Dorsal Retinal Pigment Epithelium Differentiates as Neural Retina in the Microphthalmia (mi/mi) Mouse

Abstract

purpose. Microphthalmia, a bHLH-zip transcription factor associated with the onset and maintenance of pigmentation, identifies the retinal pigment epithelial (RPE) compartment during optic vesicle and optic cup development. To determine a role for microphthalmia (mi) during eye development, the effects of an mi loss of function mutation on RPE and neural retinal were investigated in the mi/mi mouse.

methods. A series of embryonic and postnatal mi/mi and wild-type eyes were sectioned and labeled with neural retina– and RPE cell type–specific antibodies. Photoreceptor loss was quantified by counting the number of photoreceptor nuclei spanning the outer nuclear layer throughout postnatal retinal development.

results. Early neural retinal differentiation is not affected in the mi/mi mouse. The mi/mi ventral retinal pigment epithelial layer begins to develop normally, but does not pigment or attain a differentiated cuboidal morphology. The dorsal region of mi/mi retinal pigment epithelium expands and forms an ectopic retina, which develops all major retinal cell types along a similar time course as the wild type. After birth, mi/mi photoreceptors begin to form rosettes, outer segments fail to elongate, and over an extended time period, the retina degenerates.

conclusions. Together these results suggest that early retinal development can proceed normally in the mi/mi mutant, but later retinal histogenesis is dependent on the presence of a differentiated retinal pigment epithelium. Most importantly, loss of mi function permits a change in cell fate from RPE to retina in the dorsal eye.

In rodent eye development, the optic vesicles invaginate to begin forming the optic cups by embryonic day (E) 11. ¹ At this stage, the developing eye is compartmentalized, and the outer optic cup, which gives rise to the retinal pigment epithelium (RPE), expresses the transcription factors otx-2 and microphthalmia (mi). ^{2 3 4} Once the optic cup is fully formed, the RPE further differentiates by reducing its mitotic rate, thinning, and beginning to express pigment. ⁵ The program of RPE development in many species can be altered by growth factor stimulation, causing it to transdifferentiate into neural retina by losing RPE gene expression and initiating the expression of neural retina–specific genes. ^{6 7} Amphibian RPE transdifferentiates throughout life, but fish, chick, and rodent RPE do not. ^{8 9 10 11} Rodent RPE, the most developmentally restricted, transdifferentiates only during a narrow developmental window. ^{7 8}  

Mi, bHLH-zip transcription factor, which is associated with the onset and maintenance of pigmentation, may play a role in transdifferentiation and early specification of the outer optic cup neuroepithelium as RPE. Studies of avian RPE have demonstrated that mi is downregulated during transdifferentiation and that mi overexpression inhibits transdifferentiation. ¹² Mice homozygous for a 3-bp deletion in the DNA binding region of mi show numerous pigmentation defects and are microphthalmic. In a detailed histologic study, Scholtz and Chan ¹³ noted that a portion of the RPE was hyperplastic and suggested that it might represent ectopic retina. In the present study, we have confirmed and extended these observations. Using a panel of retina- and RPE-specific markers, we have shown that the hyperlplastic RPE region is an inverted ectopic retina that contains major retinal cell types. The results also indicate that the neural retina and dorsal ectopic retina develop in parallel, whereas the ventral RPE remains a monolayer of incompletely differentiated cells. After birth, outer segments (OS) do not elongate, and there is a progressive rod photoreceptor loss. Together, these results demonstrate that mi plays a key role in directing the cell fate of dorsal RPE as well as in the maturation of all RPE cells. 

Materials and Methods

Experimental Animals

The use of animals in this work was in accordance with the ARVO resolution for the Use of Animals in Ophthalmic and Vision Research and the Yale Animal Care and Use Committee. Heterozygous mice (mi/ + ) who contain a 3-bp deletion in mi that eliminates DNA binding activity were obtained from Jackson Laboratories (Bar Harbor, ME). 

Tissue Preparation

Embryos were obtained from timed pregnancies where observation of a vaginal plug was taken as day 0. mi/mi mice were identified by their lack of pigmentation and small eye size. The dams were killed by CO₂ inhalation, and the embryos were removed and immersion-fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (PFA), pH 7.4, overnight at 4°C. Postnatal animals were injected intraperitoneally with an overdose of sodium pentobarbital (65 mg/kg) and then perfused through the heart with 4% PFA. Eyes were enucleated and immersion-fixed in 4% PFA for an additional 1 to 4 hours at room temperature. 

Tissue were cryoprotected in 30% sucrose overnight at 4°C, frozen in 30% sucrose:OCT cryostat medium (1:1), and serially sectioned at 12 to 15 μm. Sections were mounted onto gelatin-coated slides and stored frozen at −20°C. Every fifth slide was stained with cresyl violet, dehydrated, and coverslipped with Permount. 

Immunocytochemistry

Frozen sections were thawed for 15 minutes at room temperature. Sections were blocked at room temperature for 1 hour with 10% normal goat serum in 0.1 M phosphate-buffered saline (PBS; 1×) containing 0.1% Triton X-100. The following monoclonal antibodies were used: 8A1, ¹¹ HPC-1, ¹⁴ RetP1, ¹⁵ and SVP-38. ¹⁶ An affinity-purified polyclonal rabbit antiserum to the transcription factor Otx-2 was a gift from Flora Vaccarino. 

The anti-mi antibody was generated by injecting a rabbit with a keyhole limpet hemocyanin (KLH) coupled Mi peptide (NH₂-T-S-S-R-R-S-S-M-S-A-E-E-H-T-E-H-A-COOH), which corresponds to the COOH-terminal of the Mi protein. ³ Rabbits were inoculated with the Mi-KLH emulsified with CFA, boosted monthly with Mi-KLH emulsified with in CFA, and bled 1 to 2 weeks after boosting. ¹⁷  

Sections were incubated in primary antibody for 24 hours at 4°C, washed in 1× PBS, and then incubated in anti-mouse IgG-Texas Red (1:200) or anti-rabbit Texas Red (1:200) for 35 minutes at room temperature (Jackson ImmunoResearch Laboratories, West Grove, PA). After washing with 1× PBS for 10 minutes, sections were coverslipped in 80% glycerol in 1× PBS. Sections were photographed, and cells counted either on a Zeiss Axiovert (Thornwood, NY) and then scanned into Adobe Photoshop, or images were imported directly into Adobe Photoshop (Adobe Systems, San Jose, CA) using a Zeiss Axioskop equipped with a Spot camera (Diagnostic Instruments, Sterling Heights, MI). 

Results

Normal mi/mi Neural Retina Differentiation Abnormal RPE Development

Embryonic neural retina and RPE development were investigated in the wild-type and mi/mi eye using morphology, transcription factor, and neural retinal cell type-specific antibodies. By E14, the wild-type and mi/mi neural retina had a ganglion cell layer (GCL) and optic nerve (ON). The inner retina began to stratify, whereas the outer retina appeared as a morphologically undifferentiated tissue (Figs. 1A 1B ). Although wild-type RPE was heavily pigmented (Fig. 1A , arrows), in mi/mi RPE pigmentation was absent, and the ventral RPE appeared similar to an undifferentiated neuroepithelium (Fig. 1B , arrows). Moreover, the dorsal RPE was expanded in mi/mi mice (Fig. 1B , arrowheads). 

At E13, a Mi polyclonal antibody labeled RPE cell nuclei in the wild type (Fig. 1C , arrowheads). In the mi/mi RPE, Mi was detected only in regions where the RPE remained a monolayer (Fig. 1D , arrowheads). The dorsal expanded region (DER) lacked Mi labeling (Fig. 1D , tissue between arrows). Otx-2 labeling was observed in cells scattered through the wild-type neural retina and in the RPE nuclei (Fig. 1E , arrows). Like Mi, Otx-2 labeling in the mi/mi eye was detected in the neural retina and in RPE regions that remained a monolayer (Fig. 1F , arrowheads). The DER contained little Otx-2 labeling (Fig. 1F) with only a few scattered otx-2–labeled cells whose distribution was similar to the neural retina (data not shown). As the retina further differentiated by E17, ganglion cells, their axons, and horizontal cells were labeled with a neurofilament (8A1) antibody (Figs. 1G 1H 1I) . The wild-type neural retina and mi/mi neural retina contained 8A1-positive ganglion cells and processes (arrows) and migrating horizontal cells (Figs. 1G 1H , arrowheads). Neurites labeled in the most scleral region of the DER in mi/mi mice (Fig. 1I , arrows). 

After birth at postnatal day 2 (P2), the wild-type, mi/mi neural retina and DER contained a well-defined GCL and inner plexiform layer (IPL), emerging inner nuclear layer (INL) and undifferentiated outer retina (Figs. 2A 2B 2C ). In addition to thinning of the mi/mi neural retina and DER, pronounced changes in the morphologic development of the mi/mi neural retina and DER were clearly visible by P11. The wild-type neural retina was stratified and OS had elongated (Fig. 2D) . The mi/mi neural retina and DER stratification was becoming disorganized, OS failed to elongate into the inner photoreceptor space, and photoreceptor rosettes with short OS were observed in the neural retina (Figs. 2E 2F , arrows). 

In both the wild-type and mi/mi neural retina, HPC-1–labeled amacrine cells in the GCL, processes in the IPL and INL, and faint horizontal cells in the OPL (Figs. 2G 2H , arrows). Amacrine cells were also labeled with HPC-1 in the DER (Fig. 2I , arrows). The unexpanded RPE was not labeled with HPC-1 (Fig. 1C , arrowheads). 

During the third postnatal week (PW3), the wild-type neural retina was fully differentiated, whereas the mi/mi neural retina stratification had progressed, but was increasingly disorganized (Figs. 3A 3B ). OS had yet to elongate outside of photoreceptor rosettes (Fig. 3B , arrows). The DER maturation, although inverted, mirrored mi/mi neural retina development and its subsequent disorganization (Fig. 3C) . 

Rods and rod OS were labeled with Ret-P1 in the wild-type neural retina (Fig. 3D) , but only rods were labeled in the mi/mi neural retina and DER (Figs. 3E 3F) . The wild-type neural retina (Fig. 3G) , mi/mi neural retina (Fig. 3H) , and DER (Fig. 3I) contained otx-2–labeled bipolar cell nuclei, although the mi/mi neural retina and ectopic INL contained fewer labeled cells (Figs. 3G 3H 3I , arrows). By P21, two bands of SVP38 labeling indicated synaptic connections in the OPL and INL of the wild-type (Fig. 3J , arrows) and mi/mi neural retina (Fig. 3K , arrows) and the DER (Fig. 3L , arrows). Taken together, these data indicate that the DER differentiated as neural retina, which contained defined retinal cell types and synaptic layers, whereas the ventral RPE, which did not label with neural retina markers, was arrested in an morphologically undifferentiated state. 

mi/mi Photoreceptor Degeneration

To quantify the changes observed in mi/mi retina, we counted the number of photoreceptor nuclei spanning the ONL of postnatal wild-type and mi/mi mice (Fig. 4) . Progressive retinal disorganization was followed by photoreceptor degeneration at later ages. Early in postnatal development, the mi/mi neural retina contained at least 50% fewer photoreceptors spanning the ONL compared with the wild type. A progressive decrease in photoreceptor number in the mi/mi ONL was apparent by PW3 to PW5 and continued over the next 32 weeks (8 months). When the photoreceptor number was determined between 8 to 16 weeks, there was on average 1 nuclei spanning the ONL compared with 10.9 in the wild type (Fig. 4) . A similar pattern was observed in the DER. Photoreceptor loss was observed throughout the retina, although it occurred first in the central ONL. By 32 weeks, the ONL was completely absent, and the remaining retinal layers were either missing or severely disorganized (data not shown). 

Discussion

These results demonstrate that when Mi is unable to bind DNA, early retinal cell differentiation proceeds normally, but RPE development is abnormal. The most striking change is the differentiation of dorsal RPE into ectopic neural retina. In the ectopic and normally situated mi/mi neural retina, neural retinal cell-type specific markers are expressed with an onset and distribution pattern similar to the wild type, whereas ventral RPE remains a monolayer throughout development. Although early neural retina development is not influenced in the mi/mi mouse, after birth, the retina becomes disorganized, OS fail to elongate, and significant photoreceptor loss is observed. 

Scholtz and Chan ¹⁴ previously reported morphologic defects associated with the mi/mi mutation during mouse eye development and focused on determining the cause of microphthalmia as well as gross morphologic changes occurring in lens and retina. The results in this study not only confirm those of Scholtz and Chan, ¹⁴ but also add new data concerning specific cellular differentiation and degeneration events in developing and adult mi/mi eye. For instance, the presence of an RPE monolayer surrounding the DER argues that the neural retinal has not folded in on itself, which is likely the case with the ectopic retina observed by Scholtz and Chan near the optic disc. Furthermore, transcription factor labeling indicates that a cell fate change in the DER occurs early in optic cup development. Retinal cell-type–specific antibodies show that the DER matures in a pattern similar to the wild type and the normally situated mi/mi neural retina. In this study, the early postnatal neural retinal disorganization and photoreceptor degeneration was characterized and quantified by cell counts. 

Neural Retina Development

During early eye development, RPE likely facilitates but is not required for neural retina differentiation. For example, in vitro studies have shown that pigment-epithelial derived growth factor (PEDF) induces neuronal phenotypes in Y79 cells and increases RPE pigment granule maturation. ^{18 19} Despite PEDF’s possible role as a neuronal differentiation factor, other in vitro evidence suggests that RPE is not required for retinal cell generation. Embryonic retina cell cultures lacking RPE cells differentiate into phenotypes normally generated first in retinal development. ^{20 21} Progenitor cells cultured from later ages produce later-born cell types. ^{20 21} Therefore, it is not surprising that although the RPE remains immature in the mi/mi mouse, the neural retina continues to develop. Data presented here do not preclude the possibility that other RPE-derived factors, missing in the mi/mi mouse, influence overall retinal histogenesis. 

RPE Development

In addition to altered gene expression and pigmentation loss, the mi/mi RPE does not differentiate morphologically. In the DER, Otx-2 and Mi labeling patterns are similar neural retina, whereas ventrally otx-2 and Mi labeling corresponds to wild-type RPE. Studies in avian RPE have demonstrated that changes in mi expression can change the response of RPE to growth factor induced transdifferentiation. ¹³ This suggests that, in the mi/mi eye, tissue compartmentalization is disrupted and regulation of RPE phenotype is dependent on mi expression. 

One possibility for differences in the dorsal/ventral differentiation of mi/mi RPE is that when mi is nonfunctional, RPE competence to form neural retina is dependent on spatial location (i.e., there is an intrinsic difference in dorsal versus ventral RPE). Spatial location could be dictated by differential transcription factor or growth factor receptor expression. In developing Xenopus eye, the transcription factor Xbr-1, a member of a novel class of homeobox genes, is differentially expressed in dorsal ciliary margin. ²² Similar dorsal/ventral-specific distributions of other transcription factors and/or growth factor receptors may explain how differing phenotypes observed in mi/mi RPE are generated. This hypothesis predicts that dorsal and ventral RPE would respond differently to environmental cues. The dorsal RPE layer has the capacity to generate neural retinal, whereas ventral RPE does not. 

Alternatively, dorsal and ventral RPE may have equal competence to respond to exogenous stimuli regardless of location. Thus, the different phenotypes seen in the mi/mi RPE would be influenced by differential dorsal/ventral expression of various enzymes and growth factors. For example, if bFGF (a signal for transdifferentiation) is expressed in a gradient with high levels found dorsally, then ventral eye may not receive signals to form neural retina. There also may be complex interactions between other extrinsic factors involved in dorsal ventral patterning. The bone morphogenetic proteins are necessary for initial optic cup formation and specification of the dorsal axis. ²³ The ventral RPE may be influenced by retinaldehyde dehydrogenase and retinoic acid, which are important in the formation of ventral characteristics. ^{24 25} Thus, if dorsal and ventral RPE have equal competence, then we would predict that experimental manipulation of external dorsal/ventral influences on the RPE so the influences are evenly distributed, the ventral RPE would generate neural retina. These hypotheses are currently being tested. 

The ectopic retina in the RPE layer of the mi mutant indicates that, in mammals, the RPE has the ability to transdifferentiate in vivo and that mi is essential for RPE differentiation. Determining the molecular signals and growth factor interactions necessary for changing the competence of mammalian RPE will help to determine how a nascent neural retina can be generated from RPE in mammals over a longer developmental window. 

 Supported by National Institutes of Health Grant EY06890 (KMB) and the Foundation Fighting Blindness. CJB is a Senior Investigator of Research to Prevent Blindness, Inc.

 Submitted for publication June 4, 1999; revised September 17, 1999; accepted October 7, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

E13/14 wild-type and mi/mi neural retina stained with cresyl violet and labeled with Mi and Otx-2 antibodies. E17 wild-type and mi/mi neural retina labeled with an 8A1 antibody. (A) In the wild type, the outer optic cup has differentiated into a single layer pigmented epithelium that lines the most outer part of the neural retina (arrows). In the neural retina, the ganglion cell layer (GCL) appears centrally and ganglion cell axons form a well-defined optic nerve (ON). (B) The mi/mi eye shows a dorsal hyperproliferation of the outer optic cup (arrowheads) and a slight ventral thickening (arrows). (C) An antibody to mi indicates that in the wild-type mouse, label is present in the RPE cell nuclei (arrowheads) with no staining in the neural retina. (D) In the mi/mi eye, the Mi antibody does not label in the neural retina or the dorsal expanded region of the RPE (arrows), but does label RPE that is a single cell layer. (E) An antibody to Otx-2 labels a row of RPE nuclei and cell nuclei in the central neural retina in the wild type (arrows). (F) The mi/mi eye has Otx-2 labeling in the RPE monolayer and in scattered cell nuclei in the central neural retina (arrowheads). (G) In the E17 wild-type neural retina, 8A1 labels the GCL (arrows) and migrating horizontal cells (arrowheads). (H) 8A1 labels the same cell types in the E17 mi/mi neural retina as seen in the wild-type neural retina (arrows and arrowheads). (I) At E17, masses of neurites in the most scleral (s) part of the dorsal thickened RPE label with 8A1, which is normally never expressed in the RPE layer (arrows). Nr, neural retina. Scale bars, (A, B) 100 μm; (C, D) 25 μm; (E, F) 50 μm; (G, H) 8μ m; (I) 12 μm.

Figure 2.

 

Cresyl violet staining of a P2 and P11 wild-type and mi/mi retina and HPC-1 labeling of P11d wild-type and mi/mi neural retina. (A) The wild-type neural retina at P2 has a fully differentiated GCL and inner plexiform layer (IPL). The RPE layer lines the most outer part of the neural retina as single pigmented layer. (B) The mi/mi neural retina is structurally similar to wild type, with a stratified GCL and IPL. (C) The mi/mi dorsal expanded region continues to develop along a neural retina pathway, with cellular stratification becoming more obvious compared to earlier ages (GCL and IPL). (D) By P11, the wild-type neural retina contains a differentiated outer nuclear layer (ONL), outer plexiform layer (OPL), inner nuclear layer (INL), IPL, and GCL. At this age, outer segments (OS) started to elongate. (E) The mi/mi mutant type neural retina has all the cellular and synaptic layers. In the ONL, photoreceptor rosettes are beginning to form (arrows). (F) By P11 the dorsal expanded region there is a clear ONL, OPL, INL, IPL, and GCL. (G) Amacrine cells in the wild-type neural retina label with HPC-1 along the edges of the INL and processes in the IPL (arrows). (H) Labeling for HPC-1 is similar in the mi/mi neural retina (arrows). (I) HPC-1 labeling was found in the INL and IPL of the dorsal expanded region, but no labeling was detected when the RPE thinned to a monolayer (arrows). Scale bar, (A through F) 50 μm; (G through I) 12 μm.

Figure 3.

 

P17 wild-type and mi/mi eyes stained with cresyl violet and labeled with Otx-2 and Ret-P1, and at P21 labeled with SVP38. (A) P17 wild-type retina is fully stratified (ONL, INL, GCL), contains well-formed outer segments (OS), and has a fully differentiated RPE. (B) The mi/mi neural retina is also stratified with a clear ONL, INL, and GCL, but the OS fail to elongate, and photoreceptor rosettes with small OS are common (arrows). (C) Neural retinal stratification of cell and synaptic layers is complete by P17 in the mi/mi dorsal expanded RPE. Clearly defined ONL, INL, and GCL are present. (D) Rod opsin in the wild-type neural retina is localized to the ONL and rod outer segments (ROS). (E) In the mi/mi eye, the neural retina primarily labels for rod opsin in the ONL. (F) The thickened dorsal RPE is oriented such that an ectopic Ret-P1–positive ONL layer is adjacent to the appropriately placed neural retinal layer. (G) Otx-2 in the wild-type neural retina has labeling in RPE (arrowheads) and bipolar cell nuclei in the INL (arrows). (H) Although fewer cells are indicated, bipolar cells label for otx-2 in the mi/mi INL (arrows). (I) In the dorsal expanded region, Otx-2 also labels bipolar cell in an ectopic INL. (J) Labeling with SVP38 is restricted to the OPL and IPL in the wild-type neural retina. (K) In the mi/mi eye, the neural retina has SVP38 labeling in the OPL and IPL. (L) Two bands of SVP38 labeling are detected in the an ectopic OPL and IPL in expanded region of the RPE that has formed neural retina (arrows). Scale bars, (A, G, J) 50 μm; (B, D, E, H, K) 50 μm; (C, F, I, L) 50 μm.

Figure 4.

 

The number of photoreceptor nuclei spanning the ONL of wild-type and mi/mi neural retina. ONL nuclei were counted in five age groups (n = 3, except at PW1 to PW2 n = 4). Averages are reported on the graph, Error bars, SD.

The authors thank Stephen Viviano and Adrienne LaRue for their outstanding technical assistance; Flora Vaccarino for the gift of the Otx-2 antibody; and Julian Martinez, Laurence Leconte, and Ken Wikler for critical comments on the manuscript.