In Vivo Vascular Injury Readouts in Mouse Retina to Promote Reproducibility

Summary

Here, we present three data analysis protocols for fluorescein angiography (FA) and optical coherence tomography (OCT) images in the study of Retinal Vein Occlusion (RVO).

Abstract

Advancements in ophthalmic imaging tools offer an unprecedented level of access to researchers working with animal models of neurovascular injury. To properly leverage this greater translatability, there is a need to devise reproducible methods of drawing quantitative data from these images. Optical coherence tomography (OCT) imaging can resolve retinal histology at micrometer resolution and reveal functional differences in vascular blood flow. Here, we delineate noninvasive vascular readouts that we use to characterize pathological damage post vascular insult in an optimized mouse model of retinal vein occlusion (RVO). These readouts include live imaging analysis of retinal morphology, disorganization of retinal inner layers (DRIL) measure of capillary ischemia, and fluorescein angiography measures of retinal edema and vascular density. These techniques correspond directly to those used to examine patients with retinal disease in the clinic. Standardizing these methods enables direct and reproducible comparison of animal models with clinical phenotypes of ophthalmic disease, increasing the translational power of vascular injury models.

Introduction

Neurovascular disease is a major healthcare problem responsible for ischemic strokes, a leading cause of mortality and morbidity, and retinal vascular diseases that lead to vision loss^1,2. To model neurovascular disease, we employ a mouse model of retinal vein occlusion (RVO). This model is noninvasive and utilizes similar in vivo imaging techniques to those used to examine people with retinal vascular disease in a clinical setting. The use of this model thus increases the translational potential of studies utilizing this model. As with all mouse models, it is critical to maximize reproducibility of the model.

Retinal vascular diseases are a major cause of vision loss in people under the age of 70. RVO is the second most common retinal vascular disease after diabetic retinopathy³. Clinical features characteristic of RVO include ischemic injury, retinal edema, and vision loss as a consequence of neuronal loss^3,4. Mouse models of RVO using laser photocoagulation of major vessels have been developed and refined to replicate key clinical pathologies observed in human RVO^5,6,7. Advancements in ophthalmic imaging also allow for replication of noninvasive diagnostic tools used in humans, namely, fluorescein angiography (FA) and optical coherence tomography (OCT)⁶. Fluorescein Angiography allows for the observation of leakage due to the breakdown of the blood-retinal barrier (BRB) as well as blood flow dynamics in the retina, including sites of occlusion, using the injection of fluorescein, a small fluorescent dye^8,9. OCT imaging allows for the acquisition of high-resolution cross-sectional images of the retina and the study of the thickness and organization of retinal layers¹⁰. Analysis of FA images has historically been largely qualitative, which limits the potential for direct and reproducible comparison between studies. Recently, a number of methods have been developed for the quantification of layer thickness in OCT imaging, though there is currently no standardized analysis protocol and the site of OCT image acquisition varies¹¹. In order to properly leverage these tools, standardized, quantitative, and replicable data analysis methodology are needed. In this paper, we present three such vascular readouts used to evaluate pathological damage in a mouse model of RVO-fluorescein leakage, OCT layer thickness, and disorganization of retinal layers.

Protocol

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: Imaging was done on 2 month old C57BL/6J male mice that weighed approximately 23 g.

1. Preparation of reagents for retinal imaging

1. Preparation of injectable fluorescein solution.
   NOTE: Fluorescein is very light-sensitive. Protect from light and use it shortly after preparation.
   1. Dilute fluorescein to a concentration of 1% in sterile saline.
2. Preparation of Ketamine/Xylazine
   1. Dilute Ketamine and Xylazine in sterile saline accordingly for the following concentration: Ketamine (80-100 mg/kg) and Xylazine (5-10 mg/kg).
3. Sterile saline
   1. Prepare a 5 mL syringe with a 26 G needle with sterile saline.

2. OCT and fluorescein imaging

1. Turn the retinal imaging microscope lightbox, the OCT machine, and the heated mouse platform ON.
2. Turn the computer ON and open the imaging program.
3. Add one drop of phenylephrine and tropicamide to each eye.
4. Inject 150 µL of anesthesia (Ketamine (80-100 mg/kg) and Xylazine (5-10 mg/kg)) intraperitoneally (IP). Determine the depth of anesthesia by toe pinch and wait until the animal is unresponsive. Apply ophthalmic ointment or artificial tears to both the eyes.
5. Accommodate the mouse on the platform.
6. Adjust the height and angle of the platform until the view of the retinal fundus is clear and focused. Take a picture of the fundus.
7. Open the imaging and OCT software. In the OCT program, adjust nudge to 5.
8. Take an OCT image at 75 µm distal from the burn. Repeat for the other three quadrants of the retina.
9. Inject 100 µL of 1% fluorescein IP.
10. Switch the camera to a 488 nm filter. Increase the camera gain to 5.
11. Take a picture of the fundus at exactly 5 min after fluorescein injection.
    ​NOTE: Avoid prolonged exposure of the eye to the camera light at maximum setting, as fluorescein can exacerbate retinal photodamage. Keep light source off until the 5 min wait time has elapsed and the mouse is ready for imaging.

3. Aftercare

1. Inject 1 mL of sterile saline IP. Apply lubricant eye drops to both the eyes. Apply ophthalmic ointment or artificial tears to both the eyes.
2. Observe the mouse as it recovers from anesthesia. Return to the cage with other animals only when fully recovered, generally after around 40 min.

4. Assessment for exclusion criteria

1. Open the fundus image taken at 24 h post-procedure to assess for exclusion criteria. Exclude the eye if any of the following criteria are identified.
2. Assess whether the image has zero occlusions
   1. Evaluate the image for the number of occluded vessels.
      NOTE: A successful occlusion usually has some purple pigmentation on or around the burn, very thin or discontinuous vessel through the burn, faint or non-existent vessel appearance outside the burn area, and retinal discoloration from hypoxia. If the entire vessel can be seen through the white burn by the laser, the vessel failed to occlude. Sometimes the vessel will appear partially obstructed, but if it looks uninterrupted outside the burn, the vessel likely didn't occlude.
   2. For ambiguous cases, use FA imaging at the same time point to evaluate occlusions. In these images, an occlusion will appear as a break in the continuity of a vessel, often with a tapering of the surrounding vessel.
   3. If zero occlusions are identified, exclude the eye from analysis, as the RVO is considered ineffective.
      NOTE: Occlusions typically resolve by 48-72 h post-RVO, and the presence of occlusions should no longer be used as an exclusion criterion at these time points.
3. Assess the fundus and OCT images for excessive retinal detachment
   NOTE: Subretinal fluid accumulation is common after induction of RVO, and causes separation of neural retina from RPE. Exclusionary criteria for excessive retinal detachment are defined as follows: OCT will either be completely unviewable, or some layers will appear incredibly distorted. Image quality is poor, with a loss of resolution of outer plexiform and RPE layers. The separation between the neural retina and the choroid is greater than what the OCT field of view allows. On the fundus image, the retina tone will be nearly completely white, with some purple blotching. Part of the retina may appear distorted and out of focus. This is because it has detached and is at a different focal distance than the rest of the retina.
   1. If the assessment of the images from an eye determines peripheral or complete detachment of the retina, exclude the eye from the analysis.
4. Exclude images with evidence of corneal cataract
   NOTE: A corneal cataract appears as an opaque white dot on the mouse's cornea. Cataracts typically occur due to insufficient lubrication of the eyes while the animal is anesthetized and can be largely avoided by taking care to apply eye ointment generously. Cataracts can generally be identified before imaging by inspecting the animal. Mice that have developed cataracts should be excluded from the dataset without needing to undergo the imaging process. In imaging, cataracts will obscure the retina from the camera, and the OCT will appear warped.
5. Assess the image for excessive hemorrhage
   ​NOTE: Excessive hemorrhage can be identified as amounts of red fluid in the image, usually obscuring retinal background, vessel, and burn. These areas of red fluid will be a brighter, opaquer red than the purple splotches that are normal in successful RVO. Hemorrhages show up at the ganglion cell layer on OCT imaging and interfere with the ability to visualize other retinal layers beneath the hemorrhage.
   1. If the image is determined to have an excessive hemorrhage, exclude the eye from the analysis.

5. Fluorescein image processing

1. Open the fluorescein image in the image processing software.
2. Duplicate the image
3. Using a selection tool, carefully trace the major vessels.
   1. The major vessels are the thicker veins and arteries radiating out from the optic disc. Ignore any vessels branching out from these vessels.
   2. If leakage prevents the outline of the vessel from being seen near the occlusion site, trace through the leakage in the approximate location of the vessel (maintain thickness, connect the last visible point to the next visible point).
4. In the first image, delete the selection, leaving only the background. Save this masked image.
5. Move the selection to the second image, invert the selection and delete, isolating the vessels. Save this masked image.
6. Open the two images in ImageJ. Open the background image and measure the integrated density.
7. Open the vessel's image, select the outline of the vessels, and then measure the mean intensity.
8. Divide the integrated density of the background by the mean intensity of the vessels, generating the leakage ratio for the eye.
9. Record this leakage ratio for each eye in an experimental cohort.
10. To further control for background, normalize experimental eyes to the mean leakage ratio of uninjured control eyes.
    ​NOTE: In order to create a standardized quantification of fluorescein leakage in the FA image, this calculation uses a ratio of the background density (where the leakage will be present) with the brightness of the major vessels to create results that control for the variation in brightness from image to image and can be reliably quantified. Eyes that are undamaged have no leakage and should theoretically have ratios of zero. The ratios calculated from these undamaged control eyes, therefore, represent background noise, and this value is used to further normalize experimental values.

6. Retinal layer thickness

1. Open the OCT image in the image processing software.
2. Trace the borders of the ganglion cell layer, inner plexiform layer, inner nuclear layer, outer plexiform layer, photoreceptor layer, and RPE layer. Measure the mean thickness of each layer.
3. Repeat for OCT images from the other three quadrants of the retina. Average the mean layer thicknesses across the four quadrants to obtain the mean thickness of each retinal layer for the eye.
4. Repeat for each eye in the experimental cohort.

7. Disorganization of retinal inner layers (DRIL)

1. Open the OCT image in ImageJ.
2. Using the line tool, measure the distance where the upper border of the outer plexiform layer is indistinct.
   NOTE: It is important to differentiate between DRIL and areas of poor layer visibility caused by imaging artifacts. Poor OCT image quality may invalidate an eye for DRIL analysis if sufficient image resolution is not possible. Images with DRIL will typically have other regions or retinal layers that are clearly resolved and organized, which can be a good indicator of sufficient image quality.
   1. Measure horizontally from the latitude where the disorganization begins to the latitude where the upper border of the outer plexiform layer becomes visible again, if at all. Even if the outer plexiform layer shifts upward or downward vertically, measure perfectly horizontally.
   2. There may be multiple areas of disorganization separated by areas with no disorganization. Measure these individually and calculate the sum of the distances.
3. Divide the length of disorganization by the total length of the retina visible in each OCT image to obtain the ratio of disorganization for the image.
4. Repeat the measurement and calculation for OCT images from the other three quadrants of the retina.
5. Take the mean of the ratios of disorganization from the four OCT images. This number represents the average disorganization for the whole retina. Repeat for each eye in the experimental cohort.

Results

These analysis methods allow for the quantification of retinal pathology captured by FA and OCT imaging. The experiments from which the representative data is extracted used C57BL/6J male mice who either served as uninjured controls or underwent the RVO procedure and received either Pen1-XBir3 treatment eyedrops or Pen1-Saline vehicle eyedrops. The RVO injury model involved the laser irradiation (532 nm) of the major veins in each eye of an anesthetized mouse following a tail-vein injection of rose bengal, a photoactivat...

Discussion

Noninvasive rodent retinal imaging presents an avenue to study pathology and develop interventions. Previous studies have developed and optimized a mouse model of RVO, limiting variability and allowing for reliable translation of common clinical pathologies in the murine retina^5,7,13. Developments in ophthalmic imaging technology further allow for the use of clinical in vivo imaging techniques such as FA and OCT in expe...

Materials

Name	Company	Catalog Number	Comments
AK-Fluor 10%	Akorn	NDC: 17478-253-10	light-sensitive
Carprofen	Rimadyl	NADA #141-199	keep at 4 °C
GenTeal	Alcon	00658 06401
Image J	NIH
InSight 2D	Phoenix Technology Group		OCT analysis software
Ketamine Hydrochloride	Henry Schein	NDC: 11695-0702-1
Phenylephrine	Akorn	NDCL174478-201-15
Phoenix Micron IV	Phoenix Technology Group		Retinal imaging microscope
Phoenix Micron Meridian Module	Phoenix Technology Group		Laser photocoagulator software
Phoenix Micron Optical Coherence Tomography Module	Phoenix Technology Group		OCT imaging software
Phoenix Micron StreamPix Module	Phoenix Technology Group		Fundus imaging and acquisition targeting
Photoshop	Adobe
Refresh	Allergan	94170
Tropicamide	Akorn	NDC: 174478-102-12
Xylazine	Akorn	NDCL 59399-110-20