|Molecular Vision 2003;
Received 24 October 2002 | Accepted 8 February 2003 | Published 15 February 2003
Development of the cone photoreceptor mosaic in the mouse retina revealed by fluorescent cones in transgenic mice
Department of Ophthalmology and Visual Science, Yale University School of Medicine, New Haven, CT
Correspondence to: Yijian Fei, MD, Department of Internal Medicine, Yale University School of Medicine, LMP 2073, PO Box 208029, New Haven, CT, 06520; Phone: (203) 785-6044; FAX: (203) 785-7068; email: firstname.lastname@example.org
Purpose: Normal function of the retina relies on the orderly stereotyped organization of different neurons and their synaptic connections. How such neural organization is patterned during development remains poorly understood due to the paucity of adequate developmental markers. This study was to examine the spatial organization and development of cone photoreceptors quantitatively in the mouse retina.
Methods: A transgenic approach was used to generate a living cone cell marker by driving GFP expression in mouse cones with the human red/green opsin gene 5' sequences. The spatial organization and development of the cones in the mouse retinas were examined quantitatively with epifluorescence and scanning laser confocal microscopy. Cone specific GFP expression in the developing retinas was verified with peanut agglutinin (PNA) staining. Developmental expression of mouse cone opsin genes was determined with RT-PCR.
Results: The fluorescent retinal cells expressing GFP can be visualized as early as on embryonic day E15. Following up morphological differentiation of these cells revealed features that were consistent with the typical morphology of the mouse cones. Double labeling with cone specific PNA showed that these cells were co-labeled starting from postnatal day P1, and that a subpopulation of PNA positive cones expressed the GFP. The fluorescent cell densities had a similar ventral and dorsal distribution from E15 to P2, increased dramatically in the ventral by P6, and in the dorsal from P7. Nearest neighbor distance analysis demonstrated that this subpopulation of cones was organized into a regular mosaic pattern with a regularity index of 4.82 in the central and 3.55 in the peripheral retina. Quantitative pattern assessment of the developing cones revealed that the fluorescent cells appeared to be distributed in a non-random array before birth. The regularity of the cone array began to rise on P7, in parallel with the onset of mouse green opsin gene expression and the development of cone pedicles. The regular pattern of cone mosaic organization was basically formed by P10, coinciding with the timing of the cone pedicle maturation.
Conclusions: The cones in the mouse retina are organized in a regular mosaic pattern. Patterning the cone mosaic appears to follow a two phase developmental process involving regulated opsin gene expression and cone pedicle maturation: an early phase where a non-random array emerges during cone differentiation, and a late phase where the regular mosaic pattern is mature at the time when cone synaptic contacts are being formed.
The neural retina in vertebrates exhibits a remarkable laminar architecture of alternate cellular and synaptic layers, and a striking pattern of mosaic organization where neurons of the same type lying in each corresponding cellular layer are distributed in a non-random array intermingled with other neuronal cells [1-4]. The mosaic organization of photoreceptor cells in most vertebrate retinas has been used as a model system for the study of neural patterning and organization [4-7], and it is believed to be a developmental strategy of the central nervous system for facilitating connections of the visual neural circuitry and efficient sampling and processing of visual signals [1,8,9]. Although the cone photoreceptor mosaic has been examined in a variety of species [1,2,6,10-14], few studies have described the development of the mosaic pattern quantitatively, and little is known about the mouse due to the lack of adequate developmental markers. A quantitative study of the spatial organization and mosaic development of cone photoreceptors in the mouse retina is particularly important because there is an increasing number of spontaneous and induced mouse mutants that offer unique views of the retinal structure, function and disease [15-22]. Understanding the cellular organization of the mouse retina and the normal developmental process of the mouse cone mosaic will be critical for studying the pathological processes in mutant mouse models of human retinal degeneration.
Investigating the development of mouse cone mosaic requires early appearing, cell specific markers to identify the developing cone cells. However, currently used markers for the mouse cones [23-26] appear to be inadequate for this task. In a previous study, transgenic mice with the living cone cells labeled by a transgenic green fluorescent protein (GFP) marker were created . This transgenic cone marker was capable of selectively labeling the entire cone cell beginning from the early developmental stage. The goal of the present study was using the transgenic mice to study the development and patterning of cone photoreceptors in the mouse retina.
Generation and use of the transgenic mice
Procedures for creating the transgenic mice with GFP expression in the cones were described elsewhere . The transgene was composed of a 6.8 kb regulatory sequence 5' of the human red and green opsin genes containing the promoter and locus control region isolated from the pR6.5 lacZ plasmid  (a generous gift from Dr. Jeremy Nathans, Johns Hopkins University,) and the GFP reporter cassette. Transgenic mice were identified by polymerase chain reaction (PCR) analysis of mouse tail DNA. Mice used in this study did not carry the rd allele from the background SJL strain. Prenatal mice were from timed pregnancies of the transgenic mice. The morning when the copulation plug was observed was taken as day 0. Mice used in this study were cared for and handled in accordance with the Yale Animal Care and Use Committee guidelines and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
Preparation of retinal wholemounts and sections
Mice used in this study were killed by inhalation of overdose of anesthetic, Isofluorane (Fort Dodge Animal Health, Fort Dodge, Iowa) or Methoxyflurane (Pitman-Moore Inc., Mundebin, IL). The orientation of mouse eyes was marked through lightly burning the cornea at 12 o'clock position with a fine tipped surgical cautery before the eyes were enucleated. The neural retinas were dissected free in ice cold Hanks' buffer (Ca2+ and Mg2+ free) under a dissection microscope. For examination of the live cone cells, some retinas were freshly dissected from the enucleated eyes without fixation. For better preservation of the retinal architecture, other retinas were fixed before dissection by immersing the enucleated eyes in ice cold 4% paraformaldehyde in phosphate buffered saline (PBS, pH 7.4). The time of fixation was 30-60 min for the prenatal mouse eyes, and overnight for the postnatal eyes. The fixed mouse eyes were rinsed with PBS 3 times, 30-40 min each. The dissected neural retinas were flat mounted on glass slides with the photoreceptor side up, coverslipped with either PBS (for live retinas) or 90% glycerol in PBS (for fixed retinas). For sections, the fixed retinas from adult transgenic mice were embedded in 5% agarose. Cross sections (50-200 μm) were cut with a vibratome (Electon Microscopy Sciences, Fort Washington, PA). The nuclei in some retinal sections were counterstained with propidium iodide (2 μg/ml) for about 1 h, and then rinsed in PBS for 30-60 min.
Double labeling of retinal wholemounts with peanut agglutinin (PNA)
To determine the specificity of GFP labeling of the cones during development, cone cell specific PNA  was used to stain the retinal wholemounts. The fixed retinas were incubated in PBS containing 10% normal goat serum and 0.2% Triton X-100 for 1-3 h, rinsed with PBS, and reacted with 1:10 diluted rhodamine-conjugated PNA (Vector Laboratories, Burlingame, CA) overnight. PNA stained retinas were rinsed with PBS, and mounted on glass slides with 80% glycerol in PBS containing 0.4% phenylendiamine.
Epifluorescence and confocal microscopy
To detect the GFP signal from mouse cone cells expressing the GFP transgene, retinal preparations were first examined with a Zeiss microscope equipped with a Micromax CCD camera (Princeton Instruments, Trenton, NJ) and standard HQ FITC filter sets (Chroma, Brattleboro, VT). Some images of the retinal wholemounts were also taken with differential interference contrast (DIC) optics to view the overall profiles of both the fluorescent and non-fluorescent retinal cells. A Biorad MRC 600 scanning laser confocal microscope (Hercules, CA) fitted with standard FITC and rhodamine filter sets was used to collect 3 dimensional images from the retinal preparations. The digital images were analyzed with IPLabs software (Scanalytics Inc., Fairfax, VA). Montages were assembled from overlapping images taken from the retinal wholemounts based on common reference points. IMARIS software (Bitplane AG, Zürich, Switzerland) was used to process the confocal 3D data sets on a SGI silicon graphics octane 2 work station (Silicon Graphics, Inc., Mountain View, CA) for 3D view of the cone cell morphology. Image manipulations involved adjusting contrast, adding scale bars and labels. None of the values in the image files were manipulated. The final figures were composed with Adobe Photoshop 5.5.
Sampling and quantitative analysis of the cone mosaic
Transgenic mice used for quantitative analysis of cone mosaic were chosen from line 5933, since this line had the highest level of GFP expression in the mouse cones . Three retinas of three mice were analyzed at each age (E19, P2, P6, P7, P9, P10 and adult). To examine the topographic distribution of the fluorescent cones during development, the numbers of the fluorescent cones were counted in contiguous sampling windows (300 μm x 132 μm/window for adult and 100 μm x 132 μm/window for developing mice) from the dorsal periphery through the optic disc to the ventral periphery. The cone density data were analyzed with IGOR Pro software (WaveMetrics, Lake Oswego, OR). Spatial density profiles of the fluorescent cones were plotted as the average cone density (cones/mm2) against the retinal eccentricity (mm). The nearest neighbor distance analysis [1,7,28-32] was used to quantify the regularity of the cone mosaic pattern. Cone nearest neighbor distances were measured from center to center of the nearest cone cell bodies using IPLabs software. Fields for this analysis were chosen from the central area of the dorsal retina, where the fluorescent cones had the highest density similar to that in non-transgenic mouse  and GFP expressing cones approached 100% of the total cones revealed by counterstaining with PNA . The far periphery of the dorsal retina where the cell density is quite low was also sampled in adult retinas to examine whether the regularity of cone mosaic pattern in the central retina would be different from that in the peripheral retina. The distribution of the nearest neighbor distances was analyzed with IGOR Pro software (WaveMetrics, Lake Oswego, OR) and fitted to the normal Gaussian distribution. The statistical difference between the sample distribution and the normal distribution was determined with the χ2 test . The regularity index (RI) or conformity ratio , was calculated as the ratio of the mean nearest neighbor distance to the standard deviation, which provides a good test for mosaic regularity [1,28,29,32].
RNA preparation and reverse transcription-PCR
Total RNA was isolated from mouse retinas from E13 to P10 with TRIzol reagent (Life Technologies, Gaithersburg, MD) following the manufacturer's instructions. Aliquots of the RNA samples with or without DNase I (RNase free) treatment were used for RT-PCR. The mouse retinal mRNA was reverse transcripted into cDNA using the ThermoScript RT-PCR system (Life Technologies, Gaithersburg, MD) with the Oligo [dT]20 primer, and amplified by PCR according to the manufacturer's protocols. PCR primers used to amplify the mouse blue and green cone opsin (short and middle wavelength sensitive opsin, respectively) gene cDNA [34,35] and the PCR cycling profiles were described in the previous study .
Cone specific, consistent pattern of GFP transgene expression
It was previously shown that the GFP expression reliably marked the cones in the adult mouse retina . Examination of retinal wholemounts and sections further demonstrated that the GFP labeling was restricted to the photoreceptors and that no other retinal cells expressed detectable levels of GFP (Figure 1). These fluorescent photoreceptor cells had the majority of their somata located in the outer border of the outer nuclear layer (ONL) and their pedicles lying in the inner part of the outer plexiform layer (OPL; Figure 1C). A 3D image of one of the fluorescent cells from a flat mounted retina revealed a conical shaped outer segment, a thick inner segment and a large flattened synaptic pedicle connected to the cell body via an axon (Figure 2A). All these features are consistent with the morphology of a cone cell in the mouse retina . Double labeling with cone specific PNA marker showed co-localization of the staining in the outer segment (Figure 2B), which further confirms that the fluorescent photoreceptor cells are indeed the cones. All three transgenic lines generated had a similar dorsal-ventral graded pattern of GFP expression despite of a great variation in the number of cones expressing GFP. This characteristic expression pattern has been faithfully transmitted from early generation (F1) to later ones (F4). Figure 3 shows the representative patterns of fluorescent cones labeled by GFP in the F1 (Figure 3A,B) and F4 (Figure 3C,D) mouse retinas of the same 5933 line. Both mice exhibited very similar spatial patterns of the fluorescent cone distributions in the dorsal (A and C) and the ventral (B and D) retinas. The mosaic organization of cones in the mouse retina is readily appreciated in Figure 3, where the fluorescent cones appeared to be somewhat evenly spaced.
Early emergence and morphological differentiation of the fluorescent cells
To determine when the fluorescent cells could be detected during retinal development, retinal wholemounts from the transgenic lines at each embryonic day starting from E13 when cone genesis begins  were examined. The first fluorescent cells that could be reliably detected in the mouse retinas from line 5933 appeared at embryonic day E15 in both live (Figure 4) and fixed retinas. These fluorescent cells were situated in the outer surface of the neuroblast layer of the embryonic retina, and most of them had a sclerally oriented process with a bulbous enlargement at the end. Similar fluorescent cells were also observed in the E15 retinas of another mouse line 5922 (Figure 5), where the fluorescent cell had a cell body bearing a short process oriented sclerally with a bulb-like enlargement (Figure 5A, arrow). Optical sectioning through the cell bodies with confocal microscopy revealed that the fluorescent cells had a large, elongated or oval nucleus. The nuclei usually contain 2-3 darker areas that appear to represent the characteristic dense heterochromatin clumps of the cone nuclei in mouse retina , which exclude the GFP fluorescence and produce these dark areas (Figure 5B, arrowheads).
To follow the morphological differentiation of the fluorescent cells, the developing mouse retinas starting from E17 were examined. By E19, the fluorescent cells developed a vitreally oriented inner process. The sclerally directed outer process became much thicker, it first narrowed down away from the elongated cell body and then expanded into a bulbous structure, representing a primitive inner segment (Figure 6A). By P2, the inner segments were evident and the axons became thicker (Figure 6B). Apparent outer segment development appeared consistently in the P4 mouse retina, where a conical structure extended sclerally from the tip of the photoreceptor inner segment (Figure 6C, arrowhead). At this stage, only a few axons had terminal enlargements. The synaptic pedicles with apparent fine basal processes were observed in some cones by P7 (Figure 6D, arrowhead). Thereafter, the fluorescent cells appeared to be maturing with growing outer and inner segments and enlarging pedicles (Figure 6E,F). An adult-like morphology of the cones basically developed by P10. The fluorescent cones were fully mature by P20 (Figure 6G), with an overall morphology comparable to that of the adults.
To confirm that the developing fluorescent cells are cones, the cone specific PNA marker  was used to stain retinal wholemounts of the early postnatal developing mice. PNA staining demonstrated that every fluorescent cell in the P1 (Figure 7A), P4 (Figure 7B) and P10 (Figure 7C) mouse retinas was indeed a cone cell co-labeled by the PNA marker (arrows), despite that some cones with PNA staining did not show detectable GFP (arrowheads). Attempts were made to quantify the numbers of PNA positive cones that express GFP during development. However, accurate counting of the PNA positive cells in the developing retina was problematic, particularly at the early developmental stages, because of the weak signal, less discrete patchy staining, high background due to binding of PNA to the interphotoreceptor matrix, inner synaptic layer and retinal vasculature during development , and the inability of PNA to reveal the cell morphology. These problems restricted the counting of PNA labeled cones during development to P10. At this stage, as cones are almost mature, the patchy staining signal of PNA becomes stronger and more discrete, counting PNA+ cells is less ambiguous. The number of PNA+ cells expressing GFP is about 91-95% in the dorsal and 32-40% in the ventral of the P10 retina.
Spatial distribution of the fluorescent cones in the mouse retina during development
Figure 8 shows the density profiles of the fluorescent cones in developing and mature mouse retinas. These density data were from three retinas of 3 mice at each developmental time point. During embryonic stages, the fluorescent cells had a similar ventral and dorsal distribution with peak densities at eccentricity 0.6-1.1 mm. Similar distribution pattern was observed by postnatal day P2. At this stage, the cone density increased only slightly with peaks at eccentricity 0.9-1.2 mm, although the total number of cones expressing GFP increased about 1.8 fold due to the expansion of the retinal area. From P2 to P6, the number of fluorescent cones rose steeply in the ventral retina with the highest density occurring at eccentricity 1.2 mm ventral to the optic disc and 0.7-1.4 mm dorsal to the optic disc (Figure 8A). A striking feature of the cone density profile in the P6 mouse retina was the asymmetric distribution of fluorescent cones along the ventral-dorsal meridian of the retina, with about 3 times more fluorescent cones in the ventral than in the dorsal. From P6 to P7, the cone density rose only slightly in the ventral retina, but it dramatically increased about 5 fold in the dorsal retina (Figure 8 A). From P7 to P10, the cone density increased rapidly across the entire retina, but predominantly in the dorsal retina (about 14 fold in ventral and 23 fold in dorsal; Figure 8A,B). By P10, the density profile of fluorescent cones was very similar to that of the adult in the ventral retina with peak densities at eccentricity 1.2-1.4 mm, but reached only about 90% of the adult level in the dorsal retina (Figure 8B). At this stage, the estimated total number of fluorescent cones per retina was about 92% of the adult fluorescent cone number. In the adult mouse, the overall fluorescent cone density in the dorsal retina is about 3.4 times that in the ventral retina. The cone densities gradually rose with eccentricity until peak densities (about 10300 cells/mm2 in the dorsal and 3100 cells/mm2 in the ventral) were reached around 1.2 mm eccentricity, and thereafter the densities fell rapidly toward the far peripheral retina.
Quantitative analysis of the spatial organization of the cones in adult mouse retina
The mosaic organization of cones varies widely among vertebrates, even between very closely related species [6,31,38,39]. To quantitatively analyze the spatial pattern of cones in the mouse retina, the nearest neighbor distance (NND) analysis was performed. Figure 9 shows the NND distributions of the fluorescent cones in the adult mouse central (Figure 9A) and far peripheral (Figure 9B) retina. The distributions appeared to be symmetric around the mean and fitted well to the normal Gaussian distribution (p>0.10), although there was a consistent tail towards larger distances. This indicates a regular distribution of the inter-cone distances, and thus a regular spacing of the cone cells. The average center to center distance between cone cells was 6.46 ± 1.34 μm (mean and standard deviation) in the central and 8.97 ± 2.53 μm in the peripheral retina. The regularity index (RI), expressed as the ratio of the mean NND to the standard deviation, was used to quantify the regularity level of the cone spatial patterns; The higher the RI, the more regular the spatial pattern [1,28,29,32]. The regularity index of the cones was 4.82 in the central, and 3.55 in the peripheral retina. These relatively high RI suggest that the fluorescent cones are organized into a non-random regular pattern, and that the mosaic array of cones in the central retina is more regular than that in the far peripheral retina.
Patterning of the cone mosaic during mouse retinal development
To study the developmental patterning of the mouse cone mosaic, retinal wholemounts from mice at ages from E15 to P10 were examined. Three mice at each age were examined and sampled for the NND analysis. From E15 to E18, there were only a few bright fluorescent cells (20-90) per retina, sparsely scattered over both the dorsal and ventral retinal surfaces. On E19, the number of fluorescent cells increased significantly. At this stage, these differentiating fluorescent cells seemed to be somewhat evenly spaced on the retinal surface (Figure 10A). The distribution of nearest neighbor distances of these cells appeared to be symmetric around the mean inter-cell distance (87.86 μm), and closely fitted to a broad Gaussian distribution with a RI of 2.60 (Figure 10, E19). χ2 test of the statistical difference between the normal distribution and the actual NND distribution showed no significant difference (p>0.10). These results indicate that a random distribution is less likely to be the description of the NND data, and that the lower RI might reflect a lower regularity of the pattern. Taken together, these findings suggest that the differentiating cones in the mouse retina appeared to be organized in a non-random array with a lower level of regularity before birth. By P6, the spatial arrangement of the fluorescent cones showed a similar pattern (Figure 10B, P6). On P7, a more regular mosaic array began to emerge with a higher RI of 3.00 and a NND data well fitted to the normal distribution (p>0.10; Figure 10C, P7). On P9, the mosaic pattern seemed to be more regular with an increased RI of 3.71 (Figure 10D, P9). An adult-like cone mosaic pattern is formed by P10, where the NND distribution of the fluorescent cones fitted well to the normal distribution (p>0.10) and the regularity index of 4.56 was very close to that of the adult (Figure 10E, P10).
Development and patterning of the cone pedicles during mosaic formation
To explore whether the formation of the cone mosaic in the mouse retina is associated with cell-cell interactions via cone synaptic connections, the morphological development and patterning of the fluorescent cone pedicles was examined with confocal microscopy. At embryonic stages, the pedicles were not formed. From birth to P6, most of the developing cones had axonal processes with varying sizes of terminal enlargements. By P7, some cones developed pedicles with fine basal processes. These pedicles were sparsely distributed in the outer plexiform layer (OPL) of the retina. Cone pedicles were observed in more fluorescent cones by P9 (Figure 11A, arrows). At this stage, some cones already had adult-like pedicles with basal processes contacting neighbor cones (Figure 11A, arrowheads), but the patterning of the cone pedicles in the OPL was obviously incomplete. By P10, the cone pedicles appeared to be basically mature (Figure 11B, arrows) compared with those of the adults (Figure 11C, arrows), and almost every pedicle developed robust basal processes that contact the neighboring pedicles (Figure 11B, arrowheads). The density of the pedicles and the pattern of the pedicle arrays in the OPL were very similar to those of the adults (Figure 11C).
Temporal expression of the mouse cone opsin genes during development
To test whether the timing of the cone mosaic development in the mouse retina is temporally correlated to the appearance of the endogenous mouse cone specific gene expression, the temporal expression of the mouse blue and green opsin genes was examined at the transcription level. In parallel with transgene expression, the strong message of blue opsin gene can be reproducibly detected at E15 (Figure 12), when the first fluorescent cells were identified in the mouse retina. The transcriptional expression of blue opsin gene persisted from E15 throughout postnatal stages to adulthood in the mouse retina. In contrast, the expression of the green opsin gene was not detectable until postnatal day 7 (Figure 12), when the regular mosaic arrays of cones began to emerge, and then persisted to adulthood. Younger retinas at earlier developmental stages were also examined. Although a very faint band similar to the blue opsin transcript could be observed at E13-14, unambiguous strong signal that can be reproducibly detected did not appear until E15. While strong GFP fluorescence signal was not observed until E15, it is possible that a lower level of GFP expression undetectable with the epifluorescence microscope, particularly at the message level, was present in the E13-14 retina.
The differentiation, morphogenesis and spatial patterning of most retinal neurons take place during early developmental stages. Understanding of these developmental events in cone photoreceptors relies on the development of early appearing, cone specific markers capable of revealing the entire structures of the cone cells. Currently used cone specific antibodies, for example, the anti-cone opsin antibodies [24,25,40], have been a valuable tool for identification of the cones. Unfortunately, these antibodies do not appear to readily stain the early developing cones in the mouse retina until the cone outer segments begin to develop . Further, these antibodies, like PNA, another commonly used cone marker , are unable to label the entire cone cell. While in situ hybridization with cone specific cDNA probes has provided a very useful tool for studying the photoreceptor patterning in vertebrate retinas [2,11,12,40], the diffuse hybridization signals often make it difficult to perform a quantitative analysis on the number and spacing of the cones during mosaic patterning. In addition, in situ hybridization is unable to reveal the morphology of cells. Other markers such as cytochrome oxidase  and neuron specific enolase  can also readily label the early postnatal cones. However, these markers are not cone cell specific. These technical limitations constrain our understanding of the early events of cone development. It is demonstrated in this study that the transgenic GFP marker can reliably label the developing cones in the mouse retina. In contrast to other cone markers such as PNA, the strong signal of GFP provides a full view of the entire cone cell morphology in both living and fixed retinas. This allows a reliable measurement of the center to center inter-cone distances and unambiguous counting of individual cones for quantitative analysis of the cone mosaic. In addition, this transgenic marker also allows the visualization of the developing mouse cones much earlier than other cone markers do. These advantages of the GFP marker make it possible for the first time to quantitatively assess the mosaic organization and development of cones in the mouse retina. Several new observations are made in this study: First, it demonstrates that a subpopulation of cones is organized into a regular mosaic pattern in the mouse retina. Second, it reveals a previously unidentifiable early phase of mouse cone patterning: a nonrandom array of the cones before birth. Third, it shows that the cone mosaic development appears to follow a characteristic timetable associated with mouse opsin gene expression and the maturation of cone synaptic pedicles.
One disadvantage in using this transgenic marker, however, is that not every cone in the mouse retina expresses GFP, particularly in the ventral retina. It is obvious that sampling the GFP labeled cones from the ventral retina would be inadequate for quantitative analysis of the cone mosaic. To overcome this problem, sampling of the fluorescent cones in this study, was chosen from the central area of the dorsal retina, where the PNA positive cones that express GFP reached 100% in the adult, and about 91-95% in the developing (P10) retinas. Thus the quantitative NND analysis of the GFP cones in this region is a reasonable representation of the native cones in the dorsal retina. Because the cones are fairly uniformly distributed in both the dorsal and the ventral regions of the mouse retina [24,33,40], we speculate that the cone mosaic pattern derived from the dorsal retina also applies to the ventral retina. It should be pointed out that those few GFP negative cones, escaped from the NND analysis in the dorsal retina, could increase the NND of a few cones and slightly reduce the regularity of the analyzed patterns by introducing those tails seen in the NND distributions. Therefore, it is possible that the native cone mosaic pattern is slightly underestimated by analysis of the fluorescent cones in this study. It would be interesting to simultaneously analyze the NND of PNA positive cells to quantify the mosaic pattern of the native cones in the mouse retina. Unfortunately, the inability of PNA to readily label the cell bodies of mouse cones makes it impossible to accurately measure the center to center distances of the cell bodies for the quantitative NND analysis.
Although the transgenic lines had different numbers of cones expressing GFP, which is presumably due to variations in transgene copy numbers or position effect variegation, all shared similar dorsal-ventral graded pattern of GFP expression. The parallel temporal expression of the transgene and the endogenous opsin gene, and the similarity of the spatial developmental pattern of GFP labeling to that of the mouse cone opsin staining reported previously [24,40,42,43], indicate that the expression of the transgene and the endogenous genes might be regulated in a similar manner.
Differentiation and morphogenesis of the cones in mouse retina appear to occur in a sequential manner. Most of the early differentiating cones seem to first grow out a sclerally oriented process representing the early inner segment and then a vitreally descending process before birth. The morphology of the fluorescent embryonic cells is consistent with that of the immature photoreceptors in the embryonic mouse retinas observed by Hinds and Hinds  with electron microscopy. The location of these fluorescent cells on the outer surface of the neuroblast layer and the timing of their appearance during the cone genesis, together with the cone specific pattern of the transgene expression, further suggest that these fluorescent cells are more likely the early differentiating cones. Unfortunately, there are hardly any markers available to label the mouse cones before birth. Nevertheless, following up the morphological development of the fluorescent cells in early postnatal retinas revealed that these cells had the typical morphology of mouse cones described in previous studies [23,25,26,42,45-47]. Double labeling with PNA in postnatal mouse retinas starting from P1 confirmed that these developing fluorescent cells are indeed the cones.
The mosaic organization of cones has been documented in a number of species ranging from fish to primates including humans [1,2,6,10,12-14]. However, the nature of the cone mosaic appears to be species dependent, and varies from highly regular to irregular . There is a lack of quantitative study on the spatial organization of cones in the mouse retina. In this study, quantitative pattern analysis revealed that the cones in mouse retina are organized in a regular mosaic pattern. While the overall cone mosaic in the mouse retina seems to be less precise than that in monkey and cat , this less precision could be due to the underestimation of the cone mosaic in this study. Nevertheless, the findings in this study suggest that the spatial organization of cones in the mouse retina follows a general organization principle shared by cones and other retinal neurons in many other vertebrate species, in spite of the fact that mouse is unusual in having only a single type of cone  and does not appear to have color vision . In addition, knowledge of this normal pattern of cone organization in the mouse retina may be useful for examining potential pathological alterations in the organization of the cone photoreceptors in retinal diseases of mutant mice.
Few studies have described the development of retinal mosaic quantitatively. This study provides a quantitative assessment of the cone mosaic development in the mouse retina. It appears that the patterning of the cone mosaic in mouse retina is largely late embryonic and early postnatal developmental events with a characteristic timetable. Interestingly, the distribution of the fluorescent cones appears to exhibit a non-random pattern before the mouse is born, a previously unidentifiable early phase of cone patterning. During postnatal development, the peak fluorescent cone densities are clustered around the ventral and dorsal centers. There seems to be two distinct patterns of rapid development of the fluorescent cones around P6-P7: one occurs in the ventral retina by P6, and the other appears in the dorsal retina starting from P7. In parallel with the onset of mouse green opsin gene expression and appearance of the immature cone pedicles, the mosaic pattern with higher regularity begins to emerge by P7. Coinciding with the timing of cone pedicle maturation, the regular pattern of cone mosaic is basically formed by P10.
It remains unknown in general how a retinal mosaic is developed and what mechanisms control the precision of regular spacing of the mosaic cells. The creation of the mosaic organization is believed to be a multi-step process of spatiotemporal neural patterning involving a number of coordinated molecular and cellular events that include cell genesis, cell fate determination, activation of the transcriptional network for retinal patterning genes, expression of cell specific markers, cell migration and differentiation [4,5,7,30,48]. The results of this study appear to be consistent with the idea of a two phase process of cone mosaic development in the mouse retina. The early phase involves the development of a non-random array of cones that occurs very early in development. The late phase is characterized by a gradually increased regularity of the cone arrays that reaches the adult pattern when the cone pedicles are maturing. The parallel expression of the transgene and the blue opsin gene at the stage of cone genesis, and the appearance of a orderly cone array in the early phase of development, suggest that the mosaic organization of mouse cones might be pre-patterned during the embryonic stages. Similar concept was proposed for mosaic patterning of cones in the monkey retina [49,50]. Since cell migration [30,51], programmed cell death  and differential retinal expansion  occurring with the developing retina could change the initial pattern of cone arrays, the final mosaic pattern of mouse cones might require a refining process through certain mechanisms to ensure a regular spacing of cones in the mature mosaic pattern. In this sense, the late phase of mouse cone mosaic development might reflect a refining process of the cone mosaic pattern. Alternatively, the apparent increase in the mosaic regularity in this late phase might be due to the increasing numbers of fluorescent cones at this developmental phase. Nevertheless, the remarkable coincidence between the maturation of the mosaic and development of the cone pedicles, suggests that synaptic contact mediated cone-cone interactions might be involved. Such interactions may feedback to finely tune the cell positioning through short distance lateral cell movement. If it is possible to image the retinas overtime during the late phase of development, and maintain the integrity of the live tissue, the GFP reporter in these transgenic mice could potentially be valuable in addressing these issues of cone photoreceptor development in vivo.
This work was supported by NEI grant EY08362 and The Matilda Zeigler Foundation awarded to Dr. Thomas Hughes. I thank Dr. Hughes for his generous support, encouragement and critical reading of the initial drafts of this manuscript; Drs. Jeremy Nathans and Yanshu Wang for kindly sharing the pR6.5 lacZ plasmid; Drs. James Howe and Douglas Gregory for helpful discussions and reading of the first draft of this manuscript.
1. Wassle H, Riemann HJ. The mosaic of nerve cells in the mammalian retina. Proc R Soc Lond B Biol Sci 1978; 200:441-61.
2. Raymond PA, Barthel LK, Curran GA. Developmental patterning of rod and cone photoreceptors in embryonic zebrafish. J Comp Neurol 1995; 359:537-50.
3. Galli-Resta L. Patterning the vertebrate retina: the early appearance of retinal mosaics. Semin Cell Dev Biol 1998; 9:279-84.
4. Cook JE, Chalupa LM. Retinal mosaics: new insights into an old concept. Trends Neurosci 2000; 23:26-34.
5. Cepko CL. The patterning and onset of opsin expression in vertebrate retinae. Curr Opin Neurobiol 1996; 6:542-6.
6. Galli-Resta L, Novelli E, Kryger Z, Jacobs GH, Reese BE. Modelling the mosaic organization of rod and cone photoreceptors with a minimal-spacing rule. Eur J Neurosci 1999; 11:1461-9.
7. Stenkamp DL, Cameron DA. Cellular pattern formation in the retina: retinal regeneration as a model system. Mol Vis 2002; 8:280-93 <http://www.molvis.org/molvis/v8/a35/>.
8. Ahnelt PK. The photoreceptor mosaic. Eye 1998; 12:531-40.
9. Jeyarasasingam G, Snider CJ, Ratto GM, Chalupa LM. Activity-regulated cell death contributes to the formation of ON and OFF alpha ganglion cell mosaics. J Comp Neurol 1998; 394:335-43.
10. Muller B, Peichl L. Topography of cones and rods in the tree shrew retina. J Comp Neurol 1989; 282:581-94.
11. Bruhn SL, Cepko CL. Development of the pattern of photoreceptors in the chick retina. J Neurosci 1996; 16:1430-9.
12. Bumsted K, Jasoni C, Szel A, Hendrickson A. Spatial and temporal expression of cone opsins during monkey retinal development. J Comp Neurol 1997; 378:117-34.
13. Wikler KC, Rakic P, Bhattacharyya N, Macleish PR. Early emergence of photoreceptor mosaicism in the primate retina revealed by a novel cone-specific monoclonal antibody. J Comp Neurol 1997; 377:500-8.
14. Roorda A, Williams DR. The arrangement of the three cone classes in the living human eye. Nature 1999; 397:520-2.
15. Olsson JE, Gordon JW, Pawlyk BS, Roof D, Hayes A, Molday RS, Mukai S, Cowley GS, Berson EL, Dryja TP. Transgenic mice with a rhodopsin mutation (Pro23His): a mouse model of autosomal dominant retinitis pigmentosa. Neuron 1992; 9:815-30.
16. Chen J, Makino CL, Peachey NS, Baylor DA, Simon MI. Mechanisms of rhodopsin inactivation in vivo as revealed by a COOH-terminal truncation mutant. Science 1995; 267:374-7.
17. Lem J, Makino CL. Phototransduction in transgenic mice. Curr Opin Neurobiol 1996; 6:453-8.
18. Tsang SH, Gouras P, Yamashita CK, Kjeldbye H, Fisher J, Farber DB, Goff SP. Retinal degeneration in mice lacking the gamma subunit of the rod cGMP phosphodiesterase. Science 1996; 272:1026-9.
19. Fei Y, Ripps H, Peachey N, Matragoon S, Wei S, Wang Y, Liu C, Gottesman ME, Blaner WS, Liou GI. Targeted disruption of the interphotoreceptor retinoid-binding protein (IRBP) locus. Invest Ophthalmol Vis Sci 1997; 38:S699.
20. Furukawa T, Morrow EM, Li T, Davis FC, Cepko CL. Retinopathy and attenuated circadian entrainment in Crx-deficient mice. Nat Genet 1999; 23:466-70.
21. Jacobs GH, Fenwick JC, Calderone JB, Deeb SS. Human cone pigment expressed in transgenic mice yields altered vision. J Neurosci 1999; 19:3258-65.
22. Haverkamp S, Wassle H. Immunocytochemical analysis of the mouse retina. J Comp Neurol 2000; 424:1-23.
23. Blanks JC, Johnson LV. Selective lectin binding of the developing mouse retina. J Comp Neurol 1983; 221:31-41.
24. Wang Y, Macke JP, Merbs SL, Zack DJ, Klaunberg B, Bennett J, Gearhart J, Nathans J. A locus control region adjacent to the human red and green visual pigment genes. Neuron 1992; 9:429-40.
25. Szel A, Rohlich P, Caffe AR, Juliusson B, Aguirre G, Van Veen T. Unique topographic separation of two spectral classes of cones in the mouse retina. J Comp Neurol 1992; 325:327-42.
26. Rich KA, Zhan Y, Blanks JC. Migration and synaptogenesis of cone photoreceptors in the developing mouse retina. J Comp Neurol 1997; 388:47-63.
27. Fei Y, Hughes TE. Transgenic expression of the jellyfish green fluorescent protein in the cone photoreceptors of the mouse. Vis Neurosci 2001; 18:615-23.
28. Hutsler JJ, Chalupa LM. Neuropeptide Y immunoreactivity identifies a regularly arrayed group of amacrine cells within the cat retina. J Comp Neurol 1994; 346:481-9.
29. Cook JE. Spatial properties of retinal mosaics: an empirical evaluation of some existing measures. Vis Neurosci 1996; 13:15-30.
30. Galli-Resta L, Resta G, Tan SS, Reese BE. Mosaics of islet-1-expressing amacrine cells assembled by short-range cellular interactions. J Neurosci 1997; 17:7831-8.
31. Ahnelt PK, Fernandez E, Martinez O, Bolea JA, Kubber-Heiss A. Irregular S-cone mosaics in felid retinas. Spatial interaction with axonless horizontal cells, revealed by cross correlation. J Opt Soc Am A Opt Image Sci Vis 2000; 17:580-8.
32. Wassle H, Dacey DM, Haun T, Haverkamp S, Grunert U, Boycott BB. The mosaic of horizontal cells in the macaque monkey retina: with a comment on biplexiform ganglion cells. Vis Neurosci 2000; 17:591-608.
33. Jeon CJ, Strettoi E, Masland RH. The major cell populations of the mouse retina. J Neurosci 1998; 18:8936-46.
34. Chiu MI, Zack DJ, Wang Y, Nathans J. Murine and bovine blue cone pigment genes: cloning and characterization of two new members of the S family of visual pigments. Genomics 1994; 21:440-3.
35. Sun H, Macke JP, Nathans J. Mechanisms of spectral tuning in the mouse green cone pigment. Proc Natl Acad Sci U S A 1997; 94:8860-5.
36. Carter-Dawson LD, LaVail MM. Rods and cones in the mouse retina. I. Structural analysis using light and electron microscopy. J Comp Neurol 1979; 188:245-62.
37. Carter-Dawson LD, LaVail MM. Rods and cones in the mouse retina. II. Autoradiographic analysis of cell generation using tritiated thymidine. J Comp Neurol 1979; 188:263-72.
38. Kouyama N, Marshak DW. The topographical relationship between two neuronal mosaics in the short wavelength-sensitive system of the primate retina. Vis Neurosci 1997; 14:159-67.
39. Martin PR, Grunert U, Chan TL, Bumsted K. Spatial order in short-wavelength-sensitive cone photoreceptors: a comparative study of the primate retina. J Opt Soc Am A Opt Image Sci Vis 2000; 17:557-67.
40. Applebury ML, Antoch MP, Baxter LC, Chun LL, Falk JD, Farhangfar F, Kage K, Krzystolik MG, Lyass LA, Robbins JT. The murine cone photoreceptor: a single cone type expresses both S and M opsins with retinal spatial patterning. Neuron 2000; 27:513-23.
41. Kageyama GH, Wong-Riley MT. The histochemical localization of cytochrome oxidase in the retina and lateral geniculate nucleus of the ferret, cat, and monkey, with particular reference to retinal mosaics and ON/OFF-center visual channels. J Neurosci 1984; 4:2445-59.
42. Szel A, Rohlich P, Mieziewska K, Aguirre G, van Veen T. Spatial and temporal differences between the expression of short- and middle-wave sensitive cone pigments in the mouse retina: a developmental study. J Comp Neurol 1993; 331:564-77.
43. Shaaban SA, Crognale MA, Calderone JB, Huang J, Jacobs GH, Deeb SS. Transgenic mice expressing a functional human photopigment. Invest Ophthalmol Vis Sci 1998; 39:1036-43.
44. Hinds JW, Hinds PL. Differentiation of photoreceptors and horizontal cells in the embryonic mouse retina: an electron microscopic, serial section analysis. J Comp Neurol 1979; 187:495-511.
45. Olney JW. An electron microscopic study of synapse formation, receptor outer segment development, and other aspects of developing mouse retina. Invest Ophthalmol 1968; 7:250-68.
46. Blanks JC, Adinolfi AM, Lolley RN. Synaptogenesis in the photoreceptor terminal of the mouse retina. J Comp Neurol 1974; 156:81-93.
47. Jansen HG, Sanyal S. Development and degeneration of retina in rds mutant mice: electron microscopy. J Comp Neurol 1984; 224:71-84.
48. Adler R. A model of retinal cell differentiation in the chick embryo. Prog Retinal Eye Res 2000; 19:529-57.
49. Wikler KC, Rakic P. Relation of an array of early-differentiating cones to the photoreceptor mosaic in the primate retina. Nature 1991; 351:397-400.
50. Wikler KC, Rakic P. An array of early differentiating cones precedes the emergence of the photoreceptor mosaic in the fetal monkey retina. Proc Natl Acad Sci U S A 1994; 91:6534-8.
51. Reese BE, Necessary BD, Tam PP, Faulkner-Jones B, Tan SS. Clonal expansion and cell dispersion in the developing mouse retina. Eur J Neurosci 1999; 11:2965-78.
52. Young RW. Cell death during differentiation of the retina in the mouse. J Comp Neurol 1984; 229:362-73.
53. Reichenbach A, Schnitzer J, Reichelt E, Osborne NN, Fritzsche B, Puls A, Richter U, Friedrich A, Knothe AK, Schober W, et al. Development of the rabbit retina, III: Differential retinal growth, and density of projection neurons and interneurons. Vis Neurosci 1993; 10:479-98.