Extraocular Source of Oligodendrocytes Contribute to Retinal Myelination and Optokinetic Responses in Zebrafish

Abstract

Purpose: There is no myelination in most mammalian retinas, and if it does happen, it is always accompanied by eye disease. Although lower vertebrates are born with myelin, the precise temporal dynamics of myelination in which oligodendrocytes (OLs) are involved, the origin of OLs, their behaviors in myelination, and their function in retinas have not yet been clearly elaborated. Therefore, we focus on these aspects to study the oligodendrocytes and myelin sheath in the zebrafish retina.

Methods: Retinal whole mount, immunohistochemistry, and optic nerve retrograde labeling were performed to monitor the myelination. Taking advantage of whole eye eversion and transplantation techniques, we studied the retinal origin of OLs. By optic nerve transplantation, we can observe single OLs in zebrafish retina. The optokinetic reflex (OKR) behavior test and the lysophosphatidylcholine (LPC)-induced retinal demyelination model were used to test the function of the myelin.

Results: First, we demonstrated that myelination starts at 28 dpf in zebrafish retinas. Second, we directly proved that all the OLs in zebrafish retinas migrated from the optic nerve rather than from a domestic source. Third, we found that compared with adult OLs, younger OLs tend to generate longer but a fewer number of internodes. Finally, we found that the myelin in zebrafish eyes is functionally relevant to the elegant OKR.

Conclusions: Our data suggest that the extraocular source of OLs first appeared at 28 dpf in zebrafish retina and then gradually developed with age, which contribute to optokinetic responses.

Normally, intraretinal myelination is absent in most mammals, including human beings. Myelinated retinal nerve fibers appear white or as an opaque patch in the human retina, which is present in approximately 1% of all eyes and may be associated with myopia and amblyopia.¹ Different from mammals, some myelin exists in some lower vertebrates, such as chickens, lizards, turtles, and goldfish.^2–5 By immunohistochemistry and transmission electron microscopy (TEM), previous studies described myelination and different developmental states of oligodendrocytes (OLs) in the above animals' retinas.^2–5 Through an injection of fluorescent dye DiI into the third ventricle, previous studies have also shown that some OLs can migrate into chick retinas from external to the eye.³ However, whether all OLs in retinas are from external to the eye was not yet clearly demonstrated. It is interesting to learn whether all OLs in retinas migrated from the optic nerve or whether some migrated from a domestic source. 

Previous studies have shown that myelin formed by OLs can be regulated by age, and internode lengths significantly decrease with age.^6,7 Oligodendrocyte precursors continue to proliferate and generate myelinating OLs into adulthood, and late-born OLs generate many more internodes, much shorter in length, than early-born OLs in mice optic nerves.^8,9 It is important to identify whether the OLs at different ages in retinas may have different behaviors regarding the number and lengths of internodes. 

Comparing the compact myelin sheath–wrapped axons in other parts of the central nervous system (CNS), the retina ganglion cell (RGC) axons are loosely wrapped with a single loose lamellar sheet in zebrafish retinas.¹⁰ The compact myelin sheath, formed by OLs, plays an important role in the rapid conduction of action potentials in the CNS, providing trophic support for neuronal somas and maintaining the normal physiological and structural features of axons.¹¹ It is necessary to study how one myelin lamella functions in zebrafish retinas. 

In this study, using olig2 transgenic zebrafish as an animal model, we aimed to investigate quantitatively the temporal dynamics of myelination, the source of OLs, their behavior in myelination, and their function in the myelinated retina. This study may provide more details for understanding intraretinal myelin and its physiological characteristics. 

Materials and Methods

Animals

Two zebrafish lines, wild-type (WT) and Tg(olig2:egfp),¹² were used in the experiment. All zebrafish were raised under standard conditions at 28.5°C with a 14-hour:10-hour light-dark cycle and a two times per day diet. All procedures were carried out in strict accordance with the provisions issued by the University of Science and Technology of China (USTC) Animal Resources Center and University Animal Care and Use Committee. The procedures were compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The protocol was approved by the Committee on the Ethics of Animal Experiments of the USTC (Permit Number: USTCACUC1103013). 

Retinal Whole Mount Preparation and Imaging

Zebrafish were anesthetized with 0.03% tricaine methane-sulfonate (MS-222; Sigma, St. Louis, MO, USA) in 0.68% saline, put into the small paraffin wax groove, and set belly up. The thoracic cavity was exposed and distracted by two needles. The venous sinus was broken and the heart was injected with 0.68% saline followed by 4% paraformaldehyde (PFA) (pH 7.4). After cardiac perfusion, the connective tissue around the eye was removed with jewelry #5 forceps (F.S.T, Heidelberg, Germany). The eye was cut and put into 4% PFA for an additional 0.5 hour post-fixation. After washing in PBS, the cornea, lens, sclera, choroid, and iris were torn off with forceps, respectively, and the optic nerve was cut off with scissors. The retina was transferred to a glass slide with an RGC layer upward and cut into quadrants. Excess solution was removed and then 100 μL 75% glycerin was dropped onto the retina. The retina was slowly covered with a coverslip and sealed with nail polish. 

The retina was imaged using a BX60 microscope (Olympus, Tokyo, Japan) with DP72 under a ×20 objective lens (Olympus). The images, 1360 × 1024 pixels, were taken three times with different focuses, and IPP6 software (Media Cybemetics, Bethesda, MD, USA) was used to expand the depth. Then, all the pictures were merged with Photoshop CS2 (Adobe Systems, Inc., San Jose, CA, USA). As the density of olig2⁺ cells is uneven, we manually counted the whole retina to obtain the total number. For the RGC number calculation, we chose three fields in each retina quadrant, calculated the density of the RGCs, and calculated the area with IPP6 software; finally, we can obtain the total number by density × area. 

Optic Nerve Retrograde Labeling

Anesthetized with MS222, each fish was put onto a wet tissue paper. After removing the connective tissue around the eye, pulling the eye out of orbit, and putting two small balls of paper into the space between the eye and the orbit, the optic nerve of the zebrafish was exposed. The optic nerve was cut with Vannas Spring Scissors (F.S.T.) and a piece of biocytin (B4261; Sigma) or Texas Red dextran (D-3328; Invitrogen, Carlsbad, CA, USA) crystal was put behind the stump. After absorbing for 2 hours, the fish was perfused with 4% PFA. The retina was soaked in Streptavidin-Cy3 (S6402; Sigma) for detecting biocytin and was imaged under a BX60 microscope (×20 objective; Olympus). 

Western Blot

Zebrafish eyeballs at different ages were lysed with SDS sample buffer (Beyotime, Shanghai, China), centrifuged for 1 minute at 9,000g, and the supernatant was collected. The 2X SDS-page loading buffer was equally added into the supernatant and then boiled for 10 minutes. Equal amounts of protein were separated by 12% SDS-PAGE and transferred onto a polyvinylidene difluoride membrane (Millipore, Bedford, MA, USA). The membrane was incubated in 5% milk (Sangon Biotech, Shanghai, China) for blocking, in the anti-myelin basic protein (MBP) antibody (1:1000, Abmart, Shanghai, China) and anti-β-actin (Cell Signaling, Beverly, MA, USA) for antigen-antibody recognition, and then in horseradish peroxidase–conjugated goat anti-rabbit IgG (1:10,000; Sango, Shanghai, China) for signal detection. Enhanced chemiluminescence detection kits (Thermo Fisher Scientific, Waltham, MA, USA) were used to visualize protein and films (Kodak, Rochester, NY, USA) were used to expose the membrane. 

Transplantation and Eye Eversion Experiment

Before optic nerve transplantation, two lines of zebrafish (2-month-old young zebrafish and 4-month-old adult zebrafish), the WT line and the Tg(olig2:egfp) line, were raised at 20°C for 1 week. By exposing the optic nerve, as previously described, a section of the optic nerve was cut in both lines and the olig2 optic nerve was transplanted into the gap between two optic stumps of the WT line. The small balls of paper in the space between the eye and the orbit were taken out. After the small balls of paper were taken out, the zebrafish eye were relieved from the state of pulling out of the orbit. Ensure the correct orientation of the eye in dorsal, ventral, nasal, temporal position (Supplementary Video S3). At last, the zebrafish were returned to the system water and raised at 20°C with a 14-hour:10-hour light-dark cycle and a two times per day diet. 

Two lines of zebrafish mentioned above at 20 day post-fertilization (dpf) were used in the eye transplantation. After anesthesia and removing the connective tissue around the eyes, the eye was pulled out of orbit with forceps and the eye muscle was pinched off. The optic nerve and vessels of both zebrafish lines were crushed for 30 seconds before being cut. Then, the optic nerves and vessels along the optic nerve head were cut and the eyes of these two fish were exchanged (see Supplementary brief method article including Supplementary Videos S1, S2 for more information of transplanted eye and Supplementary Video S4 for operations steps). These zebrafish were raised in an embryo medium solution for 3 days, after which the embryo medium was gradually replaced with system rearing water. 

An olig2 zebrafish at 20 dpf was anesthetized and connective tissue around its eye was removed. The eye was pulled out of orbit with forceps and eye muscle was pinched off. The optic nerve was pinched off and the eye was flipped 90° outside (see Supplementary Video S5 for detail). The zebrafish was checked under fluorescence microscopy to make sure there was no fluorescent signal behind the eyeball. Then, the zebrafish was put back into the embryo medium and raised in the embryo medium for 3 days under standard conditions, after which the embryo medium was gradually replaced with system rearing water. 

Retina whole mounts were imaged under a confocal microscope (×20 objective, FV 1000; Olympus). For quantifying the internode number and lengths, confocal images were collected (×60 objective). The Imaris 7.1.0 software (Bitplane, Zurich, Switzerland) and our custom-made software were used for the quantitative analysis of internode lengths and number, respectively (see Supplementary Methods for details). 

The Optokinetic Reflex (OKR) Behavior Test and Drug Application

To make the eye demyelination model, we injected 200 nL 1% lysophosphatidylcholine (LPC) (L0906; Sigma) in 0.68% saline into the eyes of zebrafish. Anesthetized with MS222, zebrafish were put on the white wet wipes. A small hole was poked in the posterior cornea adjacent to the lens with a glass needle. The 200 nL 1% LPC was microinjected into zebrafish eyes through the small hole using a microinjector. 

Aiming to test whether demyelination in the eye can affect visual function, we conducted the OKR behavior test. As a previous study described, after anesthesia, we fixed the fish on a sponge in a transparent cylindrical glass tank with needles and poured system water into the tank to awaken it.¹³ Zebrafish eye movements at a cycle of 12 and the speed of 12 or 24 rpm, were recorded with a digital camera (A640; Canon, Tokyo, Japan). Video (a frame rate of 29 frames per second) analysis and data processing were based on our custom-made software. 

Transmission Electron Microscopy

After perfusion with 4% PFA, LPC- or saline-treated and untreated zebrafish eyes were then fixed in 2.5% glutaraldehyde solution in 0.1 M phosphate buffer (pH 7.4) overnight at 4°C. The following steps were standard electron microscopy sample preparation steps, as a previous study described.¹⁴ Transverse ultra-thin sections were mounted onto copper grids and imaged at ×25,000 by TEM (JEM-1230; JEOL, Tokyo, Japan). All axons in each image were counted and analyzed for the diameter of myelinated and unmyelinated axons at three different orientations. The G-ratio (the ratio of the inner axonal diameter to the total outer diameter) and the myelin thickness were calculated in all myelinated axons. 

Immunohistochemistry

Cryosections of WT zebrafish were stained with customized rabbit anti-MBP (1:1000; Abmart) and α-tubulin (1:1000; Invitrogen) and then by donkey anti-rabbit Alexa fluor 488 and goat anti-mouse Alexa fluor 568 (1:1000; Invitrogen). All cryosections were mounted in 75% glycerol and imaged by fluorescence microscopy (BX61; Olympus). 

After perfusion with 2% PFA, zebrafish were further fixed in 2% PFA for 6 hours at 4°C. Retinas were taken out and blocked in 10% PBST (10% serum, 1% BSA, 0.3% Triton X-100 in PBS), incubated with mouse anti-Nav 1.6 (1:1000, S8809; Sigma) overnight at 4°C, washed three times (5 minutes per time) with PBS, incubated with goat anti-mouse Alexa fluor 568 (1:1000; Invitrogen) overnight at 4°C, washed three times with PBS, and mounted in 75% glycerol. The images of OLs and sodium channel Nav 1.6 near the outer limit of the retinal vessels were collected (×60 objective, FV1000; Olympus). 

Statistical Analysis

The data were presented as mean values ± SE and analyzed with 1-way ANOVA by Tukey's test in GraphPad Prism version 6 (Prism, La Jolla, CA, USA). The criterion for significance was set at P = 0.05 where *, **, and *** represent P < 0.05, P < 0.01, and P < 0.001, respectively. The comparison of internode length and G-ratio distributions were done by the K-S nonparametric test in GraphPad Prism version 6. 

Results

The Temporal Developmental Dynamics of Myelination in Zebrafish Retinas

We first checked the myelination in zebrafish retinas at 7, 14, and 28 dpf by immunohistochemistry with MBP and α-tubulin antibodies. The MBP-positive signal was absent at 7 and 14 dpf, but it was visible at 28 dpf (Fig. 1A). Similar with MBP immunohistochemistry, there were no olig2⁺ OLs in zebrafish retinas at 7 and 14 dpf, but a small number of OLs and a small amount of myelin appeared in olig2 zebrafish retina at 28 dpf (Fig. 1B). With the retina growing to keep pace with age, the number of OLs continued to increase, ranging from 37 ± 5 (mean ± SEM, n = 44) to 7989 ± 188 (mean ± SEM, n = 8), respectively, at age 28 dpf to 6 months (Figs. 1B–D). The density of OLs increased with age until postfertilization at 3 months, at which time it showed no statistically significant difference compared with 6 months postfertilization (Supplementary Fig. S1D). In addition, we found the expression of MBP increased along with development and remained stable at 3 months, as indicated by the Western blot data (Figs. 1E, 1F). 

Figure 1

 

Development of myelination in zebrafish retinas. (A) Cryosections of 7- to 28-dpf zebrafish retinas were stained with MBP and a tubulin antibody. Myelin basic protein appeared at 28 dpf. (B, C) Whole-mount olig2 zebrafish retinas at the age of 20 to 36 dpf and 1.5 to 6.0 months. The olig2⁺ cells, outside the nerve fiber layer, were not OLs at 7 and 14 dpf. (D) Quantification of the number of OLs (n = 10, 10, 44, 21, 14, 13, 10, and 8, respectively) in zebrafish retinas from 7 dpf to 6 months. (E, F) Myelin basic protein Western blot of zebrafish retinas and its quantitative analysis at age 1.5 to 6.0 months. Scale bars: (A) = 100 μm; (B, C) = 200 μm.

To furtherly confirm the time when OLs appeared in the zebrafish retinas, we counted the number of OLs from 20 dpf to 36 dpf in detail and discussed the relationship among the number, body lengths, and ages of OLs. There were progressive increases in the mean body length of zebrafish and the average number of OLs, from 6.21 mm (n = 37) and 0.42 (n = 37) at 20 dpf to 6.61 mm (n = 40) and 6 (n = 35) at 24 dpf to 7.09 mm (n = 38) and 27.43 (n = 36) at 28 dpf to 8.02 mm (n = 21) and 101.57 (n = 21) at 36 dpf (Supplementary Figs. S1A, S1B). In fact, zebrafish body lengths were longer than 7.00 mm once OLs appeared in their retinas (Supplementary Fig. S1C). The zebrafish with shorter body lengths (e.g., 5.15 mm) did not have OLs in their retinas, even though they were at the age of 36 dpf (Supplementary Figs. S1B, S1D). These data indicate that zebrafish at 28 dpf with body lengths longer than 7 mm have OLs and myelin in their retinas. 

Similar with the OLs, by taking advantage of the retrograde labeling technique, we found the number of RGCs also continued to increase at the age of 1.5 to 6.0 months (Supplementary Fig. S2). The ratio between RGCs and OLs decreased with age until 3 months postfertilization, which indicate the gradually development of myelin in zebrafish retina (Supplementary Fig. S2). 

The Origins of OLs in Zebrafish Retinas

A large number of stem cells exist in zebrafish retinas, such as the retinal stem cells in the ciliary marginal zone.¹⁵ Besides, scattered Müller glia can function as multipotent retinal stem cells to regenerate other types of retinal neurons and photoreceptors following injury.^16–20 Stem cells from the developing retinas can differentiate into myelinating OLs.²¹ 

To determine whether the OLs in zebrafish retinas are from external to the eye or from the retinal stem cells, we first performed the eye eversion experiment. The optic nerve of the zebrafish's right eye was cut at 20 dpf along the optic nerve head and flipped outside by 90° (Fig. 2A). At 2 weeks after eye eversion, the retina was whole-mounted and the number of OLs was counted (Figs. 2B, 2C). We found no OLs in zebrafish right eyes, whereas there was a small amount of OLs in the control left eyes. This may indicate that the OLs in zebrafish retinas are from external to the eye. To furtherly verify this point, we then carried out the eye transplantation experiment. We transplanted WT zebrafish eyes into the olig2 zebrafish and vice versa (Fig. 2D). We found olig2 retinas of WT zebrafish did not have OLs, whereas OLs appeared in the WT retinas of olig2 zebrafish, and their number showed no significant difference compared with the control eye at 3 and 5 weeks after transplantation (Figs. 2E, 2F). This strongly supported the notion that all of the OLs in zebrafish eyes are from external to the eye. 

Figure 2

 

The origin of OLs in zebrafish retinas. (A) An illustrative picture shows the normal and eye eversion zebrafish. (B, C) Whole mount and quantification of the number of OLs in normal olig2 zebrafish retinas and the retinas after eye eversion 2 weeks later. (D) An illustrative picture shows the exchange of eyeballs in WT and olig2 zebrafish. (E) Whole-mount retinas show OLs in normal olig2 zebrafish eyes, olig2 zebrafish with WT eyes, and WT zebrafish with olig2 eyes at 3 weeks after eye eversion. (F) Quantification of the number of OLs in normal olig2 zebrafish eyes and olig2 zebrafish with WT eyes (n = 5 and n = 7, respectively) at 3 and 5 weeks after eye eversion. Scale bars: (B) = 150 μm, (E) = 200 μm.

The Behaviors of OLs in Zebrafish Retinas

By transplanting a section of the olig2 optic nerve into WT zebrafish optic nerves, we can observe a small number of OLs at different developmental stages, such as oligodendrocyte precursor cells (OPCs) (Fig. 3A), immature OLs (Fig. 3B), and mature OLs (Fig. 3C), in zebrafish retinas. Through whole-mount retinal immunohistochemistry stained with MBP and a tubulin antibody, we found some OLs from the transplanted optic nerves could form mature myelin in zebrafish retinas (Fig. 3C). 

Figure 3

 

Oligodendrocytes at different developmental stages in zebrafish retinas. (A) The OPC was migrating out of the optic nerve head. The white dashed circles stand for optic nerve head. (B, C) Oligodendrocytes at different developmental stages: immature and mature OLs in zebrafish retinas stained with MBP and a tubulin antibody. The immature OLs are not colocalized with MBP, but the mature OLs are colocalized with MBP. Scale bars: 20 μm.

To test whether the OLs at different ages exhibit different behaviors, we transplanted adult (4 months) olig2 zebrafish optic nerves and young (2 months) olig2 zebrafish optic nerves into WT zebrafish, respectively. After comparing the morphology of adult OLs with young OLs in zebrafish retinas, we found young OLs possessed longer but fewer numbers of internodes than adult OLs (the average length of an internode is 46 μm in young OLs and 29 μm in adult OLs; the average number of internodes is 36 in young OLs and 51 in adult OLs) (Figs. 4A–C). As retinas continue to grow with age, OLs continue to migrate from the optic nerve head to the retinal vessel along the optic nerve (Figs. 1B, 1C). The OLs close to the outer limit of the retinal vessels in young and adult zebrafish retinas represent young and adult OLs, respectively. We counted the number of Nav1.6 per OL in zebrafish retinas close to the outer limit of the retinal vessels by whole-mount retinal with Nav1.6 immunohistochemistry. Similar to a previous result, young OLs tended to generate a fewer number of internodes (Supplementary Fig. S3). These data indicated OLs at different ages exhibit different behaviors. However, the total lengths of myelin sheaths synthesized by young and adult OLs are similar (∼1660 μm for young OL and ∼1480 μm for old OL). After demyelination induced by LPC, the number of OLs was greatly decreased (Figs. 6C, 6D) and most OLs involved in remyelination were adult OLs migrated into the retinas along the optic nerve. We found the number of OLs in completely remyelinated zebrafish retinas (2 months after LPC-induced demyelination) had no significant difference compared with the control group (Supplementary Fig. S4). This indicated the ability to form myelin by young and adult zebrafish is similar. 

Figure 4

 

Oligodendrocytes at different stages have different behaviors. (A) The morphology of adult and young OLs in zebrafish retinas. (B) Distributions of internode lengths for adult OLs and young OLs (n = 2539 and n = 713 internodes, respectively). The K-S test shows that the internode length distribution was significantly different for young and adult OLs (P < 10⁻⁴). (C) Quantification of the number of internodes in adult and young OLs (n = 87 and n = 47, respectively). (D) Quantification of the lengths of internodes in adult OLs at 2 and 3 months after transplantation (n = 1354 and n = 1185, respectively). Scale bar: 20 μm.

In comparison with the old OLs, the young OLs tended to have longer internodes. Does this mean the internodes will shrink with time? To answer this question, we compared the lengths of internodes at 2 and 3 months after optic nerve transplantation and found that it showed no significant change with time (Fig. 4D). This indicated the internode length is mainly determined by the age at which it forms, which is consistent with a previous study on the optic nerves of mice.⁸ 

The Function of OLs in Zebrafish Retinas

Oligodendrocytes can form single loose lamellar myelin sheets in zebrafish retinas. We used TEM to examine the state of myelination in zebrafish retinas during development (Fig. 5A). We found that, although the diameter of the axon is increased with age, the unmyelinated axon diameter remains stable (Figs. 5B, 5C). Through a comparison of the diameters of myelinated and unmyelinated axons, we found myelinated axons tend to have a larger diameter and their diameter is increased with age (Fig. 5C). The increase in fiber diameter is not accompanied by a significant change in the G-ratio between 1.5 (0.658 ± 0.007, n = 85) and 2 months (0.661 ± 0.009, n = 111) (K-S test, P = 0.983), but the G-ratio has a slight increase at 3 months after fertilization (0.691 ± 0.003, n = 265) (K-S test, P < 0.001) (Fig. 5D). Although the mean G-ratio is increased, it still remains at 0.6 to 0.7 (Fig. 5D), as with most myelinated axons.²² The thickness of myelin is increased with age (Fig. 5E), which indicates the growth rate of the axon diameter is faster than the thickness of the myelin. Instead of opaque compact myelin, one myelin lamella maintains the transparency of the retina and maintains the G-ratio at 0.6 to 0.7, which may play a role in visual function. 

Figure 5

 

The relationship between myelination and the axon diameter. (A) Electron micrographs of RGC axons in zebrafish retinas from 1.5 to 3.0 months. (B, C) Quantification of the axon diameter (n = 517, 295, and 311 at 1.5, 2.0, and 3.0 months, respectively), myelinated (axons: n = 85, 140, and 251; zebrafish: n = 3, 3, and 3; images: n = 10, 11, and 13 at 1.5, 2.0, and 3.0 months), and unmyelinated axon (axons: n = 432, 155, and 60; zebrafish: n = 3, 3, and 4; images: n = 10, 11, and 16 at 1.5, 2.0, and 3.0 months) diameter in zebrafish retinas at the age of 1.5 to 3.0 months. (D) A scatterplot of axon diameter versus G-ratio for myelinated axons in 1.5-, 2.0-, and 3.0-month-old zebrafish retinas. (E) Quantification of the thickness of myelin in zebrafish retinas (axons: n = 82, 67, and 144; zebrafish: n = 3, 3, and 4; images: n = 10, 11, and 16 at 1.5, 2.0, and 3.0 months). Scale bar: 200 nm.

Figure 6

 

The OLs in zebrafish retinas are functional. (A) Cryosections of normal (right eye) and demyelinated (left eye) zebrafish retinas were immunolabeled with MBP and a tubulin antibody. (B) Myelin basic protein Western blot of normal and demyelinated zebrafish retinas. (C) Whole mount of normal and demyelinated olig2 zebrafish retinas. (D) Quantification of the number of OLs in the normal and demyelinated olig2 zebrafish retinas. (E) Electron micrographs of RGC axons in normal and LPC-induced demyelinated zebrafish retinas. (F, G) Optokinetic reflex behavior test of saline-treated and LPC-induced demyelinated zebrafish at the speed of 12 (n = 19, 18) and 24 rpm (n = 38, 41). Scale bars: (A) = 100 μm, (C) = 200 μm, and (E) = 200 nm.

To test whether the myelin in zebrafish retinas has a function, we first established an LPC-induced demyelination model in zebrafish retinas. Using Western blot and immunohistochemistry, we found the expression of MBP decreased remarkably a week after LPC-induced demyelination and the number of OLs was greatly decreased (Figs. 6A–D). Electron microscopy also showed a large number of demyelinated axons (Fig. 6E). Through retrograde labeling, we found the number of RGCs did not have significant change a week after LPC treatment, which indicates axons retain integrity in zebrafish retinas (Supplementary Fig. S5). The OKR consists of a fast saccade and smooth pursuit eye movements, which happens when the eye follows an object that moves across a visual field and it resets the eyes once the object has left the visual field. By using the OKR behavior test, we found the value of the gain (angular velocity of the eye/angular velocity of the rotated grating) at the speed of 12 (Fig. 6F) or 24 (Fig. 6G) rpm was slightly but significantly decreased a week after LPC-induced demyelination in zebrafish retinas. This indicated myelin contributes to eye movement functions in zebrafish eyes. 

Discussion

Myelination in Zebrafish Retinas

In zebrafish, OPCs begin to migrate into the anterior spinal cord at 49 hours after fertilization and myelination starts at 3 dpf, at which time the Mauthner axons, ventral axons, and peripheral axons are surrounded by loose myelin.^12,23 The myelinated axons can be seen at 7 dpf in zebrafish optic nerves.²⁴ Now that the myelination starts as early as 7 dpf in zebrafish optic nerves, we speculated OLs, which migrate into the retina along the optic nerve, should appear soon after 7 dpf, as well as the start of myelination in zebrafish retinas. Actually, this was not the case until 28 dpf, as some myelinated axons appeared in zebrafish retinas and some zebrafish with shorter body lengths did not even have OLs in their retinas at 28 dpf (Fig. 1, Supplementary Fig. S1). The reasons for the late appearance of OLs are unclear, although there may be two possibilities. First, some signals or physical barriers at the retinal optic nerve junction could disturb the migration of OLs during the early development of zebrafish retinas. Second, the zebrafish retinas may lack some necessary signals for the migration of OLs at the early stage of development. As numerous signals, including secreted growth factors, morphogens, and guidance cues, as well as cell adhesion molecules, can affect the migration of OLs, what molecules can affect OLs migrating into the retinas must be identified in the future.²⁵ 

Previous studies have shown that axons are myelinated in some lower vertebrate retinas, such as in chickens, lizards, and goldfish.^2–5 Despite a lack of transgenic lines, they simply described myelination and the different developmental states of OLs by immunohistochemistry and TEM evidence. Using olig2 zebrafish,¹² we can quantitatively observe the temporal dynamics of myelination in zebrafish retinas. We found OLs exhibit an outward diffused distribution along the optic nerve toward the outer limit of retinal vessels. The numbers of OLs and RGCs increased with age. The ratio of RGCs versus OLs decreased with age and remains stable at 3 months, as well as the density of OLs (Fig. 1, Supplementary Fig. S2). 

The Origins of OLs in Zebrafish Retinas

Previous studies have shown OLs can migrate along the optic nerve and move into the retina.³ The reason most mammalian retinas lack myelin is not because ganglion cell axons themselves cannot be myelinated in the retina,^26–28 but because a large number of migratory OLs are kept out of the retina by the retinal optic nerve junction.^28,29 The concentrated astrocytes that highly express glial fibrillary acidic protein at the retinal optic nerve junction may be responsible for the inhibition of migratory OLs.^29,30 As there is a large number of pluripotent stem cells in the retina^15,31 and stem cells from the developing retina can differentiate into myelinating OLs in vitro,²¹ previous data cannot exclude the possibility that some OLs are generated in the retina. Zebrafish have a strong ability to regenerate in the CNS, which makes eye and optic nerve transplantation possible. Owing to the OL transgenic lines, zebrafish eye transplantation is an ideal model to study the origins of OLs in the retina. By observing whether OLs could appear in WT-olig2 zebrafish or olig2-WT zebrafish, we have directly proven that the OLs in zebrafish retinas are indeed from external to the eye (Fig. 2). 

The Behaviors of OLs in Zebrafish Retinas

By taking an in vivo image, the behavior of a single OL can be observed,^32–37 but this observation can be confined only to the transparent zebrafish embryo. The high density of myelinating cells in the CNS limits the ability to observe specific myelin internodes corresponding to particular OLs. Using the optic nerve transplantation model, which makes observing a small number of scattered OLs in zebrafish retinas possible and provides us with a good model to study the behaviors of OLs at different ages in zebrafish, we can clearly observe myelin internodes associated with each OL at different ages and analyze the number and lengths of internodes (Fig. 4). Our data showed young OLs make approximately 36 internodes with a mean length of 46 μm, whereas adult OLs make approximately 51 internodes with a mean length of 29 μm, which indicates young OLs tend to have longer but a fewer number of internodes (Fig. 4, Supplementary Fig. S3). Besides, the total lengths of myelin sheaths synthesized by young and adult OLs are similar and their abilities to form myelin are unchanged (Fig. 4, Supplementary Fig. S4). The lengths of internodes do not shrink with age, which indicates internode length does not relate to the length of time it has existed, but it is determined by the time it formed (Fig. 4). The behaviors of OLs in zebrafish retinas are similar to the behaviors of OLs in the optic nerves of mice.^8,9 

The lengths and number of internodes made by each OL varied greatly in different CNS regions.³⁸ We found that, in comparison with the spinal cord,^32,33,37 retinas tend to have longer and a greater number of internodes. Oligodendrocytes from different brain regions express divergent properties in self-renewing divisions and respond to inducers of OL generation, which appear to be cell intrinsic and also seem to be correlated with an intracellular redox state.³⁹ This led us to hypothesize that the difference in the number and lengths of OLs in different CNS regions are modulated by cell intrinsic properties. 

The Function of OLs in Zebrafish Retinas

Myelinated optic nerve fibers terminate at lamina cribrosa in healthy people. If OLs are abnormally migrating into the retina, myelinated retinal nerve fibers will appear in the human eye, which may be associated with visual defects, such as myopia and amblyopia,¹ whereas OLs exist in zebrafish retinas. Different from the compact myelin in patients with myelinated retinal nerve fibers, the axon is wrapped with a single loose lamellar sheet in zebrafish retinas. Nervous systems have evolved two basic mechanisms to increase the conduction speed of the electrical impulse. First, an increase in the diameter of the axons and second, wrap axons with myelin.⁴⁰ The diameters of axons in zebrafish eyes are far smaller than those of the axons in mammal eyes, and their optic nerves (data not shown), which signals a low conduction speed. To increase the conduction speed, myelin is needed in zebrafish eyes. The compact myelin is opaque, which may opacify the retina and impair vision, whereas transparency of one myelin lamella is much better. For these reasons, one myelin lamella may be a good choice for zebrafish in evolution. Previous studies have shown myelin and OLs play a positive role in the axonal diameter.^41,42 We found the diameter of myelinated axons increased with age, whereas the diameter of unmyelinated axons seems to be maintained at a stable value, which implied the myelin in zebrafish eyes might promote the expansion of the axonal diameter (Fig. 5). With the increase in the axonal diameter, myelin thickness was also increased, which maintains the G-ratio in zebrafish eyes at 0.6 to 0.7, as with most myelinated axons (Fig. 5).²² 

Although retinal myelination has been reported in some lower vertebrates, such as chickens, lizards, turtles, and fish,^2–5 no study has related the functional characterization of myelin in retinas. Lysophosphatidylcholine can induce rapid and highly reproducible demyelination in the CNS without producing much damage to the axons.^43,44 By injecting LPC into olig2 zebrafish eyes, we can quantitatively observe the process of demyelination and remyelination, which makes an olig2 zebrafish-based retinal demyelination model a good tool for studying demyelination and remyelination. As zebrafish have a strong reproductive capacity and low maintenance costs, a zebrafish retinal demyelination model provides a good platform for remyelination drug screening. Through the LPC-induced demyelination model and OKR behavior test, we found the gain values were slightly but significantly decreased after retinal demyelination, which indicated the myelin in zebrafish retina is functional (Fig. 6). This may be the reason for the reservation of retinal myelin in evolution. 

Acknowledgments

The authors thank Yubin Huang for the OKR behavior test equipment, data analysis software support, and the software for the morphological analysis of internode provision. 

Supported by the 973 MOST Grant (Grant No. 2011CB504402, 2012CB947602) and the National Natural Science Foundation of China (Grant No. 91132724, U1332136). 

Disclosure: C. Tian, None; S. Zou, None; B. Hu, None