Here, we describe an optimized protocol for retinal vein occlusion using rose bengal and a laser-guided retinal imaging microscope system with recommendations to maximize its reproducibility in genetically modified strains.
Mouse models of retinal vein occlusion (RVO) are often used in ophthalmology to study hypoxic-ischemic injury in the neural retina. In this report, a detailed method pointing out critical steps is provided with recommendations for optimization to achieve consistently successful occlusion rates across different genetically modified mouse strains. The RVO mouse model consists primarily of the intravenous administration of a photosensitizer dye followed by laser photocoagulation using a retinal imaging microscope attached to an ophthalmic guided laser. Three variables were identified as determinants of occlusion consistency. By adjusting the wait time after rose bengal administration and balancing the baseline and experimental laser output, the variability across experiments can be limited and a higher success rate of occlusions achieved. This method can be used to study retinal diseases that are characterized by retinal edema and hypoxic-ischemic injury. Additionally, as this model induces vascular injury, it can also be applied to study the neurovasculature, neuronal death, and inflammation.
Retinal vein occlusion (RVO) is a common retinal vascular disease that affected approximately 28 million people worldwide in 20151. RVO leads to vision decline and loss in working aged adults and elders, representing an ongoing sight-threatening disease estimated to increase over the proximate decade. Some of the distinct pathologies of RVO include hypoxic-ischemic injury, retinal edema, inflammation, and neuronal loss2. Currently, the first line of treatment for this disorder is through the administration of vascular endothelial growth factor (VEGF) inhibitors. While anti-VEGF treatment has helped ameliorate retinal edema, many patients still face vision decline3. To further understand the pathophysiology of this disease and to test potential new lines of treatment, there is a need to constitute a functional and detailed RVO mouse model protocol for different mouse strains.
Mouse models have been developed implementing the same laser device used in human patients, paired with an imaging system scaled to the correct size for a mouse. This mouse model of RVO was first reported in 20074 and further established by Ebneter and others4,5. Eventually, the model was optimized by Fuma et al. to replicate key clinical manifestations of RVO such as retinal edema6. Since the model was first reported, many studies have employed it using the administration of a photosensitizer dye followed by photocoagulation of major retinal veins with a laser. However, the amount and type of the dye that is administered, laser power, and time of exposure vary significantly across studies that have used this method. These differences can often lead to increased variability in the model, making it difficult to replicate. To date, there are no published studies with specific details about potential avenues for its optimization.
This report presents a detailed methodology of the RVO mouse model in the C57BL/6J strain and a tamoxifen-inducible endothelial caspase-9 knockout (iEC Casp9KO) strain with a C57BL/6J background and of relevance to RVO pathology as a reference strain for a genetically modified mouse. A previous study had shown that non-apoptotic activation of endothelial caspase-9 instigates retinal edema and promotes neuronal death8. Experience using this strain helped determine and provide insight towards potential modifications to tailor the RVO mouse model, which can be applicable to other genetically modified strains.
This protocol follows the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and vision research. Rodent experiments were approved and monitored by the Institutional Animal Care and Use Committee (IACUC) of Columbia University.
NOTE: All experiments used two-month-old male mice that weighed approximately 20 g.
1. Preparation and administration of tamoxifen for inducible genetic ablation of floxed genes
NOTE: Retinal vessel diameter can be affected by the weight of the animal. Make sure that all animals used for an experiment are of similar weights.
2. Preparation of reagents for laser photocoagulation
3. Laser setup
4. Mouse tail vein injection of rose bengal
5. Occlusion of major veins
6. Establishing the number of veins occluded at day 0
7. Aftercare
8. Assessment of retinal edema by optical coherence tomography (OCT)
NOTE: This step can be done at the investigator's time point of interest. The peak of retinal edema for a C57BL/6J mouse is 1 day after the RVO procedure. This time point might vary depending on the background of the mouse.
The RVO mouse model aims to successfully achieve occlusions in the retinal veins, leading to hypoxic-ischemic injury, breakdown of the blood retinal barrier, neuronal death, and retinal edema8. Figure 1 shows a timeline of steps to ensure reproducibility, a schematic of the experimental design, and outlines steps that can be further optimized depending on the experimental questions. The three main steps that can be modified are the waiting time after rose bengal administration, the baseline laser power, and the experimental laser output. In this report, C57BL/6J mice, as well as WT and KO littermates from an inducible endothelial cell caspase-9 knockout mouse line (iEC Casp9), were used to determine the optimal settings across different strains.
The wait time from rose bengal injection to laser irradiation can alter the success of photocoagulation in the veins. Too short a wait time may result in low rose bengal concentration within the veins, whereas too long a wait time can lead to rose bengal being cleared from the retinal circulation. Both situations can lead to poor photocoagulation and unsuccessful occlusions. When testing the number of occlusions obtained immediately after laser irradiation, comparing animals that were lasered 10 and 20 min after rose bengal administration revealed that there were no differences in the number of occlusions achieved (Figure 2B). However, the number of occlusions sustained up to 1 day after RVO significantly decreased in animals that were lasered 20 min after rose bengal administration independently of genotype. This result suggests that when studying acute RVO-induced injury, the waiting time after rose bengal administration can impact occlusion stability. Early reperfusion of veins (prior to 24 h after injury) could impact the development of retinal edema and therefore, should be controlled by determining the correct waiting time from rose bengal administration to laser irradiation.
In principle, successful photocoagulation leading to occlusion is driven by laser power. Although this is such an important part of the process, it is also one of the greatest sources of variability in the model and should be optimized for consistency. To accomplish this, it is recommended to measure the laser output during setup before the mice are injected with rose bengal. The recommended output for the baseline laser power is between 13.0 and 15.0 mW. Low baseline laser power, such as 11.5 mW, without modifying the experimental power (100 mW), resulted in no occlusions, as shown in Figure 3. In contrast, a baseline laser output of 13.5 mW with an experimental power of 100 mW resulted in successful occlusions. In cases where the laser output was below 13.0 mW, the experimental power was increased to 110 mW to achieve the same successful occlusions as with higher baseline laser output. Typically, 100 mW is the standard experimental power; however, if the laser output is below 13.0 mW, it can be compensated by modifying the experimental power with the recommended ranges in Table 1.
Four main types of occlusions have been noted to occur after laser photocoagulation of the veins. These types of occlusions are summarized in Figure 4A and have been classified according to the amount of blood flow; fully occluded vessels (no blood flow), partially occluded vessels (mostly blocked with occasional flow), partially reperfused (uninterrupted steady blood flow with hindrance), and fully reperfused vessels (no obvious obstruction whatsoever). To investigate if the types of occlusions change according to the genotype and determine the time spent per occlusion state, 10 min videos after laser irradiation were evaluated. This assessment helped determine that the irradiated vasculature of iEC Casp9 mice spend more time in partially reperfused and partially occluded states than C57BL6/J, which spend more time in fully occluded states (Figure 4B).
Figure 5 demonstrates how the occlusion state of the vessels changes rapidly within the first 10 min after laser photocoagulation. Once these initial 10 min have passed, the occlusions stabilize and are maintained up to a 24 h time point. Thus, to assess the accurate initial number of occlusions per eye, it is recommended to wait for 10 min after irradiation. A previous assessment of this model determined that most occlusions reperfuse by 8 days after irradiation8; however, the rate of occlusions that reperfuse per day can vary by strain and must be determined in each experimental model. The model presented here is of acute injury and intended to be used to understand pathways that lead to edema, which develops within 24 h after occlusions. Another characteristic of RVO is flame-shaped hemorrhages, which can be observed at 24 h post injury, as depicted in Figure 5.
Follow-up at 24 h post RVO may reveal other ophthalmic pathologies that can occur as a result of the RVO method. Some include but are not limited to subretinal hemorrhage (characterized by continuous blood patch), retinal detachment, fully ischemic retina (no blood flow in veins and arteries), and cataracts. Figure 6 shows fundus images with corresponding OCT as examples of these except in eyes where cataracts form (Figure 6F), as OCT cannot be performed in the presence of a cataract. Figure 6A shows examples of what a fundus image and OCT of an uninjured eye look like for reference.
The main morphologic pathology of RVO in this model is retinal edema. To assess the level of retinal edema, it is recommended to take OCT images before the day of the RVO procedure for a baseline reading and at the time point of interest. Figure 7 shows OCT quantification of retinal edema in injured eyes. Another measure used to determine the state of neuronal layers is to assess the disorganization of retinal inner layers (DRIL). This is a measure used in the clinical setting that represents capillary nonperfusion, another hallmark of RVO5,9,10. An example of this assessment can be found in Figure 7B. Figure 7C shows an example of an OCT image with the corresponding labels for each retinal layer.
Figure 1: Timeline and schematic of the RVO mouse model. (A) Timeline of events from rose bengal administration to the imaging of occluded veins. (B) Summarized representation of the method to achieve successful retinal vein photocoagulation. Red boxes represent important steps in the process that are highly variable and that can be optimized per mouse model and question of interest. (C) Retinal major veins (V) are wider and darker compared to arteries (A). Each major vein will be irradiated with a guided laser of 532 nm, spot size 50 µm, power 100 mW, duration 1 s, total energy 0.3 J, and radiant exposure 15278.87 J/cm2, at an average distance of 375 µm from the optic nerve. (D) Laser application causes a vaporization bubble visible on fundus imaging of approximately 150 µm and covers <4% of the total retinal area. Numbers represent the suggested location and direction of movement (arrows) of the laser beam when irradiating vessels. Abbreviations: A = artery; V = vein; ON = optic nerve. Please click here to view a larger version of this figure.
Figure 2: The time of photo-occlusion relative to rose bengal administration is critical for successful photocoagulation. (A) Fundus retinal images of iEC Casp9 WT and iEC Casp9 KO photo-occluded 10 and 20 min after rose bengal administration. White circles represent veins that had successful occlusions. (B) Number of occlusions immediately after irradiation (0 h) and 1 day post irradiation for 10 min and 20 min post rose bengal injection of combined genotypes. Welch's t-test, error bars indicate SEM. (C) Number of occlusions separated by genotype. Two-way ANOVA and Fisher's LSD test; error bars indicate SEM. Abbreviations: WT = wild-type; KO = knockout; SEM = standard error of the mean; ANOVA = analysis of variance; LSD = least significant difference; ns = not significant; P-RVO = post retinal vein occlusion. Please click here to view a larger version of this figure.
Figure 3: Measuring baseline and experimental laser output are critical steps for successful photocoagulation. Fundus retinal images of iEC Casp9 WT and iEC Casp9 KO 10 min after photocoagulation; photo-occluded with different baseline and experimental laser output levels. Low baseline laser output can be compensated with experimental laser output (12.8 mW, 110 mW). White circles represent veins that had successful occlusions. Abbreviations: WT = wild-type; KO = knockout. Please click here to view a larger version of this figure.
Figure 4: The RVO method results in different types of occlusions. (A) Fundus retinal images of C57BL/6J, iEC Casp9 WT, and iEC Casp9 KO 10 min after photocoagulation with different types of occlusions: fully occluded, partially occluded, partially reperfused, and fully reperfused. Insets show a focused view of a vein that resulted in a specific type of occlusion after photocoagulation. (B) Quantification of the percentage of veins occluded in the different occlusion states for each genotype during the first 10 min post irradiation. Ten-minute videos were evaluated by two investigators, blinded to genotypes, who assigned numbers to the different occlusion states (fully occluded (-2), partially occluded (-1), fully reperfused (2), and partially reperfused (1)) per duration. Abbreviations: RVO = retinal vein occlusion; WT = wild-type; KO = knockout. Please click here to view a larger version of this figure.
Figure 5: Timeline of occlusions after RVO. Fundus retinal images of C57BL/6J, iEC Casp9 WT, and iEC Casp9 KO 0, 5 min, 10 min, and 24 h after laser irradiation. The first 10 min are critical for the state of the occlusions and may change rapidly. After the initial 10 min, occlusions are stable up to at least 24 h. White circles represent veins that had successful occlusions, and yellow arrowheads depict flame-shaped hemorrhages. Abbreviations: RVO = retinal vein occlusion; WT = wild-type; KO = knockout. Please click here to view a larger version of this figure.
Figure 6: Different ophthalmic pathologies may occur after RVO. (A–E) show fundus retinal images and corresponding OCT. (A) An example of an uninjured eye that did not undergo the RVO process. (B) Subretinal hemorrhage showing blood leaking out of the vessels in the fundus image. (C) Retinal detachment seen by the blurry fold in the fundus and the lifting of the retina in the OCT. (D) Excessive edema shown by a large amount of swelling in the OCT. (E) A fully ischemic eye with completely impaired blood flow resulting in a white retina. (F) Two different examples of a cataracted eye where a clear fundus image and OCT could not be obtained. OCT scale bars: 100 µm. Abbreviations: RVO = retinal vein occlusion; OCT = optical coherence tomography. Please click here to view a larger version of this figure.
Figure 7: Quantification of OCT images. (A) Examining the layer thickness and the DRIL in unlasered control eyes and eyes that went through the RVO procedure. GCL, IPL, INL, OPL, ONL, Outer segment, and Whole Retina measurements. Statistics, Mann-Whitney test p-values: GCL: 0.0070, IPL: 0.0205, INL: <0.0001, OPL: 0.0014, ONL: 0.5582, Outer Segment: 0.44852, Whole Retina:0.0019. Error bars show SEM. (B) DRIL quantification measured from the OCT images from unlasered controls and RVO WT and KO iEC Casp9 mice as well as c57/BL6J mice that had RVO. Error bars show SEM. (C) Example OCT with the labels of each retinal layer. Abbreviations: DRIL = Disorganization of the inner retinal layers; RVO = retinal vein occlusion; WT = wild-type; KO = knockout; GCL = Ganglion Cell Layer; IPL = Inner Plexiform Layer; INL = Inner Nuclear Layer; OPL = Outer Plexiform Layer; ONL = Outer Nuclear Layer; SEM = standard error of the mean; OCT = optical coherence tomography; ns = not significant. Please click here to view a larger version of this figure.
Baseline Laser Output (mW) | Recommended Experimental Laser Output (mW) | Recommended Time Exposure (ms) |
<11.0 or >15.0 | Turn off the laser and adjust the fiber at the end that is connected to the laser control box. Unscrew it and slightly move it towards the right or left. Measure the outcome again, until it reaches a higher or lower value. | |
11.0-12.0 | 120 | 1,000 |
12.0-13.0 | 110 | 1,000 |
13.0-14.0 | 100 | 1,000 |
14.0-15.0 | 100 | 1,000 |
Table 1: Low baseline laser output can be compensated with higher experimental laser output. Variations in baseline laser output and recommended experimental laser output and exposure time.
The mouse RVO model provides an avenue to further understand RVO pathology and to test potential therapeutics. While the mouse RVO model is widely used in the field, there is a need for a current detailed protocol of the model that addresses its variability and describes the optimization of the model. Here, we provide a guide with examples from experience on what can be altered to get the most consistent results across a cohort of experimental animals and provide reliable data.
The two most essential elements of the RVO mouse model are the laser output and successful intravenous injection of the photosensitizer dye. To produce the power necessary to induce coagulation when the laser is aimed at a particular vein, the laser output has to be properly adjusted. While this can be achieved using the techniques suggested in the method, it is important to consider the differences in each laboratory's system setup. Variations of the fiber-optic cable and how it is accommodated in relation to the equipment and room temperature are some of the variables that can account for low laser output. Independent tunings to the system setup are recommended to increase the laser output.
However, this effort is inoperable without an apt tail vein technique to deliver the photosensitizer dye. Tail vein injections can be difficult to achieve, and it is a skill that takes time to develop. Poor injections can result in no occlusions; in this case, rose bengal can be administered via IP. Rose bengal administration via IP has been used to model RVO but with a longer laser irradiation time (3 s) and higher rose bengal concentration (40 mg/mL)11. To limit prolonged laser irradiation and specifically target the vasculature, tail veining is the preferred mode of administration.
This model could also be accomplished using other photoactivatable dyes such as Y eosin and sodium fluorescein12,13,14, although rose bengal is the most used photoactivatable dye4,5,6,8. All dyes have been shown to produce early features of clinical disease such as retinal hemorrhage and retinal edema15. Photoactivatable dyes have been shown to have no adverse effects on the animals, thus showing insignificant system toxicity15,16. It should also be noted that the dye chosen should have an absorption maximum compatible with the wavelength of the laser being used. Rose Bengal has an excitation at 525 nm17, sodium fluorescein at 475-490 nm18, and Y eosin at 490 nm19.
The main sources of variability in this model are the time of photo-occlusion relative to rose bengal administration and the baseline and experimental laser output. While Figure 2 shows 10- and 20-min time points for photo-occlusion, a small number of 5- and 15-min experiments were also performed (data not shown), yielding occlusions that were not as consistent as the 10 min time point. Therefore, 10 min was chosen to be the optimal wait time between rose bengal administration and photo-occlusion for this method. However, studies have reported that RVO can also be induced as early as 3 min after the administration of rose bengal5. Another way to determine the mouse strain-specific optimal waiting time from rose bengal to photo-occlusion is to monitor the relative rose bengal concentration using the fluorescence imaging mode with a tetramethylrhodamine (TRITC) filter. However, the protocol described here can be performed using retinal imaging microscopes that do not have a TRITC filter.
The other source of high variability in this model is the baseline laser output. As day-to-day levels of baseline laser output can be vastly different, it is important to assess the levels before each experiment. Standardizing the baseline laser output across studies can help potentiate and expand the use of the RVO mouse model. Readjusting the fiber cable can be enough to modify the baseline levels; however, if a 13.0 mW measurement cannot be reached, Table 1 provides a guide for compensation using the experimental power. It is important to note that because the retina is a closed system, determining the fraction of veins occluded (the number of veins occluded divided by the number of veins irradiated) is essential for understanding, controlling, and predicting the severity of the damage in the RVO model. Previous analysis of damaged readouts (DRIL and retinal thickness) correlated to the fraction of veins occluded and predicted retinal atrophy at 8 days post RVO8. Thus, the fraction of veins occluded should be considered and evaluated. It is still unclear how other intermediate occlusion states, such as partially reperfused, partially occluded, or veins that were once occluded and reperfused by 10 min, contribute to the development of retinal edema and atrophy.
Further studies of the eyes with these types of occlusions can help investigate whether a sustained occlusion is necessary for substantial damage or if even transient occlusions are important in this model. Depending on the experimental question being asked, different occlusion rates will be optimal. A 40-50% occlusion rate is ideal in most cases, meaning two or three occlusions in an eye with six veins. This ensures substantial injury, but the retina is intact and can be dissected for immunohistochemical and biochemical analyses.
To determine this, distinguishing the pathological view of a successful and representative occlusion of the RVO signature is relevant. The natural occlusion presented by RVO includes flame-shaped hemorrhages20 (not to be confused with subretinal hemorrhage), which can be observed in this model 24 h after injury. The RVO model can also lead to unwanted retinal injuries (not distinctive of RVO pathology) such as the ones shown in Figure 6B-F, if its parameters are not cautiously controlled. An approach that can be taken to avoid these, besides regulating the concentration of rose bengal and the experimental laser power, is to stop irradiating the vessels that have a clear formation of a thrombus after the first or second irradiation.
Other factors to consider when employing this model and deciding which veins to occlude are the mouse strain and the vessel diameter. Some mouse strains, such as BALB/c, most commonly known as albino, are susceptible to light damage21. Additionally, they have retinal developmental deficits that lead to defects in the decussation at the optic chiasm and visual acuity22,23. It is recommended to fully evaluate basal retinal and vascular integrity of uninjured controls for the mouse strain chosen for RVO studies. Studies have shown that vein width can interfere with the development of retinal edema and disease pathology24. Thus, animals of similar weight should be used to avoid further variability. The exclusion factors of which eyes to use in a study will also depend on the experimental question. It may be wise to include any eye that was once occluded, even if it is reperfused at the follow-up time point for signaling. However, if stable occlusions are desired, these eyes would be excluded. Figure 6 shows examples of possible exclusion criteria or eyes that could be used for pathological analysis.
To show that these laser settings were not causing damage and the effects seen were truly driven by the occlusions themselves, sham controls were done in previous studies8. The sham mice still received a tail vein injection of rose bengal but were irradiated in the parenchymal space between the major vessels instead of being irradiated on the vein. This was an important step in ensuring that the model replicated RVO injuries seen in patients instead of simply injuring tissue with a laser. These shams show no activation of caspase-9 or -7 or any of the edema seen in the mice who received normal laser irradiation to the vessels, indicating that the laser did not have any adverse effects. Having these controls is essential, especially if higher laser settings will be used to ensure that the injury being modeled is an accurate representation of the desired damage8.
The RVO mouse model can be applied to study other diseases that result from hypoxic-ischemic injuries in the retina and brain such as diabetic retinopathy, retinopathy of prematurity, and stroke. Additionally, it can serve as a model in which to study signaling pathways relevant for vascular injury development and test potential treatments that ameliorate neurodegeneration in the central nervous system. The optimization tailored in this report can limit variability in the mouse RVO model and shed light on the pathophysiology of RVO.
The authors have nothing to disclose.
Carprofen | Rimadyl | NADA #141-199 | keep at 4 °C |
Corn Oil | Sigma-Aldrich | C8267 | |
Fiber Patch Cable | Thor Labs | M14L02 | |
GenTeal | Alcon | 00658 06401 | |
Ketamine Hydrochloride | Henry Schein | NDC: 11695-0702-1 | |
Lasercheck | Coherent | 1098293 | |
Phenylephrine | Akorn | NDCL174478-201-15 | |
Phoneix Micron IV with Meridian, StreamPix, and OCT modules | Phoenix Technology Group | ||
Proparacaine Hydrochloride | Akorn | NDC: 17478-263-12 | keep at 4 °C |
Refresh | Allergan | 94170 | |
Rose Bengal | Sigma-Aldrich | 330000-5G | |
Tamoxifen | Sigma-Aldrich | T5648-5G | light-sensitive |
Tropicamide | Akorn | NDC: 174478-102-12 | |
Xylazine | Akorn | NDCL 59399-110-20 |