Oxygen-Induced Retinopathy Model for Ischemic Retinal Diseases in Rodents

Summary

Oxygen-induced retinopathy (OIR) can be used to model ischemic retinal diseases such as retinopathy of prematurity and proliferative diabetic retinopathy and to serve as a model for proof-of-concept studies in evaluating antiangiogenic drugs for neovascular diseases. OIR induces robust and reproducible neovascularization in the retina that can be quantified.

Abstract

One of the commonly used models for ischemic retinopathies is the oxygen-induced retinopathy (OIR) model. Here we describe detailed protocols for the OIR model induction and its readouts in both mice and rats. Retinal neovascularization is induced in OIR by exposing rodent pups either to hyperoxia (mice) or alternating levels of hyperoxia and hypoxia (rats). The primary readouts of these models are the size of neovascular (NV) and avascular (AVA) areas in the retina. This preclinical in vivo model can be used to evaluate the efficacy of potential anti-angiogenic drugs or to address the role of specific genes in the retinal angiogenesis by using genetically manipulated animals. The model has some strain and vendor specific variation in the OIR induction which should be taken into consideration when designing the experiments.

Introduction

Reliable and reproducible experimental models are needed to study the pathology behind angiogenic eye diseases and to develop novel therapeutics to these devastating diseases. Pathological angiogenesis is the hallmark for wet age-related macular degeneration (AMD) and for many ischemic retinal diseases among them retinopathy of prematurity (ROP), proliferative diabetic retinopathy (PDR) and retinal vein occlusion (RVO)^1,2,3,4. Human and rodent retinas follow a similar pattern of development, as both human and rodent retina are among the last tissues that are vascularized. Before the retinal vasculature has completely developed, retina receives its nutrient supply from hyaloid vasculature, which, in turn, regresses when the retinal vasculature starts to develop^1,2. In human, retinal vascular development is completed before birth, whereas in rodents the growth of retinal vasculature occurs after birth. Since the retinal vascular development occurs postnatally in rodents, it provides an ideal model system to study the angiogenesis^2,3. The newborn rodents have an avascular retina that develops gradually until complete vascular retina development is achieved by the end of third postnatal week⁴. The growing blood vessels of neonatal mouse are plastic, and they undergo regression during hyperoxia stimulus⁵.

ROP is the leading cause for childhood blindness in Western countries, as it affects almost 70% of the premature infants with birthweight under 1,250 g^6,7. ROP occurs in premature infants who are born before retinal vessels complete their normal growth. ROP progresses in two phases: in Phase I, preterm birth delays the retinal vascular growth where after in phase II, the unfinished vascularization of the developing retina causes hypoxia, which induces the expression of angiogenic growth factors that stimulate new and abnormal blood vessel growth⁸. The OIR model has been a widely used model to study the pathophysiology of ROP and other ischemic retinopathies as well as to test novel drug candidates^2,3,9. It is widely considered as a reproducible model for carrying out proof-of-concept studies for potential antiangiogenic drugs for ocular as well as non-ocular diseases. The two rodent models i.e., mouse and rat OIR differ in their model induction and disease phenotype. The rat model mimics ROP phenotype more accurately, but the mouse model provides a more robust, fast and reproducible model for retinal neovascularization (NV). In the mouse model, NV develops to the central retina. This pathological read-out is important in pharmacologic efficacy studies for many ischemic retinopathies, such as PDR, RV and exudative AMD as well as for non-ocular, angiogenic diseases such as cancer. Moreover, availability of genetically manipulated (transgenic and knockout) mice makes the mouse OIR model a more popular option. However, neither mouse nor rat OIR model creates retinal fibrosis, which is typical in human diseases.

The understanding that high oxygen levels contribute to the development of ROP in 1950s^10,11 led to the development of animal models. The first studies about the effect of oxygen on retinal vasculature were done in 1950^12,13,14 and until the 1990s there were many refinements to the OIR model. The research by Smith et al. in 1994 set a standard for the current mouse OIR model that separates hyaloidopathy from retinopathy¹⁵. A wide adoption of the method to quantify vaso-obliteration and pathological NV by Connor et al. (2009) further increased its popularity¹⁶. In this model, mice are placed at 75% oxygen (O₂) for 5 days at P7, followed by 5 days in normoxic conditions. Hyperoxia from P7 to P12 causes retinal vasculature to regress in central retina. Upon return to normoxic conditions, avascular retina becomes hypoxic (Figure 1A). Due to the hypoxic stimuli of the avascular central retina, some of the retinal blood vessels sprout towards the vitreous, forming preretinal NV, called preretinal tufts^2,3. These tufts are immature, and hyperpermeable. The amount of NV peaks at P17, after which it regresses. The retina is fully revascularized and NV is fully regressed by P23 - P25 (Figure 2A)^2,3.

The rat OIR model (using varying levels of O₂) was first described in the 1990s showing that varying O₂ levels at 80% and 40% cause more pronounced NV than under 80% O₂ constant exposure¹⁷. Later it was discovered that the intermittent hypoxia model, where O₂ is cycled from hyperoxia (50%) to hypoxia (10-12 %), causes even more NV than the 80/40% O₂ model¹⁸. In the 50/10% model, rat pups are exposed to 50% for 24 hours, followed by 24 hours in 10% O₂. These cycles are continued until P14, when the rat pups are returned to normoxic conditions (Figure 1B). As in human ROP patients, in the rat model the avascular areas develop to the periphery of retina because of immature retinal vascular plexus (Figure 3).

In both models, the main parameters that are usually quantified are the size of AVA and NV. These parameters are typically analyzed from retinal flat mounts where the endothelial cells are labeled^4,16. Previously the amount of preretinal NV was evaluated from retinal cross sections by counting blood vessel or vascular cell nuclei extending to vitreous above the inner limiting membrane. The major limitation of this approach is that it is not possible to quantify the AVAs.

Protocol

The protocol described here has been approved by the National Animal Ethics Committee of Finland (protocol number ESAVI/9520/2020 and ESAVI/6421/04.10.07/2017).

1. Experimental animals and mouse OIR model induction

NOTE: Use time-mated animals, e.g., commonly used C57BL/6J mice, to get pups born on the same day. Use fostering dams, e.g., 129 strain (129S1/SvImJ or 129S3/SvIM) lactating dams, to nurse the pups during and after the induction of hyperoxia. Alternatively, make sure that there are extra lactating dams available in case the nursing dams need to be replaced due to exhaustion. Restrict the litter size to 6-7 pups for each dam when using C57BL/6J mice/dams (if the litters are larger than that the pups tend to have restricted weight gain)¹⁶.

1. Record the weight of the animals before and after hyperoxia induction, and at the time of sacrifice.
2. Make sure that there is enough food on the bottom of the cage, so the dams have an easy access to food.
3. Add soda lime with color indicator to the bottom of the chamber to absorb excess CO₂ when a filtration system is not used.
4. Monitor the humidity and temperature inside the chamber and keep the humidity between 40 to 65%. Increase the humidity of the chamber, if needed, by placing dishes with water on bottom of the chamber (e.g., Petri dishes).
5. Calibrate the O₂ sensor with 100% O₂ and normal room air.
6. Place the P7 mice into a chamber and set up the O₂ level to 75%. Keep the mice in the chamber for 5 days, until P12. Avoid opening the chamber during the hyperoxia induction. Check the gas pressure of the O₂ cylinder and replace the cylinder when needed. Monitor the animals during the induction.
7. Take the mouse cages out of the chamber and weigh all the pups. Group the pups based on the weight so that each experimental group has similar weight distribution in pups.

2. Experimental animals and rat OIR model induction (using semi-closed system)

NOTE: Use time-mated animals to get the pups born on the same day. For rat OIR, use increased litter size, approximately 18 pups/dam, to obtain sufficient NV induction in the rat model. Pool pups from several litters to obtain enough pups to each litter.

1. Record the weight of the animals before and after induction, and at the time of sacrifice.
2. Make sure that there is enough food on the bottom of the cage, so the dams have an easy access to food.
3. Add soda lime with color indicator to the bottom of the chamber to absorb excess CO₂ when filtration system is not used.
4. Monitor the humidity and temperature inside the chamber. Absorb extra humidity (generated from multiple number of rats) by adding silica gel on the bottom of chamber.
5. Calibrate the O₂ sensor with 100% N₂ and normal room air.
6. Place the rats into the chamber at P0 (few hours after the birth). Set the O₂ level to 50% and connect O₂ cylinder to the chamber for 24 h. After that, switch the settings to 10% O₂ and connect nitrogen (N₂) cylinder to the chamber for 24 h. Continue the 24 h cycling between 50% and 10% O₂ levels for 14 days.
7. Monitor the gas consumption and the wellbeing of the animals during the study. Open the chamber during the change between 50/10% O₂ and add more food and water if needed. Change the cages of the animals to clean ones during the induction.
8. Take the rat cages out of the chamber and weigh all the pups. Group the pups based on the weight so that each experimental group has similar weigh distribution in the pups.

3. Drug administration (optional)

NOTE: Commonly used drug administration route in OIR is by intravitreal treatment (ivt), at P12-P14 for mice and at P14 for rats. Determine the treatment day based on the experimental setup. When multiple litters of pups are used in experiments, divide the treatment groups to have animals from all the litters. Preferably, inject the drug to only one eye, and keep the contralateral eye as a control.

1. Weigh the animals and make identification marks to the tail and/or ear.
2. Anesthetize the animal either with injectable anesthesia (for example mixture of ketamine and medetomidine, 30 mg/kg and 0.4 mg/kg for mice) or with inhalation anesthesia (isoflurane at 2-3.5% isoflurane and 200-350 mL/min air flow). Check the depth of the anesthesia by pinching the toes. Keep the animal on a heating pad during the treatment.
3. For local anesthesia, apply a drop of analgesic onto the eyelid. Open the eyelid carefully with forceps before performing the ivt, as mice and rats open their eyes around P14. Apply a drop of analgesic (e.g., oxybuprocaine hydrochloride) onto the cornea.
4. Apply a drop of iodine before conducting the ivt injection.
5. For the ivt injection use a glass syringe with a 33-34 G needle attached. Press the eyelids down and grap the eyeball with forceps. Make the injection posterior to the limbus, approximately in 45^° angle needle pointing towards optic nerve.
6. Avoid injecting more than 1.0 µL into the intravitreal space. Keep the needle in place for 30 s after injecting the drug to avoid reflux of the injected solution.
7. Examine the eye (e.g., with an ophthalmoscope) for any complications, such as hemorrhages or retinal damage, after removing the needle. Apply antibiotic ointment on top of cornea after the injection.
   NOTE: The ivt injection volume for mice should be 0.5 – 1.0 µL.
8. Reverse the anesthesia (for example with an α2-antagonist for medetomidine (2.5 mg/kg) and return the pup to the cage. House the litter normally until the end of the study.

4. In vivo imaging and electroretinography (optional)

1. If desired, conduct in vivo imaging on live animals during the follow-up period to record changes that develop in retina during the angiogenic responses. For example, perform fluorescein angiography (FA) or scanning laser confocal microscopy¹⁹ to visualize the vasculature (Figure 4). Use spectral domain optical coherence tomography (SD-OCT) to visualize retinal layers in vivo (Figure 4).
2. If desired, investigate functional changes in different retinal cell populations after OIR induction by using electroretinography (ERG) (Figure 5).

5. Tissue collection and preparation of retinal flat mounts

NOTE: Collect the tissues according to the desired research hypothesis. For mice, collect the samples for example at P12 (to study vaso-obliteration after the hyperoxic phase) or at the hypoxic period (P13-P17). Collect the mouse OIR samples at P17, which is the most common time point for sampling, to detect the peak in NV amount. In rat OIR, collect the samples at P18-P21 to observe the highest amount of NV (Figure 3).

1. Weigh the animals before sampling.
2. To label the retinal vasculature, deeply anesthetized animals can be transcardially perfused with FITC-dextran. (Alternatively, stain the retinal flat mounts with Isolectin later).
3. Sacrifice the animals using either overdose of anesthesia drugs (for example mixture of ketamine and medetomidine, 300 mg/kg and 4 mg/kg for mice) or CO₂ inhalation.
4. Collect the eyes of the animals by grabbing behind the eyeball with curved forceps, cut the tissue around the eyes and lift the eye out from the orbit.
5. Incubate the eyeballs in freshly made, filtered 4% paraformaldehyde (in phosphate-buffered saline, PBS) for 1-4 h. Remove the fixative and wash the eyeballs 3 x 10 min with PBS. Dissect the retinas immediately or store them in PBS at +4 °C.
   CAUTION: Paraformaldehyde is toxic by inhalation, in contact with skin and if swallowed. Please read safety data sheet before working with it.
   NOTE: Do not apply pressure to the eyeball during the sampling or any phase of the tissue processing in order to avoid retinal detachment, if cross-sections from whole eyeballs are done.
6. Prepare retinal flat mounts to quantify the amount of NV and the size of AVAs. Alternatively, process the eyeballs/retinas for histology, or RNA or protein analysis. Dissect the retina under a stereo microscope using micro scissors and forceps.
   1. Place the eyeball in PBS to keep it moist and puncture the eyeball at limbus with a needle (23G) and cut around limbus with curved micro scissors to remove iris and the cornea.
   2. Carefully place the tip of the scissors between sclera and retina and cut sclera towards the optic nerve. Do the same to the other side of the eyeball, and carefully cut/tear the sclera until the retinal cup is exposed. Pull the lens out from the retinal cup and add PBS to the cup.
   3. Remove all the hyaloid vessels, vitreous and debris without damaging the retina. Wash the retina by adding PBS to the retinal cup. Perform four incisions (at 12, 3, 6 and 9 o’clock) to the retina with straight micro scissors to make a flower-like structure. Optionally, make the cuts with surgical blade prior mounting the samples. Lift the retina using a soft paintbrush to a well-plate for staining.
7. Label the retinal vasculature using Isolectin B₄ which stains the surface of endothelial cells (if the animals were not perfused with FITC-dextran). Incubate the retinas in blocking buffer (10% NGS + 0.5% Triton in TBS) for 1 h and wash with 1% NGS + 0.1% Triton in TBS for 10 min. Incubate the retinas with fluorescent dye conjugated Isolectin B₄ (5-10µg/ml) in 1% NGS + 0.1% Triton in TBS overnight at +4 °C while protected from the light.
   NOTE: If desired, label other cells such as inflammatory cells and pericytes using specific antibodies.
8. Wash the retinas 3x for 10 min with 1% NGS + 0.1% Triton in TBS and lift retinas on a microscopic slide, inner retina facing upwards. Carefully spread out the retina using soft paintbrush and remove any remaining hyaloid vessels or debris. Add mounting medium to a cover slip and place it on top of the retina. Store retinas at +4 °C and protect from light.

6. Analysis of the flat mounts

1. Take images of the retinal flat mounts using a fluorescence microscope with 10x objective. Focus to the superficial vascular plexus and to the preretinal neovascularization. Make a tile scan image to capture the whole retina and merge the tile scans
2. Quantify the images by measuring the AVAs, area of NV and total retinal area using an image processing program (see Table of Materials).
   1. Draw the AVAs and total retinal area using a free hand drawing tool and select the neovascular areas using a selection tool. The software measures the regions of interest in pixels, and the AVA and NV areas (expressed in pixels) can be used to calculate their percentage in relation to the total retinal area. Also, some software tools are available for quantifying NV.
      NOTE: Recently, an open-source, fully automated programs for the quantification of key values of OIR images using deep learning neural networks have been introduced and provide a reliable tool for reproducible quantification of retinal AVA and NV (e.g., https://github.com/uw-biomedical-ml/oir/tree/bf75f9346064f1425b8b9408ab792b1531a86c64)^20,21.
3. If using antibodies for immunohistochemical detection of individual cell populations, quantify the number of stained cells (such as microglia, Figure 2B) from the retinal flat mounts by hand or by automated image analysis systems if desired.

7. Statistics

1. Analyze normally distributed data by Student’s t test or One-Way ANOVA followed by Dunnett’s or Tukey’s multiple comparisons test, as appropriate. Use nonparametric tests like Mann-Whitney U test or Kruskal Wallis test for non-normally distributed data. Consider differences statistically significant at the P < 0.05 level.

Representative Results

The main outcome of the model is the vascular phenotype: the size of AVAs and the amount of NV. In the mouse OIR model, the vaso-obliteration occurs in the central retina (Figure 2A), while in the rat model it develops in the periphery, i.e., similar to human ROP²² (Figure 3A). This is because the superficial vascular plexus has already developed when mice are exposed to hyperoxia, whereas in the rat model the retina is avascular at the time of OIR induction (P0). Preretinal neovascularization develops near the avascular areas, i.e., central retina in mouse, and periphery in rats (Figure 2A and Figure 3A).

Histological analysis using either cross-sections or flat mounts can be done to evaluate morphological changes in OIR retinas or the presence of cell types of interest, for example inflammatory cells (Figure 2B). In addition to retina, whole eye or vitreous samples can be collected for further gene and protein expression analyses in different time points during the OIR model. Gene or protein expression levels can be analyzed with standard methods, such as RT-qPCR or western blotting.

Optionally, non-invasive in vivo imaging can be conducted during the OIR follow-up period. Retinal and hyaloid vasculature can be visualized with FA (Figure 4A). SD-OCT can be used to evaluate structural changes in the retina (Figure 4B). Functional changes in retina can be measured by ERG (Figure 5).

Inhibitors of vascular endothelial growth factor (VEGF) are commonly used in the treatment of human angiogenic eye diseases. Thus, anti-VEGF is often used as a reference compound in OIR. Aflibercept, that works as a soluble VEGF-trap, inhibits both NV and physiological revascularization in OIR in both high and low doses (injected at P14). OIR eyes injected at P14 with high dose of aflibercept had even bigger retinal AVAs than untreated eyes (Figure 6). This suggests that aflibercept blocks also physiological retinal revascularization driven by hypoxia. Both mouse and rat models can be used to evaluate the effect of different anti-angiogenic agents on retinal NV and physiological revascularization of the retina (Figure 3B and Figure 6).

Figure 1: Graphical overview of a standard OIR study design for mice and rats. (A) Mouse OIR model was induced by exposing the mice to 75% O₂ from P7 to P12 and returned to normal room air. The peak of preretinal NV was seen at P17, which is usually the sampling point for the experiment. (B) Rat OIR model was induced by exposing the rats to alternating O₂ levels (50/10 % O₂) from P0 to P14 and returned to normoxic conditions. Rats are usually sacrificed at P20, when the amount of NV peaked. Please click here to view a larger version of this figure.

Figure 2: Vascular phenotype in the mouse OIR model. Mouse OIR model was generated as described in Figure 1. (A) The superficial vascular plexus developed in normal mice while the OIR mice were exposed to 75% O₂ from P7 to P12. During this time, vascular obliteration developed in the central retina (marked in blue at P12 and P17 in the quantified images). Mice were returned to normal room air, and the avascular retina became hypoxic, leading to functional vessel regrowth in retina and pathological NV (marked in red at P17 in the lower panel). Preretinal neovascular tufts developed between the vascular and avascular areas in central retina. The amount of NV peaked at P17 and regressed afterwards. Retina was fully revascularized and the NV regressed around P24-P25. (B) An example of immunohistochemical staining of retinal flat mount showing retinal Iba-1 stained microglia (green) and GS-IB₄ stained blood vessels (red) at P12 and P17. (C) Cross-section of a mouse OIR eye at P17, where preretinal tufts (arrows) were sprouting towards the vitreous. Also, thinning of the inner nuclear layer (INL) and outer plexiform layer (OPL) was seen. Scale bars are 1 mm in A, 50 µm in B, and 100 µm in C. ILM = inner limiting membrane, GCL = ganglion cell layer, IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer, RPE = retinal pigment epithelium. Please click here to view a larger version of this figure.

Figure 3: Vascular phenotype and its development in the rat OIR model. (A) Avascular areas (marked in blue) and NV (marked in red) developed in the periphery of the retina, similar to human ROP. (B) AVAs were seen in OIR retinas, but not in normoxic controls. (C) Quantification of the area of NV showed a trend towards a peak in the amount of the NV at P20, but the difference was not statistically significant compared to P18 and P21 in OIR. (Student’s t-test, between time matched normoxic and OIR retinas, ** p ≤ 0.01, *** p ≤ 0.001, both eyes of the animals plotted in the graph). Please click here to view a larger version of this figure.

Figure 4: In vivo imaging using fluorescein angiography (FA) and spectral domain optical coherence tomography (SD-OCT) in the rat OIR model. (A) Vascular tortuosity (arrowheads) was seen in the images taken from the central retina (top row) at P18 and P20 in OIR rats compared to normoxic P19 rats. NV (arrows) and AVAs (asterisk) developed to the periphery of the retina (middle row) as seen in P18 and P20 OIR retinas. Regressing hyaloid vessels (bottom row) were captured by FA. (B) Blood vessel growth towards the vitreous was observed as a thickening on the nerve fiber layer (NFL) and ganglion cell layer (GCL) in the OIR retinas (arrow). Please click here to view a larger version of this figure.

Figure 5: Functionality of retinal neurons can be measured using flash electroretinography (fERG). (A) a-waves were derived from retinal cone and rod photoreceptors and their amplitudes were significantly decreased in the rat OIR model. The effect increased with a higher light intensity, suggesting that the defects were affecting cone photoreceptor functions primarily. (B) B-wave amplitudes from OIR animals were decreased compared to normoxic controls. B-waves were derived from ON-bipolar cells and Müller cells. Please click here to view a larger version of this figure.

Figure 6: Intravitreally injected PBS reduces NV and AVAs in the mouse OIR model. (A) Representative images of retinal flat mounts from untreated, and aflibercept-treated (20 µg at P14) and PBS injected OIR eyes at P17. High dose of aflibercept (20 µg) increased the size of AVAs by 47% (outlined in white) and inhibited NV by 98% (arrows) compared to untreated controls. PBS injection decreased AVAs by 31% and NV by 42% compared to untreated controls. Also, puncture of the sclera resembling ivt produced similar effects as ivt injection of PBS, but the differences were not statistically significant. Arrowhead points at hyaloid vessels that were not removed during the dissection. (One-Way ANOVA, * p ≤ 0.05; ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001). Please click here to view a larger version of this figure.

Discussion

The severity of disease phenotype is dependent on both the strain and even vendor in both mouse and rat OIR models²³. This suggests that there is a wide genotypic variability in the pathology development. In general, pigmented rodents develop more severe phenotype than the albino ones. For example, the retinal vasculature of albino BALB/c revascularizes rapidly after hyperoxia and does not develop NV at all²⁴. Similarly, in rats, pigmented Brown Norway rats show more severe pathology than albino Sprague Dawley (SD) rats²⁵. SD rats are commonly used strain, and vendor-related differences in OIR phenotype have been reported within the strain. For example, SD rats from Charles River produce significantly more NV than rats from Harlan or Zivic-Miller²³. C3H/HeJ mice, in turn, that have a mutation in the retinal degeneration 1 (Rd1) gene, have thin retinas, and do not develop NV²⁶. Due to these reasons and the availability of transgenic mice lines, the inbred C57BL/6J is the most commonly used mouse strain for OIR studies. However, there is quite a high mortality among the C57BL/6J dams due to the hyperoxic exposure, so the wellbeing of the mice needs to be considered when designing the study and monitored during the experiments. Moreover, increased photoreceptor damage is seen in C57BL/6 mice compared to BALB/c mice²⁷.

In order to ensure the wellbeing of the mice and survival of the dams and pups, the litter size should be kept small (6-7 pups/dam). This is especially important when using C57BL/6 mice, which are more susceptible to hyperoxic stress^27,28. In addition, easily accessible support food (placed on the bottom of the cage) should be provided to the mice. Some researchers use surrogate dams to replace the dams in the chamber with healthy ones after the oxygen-induction. However, most researchers use surrogates only if the dam is exhausted^22,29. Surrogate dams are recommended to be used if bigger litter sizes are used. It has been published that 129S3/SvIM mice produce more NV than C57BL/6 and sustain better the changes caused by varying oxygen levels²⁸. It should be noted that the dams used in OIR are unfertile after the OIR induction and should not be used for further breeding purposes²². Postnatal weight gain of the pups affects the severity of OIR pathology. Pups with poor postnatal weight gain (<5 g at P17) show delayed expression of VEGF and thus prolonged phase of retinopathy when compared to pups with normal (5 - 7.5 g at P17) or extensive weight gain (>7.5 g at P17)²⁹. Furthermore, mice pups weighing over 5 g at P7 do not show vaso-obliterated areas after OIR induction³⁰. Thus, it is important to weigh the pups during the experiment. The contralateral untreated eye can be used as an internal control to ensure that the nutritional status of the mice does not affect the results. The situation is similar in rats; the smaller the pups are, the more severe OIR phenotype they develop³¹. Thus, large litter size (approx. 18 pups) is recommended to be used for the rat OIR model.

The OIR model induction can be done using either fully closed system, where there is a closed-loop circuit with a pump that circulates the air through a filter system (soda lime and activated carbon) back to the chamber. Another option is a semi-closed system, where ventilation ensures that the excess metabolites are removed from the chamber. To ensure that excess CO₂ is removed, soda lime (mixture of calcium hydroxide with sodium hydroxide or potassium hydroxide) can be placed on the bottom of the chamber. The removal of CO₂ is mandatory as the high CO₂ levels worsen the disease phenotype in rats³². One should also remember to recalibrate the oxygen sensor of the chamber regularly as they tend to drift over time and may provide wrong oxygen concentration from the one intended.

It has been reported that vehicle injection (PBS) alone or even just the puncture to the intravitreal space has an effect for the revascularization rate and the amount of NV (Figure 6)^33,34,35,36. It has been speculated whether the effect seen with PBS/vehicle injection could be due to changes in intraocular pressure during injection, or due to injury to ocular structures that increases levels of angiogenic growth factors among them pigment epithelium-derived growth factor^34,37. Furthermore, just a pilot subretinal injection alone ( puncture to the subretinal space) has been shown to have effects on the vascular and functional phenotype in OIR when compared to the untreated rats³⁸. These results highlight the importance of proper negative controls in the OIR studies or even systemic administration of tested compounds, as well as big enough n-numbers in each study group. The fact that vehicle injection indeed enhances revascularization rate in retina is a major limiting factor as the revascularization rate and the formation of pathological preretinal tufts are inter-related in the OIR model: if the revascularization rate is accelerated, it leads to the compensatory downregulation of neovascular tufts and vice versa. Thus, both primary outcome measures of OIR model are affected by vehicle injection.

VEGF inhibitors have revolutionized the treatment AMD and DR. OIR model was used to demonstrate their efficacy preclinically. Concerning VEGF inhibitors in OIR, a study comparing different anti-VEGFs and anti-PlGFs (placental growth factor) (injected at P12) showed that anti-VEGF alone led to small avascular areas, whereas anti-PIGF treated eyes had bigger avascular areas³³. Aflibercept had the biggest AVAs when compared to other drug treatments³³. Aflibercept is known to bind both VEGF and PlGF³⁹, it could be that inhibiting PlGF in OIR leads to blocking of physiological angiogenesis.

Noninvasive in vivo imaging provides a tool for monitoring retinal vasculature⁴⁰ and retinal layers and structure during follow-up period. Using FA, parameters like vascular density and arterial tortuosity and venous dilation (called plus disease, Figure 4A) can be measured^40,41. SD-OCT can be used for evaluating structural changes and measuring retinal thickness during OIR follow-up period^42,43. ERG is used to measure the functional changes in the retina. Different cell types produce electric potentials after light stimulus, and these signals or “waves” can be measured with ERG. Photoreceptors are producing the negative a-wave, and ON bipolar cells and Müller cells are mainly responsible for the b-wave⁴⁴. Loss of retinal function is typical in OIR mice and rats^45,46 (Figure 5). Long lasting changes persist in retina and both functional changes and changes in protein level have been reported in OIR retinas even after revascularization and NV regression^34,47.

To visualize the retinal vasculature, the live mice can be either perfused with FITC-labeled dextran or the retinal flat mounts can be stained with fluorescent dye labeled Isolectin B₄. One needs to understand the difference between the fluorescent dyes; the perfusion with FITC-dextrans labels only the lumen of the vessels, whereas Isolectin B₄ stains the surface of endothelial cells. Thus, functional blood vessels with proper lumen will only be visualized with FITC-dextran perfusion, while the total area of NV appears bigger in Isolectin B₄ stained retinas than in FITC-dextran perfused animals, because Isolectin B₄ essentially picks up even endothelial cells which have not formed functional lumens yet⁴⁸. A more recent option to visualize whole retina/eye ball in 3-D format is deep tissue imaging by two-photon fluorescence microscopy⁽⁴⁹⁾.

The rodent OIR model, as well as the animal models in general, only partially represent the characteristics of human diseases. The major difference related to OIR is that retinal neovascularization is not associated with fibrosis in rodent OIR, whereas retinal neovascularization leads commonly to fibrovascular proliferation in human neovascular retinal diseases. Furthermore, the conditions that cause OIR vs. human disease can be almost opposite. The preterm neonates with ROP require supplemental oxygen with respiratory support, experience frequently intermittent hypoxemic and hyperoxemic episodes caused by recurrent apnea, but they are not exposed to high levels of oxygen. To minimize the permanent damage, high fractions of inspired O₂ are avoided. In that respect, neither the mouse model with constant exposure to 75% O₂ for 5 days nor the 50/10 rat model resemble pathogenesis of human ROP. Furthermore, there are also differences between the rat and the mouse OIR models. The neovascularization regresses in the mouse model with re-establishment of normal vessels by P24, whereas the condition gets worse in the rat OIR model (similar to human ROP). Although, the rat OIR model shows clinically relevant features of ROP such as the delayed retinal vascular development and subsequent pathological neovascularization, its use is limited by almost complete absence of transgenic rat strains, higher maintenance costs and less NV than in the mouse OIR model.

Taken together, despite the differences between human ischemic proliferative retinopathies and rodent OIR models, the ease by which the NV can be induced, coupled with easy visualization and quantification of the retina, make the OIR models popular to study the molecular mechanisms and potential therapeutics for ischemic proliferative retinopathies. Interestingly, the hyperoxia exposure in mouse OIR model also induces bronchopulmonary dysplasia, another disease caused by supplemental oxygen therapy in human premature infants, showing that OIR model could be used to explore novel targets for both ROP and bronchopulmonary dysplasia simultaneously⁵⁰.

Materials

Name	Company	Catalog Number	Comments
33 gauge, Small Hub RN Needle	Hamilton Company	7803-05, 10mm, 25°, PS4	For intravitreal injection
Adobe Photoshop	Adobe Inc.		For image analysis
Air pump air100	Eheim GmbH & Co. KG.	143207	For inhalation anaesthesia
Anaesthesia unit 410 AP	Univentor Ltd.	2360309	For inhalation anaesthesia
AnalaR NORMAPUR Soda lime	VWR International Ltd	22666.362	For CO2 control during model induction
Attane Vet 1000 mg/g	VET MEDIC ANIMAL HEALTH OY	vnr 17 05 79	For inhalation anaesthesia
Brush			For preparation of flat mounts
Carbon dioxide gas			For sacrifice
Celeris D430 ERG system	Diagnosys LLC	121	For in vivo ERG
Cell culture dishes	Greiner Bio-One International GmbH	664 160	For preparation of flat mounts
Cepetor Vet 1 mg/mL	VET MEDIC ANIMAL HEALTH OY	vnr 08 78 96	For anaesthesia
Cover slips	Thermo Fisher Scientific	15165452	For preparation of flat mounts
O2 Controlled InVivo Cabinet, Aninal Filtrarion System and Dehumidifier	Coy Laboratory Products		Closed system for disease model induction, optional for semi-closed system
E702 O2 sensor	BioSphenix, Ltd.	E207, 1801901	For oxygen level measurement
Envisu R2200 Spectral Domain Optical Coherence Tomograph (SD-OCT)	Bioptigen, Inc.	BPN000668	For in vivo imaging
Eye spears	Beaver-Visitec International, Inc.	0008685	For intravitreal injection and in vivo imaging
Flexilux 600LL Cold light source	Mikron	11140	For intravitreal injection or tissue collection
Fluorescein sodium salt	Merck KGaA	F6377-100G	For in vivo imaging
Gas Exhaust unit (+Double 3-way valve, mouse and rat face masks, UNOsorb filter)	UNO Roestvaststaal BV	GEX 17015249	For inhalation anaesthesia
Glass syringe, Model 65 RN	Hamilton Company	7633-01	For intravitreal injection
HRA2 Retina angiograph (FA)	Heidelberg Engineering GmbH	Spec-KT-05488	For in vivo imaging
Isolectin GS-IB4, Alexa Fluor 488 Conjugate	Thermo Fisher Scientific	I21411	For labeling retinal vasculature on flat mounts
Ketaminol Vet 50 mg/mL	Intervet International B.V.	vnr 51 14 85	For anaesthesia
Medicinal Oxygen gas			For disease model induction
Mice C57BL/6JRj	Janvier Labs		Also other strains possible
Microscope slides	Thermo Fisher Scientific	J1800AMNZ	For preparation of flat mounts
Minims Povidone Iodine 5% (unit)	Bausch & Lomb U.K Limited	vnr 24 11 304	For intravitreal injection
Nitrogen gas			For disease model induction (rat)
Oftan Chlora 10 mg/g	Santen Pharmaceutical Co., Ltd.	vnr 55 01 11	For intravitreal injection
Oftan Metaoksedrin 100 mg/ml	Santen Pharmaceutical Co., Ltd.	vnr 55 03 43	For in vivo ERG
Oftan Obucain 4 mg/ml	Santen Pharmaceutical Co., Ltd.	vnr 55 03 50	For intravitreal injection
Oftan Tropicamid 5 mg/ml	Santen Pharmaceutical Co., Ltd.	vnr 04 12 36	For in vivo imaging
ProOx Model 110 O2 controller and animal chamber	BioSphenix, Ltd.	803	For disease model induction, semi-closed system, optional for closed system
ProOx Model P360 O2 controller and animal chamber	BioSphenix, Ltd.	538	For disease model induction, semi-closed system, optional for closed system
Rats CD(SD)	Charles River Laboratories		Also other strains possible
Revertor 5 mg/mL	VET MEDIC ANIMAL HEALTH OY	vnr 13 04 97	For anaesthesia reversal
Silica gel			For humidity control during model induction
Systane Ultra 10ml	Alcon	Tamro 2050250	For hydration of the eye
Systane Ultra unit 0.7ml	Alcon	Tamro 2064871	For hydration of the eye
Transfer pipette	Thermo Fisher Scientific	1343-9108	For preparation of flat mounts
VENTI-Line VL 180 PRIME Drying oven	VWR	VL180S 170301	For drying silica gel
VisiScope SZT350 Stereomicroscope	VWR	481067	For intravitreal injection or tissue collection