Fly photoreceptor synapses: their development, evolution, and plasticity

Recent studies are reviewed on the synapses of photoreceptor terminals in the first optic neuropile of the flies, Musca and Drosophila. Afferent synaptic contacts are of uniform dimensions; they have a postsynaptic tetrad with a membrane organization of P-face particles, resembling other inhibitory synapses. A distributed population of such contact sites forms progressively during synaptogenesis by the selective, sequential accretion of identified postsynaptic elements at the receptor terminal. The comparative anatomy of this synapse indicates that elements have also been added during its phylogeny from an ancestral dyad. All cells are homologs of those in other species of Diptera. The number of synaptic sites is regulated by both pre- and postsynaptic cells, in proportion to their cell surfaces; an independent size increase in the receptor terminals (procured in the Drosophila mutant gigas) produces an increase in their synaptic population. The number of sites declines with age, however, accompanied by an increase in size of those synaptic sites remaining; this occurs for both afferent and feedback photoreceptor synapses. Lastly, the number of sites changes with visual experience; the frequency of feedback synapses is larger following dark rearing during early adult life than following visual experience.

Anatomical plasticity of synapses in the lamina of the optic lobe of the fly

Insects are frequently assumed to have hard-wired nervous systems that fail to demonstrate functional plasticity. We have produced changes in synaptic frequency, and analysed their developmental time course, dynamics and reversibility, in the lamina underlying the compound eye of the fly, by exposing young adults to different visual stimuli. The class of synapse examined feeds back from L2, one of the monopolar cells found in each lamina cartridge, to photoreceptor terminals; each site is a synaptic dyad marked by the presence of a few, round vesicles surrounding a T-shaped presynaptic ribbon and, in the photoreceptor, by a subsynaptic vacuole. In control adult flies reared in normal room lighting, the frequency of synaptic profiles scored in micrographs of single sections initially increased until one day post-eclosion (E + 1), but declined thereafter. Frequencies measured in left and right eyes of the same control animals were closely matched. Experimental flies were put for one to two days into an integrating sphere illuminated continuously with square-wave, 25 Hz green light. They had one eye occluded, so providing control comparisons between flicker-reared (FR) and occluded (dark-reared, DR) eyes within the same animal. The DR eyes invariably (n greater than 22) had higher frequencies of synaptic profiles than those seeing light, regardless of age or the period of light exposure, although the detailed relative effects of FR and DR depend upon the age of the animal. The evidence suggests that exposure to light actively depresses synaptic frequency and increases its variability. The greatest difference (30%) achieved was at two to four days after eclosion and there was no difference beyond six days, so demarcating a prospective sensitive period. Rearing in DC light was equally effective as FR, so visual contrasts per se are apparently inessential. Frequency values can change rapidly. During the first 24 h post-eclosion, DR resulted in new synapses adding to L2's complement of 25-35 at a maximum rate of 4 per 6 h, whereas light exposure caused a frequency decrease after as little as 6 h. Alternating 24 h periods of light and dark during the first two days produced reversible synaptic frequency changes. Individual synaptic contacts enlarge with age but not significantly with different visual experiences. The decrease in frequency of synaptic profiles with age thus actually underestimates the true decrease in synaptic number, whereas the altered synaptic frequencies seen after differential exposure represent true differences in synaptic number.(ABSTRACT TRUNCATED AT 400 WORDS)

Development of the central nervous system of the larva of the ascidian, Ciona intestinalis L. I. The early lineages of the neural plate

The early lineages of the larval central nervous system (CNS) of the ascidian, Ciona intestinalis, have been traced using scanning electron microscopy (SEM) of embryos fixed at 12-min intervals. The CNS precursors lie superficially, exposed for a long portion (9.3 hr of 42%) of embryonic development, in the neural plate. In the 64-cell stage embryo the neural plate contains 10 cells; in all but the first vegetal division these divide with transverse cleavage planes. Synchrony is progressively lessened, but temporal sequence is always exact. Successive divisions occur initially at 30-min intervals. Our analysis confirms existing lineage descriptions for the neural plate up to the end of gastrulation and advances the lineage record through the crucial and temporally complex ninth cleavage, during which cells divide by the following rules: medial cells in each row divide first; the anterior row of vegetal daughter cells divides before their posterior siblings; the posterior row of animal daughter cells divide before their anterior siblings. All cells attain their 10th generation, but four cannot be followed by SEM. In preparation for neurulation the neural plate then comprises 76 cells, forming up to four rows each of eight vegetal hemisphere cells located on the dorsal surface of the embryo, anterior to the blastopore, and eight rows each of six animal hemisphere cells, located anterior to the rows of eight. The temporal and spatial patterns of early cleavage stages have been confirmed in vivo by observations using Nomarski optics.

Development of the central nervous system of the larva of the ascidian, Ciona intestinalis L. II. Neural plate morphogenesis and cell lineages during neurulation

We describe the lineage and morphogenesis of neural plate cells in the ascidian, Ciona intestinalis, from reconstructed cell maps of embryos at 12-min intervals during and after neurulation, between 31 and 61% of embryonic development. Neurulation commences in a posterior to anterior wave following in the wake of the ninth cleavage, when all cells, except possibly four, are in their 10th generation. The neural plate then comprises 76 cells, in up to four posterior rows each of eight vegetal-hemisphere cells, and eight anterior rows each of six animal-hemisphere cells. Two cells are lost from the neural plate to the muscle cell line during neurulation and four cells are gained from ectoderm outside the plate. All cells become wedge-shaped. Simple, stereotyped positional changes transform cells from lateral locations in the plate to posterior locations in the tube; bilateral partners shear their midline positions to form the keel, and ectodermal cells zipper up dorsally to form the capstone, of a tube which is four cells in cross section posteriorly, but more complex anteriorly. Neither cell death nor migration occur during neurulation. Divisions become asynchronous and the cell-cycle extends; 170 10th- to 12th-generation cells exist by the time the neural tube becomes completely internalized. Generally, only one further division is required to complete the lineage analysis, two at the most. Neural plate cell divisions were invariant using our observational methods, and their lineage is compared with that from recent studies of H. Nishida (1987, Dev. Biol. 121, 526-541).

Regulation of synaptic frequency: comparison of the effects of hypoinnervation with those of hyperinnervation in the fly's compound eye

At the anterior rim of the first optic neuropile, or lamina, of the housefly's (Musca domestica) compound eye, the terminals of photoreceptors (R) innervate postsynaptic neurons in variable numbers to provide a continuous range of natural hypo- and hyperinnervations. Frequencies of photoreceptor synapses have been measured from quantitative electron microscopy on single sections of the lamina's unit synaptic modules, called cartridges. These are normally innervated by six photoreceptor terminals (6R cartridges). At the lamina's edge hypoinnervated cartridges (2R-5R) are found, whereas hyperinnervated cartridges (7R, 8R) are located at the equator between dorsal and ventral eye halves. In 2R cartridges each presynaptic terminal forms up to 1.5 times the normal, 6R cartridge number of synapses, thereby offsetting the reduced number of terminals and partially conserving the input upon the postsynaptic neurons. Thus the terminals have a reserve synaptogenic capacity never normally revealed. By comparison, terminals in 8R cartridges form about the same numbers of synapses as in "normal" eye regions, so that their postsynaptic neurons have a synaptic input increased by the extra number of terminals. The number of synapses formed between input terminals and target neurons is therefore not fixed but changes as a function of the total receptor terminal complement. The size of a photoreceptor terminal covaries to a certain extent with the number of its presynaptic sites; the spacing density of presynaptic sites over the terminals' surface in a 2R cartridge compared with an 8R cartridge increases far less (only 17%) than the increase in the number of sites (43%). The pair of postsynaptic cell interneurons in each 2R cartridge also shows a decrease in axonal diameter compared with those in 8R cartridges. Thus both the pre- and postsynaptic cells show size changes correlated with changes in their synaptic engagement.

Evolutionary progression at synaptic connections made by identified homologous neurones

A comparative ultrastructural study of photoreceptor synapses formed upon homologous postsynaptic neurones in insects has been made by using serial-section electron microscopy in representative Diptera from a monophyletic series of 14 families. At all of the synaptic contacts there is a presynaptic dense bar, surmounted in phylogenetically more recent families by a presynaptic platform. Opposite the bar lies a pair of postsynaptic elements that invariably originate one each from two unique monopolar neurones L1 and L2. Both elements contain increasingly elaborate cisternae in more recent flies. Within the phylogenetic series, the postsynaptic ensemble itself changes from the original dyad to a tetradic configuration in more recent Muscomorpha by the addition of two new postsynaptic elements from an amacrine cell. This transition occurs once only in the series, which, gauged by the fossil record, covers divergences from the stem line extending back greater than 200 million years.

Anti-adhesive coating for glass microelectrodes

A simple procedure is described for coating the tips of glass micropipette electrodes with a durable thin film of a commercially available anti-adhesive agent (Antispread). This lowers the surface tension of the electrode glass, and helps reduce damage and distortion to tissue, by preventing the microelectrode from sticking permanently to cell membranes. In controlled trials, the coating actually increased the probability that nearby cells could be recorded from successively by about 30%. Electrode resistance increased only very slightly (on average by 8%) as a result of the coating. Coated electrodes yielded excellent long-term penetrations of small visual cells, and in addition allowed these to be dye-filled in the normal way. The coating may have applications in all situations where prevention of wetting by tissue is desirable.

The long visual fibers of the dragonfly optic lobe: their cells of origin and lamina connections

A pair of long visual fibers projects from each ommatidium of the compound eye of the dragonfly Sympetrum. They arise from retinular cell R7 and its slender partner R6. To investigate the segregation of lamina input pathways between these and the well-described short retinular terminals, we have examined the cell morphology and en passant synaptic involvements of R7 and 6 in the ventral lamina using serial-EM and combined Golgi-EM. In the retinula, R7 has a large apical rhabdomere with microvilli aligned along the animal's horizon, and a basal axon. R6 has a few microvilli at all depths, which align with those of neighboring rhabdomeres. In the lamina, R7 and 6 neighbor on MV, the fifth monopolar cell. The stout axon of R7 has many diffusely distributed spines which contact the neighboring terminals of R5 and 8, and MII. In one entire cartridge R7 was presynaptic to MII at 24 dyads and triads throughout the lamina and also to MV at dyads exclusively in the distal lamina. Another class of element which derives from the unidentified processes called alpha was also postsynaptic to R7 at dyads, and, mostly in the proximal lamina, was reciprocally presynaptic to R7, along with R5 and 8. R6 is slender with a single conspicuous spine. At a total of three zones in the proximal lamina and distal chiasma it was presynaptic at 38 dyads--also upon MII and MV and often in combination with other elements; it was not postsynaptic in the lamina. R6 and 7 thus provide the sole input to MV and contribute to the general retinular input to MII. They form the lowest ratios of triad:dyad synapses of all lamina retinular elements. Comparison is made with the long visual fibers of other arthropods but known examples are too diverse to detect functional commonalities.

Quantitative features of synapse formation in the fly's visual system. I. The presynaptic photoreceptor terminal

Photoreceptors of the adult fly's compound eye each form a population of stereotyped output synapses distributed over the surface of their terminal. The formation of this class of afferent synapses during development has been followed from serial electron microscopy of the same eye region in four pupal and several adult stages, all of female Musca domestica. These synapses, or tetrads, have an invariant postsynaptic composition of four members and so may provide a model for multiple-contact synapses in general. In the adult fly the four postsynaptic elements of each synapse are contributed by two interneurons, L1 and L2, and, usually, two alpha processes of an amacrine cell. These postsynaptic elements assemble at individual developing synapses by selective sequential addition. Assembly starts with L1 or L2, subsequent elements of the final tetrad adding in all conceivable permutations, at least as fast as one per 7 hr. They rarely (only once) incorporate incorrect or supernumerary elements, however. The synaptic population as a whole was also sampled during development to analyze the possible factors determining the normal precision of the size of the adult population. The number of synapses per terminal increases gradually until 74% pupal development. Thereafter it decreases so that the final number of synapses in each receptor's population is the consequence of a net loss. Synapses enlarge with age, chiefly by incorporating new elements, but the loss of synaptic sites is only partially offset by the increase in size of those that remain. Throughout all stages examined in pupal and adult life, total synaptic area is linearly proportional to the surface area of the axon terminal. Thus, from the 74% pupal development stage onward, a population of many small synapses closely spaced, on average, over the terminal's surface transforms into one characteristic of the adult with progressively fewer, larger, more widely spaced synapses.

An analysis of the number and composition of the synaptic populations formed by photoreceptors of the fly

The photoreceptors terminals of newly enclosed female flies, Musca domestica, have been sampled in the first optic neuropile (or lamina) in one of two ways: first, in large number (n = 760) from single sections and second, from serial electron micrographs of the six terminals within each of three cartridges. Both sampling methods concur in assessing the number of synapses established with the two principal monopolar relay interneurons, L1 and L2, within each cartridge. Each receptor is calculated to be presynaptic at about 200 +/- 40 (2 SE) synapses. This value considerably exceeds previous estimates, primarily because we took careful account of the appearance of synapses in different section planes. The number of these synapses correlates highly with the area of receptor terminal presynaptic membrane, so that each synapse is allotted, on average, about 1.6 micrometer2. The synapses are evenly graded in their distribution with an unexplained 23% decrease in both membrane perimeter and synapse number halfway along their receptor terminal's length. The numbers of synapses per receptor did not vary systematically within two horizontal (3 X 20 cartridge) strips of frontal, equatorial lamina sampled. Individual synapses are elongate tetrads (Burkhardt and Braitenberg, '76) with two pairs of postsynaptic elements. The first pair is invariably contributed by the interneurons L1 and L2 (one each). The second pair comes either from the alpha processes of an amacrine cell or from a glial cell. In the distal lamina, however, L3 contributes one of the two postsynaptic processes, the second being alpha or glial. The overall ratio of postsynaptic involvement at distal synapses (alpha: glial: L3) is 55%, 20%, and 12% respectively, the remainder being unidentified.

Regulation in the number of fly photoreceptor synapses: the effects of alterations in the number of presynaptic cells

At the equator of the fly's eye, between dorsal and ventral eye halves, a systematic, natural addition of photoreceptor terminal input occurs at each of the fixed populations of uniquely identifiable postsynaptic interneurons in each cartridge of the first optic neuropile, or lamina. The equatorial cartridges are identical in composition except in having seven and eight receptor terminals (7R and 8R, respectively), compared with six elsewhere (6R cartridges). The effects of this augmented presynaptic input upon the frequency of the chief afferent class of photoreceptor tetrad synapse were studied compared with control data for 6R cartridges (Nicol and Meinertzhagen, '82). The synapse population size and distribution within five depth levels of the lamina is, on average, approximately constant for all receptor terminals whether from 6R, 7R, or 8R cartridges. The overall determinant of synapse frequency is therefore presynaptic. Small (5-6%) average decreases in synapse frequency per receptor in 7R and 8R cartridges compared with 6R co-vary with similar decreases in membrane area, each synapse occupying a patch of membrane of similar area in all cases. The tetradic postsynaptic composition of synapses was also similar in all cases. Because of the augmented synaptic input to the postsynaptic neurons, a morphometric analysis was undertaken of two (L1 and L2), which receive input as a pair from every synapse. There is the same dendrite number (about 180) in 8R L1/L2 as in 6R L1/L2 but they have different branching patterns, conforming to the different number and configuration of receptor terminals. Thus in an 8R cartridge each terminal is serviced by a comb of fewer dendrites, but each dendrite is longer, fatter, and services more synapses. The area of L1/L2 dendritic membrane exposed is increased, compared with 6R cartridges, in proportion with the number of synapses it participates at postsynaptically, so that all dendrites (6R and 8R) allocate the same mean area (about 0.55 micrometer2) of postsynaptic membrane per synapse.

Synaptogenesis in the first optic neuropile of the fly's visual system

The developmental transformation of the chief afferent class of fly photoreceptor synapse has been examined from serial electron micrographs of animals fixed at 74, 81, 94 and 100% pupal development (100% pupal development being defined by the time of normal adult eclosion). Animals were selected both by their age and the conformity of their eye coloration to standards for each stage. Two animals were analysed from each stage, one in greater detail than the other. For the first, the exact coordinates of the cartridges (the synaptic columns of the first optic neuropile) from which the analyses were made were mapped and selected to be within the same region of the eye field at all stages. From all animals a portion of one or two cartridges was analysed from series of up to 100 sections and the synapse populations (greater than 80) were analysed for their fine structure and postsynaptic composition. Adult synapses are confirmed as tetrads, with two of the four postsynaptic elements invariably from two monopolar interneurons L1 and L2, one from each. The two others are usually from alpha processes of the same amacrine cell. Synapses appear during the last half of pupal development, with no obvious asynchrony of ultrastructural maturation and in parallel with those of at least one of the other synaptic classes present (which were otherwise not studied). Many adult features of synaptic ultrastructure emerge late, only by 94% pupal development. These include adult numbers of synaptic vesicles, the complete form of the presynaptic ribbon with platform and the postsynaptic cisternae of L1/L2. Prior to 94% the synapses are smaller with postsynaptic elements having a less regular geometry and with postsynaptic densities which are subsequently lost (alpha processes) or replaced by cisternae (L1/L2). At the presynaptic sites of the younger animals (74%, 81%) dyads and triads of postsynaptic elements coexist with tetrads, those of older animals having, on average, more postsynaptic processes per synapse. It is suggested that individual synapses assemble piecemeal, element by element.

The number and arrangement of elements in the lamina cartridge of the dragonfly Sympetrum rubicundulum

Five monopolar cells and two long visual fibres are a consistent component of the lamina cartridge of the ventral half of the eye of the dragonfly Sympetrum rubicundulum. They communicate with the chiasm via a cartridge axon bundle comprising a minimum of ten fibres. The arrangement of these elements is documented with respect to the ommatidial photoreceptor axon bundle innervating them. These relationship are described both within the lamina cortex and in the cross-section of the underlying cartridge.

The organization of perpendicular fibre pathways in the insect optic lobe

High resolution serial photomicrography has been used to plot the axonal projection patterns between retina, lamina and medulla in the optic lobes of various insects with differing ommatidial receptor arrangements. Observations are reported on the cabbage white and skipper butterflies, the bee, locust, fly, backswimmer and waterbug. The patterns of these fibre pathways have previously eluded non-rigorous analyses primarily because of their physical dimensions but are revealed in this study to have striking precision and uniformity between species when examined at the level of individually identifiable cells. Axon bundles of the tracts between retina and lamina or lamina and medulla project between a single ommatidium and its corresponding lamina cartridge or between corresponding lamina and medulla cartridges. Lateral interweaving of axons between adjacent bundles is absent. The bundles preserve the retinotopic order within their total array, so transferring the pattern of retinulae directly upon the lamina and thence after horizontal inversion in the chiasma upon the medulla. Within the lamina neuropile on the other hand the trajectories of the individual terminals from a bundle have patterns which are species-specific, sometimes involving lateral divergences. In species with open-rhabdomere ommatidia the terminals distribute to a group of lamina cartidges with a pattern which resembles the receptor pattern in the overlying ommatidium. In species with fused-rhabdome ommatidia the terminals of a single retinula behave less interestingly and all enter the same cartridge, within which, again, each occupies a position related to its cell body position within the retinula. Long visual fibres in both eye types penetrate the lamina and terminate in the particular medulla cartridge that connects with the lamina cartridge underlying their ommatidium. The perpendicular fibre pathways therefore project the visual field exactly upon the medulla in all species while the lack of interweaving between adjacent fibre bundles precludes their involvement in lateral interactions between pathways with differing visual axes. Uniformity of these projection patterns between cell layers and species differences in retinular terminal locations in the lamina can be correlated with different modes of axon growth between and within neuropile layers during optic lobe neurogenesis. Further discussion surrounds the question of which particular receptors give rise to which type of axon, for which no clear generalization has yet emerged.

The accuracy of the patterns of connexions of the first- and second-order neurons of the visual system of Calliphora

The axons of the primary photoreceptor cells of the compound eye of the fly interweave in a complex but definite pattern before they terminate upon the second-order neurons. Of approximately 650 short retinula axons from behind 120 facets of the eye none terminated at an incorrect lamina cartridge. Six, seven, or eight first-order terminals upon one pair of second-order cells are arranged in a rotational sequence that is related to the positions of the retinula cells within the ommatidia. Errors in location of the terminal among its neighbours occurred only ten times. The asymmetry of the receptor pattern in the dorsal half of the eye has a mirror image in the ventral half. Along the equator of the eye is a plane of symmetry which many axons necessarily cross in maintaining the appropriate connexions of their receptors. Axons which cross this plane of symmetry have somehow found their appropriate second-order cells, although to do so they must have grown through a milieu which is the mirror image of that in their own half of the eye. Each pair of second-order axons proceeding from the lamina forms a small bundle with the axons of the two long retinula cells that have the same visual axis. Between the lamina and the medulla is a chiasma (with the crossing in the horizontal plane) through which bundles from the lamina pass to project in exactly reverse order upon the medulla. No errors of projection have been found at the single neuron level in this chiasma.