Monitoring Dynamic Growth of Retinal Vessels in Oxygen-Induced Retinopathy Mouse Model

This protocol describes a detailed method for the preparation and immunofluorescence staining of mice retinal flat mounts and analysis. The use of fluorescein fundus angiography (FFA) for mice pups and image processing are described in detail as well.

Oxygen-induced retinopathy (OIR) is widely used to study abnormal vessel growth in ischemic retinal diseases, including retinopathy of prematurity (ROP), proliferative diabetic retinopathy (PDR), and retinal vein occlusion (RVO). Most OIR studies observe retinal neovascularization at specific time points; however, the dynamic vessel growth in live mice along a time course, which is essential for understanding the OIR-related vessel diseases, has been understudied. Here, we describe a step-by-step protocol for the induction of the OIR mouse model, highlighting the potential pitfalls, and providing an improved method to quickly quantify areas of vaso-obliteration (VO) and neovascularization (NV) using immunofluorescence staining. More importantly, we monitored vessel regrowth in live mice from P15 to P25 by performing fluorescein fundus angiography (FFA) in the OIR mouse model. The application of FFA to the OIR mouse model allows us to observe the remodeling process during vessel regrowth.

Retinal neovascularization (RNV), which is defined as a state where new pathologic vessels originate from existing retinal veins, usually extends along the inner surface of the retina and grows into the vitreous (or subretinal space under some conditions)¹. It is a hallmark and common feature of many ischemic retinopathies, including retinopathy of prematurity (ROP), retinal vein occlusion (RVO), and proliferative diabetic retinopathy (PDR)².

Numerous clinical and experimental observations have indicated that ischemia is the main cause of retinal neovascularization^3,4. In ROP, neonates are exposed to high-level oxygen in closed incubators to increase the survival rates, which is also an important driver for the arrest of vascular growth. After the treatment is done, the retinas of newborns experience a relatively hypoxic period⁵. Other situations are seen in the occlusion of central or branch retinal veins in RVO and damage of retinal capillaries is also observed which is caused by microangiopathy in PDR². Hypoxia further increases the expression of angiogenic factors such as vascular endothelial growth factor (VEGF) through the hypoxia-induced factor-1α (HIF-1α) signaling pathway which in turn guide vascular endothelial cells to grow into the hypoxic area and form new vessels^6,7.

ROP is a kind of vascular proliferative retinopathy in preterm infants and a leading cause of childhood blindness^8,9, which is characterized by retinal hypoxia, retinal neovascularization and fibrous hyperplasia^10,11,12. In the 1950s, researchers found that high concentration of oxygen can significantly improve the respiratory symptoms of premature infants^13,14. As a result, oxygen therapy was increasingly used in premature infants at that time¹⁵. However, concurrent with the widespread use of oxygen therapy in preterm infants, the incidence of ROP increased year by year. Since then, researchers have linked oxygen to ROP, exploring various animal models to understand the pathogenesis of ROP and RNV¹⁶.

In human, most retinal vasculature development is completed before birth while in rodents the retinal vasculature develops after birth, providing an accessible model system to study angiogenesis in the retinal vasculature². With the continuous progress of the research, oxygen-induced retinopathy (OIR) models have become major models for mimicking pathological angiogenesis resulting from ischemia. There are no specific animal species in the study of the OIR model and the model has been developed in various animal species, including kitten¹⁷, rat¹⁸, mouse¹⁹, beagle puppy²⁰, and zebrafish²¹. All of the models share the same mechanism by which they are exposed to hyperoxia during early retinal development and then returned to the normoxic environment. Smith et al. observed that exposing mouse pups to hyperoxia from P7 for 5 days induced an extreme form of vessel regression in the central retina and bringing them back to the room air at P12 gradually triggered neovascular tufts, which grew toward the vitreous body¹⁹. This was a standardized OIR mouse model also named as Smith model. Connor et al. further optimized the protocol and provided a universally applicable method to quantify the area of VO (vaso-obliteration) and NV (neovascularization) in 2009, which increased the acceptance and utilization of the model²². OIR mouse model is still the most widely used model now because of its small size, fast reproduction, clear genetic background, good repeatability, and high success rate.

In mice, retinal vascularization starts after birth with the ingrowth of vessels from the optic nerve head into the inner retina toward the ora serrata. During normal retinal development, the first retinal vessels sprout from the optic nerve head around birth, forming an expanding network (the primary plexus) that reaches the periphery around postnatal day 7(P7)²³. Then the vessels start to grow into the retina to form a deep layer, penetrate the retina, and establish a laminar network around the inner nuclear layer (INL) as in human²⁴. By the end of the third postnatal week (P21), deeper plexus development is almost completed. For the OIR mouse model, vascular occlusion always appears in the central retina because of the rapid degeneration of a large number of immature vascular networks in the central region during hyperoxia exposure. So, the growth of pathological neovascularization also occurs in the mid-peripheral retina, which is the boundary of the non-perfusion area and the vascular area. However, human retinal vessels have almost formed before birth. As for premature infants, the peripheral retina is not completely vascularized when exposed to hyperoxia^25,26. So vascular occlusion and neovascularization mainly appear in the peripheral retina^27,28. Despite these differences, the mouse OIR model closely recapitulates the pathologic events that occur during ischemia-induced neovascularization.

The induction of the OIR model can be divided into two phases²⁹: in phase 1 (hyperoxia phase), retinal vascular development is arrested or retarded with occlusion and regression of blood vessels as a result of the decline in VEGF and the apoptosis of endothelial cells^24,30; in phase 2 (hypoxia phase), the retinal oxygen supply will become insufficient under room air conditions²⁹, which is essential for neural development and homeostasis^19,31. This ischemic situation usually results in unregulated, abnormal neovascularization.

Currently, the commonly used modeling method is alternating high/low oxygen exposure: Mothers and their pups are exposed to 75% oxygen for 5 days at P7 followed by 5 days in room air till P17 demonstrated comparable results²², which is the endpoint of OIR mouse model induction. (Figure 1). In addition to simulating ROP, this ischemia-mediated pathological neovascularization can also be used to study other ischemic retinal diseases. The main measurements of this model include quantifying the area of VO and NV, which are analyzed from retinal flat mounts by immunofluorescence staining or FITC-dextran perfusion. Each mouse can be studied only once because of the lethal operation. At present, there are few methods to observe dynamic changes of retinal vasculature continuously during the process of vascular regression and pathologic angiogenesis³². In this paper, we provide a detailed protocol of OIR model induction, analysis of retinal flat mounts as well as a workflow of fluorescein fundus angiography (FFA) on mice which would be helpful to gain a more comprehensive understanding of vascular dynamic changes during two phases of the OIR mouse model.

All procedures involving the use of mice were approved by the animal experimental ethics committee of Zhongshan Ophthalmic Center, Sun Yat-sen University, China (authorized number: 2020-082), and in accordance with the approved guidelines of Animal Care and Use Committee of Zhongshan Ophthalmic Center and the Association Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.

1. Induction of mouse OIR model

1. Use mice with a lower rate of congenital malformation of the eyes, e.g., C57BL/6J mice, and mate them at a ratio of male/female = 1:2. Get the pups born on the same day and start to induce the OIR model at P7. Record the bodyweight of mouse pups strictly before modeling.
   NOTE: Note the day of birth as P0. Record the weight of each mouse regularly. The bodyweight of newborn pups is very important during the induction of OIR as the sensitivity of mice in different states to oxygen is different. Exclude the pups more than 5 g at P7 to ensure comparable results.
2. Provide a suitable living environment for nursing mothers and their pups, such as setting the temperature at 23 °C ± 2 °C, controlling the humidity at 40%-65%, alternating 12 h of light and 12 h of darkness every day, adding some cotton wool to the cage for nesting, ensuring adequate sterilized food and water, and keeping them in individually ventilated cages (IVC).
3. Monitor the level of humidity and temperature inside the chamber. Control the humidity between 40% to 65% and keep the temperature at 23 °C ± 2 °C.
4. Check the oxygen supply with oxygen sensors, maintain a constant oxygen level at 75% and control the oxygen flow rate at 0.5-0.75 L/min. Put 50 g soda lime at the bottom of the chamber to absorb excessive CO2 and maintain CO2 values below 3%²².
5. Monitor the behaviors of nursing mothers such as nest-building behavior, biting their pups, and refusing lactation at least once a day. Eliminate nursing mothers with poor motherhood.
6. Place the P7 pups (male and female) and their nursing mothers into an oxygen chamber in which the oxygen level is 75% for 5 days to P12. Avoid unnecessary opening of the chamber during the period of model induction. Ensure that there are extra surrogate mothers for replacement, in case the nursing mothers die due to lung injury while in hyperoxia.
   NOTE: To ensure the comparability of the experiment, restrict the number to 6-8 pups for each mother. Pay attention to the potential problem of oxygen toxicity, which causes the death of some nursing mothers. The signs of hyperoxic lung injury in nursing mothers include, but are not limited to, fluctuating respiratory rate, decreased activity, and decreased feeding. When the above phenomenon occurs, euthanize the nursing mother with 1% pentobarbital sodium (50 mg/kg) as soon as possible. Prepare some surrogate mothers, e.g., 129S1/SvImJ for replacement and use them only if necessary. It is not recommended to replace nursing mothers as a routine, as this will lead to frequent opening of an oxygen chamber, resulting in unstable oxygen levels and maternal aggression.
7. Bring the pups and their nursing mothers back to the room air at P12 and monitor the weight of all the pups continuously until P17. Group the pups based on the weight to ensure that each experimental group has a similar weight distribution.

2. Preparation of retinal whole mounts and immunofluorescence staining

1. Record the bodyweight of the pups. Sacrifice the pups by an overdose of anesthetic (1% pentobarbital sodium 50 mg/kg) or CO₂ inhalation. Other methods of euthanasia, such as cervical dislocation and bilateral thoracotomy, can be used if necessary.
2. Use curved scissors to release the connection between eyeballs and orbital tissue. Then, put curved forceps into the posterior part of the eyeball, clamp the optic nerve, and quickly lift the eye out from the orbit. Wash the eyeballs in a pre-cooled 1x phosphate buffer saline (PBS) to remove the hair and blood from the surface of the eyeballs.
3. Place the cleaned eyeballs in a 2 mL microcentrifuge tube filled with 4% paraformaldehyde (PFA) and incubate for 15 min at room temperature on a shaker at a speed of 12-15 revolutions per minute (rpm) (initial fixation).
   CAUTION: Paraformaldehyde is known to be allergenic, generally toxic, and extremely cytotoxic. Follow the safety instructions strictly and avoid inhalation and skin contact.
4. Use a culture dish and put a drop of 1x PBS into the central part and perform the following steps under a dissecting microscope and place one eyeball in this drop. Hold the eyeball with a pair of forceps and carefully puncture the cornea at the corneal limbus using a 1 mL syringe needle. Insert the tip of the scissors into this hole and cut off the cornea carefully along the cornea limbus. Be careful not to cut the retina.
5. Remove the iris and lens with a pair of forceps. Then place the remaining eyecup in the 4% PFA and fix again for another 45 min at room temperature on a shaker at a speed of 12-15 rpm (secondary fixation).
6. Use a culture dish and put a drop of 1x PBS into the central part. Place the fixed eyeball in this drop. Hold the eyeball with a pair of forceps. Gently separate the retina and sclera layers using two forceps. Place the tip of the scissors between the retina and sclera layers and cut the sclera toward the optic nerve. Peel the sclera off the retina and obtain the retinal cup.
   NOTE: Hold the posterior cup by the optic nerve with forceps, then use the curved end of another forceps to press down on the sclera at the optic nerve head and gently massage out the retina in a forward sweeping motion as an alternative to release the retina.
7. Use forceps to release the connection between radial hyaloid vessels and peripheral retina, clamp the root of the hyaloid vessels which is close to the optic nerve head, and cut the hyaloid vessels off carefully.
8. Use a 2 mL pipette with the tip cut off to transfer the retinal cup. Place the retinal cup into one well in a 48-well plate and wash it for 3 x 5 min with 1x PBS at room temperature on a shaker at a speed of 12-15 rpm.
9. Incubate the retinal cup in a mixed solution of 1% Triton X-100 (in PBS) and 5% normal donkey serum (in PBS) overnight at 4 °C.
   1. Alternatively, block and permeabilize retinas at room temperature for 1 h as an alternative. Change blocking serum according to the source of the secondary antibody.
10. If labeling the retinal vasculature using Isolectin B4, incubate the retina in a well of 48-well plate with 0.1% normal donkey serum (400 µL) and IsolectinB4-594 (1:400) overnight at 4 °C on a shaker at a speed of 12-15 rpm.
    NOTE: If labeling the blood vessels with other markers, such as CD31, or labeling other cells, use specific primary antibodies to label them.
11. Incubate the retina with 1:100-1:500 specific primary antibodies (in 400 µL 0.1% normal donkey serum) at 4 °C on a shaker at a speed of 12-15 rpm for 48 h. (optional)
12. After returning to the room temperature, wash the retina with 0.1% PBST (0.1% TritonX-100 in PBS) for 3 x 20 min on a shaker at a speed of 12-15 rpm.
13. Incubate the retina with 1:1,000 secondary antibodies (in 400 µL 0.1% normal donkey serum) overnight at 4 °C on a shaker at a speed of 12-15 rpm. (optional)
    1. Alternatively, incubate the retina with high-affinity secondary antibodies at room temperature for 1 h.
14. Incubate the retina with DAPI (1:1,000) at room temperature for 20-25 min to label the nucleus.
    NOTE: Test the optimal dilution ratios for all the antibodies used in steps 10-11 and 13-14 in pre-experiment.
15. Wash the retina for 3 x 30 min with 0.1% PBST on a shaker at a speed of 12- 15 rpm at room temperature.
16. Transfer the retinal cup to a clean slide with the opening facing upward. Cut the retina radially at the 3, 6, 9, and 12 o'clock positions from peripheral to central by cutting approximately 1-1.5 mm away from the optic nerve head.
17. Add a few drops of 1x PBS to rinse the retina three times. Use air-laid paper to dry and flatten the retina. Add a drop of mounting medium (see Table of Materials) to the center of the coverslip and stop adding it until the diameter of the droplet increases to half of the coverslip. Quickly turn over the coverslip and place it on top of the outspread retina. Avoid forming bubbles.
18. Take images of the retinal flat mounts or store and protect the slides from light at 4 °C.

3. Analysis and quantification of retinal flat mounts

NOTE: For the OIR mouse model, the researchers often record the area of central retinal vascular occlusion and peripheral retinal pathological neovascularization during P12-P25. Previous studies have shown that the central avascular area of the retina reaches the maximum at P12 and gradually shrinks from P13 to P17; at the same time, the retina of OIR mice reaches the peak of neovascularization area at around P17^22,29. From P17, neovessels gradually regress and functional vessels regrow into the avascular area. The retinal vasculature basically returns to normal at P25³³.

1. Take images of retinal flat mounts by a fluorescence microscope (see Table of Materials) with 10x objective lens. First, choose the DAPI channel and set the optic nerve head in the center of the visual field. Then, adjust other channels and focus on the superficial vasculature of the retina. Check Tiles in a photo software (see Table of Materials) and set the number of photos that need to be stitched. Click on Start Experiment to capture the whole retina.
2. Use an image processing program (see Table of Materials) to quantify the area of vaso-obliteration (VO) and neovascularization (NV) after immunofluorescence staining.
   1. First, click on the Magic Wand Tool and set an appropriate tolerance according to the difference in brightness and move the cursor to the background and click the mouse. Then, choose the Select Inverse to obtain a basic outline of the retina. Use the Lasso Tool to further outline the details of the retina. Using the Histogram function, record the pixel value of the whole retina and write it down or generate a table in a database program.
   2. Divide the retina image into four quadrants. In each quadrant, use the Lasso Tool to draw the VO area (Figure 2A-C), and use the Magic Wand Tool to select the NV area (Figure 2D-F). Through the pixel information in the histogram, calculate the pixel ratio of VO and NV to the whole retina, that is, the percentage of VO or NV area relative to the whole retina.
      NOTE: There is also an open-source and fully automated pipeline for the quantification of VO and NV areas in OIR images using deep learning neural networks (http://oirseg.org/), which provides a reliable and time-saving way for researchers as well as unifies the standard of the quantification³⁴.
3. Record pixel information in a spreadsheet table, which is convenient for subsequent analysis.

4. In vivo imaging with fluorescein fundus angiography (FFA)

NOTE: For OIR mice, both FITC perfusion and immunofluorescence staining can only be used for one time because of the death of experimental animals. Compared with this, one of the advantages of FFA is the observation of the dynamic changes of mouse retinal vessels during development and pathological state in vivo^35,36.

1. Weigh the pups before anesthesia.
2. Anesthetize pups by intraperitoneal injection of 0.3% pentobarbital sodium at a dose of 30-50 mg/kg.
   NOTE: For mice within 1 month, pay attention to the anesthetic doses. Use lower concentrations and doses of anesthetic to reduce the death of mice caused by anesthesia. After the pups are anesthetized, use a small heating pad to maintain the body temperature. Hypothermia not only affects pups' physiological function, but also leads to changes in crystallin and accelerates the development of cataracts.
3. Use 20 µL mydriatic eye drops (0.5% tropicamide + 0.5% phenylephrine hydrochloride) for each pup and wait for 5 min to achieve long-lasting pupil dilation (Figure 3A,B).
4. Bring the anesthetized pups in front of the imaging device (see Table of Materials). Keep the pups on a small heating pad, place the pups in a stable position, and use artificial tears regularly to maintain moisture in the cornea. Click on the mode of Infrared Fundus Imaging (IR) to adjust the optic nerve head to the center of the screen. 
   NOTE: When observing one eye of the pups, do not forget to protect the other eye. Use Hypromellose eye drops to prevent the cornea from whitening due to dryness.
5. After intraperitoneal injection of 0.15 mL 0.5% fluorescein sodium salt solution, click on the FA button and the Injection button immediately on the touch panel of the imaging device to start timing. Record the images after 3 min when the blood circulation of the retina enters the venous phase and observe the retina no less than 6-8 min.
   NOTE: After intraperitoneal injection of fluorescein sodium salt solution, the skin, mucosa, and urine of the pups show obvious yellowish green. Most of the fluorescein is excreted by the pups within a day. Injecting the fluorescein sodium intraperitoneally every other day for six times does not cause significant side effects³⁷.
6. Move the optic nerve head to the center of the image acquisition area and take the first image of the central retina. Then, move the lens of the imaging device horizontally to the nasal side of the eye until the optic nerve head is located at the midpoint of one side of the image acquisition area and take the second image. Continue to take images of the temporal, superior and inferior retina, respectively using this method (Figure 3C).
   NOTE: Take "Five-orientation" images within 12 min as the regression phase occurs. The position of the optic nerve head in the inferior image is allowed not to fall on the sideline due to the limited angle adjustment of the lens.
7. Save the images and use an image processing program for stitching.

5. Image processing of the fluorescein fundus angiography (FFA)

1. Open the imaging processing program and click on New in File to create a new canvas with a black background (Figure 4A).
2. Open an image of the central retina first in the background layer. Click on File and add the second image. Adjust the opacity of the second image to 60%, move and resize the second image until the same parts of the two images highly overlap. Click on the Switch Between Free Transform and Warp Modes button and make subtle adjustments to the vessels if necessary. Then, turn the opacity of the second image back to 100% (Figure 4A,B).
3. Select two images at the same time and click on Auto-Blend Layers. Check Panorama as the blend method as well as select the following two sentences. Click on OK and finish the image stitching of the first two images (Figure 4C,D).
4. Take the first two stitched images as a whole, add the third image, and continue to blend. Repeat the methods above to complete the stitching of five images (Figure 4E).
5. Use the Crop Tool to cut images of FFA at different time points to a uniform size and observe dynamic changes of retinal vasculature from P15 to P25 in both normal and OIR pups.

6. Statistical analysis

1. Present values as mean ± standard deviation (s.d.).
2. Use the Student's t-test to compare two independent samples. Use One-Way ANOVA to compare multiple sets of data and combine with Dunnett or Tukey's test, which is a commonly used multiple comparison test.
3. For non-normally distributed data, use Mann-Whitney U test or Kruskal Wallis test. Consider significant statistical differences when P < 0.05.

In the OIR mouse model, the most important and basic result is the quantification of the VO and NV area. After living in the hyperoxia environment for 5 days from P7, the central retina of the pups showed the largest non-perfusion area. Under the stimulation of hypoxia in another 5 days, retinal neovascularization was gradually produced which fluoresced more intensely than surrounding normal vessels. After P17, the fluorescence signal of pathological neovascularization regressed rapidly as the remodeling of the retina (Figure 5A). By controlling the litter size and the postnatal weight gain of the pups, the area of the VO and NV of the OIR mouse model showed good repeatability and stability and the peak of retinal neovascularization occurred at P17, which was in line with the previous studies (Figure 5B,C).

FFA is an ideal tool for studying retinal vasculature. Given the application of FFA in vivo, it shows a great reduction in the waste of experimental animals as well as displays the dynamic changes of the retinal vessels with time. In previous studies, FFA was not often used in mice pups and was presented in a single-view image, which was difficult for further study. In this protocol, the "Five-orientation" images of the retina vasculature were stitched together using an image processing software to display a wider field of the retina at one time, which was helpful for subsequent analysis, if needed (Figure 4). Besides, the OIR mouse pups showed a prolonged eye opening so the FFA images were taken from P15 to meet the requirements of animal ethics. In the retina of the OIR mouse model, the diameter of blood vessels increased evidently and became highly tortuous when comparing to normal mice. Besides, the FFA showed a similar trend of dynamic changes of retinal vasculature with immunofluorescence staining with isolectin B4-594 from P15-P25 without the death of the pups (Figure 6).

Figure 1: Cartoon schematic of OIR mouse model. OIR mouse model was induced by keeping pups and their nursing mothers in a room for some time (P0-P7). At P7, both of them were exposed to 75% oxygen for 5 days, which inhibited retinal vessel growth and caused significant vessel loss in the central retina. Mice were then brought back to room air at P12 and the avascular retina started becoming relatively hypoxic, triggering both normal vessel regrowth and a pathological response around the mid-peripheral retina. The maximum neovascularization (NV) was seen at P17. Then, pathological neovascularization underwent a process of spontaneous regression. The retinal vascular system was back to normal again at around P25. Please click here to view a larger version of this figure.

Figure 2: Measurement of vaso-obliteration (VO) and neovascularization (NV) in the mouse retina. (A) Image of 10x P12 OIR retinal whole-mount stained for endothelial cells with isolectin B4-594. (B) Screenshot of a retina with the avascular area selected. Tools necessary to make this measurement is highlighted with white arrows: Magic Wand Tool and Lasso Tool. (C) Highlight the avascular area of the retina and save the image as a copy. (D) Image of 10x P17 OIR retinal whole-mount stained for endothelial cells with isolectin B4-594. (E) Screenshot of a retina with neovascular tufts selected. Use Magic Wand Tool and set an optimal Tolerance to highlight NV. Set the tolerance to 3-5 and check the anti-alias and contiguous boxes. (F) Save the neovascularization area only as a copy. Scale bars represent 1,000 µm. Please click here to view a larger version of this figure.

Figure 3: Acquisition of the "Five-orientation" images in the mouse retina. (A) The normal mouse pupil. (B) Mouse pupil in mydriasis. (C) The "Five-orientation" images of the central, nasal, temporal, superior, and inferior area of the retina were collected, respectively (P17 pups in room air). Scale bars represent 500 µm. Please click here to view a larger version of this figure.

Figure 4: General workflow of stitching the "Five-orientation" images from fluorescein fundus angiography (FFA). (A) Create a new canvas with a black background and open the FFA image of the central retina. (B) Open an FFA image of the temporal retina and adjust the opacity of the second image to 60%; move and resize the image until the same parts of the two images highly overlap. Click on Switch Between Free Transform and Warp Modes to make subtle adjustments if necessary. Turn the opacity of the second image back to 100%. (C) Select two images at the same time and click on Auto-Blend Layers. (D) Use Panorama as the blend method to finish the image stitching of the first two images. (E) Continue to stitch images by repeating the methods above to complete the stitching of all the images. Please click here to view a larger version of this figure.

Figure 5: Quantification of vaso-obliteration (VO) and neovascularization (NV) in the retina of the OIR mouse model. (A) Image of 10x OIR retinal whole-mounts stained for endothelial cells with isolectin B4-594 from P12 to P25. After being exposed to 75% oxygen for 5 days, pups and their nursing mothers were brought back to the room air at P12 at which the area of vaso-obliteration reached the maximum. The relative hypoxia in the central retina led to vessel regrowth in this area as well as pathological angiogenesis in the mid-peripheral retina. At P17, pre-retinal neovascular tufts reached the maximum and then shrank quickly. NV regressed completely and the retina seemed to be normal at around P25. (B) Quantification of the area of VO showed a peak at P12 and disappearance at around P25. (C) Quantification of the area of NV showed a peak at P17 and regression at around P25. Scale bars represent 1,000 µm in A. (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.

Figure 6: In vivo imaging of fluorescein fundus angiography (FFA) in the OIR mouse model. In the retina of the OIR mouse model, the diameter of blood vessels increased evidently and became highly tortuous when comparing to normal mice. Besides, the FFA showed a similar trend of dynamic changes of retinal vasculature with immunofluorescence staining with isolectin B4-594 from P15-P25 without the death of mice pups. Scale bars represent 500 µm. Please click here to view a larger version of this figure.

The susceptibility of mice to OIR is affected by many factors. The pups of different genetic background and strains cannot be compared. In BALB/c albino mice, vessels regrow into the VO area rapidly with significant reduced neovascular tufts³⁸, which bring some difficulties to the research. In C57BL/6 mice, there is increased photoreceptor damage when compared to BALB/cJ mouse strain^39,40. The same goes for different types of transgenic mice^41,42,43. Besides, C57BL/6 mice display a lower level of angiogenesis when compared to 129S3/SvIM mice⁴⁴.

Postnatal weight gain (PWG) is also important to consider⁴⁵ and is one of the indicators to evaluate the nutritional status of newborns. It has also become a reliable method to predict ROP, which attracts the attention of many animal modelers⁴⁶. PWG affects the response of mice to hyperoxia and hypoxia. At P7, pups with increased body weight (>5 g) show an insufficient vaso-obliteration and retinal neovascularization, while pups with decreased body weight (<5 g) show obvious response to hyperoxia and hypoxia. Besides, at P17, pups with poor (<5 g) and extensive (>7.5 g) weight gain show a decreased NV. However, pups with poor weight gain (<5 g) have significantly prolonged vaso-obliteration (VO) and neovascularization (NV) stage with a delay in the occurrence of NV peak⁴⁵. Therefore, it's necessary to record and control the PWG of pups at P7 and P17 and eliminate pups with low PWG (< 6 g at P17) to ensure the repeatability and comparability of the experiment.

The litter size has a greater impact on PWG, and some researchers suggest it should be limited to 6-8 pups/dam to meet the requirements for PWG^22,31. The state of the nursing mother needs a consideration as well. Nursing mothers are more likely to die from lung damage in a hyperoxic environment⁴⁷. If nursing mothers die or neglect their pups during and after the induction of OIR, pups will easily lose weight or even die due to the lack of nutrition³². Therefore, it is necessary to ensure that there are enough surrogate mothers to replace them. However, these surrogate mothers are suggested to be used only when the mother expires, which usually happens during the period of hyperoxia exposure or return to the room air²². Providing adequate food for nursing mothers is also helpful to improve the nutritional status of their pups.

A useful note to prepare the retinal flat mounts is that an optimal time of fixation is usually necessary for further long-time staining. As mice of P12-P25, a 15 min + 45 min fixation at room temperature is recommended²⁹. Fixing the retina at 4 °C overnight is an alternative if time is limited. Besides, the permeable and blocking buffer with a higher concentration of 1% Triton X-100 and 5% normal donkey serum effectively reduce the background of immunofluorescence staining according to our experience.

Isolectin B4 staining and FITC-dextran perfusion are commonly used methods to visualize and quantify the neovascular^48,49. A major limitation of these two methods is that the mice must be sacrificed. So, the methods for in vivo imaging and quantification of NV are needed²⁹. Paques et al. developed a technique named topical endoscopy fundus imaging (TEFI), which provides high-resolution digital photographs of the retina in live mice⁵⁰. The TEFI can detect retinal vascular changes as early as P15 and the images obtained are in accordance with the conventional methods of assessment. Mezu-Ndubuisi et al. then provided the methods for in vivo retinal vascular oxygen tension (PO₂) measurements and fluorescein angiography (FA), improving the understanding of retinal vascular changes and oxygenation alterations due to ROP and other ischemic retinal diseases³⁷. Although neither TEFI nor FA is as accurate as conventional methods, they reduce the death of experimental animals and can be performed repeatedly. Besides, they allow each mouse to serve as its own control, thus making the OIR data more comparable. In this paper, an improved method of FFA imaging and image stitching is provided. Performing FFA on pups within 1 month is not easy because excessive anesthesia and hypothermia directly cause the death of the pups. Thus, try to use the minimum dose of anesthesia and pay special attention to maintaining the body temperature of pups throughout and after the process by using a small heating pad. Always moisten the ocular surface with saline and Hypromellose in case of failure of the following observation.

In summary, the OIR mouse model is a very common and widely used model of retinal ischemia and pathological neovascularization. One of the major problems of this model is that the neonatal mice pups are essentially healthy and do not have metabolic instability or respiratory problems when compared to prematurely born infants. Another difference between the OIR mouse model and humans is that there is always fibrovascular proliferation in human retinal neovascularization whereas the retinal neovascular is not associated with fibrosis in the OIR mouse model⁵¹. To make better use of this model and acquire more information, a detailed description of using FFA to monitor the dynamic changes of OIR retinal vasculature is provided, including the methods of taking "Five-orientation" images and image processing. It is believed that FFA will become an effective method partially or fully to replace the immunofluorescence staining to observe and evaluate the morphology and function of retinal vasculature⁴⁹. Although the OIR mouse model doesn't fully resemble the microenvironment and pathogenesis of various ischemic retinopathy in humans, it provides us with an opportunity to conduct drug and transgenic experiments as well as to explore the mechanism of pathological angiogenesis on the ischemic retina⁵¹.

The authors have nothing to disclose.

We thank all the members from our lab and Ophthalmic Animal Laboratory of Zhongshan Ophthalmic Center for their technical assistance. We also thank Prof. Chunqiao Liu for experimental support. This work was supported by grants from the National Natural Science Foundation of China (NSFC: 81670872; Beijing, China), the Natural Science Foundation of Guangdong Province, China (Grant No.2019A1515011347), and High-level hospital construction project from State Key Laboratory of Ophthalmology at Zhongshan Ophthalmic Center (Grant No. 303020103; Guangzhou, Guangdong Province, China).