Nanobiotechnol (2007) 3:107–115 DOI 10.1007/s12030-008-9001-5
Novel Drug Delivery System for Age-related Macular Degeneration Using Nanotechnology Yasuhiro Tamaki
Published online: 26 February 2008 # Humana Press Inc. 2008
Abstract Age-related macular degeneration (AMD) is a leading cause of legal blindness in developed countries. Even with the recent advent of several treatment options such as photodynamic therapy (PDT) and anti-vascular endothelial growth factor (VEGF) therapy for the treatment of exudative AMD, characterized by choroidal neovascularization (CNV), their efficacy is still limited. Thus, in this review article, we investigated novel drug delivery system for AMD using nanotechnology. Polyion complex (PIC) micelle has a size range of several tens of nanometers formed through electrostatic interaction, and accumulates in solid tumors through enhanced permeability and retention (EPR) effect. The distribution of the PIC micelle encapsulating fluorescein isothiocyanate-labeled poly-L-lysine (FITC-P(Lys)) in experimental CNV in rats was investigated. PIC micelle accumulates in the CNV lesions and is retained in the lesion for as long as 168 h after intravenous administration. PIC micelles can be used for achieving effective drug delivery system to CNV. Although PDT is a main treatment option for CNV, most patients require repeated treatments. For effective PDT against AMD, the selective delivery of photosensitizer to the CNV lesions and an effective photochemical reaction at the CNV site are necessary. The characteristic dendritic structure of the photosensitizer prevents aggregation of its core sensitizer, thereby inducing a highly effective photochemical reaction. A supramolecular nanomedical device, i.e., a novel
Y. Tamaki (*) Department of Ophthalmology, University of Tokyo School of Medicine, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan e-mail:
[email protected]
dendritic photosensitizer encapsulated by a polymeric micelle formulation was employed for an effective PDT for AMD. With its highly selective accumulation on CNV lesions, this treatment resulted in a remarkably efficacious CNV occlusion with minimal unfavorable phototoxicity. Our results will provide a basis for an effective approach to PDT for AMD. Keywords age-related macular degeneration . choroidal neovascularization . drug delivery system . nanotechnology . polyion complex micelle . photodynamic therapy . dendritic photosensitizer
Accumulation of Polyion Complex Micelle to Choroidal Neovascularization Exudative age-related macular degeneration (AMD) characterized by choroidal neovascularization (CNV) is a major cause of visual loss in developed countries [1, 2]. Although photocoagulation of the entire CNV is an effective treatment option for exudative AMD proved in large randomized control studies performed by the Macular Photocoagulation Study Group [3], permanent central visual loss is inevitable immediately after photocoagulation. Thus, alternative treatments for CNV with minimal damage to the healthy retina are being developed. Recently, photodynamic therapy (PDT) [4] and anti-vascular endothelial growth factor (VEGF) therapy [5, 6] are main choices for CNV treatments; however, their efficacy is still limited. To develop a pharmacological therapy for CNV with minimal systemic adverse effect, it is necessary to achieve a high local concentration of the drug [7]. Macromolecules can accumulate and prolong their retention in perivascular regions of solid tumors to a greater extent
108
Tamaki
Fig. 1 Preparation and schematic structure of PIC micelle encapsulating FITC-P(Lys)([20], with permission from Elsevier). PIC micelle consists of inner complex core formed by P(Asp) segment and FITC-P(Lys), and outer PEG shell
than in normal tissues because newly formed vessels in solid tumors exhibit high substance permeability compared with those in normal tissues, and the lymph systems in tumor tissue are incomplete [8, 9]. This effect is known as the enhanced permeability retention (EPR) effect [10]. CNV membranes bear high permeability and several studies have demonstrated that macromolecules accumulate in experimental CNV presumably through the EPR effect [7]. The size of the molecules is an important factor in exerting the EPR effect. Polymeric micelles have a size range of several tens of nanometers with a considerably narrow distribution, similar to that of viruses and lipoproteins [8]. Thus, they accumulate in solid tumors through the EPR effect [11, 12]. In addition, compared with the drug delivery system based on macromolecule conjugates, polymeric micelle can stably encapsulate chemical compounds with high efficiency [8]. Together with their high drug-loading capacity, polymeric micelles are expected to become a novel drug delivery system. We have recently developed a novel type of polymeric micelle formed through electrostatic interaction (polyion complex [PIC] micelle) [8, 13]. Unlike polyion complexes formed from an oppositely charged pair of simple homopolymers or statistical copolymers, PIC micelles from charged block copolymers are totally watersoluble and are narrowly distributed. To investigate whether PIC micelle can be used for treatment of CNV, the distribution of the PIC micelle in blood and in experimental CNV in rats were examined.
(PEG-P(Asp); PEG Mw =5,000 g/mol, polymerization degree of P(Asp) segment=78) were dissolved in 10.0 and 5.0 ml of phosphate-buffered saline (PBS), respectively (Fig. 1). PIC micelle solution was prepared by mixing the same volume (5.0 ml) of FITC-P(Lys) and PEG-P(Asp) solutions, in which a molar ratio of Lys and Asp residues was adjusted to unity. As a control, 5.0 ml of FITC-P(Lys) solution was diluted in 5.0 ml of PBS. Both PIC micelle and control solutions included the same concentration (5.0 mg/ml) of FITC-P(Lys). The average diameter and polydispersity index of PIC micelles were evaluated by dynamic light-scattering (DLS) measurement at 25°C, using a light-scattering spectrophotometer (DLS-7000, Otsuka Electronics, Osaka, Japan) with a vertically polarized incident beam at 632.8 nm supplied by a He/Ne laser. Fig. 2 shows the size distribution of PIC micelle obtained
Preparation of PIC Micelle Encapsulating Fluorescein Isothiocyanate-labeled Poly(L-lysine) One hundred milligrams of fluorescein isothiocyanatelabeled poly( L -lysine) (FITC-P(Lys); polymerization degree=105, FITC=0.004 mol/mol of Lys) and 48.6 mg of poly(ethylene glycol)-block-poly(α,β-aspartic acid)
Fig. 2 DLS histogram of PIC micelle encapsulating FITC-P(Lys) ([20], with permission from Elsevier). The average diameter and polydispersity index of PIC micelles were evaluated by dynamic light-scattering measurement at 25°C, using a light-scattering spectrophotometer. PIC micelle had narrowly dispersed with size around 50.7 nm
Novel Drug Delivery System for Age-related Macular Degeneration
109
Fig. 3 Accumulation of PIC micelle to CNV lesion ([20], with permission from Elsevier). The frozen sections of the CNV lesions were observed under a fluorescent microscope 1 (a), 4 (b), 24 (c) and 168 (d) hours after rats received PIC micelle incorporated FITC-P(Lys). Note that FITC-P(Lys) initially diffuses to the CNV lesion and choriocapillaris (a and b), and becomes confined to the CNV lesion thereafter (c and d). Bright fluorescence was observed in the CNV lesions up to 168 h (d). One hour after rats received intravenous injection of free FITC-P(Lys), FITC-P(Lys) distributes to the CNV lesion and choriocapillaris (e). Arrowheads (in a to e) indicate CNV lesions. Most rats died 1 h after free FITC-P(Lys) administration. In the laser-non-treated eyes, fluorescence was observed in the choriocapillaris 1 h after rats received PIC micelle (f) and the fluorescence became invisible 4 h after the injection of PIC micelle (g). Note that PIC micelle effectively accumulated to the CNV lesion throughout the studied period. Scale bar: 25 μm
from histogram analysis of DLS measurements. It was clear that the prepared PIC micelle had a unimodal size distribution. Also the average diameter and polydispersity index of PIC micelle were determined to be 50.7 nm and 0.046 by using cumulant approach of DLS measurement, indicating the formation of PIC micelle with extremely narrow size distribution. Accumulation of PIC Micelle to CNV Lesions In PIC micelle group, experimental CNV was created as previously described [14, 15]. Laser photocoagulations were applied to each eye of male Brown Norway (BN) rats between the major retinal vessels around the optic disc with a diode-laser photocoagulator and a slit lamp delivery system at a spot size of 75 μm, duration of 0.05 s, and intensity of 200 mW. To investigate the accumulation of PIC micelle to the CNV lesions, 50 photocoagulations were applied to the right eye of BN rats. The left eyes served as non-photocoagulated controls. By tail injection, 400 μl of PIC micelles encapsulating 5.0 mg/ml FITC-P(Lys) or
Fig. 4 Concentration of FITC-P(Lys) in retina/choroid([20], with permission from Elsevier). Concentration of FITC-P(Lys) in retina/ choroid after rats received equivalent doses of FITC-P(Lys) of PIC micelle-incorporated FITC-P(Lys). Note that FITC-P(Lys) was detected in retina/choroid from the laser-treated eyes of PIC micelle group as early as 1 h and the concentration became peak at 4 h and was still evident 168 h after intravenous administration, whereas the concentration of FITC-P(Lys) in free FITC-P(Lys) group was significantly lower compared to that in PIC micelle group 1 h after injection. Most rats died 1 h after free FITC-P(Lys) administration. Error bars indicate SD. *p>0.05
110 Fig. 5 Dendrimer porphyrin incorporated into supramolecular nanocarrier
Fig. 6 Accumulation of DP after administration of DPincorporated micelles to rats with experimental CNV ([22], with permission from ACS Publications). (a–e) Accumulation of the DP-incorporated micelle (a–c) and free DP (d and e) to experimental CNV. When the DP-incorporated micelle was administered, the accumulation of DP in CNV lesions was observed as early as 15 min after the injection, and peaked at 4 h, and was still evident 24 h after the injection. Whereas, after the free DP injection, DP was also recruited to CNV lesions for up to 4 h, but disappeared within 24 h. (f and g) Immunohistochemistry showing the localization of the micelle and DP and Factor VIII-positive endothelial cells. Note that DP is present within the Factor VIII-positive endothelial cells when the DP-incorporated micelle is administered, whereas DP was present mainly outside the Factor VIII-positive endothelial cells when free DP was administered. PIC polyion complex, DP dendrimer porphyrin, CNV choroidal neovascularization
Tamaki
Novel Drug Delivery System for Age-related Macular Degeneration
111
400 μl of free FITC-P(Lys) at a concentration of 5.0 mg/ml was administered into rats 7 days after photocoagulation. In PIC micelle group, fluorescent staining was observed in the CNV lesions and also in the choriocapillaris for up to 4 h (Fig. 3a and b). Twenty-four hours after the administration, accumulation of FITC-P(Lys) to the CNV lesion was more evident and fluorescence became invisible outside the photocoagulated lesion, including the choroidal and retinal vasculature (Fig. 3c). The fluorescence was observed for up to 168 h (Fig. 3d). One hour after rats received intravenous injection of free FITC-P(Lys), FITC-P(Lys) distributes to the CNV lesion and choriocapillaris (Fig. 3e). Most rats died 1 h after free FITC-P(Lys) administration. In the non-laser-treated eyes, the fluorescence was visible in the choroidal vessels for up to 4 h (Fig. 3f), and the fluorescence became invisible at 24 h and thereafter (Fig. 3g). Light microscopic analysis revealed no abnormalities in other retinal structures. In free FITC-P(Lys) group, most rats died 1 h after free FITC-P(Lys) administration, suggesting that P(Lys) has toxicity. When evaluated 1 h after intravenous administration, fluorescence was observed in the CNV lesion and choriocapillaris. To reduce
the toxic effect of free FITC-P(Lys), rats received a lower dose (10 mg per injection) of free FITC-P(Lys). However, all rats died before 2 h after the administration.
Fig. 7 Efficacy of PDT laser after administration of the DP-incorporated micelle ([22], with permission from ACS Publications). (a) Representative images of fluorescein angiograms in control, PDT-laser-irradiated eye after free DP was administered (DP) and PDT-laser-irradiated eye after the DP-loaded micelle was administered (PIC micelle). (b) Immunostaining of the CNV lesion with factor VIII antibody (control, arrowheads; CNV lesion) (c) Immunostaining of the CNV lesion with factor VIII antibody (PDT laser-irradiated eye after treatment with DPincorporated micelle DP-loaded micelle, arrowhead; normal retinal and
choroidal vessels, autofluorescence from retinal pigment epithelium) (d) Transmission electron microscopy of the CNV lesion of the PDT laserirradiated eye after the DP-incorporated micelle was administered. Note that the enhanced accumulation of DP-incorporated micelles in CNV lesions resulted in a significantly pronounced photodynamic effect, whereas almost all CNV lesions showed strong hyperfluorescence and the CNV endothelial cells appeared normal when free DP was administered. CNV choroidal neovascularization, DP dendrimer porphyrin, PIC polyion complex
Concentration of FITC-P(Lys) in the Laser-Treated Eyes In the PIC micelle group, FITC-P(Lys) was detected in the retina/choroid from the laser-treated eyes as early as 1 h and the concentration peaked at 4 h and the residual FITC-P (Lys) was still evident 168 h after intravenous administration (Fig. 4). Concentration of FITC-P(Lys) was below detectable level in the non-laser-treated eyes. In the free FITC-P(Lys) group, the concentration of FITC-P(Lys) in free FITC-P(Lys) group was significantly lower compared to that in the PIC micelle group when evaluated 1 h after intravenous administration (Fig. 4). In this study, FITC-P(Lys) was retained in the CNV lesion for as long as 168 h after intravenous injection of PIC micelle encapsulating FITC-P(Lys) as demonstrated by histological analysis and confirmed by the measurement of the concentrations in the retina/choroid of the laser-treated eyes. Further, it was demonstrated that a significantly
112
higher amount of FITC-P(Lys) accumulated to the CNV lesions 1 h after the injection in the PIC micelle group compared to the free FITC-P(Lys) group. It is of note that the amount of FITC-P(Lys) in the non-laser-treated eyes was below the detectable level. In addition, because CNVs are highly permeable similar to the newly formed vessels in solid tumors, it is plausible to speculate that PIC micelle is likely to accumulate in CNV lesions presumably through the EPR effect. This idea is consistent with recent studies that have demonstrated that macromolecule accumulates in the CNV lesion in rabbits [16]. In summary, it has been demonstrated that PIC micelle effectively accumulates in the CNV lesion. The distribution of drug-loaded polymeric micelles in the body may be determined mainly by their size and surface properties and are less affected by the properties of loaded drugs if they are embedded in the inner core of the micelles. Considering that PIC micelles are demonstrated to be able to reserve a variety of drugs, enzymes [17], and DNA in the core and can serve as non-viral gene delivery vector [9, 18, 19], PIC micelles have a great potential for achieving effective drug targeting to CNV.
Tamaki
quite different patterns of accumulation. When the DPincorporated micelle was administered, the accumulation of DP to the CNV lesions was observed as early as 15 min after the injection, and peaked at 4 h, and was still evident 24 h after the injection. It is interesting to note that immunohistochemical analysis demonstrated that DP was present within the factor VIII-positive endothelial cells when the DP-incorporated micelle was administered. In contrast, after the free DP injection, DP concentrated in the CNV lesion up to 4 h, but disappeared within 24 h. Immunohistochemical analysis demonstrated that DP was present mainly outside the factor-VIII positive endothelial
Photodynamic Therapy for Exudative AMD Using a Supramolecular Nanocarrier Loaded with a Dendritic Photosensitizer Photodynamic therapy (PDT), which utilizes cytotoxic singlet oxygen produced by photoirradiation of a photosensitizer (PS), is a main treatment option for CNV, and is effective at reducing the relative risk of visual acuity loss; however, it requires expensive treatments and is ineffective for some groups of patient [21]. We have developed an effective PDT for AMD employing a supramolecular nanomedical device, i.e., a novel dendritic PS, dendrimer porphyrin (DP), incorporated into supramolecular nanocarrier, a polyion complex (PIC) micelle through electrostatic interaction with oppositely charged block copolymers (Fig. 5). Biodistribution of Dendrimer Porphyrin-incorporated Micelle To investigate the biodistribution, the free DP or DPincorporated micelle was intravenously administered to rats with photocoagulation-induced CNV. After a single injection, the DP-incorporated micelle showed a prolonged circulation in the blood, whereas the concentration of free DP in blood gradually declined. The accumulation of DP in the CNV lesions was observed by histological analysis. Both the free DP and DP-incorporated micelle were clearly recruited to the CNV lesions. However, they exhibited
Fig. 8 Skin phototoxicity([22], with permission from ACS Publications). Rats were shaved and depilated at least 24 h prior to exposure to light to minimize skin irritation associated with this procedure. Rats were intravenously injected with 400 μl of DP-incorporated micelle or 400 μl of Photofrin® at a concentration of 1.5 mg/ml as a control, and exposed to broad-band visible light for 10 min using a Xenon lamp (150 W) equipped with a filter passing light of 377–700 nm (estimated incident light irradiance; approximately 30 mW/cm2) 4 h after the injection. Skin phototoxicity was not observed macroscopically when the rats were exposed to broadband visible light 4 h after the injection of a DP-incorporated micelle (top panel), despite the relatively high serum levels of the compound. In sharp contrast, severe phototoxicity was observed after the injection of Photofrin® (bottom panel)
Novel Drug Delivery System for Age-related Macular Degeneration
113
Fig. 9 Fundus photography, fluorescein angiography, and indocyanine angiography before (a–c) and 1 week after (d–f) PDT with verteporfin (Visudyne®) in monkey ([23], with permission from the Japanese
Ophthalmological Society). One week after PDT, normal choriocapillaris in the area irradiated with PDT laser was almost occluded (f)
cells when compared to those cells administered the DPincorporated micelle (Fig. 6). These observations agree with our previous finding in vitro that free DP exhibits a lowered cellular uptake because of its negatively charged periphery. Therefore, the DP-incorporated micelle is a very effective vehicle for delivery to factor VIII-positive endothelial cells for efficient photodynamic treatment.
Photodynamic Effect of Dendrimer Porphyrin-incorporated Micelle
Fig. 10 Fundus photography, fluorescein angiography, and indocyanine angiography before (a–c) and 1 week after (d–f) PDT with DPincorporated micelle in monkey ([23], with permission from the
Japanese Ophthalmological Society). No harmful effect on normal retina and choroid was observed after PDT using DP-incorporated micelle in monkeys
Indeed, the enhanced accumulation of micelle-encapsulated DP into CNV lesions resulted in a significantly pronounced photodynamic effect, which was evaluated using fluorescein angiography (Fig. 7). When the PDT laser was applied
114
Tamaki
Fig. 11 In vitro phototoxicity ([22], with permission from ACS Publications). The in vitro cytotoxicity of free DP and DP-incorporated micelle was assessed using Lewis lung carcinoma (LLC) cells. In a darkened room, photosensitizers with different concentrations in medium (Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum) were added to the cell solution in 96-well culture plates (n=4). After 6 h of incubation, the medium was exchanged, and then the plates were photoirradiated with broad-band visible light using a Xenon lamp (150 W) equipped with a filter passing light of 377– 700 nm (Fluence: 18 J/cm2). After 24 h of incubation, the viability of photoirradiated and nonphotoirradiated cells was evaluated by the 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. In vitro cytotoxicity of free-DP (circle) and DP-incorporated micelle (square) in the presence (open symbol) or absence (closed symbol) of photoirradiation is shown. Encapsulation of DP into the micelle resulted in a 130- ~ 280-fold increase in phototoxicity in vitro
15 min after the injection of a DP-incorporated micelle, 78% of the CNV lesions showed no fluorescein leakage at day 1. At day 7, the hypofluorescence persisted, suggesting the leakage from the CNV lesions was still reduced. Similar results were observed when PDT was performed 4 h after the injection. In sharp contrast, neither the administration of DP alone, regardless of whether it was in a micelle or as free DP, nor the application of the PDT laser alone affected the CNV activity (i.e., a similar proportion of the CNV lesions showed leakage [70–80%]) (Table 1). Immunohistochemical analysis with factor VIII demonstrated that CNV endothelial cells were destroyed by the PDT laser compared to the untreated control lesion when the lesion was treated with the laser after the administration of a DP-incorporated micelle.
Transmission electron microscopy revealed that the vessels in the CNV lesion regressed to collagen tubes without endothelial cells or were occluded by erythrocytes. Importantly, endothelial cells of the overlying normal blood vessel in retina and choroid in the treated lesions were not destroyed, even after the application of the PDT laser at maximum energy after the administration of the DPincorporated micelle, probably because neither the PIC micelle nor free DP was taken up into endothelial cells in normal blood vessels in fluid (Fig. 7). Moreover, skin phototoxicity was not observed macroscopically when the rats were exposed to broadband visible light 4 h after the injection of the DP-incorporated micelle, despite the relatively high serum levels of the compound, which is in sharp contrast to the results observed after the injection of Photofrin® (Fig. 8). Furthermore, it had almost no harmful effect on normal retina and choroid after PDT in monkeys (rhesus macaque) (Figs. 9 and 10). In contrast to the suppressive effect after administration of the DP-incorporated micelle, almost all CNV lesions showed a strong hyperfluorescence (i.e., no occlusion of CNV) and the CNV endothelial cells appeared normal when free DP was applied under identical PDT conditions. This seems to be consistent with the observation that encapsulation of DP into the micelle resulted in 130- ~ 280-fold increase in photocytotoxicity in vitro (Fig. 11). Since the DP-incorporated micelle is assumed to gradually dissociate into the constituent DP and block copolymer in the body, long-term phototoxicity is avoidable after PDT using this micelle. In summary, a novel concept for the treatment of CNV using the combination of a dendrimer photosensitizer and polymeric micelles was presented. The DP-incorporated micelle achieved a highly effective accumulation of DP in the CNV lesions, and lower power energy of light was sufficient to occlude the CNV lesions. This might be attributed to a unique characteristic of DP; the aggregation of core porphyrin is sterically prevented even at a much higher concentration. Our data suggest that DP-incorporated micelles significantly enhanced the efficacy of PDT while circumventing side effects to the normal retinal and choroidal vessels and skin. Our results will provide a basis for an effective approach to PDT for AMD.
Table 1 CNV closure rate after PDT([22], with permission from ACS Publications). Light Fluence
Closure rate at day 1 (%) Control
0 J/cm2 5 J/cm2 50 J/cm2
31.7±13.4
Closure rate at day 7 (%)
DP
DP-incorporated micelle
0.25 h
0.25 h
4h
20.9±4.2 19.4±5.6 20.8±4.2
33.3±0 77.7±5.6 77.1±2.9
60.0±10.0 72.2±15.5
Control
15.0±3.5
DP
DP-incorporated micelle
0.25 h
0.25 h
4h
25.0±8.3 33.3±8.3 18.8±2.1
25.0±8.3 80.5±2.8 83.4±5.2
81.7±1.7 77.8±2.8
Novel Drug Delivery System for Age-related Macular Degeneration
References 1. Ferris FL 3rd, Fine SL, Hyman L. Arch Ophthalmol 1984; 102:1640–2. 2. Bressler NM, Bressler SB, Fine SL. Surv Ophthalmol 1988; 32:375–413. 3. Macular Photocoagulation Study Group. Arch Ophthalmol 1991;109:1242–57. 4. Treatment of Age Related Macular Degeneration with Photodynamic Therapy (TAP) Study Group. Arch Ophthalmol 1999; 117:1329–45. 5. The Eyetech Study Group. Ophthalmology 2003;110:979–86. 6. Heier JS, Antoszyk AN, Pavan PR, et al. Ophthalmology 2006;113:633–42. 7. Kimura H, Yasukawa T, Tabata Y, Ogura Y. Adv Drug Deliv Rev 2001;52:79–91. 8. Kataoka K, Harada A, Nagasaki Y. Adv Drug Deliv Rev 2001; 47:113–31. 9. Kakizawa Y, Kataoka K. Adv Drug Deliv Rev 2002;54:203–22.
115 10. Matsumura Y, Maeda H. Cancer Res 1986;46:6387–92. 11. Nakanishi T, Fukushima S, Okamoto K, et al. J Control Release 2001;74:295–302. 12. Nishiyama N, Kato Y, Sugiyama Y, Kataoka K. Pharm Res 2001;18:1035–41. 13. Harada A, Kataoka K. Macromolecules 1995;28:5294–9. 14. Yanagi Y, Tamaki Y, Obata R, Muranaka K, Homma N, Matsuoka H, Mano H. Invest Ophthalmol Vis Sci 2002;43:3495–9. 15. Yanagi Y, Tamaki Y, Inoue Y, Obata R, Muranaka K, Homma N. Invest Ophthalmol Vis Sci 2003;44:751–4. 16. Yasukawa T, Kimura H, Tabata Y, Miyamoto H, Honda Y, Ikada Y, Ogura Y. Invest Ophthalmol Vis Sci 1999;40:2690–6. 17. Harada A, Kataoka K. J Control Release 2001;72:85–91. 18. Harada–Shiba M, Yamauchi K, Harada A, Takamisawa I, Shimokado K, Kataoka K. Gene Ther 2002;9:407–14. 19. Harada A, Togawa H, Kataoka K. Eur J Pharm Sci 2001;13:35–42. 20. Ideta R, Yanagi Y, Tamaki Y, et al. FEBS Lett 2004;557:21–5. 21. Meads C, Hyde C. Br J Ophthalmol 2004;88:212–7. 22. Ideta R, Tasaka F, Jang WD. Nano Lett 2005;5:2426–31. 23. Tamaki Y. J Jpn Ophthalmol Soc 2007;111:232–69.