In vivo imaging of choroidal angiogenesis using fluorescence-labeled cationic liposomes

Synthesis and analytical characterization of fluorophore-labeled cationic liposomes

1,2 Dioleoyl-3-trimethylammonium propane (DOTAP) was from Merck Eprova (Schaffhausen, Switzerland), and 1,2 dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1,2 dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Rh-DOPE) for making CL-Rh were from Avanti Polar Lipids, Inc. (Alabaster, AL). (Oregon Green 488) 1,2 dihexadecanoyl-sn-glycero-3-phosphoethanolamine (OG-DHPE) for making CL-OG was obtained from Invitrogen (Karlsruhe, Germany). Indocyanine green-1,2 di (cis-9-octadecenoyl)-sn-glycero-3-phosphoethanolamine (ICG-DOPE) for making CL-ICG was synthesized by Molecular Probes (Invitrogen, Karlsruhe, Germany, custom synthesis). Liposomes with a total lipid concentration of 10 mM were prepared by the lipid film method (see Table 1 for compositions). To prepare the liposomes, the respective lipids and lipid-coupled dyes were dissolved in 15 ml chloroform in a round bottom flask. The mixture was gently warmed to 40 °C, and the solvent was evaporated at reduced pressure in a rotary evaporator. The resulting lipid film was dried under vacuum for about 60 min, and then it was resuspended in an aqueous solution containing 10% trehalose. The resulting suspension of the multilamellar vesicles was extruded (10 ml Lipex extruder; Northern Lipids Inc., Vancouver, BC, Canada) five times through a 0.2 µm polycarbonate membrane (GE Osmonics, Minnetonka, MN), yielding unilamellar liposomes. To prepare 10 ml of neutral liposomes, 0.095 mmol DOPC and 0.005 mmol OG-DHPE were dissolved in chloroform and treated as described above.

The particle size and size distribution of the liposomal suspensions were analyzed with photon correlation spectroscopy (Malvern Zetasizer 3000; Malvern Instruments, Herrenberg, Germany), and the results are listed in Table 1. The concentration and purity of the liposomal components were checked with high-performance liquid chromatography (HPLC)-ultraviolet/visible spectroscopy.

The fluorescence properties of all formulations were optimized and checked with fluorescence spectroscopy using a fluorescence spectrophotometer (HORIBA Jobin Yvon GmbH, Munich, Germany). Liposomal formulations were diluted 1:50 (CL-ICG) or 1:2,000 (CL-OG) in trehalose. The excitation wavelength was set to 805±5 nm (ICG) or 490±5 nm (OG), and the emission was measured at 825±5 nm (ICG) or 534±5 nm (OG).

Mouse RPE was isolated from a freshly prepared posterior eye cup after the retina was removed and appropriately diluted. The absorption spectrum demonstrates the amount of light absorbed by RPE in different wavelengths and thus estimates the reduction in fluorescence emitted by various fluorophores, when they are located behind the RPE in vivo.

Laser-induced choroidal neovascularization model

All animal procedures adhered to the animal care guidelines of the Institute for Laboratory Animal Research (Guide for the Care and Use of Laboratory Animals) in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the local animal welfare committee.

The laser-induced CNV model in mice was performed as previously described [19]. Briefly, C57BL/6J mice (Charles River Laboratories, Hamburg, Germany) were anesthetized with an intraperitoneal injection of a ketamine (100 mg/kg) and xylazine (5 mg/kg) mixture. Then 0.5% tropicamide (Pharma Stulln, Stulln, Germany) and 0.5% phenylephrine (Ursapharm, Saarbrücken, Germany) eye drops were used to dilate pupils. Three laser spots per eye were induced with an argon laser (100 ms, 100 μm, and 150 mV, 532 nm, Visulas 532s, Carl Zeiss Meditec AG, Jena, Germany). At least three mice were used per time point and treatment.

In vivo imaging with scanning laser ophthalmoscope

Fluorophore-labeled liposomes were injected into the tail vein (10 mM, 150 μl per mouse). Mice were anesthesized, and the pupils dilated as described above for CNV induction by laser. CNV were localized in the infrared mode of a digital SLO (HRA1, Heidelberg Engineering, Heidelberg, Germany). Fundus images were taken with filter sets for fluorescein (488 nm excitation, 530 nm emission) and ICG (795 nm excitation, 830 nm emission). Images were taken before and at different time points after intravenous injections of liposomal formulations and analyzed with ImageJ. The fluorescence intensity of a CNV (I_L) was calculated by selecting a region of interest covering the laser CNV and dividing the sum of the intensity values of each pixel in that area by the number of the pixels. In the same way, background fluorescence (I_B) was averaged from three areas of identical size and shape as the CNV area that were selected in close proximity to the CNV. An accumulation index was defined as I_L divided by I_B. Values were tested for statistical significance with ANOVA, and differences were assumed to be significant when p<0.05.

Sclerochoroidal flatmounts and microscopy

Immediately after the in vivo SLO images were taken, eyes were fixed in 4% formalin buffered with PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄). After the anterior eye part and the retina were removed, the sclera and choroid complex including the RPE was flatmounted and examined with epifluorescence microscopy with 480 nm excitation and 510 nm emission filters without additional staining. Since ICG signals are detectable only in the infrared channel, which is unavailable in microscopes, only CL-OG injected eyes were included. Images were taken, and the selectivity in CNV lesions was quantified in the same way as in the SLO images. Three control areas were selected in healthy RPE outside the CNV area.

Histological localization of cationic liposome and fluorescein isothiocyanate-lectin

Mice with L-CNV received an intravenous application of cationic liposomes labeled with rhodamine (CL-Rh) at d5 or d14 after laser coagulation (4 µl/g). The liposomes were allowed to circulate for 60 min before the animals received 100 µl fluorescein isothiocyanate (FITC) lectin (2 mg/ml, Vector Labs, Burlingame, CA) intravenously. After 5 min, the animals were perfused with 1% paraformaldehyde. Eyes were removed and embedded in optimum control temperature medium (TissueTek, Sakura, Japan) and quick-frozen in a Leica cryostat (Leica, Wetzlar, Germany). Then 50 µm sections were cut, dried for 30 min at room temperature, fixed in 4% buffered formalin, and stained with 10 µM ToPro-3 (Invitrogen, Karlsruhe, Germany) for 1 min at room temperature. After being washed with PBS, sections were mounted in Dako faramount aqueous (Dako, Hamburg, Germany), left to dry for 60 min at room temperature in the dark, and analyzed with a Leica TCS SP2 confocal microscope (Leica, Heidelberg, Germany) at excitation wavelengths of 488 nm, 543 nm, and 633 nm.

Optimization of fluorophore-labeled liposomes

The fluorophores OG, ICG, and Rh were stably linked to CL as they were covalently bound to phospholipids. This was confirmed with high-performance liquid chromatography analysis that did not detect any unbound fluorophore at any time during the study. Fluorescence intensity was optimal at 1 mol% ICG and 5 mol% OG in total liposomes due to intraliposomal quenching at higher fluorophore ratios (Figure 1). The inset in Figure 1 shows the respective molecular structure; OG differs from fluorescein only by having two F atoms at the aromatic ring system. In both structures, “R” indicates where the linker to the lipid moiety has been attached. Quenching of Rh was similar to OG in CL. Therefore, Rh was also used at a concentration of 5 mol%.

In vivo short-term kinetics of fluorophore-labeled liposomes (scanning laser ophthalmoscope imaging)

The distribution of CL labeled with ICG or OG in the fundus was followed by SLO after intravenous injection. Figure 2 shows representative images at various time points after application of labeled CL-ICG or CL-OG 14 days after laser coagulation. No specific signal was detected before CL-ICG or CL-OG was injected. Circulation of CL formulations in the blood plasma was faintly visible for 2–3 min after application based on the respective fluorescence signal (Figure 2B). After that, fast accumulation of fluorescence at the CNV site was observed. Typically, the signal due to CL-OG was visible between 20 and 120 min after application. The CL-ICG signal was stronger and therefore was visible as early as 5 min after application and was observed for several hours, in some cases even for 24 h. The CL-ICG and CL-OG peak signal was observed between 30 and 90 min so that a measuring time point of 60 min after injection was chosen in the following experiments. CL-ICG resulted in a more intense and precise image compared to CL-OG (Figure 2C). In contrast, neutral liposomes did not accumulate at the laser site but remained much longer in circulation (Figure 3) similar to ICG or fluorescein in angiography.

Kinetics of choroidal neovascularization formation detected with scanning laser ophthalmoscope

CL-OG, CL-ICG, and neutral liposomes labeled with OG were intravenously applied to mice on various days after laser coagulation. SLO images were recorded 60 min after injection. The CL-ICG yielded a clear signal five days after laser treatment. Both CL formulations showed a significant enrichment at the CNV site at d10 and d14 compared to neutral liposomes (Figure 3). The highest value was reached by CL-ICG at d14, where a 3.6-fold accumulation in the CNV was found. The CL-OG showed a weaker signal than CL-ICG probably because of absorption of the OG fluorescence by the RPE (compare Figure 3C). This shows again that CL-ICG’s signal was superior to that of CL-OG.

In the d1 CL-ICG image (Figure 3), the central laser burn was a black hole surrounded by CL-ICG signal (accumulation <1) probably due to activation of endothelial cells through local inflammation. This hole was gradually filled in with blood vessels during the course of the study as illustrated in the CL-ICG images suggesting that the accumulation of CL correlates with the growth of choroidal neovessels within the CNV.

The SLO images obtained with CL represented the morphology of CNV lesions well at d10 and d14, which appeared not as a homogenous distribution but as hot spots whose structure resembled blood vessels (Figure 2 and Figure 3). Neutral liposomes showed no accumulation in the CNV lesions (Figure 3), but remained in the circulation even 60 min after the intravenous injection similar to ICG or fluorescein in angiography. This demonstrates that the cationic charge is important for the accumulation of CL, which depends on the biologic status of the growing CNV.

Confirmation of laser-choroidal neovascularization development with flatmount analysis

The choroids of the mice that were imaged with OG-labeled liposomes were flatmounted and investigated with fluorescence microscopy. Accumulation indices were computed analogous to the procedure with SLO and are shown in Figure 4B. Similar to the SLO data in Figure 3B, there was a significantly increased accumulation of CL-OG compared to neutral liposomes labeled with OG at d10 and d14 confirming that CL bound specifically to the CNV site. The CL-OG signal was much weaker at d1 and d5, accompanied by homogeneous background staining. Flatmounts from mice treated with neutral liposomes showed background staining as well. Background staining is typically derived from tissues such as muscle showing autofluorescence at the wavelength used for detecting OG.

Histological localization of cationic liposomes

To detect CL in histological sections, CL-Rh was applied intravenously, and blood vessels were labeled with fluorescein cyanate (FITC) lectin. Figure 5 shows representative images of laser-CNV at d5 and d14 after laser coagulation. The images show a well defined vessel structure in the retina and in the choroid. Staining with ToPro-3 allowed the nuclear layers of the retina that were deformed at the laser site to be identified. CL were exclusively found at the laser site. However, CL were not confined to the vessels but showed some spotty distribution around the vessels within the area of the CNV. There was no non-specific binding of CL to resting blood vessels, to photoreceptors, or to other structures. The amount of CNV labeled with CL-Rh increased from d5 to d14 confirming the results from the SLO and flatmount analysis.