This protocol provides a comprehensive dissection and analysis guide for the use of deep ocular landmarks, s-opsin immunohistochemistry, Retistruct, and custom code to accurately and reliably orient the isolated mouse retina in anatomical space.
Accurately and reliably identifying spatial orientation of the isolated mouse retina is important for many studies in visual neuroscience, including the analysis of density and size gradients of retinal cell types, the direction tuning of direction-selective ganglion cells, and the examination of topographic degeneration patterns in some retinal diseases. However, there are many different ocular dissection methods reported in the literature that are used to identify and label retinal orientation in the mouse retina. While the method of orientation used in such studies is often overlooked, not reporting how retinal orientation is determined can cause discrepancies in the literature and confusion when attempting to compare data between studies. Superficial ocular landmarks such as corneal burns are commonly used but have recently been shown to be less reliable than deeper landmarks such as the rectus muscles, the choroid fissure, or the s-opsin gradient. Here, we provide a comprehensive guide for the use of deep ocular landmarks to accurately dissect and document the spatial orientation of an isolated mouse retina. We have also compared the effectiveness of two s-opsin antibodies and included a protocol for s-opsin immunohistochemistry. Because orientation of the retina according to the s-opsin gradient requires retinal reconstruction with Retistruct software and rotation with custom code, we have presented the important steps required to use both of these programs. Overall, the goal of this protocol is to deliver a reliable and repeatable set of methods for accurate retinal orientation that is adaptable to most experimental protocols. An overarching goal of this work is to standardize retinal orientation methods for future studies.
An important and sometimes overlooked aspect of retinal neuroscience is the proper orientation and analysis of the isolated whole-mount retina, whether it be the orientation of a retina in an electrophysiology recording chamber or on a histological slide. This is particularly important for studies involving the mouse retina, which is currently the most widely used model for investigations of the mammalian visual system. Recent discoveries reveal that the mouse retina is not spatially uniform but has density and size gradients of functionally-distinct retinal cell types, such as melanopsin ganglion cells, transient OFF-alpha ganglion cells, and cone opsins1,2,3,4,5. Consequently, the method used to determine the orientation of the retina may affect the experimental results involving cell type or opsin distributions2,3,6, direction tuning of direction-selective ganglion cells7,8,9, and topographic patterns of retinal degeneration10,11,12,13,14. In fact, not reporting how retinal orientation is reported can cause discrepancies in the literature and confusion when attempting to compare data between studies. It is therefore vital that researchers report the method for identifying the orientation of the retina so that results of such studies can be accurately interpreted.
Retinal orientation is commonly identified by scoring the dorsal, ventral, nasal or temporal cornea prior to ocular enucleation1,3,12,15,16,17,18,19 or by cutting or staining deep anatomical eye landmarks such as the extraocular muscles6,7, the choroid fissure20,21, or the s-opsin gradient2,3. The rectus muscles can be used to identify the dorsal, ventral, nasal, and temporal retina by making a deep relieving cut that bisects the attachment of the either the superior rectus, inferior rectus, medial rectus, or lateral rectus muscle, respectively. However, for most experiments, using one rectus muscle is sufficient for orienting the retina22. The choroid fissure, which is a remnant of eye development, can be viewed as a faint horizontal line on the back of the eye. Each end of this line terminates at either the nasal or the temporal pole of the globe23. Finally, s-opsin expression is asymmetrically distributed to the ventral retina in mice, and s-opsin antibodies can be used to reveal the ventral retina in immunohistochemical experiments1.
Recent work by Stabio, et al.22 demonstrated that superficial ocular landmarks such as corneal burns are a less reliable method for orienting the retina in anatomical space, most likely due to human error and variability in making the corneal burn when using the temporal and medial canthi as reference points. In contrast, deep landmarks, such as the superior rectus muscle, choroid fissure, and the s-opsin gradient, have been shown to be more reliable and accurate landmarks for orienting the retina22. However, the identification of these anatomical landmarks requires unique dissection steps that are not described in detail in the literature. Thus, the goal of this protocol is to provide a comprehensive tutorial on how to use the superior rectus muscle, choroid fissure, and s-opsin gradient to accurately identify the spatial orientation of the mouse retina. In addition, we have included a comparison of the effectiveness of two s-opsin antibodies, as well as a protocol for s-opsin immunohistochemistry.
One additional challenge to studies relying on precise retinal orientation is the large relieving cuts required to flatten wholemount retinas on a recording chamber, dish, or slide. This can introduce challenges for the analysis of what is naturally a three-dimensional structure when it is imaged as a flat two-dimensional structure. A program called Retistruct24 can be used to return a flat wholemount retina to its three-dimensional structure before the data collected from it is analyzed. Thus, a section of this protocol is dedicated to highlighting the steps that are necessary for using the Retistruct software to reconstruct s-opsin immunostained mouse retinas. We have also included a section of protocol for using our custom MATLAB script, which was developed to accurately rotate and orient mouse retinas stained with s-opsin.
All methods described here have been approved by the Institutional Animal Care and Use Committee (IACUC) of The University of Akron.
1. Using the Superior Rectus Muscle Landmark to Identify Retinal Orientation
NOTE: The superior rectus muscle is a landmark for the dorsal retina (Table 1). If the experiment does not require the marking of the dorsal retina, skip step 1 and continue to step 2.
2. Using the Choroid Fissure Landmark to Identify Retinal Orientation
NOTE: The choroid fissure is present on the sclera on the back of the eye, and runs from the temporal pole to the nasal pole (Figures 2B and 2C; Table 1).
3. Labeling the S-opsin Gradient in the Mouse Retina
NOTE: The s-opsin photopigment expression is asymmetrically distributed to the ventral retina1, making it an excellent marker for the ventral half of the retina. This method is only useful for fixed and immunostained tissue (Table 1). The following steps can be applied to retinas that have been dissected using any of the aforementioned methods.
4. Using Reconstructed Retinas Immunostained with S-opsin to Identify Retinal Orientation
A single relieving cut that bisects the superior rectus muscle accurately and reliably identifies the dorsal retina (Figure 1). The choroid fissure accurately and reliably identifies the nasal and temporal retina with deep relieving cuts along the temporal and nasal choroid fissure (Figure 2). In this example, a relieving cut has also been made in the dorsal retina in order to identify the dorsal/ventral axis of the retina (Figure 2D, vertical arrow). The steps of these processes are shown for purpose of replication by future dissectors. A combination of s-opsin immunohistochemistry (Figure 3A and 3D), reconstruction with Retistruct software (3B, 3E) and accurate rotation with a custom MATLAB code (3C, 3F) allows for the identification of the ventral and dorsal halves of the retina, as well as the nasal and temporal poles if it is known whether the retina is from a right or left eye (Figure 3). We also compared two commonly used s-opsin primary antibodies for effectiveness in labeling s-opsin cones (Figure 4A-D): Both the goat anti-s-opsin primary antibody and the rabbit anti-s-opsin primary antibody effectively label s-opsin cones (Figure 4E) in the same mouse.
Relieving cuts were identified on s-opsin immunostained reconstructed retinas and their locations were compared to the orientation determined by the s-opsin gradient. Using our custom MATLAB code (see Supplemental Materials), retinas were accurately rotated so that the highest concentration of s-opsin staining is located ventrally, thus placing true dorsal at 90° (for superior rectus), true nasal at 0° (for nasal choroid fissure) and true temporal at 180° (for temporal choroid fissure). The value of each individual relieving cut angle was determined using the angle tool in ImageJ after retinas were rotated according to the s-opsin gradient. An average angle was calculated for each relieving cut type and the average value of each relieving cut type was then plotted on a polar plot (Figure 6). On average, superior rectus muscle cuts identified the dorsal pole at 96.3 ± 4.3° (n = 11) (Figure 6). The nasal choroid fissure identified the nasal pole at 6.7 ± 5.8° and the temporal choroid fissure identified the temporal pole at 172.0 ± 4.4° (n = 9; Figure 6).
Figure 1: Using the superior rectus muscle to accurately identify the dorsal retina of a right eye. (A) An example of a dorsal corneal burn near the corneal-scleral border made with a cautery tip pen (white arrow). The superior rectus muscle is also visible in this view (white arrow). (B) An example of a whole mounted retina with a relieving cut made in the dorsal retina by bisecting the superior rectus muscle. Arrow depicts the deep relieving cut made in the dorsal retina by bisecting the superior rectus muscle. The retina is stained with primary antibody goat anti-s-opsin (see Table of Materials) and secondary antibody donkey anti-goat Alexa 594 (see Table of Materials; excitation: 590 nm; emission: 620 nm) (cyan). Retina was imaged with an epifluorescent microscope with a Texas Red filter (595 nm). (C) A retina reconstructed in Retistruct and rotated with a custom MATLAB code (see Supplemental Materials) with the superior rectus muscle relieving cut visible (white arrow). D: dorsal, V: ventral, T: temporal, N: nasal. Scale bars = 1 mm. Please click here to view a larger version of this figure.
Figure 2: Using the choroid fissure to accurately identify the nasal and temporal poles of the retina of a right eye. (A) An example of a dorsal corneal burn near the corneal-scleral border made with a cautery tip pen. (B) The choroid fissure visible on the back of the eye on the sclera (white arrow). The dorsal corneal burn is also visible in this view, located about 90° from the temporal choroid fissure. (C) The choroid fissure visible on the back of the eye on the sclera, traveling from the optic nerve to the corneal-scleral border. (D) A retina stained with goat anti-s-opsin (see Table of Materials) and secondary antibody donkey anti-goat Alexa 594 (see Table of Materials; excitation: 590 nm; emission: 620 nm) (cyan) with choroid fissure cuts (horizontal arrows) and the dorsal relieving cut (vertical arrow). Retina was imaged with an epifluorescent microscope with a Texas Red filter (595 nm). (E) A retina reconstructed in Retistruct and rotated with a custom MATLAB code (see Supplemental Materials) with the dorsal relieving cut and the nasal and temporal choroid fissure cuts visible (white arrows). D: dorsal, V: ventral, T: temporal, N: nasal. Scale bars = 1 mm. Please click here to view a larger version of this figure.
Figure 3: Using the s-opsin gradient to identify all four poles of the retina. (A) An example of a retina dissected from a right eye that has been immunostained to label s-opsin and imaged with an epifluorescent microscope with a Texas Red filter (595 nm). The cuts in this retina are arbitrary since the topographical orientation is determined by the s-opsin gradient. (B) The results of reconstructing the retina in A with Retistruct. Notice that the s-opsin gradient is not correctly aligned because the retina has not been run through the custom MATLAB code (See Supplemental Materials). (C) The results of rotating the retina in A with the custom code. The retina has been rotated so that the highest concentration of s-opsin staining is located at the bottom and identified as the ventral retinal. Because the retina is from a right eye, the temporal pole is located 90° counterclockwise from the dorsal pole and the nasal pole is located 90° clockwise from the dorsal pole. (D) An example of a retina dissected from a left eye that has been immunostained to label s-opsin and imaged with a Texas Red filter (595 nm). The cuts in this retina are arbitrary since the topographical orientation is determined by the s-opsin gradient. (E) The results of digitally reconstructing the retina in D with Retistruct. Notice that the s-opsin gradient is not correctly aligned because the retina has not been rotated by the custom code. (F) The results of rotating the retina in D with the custom code. The retina has been rotated so that the highest concentration of s-opsin staining is located at the bottom and identified as the ventral retinal. Because the retina is from a left eye, the nasal pole is located 90° counterclockwise from the dorsal pole and the temporal pole is located 90° clockwise from the dorsal pole. D: dorsal, V: ventral, T: temporal, N: nasal. Scale bars = 1 mm. Please click here to view a larger version of this figure.
Figure 4: Comparison of two primary s-opsin antibodies in labeling s-opsin cones. (A) A retina stained with the goat anti-s-opsin primary antibody (see Table of Materials). (B) The other retina of the same mouse stained with rabbit anti-s-opsin primary antibody (see Table of Materials). (C) A representative region (0.1 x 0.1 mm2) from a retina stained with the goat anti-s-opsin primary antibody. Image taken on an epifluorescent microscope at 40X magnification. (D) A representative region (0.1 x 0.1 mm2) from a retina stained with rabbit anti-s-opsin (see Table of Materials), a primary antibody alternative. Image was taken on an epifluorescent microscope at 40X magnification. (E) Both antibodies label the same number of s-cone outer segments because there is no significant difference in the number of immunopositive s-cones that are stained by goat anti-s-opsin and rabbit anti-s-opsin at any of the tested retinal eccentricities (n = 2; ANOVA with post hoc Bonferroni test; p >0.05). Scale bars = 1 mm (A-B); 25 µm (C-D). Please click here to view a larger version of this figure.
Figure 5: A visual guide for using the Retistruct software to reconstruct retinas immunostained with s-opsin. (A) A retina opened in Retistruct with the outline visible and a "Tear" added. Points of the "Tear" are indicated with superimposed white arrows. All cuts in this retina are arbitrary, as no particular landmark was used to mark retinal orientation during dissection. Important buttons are outlined in red. (B) A retina with all "Tears" added and the dorsal retina identified with "D" on the edge of the retina. Notice that the "Reconstruct Retina" button is now visible. Important buttons are outlined in red. (C) The process of reconstructing a retina. The polar plot of the reconstructed retina will appear on the right, showing the relieving cuts in cyan (blue arrows superimposed to clarify cut locations). (D) The final result of running a retina through Retistruct. The original wholemount retina remains on the left and the reconstructed retina appears on the right. The relieving cuts are visible in cyan (white arrows superimposed to clarify cut locations). Please click here to view a larger version of this figure.
Figure 6: The superior rectus muscle and choroid fissure can be used to accurately orient the mouse retina. A polar plot of the angles obtained from either superior rectus muscle relieving cuts or choroid fissure cuts in retinas that have been reconstructed with Retistruct. Relieving cuts were identified on s-opsin immunostained reconstructed retinas and their locations were compared to the location of the s-opsin gradient. Using the custom MATLAB code to accurately rotate the retinas so that the highest concentration of s-opsin staining is located ventrally, true dorsal (90° for superior rectus), true nasal (0° for nasal choroid fissure) and true temporal (180° for temporal choroid fissure) were determined for each retina. The value of each individual relieving cut angle was determined in ImageJ and an average angle was calculated for each relieving cut type. Superior rectus muscle cuts identified the dorsal pole at 96.3 ± 4.3° (n = 11). The nasal choroid fissure identified the nasal pole at 6.7 ± 5.8° and the temporal choroid fissure identified the temporal pole at 172.0 ± 4.5° (n = 9). Please click here to view a larger version of this figure.
Deep Landmark | Corneal Burn Location | Pole of Retina Identified | Experimental Application |
Superior Rectus | Dorsal | Dorsal | Live or Fixed |
Nasal Choroid Fissure | Dorsal | Nasal | Live or Fixed |
Temporal Choroid Fissure | Dorsal | Temporal | Live or Fixed |
S-opsin Gradient | None | Dorsal, Ventral, Nasal, Temporal | Fixed |
Table 1: Deep landmarks, the pole of the retina they identify, and whether they can be used for live or fixed tissue application.
There has been no comprehensive, standardized protocol for determining and labeling the orientation of the isolated mouse retina in anatomical space. The protocol detailed here attempts to fill this void by standardizing and detailing how to use deep anatomical landmarks as reference points to reliably identify retinal orientation. It has been shown that the deep anatomical landmarks in this protocol provide a more accurate and reliable method for orienting the mouse retina than superficial landmarks such as corneal burns22. Thus, studies that have relied on corneal burns for retinal orientation may have had greater errors in orientation than studies that have relied on landmarks such as the rectus muscles and choroid fissure. This discrepancy highlights the need for and significance of this standardized protocol with respect to interpreting results and making comparisons between studies that depend upon accurate retinal orientation. Overall, a standardized protocol will provide a common method for vision researchers to follow, thus eliminating the presence of a confounding variable in data acquisition that may occur with the use of non-standardized methods for identifying retinal orientation.
The methods presented here are easily repeatable and applicable to many types of experimental protocols. In fact, one of the greatest advantages of this protocol is its adaptability. Because the choroid fissure, s-opsin expression, and rectus muscle landmarks have all been found to reliably identify retinal orientation22 the landmark that best suits the experimental parameters can be chosen to optimize data acquisition (Table 1). Additionally, methods of dissection can be combined so as to further clarify the orientation of the retina. For example, choroid fissure cuts can be combined with s-opsin immunohistochemistry in order to orient all four poles of the retina: nasal and temporal hemispheres can be identified by the choroid fissure cuts, and s-opsin immunohistochemistry can identify ventral and dorsal hemispheres. Yet, the adaptability of this protocol may be constrained by the time-sensitive nature of physiology experiments. Because the time it takes to identify a landmark, make a corneal burn, and execute a relieving cut could result in significant tissue death in ex vivo experiments, some of these dissection methods may be less than optimal. Fortunately, once a dissector has become familiar with either the choroid fissure or superior rectus muscle dissection method, identifying the deep landmarks and making the relieving cuts quickly become a part of the dissection routine and do not significantly add to the length of dissection. Although we do acknowledge that the steps detailed here can add on time to extremely time-sensitive experiments, we suggest using the s-opsin gradient for post hoc retinal orientation when the viability of the tissue is no longer an issue (Figure 3). Staining the retina for s-opsin is an effective way to orient the retina, as it can identify all four poles: s-opsin staining divides the retina into dorsal and ventral poles and allows for identification of the nasal and temporal poles depending on whether the retina is from a right or left eye (Figure 3). Therefore, we believe this protocol delivers a reliable and repeatable set of methods for accurate retinal orientation that can fulfill any experimental parameters.
As with any modified retinal dissection, the validity of the dissection method is limited by the accuracy of the dissector and the quality of the tissue that has been isolated. If any tissue is lost during dissection or a retina is too mangled for accurate reconstruction, Retistruct and the MATLAB program will not be able to reliably reconstruct or orient the retina. It is therefore important to practice the dissection method before using it for data-collecting experiments. While the types of dissections explained here are not difficult, they must be practiced to ensure the repeatability of identifying retinal orientation with a particular landmark. Furthermore, it is essential that the dissector practice visually identifying the anatomical landmarks prior to beginning data collection to make certain that the correct landmark is being used. One way to check the accuracy of a particular dissector is to make either choroid fissure cuts or superior rectus muscle cuts and then compare the location of the cuts to the s-opsin gradient, since it is a fixed marker and thus is not dependent on the accuracy of dissection. Potential dissectors can also compare their reconstructed retinas to the examples of reconstructed retinas with accurate landmark cuts are shown in Figure 1 and Figure 2. Essentially, a potential dissector should perform the steps outlined in this protocol for a particular dissection type, whether it be the superior rectus muscle or choroid fissure method, and compare the results to the s-opsin gradient to establish the validity of a particular dissector. Because if the dissector is unsure about the location of the landmark, it may result in an inaccurate orientation of the retina that will, by default, affect data collection and interpretation.
The authors have nothing to disclose.
We would like to thank Brittany Day and Jessica Onyak for their technical assistance and Dr. Liu for kindly letting us use his epifluorescent microscope. Acknowledgements of Support: NIH R15EY026255-01 and the Karl Kirchgessner Foundation.
0.1 M Phosphate Buffered Saline | Sigma-Aldrich | P5244 | |
Axioplan2 Epifluorescent Microscope | Zeiss | N/A | |
Clear Nailpolish | N/A | N/A | |
Corning LSE Low Speed Orbital Shaker | Sigma-Aldrich | CLS6780FP | |
Costar TC-Treated 24-well Plates | Sigma-Aldrich | CLS3524 | |
Dissection Microscope | Olympus | SZ51 | |
Donkey anti-Goat Alexa 594 | Life Technologies | A11058 | |
Donkey anti-Rabbit Alexa 594 | Life Technologies | A21207 | |
Donkey Normal Serum | Millipore | 566460 | Use at 5.2% (52 μL with 86 μL of 20% Triton X-100 and 863 μL of 0.1M PBS for 1 mL of blocking solution) |
Fisherbrand Superfrost Plus Microscope Slides | Fisher Scientific | 12-550-15 | |
Goat anti-s-opsin | Santa Cruz Biotechnologies | sc-14363 | Not commerically available as of 2017 |
Graefe Curved Forceps | Fine Science Tools | 11052-10 | |
ImageJ or FIJI | National Institute of Health | N/A | Freely available software |
Low Temperature Cautery Ophthalmic Fine Tip Cauterizer | Bovie Medical Corporation | AA00 | |
MATLAB | MathWorks | N/A | At least version 2007b or later |
Micro Cover Glasses | VWR International | 48393-241 | |
Micro Slide Trays | VWR International | 82020-913 | |
Moira Ultra Fine Forceps | Fine Science Tools | 11370-40 | |
Nitrocellulose membrane | Millipore | HAWP04700 | |
Paraformaldehyde | Electron Microscopy Sciences | 15714-S | Use at 4% (25 μL and 875 μL of 0.1 M PBS for 1 mL of fixative) |
PrecisionGlide Needle 20G (0.90mm x 25mm) | BD PrecisionGlide | 305175 | |
Pyrex Glass Petri Dish | Sigma-Aldrich | CLS3160152 | |
R | The R Project for Statistical Computing | N/A | Freely available software; version 3.4.3 or later |
Rabbit anti-s-opsin | Millipore | ABN1660 | |
Retiga R3 Microscope Camera | Qimaging | 01-RET-R3-R-CLR-14-C | |
Retistruct | N/A | N/A | Freely available software compatiable with Windows 7 or Windows 10 |
Shandon Aqua-Mount Slide Mounting Media | Fisher Scientific | 14-390-5 | |
Triton X-100 | Sigma-Aldrich | T8787 | Use 1.7% (86 μL of 20% Triton-X with 52 μL of Donkey Normal Serum and 863 μL of 0.1 M PBS for 1 mL of blocking solution) |
Vannas Spring Dissection Scissors | Fine Science Tools | 15000-03 | |
5MP USB Microscope Digital Camera | AmScope | MU500 | To be used with the Olympus Dissection Microscope |