Regenerated retinal ganglion cell axons form normal numbers of boutons but fail to expand their arbors in the superior colliculus

Regenerated retinal ganglion cell (RGC) axons can re-form functional synapses with target neurons in the superior colliculus (SC). Because preterminal axon branching determines the size, shape and density of innervation fields, we investigated the branching patterns and bouton formation of individual RGC axons that had regrown along peripheral nerve (PN) grafts to the SC. Within the superficial layers of the SC, the regenerated axons formed terminal arbors with average numbers of terminal boutons that were similar to the controls. However, axonal branches were shorter than normal so that the mean area of the regenerated arbors was nearly one-tenth that of control arbors and the resulting fields of innervation contained greater than normal numbers of synapses concentrated in small areas of the target. Our results have delineated a critical defect in the reconstitution of retino-collicular circuitry in adult mammals: the failure of terminal RGC branches to expand appropriately. Because recent studies have documented that brain-derived neurotrophic factor (BDNF) can specifically lengthen RGC axonal branches not only during development in the SC but also within the adult retina after axotomy, the present quantitative studies should facilitate experimental attempts to correct this deficit of the regenerative response. © 1998 Chapman and Hall     

Axonal sprouts of the hypoglossal nerve implanted in the superior cervical ganglion of adult rats establish synaptic contacts under long-lasting GABA effect. An experimental degeneration study

Experimental degeneration was used in this study to determine if the hypoglossal nerve implanted already in the superior cervical ganglion of adult rat under GABA treatment has established morphologically-identifiable synapses with the dendrites of principal ganglion cells. The implanted hypoglossal nerve trunk was cut in a re-operation, and the ganglionic samples were studied by electron microscopy after 0, 6, 12, 24 and 48 h survival times. First signs of degenerative changes were found in the myelinated and non-myelinated axons alike, 6 h after axotomy. The fine-structural signs of degeneration resembled those of the preganglionic nerve fibres. Degenerating nerve terminals establishing synaptic contacts with the dendrites of the principal ganglion cells were also seen, indicating that the axonal sprouts of the implanted hypoglossal nerve established synaptic contacts with the ganglion cells. It remained, however, to be elucidated whether or not these synapses of the hypoglossal nerve are functionally active contacts while the preganglionic innervation is also present within the ganglion.     

Alterations in the morphology of ganglion cell dendrites in the adult rat retina after optic nerve transection and grafting of peripheral nerve segments

Summary Transected ganglion cell axons from the adult retina are capable of reinnervating their central targets by growing into transplanted peripheral nerve (PN) segments. Injury of the optic nerve causes various metabolic and morphological changes in the retinal ganglion cell (RGC) perikarya and in the dendrites. The present work examined the dendritic trees of those ganglion cells surviving axotomy and of those whose severed axons re-elongated in PN grafts to reach either the superior colliculus (SC), transplanted SC, or transplanted autologous thigh muscle. The elaboration of the dendritic trees was visualized by means of the strongly fluorescent carbocyanine dye DiI, which is taken up by axons and transported to the cell bodies and from there to the dendritic branches. Alternatively, retinofugal axons regrowing through PN grafts were anterogradely filled from the eye cup with rhodamine B-isothiocyanate. The transection of the optic nerve resulted in characteristic changes in the ganglion cell dendrites, particularly in the degeneration of most of the terminal and preterminal dendritic branches. This occurred within the first 1 to 2 weeks following axotomy. The different types of ganglion cells appear to vary in their sensitivity to axotomy, as reflected by a rapid degeneration of certain cell dendrites after severance of the optic nerve. The most vulnerable cells were those with small perikarya and small dendritic fields (type II), whereas larger cells with larger dendritic fields (type I and III) were slower to respond and less dramatically affected. Regrowth of the lesioned axons in peripheral nerve grafts and reconnection of the retina with various tissues did not result in a significant immediate recovery of ganglion cell dendrites, although it did prevent some axotomized cells from further progression toward posttraumatic cell death.     

Injury-associated induction of GAP-43 expression displays axon branch specificity in rat dorsal root ganglion neurons

Peripheral nerve injury results in the increased synthesis and axonal trasnport of the growth-associated protein GAP-43 in dorsal root ganglion (DRG) neurons, coincident with regenerative growth of the injured peripheral axon branches. To determine wheter the injury-associated signalling mechanism which leads to GAP-43 induction also operates through the central branches of DRG axons, we used immunocytochemistry to compare the expression of GAP-43 in adult rat DRG neurons 2 weeks after dorsal root crush lesions (central axotomy) or peripheral nerve crush lesions (peripheral axotomy). In uninjured ganglia, a subpopulation of smaller DRG neurons expresses moderate levels of GAP-43, whereas larger neurons generally do not. At 2 weeks following peripheral axotomy, virtually all axotomized neurons, large and small, express high levels of GAP-43. At 2 weeks following dorsal root lesions, no increase in GAP-43 expression is detected. Thus, the injury-associated up-regulation of GAP-43 expression in DRG neurons is triggered by a mechanism that is responsive to injury of only the peripheral, and not the central, axon branches. These findings support the hypothesis that GAP-43 induction in DRG neurons is caused by disconnection from peripheral target tissue, not by axon injury per se. © 1993 John Wiley & Sons, Inc.     

Connectivity of identified central synapses in the cricket is normal following regeneration and blockade of presynaptic activity

Cercal sensory neurons in the cricket innervate interneurons in the central nervous system (CNS) and provide a model system for studying the formation of central synapses. When axons of the sensory neurons were transected during larval development, the cell bodies and the soma-bearing portion of axons, which are located within the cercus, survived but lost their excitability for 9-10 days. During this period, the sensory neurons grew new axons and reinnervated the terminal abdominal ganglion. Physiological recordings showed that sensory neurons of known identity reestablished monosynaptic contacts with their normal postsynaptic interneuron. Moreover, each synapse exhibited a characteristic strength indistinguishable from the intact synapse in an unoperated cricket. Since this selective connectivity was apparent immediately after the excitability of the axotomized sensory neurons was restored, action potentials in the sensory neurons appear to be unnecessary for normal synaptic regeneration to occur. Consistent with this, the reinnervation process was unaffected even when action potentials in the sensory neurons were blocked by tetrodotoxin (TTX) immediately following axotomy until just before testing. During the normal course of development, the characteristic strength of individual synapses changes systematically, resulting in the developmental rearrangement of these synapses (Chiba et al., 1988). This synaptic rearrangement was also unaffected when action potentials in the sensory neurons were blocked by TTX for the last 30% of larval development. Therefore, in the cricket cercal sensory system, both regeneration of the central synapses following axotomy of the presynaptic sensory neurons and the normal rearrangement of connectivity during larval development appear not to require axonal action potentials.     

Synaptic patterns in the dorsal lateral geniculate nucleus of the monkey

Summary Synaptic junctions are found in all parts of the nucleus, being almost as densely distributed between cell laminae as within these laminae.In addition to the six classical cell laminae, two thin intercalated laminae have been found which lie on each side of lamina 1. These laminae contain small neurons embedded in a zone of small neural processes and many axo-axonal synapses occur there.Three types of axon form synapses in all cell laminae and have been called RLP, RSD and F axons. RLP axons have large terminals which contain loosely packed round synaptic vesicles, RSD axons have small terminals which contain closely packed round vesicles and F axons have terminals intermediate in size containing many flattened vesicles.RLP axons are identified as retinogeniculate fibers. Their terminals are confined to the cell laminae, where they form filamentous contacts upon large dendrites and asymmetrical regular synaptic contacts (with a thin postsynaptic opacity) upon large dendrites and F axons. RSD axons terminate within the cellular laminae and also between them. They form asymmetrical regular synaptic contacts on small dendrites and on F axons. F axons, which also occur throughout the nucleus, form symmetrical regular contacts upon all portions of the geniculate neurons and with other F axons. At axo-axonal junctions the F axon is always postsynaptic.Supported by Grant R 01 NB 06662 from the USPHS and by funds of the Neurological Sciences Group of the Medical Research Council of Canada. Most of the observations were made while R. W. Guillery was a visiting professor in the Department of Physiology at the University of Montreal. We thank the Department of Physiology for their support and Mr. K. Watkins, Mrs. E. Langer and Mrs. B. Yelk for their skillful technical assistance.     

Fate of ganglionic synapses and ganglion cell axons during normal and induced cell death

In order to understand the significance of cell death in the formation of neural circuits, it is necessary to determine whether before cell death neurons have (a) sent axons to the periphery; (b) reached the proper target organs; and (c) have established synaptic connections with them. Axon counts demonstrated that, after sending out initial axons, ciliary cells sprouted numerous collaterals at the time of peripheral synapse formation. Subsequently, large numbers of axons were lost from the nerves, slightly later than the onset of ganglion cell death. A secondary loss of collaterals later occurred unaccompanied by cell death. Measurements of conduction velocity and axon diameters indicated that all ganglion cell axons grew down the proper pathways from the start, but it was not possible to determine whether all axons had actually formed proper synapses. This was ascertained, however, in the ganglion itself where preganglionic fibres were shown to synapse selectively with all ganglion cells before cell death. During this period, degenerating preganglionic synapses were observed on normal cells. It can therefore be inferred that at least some preganglionics established proper synapses before dying and that a single synapse is not sufficient to prevent cell death. In this system neither preganglionic nor ganglionic cell death seems designed to remove improper connections but rather to remove cells that have not competed effectively for a sufficient number of synapses, resulting in a quantitative matching up of neuron numbers.     

Cardioacceleratory Neurons of the Isopod Crustacean, Ligia exotica: Visualization of Peripheral Projection onto the Heart Muscle

Innervation of the heart muscle by the cardioacceleratory neurons was morphologically and electrophysiologically examined in the isopod crustacean, Ligia exotica. Intracellular injection of neurobiotin into the first and second cardioacceleratory neurons (CA1 and CA2) revealed their peripheral axonal projections. Inside the heart, the CA1 and CA2 axons ran along the trunk of the cardiac ganglion. Finely arborized branches with many varicosities arose from the axon and projected over the heart muscle. Stimulation of either the CA1 or CA2 axon caused an overall depolarization in the muscle of a quiescent heart. The amplitude of the depolarization increased with increasing stimulus frequency. During stimulation, the membrane resistance of the heart muscle decreased. In a beating heart, the cardioacceleratory nerve stimulation caused multiple effects on the heart muscle activity and the heartbeat. The results suggest that the cardioacceleratory neurons of Ligia exotica regulate the amplitude of the heartbeat (inotropic effect) and the heart tonus (tonotropic effect) via the synaptic contacts on the heart muscle, while the heartbeat frequency (chronotropic effect) is regulated via the synapses on the cardiac ganglion neurons.     

Disruption of peripheral target contact influences the development of identified central dendritic branches in a leech motor neuron in vivo

Retrograde signaling from target tissues has been shown to influence many aspects of neuronal development in a number of developmental systems. In these experiments using embryonic leeches (Hirudo medicinalis), we examined how depriving a neuron of contact with its peripheral target affects the development of the cell's central arborization. We focused our attention on the motor neuron cell 3, which normally stimulates dorsal longitudinal muscle fibers to contract. At different locations in the periphery and in embryos of several different stages, we cut the nerve containing the growing axon of cell 3. This surgery led to dramatic overgrowth of cell 3's central dendritic branches, which normally accept synaptic contacts from other neurons, including the inhibitory motor neuron cell 1. When cell 3's peripheral axon was cut relatively early in development, its overgrown central branches eventually retracted. However, cells that were disrupted later in development retained their overextended branches into adulthood. In addition, if the axon was cut close to the ganglion early in development, depriving the cell of contact with any dorsal tissues, the central branches failed to retract and were instead retained into adulthood. Unlike cell 3, the central branches of cell 1, which has the same peripheral target muscles as cell 3, remained unchanged following all axotomy protocols. These results suggest that in at least some neurons contact with peripheral targets can influence development of the central processes that normally mediate synaptic contacts.     

Development of neurons in the abdominal ganglion of Aplysia californica. I. Axosomatic synaptic contacts.

The development of mariculture techniques for the raising of Aplysia californica in the laboratory from fertilized egg to reproductively mature adult permits the study of the developmental program whereby individual identified neurons in the abdominal ganglion acquire their specific adult properties. In this paper, we describe one of the early steps of this developmental program: the outgrowth of axonal processes by neurons of the abdominal ganglion. Axonal outgrowth is correlated with and may be triggered by the transient appearance of morphologically identifiable axosomatic contacts between the as yet undifferentiated cell body of specific neurons and an axon terminal from an incoming nerve fiber from the pleuroabdominal connective. The evidence that transient axosomatic contacts may signal neuronal differentiation is the following: (1) Axosomatic contacts have not been observed in the abdominal ganglion of adult animals, whereas they are commonly observed during the early stages of development. (2) Cells that receive axosomatic contacts are undifferentiated morphologically and do not as yet have axons. By contrast, cells with axons do not have soma contacts. (3) Individual cells that can be identified from animal to animal in the same and succeeding developmental stages receive axosomatic contacts on similar topographic postions of the cell body at one point in development. Axon outgrowth then occurs at the site of contact. Later in development, with further axon extension, these cells no longer have synaptic contacts on the cell body or axon.     

The oculomotor system of Daphnia magna

Summary The highly mobile cyclopic compound eye of Daphnia magna is rotated by six muscles arranged as three bilateral pairs. The three muscles on each side of the head share a common origin on the carapace and insert dorsally, laterally and ventrally on the eye. The dorsal and ventral muscles are each composed of two muscle fibers and the lateral muscle is composed of from two to five fibers, with three the most common number. Individual muscle fibers are spindle-shaped mononucleated cells with organized bundles of myofilaments. Lateral eye-muscle fibers are thinner than those of the other muscles but are otherwise similar in ultrastructure. Two motor neurons innervate each dorsal and each ventral muscle and one motor neuron innervates each lateral muscle. The cell bodies of the motor neurons are situated dorsally in the supraesophageal ganglion (SEG) and are ipsilateral to the muscles they innervate. The dendritic fields of the dorsal-muscle motor neurons are ipsilateral to their cell bodies; those of the ventral-muscle motor neurons are bilateral though predominantly contralateral. The central projections of the lateral-muscle motor neurons are unknown. In the dorsal and ventral muscles one motor axon synapses principally with one muscle fiber; in each lateral muscle the single motor axon branches to, and forms synapses with, all the fibers. The neuromuscular junctions, characterized by pre- and postsynaptic densities and clear vesicles, are similar in all the eye muscles.     

Cell-Cell Recognition in the Regenerating Neuromuscular System of the Cockroach

SYNOPSIS. The neuromuscular system of the cockroach containsmotor neurons and muscles that can be identified in all individualinsects When the axons of these motor neurons are damaged theyregenerate and eventually reform synapses only with the originaltarget muscles However at early times after axotomy transientinappropriate functional connections are made between regeneratingneurons and muscles that theynever normally innervate Laterthe inappropriate synapses are inactivated, the inappropriateaxon branches eliminated and the original innervation patternreformed A cellcell recognition between identified motor neuronsand muscles is required to explain these observations, particularlyin light of experiments demonstrating the absence of competitionbetween appropriate and inappropriate axon terminals withinthe muscle. A minimum biochemical requirement of such a cell-cell recognitionis the existence of molecules whose presence in muscles correlateswith the innervation by identified motor neurons Using fluoresceinlabelled plant lectins to detect muscle surface glycoproteinssuch molecules have been identified In addition, there shouldbe molecular differences among the surfaces of the axon terminalsof the various identified motor neurons Hybrid oma techniqueshave enabled us to obtain monoclonal antibodies that bind tosurfaces of axon terminals of some motor neurons and not othersThese lectin receptors and antigens are good candidate recognitionmacromolecules Other molecules essential for axonal regenerationhave been identified by their presence in embryonic and adultregenerating neurons and their absence from intact adult neurons.     

Regulation of axon arborization pattern in the developing chick ciliary ganglion: Possible involvement of caspase 3

During a certain critical period in the development of the central and peripheral nervous systems, axonal branches and synapses are massively reorganized to form mature connections. In this process, neurons search their appropriate targets, expanding and/or retracting their axons. Recent work suggested that the caspase superfamily regulates the axon morphology. Here, we tested the hypothesis that caspase 3, which is one of the major executioners in apoptotic cell death, is involved in regulating the axon arborization. The embryonic chicken ciliary ganglion was used as a model system of synapse reorganization. A dominant negative mutant of caspase‐3 precursor (C3DN) was made and overexpressed in presynaptic neurons in the midbrain to interfere with the intrinsic caspase‐3 activity using an in ovo electroporation method. The axon arborization pattern was 3‐dimensionally and quantitatively analyzed in the ciliary ganglion. The overexpression of C3DN significantly reduced the number of branching points, the branch order and the complexity index, whereas it significantly elongated the terminal branches at E6. It also increased the internodal distance significantly at E8. But, these effects were negligible at E10 or later. During E6–8, there appeared to be a dynamic balance in the axon arborization pattern between the “targeting” mode, which is accompanied by elongation of terminal branches and the pruning of collateral branches, and the “pathfinding” mode, which is accompanied by the retraction of terminal branches and the sprouting of new collateral branches. The local and transient activation of caspase 3 could direct the balance towards the pathfinding mode.     

Types of neurons and synaptic connections at hypostome-tentacle junctions in Hydra

Using transmission electron microscopy of thin sections we have examined neuronal concentrations at hypostome-tentacle junctions in Hydra littoralis. A total of 194 ganglion cells were counted in 587 serial thin sections of a single hypostome-tentacle junction. We found two distinct types of ganglion cells: those with and those lacking stereocilia. The majority of the neurons observed lacked stereocilia; in a single hypostome-tentacle junction only 37% of the ganglion cells possessed a kinocilium surrounded by rodlike stereocilia. Most of the ganglion cells (55%) were clustered together in the oral or upper epidermis of the hypostome-tentacle junction: Nineteen percent were in the lateral and 26% in the aboral or lower epidermis. The two types of ganglion cells did not differ significantly in their distribution. Both types of ganglion cell had synaptic contacts with other neurons and with epitheliomuscular cells. More than 85% of the neuroneuronal and 61% of the neuroepitheliomuscular cell synapses were located in the oral epidermis of a hypostome-tentacle junction. In addition, two-way chemical synapses and a gap junction between neurons were observed at hypostome-tentacle junctions. Our morphological evidence of synaptic connectivity in neuronal clusters at hypostome-tentacle junctions suggests that primitive ganglia are present in Hydra.     

First node of Ranvier facilitates high-frequency burst encoding

In central neurons the first node of Ranvier is located at the first axonal branchpoint, ～ 100 μm from the axon initial segment where synaptic inputs are integrated and converted into action potentials (APs). Whether the first node contributes to this signal transformation is not well understood. Here it was found that in neocortical layer 5 axons, the first branchpoint is required for intrinsic high-frequency (≥ 100 Hz) AP bursts. Furthermore, block of nodal Na(+) channels or axotomy of the first node in intrinsically bursting neurons depolarized the somatic AP voltage threshold (～ 5 mV) and eliminated APs selectively within a high-frequency cluster in response to steady currents or simulated synaptic inputs. These results indicate that nodal persistent Na(+) current exerts an anterograde influence on AP initiation in the axon initial segment, revealing a computational role of the first node of Ranvier beyond conduction of the propagating AP.     

The cytoskeletal and signaling mechanisms of axon collateral branching

During development, axons are guided to their appropriate targets by a variety of guidance factors. On arriving at their synaptic targets, or while en route, axons form branches. Branches generated de novo from the main axon are termed collateral branches. The generation of axon collateral branches allows individual neurons to make contacts with multiple neurons within a target and with multiple targets. In the adult nervous system, the formation of axon collateral branches is associated with injury and disease states and may contribute to normally occurring plasticity. Collateral branches are initiated by actin filament– based axonal protrusions that subsequently become invaded by microtubules, thereby allowing the branch to mature and continue extending. This article reviews the current knowledge of the cellular mechanisms of the formation of axon collateral branches. The major conclusions of this review are (1) the mechanisms of axon extension and branching are not identical; (2) active suppression of protrusive activity along the axon negatively regulates branching; (3) the earliest steps in the formation of axon branches involve focal activation of signaling pathways within axons, which in turn drive the formation of actin-based protrusions; and (4) regulation of the microtubule array by microtubule-associated and severing proteins underlies the development of branches. Linking the activation of signaling pathways to specific proteins that directly regulate the axonal cytoskeleton underlying the formation of collateral branches remains a frontier in the field.     

Sprouting and functional regeneration of an identified serotonergic neuron following axotomy

An identified serotonergic neuron (C1) in the cerebral ganglion of Helisoma trivolvis sprouts following axotomy and rapidly (seven to eight days) regenerates to recover its regulation of feeding motor output from neurons of the buccal ganglia. The morphologies of normal and regenerated neurons C1 were compared. Intracellular injection of the fluorescent dye, Lucifer Yellow, into neuron C1 was compared with serotonin immunofluorescent staining of the cerebral and buccal ganglia. The two techniques revealed different and complimentary representations of the morphology of neuron C1. Lucifer Yellow provided optimal staining of the soma, major axon branches, and dendritic arborization. Immunocytochemical staining revealed terminal axon branches on distant targets and showed an extensive plexus of fine fibers in the sheaths of ganglia and nerve trunks. In addition to C1, serotonin-like immunoreactivity was localized in approximately 30 other neurons in each of the paired cerebral ganglia. Only cerebral neurons C1 had axons projecting to the buccal ganglia. No neuronal somata in the buccal ganglia displayed serotonin-like immunoreactivity. Observations of regenerating neurons C1 demonstrated: Actively growing neurites, both in situ and in cell culture, displayed serotonin-like immunoreactivity; severed distal axons of C1 retained serotonin-like immunoreactivity for up to 28 days; axotomized neurons C1 regenerated to restore functional control over the feeding motor program.     

Localization of GABA- and glutamate-like immunoreactivity in the cardiac ganglion of the lobster Panulirus argus

Wholemount immunohistochemical methods were used to examine the localization of γ-aminobutyric acid (GABA) and glutamate within the cardiac system of the Caribbean spiny lobster Panulirus argus. All of the GABA-like immunoreactivity (GABAi) in the cardiac ganglion originated from a single bilateral pair of fibers that entered the heart via the two dorsal nerves. Each GABAi axon bifurcated upon entering the ganglion and gave rise to varicose fibers that surrounded the somata and initial segments of the five large motor neurons. The four small posterior cells did not appear to receive somatic contacts. Double-labeling experiments in which individual motor neurons were injected with Neurobiotin showed that their dendritic processes, which project to muscle bundles adjacent to the ganglion and are thought to respond to stretch, were also accompanied by branches of the GABAi fibers. Glutamate-like immunoreactivity (GLUi) was present in each of the motor neuron cell bodies. In some preparations, GLUi was also detected in large caliber fibers in the major ganglionic nerves. These fibers gave rise to more slender branches that innervated the cardiac muscle bundles. GLUi was also found in the small cell bodies and in fibers surrounding motor neuron somata. Taken together, these findings support previous electrophysiological, pharmacological and anatomical studies indicating that GABA mediates extrinsic inhibition and that glutamate acts as a neuromuscular and intraganglionic transmitter in this system. While axosomatic contacts may play a major role in both transmitter systems, the GABAergic inhibition also appears to involve substantial axodendritic synaptic signaling.     

Neurite outgrowth on cryostat sections of innervated and denervated skeletal muscle

To localize factors that guide axons reinnervating skeletal muscle, we cultured ciliary ganglion neurons on cryostat sections of innervated and denervated adult muscle. Neurons extended neurites on sections of muscle (and several other tissues), generally in close apposition to sectioned cell surfaces. Average neurite length was greater on sections of denervated than on sections of innervated muscle, supporting the existence of functionally important differences between innervated and denervated muscle fiber surfaces. Furthermore, outgrowth was greater on sections of denervated muscle cut from endplate-rich regions than on sections from endplate-free regions, suggesting that a neurite outgrowth-promoting factor is concentrated near synapses. Finally, 80% of the neurites that contacted original synaptic sites (which are known to be preferentially reinnervated by regenerating axons in vivo) terminated precisely at those contacts, thereby demonstrating a specific response to components concentrated at endplates. Together, these results support the hypothesis that denervated muscles use cell surface (membrane and matrix) molecules to inform regenerating axons of their state of innervation and proximity to synaptic sites.     

The neuron of the fast conducting system in hirudo medicinalis: identification and synaptic connections with primary afferent neurons.

A single neuron, located in the center of each segmental ganglion of H. medicinalis is antidromically activated by electrical stimulation of the ventral cord anteriorly and posteriorly to the ganglion, at the same threshold as the fast conducting system (FCS) and with a latency equal to the FCS conduction time. This neuron is activated trans-synaptically by tactile and photic stimulation of the skin and by stimulation of high-threshold fibres running along the cord. A spike evoked by intracellular stimulation of this neuron propagates along the FCS. Intracellular staining shows that this neuron sends two axonal branches in the anterior and posterior median connectives. Direct electrical stimulation of touch cells (T cells), as well as mechanical stimulation of the skin, lowers the threshold of and may eventually fire, the FCS neurons, not only at the level of the ganglion to which they belong, but also at the level of the neighbouring ganglia. This effect is mediated by bilateral pathwasy located in the lateral connectives. It is concluded that the FCS consists of a chain of single neurons, located in each ganglion and electrotonically coupled to each other. Touch cells project with excitatory synapses on the FCS neurons.