The development of retinal ganglion cells in a tetraploid strain of Xenopus laevis: a morphological study utilizing intracellular dye injection

Wiring the retinal circuits activated by light during early development

Background

Light information is sorted by neuronal circuits to generate image-forming (IF) (interpretation and tracking of visual objects and patterns) and non-image-forming (NIF) tasks. Among the NIF tasks, photic entrainment of circadian rhythms, the pupillary light reflex, and sleep are all associated with physiological responses, mediated mainly by a small group of melanopsin-expressing retinal ganglion cells (mRGCs). Using Xenopus laevis as a model system, and analyzing the c-fos expression induced by light as a surrogate marker of neural activity, we aimed to establish the developmental time at which the cells participating in both systems come on-line in the retina.

Results

We found that the peripheral retina contains 80% of the two melanopsin-expressing cell types we identified in Xenopus: melanopsin-expressing horizontal cells (mHCs; opn4m+/opn4x+/Prox1+) and mRGCs (2.7% of the total RGCs; opn4m+/opn4x+/Pax6+/Isl1), in a ratio of 6:1. Only mRGCs induced c-fos expression in response to light. Dopaminergic (tyrosine hydroxylase-positive; TH+) amacrine cells (ACs) may be part of the melanopsin-mediated circuit, as shown by preferential c-fos induction by blue light. In the central retina, two cell types in the inner nuclear layer (INL) showed light-mediated induction of c-fos expression [(On-bipolar cells (Otx2+/Isl1+), and a sub-population of ACs (Pax6−/Isl1−)], as well as two RGC sub-populations (Isl1+/Pax6+ and Isl1+/Pax6−). Melanopsin and opsin expression turned on a day before the point at which c-fos expression could first be activated by light (Stage 37/38), in cells of both the classic vision circuit, and those that participate in the retinal component of the NIF circuit. Key to the classic vision circuit is that the component cells engage from the beginning as functional ‘unit circuits’ of two to three cells in the INL for every RGC, with subsequent growth of the vision circuit occurring by the wiring in of more units.

Conclusions

We identified melanopsin-expressing cells and specific cell types in the INL and the RGC layer which induce c-fos expression in response to light, and we determined the developmental time when they become active. We suggest an initial formulation of retinal circuits corresponding to the classic vision pathway and melanopsin-mediated circuits to which they may contribute.

Modeling human neurodevelopmental disorders in the Xenopus tadpole: from mechanisms to therapeutic targets

The Xenopus tadpole model offers many advantages for studying the molecular, cellular and network mechanisms underlying neurodevelopmental disorders. Essentially every stage of normal neural circuit development, from axon outgrowth and guidance to activity-dependent homeostasis and refinement, has been studied in the frog tadpole, making it an ideal model to determine what happens when any of these stages are compromised. Recently, the tadpole model has been used to explore the mechanisms of epilepsy and autism, and there is mounting evidence to suggest that diseases of the nervous system involve deficits in the most fundamental aspects of nervous system function and development. In this Review, we provide an update on how tadpole models are being used to study three distinct types of neurodevelopmental disorders: diseases caused by exposure to environmental toxicants, epilepsy and seizure disorders, and autism.

Dissection, culture, and analysis of Xenopus laevis embryonic retinal tissue

The process by which the anterior region of the neural plate gives rise to the vertebrate retina continues to be a major focus of both clinical and basic research. In addition to the obvious medical relevance for understanding and treating retinal disease, the development of the vertebrate retina continues to serve as an important and elegant model system for understanding neuronal cell type determination and differentiation(1-16). The neural retina consists of six discrete cell types (ganglion, amacrine, horizontal, photoreceptors, bipolar cells, and Müller glial cells) arranged in stereotypical layers, a pattern that is largely conserved among all vertebrates (12,14-18). While studying the retina in the intact developing embryo is clearly required for understanding how this complex organ develops from a protrusion of the forebrain into a layered structure, there are many questions that benefit from employing approaches using primary cell culture of presumptive retinal cells (7,19-23). For example, analyzing cells from tissues removed and dissociated at different stages allows one to discern the state of specification of individual cells at different developmental stages, that is, the fate of the cells in the absence of interactions with neighboring tissues (8,19-22,24-33). Primary cell culture also allows the investigator to treat the culture with specific reagents and analyze the results on a single cell level (5,8,21,24,27-30,33-39). Xenopus laevis, a classic model system for the study of early neural development (19,27,29,31-32,40-42), serves as a particularly suitable system for retinal primary cell culture (10,38,43-45). Presumptive retinal tissue is accessible from the earliest stages of development, immediately following neural induction (25,38,43). In addition, given that each cell in the embryo contains a supply of yolk, retinal cells can be cultured in a very simple defined media consisting of a buffered salt solution, thus removing the confounding effects of incubation or other sera-based products (10,24,44-45). However, the isolation of the retinal tissue from surrounding tissues and the subsequent processing is challenging. Here, we present a method for the dissection and dissociation of retinal cells in Xenopus laevis that will be used to prepare primary cell cultures that will, in turn, be analyzed for calcium activity and gene expression at the resolution of single cells. While the topic presented in this paper is the analysis of spontaneous calcium transients, the technique is broadly applicable to a wide array of research questions and approaches (Figure 1).

The Xenopus retinal ganglion cell as a model neuron to study the establishment of neuronal connectivity

Neurons receive inputs through their multiple branched dendrites and pass this information on to the next neuron via long axons, which branch within the target. The shape the neuron acquires is thus the key to its proper functioning in the neural circuit in which it participates. Both axons and dendrites grow in a directed fashion to their target partner neurons by responding to a large number of molecular cues in the milieu through which they extend. They then go through the process of synaptogenesis, first choosing a neuron on which to synapse, and then the appropriate subcellular location. How a neuron acquires its unique shape, establishes and modifies appropriate synaptic connectivity, and the molecular signals involved, are key questions in developmental neurobiology. Such questions of nervous system wiring are being pursued actively with a variety of different animal models and neuron types, each with its own unique advantages. Among these, the developing retinal ganglion cell (RGC) of the South African clawed frog, Xenopus laevis, has proven particularly fruitful for revealing the secrets of how axons and dendrites acquire their final morphology and connectivity. In this review, we describe how this system can be used to understand the multiple molecular events that instruct the incorporation of RGCs into the neural circuit that controls vision.

Neural activity and branching of embryonic retinal ganglion cell dendrites

The shape of a neuron's dendritic arbor is critical for its function as it determines the number of inputs the neuron can receive and how those inputs are processed. During development, a neuron initiates primary dendrites that branch to form a simple arbor. Subsequently, growth occurs by a process that combines the extension and retraction of existing dendrites, and the addition of new branches. The loss and addition of the fine terminal branches of retinal ganglion cells (RGCs) is dependent on afferent inputs from its synaptic partners, the amacrine and bipolar cells. It is unknown, however, whether neural activity regulates the initiation of primary dendrites and their initial branching. To investigate this, Xenopus laevis RGCs developing in vivo were made to express either a delayed rectifier type voltage-gated potassium (KV) channel, Xenopus Kv1.1, or a human inward rectifying channel, Kir2.1, shown previously to modulate the electrical activity of Xenopus spinal cord neurons. Misexpression of either potassium channel increased the number of branch points and the total length of all the branches. As a result, the total dendritic arbor was bigger than for control green fluorescent protein-expressing RGCs and those ectopically expressing a highly related mutant non-functional Kv1.1 channel. Our data indicate that membrane excitability regulates the earliest differentiation of RGC dendritic arbors.

Gap-Junction Proteins in Retinal Development: New Roles for the “Nexus”

Gap-junction channels, the cytoplasmic proteins that associate with them, and the transcriptional networks that regulate them are increasingly being viewed as critical communications hubs for cell signaling in health and disease. As a result, the term “nexus,” which was the original structural name for these focal intercellular links, is coming back into use with new proteomic and transcriptomic meanings. The retina is better understood than any other part of the vertebrate central nervous system in respect of its developmental patterning, its diverse neuronal types and circuits, and the emergence of its definitive structure-function correlations. Thus, studies of the junctional and nonjunctional nexus roles of gap-junction proteins in coordinating retinal development should throw useful light on cell signaling in other developing nervous tissues.

TGFbeta ligands promote the initiation of retinal ganglion cell dendrites in vitro and in vivo

Each type of neuron develops a unique morphology critical to its function, but almost all start with the basic plan of one long axon and multiple short, branched dendrites. Though extrinsic signals are known to direct many steps in the development of neuronal structure, little is understood about the initiation of processes, particularly dendrites. We find that Xenopus retinal ganglion cells (RGCs) explanted early will extend axons and not dendrites in dissociated cultures. If RGCs develop longer in vivo prior to culturing, many now extend dendrite-like processes in vitro, suggesting that an extrinsic factor is required to stimulate dendrite initiation. Members of the transforming growth factor beta (TGFbeta) superfamily, bone morphogenetic protein 2 (BMP2), and growth and differentiation factor 11 (GDF11), can signal cultured RGCs to form dendrites. Furthermore, TGFbeta ligands have an endogenous role: blocking BMP/GDF signaling with a secreted antagonist or inhibitory receptors reduces the number of primary dendrites extended in vivo.

Spatial and temporal patterns of distribution of the gap junctional protein connexin43 during retinal regeneration of adult newt

Newts possess the ability to regenerate a functional retina after complete removal of the original retina. We performed immunoblot and immunohistochemical analyses of newt retinas at different stages of regeneration by using an antibody against a gap junction channel protein, connexin43 (Cx43). The specificity of the antibody was shown on immunoblots as well as immunohistochemical staining pattern in the normal retina. Punctate Cx43 immunolabeling was detected intensely between proliferating cell nuclear antigen-immunoreactive progenitor cells in the regenerating retinas, and the amount of this labeling tended to be prominent along both scleral and vitreal sides. The amount of Cx43 became less abundant as regeneration proceeded. This temporal loss of Cx43 during regeneration was also shown on the immunoblot analysis. Furthermore, the loss of Cx43 was observed in a spatial manner in the peripheral retina, where progenitor cells clustered at the ciliary marginal zone (CMZ) are adding new cells of all types in order toward the central retina. Immunolabeling often extended longitudinally throughout the retina when regenerating retinas became thick. Double immunolabeling with Cx43 and glial fibrillary acidic protein indicated the overlapping between the Cx43 and Müller cell processes. At the beginning of the synaptic formation, immunolabeling almost disappeared in the entire retina. However, in the completely regenerated retina, Cx43 reappeared in the distal end of Müller cells and pigment epithelial cells in the same pattern as in the normal retina. The above observations lead us to speculate that Cx43-mediated gap junctions may play an important role in regenerating events. Possible roles of Cx43 during regeneration are discussed.

Expression patterns of focal adhesion associated proteins in the developing retina

Adhesive interactions between integrin receptors and the extracellular matrix (ECM) are intimately involved in regulating development of a variety of tissues within the organism. In the present study, we have investigated the relationships between beta(1) integrin receptors and focal adhesion associated proteins during eye development. We used specific antibodies to examine the distribution of beta(1) integrin ECM receptors and the cytoplasmic focal adhesion associated proteins, talin, vinculin, and paxillin in the developing Xenopus retina. Immunoblot analysis confirmed antibody specificity and indicated that beta(1) integrins, talin, vinculin, and paxillin were expressed in developing retina and in the retinal-derived Xenopus XR1 glial cell line. Triple-labeling immunocytochemistry revealed that talin, vinculin, paxillin, and phosphotyrosine proteins colocalized with beta(1) integrins at focal adhesions located at the termini of F-actin filaments in XR1 cells. In the retina, these focal adhesion proteins exhibited developmentally regulated expression patterns during eye morphogenesis. In the embryonic retina, immunoreactivities for focal adhesion proteins were expressed in neuroepithelial cells, and immunoreactivity was especially strong at the interface between the optic vesicle and overlying ectoderm. At later stages, these proteins were expressed throughout all retinal layers with higher levels of expression observed in the plexiform layers, optic fiber layer, and in the region of the inner and outer limiting membrane. Strong immunoreactivities for beta(1) integrin, paxillin, and phosphotyrosine were expressed in the radially oriented Müller glial cells at later stages of development. These results suggest that focal adhesion-associated proteins are involved in integrin-mediated adhesion and signaling and are likely to be essential in regulating retinal morphogenesis.

Local and target-derived brain-derived neurotrophic factor exert opposing effects on the dendritic arborization of retinal ganglion cells in vivo

The dendritic and axonal arbors of developing retinal ganglion cells (RGCs) are exposed to two sources of BDNF: RGC dendrites are exposed to BDNF locally within the retina, and RGC axons are exposed to BDNF at the target, the optic tectum. Our previous studies demonstrated that increasing tectal BDNF levels promotes RGC axon terminal arborization, whereas increasing retinal BDNF levels inhibits RGC dendritic arborization. These results suggested that differential neurotrophic action at the axon versus dendrite might be responsible for the opposing effects of BDNF on RGC axonal versus dendritic arborization. To explore this possibility, we examined the effects of altering BDNF levels at the optic tectum on the elaboration of RGC dendritic arbors in the retina. Increasing tectal BDNF levels resulted in a significant increase in dendritic branching, whereas neutralizing endogenous tectal BDNF with function-blocking antibodies significantly decreased dendritic arbor complexity. Thus, RGC dendritic arbors react in opposing manners to retinal- versus tectal-derived BDNF. Alterations in retinal BDNF levels, however, did not affect axon terminal arborization. Thus, RGC dendritic arborization is controlled in a complementary manner by both local and target-derived sources of BDNF, whereas axon arborization is modulated solely by neurotrophic interactions at the target. Together, our results indicate that developing RGCs modulate dendritic arborization by integrating signals from discrete sources of BDNF in the eye and brain. Differential integration of spatially discrete neurotrophin signals within a single neuron may therefore finely tune afferent and efferent neuronal connectivity.

Gap junctional coupling between progenitor cells at the retinal margin of adult goldfish

We prepared living slice preparations of the peripheral retina of adult goldfish to examine electrical membrane properties of progenitor cells at the retinal margin. Cells were voltage-clamped near resting potential and then stepped to either hyperpolarizing or depolarizing test potentials using whole-cell voltage-clamp recordings. Electrophysiologically examined cells were morphologically identified by injecting both Lucifer Yellow (LY) and biocytin. All progenitor cells examined (n = 37) showed a large amount of passively flowing currents of either sign under suppression of the nonjunctional currents flowing through K(+) and Ca(2+) channels in the cell membrane. They did not exhibit any voltage-gated Na(+) currents. Cells identified by LY fills were typically slender. As the difference between the test potential and the resting potential increased, 13 out of 37 cells exhibited symmetrically voltage- and time-dependent current decline on either sign at the resting potential. The symmetric current profile suggests that the current may be driven and modulated by the junctional potential difference between the clamping cell and its neighbors. The remaining 24 cells did not exhibit voltage dependency. A gap junction channel blocker, halothane, suppressed the currents. A decrease in extracellular pH reduced coupling currents and its increase enhanced them. Dopamine, cAMP, and retinoic acid did not influence coupling currents. Injection of biocytin into single progenitor cells revealed strong tracer coupling, which was restricted in the marginal region. Immature ganglion cells closely located to the retinal margin exhibited voltage-gated Na(+) currents. They did not reveal apparent tracer coupling. These results demonstrate that the marginal progenitor cells couple with each other via gap junctions, and communicate biochemical molecules, which may subserve or interfere with cellular differentiation.

Functional differentiation of ganglion cells from multipotent progenitor cells in sliced retina of adult goldfish

Multipotent progenitor cells at the retinal margin of adult goldfish give rise to all cell types in the rest of the retina. We took advantage of this spatial arrangement of progenitor and mature cells in slices of peripheral retina, to investigate the appearance and maturation of voltage-activated Na(+) current. We divided the peripheral retina into three broad regions (marginal, intermediate, and mature) on the basis of their morphological development. Whole-cell patch-clamp recordings were performed in ruptured-patch mode, so that cells from which currents were recorded could be identified by Lucifer Yellow fills. No voltage-activated Na(+) current was detected in the slender, peripherally located marginal cells. Voltage-activated Na(+) currents were detected in rounded cells found alongside or near marginal cells, facing the vitreal side of the retina. Some of these "intermediate cells" had a long axon-like process which ran along the vitreal surface. Intermediate cells adjacent to the marginal region tended to have smaller Na(+) currents than intermediate cells closer to the mature region. On average, the maximum Na(+) current amplitude recorded from intermediate cells was roughly 6-fold smaller than that of mature ganglion cells. In addition, the activation threshold of the Na(+) current in intermediate cells was nearly 14 mV more positive than that of mature ganglion cells. The results indicate that voltage-activated Na(+) current, as a possible marker of retinal ganglion cells, begins to develop well before these cells migrate to their adult position within the retina.

A role for voltage-gated potassium channels in the outgrowth of retinal axons in the developing visual system

Neural activity is important for establishing proper connectivity in the developing visual system. Tetrodotoxin blockade of sodium (Na(+))-dependent action potentials impairs the refining of synaptic connections made by developing retinal ganglion cells (RGCs), but does not affect their ability to get out to their target. Although this may suggest neural activity is not required for the directed extension of RGC axons, in many species developing RGCs express additional, Na(+)-independent ionic mechanisms. To test whether the ability of RGC axons to extend in a directed fashion is influenced by membrane excitability, we blocked the principal modulators of the neural activity of a neuron, voltage-dependent potassium (Kv) channels. First, we showed that RGCs and their growth cones express Kv channels when they are growing through the brain on the way to their main midbrain target, the optic tectum. Second, a Kv channel blocker, 4-aminopyridine (4-AP), was applied to the developing Xenopus optic projection. Blocking Kv channels inhibited RGC axon extension and caused aberrant routing of many RGC fibers. With the higher doses, <25% of embryos had a normal optic projection. These data suggest that Kv channel activity regulates the guidance of growing axons in the vertebrate brain.

Gap junctional coupling between progenitor cells of regenerating retina in the adult newt

Gap junctional coupling between progenitor cells of regenerating retina in the adult newt was examined by a slice-patch technique. Retinal slices at the early regeneration stage comprised one to two layers of cells with mitotic activity, progenitor cells. These cells were initially voltage-clamped at a holding potential of -80 mV, near their resting potentials, and stepped to either hyperpolarizing or depolarizing test potentials under suppression of voltage-gated membrane currents. About half the cells showed passively flowing currents that reversed polarity around their resting potentials. The currents often exhibited a voltage- and time-dependent decline. As the difference between the test potential and resting potential increased, the time until the current decreased to the steady-state level became shorter and the amount of steady-state current decreased. Thus, the overall current profile was almost symmetrical about the current at the resting potential. Input resistance estimated from the initial peak of the currents was significantly smaller than that expected in isolated progenitor cells. In a high-K(+) solution, which decreased the resting potential to around 0 mV, the symmetrical current profile was also obtained, but only when the membrane potential was held at 0 mV before the voltage steps. These observations suggest that the current was driven and modulated by the junctional potential difference between the clamping cell and its neighbors. In addition, we examined effects of uncoupling agents on the currents. A gap junction channel blocker, halothane, suppressed the currents almost completely, indicating that the currents are predominantly gap junctional currents. Furthermore, injection of biocytin into the current-recorded cells revealed tracer coupling. These results demonstrate that progenitor cells of regenerating retina couple with each other via gap junctions, and suggest the presence of their cytoplasmic communication during early retinal regeneration.

Brain-derived neurotrophic factor differentially regulates retinal ganglion cell dendritic and axonal arborization in vivo

Expression of the neurotrophin brain-derived neurotrophic factor (BDNF) and its receptor trkB in the ganglion cell layer of the Xenopus retina during retinal ganglion cell (RGC) dendritic arborization indicates that BDNF is spatially and temporally available to influence RGC morphological differentiation (; ). BDNF promotes RGC axon arborization in vivo by acting as a target-derived trophic factor (). To determine whether BDNF also acts locally to regulate RGC dendritic development in vivo, we altered retinal neurotrophin levels at the onset of dendritic arborization and assessed the resulting arbor morphologies of RGCs retrogradely labeled with fluorescent dextrans. Injecting neurotrophins or BDNF function-blocking antibodies coupled to microspheres provided local alterations of retinal neurotrophin levels. BDNF significantly decreased RGC dendritic arbor complexity, whereas neutralizing endogenous BDNF levels with function-blocking antibodies significantly increased dendritic arbor complexity. RGCs exposed to other neurotrophins, as well as RGCs in retinae treated with BDNF but in areas not directly exposed to the neurotrophin, developed dendritic arbors that were indistinguishable from controls, indicating that exogenous BDNF acts specifically and locally. In the tectum, where RGC axons arborize, BDNF had opposite effects. BDNF significantly increased RGC axon arbor complexity and anti-BDNF reduced RGC arborization. Thus, BDNF reduces RGC dendritic arborization within the retina and increases axon arborization in the tectum. These results indicate that BDNF can differentially modulate axonal and dendritic arborization within a single neuronal population in opposing manners and raise the possibility that differential modulation by a neurotrophic factor finely tunes the morphological differentiation program of a neuron.

The neuronal architecture of Xenopus retinal ganglion cells is sculpted by rho-family GTPases in vivo

Dendritogenesis, axonogenesis, pathfinding, and target recognition are all affected in distinct ways when Xenopus retinal ganglion cells (RGCs) are transfected with constitutively active (ca), wild-type (wt), and dominant negative (dn) Rho-family GTPases in vivo. Dendritogenesis required Rac1 and Cdc42 activity. Moreover, ca-Rac1 caused dendrite hyperproliferation. Axonogenesis, in contrast, was inhibited by ca-Rac1. This phenotype was partially rescued by the coexpression of dn cyclin-dependent kinase (Cdk5), a proposed effector of Rac1, suggesting that Rac1 activity must be regulated tightly for normal axonogenesis. Growth cone morphology was particularly sensitive to dn-RhoA and wt-Cdc42 constructs. These also caused targeting errors, such as tectal bypass, suggesting that cytoskeletal rearrangements are involved in target recognition and are transduced by these pathways.

Morphological classification of ganglion cells in the central retina of chicks

Classification of retinal ganglion cells (RGCs) in the chick central retina was studied by retrograde labeling of carbocyanine dye (DiI) and intracellular filling with Lucifer Yellow. Ganglion cells were divided into 4 groups, Group Ic/Is, Group IIc/IIs, Group IIIs, Group IVc, according to sizes of somal area and dendritic field and dendritic branching pattern. Group I cells had small somal area and small dendritic field. They were further divided into 2 subgroups by complexity (subgroup Ic) and simplicity (subgroup Is) of the dendritic arborization. Group II cells had medium-sized soma and dendritic field. They were also divided into subgroup IIc and IIs by the same definitions as those of subgroup Ic and Is. Group IIIs had medium-sized soma, large and simple dendritic arborization. Group IVc in which all cells had large soma, showed large and complex dendritic arborization. Cell populations of each group were 51.8% (subgroup Ic), 21.1% (subgroup Is), 6.2% (subgroup IIc), 14.6% (subgroup IIs), 4.2% (Group IIIs), and 2.1% (Group IVc). Subgroup Ic cells, which were very similar to beta-cells in the mammalian central area, represented about a half of the ganglion cell population. Cells in subgroup Is and IIs, which were not reported in the mammalian retina, were found in the chick central retina in relatively high population (35.7%). Morphological features of chick RGCs in the central retina were considered in comparison with those of other vertebrates.

Synchronizing retinal activity in both eyes disrupts binocular map development in the optic tectum

Spatiotemporal correlations in the pattern of spontaneous and evoked retinal ganglion cell (RGC) activity are believed to influence the topographic organization of connections throughout the developing visual system. We have tested this hypothesis by examining the effects of interfering with these potential activity cues during development on the functional organization of binocular maps in the Xenopus frog optic tectum. Paired recordings combined with cross-correlation analyses demonstrated that exposing normal frogs to a continuous 1 Hz of stroboscopic illumination synchronized the firing of all three classes of RGC projecting to the tectum and induced similar patterns of temporally correlated activity across both lobes of the nucleus. Embryonic and eye-rotated larval animals were reared until early adulthood under equivalent stroboscopic conditions. The maps formed by each RGC class in the contralateral tectum showed normal topography and stratification after strobe rearing, but with consistently enlarged multiunit receptive fields. Maps of the ipsilateral eye, formed by crossed isthmotectal axons, showed significant disorder and misalignment with direct visual input from the retina, and in the eye-rotated animals complete compensatory reorientation of these maps usually induced by this procedure failed to occur. These findings suggest that refinement of retinal arbors in the tectum and the ability of crossed isthmotectal arbors to establish binocular convergence with these retinal afferents are disrupted when they all fire together. Our data thus provide direct experimental evidence that spatiotemporal activity patterns within and between the two eyes regulate the precision of their developing connections.

Large retinal ganglion cells in the pipid frog Xenopus laevis form independent, regular mosaics resembling those of teleost fishes

Population-based studies of retinal neurons have helped to reveal their natural types in mammals and teleost fishes. In this, the first such study in a frog, labeled ganglion cells of the mesobatrachian Xenopus laevis were examined in flatmounts. Cells with large somata and thick dendrites could be divided into three mosaic-forming types, each with its own characteristic stratification pattern. These are named alpha a, alpha ab, and alpha c, following a scheme recently used for teleosts. Cells of the alpha a mosaic (approximately 0.4% of all ganglion cells) had very large somata and trees, arborizing diffusely within sublamina a (the most sclerad). Their distal dendrites were sparsely branched but achieved consistent coverage by intersecting those of their neighbors. Displaced and orthotopic cells belonged to the same mosaic, as did cells with symmetric and asymmetric trees. Cells of the alpha ab mosaic (approximately 1.2%) had large somata, somewhat smaller trees that appeared bistratified at low magnification, and dendrites that branched extensively. Their distal dendrites arborized throughout sublamina b and the vitread part of a, tessellating with their neighbors. All were orthotopic; most were symmetric. Cells of the alpha c mosaic (approximately 0.5%) had large somata and very large, sparse, flat, overlapping trees, predominantly in sublamina c. All were orthotopic; some were asymmetric. Nearest-neighbor analyses and spatial correlograms confirmed that each mosaic was regular and independent, and that spacings were reduced in juvenile frogs. Densities, proportions, sizes, and mosaic statistics are tabulated for all three types, which are compared with types defined previously by size and symmetry in Xenopus and potentially homologous mosaic-forming types in teleosts. Our results reveal strong organizational similarities between the large ganglion cells of teleosts and frogs. They also demonstrate the value of introducing mosaic analysis at an early stage to help identify characters that are useful markers for natural types and that distinguish between within-type and between-type variation in neuronal populations.

Large retinal ganglion cells in the channel catfish (Ictalurus punctatus): three types with distinct dendritic stratification patterns form similar but independent mosaics

Retinal ganglion cells in the channel catfish (Ictalurus punctatus) were retrogradely labelled, and those with the largest somata and thickest primary dendrites were categorized by their levels of dendritic stratification. Three types were found, each forming a mosaic making up approximately 1% of the ganglion cell population. Using a system based on established sublaminar terminology, we call these the alpha-a (alpha a), alpha-b (alpha b), and alpha-c (alpha c) ganglion cell mosaics. Cells of the alpha a mosaic had large, sparsely branched trees in sublamina a at 10-30% of the depth of the inner plexiform layer (IPL), sclerad to those of all other large ganglion cells. Some alpha a somata were displaced into the IPL or inner nuclear layer (INL) but belonged to the same mosaic as their orthotopic counterparts. Cells of the alpha b mosaic had dendrites that branched a little more and arborized in sublamina b at 50-60% of the IPL depth. Many also sent fine branches into sublamina a, and some were fully bistratified in a and b. The alpha c cells arborized in the most vitread sublamina, sublamina c, at 80-95% of the IPL depth. The soma areas of the three types in the largest retina studied ranged between 139 microns 2 and 670 microns 2 with significant differences in the order alpha a > alpha c > or = alpha b. Analyses based on nearest-neighbour distance (NND) and on spatial auto- and cross-correlograms showed that each mosaic was statistically regular and independent of the others. Mosaic spacings were similar for each type, giving mean NNDs of 242-279 microns in the largest retina and 153-159 microns in a smaller one. Correspondences between these mosaics, previously defined large ganglion cell types in catfish, and other mosaic-forming large ganglion cells in fish and frogs are discussed along with their implications for neuronal classification, function, development, and evolution.

Gap junctions in the vertebrate retina

The vertebrate retina is a highly laminated assemblage of specialized neuronal types, many of which are coupled by gap junctions. With one interesting exception, gap junctions are not directly responsible for the 'vertical' transmission of visual information from photoreceptors through bipolar and ganglion cells to the brain. Instead, they mediate 'lateral' connections, coupling neurons of a single type or subtype into an extended, regular array or mosaic in the plane of the retina. Such mosaics have been studied by several microscopic techniques, but new evidence for their coupled nature has recently been obtained by intracellular injection of biotinylated tracers, which can pass through gap junctional assemblies that do not pass Lucifer Yellow. This evidence adds momentum to an existing paradigm shift towards a population-based view of the retina, which can now be envisaged both as an array of semi-autonomous vertical processing modules, each extending right through the retina, and as a multi-layered stack of interacting planar mosaics, bearing some resemblance to a set of interleaved neural networks. Junctional conductance across mosaics of horizontal cells is known to be controlled dynamically with a circadian rhythm, and other dynamically-regulated conductance changes are also likely to make important contributions to signal processing. The retina is an excellent system in which to study such changes because many aspects of its structure and function are already well understood. In this review, we summarize the microscopic appearance, coupling properties and functions of gap junctions for each cell type of the neural retina, the regulatory properties that could be provided by selective expression of different connexin proteins, and the evidence for gap junctional coupling in retina development.

A biphasic sequence of myelination in the developing optic nerve of the frog

We have examined the sequence of myelination along the optic nerve of the frog Litoria (Hyla) moorei from early tadpole life to adulthood. Myelinated axons were counted in electron micrographs of transverse sections taken from behind the eye, at the optic foramen and the chiasm. In tadpoles, myelinated axon numbers were significantly higher at the foramen than at the other levels. By metamorphic climax, numbers had risen at all three levels but more so behind the eye and at the chiasm to become approximately equal along the nerve. After metamorphosis, there was a dramatic increase in myelinated axon numbers, but another pattern was seen; in frogs of 5 cm and 7 cm body length, counts were significantly higher at the chiasm than at the foramen and lowest behind the eye. Thereafter, myelinated axon numbers stabilized at the chiasm but increased behind the eye and at the foramen so that in the most mature stage for this species, 9 cm adults, counts were again similar at the three levels. In addition, total axon numbers, that is, myelinated plus unmyelinated, were assessed from electron micrographs and increased from approximately 8,500 in early tadpoles to 0.65 million in fully mature adults. The proportion of axons that were myelinated showed two peaks, one before and the other after metamorphosis. Measurements of axon diameters from electron micrographs suggested that there was a critical diameter for myelination of 0.3 microns before, and of 0.5 microns after metamorphosis. The data indicate that there is a biphasic sequence of myelination of optic axons, the first phase being pre-metamorphic and the second post-metamorphic. The first phase is initiated at the foramen, and then extends both towards the eye and chiasm and continues until metamorphic climax. During the second phase, myelination originates at the chiasm, spreads towards the eye, and is complete only in the most mature adults. The critical diameter for myelination is smaller in the first phase than in the second.

Dendritic remodelling of retinal ganglion cells during development of the rat

Investigation of the morphology of ganglion cells in the cat retina has shown that a remarkable reduction in the number of dendritic spines and branches occurs during development of the alpha and beta cell classes. To learn whether dendritic remodelling represents a generalized mechanism of mammalian retinal ganglion cell development, we have examined the morphology of ganglion cells in the retina of the developing rat. The present study has concentrated on type II cells, which retain a great number of dendritic spines and branches in the adult and comprise a large proportion of the population of rat retinal ganglion cells. To reveal fine dendritic and axonal processes, Lucifer yellow was injected intracellularly in living retinae maintained in vitro. Size and complexity of the dendritic trees were found to increase rapidly during an initial stage of development lasting from late fetal life until approximately postnatal day 12 (P12). Dendrites and axons of immature ganglion cells expressed several transient morphological features comprising an excessive number of dendritic branches and spine-like processes, and short, delicate axonal sidebranches. The following developmental stage was characterized by a remarkable decrease in the morphological complexity of retinal ganglion cells and a slowed growth of their dendritic fields. The number of dendritic branches and spines of types I and II retinal ganglion cells declined after P12 to reach a mature level by the end of the first postnatal month. Thus, even cells that retain a highly complex dendritic tree into the adult state undergo extensive remodelling. These results suggest that regressive modifications at the level of the dendritic field constitute a generalized mechanism of maturation in mammalian retinal ganglion cells.

Mature, growing ganglion cells acquire new synapses in the retina of the goldfish

The goldfish retina grows throughout the animal's life, primarily by a balloon-like expansion. With this expansion, dendritic arbors of ganglion cells show scaled growth; arbors increase in size from small to large with no change in their architecture (Hitchcock & Easter, 1986; Bloomfield & Hitchcock, 1991). The study reported here showed that ganglion cell arbors acquire new synapses with this growth. Arbors from a single type of ganglion cell in retinas of small, young and large, old fish were intracellularly filled with horseradish peroxidase, examined electron microscopically, and the synapses contacting them counted and compared (small arbors vs. large arbors). The small and large arbors had similar numbers and orders of dendritic branches (i.e. similar architectures), but the large arbors were significantly larger than the small ones. The increase in arbor size was correlated with a 2.7x and 1.9x increase in the number of ribbon and conventional synaptic contacts, respectively. The addition of new synapses is proposed as a mechanism by which the signaling properties of the enlarging ganglion cells can remain constant.

Involvement of neuronal cell surface molecule B2 in the formation of retinal plexiform layers

The B2 molecule is a 220 kd neuronal cell surface protein of Xenopus, recognized by monoclonal antibody B2 (MAb B2). Immunohistochemistry using MAb B2 revealed that the B2 molecule was expressed in both the inner and outer plexiform layers within the neural retina. During development of the neural retina, the B2 molecule first appeared at stages 35/36 in the newly formed plexiform layers. When embryonic eyes were cultured in the presence of anti-B2 antiserum (Fab fragments), the formation of the retinal plexiform layers was impeded. These data suggest that the cell surface molecule B2 plays a role in the development of retinal plexiform layers.

Synchronous neurite branchings in single goldfish retinal ganglion cells

We have examined the time course of branch formation in neurites of retinal ganglion cells isolated from adult goldfish (Carassius auratus). These neurites elongate at approximately 13 microns/h, and usually branch by bifurcation of growth cones at their tips. The times elapsed between branchings in different neurites of single cells can be described by a Poisson distribution with a mean interval of approximately 2 h. As predicted by this distribution, a relatively large number of branchings occur simultaneously in different neurites of individual cells. Simultaneous branchings of neurites elongating at a common rate generate branch points that lay equidistant from their soma. Since similar branching patterns can be seen in dendrites of retinal ganglion and amacrine cells in situ, these results are consistent with the possibility that dendrites of individual neurons branch synchronously and grow at common rates during development.

Metamorphosis of the amphibian eye

For many metamorphosing amphibians, the visual system must remain functional as the animal changes from an aquatic to a terrestrial habitat. Thyroid hormone, the trigger for metamorphosis, brings about changes at all levels of the animal, and profoundly alters the visual system, from cellular changes within the eye to new central connections subserving the binocular vision that develops during metamorphosis in some species. I will survey the alterations in the visual system in the metamorphosis of several Amphibian groups, and consider the role of thyroid hormone in bringing about these transformations through action at the molecular level.

Neuropeptide Y- and substance P-like immunoreactive amacrine cells in the retina of the developing Xenopus laevis

The development of neuropeptide Y-like (NPY-LI) and substance P-like (SP-LI) immunoreactive neurons was studied in retinas of Xenopus laevis from young tadpole through to adult animals. In adult retina these neuropeptides are present in wide-field amacrine cells located in the inner nuclear layer and the ganglion cell layer of the retina. Retinal wholemount preparations and sectioned material showed that immunoreactive cells appeared during early larval life and NPY-LI occurred earlier than SP-LI cells. The primary dendritic branching of NPY-LI neurons appeared from early larval life whilst SP-LI was evident in dendrites from mid-larval stages. In postmetamorphic animals the numbers of immunoreactive cells increased in proportion to retinal area growth with a relatively constant cell density of about 35 cells/mm2 for SP-LI and 45 cells/mm2 for NPY-LI. The maturation of dendritic morphology of both NPY- and SP-LI amacrine cells appeared later in larval development than the appearance of immunoreactivity in cell somas. However, the sequence of expression of NPY- or SP-LI and their dendritic maturation was different for the two classes of amacrine cells. It is suggested that the maturation of dendritic fields of amacrine cells is complete just prior to metamorphosis, consistent with the postmetamorphic onset of electrophysiological features of ganglion cells attributed to amacrine cells.

Morphological classification and retinal distribution of large ganglion cells in the retina of Bufo marinus

The retrograde transport of horseradish peroxidase (HRP) and cobaltic-lysine complex (CLC) was used to morphologically characterize large ganglion cells (GCs) and to determine their distribution in retinal wholemounts and in sectioned material in the retina of Bufo marinus. Large GCs, amounting to about 0.5% of total GC population, were defined to be those with very large dendritic field sizes varying between 0.1 mm2 to 0.6 mm2 and cell soma sizes of between 100 microns 2 to 400 microns 2. These cells were subdivided into 3 major groups, Types I, II and III, on the basis of their dendritic field sizes, aborization patterns and the strata of dendritic branching within the inner plexiform layer (IPL). The majority of large neurons (about 90%) were classified as Type I GCs with symmetrical dendritic arbor. These cells had either bistratified branching in the scleral and vitreal sublamina of the IPL (65% of Type I Cells) or unistratified branching in the scleral (26%) or in the vitreal (9%) sublamina. Their dendritic field sizes increased linearly from the retinal centre from 0.13 mm +/- 0.02 mm2 (mean and S.D.) to 0.58 +/- 0.11 mm2 in the retinal periphery. Type II GCs (about 9% of large GC population) were characterized by an asymmetrical dendritic aborization directed towards the ciliary margin with unistratified branching in the scleral sublamina of the IPL. The mean dendritic field sizes of these cells were 0.26 +/- 0.09 mm2. Type III GCs, the least frequent (about 1%) category of large GCs had sparsely branching, elongated dendritic branching aligned approximately parallel with the nasotemporal axis of the retina. The unistratified dendritic branches of these neurons were located in the vitreal sublamina of the IPL with a mean dendritic field size of 0.42 +/- 0.11 mm2. The dendritic field sizes of Types II and III GCs did not increase with retinal eccentricity. Type I GCs were distributed unevenly across the retina, the density being greatest in the visual streak, along the nasotemporal meridian of the retina. The dendritic field sizes of these cells increased towards the retinal periphery, resulting in a constant dendritic field coverage factor across the retina. Each retinal point was covered by the dendritic fields of 4-5 adjacent GCs. In contrast, Types II and III GCs had only discontinuous dendritic coverage. The identification of morphological types of large GCs with previously described functional classes of GCs in the anuran retina is discussed.

Morphology of retinal ganglion cells in lungless salamanders (fam. Plethodontidae): an HRP and Golgi study

The family Plethodontidae consists of nearly two-thirds of all living urodeles; most of them possess highly developed visual abilities. We investigated the morphology of retinal ganglion cells (RGCs) in four representative species by means of the horseradish peroxidase method in flatmounts and in transverse sections and with the Golgi method in transverse sections. In flatmount preparations, four classes of RGCs were found, differing in dendritic arborization, dendritic field size, and stratification pattern of dendrites in the inner plexiform layer (IPL). Class-1 cells had small dendritic fields (29-44 microns 2) and arborized throughout the entire depth of the IPL. Class-2 cells had medium to large dendritic fields (75-206 microns 2) and mostly arborized in two or three laminae or in a diffuse fashion in the IPL. Class-3 cells had medium to large dendritic fields (72-200 microns 2) but sparse dendritic arborization. They only arborized in the proximal lamina of the IPL. Class-4 cells had large dendritic fields (273-626 microns 2) and branched in the most sclerad stratum of the IPL. No large differences in intraspecific soma size of the different RGC classes were detected (although interspecific soma size varied to a considerable degree) and no "giant" cells typically found in other vertebrate retinas were present. The results suggest that, with respect to the pattern of arborization and stratification of dendrites, lungless salamanders possess morphological classes of RGC similar to those found in frogs, but the morphology of RGCs in lungless salamanders seems to be simplified in comparison to frog RGCs. This simplification might be a consequence of paedomorphosis.

The development of retinal ganglion cells deprived of their targets

The influence of central targets on the morphological differentiation of retinal ganglion cells was investigated in Xenopus laevis. Since the ganglion cells mature into distinct morphological subtypes after their axons have reached their central targets, it is possible that the target tissues may influence or specify this aspect of neuronal cell development. To test this idea, Xenopus eyebuds were target-deprived by transplantation to the flank region of host embryos where they developed ectopically. The grafted eyes grew at normal rates, but could not make any projections into the central nervous system. To examine the morphological differentiation of the retinal ganglion cells their structures were revealed using an in vitro retinal preparation and intracellular injections of the dye Lucifer yellow. The elaboration and maturation of ganglion cell dendrites were found to be indistinguishable between control and transplanted eyes throughout development. Thus, the development of retinal ganglion cells into distinct morphological classes can occur even when their axons do not interact with the appropriate central targets.

Retinal ganglion cell death induced by unilateral tectal ablation in Xenopus

The survival of retinal ganglion cells (GCs) in the left eye was studied on retinal wholemounts from 2-33 weeks after the surgical removal of the right tectum in juvenile Xenopus. Two to five weeks after tectal removal, about 76% of neurons of the retinal ganglion cell (GC) layer showed signs of retrograde degeneration: swelling of their somata and chromatolysis. Neurons that were not affected by the operation were taken to be either displaced amacrine cells (DAs) or GCs not projecting to the tectum. A portion of GCs showing retrograde degeneration became pyknotic and died within the period of 2-16 weeks after operation. Counts of surviving GCs 20-33 weeks after tectal removal amounted to about 55% of the corresponding neuron number in the right intact retina of the same animal. No discernible GC loss was observed in animals where only the optic fibers were cut at their entry point to the tectum indicating that axotomy alone, followed by rapid regrowth to the target, does not adversely influence the survival of GCs. In long-surviving animals, the left optic nerve was exposed to cobaltic-lysine complex and the position of filled optic axons within the brain determined. Optic axons whose tectal target had been removed were seen to cross over to the left intact tectum via the posterior and pretectal commissures. Aberrant projections were detected to the ipsilateral tectum and the diencephalic periventricular grey in addition to an increased projection to the accessory optic nucleus. It is concluded that the removal of the tectum, the main target of optic fiber projection, induces a very substantial GC death. Since only a portion of optic fibers were able to grow to alternative targets, the surviving GCs may have also included those with main projection areas to the diencephalic visual centers.

Morphological classification of retinal ganglion cells in adult Xenopus laevis

Retrograde transport of horseradish peroxidase (HRP) was used to characterise the soma and dendritic arborization of retinal ganglion cells in adult Xenopus laevis toad. HRP was administered to the cut end of the optic nerve and the morphological characteristics of HRP-filled ganglion cells were analysed in retinal wholemount preparations using computer assisted morphometry. Ganglion cells were classified according to their soma size, dendritic branching pattern, dendritic field and the number of shaft dendrites. Ganglion cells were divided into 3 major classes on the basis of soma sizes and extent of dendritic field: large (soma size, mean 258.04 micron 2 +/- 52.03 SD; dendritic field size 0.104 mm2 +/- 0.23), medium size (126.7 micron 2 +/- 37.01; 0.041 mm2 +/- 0.013) and small (87.3 micron 2 +/- 22.69; 0.0061 mm2 +/- 0.0035). A more detailed analysis allowed 12 morphologically distinct subgroups to be identified (Types I-XII). Quantitative studies showed that large cells comprise about 1%, medium size about 8-9% and the small cells over 90% of total ganglion cell population. The number of large and medium size ganglion cells corresponded well with the number of myelinated optic fibres and the number of small neurons with the number of unmyelinated optic fibres in the optic nerve. Large ganglion cells were correlated with Class 4 and 5, medium size ganglion cells with Class 3 and small ganglion cells with Class 1 and 2 functionally characterized ganglion cells in the frog retina (Maturana et al. 1960). The retinal distribution of large ganglion cells appear to suggest certain similarities to mammalian alpha type ganglion cells.

Increase in ganglion cell size after optic nerve regeneration in the frog, Rana pipiens

Even though optic regeneration is successful in the frog, Rana pipiens, at completion considerable ganglion cell loss has occurred. To determine whether ganglion cell loss affects the size of the remaining ganglion cells, these cells were back-filled with horseradish peroxidase. The size of one class of ganglion cell 6 months to 1 year following nerve crush injury (N = 4) was compared to that of normal cells of this class (N = 4). The average area of the perikaryon was 35% larger than normal (less than 0.01). This change is interpreted to reflect the increased metabolic needs of the neuron required to maintain a larger than normal axonal arbor.

Retina of the tadpole and frog: delayed dendritic development in a subpopulation of ganglion cells coincident with metamorphosis

In this study, the morphology of tadpole retinal ganglion cells was compared to that of frogs to determine if changes in dendritic structure occur during metamorphosis. Ganglion cells were analyzed in the tadpole and frog after backfilling with horseradish peroxidase. Representative ganglion cells are present in the tadpole retina, which directly correspond to each of the 7 cell classes found in the frog. However, cells in 3 of these classes (1, 3, and 7) exist in morphologically immature states in retinas from tadpole stages St. XIV-XIX. New dendritic branches appear and the dendritic arbors of these ganglion cells expand during metamorphosis. We propose that the increased dendritic arborization may be followed by new synaptic contacts onto these cells, which contributes to the emergence of new physiological receptive field properties in the frog.

Effect of tetraploidy on dendritic branching in neurons and glial cells of the frog, Xenopus laevis

Morphological aspects of four different groups of Golgi impregnated brain cells from a tetraploid strain of Xenopus laevis frogs were compared to analogous cells in comparably sized diploid frogs. The cells examined included neurons from the telencephalon, caudal hypothalamus, and optic tectum, and radial glial cells from the optic tectum. The brains of tetraploid frogs appeared grossly normal and were the same size and contained similar cell types as diploid brains. As observed in previous studies on polyploid amphibia, somal diameters increased significantly in tetraploid cells for each of the four groups of cells examined. Also, the total length of the dendritic arbors in tetraploid brain cells increased significantly by factors ranging from 1.4 to 2.4 times the total length of the analogous processes in diploid cells. Tetraploid neurons in the telencephalon and hypothalamus increased their arbor lengths predominantly by increasing the number of dendritic branches, while maintaining the average distance between branch points in the dendritic segments. In contrast, the tetraploid large pear-shaped neurons in the optic tectum had significantly longer terminal dendritic segments than the analogous diploid neurons, although these tetraploid neurons maintained their average number of dendritic segments per cell. Tetraploid tectal radial glial cells appeared to increase both their number of branches and the lengths of their terminal segments. Thus, the mode by which tetraploid brain cells achieved longer dendritic arbors varied from cell type to cell type. These results suggest a hypothetical basis for possible effects of genomic size on vertebrate brain structure and evolution at the cellular level.

Regeneration of the frog optic nerve. Comparisons with development

Developing and regenerating frog optic axons grow within optic pathways and form connections only with optic targets. However, unlike normal development, many regenerating optic axons in the adult frog are misrouted within optic pathways, including axons that grow into the opposite retina. Many of the axons misrouted during regeneration appear to be collaterals of axons that grow in normal directions. Ganglion cell loss of up to 60% occurs after optic nerve damage, beginning prior to reinnervation of optic targets. Massive axonal collateralization also takes place near the point of nerve damage, causing the normal order found within the nerve to be lost. Collaterals are eliminated as selective reinnervation is completed, and the smaller complement of optic cell axons remaining after regeneration form an expanded projection within optic targets. Evidence is reviewed that suggests that factors involved in axonal guidance and target recognition during development remain intact in the adult frog brain. Additional conditions resulting from nerve injury causes axonal guidance to be less successful during regeneration.

Regulation in the neural plate of Xenopus laevis demonstrated by genetic markers

To follow the subsequent history of grafted tissue in experiments designed to study regulation and commitment in the amphibian neural plate, previous workers have relied on graft scars, vital dyes applied externally to cells, or xenoplastic grafts. Each of these methods has been criticized on the grounds that they do not indicate unambiguously the origins of individual cells within the operated host. To overcome these difficulties, homoplastic, genetically marked embryonic grafts were taken from the prospective spinal neuroectoderm of triploid and tetraploid Xenopus laevis frogs and transplanted to presumptive eye and prosencephalic regions of the neural plate of diploid X. laevis embryos. Orthotopic presumptive eye grafts also were done. Marked cells were scored in section either by nucleolar number or computerized nuclear size analysis. Of 28 heterotopically grafted embryos that survived to stage 41, when the retina has differentiated, prospective spinal cord neuroectoderm in eight animals gave rise to cell types unique to the eye. The remaining 20 survivors appeared to be mosaic. These results substantiate claims of regulation in the neural plate and extend these observations to the level of individual cell types, a level of resolution not previously obtained in other studies.