Detection of Choroidal Neovascularization Using Optical Tissue Transparency

Abstract

Purpose: Optical tissue transparency (OTT) provides a tool for visualizing the entire tissue block. This study provides insights into the potential value of OTT with light-sheet fluorescence microscopy (LSFM) in detecting choroidal neovascularization (CNV) lesions.

Methods: OTT with LSFM, hematoxylin and eosin (H&E) staining of paraffin sections, choroidal flatmount immunofluorescence, and optical coherence tomography angiography (OCTA) were used to obtain images of CNV. We determined the rate of change as (Data of week 1 – Data of week 2)/Data of week 1 × 100%. Finally, we compared the rate of change acquired from OTT with LSFM and the other methodologies.

Results: We found that OTT with LSFM can realize three-dimensional (3D) visualizations of the entire CNV. The results showed that the decline in the rate of change from week 1 to week 2 after laser photocoagulation was 33.05% with OTT, 53.01% with H&E staining, 48.11% with choroidal flatmount, 24.06% with OCTA (B-scan), 18.08% with OCTA (en face), 10.98% with OCTA (3D reconstruction), and 7.74% with OCTA (vessel diameter index).

Conclusions: OTT with LSFM will continue to be an invaluable resource for investigators to detect more visualized and quantified information regarding CNV.

Translational Relevance: OTT with LSFM now serves as a tool for detecting CNV in mice, and it may undergo human clinical trials in the future.

Introduction

Age-related macular degeneration (AMD) is a progressive disease that leads to severe vision loss.¹ Exudative AMD is characterized by choroidal neovascularization (CNV),² which can be induced via laser to produce a reliable model that mimics exudative AMD.³ Several methods have been employed to measure CNV lesions, including choroidal flatmount immunofluorescence, optical coherence tomography angiography (OCTA), and hematoxylin and eosin (H&E) staining. 

The principle of optical tissue transparency (OTT)⁴ involves a reduction in light absorption and light scattering,⁵ maintenance of the fluorescent protein signal, and the removal of melanin.⁶ OTT with light-sheet fluorescence microscopy (LSFM) is a novel tool for the visualization and quantification of vasculature and has been applied in the imaging of the mouse kidney, carotid artery, tail, and eyes.^7–13 In the corneal inflammatory neovascularization model, pseudocolors were applied to distinguish the adjacent spatial location of the lymphatic vessels and Schlemm's canal.¹⁰ In an oxygen-induced retinopathy model, LSFM exhibited and quantified a “knotted” morphology in pathological vascular tufts,¹¹ and in Norrie disease quantification of vascular volume revealed severe hypovascularization surrounding the retinal optic nerve.¹² Furthermore, with regard to CNV, the method showed that in a mouse transforming growth factor-beta receptor type 2 (TGFβR2) knockout model, vascular endothelial cells formed an irregular new plexus between the retina and choroid.¹³ However, OTT with LSFM imaging has not been used to quantify CNV lesions. In this study, laser-induced CNV was used to estimate the potential application value of OTT in comprehensive morphological studies of CNV. 

Methods

Mice

All C57BL/6 mice 6 to 8 weeks of age were purchased from Charles River Laboratory (Beijing, China) and housed in the Animal Center of Tongji Hospital in Wuhan under a 12-hour light/dark cycle. All animal experiments were approved by the Animal Research Institute Committee of Tongji Medical Center and conducted in strict accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. 

Laser-Induced CNV Model

After dilation with tropicamide/phenylephrine eye drops, anesthetized mice were placed on a scaffold for the subsequent procedures. Laser-induced CNV broke the Bruch’s membrane of both eyes with a 100-µm spot size, 100-ms pulse interval, and 200-mW energy (wavelength 532 nm; ZEISS, Oberkochen, Germany). Cavitation bubbles without bleeding were considered to be successful CNV lesions; lesions that resulted in hemorrhage were excluded from further analysis. Four to five laser burns were made in each eye, and the laser points used in this study were located 1 to ∼2 optic nerve head sizes around the optic nerve. 

Sampling

C57BL/6J mice were anesthetized, and the perfusion needle was then inserted into the left ventricle. The right atrium was snipped, and then room-temperature 55.6 mg/L heparin (w/v) was used. The drip rate was about 1 drop per second. The 15-minute 3-mL/min infusion was followed by 4% (w/v) paraformaldehyde (Biosharp, Hefei, China) at the same rate and time. After perfusion, the muscles and fascia on the surface of the eyeball were removed, and 3 mL fixation solution (Servicebio, Wuhan, China) was used for each eye; they were stored at 4°C for 1 to 4 days until further processing. 

OTT of the Eye

Bleaching

The eyes were washed with PBS at 4°C for 1 hour, and the PBS was replaced every 20 minutes. Subsequently, each eye was maintained at 55°C with 3 mL preheated 1:2 H₂O₂ (Changle Fu Qiang Technology Company, Wefang, China) and PBS for 3 hours, and the PBS was replaced every 30 minutes. Subsequently, each eye was placed in 1 mL of room-temperature pentraxin-2 (PTX-2)¹⁴ for 2 hours, and the buffer was changed every hour. The insulin needle was inserted into a hole about 2 mm in front of the corneal margin. 

Immunofluorescence Staining

The whole process was carried out at a constant temperature of 37°C. The eyeballs were rinsed in 1 mL osmotic solution for 2 days and 1 mL blocking solution for 1 day, followed by anti-CD31 monoclonal antibody (1:250; Abcam, Cambridge, UK) incubation for 3 days. The eyeballs were then washed with PBS/0.2% Tween-20 with 10 µg/mL heparin (PTwH)¹³ for 1 day and incubated in Alexa Fluor 647 secondary antibody (1:500; Abcam) for 2 days, followed by another PTwH washing for 1 day. 

Dehydration

The eyeballs were embedded in 1% (w/v) low-melt agarose. After cooling down, the agarose blocks were cut into about 0.4 × 0.4 × 0.4 cm³ blocks, and then the blocks were dehydrated in a series of ethanol gradients: 25%, 50%, 70%, and 100%. 

Transparency

Dehydrated agarose cubes were maintained at room temperature for 1 hour, then placed in a tube filled with ethyl cinnamate at room temperature for 6 to 8 hours. Thereafter, eyeballs were completely transparent for optical imaging (week 1 n [n1] = 18 lesions, 12 eyes; week 2 n [n2] = 16 lesions, 5 eyes). 

LSFM Imaging

Three-dimensional (3D) eye images were acquired using digital scanned light-sheet microscopy (MacroView, customized; Olympus, Tokyo, Japan). For illumination, a long working distance objective (5×/0.14; Mitutoyo, Kawasaki, Japan) was used to generate a light sheet ∼2.7 µm thick. For detection, a zoomable macro-view imaging setup (0.63× to 6.3×, combined with a 2× detection objective lens, Olympus MVPLAPO 2XC) was used with a scientific complementary metal-oxide semiconductor (sCMOS) camera (ORCA-Flash4.0 V2; Hamamatsu Photonics, Shiquoka, Japan) to obtain the fluorescence images at high speed. The camera pixel size is 6.5 × 6.5 µm. All images were acquired with a 2-µm z-step size, 120-µs line exposure time, and sampling at full resolution (2048 × 2048). 

Paraffin Sections and H&E Staining

After an eyeball was embedded in paraffin and sections were cut with a thickness of 4 µm, 20 representative sections containing the thickest injury were taken from each CNV lesion, and then an ordinary optical microscope with a digital camera (Motic BA310; Motic Microscopes, Kowloon, Hong Kong) was used to acquire five continuous CNV sections passing through the optic nerve head. ImageJ (National Institutes of Health, Bethesda, MD) was used to measure the thickness of each CNV lesion. The five thickest sections were averaged as the CNV thickness of each laser-induced injury (n1 = 24 lesions, 6 eyes; n2 = 19 lesions, 5 eyes). 

Choroidal Flat Mount Immunofluorescence

Mice were euthanized and eyeballs were enucleated and fixed with 4% paraformaldehyde for 15 minutes (Biosharp). The choroid/RPE/scleral complex was isolated and permeabilized with PBS containing 0.5% Triton X-100 (Solarbio Life Science, Beijing, China), 5% fetal bovine serum (Hycezmbio, Wuhan, China), 5% donkey serum (Servicebio), and 20% DMSO (Sigma-Aldrich, St. Louis, MO) for 3 hours at room temperature. For immunofluorescence, we detected blood vessels using anti-CD31 monoclonal antibody (1:250; Abcam) with Alexa Fluor 488 secondary antibody (1:500; Abcam) and detected the pericyte-like scaffold by anti–platelet-derived growth factor receptor-β (PDGFR-β) monoclonal antibodies (1:250; Thermo Fisher Scientific, Waltham, MA) with Alexa Fluor 405 secondary antibody (1:500; Abcam). For all procedures mentioned, the choroid complex was immersed in the primary antibody solution at 4°C for 2 days. After washing with PBS six times, the choroid complex was immersed in the secondary antibody solution at 4°C for 1 day and washed with PBS six times. Finally, five incisions were made near the center followed by mounting with antifade mounting medium (Servicebio) and coverslipping (n1 = 38 lesions, 14 eyes; n2 = 46 lesions, 12 eyes). 

Optical Coherence Tomography Angiography

Microvascular imaging of mouse retinas was performed using an ophthalmic ultra-microscopic imaging system (ISOCTA, Pontypridd, UK) that was equipped with a central scanning wavelength of 1060 nm and an A-scanning rate of 200 kHz. The scanning region was 2.5 × 2.5 mm, with the optic nerve head as the center. Ten B-scans were obtained from 1024 cross-sectional locations. After collecting images from OCTA, quantitative analysis was performed using RestrUI and VCCcular. The data we analyzed included OCTA B-scans (n1 = 20 lesions, 6 eyes; n2 = 20 lesions, 6 eyes), en face OCTA (n1 = 32 lesions, 29 eyes; n2 = 32 lesions, 29 eyes), 3D OCTA reconstruction performed using B-scan images and Imaris 9.0.1 (n1 = 20 lesions, 10 eyes; n2 = 20 lesions, 10 eyes), and OCTA vessel diameter index (n1 = 25 lesions, 25 eyes; n2 = 25 lesions, 25 eyes). 

Statistics

All statistical analyses were performed using Prism 8.2.0 (GraphPad, San Diego, CA). All data are presented as mean ± SEM. The statistical significance of the OCTA (3D reconstruction, B-scan, en face, and vessel diameter index) data was determined with paired Student's t-tests and other data with unpaired Student’s t-tests. P < 0.05 is statistically significant. 

Data and Resource Availability

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. 

Results

With a slight modification of a whole-eye clearing protocol,¹⁴ we immunostained and cleared whole eyes (Fig.) in 13 days. In the posterior LSFM view,¹⁵ we scanned the choroidal vasculature in three vertical planes defined by the long posterior ciliary arteries (LPCAs) and the inferior branch of the temporal LPCA. The scans showed blood vessel structures from different orientations (Fig., Supplementary Movie S1). We found it feasible to visualize the intact CNV with OTT with LSFM. We determined the rate of change as (Data of week 1 – Data of week 2)/Data of week 1 × 100%. OTT showed that the CNV pseudocolor volume decreased by approximately 33.05% from week 1 to week 2 after laser photocoagulation (Fig., Supplementary Movie S2). The 3D reconstruction of the OCTA showed that the CNV volume decreased 10.98% (Fig.). 

Figure.

 

Images acquired from OTT with LSFM and other methods. (A) Representative phase image of the eyeball followed by each step. Scale bar: 1 mm. RT, room temperature. (B) Representative posterior view images acquired from LSFM. The X plane is parallel to the ocular axis and LPCA, the Y plane is parallel to the ocular axis, and the Z plane is perpendicular to the X and Y planes. Scale bar: 500 µm. (C) Two-dimensional and 3D images of the CNV. Scale bar: 500 µm (left), 100 µm (right) (n1 = 18; n2 = 16). (D) Representative 3D reconstruction of OCTA on the same CNV. Scale bar: 500 µm (left), 150 µm (magnified image) (n1 = 20, n2 = 20). (E) Representative images acquired from choroidal flatmount immunofluorescence. Green indicates anti-CD31; blue indicates anti-PDGFR-β. Scale bar: 500 µm (left), 50 µm (right) (n1 = 39, n2 = 46). (F) Representative H&E staining of paraffin sections. Scale bar: 500 µm (left), 100 µm (right) (n1 = 24, n2 = 23). (G) Representative OCTA of the same position. Scale bar: 500 µm (left), 250 µm (right) (n1 = 20, n2 = 20). (H) Representative dynamic OCTA en face image from the same CNV lesion. Scale bar: 500 µm (left), 250 µm (right) (n1 = 32, n2 = 32). (I) Representative vascular skeleton of OCTA en face. Scale bar: 500 µm (n1 = 25, n2 = 25). Note that sample size n1 is for week 1 and n2 is for week 2.

The rate of change of the CNV area was also detected by choroidal flatmount immunofluorescence using the data from week 1 and week 2 after laser photocoagulation. On the immunofluorescence, the CD31⁺ region represented the vessels, and the PDGFR-β⁺ region represented pericyte-like scaffolds observed at the laser-induced lesions. Consistent with the change trend observed with OTT, the mean area of CNV damage was significantly larger in week 1 than in week 2, showing that the rate of change of new vessel area declined by approximately 48.11% from week 1 to week 2 after laser photocoagulation (Fig.). 

For the vertical dimensions of the CNV, the rate of change for H&E staining was found to be similar to that observed with choroidal flatmount immunofluorescence, and quantitative analysis of the neovasculature revealed that the thickness of the newly formed vessels decreased by approximately 53.01% from week 1 to week 2 after laser photocoagulation (Fig.). 

CNV detected by OCTA imaging suggested that the rate of change of neovascular thickness from week 1 to week 2 after laser photocoagulation based on OCTA B-scans decreased by 24.06%, whereas the rate of change of the CNV lesion area declined by approximately 18.08%, and the rate of change of the vessel diameter index decreased by 7.74% (Fig.). 

In comparison, the decrease in the rate of change based on OTT from week 1 to week 2 after laser photocoagulation was higher than that provided by OCTA, whereas H&E staining and choroidal flatmount immunofluorescences were more efficient in detecting laser-induced CNV (Table). 

Table.

 

Choroidal Neovascularization Rates of Change Statistics Observed by Different Methods

 

Discussion

OTT with LSFM has been previously applied to research on corneal and retinal neovascularization; however, in our study, we used this method to evaluate CNV. The CNV area regressed spontaneously at week 1 after laser injury, with the average CNV area peaking at week 1 and decreasing between week 1 and week 2 after laser injury.^16,17 In this study, the rates of change from week 1 to week 2 acquired from the different methods exhibited an identical direction. 

A series of factors affected the CNV rate of change between week 1 and week 2. First, OCTA detects erythrocyte flow in vivo and is affected by fibrosis artifacts,¹⁸ motion artifacts, and resolution.¹⁹ Moreover, the process involved in choroidal flatmount fluorescence cannot preserve the spherical state of the eyes and thus may affect the natural vascular structure.²⁰ However, OTT with LSFM allows the visualization of intact natural CNV (Fig.), as the area is not cut into sections, and choroidal flatmount, with a large field of view, allows thin light-sheet slice-and-scan in one step, enabling lower quenching fluorescence signals. 

There are several limitations of this study, however. Because of the small sample size, it was not possible to fully control the reliability and accuracy of the results of this study. Furthermore, a drawback of the method is that its accuracy relies on the experience of the microscopist; inexperience can cause the bleach and dehydration process to detach the retina, requiring prolongation of the perfusion/fixation time to 2 to 3 days, according to our test. 

Conclusions

This study described a new slice-and-scan, one-step 3D histological method for quantifying the progressive regression of whole laser-induced CNV. OTT with LSFM can provide more comprehensive visualization and quantification than traditional histological methodologies and could be applied to monitor the fine pathological morphology in exudative AMD in the future. 

Acknowledgments

Supported by a grant from the National Natural Science Foundation of China (81974136). 

Disclosure: X.-H. Ma, None; W.-Y. Feng, None; K. Xiao, None; Z. Zhong, None; P. Fei, None; Y. Zhao, None; X.-F. Sun, None 

Supplementary Material

Supplementary Movie S1. Ocular vessels of mice acquired by OTT with LSFM. 

Supplementary Movie S2. CNV detected by OTT with LSFM.