Pattern formation of the cone mosaic in the zebrafish retina: a cell rearrangement model

In fish retinas, cone photoreceptor cells are arranged in two-dimensional regular patterns, called cone mosaics. In the zebrafish retina, four subtypes of cone cells, which are maximally sensitive to different wavelengths of light, appear in quasi-periodic patterns. The pattern formation mechanism is unknown. Here, I develop a mathematical model to examine whether cell adhesion can explain the formation of the zebrafish mosaic. I assume that the movement of differentiated cells is responsible for generating the pattern, and that the movement rate is modified by cell adhesion. The pattern is formed if the magnitudes of cell adhesion between cell types are chosen appropriately. I determine the conditions of cell adhesion for generating the pattern. I also compare this cell rearrangement model with a previously studied model in which the pattern is formed by transitions of cell fate. The condition for obtaining the focal pattern is looser in the cell rearrangement model than in the fate transition model.     

Defect patterns on the curved surface of fish retinae suggest a mechanism of cone mosaic formation

The outer epithelial layer of zebrafish retinae contains a crystalline array of cone photoreceptors, called the cone mosaic. As this mosaic grows by mitotic addition of new photoreceptors at the rim of the hemispheric retina, topological defects, called “Y-Junctions”, form to maintain approximately constant cell spacing. The generation of topological defects due to growth on a curved surface is a distinct feature of the cone mosaic not seen in other well-studied biological patterns like the R8 photoreceptor array in the Drosophila compound eye. Since defects can provide insight into cell-cell interactions responsible for pattern formation, here we characterize the arrangement of cones in individual Y-Junction cores as well as the spatial distribution of Y-junctions across entire retinae. We find that for individual Y-junctions, the distribution of cones near the core corresponds closely to structures observed in physical crystals. In addition, Y-Junctions are organized into lines, called grain boundaries, from the retinal center to the periphery. In physical crystals, regardless of the initial distribution of defects, defects can coalesce into grain boundaries via the mobility of individual particles. By imaging in live fish, we demonstrate that grain boundaries in the cone mosaic instead appear during initial mosaic formation, without requiring defect motion. Motivated by this observation, we show that a computational model of repulsive cell-cell interactions generates a mosaic with grain boundaries. In contrast to paradigmatic models of fate specification in mostly motionless cell packings, this finding emphasizes the role of cell motion, guided by cell-cell interactions during differentiation, in forming biological crystals. Such a route to the formation of regular patterns may be especially valuable in situations, like growth on a curved surface, where the resulting long-ranged, elastic, effective interactions between defects can help to group them into grain boundaries.     

Early onset of phenotype and cell patterning in the embryonic zebrafish retina

The regular arrangement of retinal cone cells in a mosaic pattern is a common feature of teleosts. In the zebrafish, Brachydanio rerio, the retinal cone mosaic comprises parallel rows consisting of a repeating motif of four cone types. In order to elucidate the temporal and spatial aspects of the genesis of the cone mosaic in the developing retina, we generated a monoclonal antibody that specifically binds to the double cone photoreceptor of the adult. We first saw staining in the developing retina with this antibody, FRet 43, at 48 hours postfertilization, the time at which the first photoreceptor cells undergo their final mitotic division. We then injected embryonic fish with the thymidine analog, 5-bromo-2'-deoxyuridine (BrdU), confirming with a double-labeling experiment that the onset of FRet 43 antigenicity occurs within three hours of the cellular division that generates the double cone photoreceptors. Then we stained tangential sections of the 54-hour embryonic retina with FRet 43, further showing that cells devoid of staining alternate with stained pairs of cells in a pattern that is consistent with the arrangement of photoreceptors in the adult cone mosaic. These results indicate that a marker of the double cone phenotype is expressed at approximately the same time as cellular birthday and that the mosaic patterning is present within 6 hours of this expression.     

Difference in the retinal cone mosaic pattern between zebrafish and medaka: cell-rearrangement model

In fish retina, four kinds of photoreceptor cells (or cones) are two-dimensionally arranged in a very regular manner, forming cone mosaics. Mosaic pattern differs between species--two typical patterns are "row mosaic" and "square mosaic", exemplified by the cone mosaics in zebrafish and in medaka, respectively. In this paper, we study a cell-rearrangement model. Cells with pre-fixed fate exchange their locations between nearest neighbors and form regular mosaic patterns spontaneously, if the adhesive force between nearest neighbors and between next-nearest neighbors depend on their cell types in an appropriate manner. The same model can produce both row and square mosaic patterns. However, if the cell-cell interaction is restricted to nearest neighbors only, the square mosaic (medaka pattern) cannot be generated, showing the importance of interaction between next-nearest neighbors. In determining whether row mosaic (zebrafish pattern) or square mosaic (medaka pattern) is to be formed, two shape factors are very important, which control the way adhesions in different geometric relations are combined. We also developed theoretical analysis of the parameter ranges for the row mosaic and the square mosaic to have higher total adhesion than alternative spatial patterns.     

Development of a rod photoreceptor mosaic revealed in transgenic zebrafish

The number and distribution of neurons within the vertebrate retina are tightly regulated. This is particularly apparent in the highly ordered, crystalline-like arrangement of the cone photoreceptors in the teleost. In this report, using a transgenic line of zebrafish, a novel and developmentally regulated mosaic pattern of the rod photoreceptors is described. The spatial and temporal expression of EGFP, under the control of the Xenopus rhodopsin gene promoter, was nearly identical to the endogenous rhodopsin. EGFP was first detected in the ventral nasal retinal in an area of precocious neurogenesis referred to as the "ventral patch". Subsequent expression of EGFP was observed in isolated cells sporadically distributed across the dorsal and central retina. However, confocal microscopy and spatial analysis of larval eyes or retinal explants from adults revealed a precise arrangement to the rod photoreceptors. The rod terminals were arranged in regularly spaced rows with clearly identifiable telodendria linking neighboring cells. The rod inner segments projected through the cone mosaic in a predictable pattern. In the adult, the rod mosaic originated near the retinal margin where clusters of rods differentiated around the immature short single cone. In the embryo, the sporadic differentiation of the rods led to the gradual formation of the mosaic pattern. With the growing interest in neuronal stem cells, revisiting this model of neurogenesis provides an avenue to uncover mechanisms underlying the precise integration of new neuronal elements into a preexisting neural network.     

Formation of cone mosaic of zebrafish retina.

In the zebrafish retina, four types of cone photoreceptor cells (or cones) with different sensitive frequencies are arranged in a regular pattern, named "cone mosaic". A pair of small cones, one sensitive to red and the other sensitive to green, is in close contact and forms a "double cone". In addition, there are two kinds of single cones, sensitive to blue and to UV, respectively. We study characteristics of cell-differentiation rules that realize stable formation of cone mosaic. Assumptions are: undifferentiated cells are arranged in a regular square lattice, and they are one of the three types (B, U, and D cells). A D cell has two parts (G and R-parts) and takes one of the four directions. The cells change their cell type and orientation following a continuous-time Markovian chain. The state transtion occurs faster if it increases the stabilities of the focal cell, in which the stability is the sum of affinities with neighboring cells. After the transient period, the system may reach a stable pattern (pre-pattern). The pattern becomes fixed later when the cells are fully differentiated in which B cells, U cells, and D cells become blue-sensitive, UV-sensitive, and double cones, respectively. We search for the combinations of affinities between cell states that can generate the same cone mosaic patterns as in zerbrafish retina. Successful transition rules give (1) zero or small affinity with the pairs of cell states that are absent in the zebrafish cone mosaic (lambda(UR), lambda(BG)and the contact of two cells of the same type); (2) a large affinity between a part of D cells and a non-D cell (lambda(UG)and lambda(BR)); and (3) a positive affinity of an intermediate magnitude between two non-D cells (lambda(BU)) and between two parts of D cells (lambda(GR)). The latter should be of a magnitude of about 60-90% of the former. The time needed to form a regular pattern increases with the lattice size if all the cells start pre-pattern formation simultaneously. However, the convergence time is shortened considerably if the pre-pattern formation occurs only in a narrow band of morphogenetic cell layer that sweeps from one end of the lattice to the other.     

Mosaic model for color vision

Color vision in man is based upon three different cone types, which are quite likely arranged in a semi-ordered array in the retina. The model proposes that this ordering is an inherent part of the genetic code that sets up the color vision mechanism, and that the specification for each cone type (red, green or blue) also includes a specification for its place in the larger structure of which it is a part. One possible positional mosaic for the three cone types is proposed, together with its degeneracies into anomalous (red-green) color mechanisms. Assuming only one fixed probability for a degenerate transition, the population frequencies for color anomalies predicted from the model agree closely with the observed frequencies.     

Ultrastructure of the distal retina of the adult zebrafish, Danio rerio

The organization, morphological characteristics, and synaptic structure of photoreceptors in the adult zebrafish retina were studied using light and electron microscopy. Adult photoreceptors show a typical ordered tier arrangement with rods easily distinguished from cones based on outer segment (OS) morphology. Both rods and cones contain mitochondria within the inner segments (IS), including the large, electron-dense megamitochondria previously described (Kim et al.) Four major ultrastructural differences were observed between zebrafish rods and cones: (1) the membranes of cone lamellar disks showed a wider variety of relationships to the plasma membrane than those of rods, (2) cone pedicles typically had multiple synaptic ribbons, while rod spherules had 1-2 ribbons, (3) synaptic ribbons in rod spherules were ∼2 times longer than ribbons in cone pedicles, and (4) rod spherules had a more electron-dense cytoplasm than cone pedicles. Examination of photoreceptor terminals identified four synaptic relationships at cone pedicles: (1) invaginating contacts postsynaptic to cone ribbons forming dyad, triad, and quadrad synapses, (2) presumed gap junctions connecting adjacent postsynaptic processes invaginating into cone terminals, (3) basal junctions away from synaptic ribbons, and (4) gap junctions between adjacent photoreceptor terminals. More vitread and slightly farther removed from photoreceptor terminals, extracellular microtubule-like structures were identified in association with presumed horizontal cell processes in the OPL. These findings, the first to document the ultrastructure of the distal retina in adult zebrafish, indicate that zebrafish photoreceptors have many characteristics similar to other species, further supporting the use of zebrafish as a model for the vertebrate visual system.     

Clonal origins of cells in the pigmented retina of the zebrafish eye

Mosaic analysis has been used to study the clonal basis of the development of the pigmented retina of the zebrafish, Brachydanio rerio. Zebrafish embryos heterozygous for a recessive mutation at the gol-1 locus were exposed to gamma-irradiation at various developmental stages to create mosaic individuals consisting of wild-type pigmented cells and a clone of pigmentless (golden) cells in the eye. The contribution of individual embryonic cells to the pigmented retina was measured and the total number of cells in the embryo that contributed descendants to this tissue was determined. Until the 32-cell stage, almost every blastomere has some descendants that participate in the formation of the pigmented retina of the zebrafish. During subsequent cell divisions, up to the several thousand-cell stage, the number of ancestral cells is constant: approximately 40 cells are present that will give rise to progeny in the pigmented retina. Analysis of the size of clones in the pigmented retina indicates that the cells of this tissue do not arise through a rigid series of cell divisions originating in the early embryo. The findings that each cleavage stage cell contributes to the pigmented retina and yet the contribution of such cells is highly variable are consistent with the interpretation that clonal descendants of different blastomeres normally intermix extensively prior to formation of the pigmented retina.     

Light-induced rod and cone cell death and regeneration in the adult albino zebrafish (Danio rerio) retina

Light-induced photoreceptor cell degeneration has been studied in several species, but not extensively in the teleost fish. Furthermore, the continual production of rods and cones throughout the teleost's life may result in regeneration of lost rods and cones. We exposed adult albino zebrafish to 7 days of constant darkness, followed by 7 days of constant 8000 lux light, followed by 28 days of recovery in a 14-h light:10-h dark cycle. We characterized the resulting photoreceptor layer cell death and subsequent regeneration using immunohistochemistry and light microscopy. Within the first 24 h of constant light, the zebrafish retina exhibited widespread rod and cone cell apoptosis. High levels of cell proliferation within the inner nuclear layer (INL) were observed within the first 3 days of constant light, as assessed by immunodetection of proliferating cell nuclear antigen and BrdU labeling. The proliferating cells within the INL were closely associated with the radial processes of Müller glia, similar to the pluripotent retinal stem cells observed during embryonic development. Using antibodies generated against the individual zebrafish opsins, we determined that rods and the green, blue, and ultraviolet cone cells were replaced within the 28 day recovery period. While both rods and cones were replaced, the well-ordered cone cell mosaic was not reestablished.     

Characterization of Light Lesion Paradigms and Optical Coherence Tomography as Tools to Study Adult Retina Regeneration in Zebrafish

Light-induced lesions are a powerful tool to study the amazing ability of photoreceptors to regenerate in the adult zebrafish retina. However, the specificity of the lesion towards photoreceptors or regional differences within the retina are still incompletely understood. We therefore characterized the process of degeneration and regeneration in an established paradigm, using intense white light from a fluorescence lamp on swimming fish (diffuse light lesion). We also designed a new light lesion paradigm where light is focused through a microscope onto the retina of an immobilized fish (focused light lesion). Focused light lesion has the advantage of creating a locally restricted area of damage, with the additional benefit of an untreated control eye in the same animal. In both paradigms, cell death is observed as an immediate early response, and proliferation is initiated around 2 days post lesion (dpl), peaking at 3 dpl. We furthermore find that two photoreceptor subtypes (UV and blue sensitive cones) are more susceptible towards intense white light than red/green double cones and rods. We also observed specific differences within light lesioned areas with respect to the process of photoreceptor degeneration: UV cone debris is removed later than any other type of photoreceptor in light lesions. Unspecific damage to retinal neurons occurs at the center of a focused light lesion territory, but not in the diffuse light lesion areas. We simulated the fish eye optical properties using software simulation, and show that the optical properties may explain the light lesion patterns that we observe. Furthermore, as a new tool to study retinal degeneration and regeneration in individual fish in vivo, we use spectral domain optical coherence tomography. Collectively, the light lesion and imaging assays described here represent powerful tools for studying degeneration and regeneration processes in the adult zebrafish retina.     

The spectral sensitivity of the retinal photoreceptors of the Asiatic smelt <Emphasis Type="Italic">Osmerus dentex</Emphasis> (Steindachner et Kner, 1870)

The spectral sensitivity and complement of the retinal photoreceptors of the Asiatic smelt from the Sea of Japan were studied by microspectrophotometry and light microscopy. Apart from rods, one type of single cones and one type of unequal double cones were found in major parts of the retina. The dominant type of the cone pattern (mosaic) is a row pattern consisting of various linear arrangements of separate single and double cones. The absorbance maxima of rods and a majority of singe cones and double cones equaled 516, 425 and 514/565 nm, respectively. It has been established that all of the pigments are based on retinal. The findings are compared with data on the osmerid retina from the literature and discussed with respect to the adaptations to light conditions, peculiarities of behavior, and seasonal migrations of smelts.     

Photoreceptor types and their relation to the spectral and polarization sensitivities of clupeid fishes

We used compound action potential recordings from the optic nerve of anesthetized live fish to study the spectral and polarization sensitivities of the northern anchovy and the Pacific herring. The photoreceptor structure and cone mosaic type of the (illuminated) central retina was studied by microscopy. Both species showed a single peak spectral photopic sensitivity function with λ_max= 500 nm for the northern anchovy and λ_max= 520 nm for the herring. However, only the northern anchovy exhibited polarization sensitivity; the response was 180° periodic with maximum sensitivity to horizontal polarization. Similar to the bay anchovy (Fineran and Nicol 1978), the central retina of the northern anchovy showed bifid cone units with cone lamellae parallel to the cones' lengths. The herring, on the other hand, had twin cones arranged in rows with the same orientation and tangentially arranged lamellae. Our results support the hypothesis that bifid cone units act as orthogonal dichroic filters rendering anchovies polarization sensitive. The lack of polarization sensitivity in the herring suggests that twin cones may not be used in polarization sensitivity or that one orientation of polarization receptors is insufficient for the animal to detect polarization direction. Accepted: 8 December 1997     

Color opponency in cone-driven horizontal cells in carp retina. Aspecific pathways between cones and horizontal cells

The spectral and dynamic properties of cone-driven horizontal cells in carp retina were evaluated with silent substitution stimuli and/or saturating background illumination. The aim of this study was to describe the wiring underlying the spectral sensitivity of these cells. We will present electrophysiological data that indicate that all cone-driven horizontal cell types receive input from all spectral cone types, and we will present evidence that all cone-driven horizontal cell types feedback to all spectral cone types. These two findings are the basis for a model for the spectral and dynamic behavior of all cone-driven horizontal cells in carp retina. The model can account for the spectral as well as the dynamic behavior of the horizontal cells. It will be shown that the strength of the feedforward and feedback pathways between a horizontal cell and a particular spectral cone type are roughly proportional. This model is in sharp contrast to the Stell model, where the spectral behavior of the three horizontal cell types is explained by a cascade of feedforward and feedback pathways between cones and horizontal cells. The Stell model accounts for the spectral but not for the dynamic behavior of the horizontal cells.     

Modelling of the vertebrate visual system 3. topological analysis of the cone mosaic

Spatial organization of the cone mosaic of the generalized vertebrate retina consists of rows of red and green cones alternating with rows of blue and blank cones. Cone inputs to retinal elements are defined spatially by red and green unit hexagons. Topological analysis entails determining for each cone in the mosaic the number of each cone type present in the unit hexagon which the activated cone can influence via electrical coupling between cones and/or stray light. Only weighted inputs in one-half of a sextant of the unit hexagon need be designated, since all other weighted inputs can be determined by rules giving systematic transformations of all cone types from one sextant to another: these rules arise from symmetries of the cone mosaic. Four retinal types are possible depending on replacement of blank cones by specific cone types; three cone-dominant retinas, where all blank cones are replaced by a specific cone type, and two forms of a trichromatic retina, where blank cones are replaced by equal numbers of red and green cones. The weighted input is the sum of individual cone type contributions and depends on the number of each cone type in the unit hexagon which can influence the cone in question. Weighted inputs for cone-dominant retinas are readily found by replacing blank cones with the proper cone type, while weighted inputs for trichromatic retinas require use of a specified cone mosaic to determine extra red and green cones. Receptive field size of post-cone elements as well as overlap of the center and surround fields of annular organized receptive fields of retinal elements increased with increasing values for attenuation factors.     

Crumbs proteins stabilize the cone mosaics of photoreceptors and improve vision in zebrafish

Although the unique organization of vertebrate cone mosaics was first described long ago,both their underlying molecular basis and physiological significance are largely unknown.Here,we demonstrate that Crumbs proteins,the key regulators of epithelial apical polarity,establish the planar cellular polarity of photoreceptors in zebrafish.Via heterophilic Crb2a-Crb2b interactions,the apicobasal polarity protein Crb2b restricts the asymmetric planar distribution of Crb2a in photoreceptors.The planar polarized Crumbs proteins thus balance intercellular adhesions and tension between photoreceptors,thereby stabilizing the geometric organization of cone mosaics.Notably,loss of Crb2b in zebrafish induces a nearsightedness-like phenotype in zebrafish accompanied by an elongated eye axis and impairs zebrafish visual perception for predation.These data reveal a detailed mechanism for cone mosaic homeostasis via previously undiscovered apical-planar polarity coordination and propose a pathogenic mechanism for nearsightedness.     

Ontogeny of Double Cones in the Retina of Perch Fry (Perca fluviatilis,Teleostei)

The formation of double cones in the retina of fry of Perca fluviatilis has been investigated by light and electron microscopy. The retina of newly hatched fry is provided with single cones and rods, single cones being the predominant receptor type. Double cones are seen for the first time 22 days after hatching. Mitoses are observed in the periphery of the retina, but are also seen in more central parts of the retina containing differentiated receptors and a cone mosaic. The fate of the cells resulting from the centrally located mitoses is not known. No signs of longitudinal fission of differentiated single cones are seen. It is suggested that double cones in the retina of perch fry arise by fusion of single cones which associate closely and develop subsurface cisterns coextensive with the region of intimate contact in the ellipsoid. During the first few weeks after hatching, there is a gradual shift in arrangement of the cones. In the newly hatched fry, the single cones are arranged in rows. When double cones are first seen, square-pattern units appear, built up from four double cones and a single cone.