In Vivo and In Vitro Development of S- and M-Cones in Rat Retina

Abstract

Purpose.: Organ cultures of the rodent retina could provide a powerful tool in the study of cone development and differentiation. Previous attempts, however, have failed to show M-cone development in organ cultures of the mouse and rat retina. This study mimicked the in vivo dynamics of S- and M-cone development in a culturing approach for the postnatal rat retina.

Methods.: Retinas of Brown Norway rats were collected at different developmental ages (postnatal day [P]0–P270) to study cone development in vivo. For culturing, the retinas were prepared from P0 to P2 animals and allowed to develop in organ culture for 2 to 15 days. Subsequently, opsin expression was analyzed immunohistochemically and morphometrically.

Results.: In control retinas, S-opsin was already expressed at birth, whereas M-opsin was detected after P4. The maximum density of S-opsin–positive cones was reached at P10 (∼17,000 cells/mm²) and of M-opsin–positive cones, at P12 (∼14,000 cells/mm²). The number of both cone types decreased gradually thereafter to ∼1,000 S-opsin cones/mm² and ∼4,000 M-opsin cones/mm² in the adult. In culture, both cone types developed with dynamics of appearance comparable to those in vivo, with a peak density of ∼12,300 cones/mm² for S-opsin and ∼7,500 cones/mm² for M-opsin labeling.

Conclusions.: These results in rat retina showed for the first time that cone development and expression dynamics can be mimicked in organ culture. With this experimental approach, it will be possible to evaluate aspects of cone development under controlled experimental conditions and to elucidate factors crucial for proper cone differentiation.

The rod-dominated retinas of most mammalian species contain two distinct cone populations with different wavelength sensitivities. The short-wavelength–sensitive cones (S-cones) that have peak sensitivities in the blue or UV part of the spectrum are usually outnumbered by the middle-wavelength–sensitive cones (M-cones). ^1–3 The distribution of different cone subtypes has been well studied in a variety of species; however, the factors that regulate their development at the molecular level are poorly understood. All cell types of the vertebrate retina derive from the same pool of common multipotent progenitor cells according to a strictly regulated timetable: The early born ganglion cells are followed by cone, amacrine, and horizontal cells, whereas rod, bipolar, and Müller cells are generated in the later phases. ^4–6 During development, cell fate is determined by complex regulatory mechanisms, and the intrinsic program is altered by extrinsic signals. ^7–11 Our present knowledge about the regulatory factors that control photoreceptor cell fate and development has been reviewed recently in several publications. ^7–11  

Cones undergo their final mitosis in the early prenatal period, yet there is a significant delay before they express their mature phenotype; and in rodents, major events of cone development, such as outer segment formation, opsin expression, and synaptogenesis extend into the first postnatal weeks. ^12–15 Aside from some notable exceptions, ¹⁶ in most mammalian species studied so far, the first detectable visual pigment during development is rhodopsin in the rods, followed by the appearance of S- and finally M-opsin in the cones. ^1,17 The factors that control opsin expression and the development of cone patterns are largely unknown, and despite general similarities, they may differ from one species to the next. In some species, such as the rat, where the cones are evenly distributed across the retina, ¹⁸ most of the developing cones coexpress both the S- and the M-opsin (transient dual cones) in the early postnatal life. The absence of dual cones in the adult supports the hypothesis that M-cones develop via transdifferentiation from initially S-opsin–expressing cells. ¹⁹ It is still unclear whether this developmental scheme is universal among mammals, but transient dual cones have also been identified in several other species, including the gerbil, tree shrew, and human. ^1,20  

Recent experiments indicated that the differentiation of M-cones is closely related to that of S-cones and that thyroid hormone via its β2 receptor (TRβ2) has a major influence on opsin expression. ^21–23 In mouse-knockout experiments, it has been demonstrated that TRβ2 is necessary to activate M-opsin expression, and in cooperation with the retinoid X receptor γ (RXRγ), TRβ2 also suppresses S-opsin production. ^21,24 The ligands of RXRγ in S-opsin repression are still unidentified, and whether the thyroid hormone alone can regulate opsin patterning is an open question. In addition, Crx, an Otx-like homeodomain transcription factor, and RORα and -β—members of the retinoic acid receptor-related orphan receptor family—have also been shown recently to play a role in the activation of cone-specific opsin expression. ^25–27  

Although organ cultures could provide a powerful tool for the study of putative regulatory factors under defined experimental conditions, all attempts so far to culture embryonic and newborn mouse retina have resulted in incomplete photoreceptor development with the complete lack of M-opsin expression. ^24,28–30 The fact that M-cones were detectable only when retinal explants were made from P3 or older pups indicates the possibility that soluble factors were acting during this critical period that were missing from the culturing medium. Similar results with only low levels of M-opsin mRNA and the complete absence of immunohistochemically detectable M-opsin–expressing cells were described in a rat explant culture. ³¹  

We present an organ culture of the postnatal rat retina that shows S- and M-opsin expression comparable to that in age-matched in vivo controls, indicating that cone transdifferentiation occurs in vitro. Our findings support the hypothesis that this process is under control of soluble factors and will help elucidate the nature of these factors and their relevance for proper cone development. 

Materials and Methods

Tissue Processing

Animal procedures were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Retinas from Brown Norway rats (Charles River, Sulzfeld, Germany) of different postnatal ages (P0, P2, P4, P6, P8, P10, P12, P14 P18, P30, P60, P120, and P270) were analyzed as wholemounts or radial sections and serve as age-matched controls for the cultured retinas. The day of birth was denoted as P0. 

After anesthesia and decapitation of the rats, the eyes were immediately enucleated and the cornea and lens were removed. For radial sections, the remaining eye cup was immersed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4) for 30 minutes at 4°C. After fixation, the eye cups were washed several times in 0.1 M PB, cryoprotected in 30% sucrose in 0.1 M PB at 4°C overnight, embedded in cryomatrix (Thermo Fischer Scientific, Waltham, MA), and frozen in liquid nitrogen. Sections (12 μm thick) were cut on a cryostat and collected on gelatin-coated slides. After air-drying, the sections were processed immediately or stored at −20°C. For wholemount immunohistochemistry, the retinas were carefully detached from the eye cups and fixed as just described. 

Preparation of Organ Culture

Retinal organ cultures were prepared as described elsewhere. ³² Briefly, the eyes of decapitated Brown Norway rats (P0-P2) were enucleated under sterile conditions and transferred to a Petri dish containing ice-cold sterile Ames medium (Sigma-Aldrich, St. Louis, MO) enriched with glucose (20 mM). After enzymatic digestion of the eyeball, the retina, together with the attached pigment epithelium, was dissected and four radial cuts were made to flatten it. Retinas were placed on culture plate inserts (Corning Life Sciences, Lowell, MA) with the ganglion cell layer uppermost. The explants were cultured in 1:1 DMEM/F12 medium (Invitrogen Corporation, Carlsbad, CA) and supplemented with: α-tocopherol (1 μg/mL), ascorbic acid (100 μg/mL), retinol (10 μM), retinyl acetate (10 μM), hydrocortisone (20 nM), insulin (5 μg/mL), progesterone (20 nM), putrescine (100 μM), selenium (20 nM), taurine (3 mM), transferrin (1 μg/mL), triiodothyronine (2 ng/mL) and 10% fetal calf serum (Invitrogen Corporation). All supplements were cell culture grade and obtained from Sigma-Aldrich, unless otherwise stated. The cultures were kept without shaking or rotation in an incubator with an atmosphere of 5% CO₂, balanced air and 100% humidity at 37°C. The medium was changed the first day in culture and every second day thereafter. After 2 to 15 days in vitro (DIV), the cultures were fixed and analyzed as radial sections or wholemounts. The results were compared with those of age-matched control retinas. 

Immunohistochemistry

Photoreceptor phenotypes were characterized by immunostaining the cryostat sections and wholemounts of age-matched control and cultured retinas with affinity-purified JH455 and JH492 rabbit polyclonal antibodies (Table 1). ^33,35 After rehydration, the radial sections were preincubated with phosphate-buffered saline (PBS; 50 mM, pH 7.4) containing 20% normal goat serum and 0.03% Triton X-100 (Sigma-Aldrich) for 2 hours at room temperature to block nonspecific antibody binding. Subsequently, the sections were incubated overnight at 4°C with specific primary antibodies. The immunoreaction was visualized with Cy3-conjugated anti-rabbit antibody (Rockland, Gilbertsville, PA) diluted 1:1000. For the visualization of cones in wholemounts, the same protocol was performed with prolonged incubation times. 

Table 1.

 

Primary Antibodies

Antigen	Host	Source	Working Concentration	References
S-opsin (JH455)	Rabbit polyclonal	Gift of Jeremy Nathans	1:20,000	Wang et al. 33
M-opsin (JH492)	Rabbit polyclonal	Gift of Jeremy Nathans	1:20,000	Wang et al. 33
M-opsin (AB5405)	Rabbit polyclonal	Millipore Billerica, MA	1:1,000	Gaillard et al. 34
M-opsin (COS-1)	Mouse monoclonal	Gift of Ágoston Szél	1:50	Röhlich and Szél 35

For wholemounts, fixed retinas were dissected from pigmented epithelium and incubated for 2 hours in blocking solution, for 3 days in the primary antibody, and for 1 day in the secondary antibody. The specificity of M-opsin–specific reactions in culture was determined by performing double labeling experiments on retinal wholemounts using a second set of M-opsin–specific antibodies. For this, COS-1, a monoclonal hybridoma supernatant ³⁵ and AB5405, a commercial available rabbit polyclonal antibody ³⁴ were used (Table 1). The reactions were visualized by species-specific Alexa Fluor 488 and Alexa Fluor 594 secondary antibodies (Invitrogen Corporation), both diluted in 1:200. Stained retinas were mounted flat, with the photoreceptor layer facing upward and coverslipped. Controls for all reactions were performed by omitting the first antibody. 

Morphometric Analysis

The density of immunoreactive cells was determined in retinal wholemounts by confocal microscopy (LSM 410; Carl Zeiss Meditec, Oberkochen, Germany). The images were sampled at 1-μm intervals in a minimum of 14 and a maximum of 30 planes, with a ×40 water-immersion objective. For morphometric analysis, a wholemount was divided into four quadrants. From each quadrant, a central area of 0.1024 mm² was selected 1 mm away from the optic disc, resulting in four areas per retina. Images were analyzed with special software (Scion Image program for Windows Beta 4.0.2; Scion Corporation, Frederick, MD). They were first converted from the gray level pixels into black or white, with the threshold set at 200, as it proved to optimize discrimination of cells from dust and scratches. The “analyze particles” function was used to export to a text file a list of parameters for each particle present on the picture. Particles with a diameter far different from that of a photoreceptor were discarded. Data were not corrected for shrinkage. 

For validation of the counting method, randomly selected images were also counted manually by an experienced investigator. Differences between the experimental groups were tested statistically by a two-tailed unpaired t-test (Prism 4 for Windows; GraphPad Software, La Jolla, CA). All data shown represent the means and standard errors of the mean from at least four separate experiments. Representative pictures were taken from the central, most differentiated area of the retina. Confocal images of double-labeled specimens were obtained by microscope (Radiance 2100 Multi Photon Imaging System [Bio-Rad, Munich, Germany] coupled to an Eclipse E800 [Nikon, Tokyo, Japan] with LaserSharp 2000 software [Bio-RD]). Image management software (Photoshop; Adobe, San Diego, CA, and Confocal Assistant [free software, copyright Todd Clark Brelje, University of Minnesota, Minneapolis, and available at http://bipl.umn.edu/downloads]) was used for primary image processing. 

Results

Postnatal Cone Development

In the control rat retinas, three distinct phases of S- and M-opsin expression were identified during the postnatal period. The first phase corresponded to the first 10 to 12 days of postnatal development. S-opsin expression was already present at birth (∼2860 cones/mm²), whereas M-opsin was detectable only after P4 (∼750 cones/mm²). In this period, the density of both types of labeled cones increased progressively, up to a peak value at P10 (∼17,000 cones/mm²) for S-opsin and P12 (∼14,000 cones/mm²) for M-opsin (Figs. 1, 2; Table 2). In this early stage of cone development, the opsin-specific antibodies labeled the whole cell (Fig. 3B, 3D). 

Figure 1.

 

Opsin expression in retinal wholemounts at different developmental stages. The retinas were treated with JH455 and JH492 antibodies to show S-opsin (left) and M-opsin (right) labeling, respectively. The photomicrographs were taken from the central, most differentiated areas. Scale bar, 100 μm.

Figure 2.

 

Quantification of S- and M-opsin expression during retinal development. S-opsin expression was already present at birth, increased in number until P10, and decreased subsequently. M-opsin became detectable from P4 on and increased in number until P12. Although they also decrease in number subsequently, M-opsin–expressing cones strongly outnumber S-opsin–expressing cones in the adult.

Table 2.

 

In Vivo and In Vitro Time Course of S-opsin– and M-opsin–Expressing Cones

Postnatal Day	Group	S-opsin–Expressing Cones		M-opsin–Expressing Cones
		Mean ± SD	P	Mean ± SD	P
0	P0	2,861 ± 674.6		0 ± 0.0
2	P2	7,246 ± 869.9		0 ± 0.0
4	P4	9,160 ± 854.9	0.5357	752 ± 160.7	0.0822
	P2 DIV2	8,789 ± 739.1		537 ± 129.1
6	P6	11,045 ± 402.7	0.0002	2,568 ± 243.7	0.1993
	P2 DIV4	9,258 ± 225.5		2,344 ± 194.0
8	P8	13,691 ± 474.4	0.0050	5,830 ± 567.3	0.0098
	P2 DIV6	12,363 ± 393.9		4,648 ± 283.5
10	P10	16,914 ± 663.7	<0.0001	11,318 ± 380.6	<0.0001
	P2 DIV8	9,990 ± 304.9		7,500 ± 186.0
12	P12	13,350 ± 516.6	<0.0001	13,975 ± 693.9	<0.0001
	P2 DIV10	6,406 ± 264.9		7,529 ± 178.7
14	P14	10,781 ± 1,005.0	<0.0001	11,787 ± 224.1	<0.0001
	P2 DIV12	5,469 ± 240.8		7,432 ± 433.1

Figure 3.

 

In the organotypic culture of the pigmented rat retina, both types of cones are present. The retinal layers are well preserved and the morphology of cones developed in culture (B, D) was comparable to that of the age-matched controls (A, C). At P10, the opsin-specific antibodies labeled the whole cone cell body. Scale bar, 50 μm.

The second phase of opsin expression was between P12 and P30. The ratio of S- to M-opsin–positive cells became inverted, and from this time, the density of S-opsin–expressing cones was lower than that of M-opsin–expressing cones. The number of labeled cells decreased rapidly during this period, but the effect was more pronounced for S-opsin–positive cells. By the end of the second postnatal week, M-opsin–positive cones clearly outnumbered S-opsin–expressing cones (Figs. 1, 2, Table 2). In this stage, the opsin-specific antibodies labeled the cone outer segments almost exclusively (data not shown). 

After P30, the density of S-opsin–positive cones decreased further to counts <1000 cells/mm² by P270. From P60, a similar but less robust decrease was found in the number of M-opsin–expressing cones, reaching a minimum at ∼3500 cells/mm² by P270 (Figs. 1, 2). 

In Vitro Cone Development

To analyze cone development under in vitro conditions, we examined organ cultures prepared from P0 to P2 retinas every second day, from DIV2 up to DIV14 in wholemounts or radial sections. The cultured retinas showed clear stratification, and all retinal layers developed by the end of the second postnatal week. Although cultures harvested from P0 pups had more rosettelike formations, the time of explantation had no major influence on overall retinal development and morphology. Cone inner–outer segments were easily recognizable, and their morphology was well preserved (Figs. 3, 4, 5). In contrast to previous reports, ^28–31 both S- and M-opsin were expressed (Figs. 4, 5), even in cultures prepared from P0 retinas (Figs. 3, 6). As in the age-matched control retinas, the opsin-specific antibodies stained the entire photoreceptor somata at younger ages (Fig. 3). Because a careful examination of cultures explanted at P0 or P2 showed no significant difference in the number of S- and M-opsin–positive cells and because of better-preserved overall morphology, cultures harvested from P2 rats were used for statistical analysis. 

Figure 4.

 

S-opsin expression in wholemounts of cultured retinas (A). Quantification of S-opsin–expressing cones (B). Despite the lower density in culture, the expression pattern of S-opsin was similar to that in the age-matched control retinas. **P < 0.01, ***P < 0.0001. Scale bar, 50 μm.

Figure 5.

 

M-opsin expression in wholemounts of cultured retinas (A). Quantification of M-opsin–expressing cones (B). Although in present in lower number, M-opsin–expressing cones developed in the culturing approach and their quantity changed similar to that in age-matched control retinas. **P < 0.01, ***P < 0.0001. Scale bar, 50 μm.

Figure 6.

 

Double labeling for M-opsin in wholemounts of cultured rat retinas. P0 DIV7 retina (A–C) immunostained with AB5405 antibody (A) and COS-1 antibody (B); merged images (C). Only a few structures were labeled with COS-1. P0 DIV14 retina (D–F) immunostained with AB5405 antibody (D) and COS-1 antibody (E); merged images (F). Outer segments are clearly distinguishable, showing a high degree of co-localization. Scale bar, 25 μm.

The density of S-opsin–expressing cones in the organ culture increased to a peak of ∼12,400 cells/mm² by day 6 in vitro and decreased to below 5,500 cells/mm² after 12 DIV (Fig. 4, Table 2). Compared with age-matched control retinas, the peak density of S-opsin–positive cones in culture was lower and the decrease in the number of cells occurred earlier. However, the dynamic of alteration was similar (Fig. 4B). 

Likewise, the density of M-opsin–expressing cones increased during the first days in culture, reaching a plateau of ∼7500 cells/mm² after 8 DIV (Fig. 5, Table 2). As observed for S-opsin–expressing cones, the peak density of M-opsin–positive cones in culture was lower than in age-matched control retinas, but the dynamic of alteration was similar (Fig. 5B). 

To unequivocally confirm the presence of M-opsin expression in our approach, we performed double staining on P0 DIV7 and P0 DIV14 cultures with two other antibodies with confirmed M-opsin specificity (Fig. 6, Table 1). The monoclonal COS-1 recognized M-opsin a few days later than did AB5405 and resulted in slightly weaker reactions at younger ages, but both antibodies labeled the same population of cells as demonstrated at P0 DIV14 (Fig. 6, lower right). 

Discussion

Morphogenesis and opsin expression of cones in rat retinas appear to occur in a sequential manner according to a well-defined timetable. Rhodopsin is detectable first, soon followed by S- and finally by M-opsin expression. ^1,17  

Despite some important differences, our results on cone development of pigmented rats are in good agreement with observations reported earlier in the albino rat. ¹⁹ Although in albino rats the first S- and M-opsin immunoreactivities appeared only at P5 and P9, we detected S-opsin at birth and M-opsin at P4. In addition to the earlier appearance of the first opsin-expressing cells, we counted higher densities for both cone types, and the maximum densities were reached a few days earlier. As no exact isodensity maps were made in either experiment, it is worth mentioning that these values represent average densities and can therefore be used as a first approximation only. There are two possible reasons for the dissimilar findings of our study and those in a previous report ¹⁹ : First, cone densities can be remarkably different across distinct rat strains. ^36–38 Second, the anti-opsin antibodies used may differ in their sensitivities, leading to different cone density detection levels. 

Our analysis of cone expression during postnatal development in the pigmented rat retina has shown that the dynamic of alteration between S- and M-opsin–expressing cones was comparable but shifted in time, resulting in a 4-day delay in the expression of M-opsin. Thus, our data support the cone developmental model of transdifferentiation proposed by Szél et al., ¹⁹ in which all cones express S-opsin only during early postnatal development, but most change their phenotype and differentiate into definitive M-cones. ^19,39  

Other studies also suggest that the default pathway in cone development is S-opsin expression and that switching on M-opsin expression requires developmental signals that are missing in organ culture systems. ^24,28–31 In cultures prepared from retinas older than P3, M-opsin expression was always detectable, independent of the culture medium. Thus, it appears that the critical period of M-cone development comprises the first two postnatal days. 

In contrast to previous organ culture approaches, we were able to demonstrate both S- and M-opsin expression in cultures prepared from retinas younger than P3. Although cone densities were lower in cultured retinas, the temporal and the spatial opsin expression patterns were similar to that of the age-matched control retinas. The expression of M-opsin was independent of the time of explantation, and no significant differences were found between retinas explanted at P0 and P2. M-opsin was detected with three different antibodies, all of which have been shown to be specific. ^33–35  

Taken together, our data strongly indicate that transdifferentiation of cones occurred in our organotypic culture approach and thus can be mimicked under controlled in vitro conditions, offering the possibility for the study of processes of cone development and the factors and signal pathways involved. 

At present, we do not know the factors responsible for the developmental switch from S- to M-opsin production in our culture system. The most likely explanations for the observed differences in cone transdifferentiation potential of rodent organotypic cultures may be the model animal used or, even more likely, the composition of the culturing medium. A detailed comparison of the various culture approaches for rodent retinas ^{24,28–30,40–42} will be necessary to identify these crucial factors. However, such an analysis is beyond the scope of the present study. 

One important candidate is the thyroid hormone, since its importance in cone development in the mouse is obvious, ^{21–25,27,43–47} and it may be important in the rat as well. However, the culture media used in all previous studies were either serum-supplemented ³¹ or contained thyroid hormone in the same or even higher concentrations than was applied in our system. ^{24,28–30,48} Moreover, the treatment of embryonic mouse retina cultures with increasing concentrations of thyroid hormone failed to activate M-opsin in a recent report by Roberts et al. ²⁴ Thus, it seems unlikely that this hormone alone can explain the differences; and this result emphasizes that attention must be drawn to other still unidentified extrinsic factors that are present in our system. 

The use of different animal models raises the question of whether the absence of M-cones in mouse organotypic cultures can be explained by the interspecies differences in retinal development. In the mouse, even in adults, most cones coexpress both S- and M-opsins, albeit in different ratios in different retinal locations. M-opsin dominates in dorsal cones, whereas S-opsin expression is more prominent in ventral retinal regions. ^49–51 This peculiar division postulates a complex regulatory process, which may be significantly different from that of the rat. However, the absence of M-cone development in a rat culture approach reported previously ³¹ supports our hypothesis that the crucial difference is the composition of the culture medium and the absence of important regulatory factors in other approaches. 

At present, a systematic testing of several candidates is in progress and, if successful, will result in the identification of one or several factors crucial for proper cone development and differentiation. Moreover, culturing of different species with different cone distributions ^1,2 or cone-rich retinas ^34,52 may help elucidate the question of whether M-cone differentiation requires different conditions in different species. 

Footnotes

 Supported by GK 794 and Charlotte and Tistou Kerstan Foundation (BA-G); EU-Marie Curie Training Site “Vision” QLGA-1999-50423 and Hungarian Scientific Research Fund OTKA K73000 (AS); Pro Retina Foundation, Germany (GP-D); and Hungarian Scientific Research Fund OTKA F61717 and OTKA K73000 (ÁL).

Footnotes

 Disclosure: B. Arango-Gonzalez, None; A. Szabó, None; G. Pinzon-Duarte, None; Á. Lukáts, None; E. Guenther, None; K. Kohler, None

The authors thank Ágoston Szél (Semmelweis University, Budapest, Hungary) for the monoclonal COS-1 antibody and for a valuable critique of the manuscript, Jeremy Nathans (Johns Hopkins University School of Medicine, Baltimore, MD) for the kind gift of JH455 and JH492 rabbit polyclonal antibodies, and Diane Blaurock for assistance in preparation of the manuscript.