Cone Differentiation with No Photopigment Coexpression

Abstract

purpose. To decide whether the transitory coexpression of cone visual pigments described in the developing rat and gerbil retina is a universal feature of dichromatic mammalian species.

methods. The rabbit, a species widely used in eye research, was selected for the study and a search made for the presence of cones that bound more than one cone antibody during the first postnatal week. To plot the densities of individual cone types and to colocalize the two visual pigments, immunocytochemistry on retinal wholemounts and consecutive tangential sections, respectively, were used.

results. The sequence in which the visual pigments began to be expressed was the same as that observed in other mammals: first, rhodopsin; second, blue pigment; and last, green pigment. The striking increase in blue cone density numbers observed in the rat, however, did not occur in the rabbit. Instead, some days after the first blue cones appeared, the green cones also started to express their visual pigment, and this cone type soon outnumbered the blue cones. Within the limits of the immunocytochemical method, it was established that unlike the developing rat, the presence of double-labeled cones was not a character of the rabbit retina.

conclusions. Visual pigment coexpression is an interesting phenomenon of retinal development, however, it is not the exclusive scenario of photoreceptor differentiation. Each species must be carefully studied before deciding whether its retinal cones synthesize both pigments during retinal development.

Each retinal photoreceptor in mammals has been thought to possess a single visual pigment that is responsible for the absorbance spectrum of the visual cell. To map the distribution of color cones on retinal samples immunocytochemistry with anti-visual pigment antibodies has been especially useful, because neither sophisticated instrumentation nor surviving retinal preparations are needed for the selective labeling. ¹  

When examining the development of cone patterns in various species, a rather uniform sequence of emerging photoreceptor phenotypes is observed. ^{2 3} The first visual cell type to appear is the rod. Some days later, the short-wave–sensitive (S) cones emerge, and last, at approximately the second postnatal week, the middle-to-long–wavelength (M/L) sensitive cone phenotype also can be detected. This sequence, as follows from the technique, is based on the onset of synthesis of the respective opsin proteins. All species examined so far, either with segregated cone fields (mouse, guinea pig, rabbit, and other species) or with homogeneous cone distribution (rat, gerbil, and others) show the same sequence, except in the primate fovea, where there is still some controversy about the preceding cone type. ^{3 4}  

Further studies in the developing rat and gerbil have revealed that much higher concentrations of S cones are produced than expected from adult densities, and visual pigment coexpression during development has also been detected. ⁵ The subsequent steep decrease of S cone numbers at postnatal weeks 2 to 3 and the temporary coexpression of S and M pigments have led us to formulate the transdifferentiation theory of cone development in these species. The proposed scenario involves the early emergence of a high number of S cones (default pathway), the majority of which would later stop synthesizing the original pigment and begin to express the M pigment. These cones comprise the definitive green (M) cones. The other cones that do not undergo this shift, comprise the definitive blue or ultraviolet (S) cones. ⁵ No data are available on the physiological significance of temporary visual pigment coexpression. 

The other relevant question is whether transdifferentiation is the only way in which the two basic cone types develop. Now we have tested the rabbit, a species that exhibits a typical divided retina with many M and a small number of S cones in the major part (superior and central regions) of the retina, whereas the most ventral crescent (blue streak) contains exclusively S cones. ⁶ There is a narrow stripe at the borderline of the two fields where a few double-labeled cones occur. ⁷ In this feature this species differs greatly from those that exhibit homogeneous cone distribution and transitory visual pigment coexpression all over the developing retina. ⁵  

Methods

Various ages (13 age groups) between postnatal days 1 and 24 of common rabbits (pigmented Dutch belted and albino New Zealand) obtained from local breeders were killed with decapitation after a prolonged ether narcosis. After enucleation, the posterior eyecup was fixed in 4% paraformaldehyde (0.1 M phosphate buffer [pH 7.4]) for 2 days. Squares (2 × 2 mm) were cut from the superior part of the eye halfway between the optic nerve head and ora serrata. For comparison, samples were also taken from the blue streak and central regions. After fixation, the retinal pieces were treated in two ways. For tangential sectioning, the retinas were not detached from the underlying choroid and sclera, and the eyeball wall was processed in its whole thickness. After dehydration, the pieces were embedded in Araldite and flatmounted, and 1-μm-thick sections were cut in a plane parallel to the retinal surface on an ultramicrotome. For wholemount immunocytochemistry, the retinas of the selected pieces were detached and collected in buffer. 

Two anti-visual pigment antibodies, COS-1 and OS-2, specific for the M/L and S pigment, respectively, were used in our studies. COS-1 was a hybridoma supernatant diluted 1:50. OS-2 was an ascites—therefore, further diluted to 1:5000. Both dilutions were the same as those used in our previous studies in mammals. ^{1 2 5} The bound antibodies were labeled with biotinylated anti-mouse antibody and the ABC technique (Vectastain; Vector, Burlingame, CA) followed by diaminobenzidine (DAB; Sigma, St. Louis, MO) as a chromogen. DAB was used in the presence of hydrogen peroxide. The wholemounts were either reacted with ABC-DAB or with fluorescein isothiocyanate (FITC)–conjugated secondary antibodies. Control reactions were performed omitting the primary and/or secondary antibodies. The reactions were inspected with a microscope (Axiophot; Carl Zeiss, Oberkochen, Germany) using Nomarski optics or the appropriate filter set for FITC, respectively. The photographs were taken by either conventional microscope cameras or a digital camera (Eastman Kodak, Rochester, NY). In the latter case, the digital images were processed with image analysis software (PhotoShop ver. 5.0; Adobe, San Jose, CA) and printed with a thermodiffusion photoprinter. 

For establishing the dual nature of individual cone cells, tangential sections taken from the outer segment level were used. Consecutive sections were reacted alternately with the two antibodies. By comparing the identical images derived from adjacent sections, each cone outer segment could be analyzed against the panel of both antibodies. Photographs were obtained from identical areas and either mounted one under the other, marking the labeled outer segments with arrows, or superimposed on one another, marking the immunopositive elements with digital coloring. 

For cell counting, two to three animals were used in each age group. From each retina at least five sample pieces were taken from the various areas specified earlier. To study the time course of the two cone populations, we carefully counted the labeled elements and calculated cone densities. Both OS-2– and COS-1–positive cone densities were plotted as a function of postnatal days, and the two curves were then diagrammatically demonstrated. In the density series, only eyes of the same side (right) were included. The applied image processing algorithms (Matlab ver. 5.2; The Mathworks, Natick, MA) were developed by the manufacturer. To avoid any bias often encountered using computerized counting techniques, we regularly controlled densities with manual counting and averaged a large number of counts obtained from the same area. 

Experimental animals in this study were managed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Results

Single-label immunocytochemistry with anti-visual pigment antibodies was used to establish the sequence of appearance of individual photoreceptor phenotypes in the rabbit retina. On wholemounted retinal pieces, similar to other mammalian species, S cones were the first cone photoreceptors to synthesize their characteristic visual pigment. From postnatal days 2 to 3 onward, scattered OS-2–immunopositive elements were seen in the samples derived from the superior–central regions. The second cone phenotype identified by visual pigment antibodies (COS-1) was that of the M-sensitive type. The first M/L cones were identified at the end of the first postnatal week, normally not earlier than postnatal day (P)6 to P7. In contrast to other species investigated in this way (rat, gerbil, and others), the density increase of S cones remained modest at all times (maximum density: 2000/mm² at P14) and no characteristic peak was detectable (Fig. 1) . Soon after reaching the maximum, the S cone density numbers started to slowly decrease and reached adult levels by the end of the first month. 

The second cone population showed a totally different density curve. After the first COS-1–positive cones appeared, the cone cell numbers increased rapidly and reached their maximum by the end of the second postnatal week (density: 12,000/mm²). After this maximum a slight decrease was observed, but the densities remained constant at approximately 9000/mm². The shape of the curves was found to be the same when counting was performed on other superior–central areas of the rabbit retina (not shown). No M cones, however, were ever detected in the blue streak, which is the ventral, crescent-shaped area of the rabbit retina (not shown). Instead, all cones located there were found to be of the S type with a density comparable to that of the M and S cones in the superior–central retina. ⁶ Although the calculated densities were not corrected with retinal growth, the two curves (Fig. 1) could be compared readily because of the presence of the same bias in both of them. Our measurements showed that the surface area of the developing retina increases considerably between P14 and P24, explaining the decrease in cone density after the peaks. 

We observed that the 10-fold overproduction of OS-2 cones was missing and that the COS-1 cone density increase was not accompanied by a simultaneous decrease of OS-2–positive cone densities, the two very features that were reported in mammals (e.g., rat) with coexpressing cones during development. 

The next step was to decide whether the observed difference in the time course of cone development is reflected in the presence of dual cones. We reacted consecutive tangential sections alternately with COS-1 and OS-2, respectively. Identical areas of adjacent sections were then compared, and double-labeled cone outer segments were sought (Figs. 2 3) . Using the residual pigment epithelial cells and/or identifiable parts of the cone mosaic for orientation, we either placed the carefully trimmed members of paired photos on one another (Figs. 2 3) or superimposed the digitalized photos of consecutive sections (not shown). Although both cone types were already present from P6 to P7 and onward, the low number of outer segment profiles on tangential sections did not allow us to locate them and document their pattern earlier than approximately P12. It is only at this age that outer segments were long enough to be encountered on two or more neighboring sections. With this approach, potential double-labeled elements could be identified easily. Including a total of 30 individual animals and almost 100 section pairs derived from various ages of our developmental series, no double-labeled cones were found in the greatest part (superior–central regions and blue streak) of the rabbit retina. It was only a narrow zone of transition at the junction of the blue streak and the superior–central area where, similar to the adult rabbit, a few dual cones were encountered. 

Discussion

Due to the inherent limits of the immunocytochemical approach, relatively little is known about the differentiation of cone subpopulations. Each species studied so far shows the same temporal pattern: S cones followed by the later-appearing M/L cones. ^{2 3 5} The only exception in which this sequence has not been confirmed yet is the primate retina. Wikler and Rakic ⁴ showed an opposite order, with precocious M/L cones occupying distinct spots of the developing mosaic. Bumsted et al., ³ in contrast, were unable to show any M/L cones that preceded S cones in the fovea, indicating that in this special retinal area, both cones might start to express visual pigments simultaneously. In this latter study, similar to subprimate mammals, the first identifiable cones in the peripheral part of the primate retina were found to be of the S type. In yet another report, Wikler et al. ⁸ presented evidence that S opsin mRNA appears before L/M opsin mRNA in the monkey retina. 

Apart from the controversy about the primate retina, the generalization about the precedence of S cones against M/L cones in mammals seems to be correct. This sequence is constant in species with different life styles and/or cone distribution patterns. A number of species have been identified exhibiting a varying degree of dorsoventral heterogeneity in their cone distribution patterns. ^{2 6 7 9 10 11} In these species, similar to those with homogeneous cone distribution (rat and gerbil ⁵ ; tree shrew [Lukáts Á and Szél Á, unpublished data, 1999]), the same sequence (rhodopsin-S cone pigment–M/L cone pigment) was found. By comparison, it would be logical to assume that the transdifferentiation mechanism taking place in the latter animal group is a general scenario for all mammals, at least in those parts of the retina that are populated by both cone types. In the present study, however, we provide evidence that transdifferentiation practically does not take place in the developing rabbit retina, clearly indicating the occurrence of a second scenario in which each cone type develops independently. The low number of species studied so far does not allow for a conclusion about whether asymmetric cone distribution and transdifferentiation are mutually exclusive features and whether further mechanisms also occur. We are also left with the question of how the double-labeled cones in the transition zone of the divided retinas come about. Experiments are under way to address these issues. 

The question of whether the ephemeral dual cones of the rat retina and the small number of dual cones in the rabbit transition zone have any visual function have not been tested yet. Even though no sensible visual role can be attributed to the temporary dual cones, there are a number of relevant reports on submammalian vertebrates. In these, the presence of more than one visual pigment within one cone reflects important spectral changes of the eye in conjunction with lifestyle changes, such as metamorphosis or migration (e.g., reference 12) The transition area, in turn, may represent a zone of perturbation, wherein the definitive phenotype of a few cones remains undetermined. Due to the negligible amount of these visual cells, it is highly unlikely that they have any important visual function. Further immunocytochemical and molecular genetic studies are needed to establish the developmental biological significance of visual pigment coexpression during mammalian retinal differentiation. 

 Supported by grants from the Hungarian Scientific Research Fund (OTKA T-029048), Hungarian Ministry of Health (ETT 537/96), Biomed2, Brussels, Belgium (IC20 CT97 0025), and Institut National de la Santé et Recherche Médicale East West, Paris, France (4E006C).

 Submitted for publication February 7, 2000; revised May 9, 2000; accepted May 10, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Time course of immunopositive cones in the developing rabbit retina as a function of postnatal days. Samples were taken halfway between the optic nerve head and ora serrata in the superior part of the retina. The density of S cones (black columns) increased slowly and did not reach a high peak. In contrast, the density of M cones increased steeply and reached a maximum at the end of the second postnatal week (shaded columns). The ratio stayed the same in the adult.

Figure 2.

 

Tangential sections of P12 rabbit retina (superior part) reacted with OS-2 (A) and COS-1 (B). The consecutive sections cut at the outer segment level show the mosaic of the same photoreceptors; therefore, the staining pattern of each cone can be established. The tiny dark dots in the upper left corner of the sections represent the granules of the pigmented epithelial cells. The larger round structures marked with arrows are the immunopositive cone outer segments. Note that the density of COS-1–labeled M cones (B, thick arrows) considerably exceeded that of the OS-2–stained S cones (A, thin arrows). All cones were labeled by only one of the two antibodies, indicating the absence of coexpressing (dual) cones. Scale bar, 10 μm.

Figure 3.

 

High-magnification micrographs derived from the previous section series showing sections reacted with OS-2 (A) and COS-1 (B). The consecutive sections have been aligned using the mosaic of rod and cone outer segments as a landmark. A few readily identifiable (rod) outer segments not labeled by any cone antibodies are marked with long vertical arrows. The oblique arrow points to an OS-2–positive cone that is not labeled by COS-1. Short vertical arrows mark COS-1–labeled outer segments that are left unstained by OS-2. All cones are labeled only by one of the two antibodies, indicating the absence of coexpressing (dual) cones. Scale bar, 5 μm.

The authors thank Howard M. Cooper, Bengt Juliusson, Malcolm von Schantz, and Theo van Veen who contributed to this work with materials and fruitful discussions.