. The rats were killed with an overdose of intravenous pentobarbital sodium. 

Induction of Choroidal Neovascularization in Rats

Dye laser irradiation (545 nm; Argon Dye Laser model 920; Coherent Medical Laser, Palo Alto, CA) was delivered through a slit lamp (Carl Zeiss, Oberkochen, Germany) with a hand-held 90-diopter lens (Nikon, Tokyo, Japan). A contact lens was used to retain corneal clarity through the photocoagulation. The laser spots were placed separately using a setting of 50-μm diameter, 0.1-second duration, and 150-mW intensity. Three laser burns were between major retinal veins in the right eye, as described previously. ³⁴  

Evaluation of Time Course of Experimental CNV

Four BN rats were used to establish the natural time course of dye laser-induced CNV. On day 0, laser photocoagulation was performed as described. The coagulated lesions in four rats were assessed on day 4 and at 1, 2, 3, 4, 5, 6, 8, 12, and 20 weeks by ophthalmoscopy, fundus photography, and fluorescein angiography. For fluorescein angiography, 0.2 ml of 10% sodium fluorescein was injected into the tail vein of anesthetized rats, and fluorescein angiograms were obtained 100 to 140 seconds after dye injection (TRC 50IA fundus camera; Topcon, Tokyo, Japan). 

Immunohistochemistry of CNV

Expression of integrin αvβ3 in neovascular vessels in CNV lesions was evaluated by immunohistochemical staining. Two rats were killed with an overdose of pentobarbital 2 weeks after photocoagulation. The enucleated eyes were immediately placed in 4% paraformaldehyde in PBS for 30 minutes. The cornea, lens, and vitreous were carefully removed from the eye and fixed overnight in 4% paraformaldehyde in PBS (pH 7.4) at 4°C. After fixation, the eyes were first washed in PBS, and then in PBS with increasing concentrations of sucrose. Eye cups were embedded in optimal cutting temperature (OCT) compound (Tissue Tek; Sakura Finetek USA, Torrance, CA) and snap frozen with liquid nitrogen. Cryostat sections of 2-μm thickness were cut at −20°C, air dried, and rehydrated with PBS. The sections then were fixed in acetone for 10 minutes at room temperature, washed with PBS, and incubated for 15 minutes with a blocking serum. The specimens were incubated for 30 minutes with the primary antibodies: anti-integrin αvβ3 (10 μg/ml) and polyclonal rabbit antibody anti-human von Willebrand factor (N-Series; Dako). They were then washed for 15 minutes with PBS. After blocking of endogenous peroxidase by 0.6% hydrogen peroxide, immunoperoxidase detection was performed with a kit (Pathostain ABC-POD [M] Kit; Wako Pure Chemicals Industries) with aminoethylcarbizole. Immunostaining for the von Willebrand factor was performed in the same manner. Finally, the slides were rinsed with tap water and placed in a glycerin-gelatin mounting medium. 

Double-immunofluorescence staining was performed by first incubating the tissue sections for 1 hour with mouse anti-human integrin αvβ3 mAb and rabbit anti-human von Willebrand factor polyclonal antibody, followed by a second incubation for 30 minutes with the corresponding fluorescent dye–conjugated secondary antibodies. The specimens were then studied with a confocal microscope (LSM 410; Carl Zeiss). 

Accumulation of FITCD-mAb in CNV Lesions

To examine the accumulation of the conjugate of FITCD-mAb in the CNV lesions, we administered the conjugate to BN rats with induced CNV 2 weeks after photocoagulation. Ten milligrams of the conjugate (n = 4) or free FITC (n = 4) with the same fluorescence intensity, counteracted with glycine, was injected intravenously. Fluorescein angiograms were taken with a fundus camera 100 to 140 seconds after injection of each substance. Twenty-four hours later, the eyes were observed with a scanning laser ophthalmoscope (SLO; Rodenstock, Munich, Germany) and the images were recorded on S-VHS videotape. 

Efficacy of Immunoconjugates for Experimental CNV

To evaluate the efficacy of MMCD-mAb, intravenous administration was initiated 2 weeks after CNV induction. Rats were treated with saline (n = 7), 1 mg per day of anti-integrin αvβ3 mAb (n = 7), 100 μg/kg per day of free MMC (n = 7), MMCD-irrelevant Ab (n = 7), MMCD+mAb (n = 7), or MMCD-mAb (n = 8) containing equal amounts of free MMC, for 3 consecutive days. The optimal dose of used MMC was determined, based on results of a preliminary study. The lesions were examined by fluorescein angiography 2 weeks after treatment. Further assessment was not undertaken, because the time course of experimental CNV indicated that fluorescein leakage begins to decrease 5 weeks after photocoagulation. The intensity of staining seen on fluorescein angiograms obtained 100 to 140 seconds after dye injection was graded by two examiners in a masked fashion using reference angiograms. ³⁵ The lesions were scored as follows: 0, no staining; 1, slightly stained; 2, moderately stained; and 3, strongly stained. When the two scores given for a particular lesion did not coincide, the higher score was used for the analysis. Such discrepant scoring occurred in only 8 of the 129 lesions analyzed, and the discrepancy was never by more than one grade. The Kruskal-Wallis test was used to analyze the in vivo data. A difference of P < 0.05 was considered to be statistically significant. Data are shown as means ± SEM. 

Histologic Studies

Seven rats treated with saline and seven with MMCD-mAb were histologically examined. Four weeks after photocoagulation, the rats were killed with an overdose of pentobarbital. Cryostat sections were cut as described above, air dried, and stained with hematoxylin-eosin for light microscopy. The thickness of lesions was measured by using sections from the central part of the lesion, which exhibited the thickest laser-induced choroidal neovascular membrane. Data obtained from the eyes of the MMCD-mAb–treated group and the control group were compared with the aid of the Wilcoxon test. A difference of P < 0.05 was considered to be statistically significant. Data are shown as means ± SEM. 

Results

Time Course of Changes in Staining Intensity Detected by Fluorescein Angiography after Photocoagulation 

The time course of changes in staining intensity detected by fluorescein angiography in nontreated rats was determined for up to 20 weeks after photocoagulation. The intensity of fluorescein staining gradually increased, reaching a peak at about 2 weeks. The peak continued for up to 4 weeks, after which the intensity began to decrease (Fig. 2) . These results coincide approximately with those recently reported elsewhere. ³⁵  

In Vitro Efficacy of Immunoconjugates

When the cytotoxic activity of MMCD-mAb against HUVECs was compared with that of free MMC, MMCD-irrelevant Ab, and MMCD+mAb, MMCD-mAb was found to be as cytotoxic as free MMC at concentrations of 20 μg/ml or more of MMC and was approximately seven times more cytotoxic than MMCD+mAb (50% inhibitory concentration [ IC₅₀]: 4 μg/ml for free MMC, 9 μg/ml for MMCD-mAb, 60 μg/ml for MMCD+mAb, and 90 μg/ml for MMCD-irrelevant Ab; Fig. 3 ). 

Immunohistochemistry of CNV

Immunoperoxidase staining for integrin αvβ3 showed almost the same distribution of positive cells as obtained with the von Willebrand factor (Figs. 4A 4B 4C 4D) . Double-immunofluorescence immunohistochemistry demonstrated extensive overlap between cells that were positive for integrin αvβ3 and those that were positive for von Willebrand factor (Fig. 4E) . Integrinα vβ3 was not expressed on the normal choroidal vessels (Fig. 4F) . 

Accumulation of FITCD-mAb in CNV Lesions

The area stained with FITC corresponding to CNV showed as fluorescein staining 100 to 140 seconds after administration of either the conjugate (Fig. 5A) or free FITC (Fig. 5B) . FITCD-mAb retained fluorescence in the area surrounding and adjacent to the CNV lesion 24 hours after intravenous administration (Fig. 5C) , whereas by this time most of the free FITC had been washed out of the CNV lesion (Fig. 5D) . 

Efficacy of Immunoconjugates on Experimental CNV

Fluorescein leakage from the CNV lesion was reduced in the eyes treated with MMCD-mAb (Fig. 6) , whereas it remained almost unchanged in the other groups. Two weeks after photocoagulation, the fluorescein leakage score was 2.05 ± 0.13 for the eyes treated with saline, 2.05 ± 0.2 for mAb, 1.95 ± 0.21 for free MMC, 2.00 ± 0.15 for MMCD-irrelevant Ab, 1.95 ± 0.16 for MMCD+mAb, and 2.00 ± 0.15 for MMCD-mAb. There were no significant differences in the scores of the groups before treatment (Fig. 7A) . Two weeks after treatment, the corresponding scores were 2.00 ± 0.17, 2.05 ± 0.2, 1.95 ± 0.21, 2.00 ± 0.15, 1.95 ± 0.16, and 1.25 ± 0.09. These results demonstrated that MMCD-mAb showed a statistically significant reduction in fluorescein staining (P < 0.01; Fig. 7B ). 

Histologic Studies

Four weeks after photocoagulation, neovascular membranes in the control eyes were situated in the subretinal space without intrusion into the retina and included numerous new vessels (Fig. 8A and arrows in 8C). In contrast, 2 weeks after treatment with MMCD-mAb, thin proliferative membranes with few new vessels were seen beneath the retina (Fig. 8B and arrows in 8D). The thickness of the lesions, which was evaluated morphometrically in serial sections, was 56.1 ± 4.5μ m for the control eyes and 33.0 ± 3.4 μm for the MMCD-mAb–treated eyes. Thus, the lesions in the eyes treated with MMCD-mAb showed a statistically significant reduction in thickness (P < 0.01; Fig. 9 ). 

Discussion

CNV is the main blinding factor in AMD. Photocoagulation, vitreous surgery, and radiation have been attempted for the treatment of CNV, but no satisfactory clinical therapy has been established that can preserve the visual function of patients with AMD. Systemic administration of anti-angiogenic agents, such as interferon-α and thalidomide, has not always sufficiently inhibited the development of CNV membranes, because of their side effects at the drug concentrations effective for the targeted tissues with neovascularization. However, selective delivery of these drugs to the targeted tissue could lead to the use of smaller doses for treatment and result in a reduction in undesirable side effects. 

mAbs have attracted a great deal of attention as mediators for targeted delivery of drugs. Many researchers have investigated the antibody-mediated delivery of cytotoxic agents in the field of cancer therapy. ^{36 37 38 39 40} Recently, some antigens expressed specifically in proliferating vascular endothelial cells have been considered as candidates for drug targeting. ^{39 40} Antibodies against these antigens may be useful as a mediator for drug targeting of CNV. Among them is integrin αvβ3, the endothelial cell receptor for von Willebrand factor, fibrinogen, and fibronectin. Integrin αvβ3 is minimally, if at all, expressed on resting, or normal, blood vessels, but is significantly upregulated on vascular cells within human tumors, ^{30 41 42} within granulation tissue, ^{41 42} or in response to certain growth factors in vitro ^{43 44} and in vivo. ^{41 42 45} Basic fibroblast growth factor and tumor necrosis factor-α were found to stimulate integrin αvβ3’s expression on developing blood vessels in the chick chorioallantoic membrane. ^{32 42 43} The highly restricted expression of integrin αvβ3 and the upregulation of its expression during angiogenesis suggest that it may play a critical role in the angiogenic process. Our immunohistochemical examination identified the expression of integrin αvβ3 on the endothelial cells in CNV, but not on normal choroidal vessels 2 weeks after photocoagulation. In this study, therefore, we used the anti-integrinα vβ3 antibody as a mediator of active targeting of CNV. 

The perfect antibody for mediating drug targeting should not only have high affinity for the targeted tissue but also should show no cross-reactivity with normal tissues. One of the major limitations of cancer chemotherapy is the indiscriminate toxicity of anticancer drugs toward both cancer cells and proliferating normal cells. To overcome this limitation, immunoconjugates were synthesized by using intermediate carriers of consisting water-soluble polymers, such as dextran, poly(ethylene glycol), and poly(vinyl alcohol). In our study, we used MMC to evaluate the effect of the antibody as a mediator of drug targeting, because MMC was expected to exhibit definite results because of its intense cytotoxicity. In cytotoxicity testing with a 1-hour exposure time, MMCD-mAb inhibited the growth of HUVECs as strongly as did free MMC. In contrast, both MMCD-irrelevant mAb and MMCD+mAb showed less cytotoxicity for HUVECs than did free MMC. Recent work in our laboratory has indicated that the inhibition by immunoconjugates of the proliferation of HUVECs was enhanced specifically by the mediatory effect of the anti-CD105 antibody, another type of antibody against neovascular vessels, as well. ⁴⁶ These findings suggest that MMCD-mAb preserves the bioactivity of MMC and becomes effective only after binding to the endothelial cell surface by antigen–antibody interaction, a prolonged accumulation, and release of MMC by chemical hydrolysis. This means that the injury to targeted cells is not directly caused by antigen–antibody interaction but is induced by drugs released during the prolonged retention in the targeted tissue. 

The pharmaceutical modification of drugs provides advantages over free drugs. Chemical conjugation of drugs with water-soluble polymers has been found to alter the pattern of drug distribution in the body, resulting in not only increased therapeutic efficacy but also diminished side effects. ^{36 47 48 49 50 51} Noguchi et al. ³⁶ suggested that dextran combined with a carboxyl group (molecular weight: 70,000) is most suitable for treatment, because this agent has a long circulating half-life and is taken up less by the reticuloendothelial system due to its polyanionic characteristics. MMCD enables the half-life of MMC in the blood (approximately 8.6 minutes) to be prolonged. MMCD with average molecular weights of 70,000 and 500,000 has corresponding serum half lives of approximately 4 and 8 hours. ⁴⁹  

Moreover, macromolecules themselves tend to accumulate and be retained for longer periods in tissues with hyperpermeable vasculature and immature lymph systems than in normal tissues (EPR effect). ^{22 23 24 25} In this connection, it has been demonstrated that passive targeting of anti-tumor agents to the tumor site can be achieved by increasing their molecular size. ⁵⁰ Tolentino et al. ⁵² have reported that dextrans with a molecular weight of more than 70,000 can pass through the leaky vessels of the CNV and accumulate in the subretinal space adjacent to CNV up to 2.5 weeks. Our study recognized the accumulation of the immunoconjugate composed of mAb and FITC-dextran in the CNV, which may result not only from the high affinity of the mAb with the newly formed vessels but also partly from the EPR effect. Moreover, because of the absence of lymphatics, the efflux of drugs released from the perivascular region in tumor tissue is supposed to be poorer than in tissues with other neovascular events, such as wound healing and inflammation. The combination of these properties could increase therapeutic efficacy and reduce adverse effects, thereby resulting in the successful clinical use of anti-angiogenic agents. Using the concept that CNV membranes appear to have anatomic features similar to those of tumor tissues, we have already demonstrated that this passive targeting system may be applied to AMD. ²⁷ The immunoconjugates produced by binding an antibody against neovascular vessels to these large substances are expected to further enhance targetability through prolonged retention. 

The efficacy of MMCD-mAb was also evaluated in vivo, where it was shown to inhibit the development of experimental CNV in rats. Friedlander et al. ³² reported that the systemic administration of the peptide antagonist of integrin αvβ3 inhibited retinal vasculogenesis in a mouse model. However, in our study, the administration of anti-integrin αvβ3 antibodies themselves could not attenuate the development of laser-induced CNV, and the efficacy of MMCD in vivo was not affected by the addition of unconjugated mAb. The dose of free MMC used in our study was supposed to be insufficient for passive targeting, based on both the prolonged plasma half-life and the EPR effect of high molecules, because neither MMCD-irrelevant Ab nor MMCD+mAb with an equal amount of MMC inhibited the progression of CNV. These findings suggest that the efficacy of MMCD-mAb may depend on the property of active targeting, based on its high affinity for the endothelial cells as well as its prolonged retention of the conjugate associated with the EPR effect in CNV. 

In this study, MMC was used to determine the efficacy of an antibody-mediated drug delivery, but other agents with less retinal toxicity should be used for the clinical treatment of human AMD. All anti-angiogenic agents with binding sites such as an amino group or a carboxyl group in parts not related to drug activity can be used for immunoconjugation. 

In conclusion, the conjugation of anti-angiogenic agents, bound to water-soluble polymer and mAbs that selectively target neovascular vessels, will enable these agents to be used more effectively and more safely because of prolonged circulation, active targeting, and slow release of the agents in active form. Although the retinal toxicity of MMC remains to be determined, mAb-mediated drug targeting may be beneficial for the treatment of ocular angiogenesis involving AMD. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May 2000.

 Supported in part by a grant from Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan, 1999.

 Submitted for publication March 9, 2001; revised June 1, 2001; accepted June 15, 2001.

 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 1.

 

Chemical structure of MMCD-mAb.

Figure 2.

 

Time course of changes in intensity of fluorescein staining detected on angiograms taken after dye laser photocoagulation in four nontreated rats. Each point represents the mean of the scores of 12 lesions. The score rapidly increased and reached its peak approximately 2 weeks after laser photocoagulation.

Figure 3.

 

Inhibition of HUVEC growth. Data are mean ± SEM.

Figure 4.

 

Immunolocalization of integrin αvβ3 (A, C) and the von Willebrand factor (B, D) in CNV 2 weeks after photocoagulation. The sections were stained with aminoethylcarbizole as the red chromogen (A, B). Double-immunofluorescence staining was performed in another membrane (C, D). The same field is displayed in a combined image of both fluorescence stainings (E), with immunofluorescence colocalization shown when stained for both integrin αvβ3 and the von Willebrand factor. (F) Integrin αvβ3 was not expressed on the normal choroidal vessels in the same section as is shown in (A). Original magnification, (A, B, F)× 400; (C, D, E) ×200.

Figure 5.

 

The accumulation of FITCD-mAb in the CNV lesion, at 100–140 seconds (A, B) and 24 hours (C, D) after the intravenous administration of FITCD-mAb (A, C) or free FITC (B, D). The eyes were observed with a fundus camera (A, B) and with a scanning laser ophthalmoscope (C, D). Both lesions in (A) and (B) were scored as 1.

Figure 6.

 

Fluorescein angiograms, obtained 100 to 140 seconds after dye injection, before (A) and 2 weeks after (B) treatment with MMCD-mAb. MMCD-mAb reduced the leakage from the CNV. The lesions in (A) and (B) were scored as 2 and 0, respectively.

Figure 7.

 

Fluorescein leakage scores 2 weeks after photocoagulation (A) and 2 weeks after treatment (B). There were no significant differences in the scores of the groups before treatment. MMCD-mAb significantly inhibited the fluorescein leakage from CNV (Kruskal-Wallis test; *P < 0.01).

Figure 8.

 

Histologic appearance of neovascular membranes 2 weeks after treatment with saline (A, C) or MMCD-mAb (B, D). (A) and (C), as well as (B) and (D), are the same sections at different magnifications. CNV membranes, both in the control eye (A) and in the treated eye (B), were above the normal RPE layer. The control eye showed numerous new vessels (arrows) in the subretinal space (C). In contrast, in the eye treated with MMCD-mAb, a thin proliferative membrane with few new vessels (arrowheads) was seen beneath the retina (D). The sections were stained with hematoxylin and eosin. Scale bar, 50 μm. Original magnification, (A, B) ×200; (C, D) ×400.

Figure 9.

 

Effect of MMCD-mAb on thickness of the CNV lesions 2 weeks after treatment. Each column shows the mean ± SEM of measurements taken from 15 lesions. The lesions in eyes treated with MMCD-mAb were significantly thinner than those in control eyes (Wilcoxon test;* P < 0.01).