Multimodal Retinal Imaging Evaluation of Macular Degeneration Persistent Disease Activity Animal Model in Watanabe Heritable Hyperlipidemic Rabbits

Abstract

The Watanabe Heritable Hyperlipidemic (WHHL) rabbit, a model for familial hypercholesterolemia, offers a unique opportunity to study lipid metabolism disorders and their ocular effects. This study employed multimodal imaging, including color fundus photography, fluorescein angiography, indocyanine green angiography, photoacoustic microscopy (PAM), and optical coherence tomography (OCT), to longitudinally assess WHHL rabbit eyes over one year. Given its relevance to persistent disease activity (PDA) in age-related macular degeneration (AMD), WHHL rabbits were evaluated as a potential model for PDA-associated AMD. Choroidal neovascularization (CNV) was induced in WHHL rabbits via subretinal injection of VEGF and Matrigel, followed by bevacizumab (Bev) treatment. A progressive lipid layer developed in the cornea by month 7. Histological analysis revealed significant outer nuclear layer cell loss in WHHL rabbits compared to wild-type (WT) controls. OCT imaging demonstrated increased CNV thickness in WHHL rabbits, consistent with PDA. Bev treatment reduced CNV leakage by 90.13% in WT rabbits but only 16% in WHHL rabbits, indicating treatment resistance. PAM at 780 nm effectively distinguished CNV from surrounding microvasculature, while OCT provided detailed retinal imaging. Although rabbits lack a fovea, the WHHL model demonstrates key features of wet AMD, including PDA and reduced anti-VEGF response. Our findings support WHHL rabbits as a promising preclinical model for studying lipid-related retinal diseases, AMD pathophysiology, and treatment resistance, advancing research into novel therapeutic strategies.

Introduction

Watanabe heritable hyperlipidemic (WHHL) rabbits are a specialized breed of laboratory animal widely employed in cardiovascular disease research. Developed in Japan during the early 1980s by Watanabe et al.(Shiomi, 2020; Shiomi et al., 1992; Watanabe, 1980; Watanabe et al., 1985), these rabbits exhibit genetically predisposed high levels of cholesterol and triglycerides, which are critical lipids in the blood(Fan et al., 2000). This unique characteristic makes WHHL rabbits an invaluable model for studying hyperlipidemia, a condition marked by elevated blood lipid levels, and its implications in atherosclerosis—the accumulation of fatty deposits within arterial walls(Fan et al., 2015; Niimi et al., 2020). The WHHL model has facilitated numerous investigations into the effects of dietary interventions, exercise, and pharmaceutical treatments aimed at preventing or managing atherosclerosis(Getz and Reardon, 2012; Napoli et al., 2006). Furthermore, these rabbits serve as a platform for testing the efficacy and safety of emerging therapeutics targeting hyperlipidemia and related cardiovascular conditions. Recent genomic research has enhanced the utility of WHHL rabbits for gene therapy applications. By establishing the genomic framework of WHHL rabbits, many studies have laid the groundwork for developing targeted gene therapies to address hyperlipidemia, potentially leading to novel treatment strategies(Cao et al., 2021; Hoekstra and Van Eck, 2024; Preta, 2023; Yuan et al., 2019).

While WHHL rabbits are primarily recognized for their contributions to cardiovascular research, emerging studies suggest a connection between lipid dysregulation, inflammation, and macrophage activation in various ocular conditions, including age-related macular degeneration (AMD) and diabetic retinopathy(Fu et al., 2024). For instance, research published highlighted elevated lipid levels in the retinas of WHHL rabbits, which could predispose them to AMD(Kouchi et al., 2006). Similarly, Jenkins and Shan et al. indicated that hyperlipidemia could lead to retinal vascular damage, a defining feature of diabetic retinopathy(Jenkins et al., 2022; Shan et al., 2022). However, despite these findings, the morphological and functional changes in the retina of WHHL rabbits throughout their lifespan remain inadequately explored.

To investigate the dynamic changes and structural characteristics of the retina in rabbits, various optical imaging modalities have been employed, including color fundus photography (CFP)(Luis Arias and Jordi Monés, 2010; Venkatesh et al., 2023), fluorescein angiography (FA)(Ricardi et al., 2024), indocyanine green angiography (ICGA)(Choi et al., 2023; Ebrahimiadib et al., 2023), optical coherence tomography (OCT)(Chua et al., 2024; Kaizu et al., 2017), and scanning laser ophthalmoscopy (SLO)(Hammer et al., 2006; Li et al., 2019; Mainster et al., 1982; Nguyen et al., 2021a; Nguyen et al., 2021b; Nguyen et al., 2023a; Nguyen et al., 2024a; Nguyen et al., 2024b; Nguyen et al., 2021d; Nguyen et al., 2019; Tian et al., 2017; Nguyen et al., 2023b; Webb and Hughes, 1981). These techniques offer high-resolution images and excellent contrast for visualizing retinal and choroidal structures. Nevertheless, traditional imaging modalities are limited in their capacity to provide a comprehensive three-dimensional view of the deeper choroidal layers and can be susceptible to motion artifacts, complicating the assessment of subretinal fluid accumulation and other pathological changes. Photoacoustic microscopy (PAM) has emerged as a promising imaging technology, capable of non-invasively visualizing various retinal structures with remarkable depth of penetration and high resolution(Beard, 2011; de La Zerda et al., 2010; Jeon et al., 2017; Zhang et al., 2018). By integrating the contrast of optical imaging with the depth capabilities of ultrasound, PAM enables the visualization of both superficial and deep tissue structures, including blood vessels and cellular components. Additionally, PAM can be synergistically combined with other imaging modalities, such as OCT, to enhance overall imaging capabilities(Li et al., 2019; Nguyen et al., 2021a; Nguyen et al., 2021b; Nguyen et al., 2023a; Nguyen et al., 2024a; Nguyen et al., 2024b; Nguyen et al., 2021d; Nguyen et al., 2019; Tian et al., 2017; Nguyen et al., 2023b).

Our research aims to provide a comprehensive evaluation of the WHHL retina through a combination of imaging and functional assessments. Utilizing multimodal imaging techniques including CFP, FA, ICGA, OCT, and PAM, we monitor dynamic changes in the retina of WHHL rabbits over time. Furthermore, we assess retinal function using electroretinography (ERG). Additionally, the response to anti-VEGF treatment is evaluated in WHHL rabbits using a choroidal neovascularization (CNV) model induced by subretinal injection of Matrigel and VEGF-165^36,(Qiu et al., 2006), followed by intravitreal injection of bevacizumab (Bev), and anti-VEGF antibody to remove the CNV. This study aims to deepen our understanding of retinal pathophysiology in WHHL rabbits and explore the potential implications for this novel animal model of persistent disease activity in AMD. Despite regular anti-VEGF therapy, some patients continue to exhibit disease activity, characterized by persistent or recurrent fluid accumulation, hemorrhage, or vision loss.

Methods

Animal Model Preparation:

All rabbit experimental procedures were conducted in alignment with the ARVO (The Association for Research in Vision and Ophthalmology) Statement for the Use of Laboratory Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care & Use Committee (IACUC) of the University of Michigan (Protocol PRO00010388).

Six New Zealand White WHHL (Watanabe Heritable Hyperlipidemic) rabbits 2 to 3 months of age, both genders (male and female) and weighing 1.9 to 2.45 kg were obtained from the Center for Advanced Models and Translational Sciences and Therapeutics (CAMTraST) at the University of Michigan Medical School. All rabbits were housed in a temperature and humidity-controlled room with a 12-hour light-dark cycle and received standard laboratory animal food with free access to water. The wild-type (WT) New Zealand White rabbits used as controls were 3 months of age and both genders (male and female) matched to the WHHL rabbits. They were sourced from CAMTraST. These controls were included to ensure comparability and eliminate confounding factors related to age, sex, or environmental exposures.

Prior to the procedure, each rabbit was anesthetized with ketamine (40 mg/kg IM, 100 mg/mL) (JHP Pharmaceuticals, Rochester, MI, USA) and xylazine (5 mg/kg IM, 100 mg/mL) (AnaSed^®, Boise, ID, USA) injected intramuscularly (IM). Next, the pupils were dilated by administering tropicamide 1% ophthalmic and phenylephrine hydrochloride 2.5% ophthalmic eye drops. For topical anesthesia, 0.5% topical tetracaine was administered. PBS was provided each minute to minimize corneal dryness. Pre, during, and post procedure, vitals including mucous membrane color, temperature, heart rate, respiratory rate, and oxygen saturation were monitored every 15 minutes with a pulse oximeter (V8400D Capnograph & SpO2 Digital Pulse Oximetry, Smiths Medical, MN, USA) until the rabbits became alert and ambulatory.

Choroidal Neovascularization (CNV) Model:

Choroidal neovascularization (CNV) was induced in the rabbits via subretinal injection of a solution composed of Matrigel (20 μL, Corning, NY, USA) and VEGF (7.5 mL, 100 μg/mL, Shenandoah Biotechnology, Warminster, PA, USA). The solution was created by mixing the Matrigel with VEGF and then kept refrigerated at 4°C. Prior to the injection, the pupils were dilated using tropicamide 1% ophthalmic (Akorn, Decatur, IL, USA) and phenylephrine hydrochloride 2.5% ophthalmic eye drops (Bausch & Lomb Pharmaceuticals, Tampa, FL, USA).

First, the superior rectus muscle was removed with surgical scissors before a 26G needle was inserted to form a scleral tunnel 3.5 mm in distance from the limbus to the posterior segment. A silicone contact lens (Volk Optical Inc., Mentor, OH, USA) was applied to the cornea using Gonak Hypromellose Ophthalmic Demulcent Solution 2.5% (Akorn Inc., Lake Forest, IL, USA), which provided coupling between the incident light and the cornea. A 50 mL Hamilton syringe, fitted with a 30G needle and a blunt tip, was filled with 27.5 μL of the Matrigel and VEGF (M&V) solution and carefully inserted into the scleral tunnel. Using an operating microscope for visualization, the needle tip was gently pushed until reaching the retinal tissue, where the M&V solution was injected into the retinal space with real-time OCT image guidance.

CNV Treatment with Avastin:

The induced CNV was treated with bevacizumab (Bev, Avastin, 25 mg/mL, 50 μL, Fargon Sterile Services, Wichita, KS, USA) via intravitreal injection 3 days after initial subretinal injection. The treatment injection was performed using a 30G ½ inch needle under a microscope at 3.5–4.0 mm from the limbus to the posterior segment of the vitreous cavity.

Imaging of CNV Model:

Prior to the subretinal injection, each New Zealand White WHHL rabbit was imaged with fundus photography, fluorescein angiography (FA), indocyanine green angiography (ICGA), PAM, and OCT. This pre-treatment imaging allowed for a pre-post assessment. After the formation of CNV and subsequent treatment with bevacizumab, each rabbit was followed up for 28 days with each imaging modality for treatment monitoring and evaluation.

In vivo fundus photography, red free (RF), fluorescein angiography (FA), and indocyanine green angiography (ICGA):

A pediatric Barraquer wire speculum was used to hold open the rabbit eyelid during imaging. Fundus imaging of the CNV and retinal blood vessels was acquired with the Topcon 50 EX system (TRC 50 EX, Topcon Corporation, Tokyo, Japan). Red-free (RF) imaging was performed using the same Topcon 50EX imaging system, with adjustments made to the filters to optimize the wavelengths for clear visualization of retinal features such as blood vessels and structural abnormalities in a contrast-rich red-free image. This imaging technique enhances the differentiation of vascular structures from surrounding tissues. For FA imaging, 0.2 mL of fluorescein sodium solution was injected intravenously in the marginal ear vein of the anesthetized rabbit. ICGA imaging was performed by injecting 0.2 mL of ICG dye into the marginal ear vein. To monitor leakage from induced neovascularization, FA and ICGA contrast was imaged for 5 to 15 minutes post-injection. ImageJ software was used to quantify the fluorescence intensity change over time, with hyperfluorescent CNV lesions being determined using the freehand drawing feature. Background intensity noise was calculated as an average of measurements from several images at different timepoints and subtracted from the values obtained from regions of CNV [3].

In vivo multimodal PAM and OCT imaging for CNV:

The dual-modality imaging system consisting of spectral-domain OCT (SD-OCT) and photoacoustic microscopy (PAM)(Tian et al., 2017). OCT is capable of detecting backscattered photons via low-coherence interferometry, and PAM offers high imaging resolution coupled with high depth of penetration. The SD-OCT and PAM systems have aerial lateral resolution of 3.8 μm and 4.1 μm, respectively, with an imaging depth of up to 1.9 mm. Further details of the imaging system have previously been described by Tian et al.(Tian et al., 2017). Modifications including an ocular lens following an additional scan lens, and a dispersion compensation glass (DCG) mounted on the reference arm were made to a commercially available SD-OCT system from Thorlabs (Ganymede-II-HR, Thorlabs, Newton, NJ, USA). The illumination source for PAM was an optical parametric oscillator (OPO) (NT-242, Ekspla, pulse duration of 3–6 ns, wavelength tunable from 405–2600 nm). A custom-built, needle-shaped ultrasonic transducer with a center frequency of 27.0 MHz (Optosonic Inc., Arcadia, CA, USA) in direct contact with PBS on the rabbit conjunctiva detected PA signals produced by endogenous chromophores including hemoglobin and melanin. The PA signal was then digitized using a high-speed digitizer with a sampling rate of 200 MS/s (PX1500–4, Signatec Inc., Newport Beach, CA, USA) and amplified using a low-noise amplifier (gain 57 dB, AU-1647, L3 Narda-MITEQ, NY, USA). The combination of two superluminescent light-emitting diodes with center wavelengths of 845 nm and 932 nm resulted in a 905 nm center excitation wavelength of the OCT system.

For in vivo imaging, the head and body of the anesthetized rabbit were placed on two independent stabilization platforms to reduce motion artifacts, with the body also placed on top of a water-circulation blanket (TP-700, Stryker Corporation, Kalamazoo, MI, USA) to maintain optimal body temperature. The region of interest (ROI) was targeted using the integrated charge-coupled device (CCD) camera during in vivo experiments, and the reference arm was calibrated to enhance image quality. B-scan OCT images were first acquired with a resolution of 512 × 1024 A-lines and an acquisition rate of 36 kHz. Matlab2019b (MathWorks, Natick, MA, USA) and the ultrasonic transducer were used to combine the PAM system with the OCT system for multimodal imaging. Three-dimensional volumetric PAM images were obtained through raster scanning with an optical scanning galvanometer.

Electroretinography (ERG):

Following 60 minutes of full dark adaptation, each New Zealand White WHHL rabbit was anesthetized with ketamine (40 mg/kg IM, 100 mg/ml) (JHP Pharmaceuticals, Rochester, MI, USA) and xylazine (5 mg/kg IM, 100 mg/mL) (AnaSed^®, Boise, ID, USA) injected intramuscularly (IM) under dim red light. Tropicamide 1% ophthalmic and phenylephrine hydrochloride 2.5% ophthalmic eye drops were administered to dilate the pupils. ERG-Jet contact lens electrodes (The Electrode Store, Enumclaw, WA, USA) were applied to the corneas of the left and right eyes of each anesthetized rabbit. Gonak Hypromellose Ophthalmic Demulcent Solution 2.5% (Akorn Inc., Lake Forest, IL, USA) was applied to maintain corneal hydration, and a ground electrode was placed under the skin.

ERG recordings were acquired using the LKC UTAS 3000 electrophysiology system (LKC Technologies, Gaithersburg, MD, USA) in a Ganzfield configuration. ERG recording was carried out using a xenon white flash, 1000 Hz sampling frequency, 0.312 to 300.000 Hz cut-off filter, 500 ms recording time, 10 ms baseline prior to flash, and without a notch filter. ERG responses were amplified at 2500 gain at 0.312 to 500.000 Hz and digitized at a rate of 2000 Hz. Scotopic ERGs were recorded at a dim flash intensity of 0.01 cd.s/m² to obtain the rod-isolated ERG and at 3.0 cd.s/m² to obtain the combined rod-cone ERG. The rabbits were then light adapted to a white 32 cd.s/m² rod-suppressing background for 10 minutes before recording photopic ERGs at a flash intensity of 3.0 cd.s/m².

The UTAS EMWIN software (LKC Technologies, Gaithersburg, MD, USA) was used for quantitative analysis of the ERG recordings. A-wave amplitude was measured from the prestimulus baseline to the trough of the a-wave, and the a-wave implicit time was measured from the flash onset to the trough of the a-wave. B-wave amplitude was measured from the trough of the a-wave to the peak of the b-wave, and the b-wave implicit time was measured from the flash onset to the b-wave peak.

Histological Analysis:

CNV and ocular damage were determined with histological analysis. The rabbits were euthanized 28 days after the treatment with bevacizumab using pentobarbital euthanasia solution (Beuthanasia-D, 0.22 mg/kg, 50 mg/mL, VetOne, ID, USA) administered intravenously at the marginal ear vein. The whole eyes were immersion-fixed with Davidson’s (Hartmann’s) fixative for 24 hours before being rinsed in PBS and dehydrated with a series of graded alcohols. After removal, separation, and dissection of the anterior cap (cornea, iris, and lens), 5 mm samples were embedded in paraffin and sectioned into 4 μm thick sections using Leica Autostainer XL (Leica Biosystems, Nussloch, Germany). These sections were subsequently stained with hematoxylin and eosin (H&E) before being imaged under the Leica DM6000 light microscope (Leica Biosystems, Nussloch, Germany).

In vivo biosafety, TUNEL assay and immunohistochemical analysis:

To assess potential toxicity in animals, we conducted a comprehensive in vivo biosafety analysis encompassing several key methodologies. First, we tracked changes in animal body weight as a vital indicator of physiological response. Each animal underwent pre- and post-treatment weight assessments, with subsequent monitoring over a 28-day period. Body weight data were meticulously analyzed and graphed using Origin software (Origin 9.0, OriginLab, MA, USA) to discern any discernible trends or anomalies. Additionally, to investigate tissue-level effects, standard hematoxylin and eosin (H&E) staining along with Terminal deoxynucleotidyl transferase dUTP Nick End Labeling (TUNEL) assay were employed. At the 28-day mark post-treatment, animals were humanely euthanized, and their eyes were carefully excised for analysis. The harvested tissue was promptly fixed in Davidson’s fixative solution (Fishersci, USA) for 24 hours, followed by immersion in 50% alcohol for 8 hours and subsequent transfer to 70% alcohol for an additional 24 hours before the sectioning process commenced.

For histological analysis, tissue samples were prepared by halving at the center of the optic nerves and embedding in paraffin. Subsequent sections, with a thickness ranging from 6 to 8 μm, were obtained using precision microtome equipment. These sections were then subjected to H&E staining to visualize tissue morphology. Microscopic examination of the stained slides was conducted utilizing a high-resolution DM6000 Leica Microscope (DM6000, Leica Biosystems, Nussloch, Germany), with digital images captured using an integrated DF450C camera. This rigorous methodology allowed for detailed evaluation of any histopathological alterations or cellular apoptosis within the ocular tissues.

To confirm the presence of choroidal neovascularization (CNV) lesions, we conducted immunohistochemistry (IHC) using two distinct antibodies. An anti-alpha smooth muscle actin (α-SMA) antibody (Abcam, Burlingame, CA, USA) was utilized to stain smooth muscle cells in the vessel walls. CD31 (PECAM-1) is an endothelial cell marker that helps visualize the formation of new blood vessels. This staining provides clear visualization of endothelial structures, enabling better analysis of vascular changes associated with AMD. The stained samples were carefully examined using a Leica DM6000 microscope (Leica Biosystems, Nussloch, Germany). High-resolution digital images of the IHC-stained tissues were captured using a DF450C camera integrated with the Leica Application Suite software (LAS X, Leica Biosystems, Nussloch, Germany). This comprehensive approach enabled detailed visualization and confirmation of CNV lesions at the cellular level.

Statistical Analysis:

Statistical evaluation was performed using a one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test to determine significant differences between treatment groups. The analysis was conducted using Origin software (OriginLab Corporation, MA, USA). Data are presented as mean ± standard deviation, with statistical significance defined as p < 0.05.

Results

Comprehensive ophthalmic evaluation:

A comprehensive ophthalmic evaluation of the anterior segments was performed on both WHHL and control (WT) groups using a slit-lamp (Fig. S1). Figure 1 shows the fundus images of the anterior segments captured at various time points over the period of 1 year. The control group exhibited normal ocular structures, including the cornea, limbus, and conjunctiva (Figure 1a). Similarly, the WHHL group displayed normal anterior segments from months 3 to 5. However, in the WHHL rabbits, a significant lipid layer originating from the peripheral cornea and limbus and progressively involving the more central cornea was observed (Figure 1b). This abnormal lipid layer appeared at month 7 and persisted through 12 months. No evidence of vitreous or lens opacity was found in any of the groups.

Figure 1. Ophthalmic examination:

(a) Wild-type (WT) group. (b) WHHL group. Anterior segment examinations were conducted at various time points, starting from month 3 and continuing through month 12. The WT group displayed normal anterior segment structures, including the limbus and cornea. In the WHHL group, no abnormalities were observed at months 3 and 5. However, beginning at month 7, a lipid layer (indicated by red arrows) was detected, which persisted and extended through month 12.

Color fundus photography, red free (RF), fundus autofluorescence (FAF), fluorescein angiography (FA), and indocyanine green angiography (ICGA) imaging:

The anatomy of the posterior segments was monitored using various imaging modalities, including color fundus photography, red-free (RF) imaging, fundus autofluorescence (FAF), fluorescein angiography (FA), and indocyanine green angiography (ICGA) (Figure 2 and Supplementary Fig. S2). The structure and morphology of the retinal vessels (RVs), choroidal vessels (CVs), optic nerves, and capillaries were clearly visualized and found to be comparable between the WT and WHHL groups, as shown in Figures 2a, 2b, 2d, and 2e. No signs of retinal vein or artery tortuosity were observed, and no leakage was detected on FA (Fig. 2d) and ICGA (Fig. 2e). These findings confirmed that WHHL rabbits have normal vascular morphology. However, several hyperfluorescent transmission areas were detected on FAF images starting at 3 months and persisting up to 12 months in the WHHL group (Figure 2c). FAF images from the control (WT) group demonstrate a normal retinal fluorescence pattern, with no signs of abnormal or hyperfluorescent transmission signals. These images indicate a healthy retinal structure, as no irregularities or unusual fluorescence are observed, suggesting the absence of retinal damage or pathological changes (Supplementary Fig. S2c). By quantifying the size of retinal vessels at different locations along the retinal blood vessels and optic disc (OD) (Supplementary Fig. S3), we found that the vessel diameter (VD) of retina vessels measured from WHHL rabbits was thinner than that of the one measured from WT rabbits (VD_Artery = 180.65±16.53 (μm) for WT vs 164.39±18.89 (μm) for WHHL respectively; p<0.01). However, the optic discs were not significantly different between these groups (OD = 11,046.49±1141.18 (μm) for WT vs 10,882.47±416.78 (μm) for WHHL, respectively; p>0.05).

Figure 2. Retinal imaging of WHHL rabbits over time:

(a) Color fundus photography showing normal retinal structures such as retinal vessels (RVs), choroidal vessels (CVs), and the nerve fiber layer (NFL), with no visible abnormalities. (b) Red-free imaging enhancing the visualization of retinal vessels and nerve fibers. (c) Fundus autofluorescence (FAF) highlighting hyperfluorescent transmission areas (white arrows), first observed at month 3 and persisting through month 12, indicating possible retinal changes. (d) Fluorescein angiography (FA) showing detailed vasculature with no significant leakage. (e) Indocyanine green angiography (ICGA) highlighting the deeper choroidal vessels. Imaging was performed at different time points from months 3 to 12, with the FAF revealing the earliest signs of retinal alterations.

Multimodal PAM and OCT visualization of retinal structure:

Figure 3 and Figure S4 show the OCT images obtained from WHHL and WT rabbits. The OCT images demonstrate different retinal structures: inner limiting membrane (ILM), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor layer (PL), retinal pigment epithelium (RPE), and choroidal layer (CL). Importantly, we found strong hyper-reflective signal from several portions in the sclera layer beneath the CL. We found that WHHL rabbits exhibit several hyperreflective areas at the retinal pigment epithelium (RPE) layer on OCT images, possibly due to underlying lipid metabolism abnormalities and melanin pigmentary changes (Figs. 3a–3f). In contrast, wild-type (WT) rabbits have normal lipid metabolism, which prevents such changes, leading to a lack of hyperreflective areas in their OCT images (Figs. S4a–S4f). Figures 3g–3i illustrate 3D volumetric OCT images. These images show the entire structure of the retina and nerve fiber layer. Additionally, using ImageJ to quantify OCT thickness, the distance between the retina and choroidal vessels was measured to be 254 μm. The total retinal thickness in WHHL rabbits was found to be 10% thinner compared to WT rabbits (Thickness = 230.40±11.03 μm for WHHL vs. 254.16±11.29 μm for WT, p<0.05). Moreover, the outer nuclear layer (ONL) thickness in WHHL rabbits was reduced by 7% compared to WT (ONL = 68.04±4.94 μm for WHHL vs. 72.99±4.24 μm for WT, p<0.05).

Figure 3. OCT imaging of WHHL rabbits:

(a–f) 2D OCT images obtained at various time points from months 3 to 12, illustrating distinct retinal layers including the inner limiting membrane (ILM), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor layer, retinal pigment epithelium (RPE), choroid layer, and sclera. Hyperreflective zones were observed at the ellipsoid zone (EZ) starting from month 3 (red arrows) and persisted throughout the 12-month period, indicating potential photoreceptor disruption. High magnification (ai-fi) of the hyperreflective zones, isolated within the dotted red rectangles shown in panels a–f, clearly reveals the areas of photoreceptor and RPE degeneration. (g–i) Selected 3D OCT images show the overall retinal structure and CNV progression at 3 months (g), 6 months (h), and 12 months (i), providing comprehensive visualization of retinal thickness and architecture over time.

Figure 4 presents the PAM images captured at an excitation wavelength of 578 nm, selected due to hemoglobin’s strong optical absorption at this wavelength. Before acquiring the PAM images, the structure of retinal vasculature was examined using fundus photography (Fig. 4a), fluorescein angiography (FA, Fig. 4b), and indocyanine green angiography (ICGA, Fig. 4c). These images display normal retinal vasculature (RVs), choroidal vasculature (CVs), nerve fiber layer (NFL), and optic nerve. Figures 4d–4f show the corresponding PAM images obtained across the scanning area highlighted in Figure 4a. The morphology of RVs, CVs, and capillaries is clearly visible with high contrast and resolution. The 3D PAM images illustrate the entire retinal structure, with RVs positioned at the top of the en face PAM image, consistent with the OCT images (Figs. 3g–4i).

Figure 4. In vivo PAM imaging of WHHL rabbits:

(a) Color fundus photography showing normal retinal anatomy including retinal vessels (RVs), the optic nerve, and the nerve fiber layer (NFL), (b) Fluorescein angiography (FA) highlighting retinal vasculature, and (c) Indocyanine green angiography (ICGA) showing deeper choroidal vessels (CVs). These images confirm the normal structure of the retina. (d–f) Corresponding maximum intensity projection (MIP) PAM images obtained from the scanning area (white dotted rectangle in (a) provide high-resolution, high-contrast visualization of RVs and CVs. PAM imaging demonstrates clear delineation of the vasculature, offering superior contrast compared to traditional methods. (g–i) 3D volumetric renderings of the PAM data show the complete structure of retinal vessels from a 3D perspective, allowing a comprehensive view of the retinal vascular architecture.

Electroretinography (ERG) Analysis:

To assess retinal function in the WT and WHHL rabbits, we used full-field electroretinography (ff-ERG), a non-invasive technique that measures the electrical response of the retina to light stimulus. Figure 5a shows scotopic ERG waveforms at 0.01 cd.s/m² excitation for both WT and WHHL rabbits. The results did not reveal significant differences between the WT and WHHL groups. The average signal amplitudes (Fig. 5ai) and the implicit time of the b-wave (Fig. 5aii) displayed slight variations across all responses in both groups at 6 and 12 months, indicating no evidence of retinal functional impairment in WHHL rabbits. Similarly, under the scotopic 3.0 cd.s/m² condition, the average signal amplitudes and b-wave implicit times in the WHHL group exhibited only minor fluctuations when compared to the WT group (Fig. 5b, Fig. 5bi, Fig. 5bii). Although no significant differences were observed in b-wave amplitudes between the two groups, there was a noticeable difference in implicit times. The implicit times in the WT group were consistently shorter than those in the WHHL group at both 6 and 12 months, with the mean implicit time difference estimated at 28% at 6 months (Time = 57.67±3.62 ms for WHHL vs. 45.17±0.29 ms for WT, p<0.01) and 25% at 12 months (Time = 56.17±3.75 ms for WHHL vs. 45.00±0.50 ms for WT, p<0.01). The longer implicit times in WHHL rabbits suggest that while the overall strength of the retinal response (as measured by amplitude) remains similar, the speed of the response is slower. This delay in implicit time could be indicative of subtle functional impairments within the rod-mediated pathways of WHHL rabbits, even in the absence of overt structural damage or degeneration.

Figure 5. Comparison of ERG data between WT and WHHL rabbits:

(a) Electroretinograms (ERGs) showing rod-isolated ERG waveforms. The waveforms reflect the electrical response of the retina to scotopic (low light) stimulation, with no significant differences observed between the two groups. (ai) Graph showing the mean ERG b-wave amplitudes for WT and WHHL rabbits at 6 and 12 months, indicating minimal fluctuations in amplitude that are not statistically significant (N=3). (aii) The average rod implicit times are shown, demonstrating that there is no significant change in response time between the two groups (N=3). (b) Combination of rod and cone ERG waveforms, assessing both photoreceptor types simultaneously. (bi) Graph of mean b-wave amplitudes showing that amplitude fluctuations are minor and not statistically significant across both groups (N=3). (bii) Mean rod and cone implicit times, with slight variation noted between WT and WHHL rabbits (N=3). (c) ERG waveforms of cone responses under photopic (light-adapted) conditions, assessing cone function specifically. (ci) Quantification of b-wave amplitudes shows minimal differences between WT and WHHL groups (N=3). (cii) Mean implicit times of cone responses, demonstrating no significant changes. (d) 32 Hz flicker ERG waveforms, used to evaluate high-frequency cone response. (di) Quantification of b-wave amplitudes shows comparable responses in WT and WHHL rabbits (N=3). (dii) The implicit time of the flicker response is also shown, with no significant variation observed (N=3). Data are presented as mean ± standard deviation.

We also performed light-adapted ERG to assess cone cell function. Under light excitation, there were no significant differences in b-wave amplitudes or mean implicit times between the WT and WHHL groups at 6 and 12 months (Figs. 5c, 5ci, 5cii). The lack of significant differences in b-wave amplitude and implicit time between WT and WHHL rabbits under photopic 3.0 cd.s/m² conditions suggests that cone-mediated responses remain intact in WHHL rabbits, indicating minimal impact of hyperlipidemia on cone function over time. Unlike rod cells, which are more prone to changes from metabolic stress, cone cells—responsible for photopic (daylight) vision—may be less susceptible to the effects of hyperlipidemia seen in WHHL rabbits.

Additionally, since WHHL rabbits primarily model cardiovascular conditions such as atherosclerosis and lipid metabolism disorders, these factors might impact the rod cells and retinal vasculature more than the cone cells. This could explain why under scotopic (low-light) conditions, some differences were observed, but under photopic conditions (which predominantly test cone function), there is little to no difference between WT and WHHL rabbits.

Furthermore, when subjected to 32 Hz flicker stimulation, which specifically evaluates the temporal response and adaptability of cone cells, both groups displayed similar amplitudes and mean implicit times (Figs. 5d, 5di, 5dii). This result highlights that the temporal processing of visual stimuli, mediated by cone pathways, remains stable in WHHL rabbits, even at higher stimulation frequencies. The preservation of flicker responses implies that despite potential systemic and retinal vascular changes associated with hyperlipidemia in WHHL rabbits, cone-driven retinal circuits continue to function normally. This finding is important because 32 Hz flicker ERG is highly sensitive to cone dysfunction and can detect early changes in diseases that primarily affect cone activity. The lack of significant differences between the groups suggests that cone photoreceptor function, and their ability to respond to rapidly flickering stimuli, is not significantly compromised by the metabolic or vascular conditions present in WHHL rabbits. Thus, retinal degeneration or dysfunction in WHHL rabbits might predominantly affect rod-mediated processes, while cone function remains relatively unaffected over time.

Anti-VEGF Response Evaluation:

To investigate the potential of anti-VEGF therapy for treating choroidal neovascularization (CNV) in WHHL rabbits, we conducted CNV treatments in WHHL rabbits. The CNV models were generated via subretinal injections of VEGF-165 mixed with Matrigel solution(Li et al., 2019; Nguyen et al., 2021c; Nguyen et al., 2022). Three days post-injection, bevacizumab (Bev), an anti-VEGF agent, was intravitreally injected into the rabbits’ vitreous cavity. The treatment process was monitored using various imaging modalities, including color fundus photography, fluorescein angiography (FA), PAM, and OCT. Figure 6 illustrates the color fundus images (Fig. 6a) and FA images at different stages: early phase (Fig. 6b), middle phase (Fig. 6c), and late phase (Fig. 6d) of CNV, captured before and after Bev treatment at different time points: days 3, 7, 14, 21, and 28. The color fundus images provide clear visualization of the retinal vessels (RVs), choroidal vessels (CVs), and optic nerve, revealing minor structural changes around the site of subretinal injection. These changes may be attributed to the side effects of Matrigel or potential inflammation, as previously described by Qiu et al.(Qiu et al., 2006). Additionally, mild hemorrhage was observed before treatment, persisting for up to 3 days post-treatment, but resolved by one week.

Figure 6. Multimodal imaging monitoring anti-VEGF treatment in WHHL rabbits:

(a) Color fundus photography obtained before and after treatment at various time points (3, 7, 14, and 28 days), showing the morphology of retinal vessels (RVs), optic nerve, and the location of choroidal neovascularization (CNV). The CNV lesion, induced by subretinal injection of VEGF-165 with Matrigel, is visible near the optic nerve. (b–d) Fluorescein angiography (FA) images acquired at different phases: Early phase (b), middle phase (c), and late phase (d). Red arrows indicate sites of leakage, confirming the development of CNV. FA reveals leakage from the CNV at each phase, with the leakage slightly decreasing after treatment with bevacizumab (Bev). Notably, while the CNV lesion reduced in size slightly, the leakage and CNV persisted up to 28 days post-treatment, indicating only a modest response to anti-VEGF therapy. This suggests that Bev treatment inhibited CNV progression to a limited extent in WHHL rabbits. (e) Image segmentation used to quantify fluorescent leakage intensity (FLI). (f) A graph of normalized fluorescent intensity (FLI) as a function of treatment time (N=3) obtained from 3 different groups: WHHL rabbits treated with Bev, young WT rabbits treated with Bev, and control group (WT rabbits untreated with Bev). Note that the data from the young WT and control groups were adapted from our previous publication (Nguyen et al., 2022). (g) A panel showing normalized vessel density (VD) from the same three groups (N = 3 per group). Data presents as mean±standard deviation, *p<0.05.

FA images provide detailed insights into the retinal and choroidal vasculature as well as the progression of newly formed CNV. Notably, leakage was consistently observed in all FA images, confirming the presence of active CNV. While the bevacizumab treatment showed some effect, it only slightly inhibited CNV growth, as evidenced by the persistent leakage across the treatment period. Using image segmentation (Fig. 6e), we quantified the fluorescent leakage intensity (FLI). Initially, the FLI showed only a slight change, decreasing by about 11–15% between days 3 and 7 after treatment. Following this period, the FLI exhibited a more significant rise, which remained elevated and sustained up to 28 days (Fig. 6f). Additionally, the mean CNV density showed no significant variation compared to the pretreatment levels (Fig. 6g), with the CNV density remaining above 84% at 28 days post-treatment (VD = 1.00±0.00 pretreatment vs. 0.84±0.11 (a.u.) post treatment at days 28). This limited reduction in leakage intensity suggests that Bev was not highly effective in completely suppressing the vascular activity associated with CNV in the WHHL model. In contrast, we have published that CNV rapidly regresses in this same subretinal injection of Matrigel + VEGF model in wild-type young animals(Nguyen et al., 2022). The persistent leakage, as observed through fluorescein angiography, is indicative of ongoing active neovascularization, which points to only partial inhibition of VEGF activity. Moreover, the relatively stable CNV density after treatment, remaining at more than 84% of the original density, implies that the Bev treatment did not significantly shrink the size or reduce the extent of the CNV lesions. This highlights the challenge of treating CNV in WHHL rabbits, a model known for its lipid dysregulation, which may influence inflammation, macrophage activation, and the vascular response to anti-VEGF therapies.

The combination of color fundus and FA imaging proves invaluable in tracking the progression of CNV and the response to Bev. The color images highlight the overall retinal morphology, including the vascular and optic nerve structures, while the FA images allow for a more precise evaluation of CNV activity, especially in terms of leakage and new vessel formation. Despite the treatment’s mild efficacy, the findings suggest that anti-VEGF therapy like Bev may not fully halt CNV progression in WHHL rabbits, indicating the need for further exploration of alternative or combination therapies to achieve better outcomes.

To further assess the structural changes associated with newly developed CNV, we performed OCT imaging at various time points. Figure 7 presents selected B-scan OCT images captured before treatment (Fig. 7a) and post-treatment on days 7 (Fig. 7b), 14 (Fig. 7c), and 28 (Fig. 7d). These images clearly outline the location and margins of the CNV lesions, indicated by yellow dotted lines. Additionally, Figures 7e–7h show the corresponding 3D OCT reconstructions, which reveal the entire retinal structure, including the retinal vessels and nerve fiber layer. However, detecting CNV lesions in the 3D OCT images proved challenging due to their subtle appearance in this modality. Quantification of the CNV lesion thickness using OCT data revealed a reduction of approximately 18% after treatment, which persisted up to 28 days compared to the pretreatment state (thickness = 72.72±5.72 μm pretreatment vs. 59.35±3.05 μm post-treatment at day 28). OCT imaging provided critical insights into the structural response of the retina to Bev treatment. The B-scan images were particularly useful in clearly identifying the CNV boundaries, which is essential for monitoring lesion size and evaluating the therapeutic effect. While the 3D OCT images offer a comprehensive view of the retinal architecture, they were less effective at highlighting the CNV lesions, possibly due to the diffuse nature of the vascular changes associated with CNV. The reduction in CNV lesion thickness suggests that the anti-VEGF therapy had some effect in limiting the structural growth of CNV, but the persistence of lesions even after 28 days indicates that the treatment was only partially effective. This highlights the need for more precise treatment strategies or additional therapeutic interventions to achieve full regression of CNV in this model. OCT’s ability to track changes in lesion thickness over time provides a valuable tool for longitudinal studies of CNV and for assessing the efficacy of different therapeutic approaches in preclinical models like WHHL rabbits.

Figure 7. OCT imaging of anti-VEGF treatment in WHHL rabbits:

(a) B-scan OCT images obtained before treatment, showing the location and margin of choroidal neovascularization (CNV) outlined by the yellow dotted line. The CNV lesion is clearly visible, indicating its extent prior to treatment. (b–d) Selected 2D OCT images captured at days 7 (b), 14 (c), and 28 (d) post-bevacizumab (Bev) treatment. Although the yellow dotted lines mark the CNV margins, there is no significant reduction in CNV thickness over time, suggesting limited efficacy of the treatment. (e–h) Enface 3D OCT images obtained pre-treatment (e) and at days 7 (f), 14 (g), and 28 (h). These 3D OCT renderings clearly depict the structure of retinal vessels (RVs) and the nerve fiber layer (NFL). However, the CNV is challenging to distinguish in the 3D images, highlighting the limitations of 3D OCT in detecting fine details of CNV lesions compared to 2D B-scan OCT.

Although both FA and OCT imaging provide valuable insights into CNV progression, distinguishing CNV from surrounding microvasculature remains challenging. To overcome this, we employed PAM to better visualize the margins and location of CNV while differentiating them from native blood vessels. PAM’s unique advantage lies in its ability to enhance sensitivity and image contrast through the use of contrast agents. By using different agents, we can adjust the excitation wavelength based on their optical absorption properties, where hemoglobin has minimal absorption, thus allowing for clearer visualization of target tissues. In this study, we utilized indocyanine green (ICG), an FDA-approved contrast agent commonly used in ophthalmology. PAM images were captured using two wavelengths: 578 nm and 780 nm. At 578 nm, hemoglobin within retinal vessels strongly absorbed the incident light, providing high-contrast images of the entire retinal structure, including retinal vessels (RVs), choroidal vessels (CVs), and capillaries (Figs. 8a–8c). However, at 780 nm, hemoglobin’s absorption was minimal, resulting in weak PAM signals (Supplementary Fig. S5).

Figure 8. In vivo PAM imaging monitoring CNV progression after treatment with anti-VEGF therapy:

(a–c) Maximum intensity projection (MIP) PAM images obtained at 578 nm at different time points: Day 7 (a), Day 14 (b), and Day 28 (c) post-treatment. These images provide a high-resolution view of retinal vessels (RVs) and choroidal vessels (CVs), clearly illustrating the vascular architecture and any changes associated with CNV progression over time. (d–f) PAM images obtained at 780 nm following intravenous injection of indocyanine green (ICG) at a dose of 0.5 mL and a mass concentration of 2.5 mg/mL. The strong PAM signals observed in these images confirm the location of the CNV, while minimal signals from the surrounding blood vessels indicate low intrinsic absorption of hemoglobin at this wavelength. This contrast enhances the visualization of CNV relative to the surrounding vasculature. (g–l) Overlay 3D PAM images demonstrate the spatial relationship between CNV and the surrounding retinal and choroidal vessels. The pseudo-green color highlights the CNV location, situated between the retinal and choroidal vessels, effectively illustrating the extent and orientation of the neovascularization in relation to normal retinal anatomy. This multi-wavelength PAM imaging approach allows for detailed monitoring of CNV progression and response to anti-VEGF treatment.

To further enhance contrast, we performed intravenous injections of ICG dye at a dose of 0.5 mL and a concentration of 2.5 mg/mL. PAM images were obtained 15 minutes post-injection to minimize interference from free dye in the bloodstream. The resulting PAM signals were significantly stronger at days 7, 14, and 28 post-treatments (Figs. 8d–8f). When compared to pre-injection images, the PAM signal intensity increased by 18-fold (PAM_Signal = 5.57±0.83 a.u. pre-injection vs. 99.04±3.63 a.u. post-injection at day 14, N=3). Notably, the PAM signal at day 7 was approximately 3% lower than that at day 14, likely due to CNV regression caused by the anti-VEGF treatment.

Overlaying the 3D PAM images allowed us to clearly distinguish CNV from the surrounding retinal vasculature (Figs. 8g–8i), a significant improvement over traditional imaging modalities like FA and OCT. The enhanced contrast achieved through PAM, especially with ICG, provided a detailed view of the boundaries of the CNV, enabling more accurate assessment of lesion size and treatment efficacy. This demonstrates PAM’s potential as a complementary imaging tool for studying CNV and assessing treatment outcomes. PAM with contrast agents like ICG offers the ability to distinctly visualize and quantify CNV, making it an invaluable asset for preclinical CNV research and potentially clinical applications. Moreover, PAM images achieved from different wavelengths and contrast agents further enhances its versatility for monitoring retinal and choroidal diseases.

Histological Analysis:

To compare the retinal architecture of the WT and WHHL rabbits, we performed histological analysis using standard H&E staining. Figure 9a shows the H&E image of the WT group, where the cellular morphology and retinal layers appear normal. Each layer including inner limiting membrane (ILM), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor layer (PL), retinal pigment epithelial layer (RPE), choroid layer (CL), and sclera was clearly visible without any abnormalities. Similarly, in the WHHL group (Fig. 9b), the overall retinal architecture and cell morphology appeared normal, but we observed a noticeable reduction in the density of ONL cells compared to the WT group. In the WHHL group treated with Bev (Fig. 9c), there were subtle changes in the photoreceptor and choroid layers, including the formation of CNV, which is characteristic of the disease progression.

Figure 9. In vivo biosafety and histological analysis of the retina of rabbits:

(a) A panel displaying body weight measurements obtained from three different groups: wild-type (WT), WHHL, and WHHL rabbits treated with bevacizumab (WHHL-Bev). This analysis provides insight into the overall health and wellbeing of the rabbits across different treatment groups. (b) H&E staining of retinal sections from the WT group, illustrating normal cellular morphology and well-preserved retinal architecture, including distinct layers such as the inner limiting membrane (ILM), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor layer (PL), retinal pigment epithelial layer (RPE), choroidal layer (CL), and sclera. (c) H&E image of WHHL rabbits, showing similar retinal architecture to that of the WT group, although some variations may be observed. These images clearly delineate the layers of the retina, indicating that the structural integrity is generally maintained in WHHL rabbits. (d) H&E image of choroidal neovascularization (CNV) model in WHHL rabbits obtained at day 28 post treatment with Bev, with black arrows indicating the location of CNV. This image reveals slight alterations in the photoreceptor (PL) and retinal pigment epithelial (RPE) layers, likely resulting from the subretinal injection of VEGF and Matrigel. These findings underscore the histological effects of the treatment while highlighting the need for ongoing monitoring of retinal integrity in the context of anti-VEGF therapy.

To further confirm the presence of CNV, we performed immunofluorescence analysis using antibodies against α-SMA (alpha-smooth muscle actin) and CD31, both markers commonly associated with vascular structures. Figure 10 illustrates these results, where we detected the strongest α-SMA-positive signal in the WHHL-Bev group (Fig. 10c), indicating the presence of smooth muscle cells, which are commonly associated with neovascular structures in CNV. In contrast, the WT group (Fig. 10a) and untreated WHHL group (Fig. 10b) showed significantly lower α-SMA signals, indicating a lack of extensive neovascularization.

Figure 10. Immunofluorescence analysis of retinal sections:

(a) Immunofluorescence image of the wild-type (WT) group, showing minimal α-SMA (alpha-smooth muscle actin, pseudo red color) and CD31 (pseudo green color) expression, indicative of low levels of neovascularization. (b) Immunofluorescence image of the WHHL group, which also exhibits reduced α-SMA and CD31 signals compared to the treated group. This suggests that, despite the presence of choroidal neovascularization (CNV), the overall vascular development remains limited in the WHHL rabbits. (c) Immunofluorescence image of the WHHL group treated with bevacizumab (Bev), demonstrating a significant increase in α-SMA and CD31 signals, confirming enhanced progression of CNV (indicated by white arrows). The elevated expression levels of these markers in the treated group indicate the presence of newly formed blood vessels, highlighting the effect of anti-VEGF therapy on CNV dynamics. This analysis underscores the importance of these markers in visualizing and quantifying neovascularization within the retina, providing valuable insights into the response of WHHL rabbits to anti-VEGF treatment.

In addition to α-SMA, CD31, a marker for endothelial cells, was used to further assess neovascularization development. No significant CD31-positive signal was detected in either the WT or untreated WHHL groups, suggesting the absence of abnormal blood vessel formation. However, in the WHHL group treated with Bev, we found strong CD31 signals (Fig. 10c), providing clear evidence of CNV development in this group. These results support the conclusion that while WHHL rabbits exhibit some retinal changes without treatment, the Bev-treated group shows more pronounced CNV growth, as confirmed by both α-SMA and CD31 markers.

The combined H&E and immunofluorescence data offer a comprehensive view of the retinal structural changes in WHHL rabbits. While H&E staining allows for detailed examination of retinal layers and cell morphology, immunofluorescence adds a deeper understanding of the molecular markers associated with CNV progression. The strong α-SMA and CD31 signals in the Bev-treated WHHL group highlight the role of neovascularization in the disease, further confirming the formation of CNV. This suggests that while anti-VEGF treatment might limit some aspects of CNV progression, significant vascular changes still occur, highlighting the complexity of treating CNV in this model.

TUNEL Assay and Immunohistochemical Analysis:

The TUNEL (Terminal deoxynucleotidyl transferase dUTP nick end labeling) assay is a vital technique for detecting DNA fragmentation that results from apoptotic signaling cascades. This method was employed to evaluate the efficacy of Bev treatment on CNV in WHHL rabbits. The TUNEL assay was performed on ocular tissue samples post-treatment to quantify apoptotic cell death within the neovascular membranes. The results shown in Fig. 11c indicated an insignificant increase in TUNEL-positive cells in the Bev-treated group compared to WT (Fig. 11a) and WHHL group (Fig. 11b), suggesting enhanced apoptosis in the neovascular regions. This apoptotic effect is consistent with the response to Bev treatment, which involves the inhibition of VEGF-mediated survival signals in endothelial cells, leading to regression of pathological neovascularization. Therefore, the TUNEL assay provided critical insights into the cellular responses to Bev therapy in this model, highlighting its potential for effectively managing CNV.

Figure 11. TUNEL staining analysis for apoptosis detection in retinal sections:

(a) TUNEL images of the wild-type (WT) group, demonstrating no evidence of apoptotic cells, indicating healthy retinal cell integrity in this group. (b) TUNEL images of the WHHL group, revealing a noticeable presence of TUNEL-positive cells, which suggests increased apoptosis in the subretinal space compared to the WT group. This indicates potential retinal stress or degeneration associated with the WHHL phenotype. (c) TUNEL images of the WHHL group treated with bevacizumab (Bev), showing minimal apoptotic cells in the subretinal space. This suggests that anti-VEGF treatment may have a protective effect on retinal cells, reducing the level of apoptosis observed in untreated WHHL rabbits. The green, fluorescent color represents TUNEL-positive cells, indicating sites of DNA fragmentation characteristic of apoptosis, while the blue, fluorescent color shows cell nuclei stained with DAPI, serving as a counterstain to visualize overall cell density and morphology. Scale bar = 25 μm. This analysis provides crucial insights into the effects of CNV and anti-VEGF treatment on retinal cell viability.

Discussions

In this study, we utilized the WHHL rabbit model, known for its genetic predisposition to hyperlipidemia, to investigate lipid-related retinal abnormalities and evaluate the effects of anti-VEGF treatment on choroidal neovascularization (CNV). Using a combination of imaging techniques, including color fundus photography, fluorescein angiography (FA), and indocyanine green angiography (ICGA), PAM and OCT imaging, we were able to track the progression of retinal changes and neovascular growth in WHHL rabbits over time.

Choroidal neovascularization (CNV), a hallmark of wet AMD and the leading cause of vision loss with AMD, was induced in WHHL rabbits via subretinal injection of VEGF-165 mixed with Matrigel. Bevacizumab (Bev), a commonly used anti-VEGF agent, was administered to evaluate its efficacy in inhibiting CNV progression. In young wild-type rabbits, Bev is very effective at causing regression of CNV, with 90.13 % regression by 28 days post treatment(Nguyen et al., 2022) In contrast, in WHHL rabbits, Bev treatment showed much less efficacy in causing CNV regression with 18 % regression by 28 days, as evidenced by a decrease in CNV lesion thickness on optical coherence tomography (OCT) images. FA and ICGA revealed that leakage and neovascular activity persisted. Quantitative analysis confirmed that CNV density remained above 80% post-treatment, suggesting that Bev only moderately inhibited CNV development in this model. While 14 month old rabbits have also been shown to have reduced response to Bev treatment(Nguyen et al., 2022), this model is both time and financially challenging to age the rabbits. For WHHL rabbits, 2–4 months of age rabbits can be utilized to achieve the same model of CNV persistent disease activity. Persistent disease activity is a common problem facing patients with AMD and can affect up to 50% of patients with current anti-VEGF therapies(Jaffe et al., 2017; Rosenfeld et al., 2011). However, current animal models of CNV often fail to recapitulate this persistent disease activity with anti-VEGF treatment, which serves as a major barrier to the development of novel therapies to treat persistent disease activity. This WHHL animal model could serve as a rapid, cost-effective model for the development of novel therapeutics for AMD persistent disease activity.

These findings highlight the association of lipid dysregulation, inflammation, and macrophage activation in retinal pathologies. WHHL rabbits, due to their hyperlipidemic background, may exhibit an altered response to anti-VEGF therapy compared to typical CNV models. The persistent neovascular activity post-treatment suggests that lipid-related factors may interfere with the efficacy of anti-VEGF therapies, necessitating alternative or combination treatment strategies to fully suppress CNV in hyperlipidemic conditions. This could include therapies targeting additional angiogenic, lipid, or inflammation pathways beyond VEGF.

Another key finding in this study was the development of a lipid layer in the cornea of WHHL rabbits, beginning around month 7 and progressively enlarging into the cornea. This observation aligns with the systemic lipid dysregulation characteristic of the WHHL model and reflects the potential for lipid deposition in ocular tissues, a phenomenon observed in human patients with hyperlipidemia. These changes were not seen in wild-type (WT) controls, highlighting the unique susceptibility of WHHL rabbits to lipid-related ocular changes.

Histological analysis revealed that WHHL rabbits exhibited reduced outer nuclear layer (ONL) cell density compared to WT rabbits, indicating early signs of retinal degeneration. The ONL, which contains photoreceptor cell bodies, is crucial for visual function, and its thinning suggests that hyperlipidemia may accelerate photoreceptor loss. This finding adds to the growing body of evidence linking lipid metabolism disorders to retinal degeneration, such as age-related macular degeneration (AMD), where lipid deposits (drusen) are implicated in disease progression.

The use of multiple imaging techniques provided comprehensive insights into both retinal structure and CNV progression. Color fundus photography was effective in visualizing the overall morphology of retinal vessels, the optic nerve, and areas affected by lipid deposition. FA and ICGA were critical in identifying CNV leakage and determining the extent of vascular changes. However, we found that these imaging modalities had limitations in differentiating CNV from surrounding microvasculature, highlighting the need for more specific imaging techniques.

In this regard, photoacoustic microscopy (PAM) proved to be a valuable complementary tool. PAM’s ability to visualize CNV and distinguish it from native vasculature, particularly when combined with contrast agents like ICG, offered a more detailed view of CNV development and regression. The use of different excitation wavelengths in PAM allowed us to improve contrast and sensitivity, providing clearer delineation of CNV structures. This suggests that PAM, in conjunction with traditional imaging modalities, could enhance the monitoring and evaluation of CNV.

Despite the novel findings, there are limitations to this study that could be addressed in future research. First, while the WHHL rabbit model effectively mimics lipid dysregulation, it may not fully represent all aspects of complex human retinal diseases such as AMD. Further studies incorporating other models or extending the treatment duration may help in understanding the long-term effects of anti-VEGF therapy in hyperlipidemic conditions. Additionally, while bevacizumab is an established anti-VEGF agent, exploring other treatments or combination therapies, such as anti-inflammatory or lipid-lowering agents, may yield better outcomes in inhibiting CNV. One limitation of using rabbits as a model for AMD is the absence of a macula, a specialized retinal region responsible for central vision in humans. The macula’s unique structure and function make it a critical focus in the study of AMD, and its absence in rabbits limits the direct translatability of findings to human disease. Despite this limitation, rabbits are widely used in preclinical research due to their relatively large eyes (approximately 2/3 the size of human eye(Bozkir et al., 1997; Hughes, 1972; Nguyen et al., 2019), ease of handling, and well-characterized retinal physiology. They are particularly valuable for evaluating ocular drug delivery systems, retinal pharmacokinetics, and localized therapeutic effects. However, the lack of a macula necessitates caution when extrapolating results to the human condition. Future studies will need to confirm these findings in macular-bearing models, such as non-human primates, to better assess the relevance and efficacy of interventions for AMD. Acknowledging this limitation is essential to contextualize the findings and ensure appropriate interpretation of the study’s implications.

Conclusion

In conclusion, the WHHL rabbit model offers valuable insights into the impact of lipid dysregulation on retinal pathology, structure, and function. Our findings suggest that lipid dysregulation contributes to both retinal degeneration and the development and response to treatment of CNV, with partial resistance to anti-VEGF therapy observed in this model. The combination of imaging techniques, including PAM, FA, and OCT, provided a comprehensive assessment of CNV progression and retinal changes, underscoring the need for multifaceted therapeutic approaches to effectively manage lipid-related retinal diseases. Future research should focus on developing novel combination treatments to improve outcomes in patients with persistent disease activity.

Highlights:

• The Watanabe Heritable Hyperlipidemic (WHHL) rabbit demonstrates key features of persistent disease activity (PDA) in AMD, including chronic CNV and treatment resistance.

• Longitudinal evaluation using fundus photography, FA, ICGA, PAM, and OCT provided detailed insights into CNV progression and retinal changes over one year.

• A progressive lipid layer developed in the cornea by month 7, and histological analysis revealed significant outer nuclear layer cell loss in WHHL rabbits compared to WT controls.

• Bevacizumab treatment resulted in a 90.13% reduction in CNV leakage in WT rabbits but only a 16% reduction in WHHL rabbits, demonstrating resistance to anti-VEGF therapy.

• Despite lacking a fovea, WHHL rabbits serve as a valuable preclinical model for studying lipid metabolism disorders, AMD pathophysiology, and novel therapeutic strategies.