Development of corticogeniculate synapses in the cat

BDNF treatment and extended recovery from optic nerve trauma in the cat

PURPOSE

We examined the treatment period necessary to restore retinal and visual stability following trauma to the optic nerve.

METHODS

Cats received unilateral optic nerve crush and no treatment (NT), treatment of the injured eye with brain-derived neurotrophic factor (BDNF), or treatment of the injured eye combined with treatment of visual cortex for 2 or 4 weeks. After 1-, 2-, 4-, or 6-week survival periods, pattern electroretinograms (PERGs) were obtained and retinal ganglion cell (RGC) survival determined.

RESULTS

In the peripheral retina, RGC survival for NT, eye only, and eye + cortex animals was 55%, 78%, and 92%, respectively, at 1 week, and 31%, 60%, and 93%, respectively, at 2 weeks. PERGs showed a similar pattern of improvement. After 4 weeks, RGC survival was 7%, 29%, and 53% in each group, with PERGs in the dual-treated animals similar to the 1- to 2-week animals. For area centralis (AC), the NT, eye only, and eye + cortex animals showed 47%, 78%, and 82% survival, respectively, at 2 weeks, and 13%, 54%, and 81% survival, respectively, at 4 weeks. Removing the pumps at 2 weeks resulted in ganglion cell survival levels of 76% and 74% in the AC at 4 and 6 weeks postcrush, respectively. The PERGs from 2-week treated, but 4- and 6-week survival animals were comparable to those of the 2-week animals.

CONCLUSIONS

Treating the entire central visual pathway is important following optic nerve trauma. Long-term preservation of central vision may be achieved with as little as 2 weeks of treatment using this approach.

Synaptic development of the mouse dorsal lateral geniculate nucleus

The dorsal lateral geniculate nucleus (dLGN) of the mouse has emerged as a model system in the study of thalamic circuit development. However, there is still a lack of information regarding how and when various types of retinal and nonretinal synapses develop. We examined the synaptic organization of the developing mouse dLGN in the common pigmented C57/BL6 strain, by recording the synaptic responses evoked by electrical stimulation of optic tract axons, and by investigating the ultrastructure of identified synapses. At early postnatal ages (<P12), optic tract evoked responses were primarily excitatory. The full complement of inhibitory responses did not emerge until after eye opening (>P14), when optic tract stimulation routinely evoked an excitatory postsynaptic potential/inhibitory postsynaptic potential (EPSP/IPSP) sequence, with the latter having both a GABA(A) and GABA(B) component. Electrophysiological and ultrastructural observations were consistent. At P7, many synapses were present, but synaptic profiles lacked the ultrastructural features characteristic of the adult dLGN, and little gamma-aminobutyric acid (GABA) could be detected by using immunocytochemical techniques. In contrast, by P14, GABA staining was robust, mature synaptic profiles of retinal and nonretinal origin were easily distinguished, and the size and proportion of synaptic contacts were similar to those of the adult. The emergence of nonretinal synapses coincides with pruning of retinogeniculate connections, and the transition of retinal activity from spontaneous to visually driven. These results indicate that the synaptic architecture of the mouse dLGN is similar to that of other higher mammals, and thus provides further support for its use as a model system for visual system development.

Combined application of BDNF to the eye and brain enhances ganglion cell survival and function in the cat after optic nerve injury

PURPOSE

To determine whether application of BDNF to the eye and brain provides a greater level of neuroprotection after optic nerve injury than treatment of the eye alone.

METHODS

Retinal ganglion cell survival and pattern electroretinographic responses were compared in normal cat eyes and in eyes that received (1) a mild nerve crush and no treatment, (2) a single intravitreal injection of BDNF at the time of the nerve injury, or (3) intravitreal treatment combined with 1 to 2 weeks of continuous delivery of BDNF to the visual cortex, bilaterally.

RESULTS

Relative to no treatment, administration of BDNF to the eye alone resulted in a significant increase in ganglion cell survival at both 1 and 2 weeks after nerve crush (1 week, 79% vs. 55%; 2 weeks, 60% vs. 31%). Combined treatment of the eye and visual cortex resulted in a modest additional increase (17%) in ganglion cell survival in the 1-week eyes, a further significant increase (55%) in the 2-week eyes, and ganglion cell survival levels for both that were comparable to normal (92%-93% survival). Pattern ERG responses for all the treated eyes were comparable to normal at 1 week after injury; however, at 2 weeks, only the responses of eyes receiving the combined BDNF treatment remained so.

CONCLUSIONS

Although treatment of the eye alone with BDNF has a significant impact on ganglion cell survival after optic nerve injury, combined treatment of the eye and brain may represent an even more effective approach and should be considered in the development of future optic neuropathy-related neuroprotection strategies.

Do first order and higher order regions of the thalamic reticular nucleus have different developmental timetables?

The thalamic reticular nucleus (TRN) can been subdivided into sectors based on thalamic and cortical input. Additionally, in carnivores the visual sector of the TRN can be subdivided into first order (perigeniculate nucleus: PGN) and higher order (TRN) regions. This report examines whether TRN development reflects the nature of its higher order visual connections. 170 cells from 12 kittens aged between postnatal day 0 (P0) and P125 were fully analysed after single cell injections in 400-500 microm fixed brain slices. TRN cells have a period of exuberant dendritic branching that peaks between P3 and P12, around the time of eye opening (P7), followed by branch pruning until P68. Similarly, most dendritic appendages are added between P12 and P22 followed by pruning, which is also largely complete by P68. Most branch points occur within the first 10-30% of the dendritic arbor, peaking between 10 and 20% (roughly equivalent to 100 mum from the soma), while appendages were concentrated between 20 and 30% of the arbour; appendages tend to be distributed over a larger proportion of the arbor up to P14 compared to later ages. TRN and PGN maturation were not significantly different. The present data suggest that clear distinctions cannot be made between the maturation of first and higher order pathways and indicate that GABAergic cells of the ventral thalamus may mature earlier than relay cells of the dorsal thalamus. Furthermore, dendritic development in the TRN may be less dependent on extrinsic factors than an intrinsic growth pattern or factors other than a functional hierarchy within the visual pathway.

Does the development of the perigeniculate nucleus support the notion of a hierarchical progression within the visual pathway?

The development of the visual pathway has been extensively studied. However, despite of the importance of the perigeniculate nucleus within this pathway, there is a lack of information concerning its development. The present study examined the dendritic development of perigeniculate nucleus cells using single cell injections in 400-500 microm thick fixed brain slices from kittens of different ages between postnatal day 0 and postnatal day 125. A total of 189 perigeniculate nucleus cells were reconstructed from serial sections for qualitative and quantitative analysis. Cells during the first month were characterized by an abundance of branch points and appendages. There was a significant (P>0.05), albeit variable, increase in the number of branch points and appendages up to about postnatal day 12 after which the numbers were rapidly reduced over the next two weeks. Similarly, appendage numbers significantly increased over the first two weeks until postnatal day 17 and then fell to near adult levels by postnatal day 34. The majority of branch points and appendages occur within 100-200 microm of the soma (10-30% of the dendritic diameter). The data indicate that perigeniculate nucleus dendritic maturation lags shortly behind that of the retina but may precede that of its dorsal thalamic target, the lateral geniculate nucleus. Thus, it may be that the earlier maturation of the perigeniculate nucleus and its inhibitory input is a necessary requirement for the proper development of retinogeniculate and corticothalamic topographic maps within the dorsal lateral geniculate nucleus and perigeniculate nucleus.

Role of retinal afferents in regulating growth and shape of the lateral geniculate nucleus

Segregated binocular maps in the dorsal lateral geniculate nucleus (LGN) develop from a stage where they initially overlap. Sophisticated computational models have been used to describe the dynamics of three-dimensional LGN shape changes that play a role in segregation. These models have revealed specific nuclear growth vectors associated with the process of ocular segregation in the LGN (Williams and Jeffery [2001] J Comp Neurol 430:332-342.). In this study, we use similar techniques to determine whether retinal innervation contributes to the dynamics of shape maturation in the ferret LGN. In this animal, 90% of the retinal innervation of the mature LGN comes from the contralateral eye. If one eye is removed before segregation, the projection from the remaining eye remains diffuse and nuclear growth is stunted. Here, we quantify this effect and show that removing the contralateral projection before segregation has a profound impact on LGN size but changes its ultimate shape by only 12%. The impact on shape on the other side of the brain where the ipsilateral projection is removed, which accounts for only 10% of its innervation in maturity, is less than 2%. Hence, retinal innervation plays a minor role in determining mature LGN shape. Although in both hemispheres, the ultimate shape of the nucleus is close to normal, removal of the larger projection disrupts normal growth vectors, with almost none being present in the 5 days after enucleation, when the normal nucleus expands markedly. Hence, the effect of enucleation is to delay shape maturation. Growth vectors absent after removal of the smaller projection are mainly confined to those in what would be the binocular region.

Ultrastructural synaptic organization of axon terminals in the ventral part of the cat oral pontine reticular nucleus

In an attempt to contribute to the current knowledge of the brainstem reticular formation synaptic organization, the ultrastructure and distribution of synaptic terminal profiles on neurons in the ventral part of the oral pontine reticular nucleus (vRPO), the rapid eye movement (REM) sleep-induction site, were studied quantitatively. Terminals with asymmetric contacts and rounded vesicles were classified according to vesicle density as type I or II (high or low density, respectively). The area, apposed perimeter length, and mitochondrial area of type I terminals, on average, were significantly smaller than those of type II terminals. Type III and IV terminals had symmetric contacts and oval and/or flattened vesicles; type III terminals formed synapses between them and on initial axons. Type V and VI terminals showed characteristics intermediate to those of asymmetric and symmetric synapses. Interestingly, some terminal features were related to both terminal area and postsynaptic dendritic diameter. The percentages of different synapses sampled on somata were as follows: asymmetric synapses (usually formed by type II terminals; mean +/- S.D.), 26.4% +/- 3%; symmetric synapses, 46.7% +/- 5.2%; and intermediate synapses, 26.9% +/- 6.1%. The percentages of different synapses sampled on dendrites were asymmetric synapses, 62.1% +/- 9%; symmetric synapses, 25.6% +/- 8.1%; and intermediate synapses, 12.3% +/- 1.7%. Comparison between large- and small-diameter dendrites revealed that the percentages of symmetric synapses and type II terminals decreased, whereas the percentages of type I terminals increased as postsynaptic dendritic diameters became smaller. Synaptic density was approximately four times lower on somata than on dendrites. The vRPO synaptic organization reflects some patterns that are similar to those found in other regions of the central nervous system as well as specific synaptic patterns that are probably related to its functions: the generation and maintenance of REM sleep and the control of eye movement or limb muscle tone.

Comparison of the laminar distribution of input from areas 17 and 18 of the visual cortex to the lateral geniculate nucleus of the cat

The feedback from area 18 of the cat visual cortex to the lateral geniculate nucleus has been investigated by labeling and reconstructing seventeen axons of known receptive field position and eye preference. The distribution of boutons from each axon was quantified with respect to the compartments of the geniculate complex, and the results were compared with an equivalent analysis of fourteen area 17 axons. Area 18 axons form large, sparse arborizations that extend up to 1.9 mm laterally (1170 +/- 85 microm; mean +/- SEM), with a core of relatively dense innervation spanning on average 600 +/- 70 microm (mean +/- SEM). Thus, they have the potential to influence the transmission of visual information from well beyond their own classical receptive fields. In this respect, they are surprisingly similar to the axons from area 17, despite the fact that the two cortical areas have very different retinotopy. However, there are important differences between the pathways. Area 18 axons project more heavily to the C layers and medial interlaminar nucleus. Whereas the input from both areas to the A layers is biased toward the layer appropriate to the eye preference of each axon, the area 18 input to magnocellular layer C is not. The distribution of area 18 boutons favors the bottom of their preferred A layer, and the area 17 boutons favor the top. These differences mirror those seen in the afferent pathways, suggesting that each cortical area preferentially targets the cells from which it receives input. Finally, their greater diameter suggests that area 18 axons provide the earliest feedback signal in the corticogeniculate loop.

The role of retinal waves and synaptic normalization in retinogeniculate development

The prenatal development of the cat retinogeniculate pathway is thought to involve activity-dependent mechanisms driven by spontaneous waves of retinal activity. The role of these waves upon the segregation of the dorsal lateral geniculate nucleus (LGN) into two eye-specific layers and the development of retinotopic mappings have previously been investigated in a computer model. Using this model, we examine three aspects of retinogeniculate development. First, the mapping of visual space across the whole network into projection columns is shown to be similar to the mapping found in the cat. Second, the simplicity of the model allows us to explore how different forms of synaptic normalization affect development. In comparison to most previous models of ocular dominance, we find that subtractive postsynaptic normalization is redundant and divisive presynaptic normalization is sufficient for normal development. Third, the model predicts that the more often one eye generates waves relative to the other eye, the more LGN units will monocularly respond to the more active eye. In the limit when one eye does not generate any waves, that eye totally disconnects from the LGN allowing the non-deprived eye to innervate all of the LGN. Thus, as well as accounting for normal retinogeniculate development, the model also predicts development under abnormal conditions which can be experimentally tested.

Muscarinic receptor subtypes in the lateral geniculate nucleus: a light and electron microscopic analysis

Neural activity in the dorsal lateral geniculate nucleus of the thalamus (DLG) is modulated by an ascending cholinergic projection from the brainstem. The purpose of this study was to identify and localize specific muscarinic receptors for acetylcholine in the DLG. Receptors were identified in rat and cat tissue by means of antibodies to muscarinic receptor subtypes, ml-m4. Brain sections were processed immunohistochemically and examined with light and electron microscopy. Rat DLG stained positively with antibodies to the m1, m2,and m3 receptor subtypes but not with antibodies to the m4 receptor subtype. The m1 and m3 antibodies appeared to label somata and dendrites of thalamocortical cells. The m1 immunostaining was pale, whereas m3-positive neurons exhibited denser labeling with focal concentrations of staining. Strong immunoreactivity to the m2 antibody was widespread in dendrites and somata of cells resembling geniculate interneurons. Most m2-positive synaptic contacts were classified as F2-type terminals, which are the presynaptic dendrites of interneurons. The thalamic reticular nucleus also exhibited robust m2 immunostaining. Cat DLG exhibited immunoreactivity to the m2 and m3 antibodies. The entire DLG stained darkly for the m2 receptor subtype, except for patchy label in the medial interlaminar nucleus and the ventralmost C laminae. The staining for m3 was lighter and was distributed more homogeneously across the DLG. The perigeniculate nucleus also was immunoreactive to the m2 and m3 subtype-specific antibodies. Immunoreactivity in cat to the m1 or m4 receptor antibodies was undetectable. These data provide anatomical evidence for specific muscarinic-mediated actions of acetylcholine on DLG thalamocortical cells and thalamic interneurons.

Stable properties of spontaneous EPSCs and miniature retinal EPSCs during the development of ON/OFF sublamination in the ferret lateral geniculate nucleus

Retinal projections to the lateral geniculate nucleus (LGN) in ferrets progressively segregate into eye-specific laminae and subsequently into sublaminae that receive inputs from either ON-center or OFF-center afferents. To study the development of synaptic efficacy during a period of activity-dependent growth and reorganization in the CNS, we recorded spontaneous EPSCs (sEPSCs) from cells of the LGN during ON/OFF sublamination. We also examined retinal inputs specifically by stimulating the optic tract in the presence of strontium and recording evoked miniature EPSCs (emEPSCs). The rise times, areas, half-widths, and decay times of sEPSCs and emEPSCs and interevent intervals of sEPSCs recorded at the beginning of ON/OFF sublamination were not different from those recorded after its completion. Typically EPSC areas were small (10-20 fC) but varied greatly both within and between neurons. The frequency of sEPSCs was also quite variable, ranging from 0.2 to 5 Hz. sEPSCs were equivalent to miniature EPSCs recorded in the presence of tetrodotoxin, and both sEPSCs and emEPSCs were CNQX-sensitive. No difference was observed between sEPSCs recorded at room temperature and those recorded at 34 degreesC, and strontium could be substituted for calcium with no effect on sEPSC shape. These data argue for a remarkable stability in the components of at least AMPA-mediated synaptic transmission during a period of major synaptic rearrangement in the LGN.

Correlation-based development of ocularly matched orientation and ocular dominance maps: determination of required input activities

We extend previous models for separate development of ocular dominance and orientation selectivity in cortical layer 4 by exploring conditions permitting combined organization of both properties. These conditions are expressed in terms of functions describing the degree of correlation in the firing of two inputs from the lateral geniculate nucleus (LGN), as a function of their retinotopic separation and their "type" (ON center or OFF center and left eye or right eye). The development of ocular dominance requires that the correlations of an input with other inputs of the same eye be stronger than or equal to its correlations with inputs of the opposite eye and strictly stronger at small retinotopic separations. This must be true after summing correlations with inputs of both center types. The development of orientation-selective simple cells requires that (1) an input's correlations with other inputs of the same center type be stronger than its correlations with inputs of the opposite center type at small retinotopic separation; and (2) this relationship reverse at larger retinotopic separations within an arbor radius (the radius over which LGN cells can project to a common cortical point). This must be true after summing correlations with inputs serving both eyes. For orientations to become matched in the two eyes, correlated activity within the receptive fields must be maximized by specific between-eye alignments of ON and OFF subregions. Thus the correlations between the eyes must differ depending on center type, and this difference must vary with retinotopic separation within an arbor radius. These principles are satisfied by a wide class of correlation functions. Combined development of ocularly matched orientation maps and ocular dominance maps can be achieved either simultaneously or sequentially. In the latter case, the model can produce a correlation between the locations of orientation map singularities and local ocular dominance peaks similar to that observed physiologically. The model's main prediction is that the above correlations should exist among inputs to cortical layer 4 simple cells before vision. In addition, mature simple cells are predicted to have certain relationships between the locations of the ON and OFF subregions of the left and right eyes' receptive fields.

Corticocortical communication via the thalamus: ultrastructural studies of corticothalamic projections from area 17 to the lateral posterior nucleus of the cat and inferior pulvinar nucleus of the owl monkey

Electron microscopic anterograde autoradiography has been used to analyze the morphology and postsynaptic relationships of area 17 cortical terminals in the lateral division of the lateral posterior nucleus (LPl) of the cat and medial division of the inferior pulvinar nucleus (IPm) of the owl monkey. Such terminals are thought to arise exclusively from layer 5 in the cat and primate (Lund et al. [1975] J. Comp. Neurol. 164:287-304; Abramson and Chalupa [1985] Neuroscience 15:81-95). All labeled terminals in both nuclei exhibited the morphology of ascending "lemniscal" afferents. That is, they contained round vesicles, were large, made asymmetrical synaptic and filamentous nonsynaptic contacts, and were classified as RLs. These cortical RLs also exhibited the postsynaptic relationships of lemniscal afferents. Thus, they were presynaptic to large dendrites within glial encapsulated glomeruli, where a majority was involved in complex synaptic arrangements called triads. They also were found adjacent to terminal profiles with pleomorphic vesicles but never adjacent to small terminals containing round vesicles. Our results suggest that the layer 5 projection from area 17 provides a functional "drive" for some LPl and IPm neurons. Information carried over this "re-entrant" pathway (Guillery [1995] J. Anat. 187:583-592) could be modified within the LPl and IPm by both cortical and subcortical pathways and subsequently conveyed to higher visual cortical areas, where it could be integrated with messages carried through the well-documented corticocortical pathways (Casagrande and Kaas [1994] Cerebral cortex New York: Plenum Press).

Relative numbers of cortical and brainstem inputs to the lateral geniculate nucleus

Terminals of a morphological type known as RD (for round vesicles and dense mitochondria, which we define here as the aggregate of types formerly known as RSD and RLD, where "S" is small and "L" is large) constitute at least half of the synaptic inputs to the feline lateral geniculate nucleus, which represents the thalamic relay of retinal input to cortex. It had been thought that the vast majority of these RD terminals were of cortical origin, making the corticogeniculate pathway by far the largest source of input to geniculate relay cells. However, another source of RD terminals recently identified derives from cholinergic cells of the brainstem parabrachial region. (These cells also contain NO.) We used techniques of electron microscopy to determine quantitatively the relative contribution of cortex and brainstem to the population of RD terminals. We identified corticogeniculate terminals by orthograde transport of biocytin injected into the visual cortex and identified brainstem terminals by immunocytochemical labeling for choline acetyltransferase or brain NO synthase (the synthesizing enzymes for acetylcholine and NO, respectively). We estimated the relative numbers of corticogeniculate and brainstem terminals with a two-step algorithm: First, we determined the relative probability of sampling each terminal type in our material, and then we calculated what mixture of identified corticogeniculate and brainstem terminals was needed to recreate the size distribution of the parent RD terminal population. We conclude that brainstem terminals comprise roughly one-half of the RD population. Thus, the cortical input is perhaps half as large and the brainstem input is an order of magnitude larger than had been thought. This further suggests that the brainstem inputs might play a surprisingly complex and subtle role in the control of the geniculocortical relay.

Morphological diversity in terminals of W-type retinal ganglion cells at projection sites in cat brain

The morphological features of retinal terminals in cat brain were examined at sites where projections of W-type ganglion cells predominate. These included the parvicellular C laminae of the dorsal lateral geniculate nucleus, the ventral lateral geniculate nucleus, stratum griseum superficiale of the superior colliculus, and the suprachiasmatic nucleus. Positive identification of retinal terminals was achieved following anterograde transport of intravitreally injected native or wheat germ agglutinin-conjugated horseradish peroxidase. In contrast to the classic features of retinal terminals as defined from sites where X- and Y-type ganglion cells predominate, i.e. round synaptic vesicles, large profiles, and pale mitochondria, substantial numbers of terminals in W-cell rich areas were found to contain dark mitochondria. Synaptic vesicles, although consistently round, were typically smaller in terminals with dark mitochondria than in those with pale mitochondria. These findings indicate a diversity among terminals of W-cells and suggest that such terminals cannot be distinguished on the basis of limited morphological criteria.

Spatial frequency tuning of orientation-discontinuity-sensitive corticofugal feedback to the cat lateral geniculate nucleus

1. The influence of spatial frequency on the inhibitory component of the effects mediated by feedback from the visual cortex has been examined in X and Y cells in the A laminae of the feline dorsal lateral geniculate nucleus (dLGN). Experiments utilized a concentric, bipartite visual stimulus centered over the receptive fields of the cells studied. The responses of dLGN cells to selective stimulation of receptive field centre (with the inner window) were compared with those to stimulation of centre and surround mechanisms (both inner and outer window), with the stimuli either in or out of orientation alignment. 2. With these same stimuli, layer VI cells in the visual cortex showed a marked increase in response magnitude when the inner and outer components of the stimulus were in orientation alignment, and presented at the preferred orientation. In the case of dLGN X and Y cells we observed an enhancement of the surround antagonism of the centre response when the inner and outer sections of the stimulus were in orientation alignment. 3. The effects of varying spatial frequency on these responses were examined in dLGN cells in the presence of corticofugal feedback. With the stimulus sections in orientation alignment, surround stimulation produced a powerful and significant reduction in the response to stimulation of centre mechanism alone with the most marked effects for stimuli in the range 0.1-0.85 cycles per degree (c.p.d.). The reduction produced by surround stimulation in the range 0.1-0.5 c.p.d. was notably more potent in X cells than in Y cells. 4. The responses to the same stimuli were examined in dLGN cells with the corticofugal feedback inactivated. Comparison of data from cells studied with and without feedback revealed a significant decrease in surround-mediated attenuation of the centre response in Y cells for spatial frequencies in the range 0.1-0.85 c.p.d. For X cells the decrease in strength of the surround antagonism was also clear and significant but only seen in the range 0.1-0.5 c.p.d. 5. The influence of the orientation alignment of inner and outer stimulus sections revealed a marked difference between cells studied with and without feedback. In the presence of feedback fully aligned stimuli enhanced surround antagonism of centre responses for spatial frequencies in the range 0.1-0.5 c.p.d., in X and Y cells. In the absence of corticofugal feedback this alignment effect was essentially eliminated. 6. These data show that surround antagonism of the centre response is influenced by orientation alignment of the stimulus sections at low spatial frequencies and in the presence of corticofugal feedback. They support a cortically driven enhancement of the inhibitory mechanisms reinforcing surround mechanisms in the dLGN. We propose that feedback enhances a low spatial frequency cut-off in the dLGN, that this effect is maximal for a continuous iso-orientated contour, but diminished whenever there is an orientation discontinuity. The hyperpolarizing influence underlying this effect may contribute to the recently described synchronizing influence of the direct corticofugal contacts onto relay cells. We suggest feedback of the cortical level of analysis refines the transfer of the visual input at geniculate level in a stimulus-context-dependent fashion.

Quantitative electron microscopic analysis of synaptic input from cortical areas 17 and 18 to the dorsal lateral geniculate nucleus in cats

Cortical feedback is the largest extraretinal projection to the lateral geniculate nucleus. This input is thought to modulate the transfer of visual information in a state-dependent manner. The quantitative distribution and synaptology of axon terminals arising from different cortical areas is still an unsolved question. To address this problem, the synaptic termination pattern of corticogeniculate axons from cortical areas 17 and 18 entering the lateral geniculate nucleus of the cat was examined. The Phaseolus vulgaris leucoagglutinin anterograde tract tracing method was used for the labeling of corticogeniculate terminals. Postsynaptic targets were characterized by postembedding gamma-aminobutyric acid (GABA) immunocytochemistry. In both laminae A and A1, labeled corticogeniculate axons from area 17 established synaptic contacts with GABA-immunopositive, interneuronal dendritic profiles more frequently (17.5% of all axons) than did labeled axon terminals from area 18 (7% of axons). Conversely, 76% of labeled corticogeniculate axons from area 17, as opposed to 87% of labeled axons from area 18, terminated on GABA-immunonegative relay cell dendrites. Furthermore, the mean diameter of GABA-negative relay cell dendrites postsynaptic to labeled axons from area 17 was significantly smaller than the diameter of relay cell dendrites synapsing with labeled terminals from area 18. These results indicate that the corticogeniculate axons from cortical areas 17 and 18 exhibit different synaptic termination patterns in the dorsal lateral geniculate nucleus of the cat, suggesting that these two projections may subserve different functions in visual information processing.

GABAergic projection from the basal forebrain to the visual sector of the thalamic reticular nucleus in the cat

We examined the projection from the basal forebrain to thalamic and cortical regions of the visual system in cats, with particular reference to the visual sector of the thalamic reticular nucleus, the lateral geniculate nucleus, and the striate cortex. First, we made injections of wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP) into the visual sector of the thalamic reticular nucleus and found cells labeled by retrograde transport in the lateral nucleus basalis magnocellularis. Injection of biocytin into the basal forebrain resulted in the anterograde labeling of a dense band of fibers and terminals within the entire thalamic reticular nucleus; this labeling extended through the visual sector including the perigeniculate nucleus. No orthograde labeling was found in the lateral geniculate nucleus. Next, we addressed the issue of putative neurotransmitters used by this pathway using a variety of immunocytochemical and histochemical markers. In this fashion, we identified two populations of cells in the nucleus basalis magnocellularis of the cat; large cholinergic cells that contain choline acetyltransferase, NADPH-diaphorase, and calbindin and that project to striate cortex and smaller cells that contain gamma-aminobutyric acid (GABA), glutamic acid decarboxylase, and parvalbumin and that project to the visual sector of the thalamic reticular nucleus. We also examined at the electron microscopic level terminals in the visual sector of the thalamic reticular nucleus that were labeled from a biocytin injection in the basal forebrain. Most of these terminals form symmetric contacts onto dendrites and were revealed by postembedding immunocytochemical staining to be positive for GABA.

Ultrastructural studies of the primate lateral geniculate nucleus: morphology and spatial relationships of axon terminals arising from the retina, visual cortex (area 17), superior colliculus, parabigeminal nucleus, and pretectum of Galago crassicaudatus

The electron microscopic autoradiographic tracing method has been used to examine the morphology and postsynaptic relationships of five projections (retina, cortical area 17, superior colliculus (tectal), parabigeminal nucleus, and pretectum) to the dorsal lateral geniculate nucleus of the greater bush baby Galago crassicaudatus. Retinal terminals have been examined in the contralaterally innervated layer of each of the three matched pairs [parvi- (X-cell), magno- (Y-cell), and koniocellular (small, W-cell)] of geniculate layers. These terminals are large and contain pale mitochondria and round vesicles (RLPs). RLPs are presynaptic to juxtasomatic regions of parvi- and magnocellular neurons. In contrast, RLPs innervate more distal regions of koniocellular neurons. Labeled cortical, tectal, and parabigeminal terminals are relatively small and contain round vesicles and dark mitochondria. Cortical terminals in each of the three representative layers are presynaptic to small diameter dendrites. No convergence of cortical and retinal terminals has been seen in any layer. Labeled tectal and parabigeminal terminals are found primarily in the koniocellular layers, but the latter are also seen in all other layers. Tectal and parabigeminal terminals have been observed converging with retinal terminals on dendrites of some koniocellular neurons. Labeled pretectogeniculate terminals contain densely packed pleomorphic vesicles, dark mitochondria, and a dark cytoplasmic matrix. These terminals, which are present in each of the representative layers, are presynaptic to conventional dendrites and profiles containing loosely dispersed pleomorphic vesicles and a pale cytoplasmic matrix.

Cytochemical polarity in lateral geniculate interneurons

To determine the cytochemical composition of presynaptic dendrites, we have examined the distribution of synapsin 1, calcium and calmodulin-dependent protein kinase II (CaM-II), microtubule-associated protein 2 (MAP-2) and spectrin in cat lateral geniculate (LGN) class III cells by immune-EM. Special attention was paid to the dendrites of these interneurons because they are both pre- and postsynaptic. The dendritic proteins MAP-2 and RBC spectrin were not observed in interneuron dendrites but these proteins were localized in relay cell dendrites. The synaptic vesicle-associated protein synapsin 1 was present in all synaptic vesicle containing profiles, including dendritic terminals. CaM-II, the major postsynaptic density protein, was found in all dendrites. Thus, the LGN interneuron dendritic compartment displays both axonal and dendritic cytochemical properties. The results suggest the possibility of unique molecular interactions in interneuron dendritic terminals.

Development of corticotectal synaptic terminals in the cat: a quantitative electron microscopic analysis

We studied the development of corticotectal synaptic terminal boutons and synapses by making injections of wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP) into area 17 of visual cortex in kittens ranging from newborn to 12 weeks of age and in adults. The location and extent of the injection site, and labeled corticotectal axon terminals in the superficial layers of the superior colliculus were demonstrated histochemically with the cobalt-glucose oxidase diaminobenzidine reaction. During the first 2 weeks after birth, the majority of labeled profiles resembled axonal growth cones, or structures intermediate in morphology between growth cones and synaptic terminals, while very few corticotectal axon terminals forming well-defined synaptic contacts were observed. Labeled synaptic terminals in kittens at 1 and 2 weeks of age were small, contained very few synaptic vesicles, which were usually restricted to the contact zone, and exhibited few mitochondria. By 4 and 6 weeks after birth, a well-developed population of synaptic terminals was established; however, growth cones and intermediate profiles were still numerous. At 8 weeks of age synaptic terminals were morphologically mature, and growth cone-like profiles were no longer observed. To study quantitative changes in synapse development we used the disector method to obtain unbiased estimates of the density and number of corticotectal synaptic terminals and synapses; both the density and number of terminals and synapses increased steadily throughout postnatal development. These results suggest that the corticotectal projection develops by the progressive elaboration of synapses, as opposed to synapse overproduction and subsequent elimination.

Orientation sensitive elements in the corticofugal influence on centre-surround interactions in the dorsal lateral geniculate nucleus

In a previous study, we have shown that the corticofugal projection to the dLGN enhances inhibitory mechanisms underlying length tuning. This suggests that the inhibitory influences deriving from the corticofugal feedback should exhibit characteristics that reflect the response properties of orientation-tuned layer VI cells. Here we report data obtained from experiments using a bipartite visual stimulus, with an inner section over the dLGN cell receptive field centre and an outer section extending beyond it. For both X and Y cells there was a modulation of the strength of the surround antagonism of centre responses that was dependent on the orientation alignment of contours in the two components of the stimulus. Layer VI cells showed maximal responses when the two components were aligned to the same orientation; dLGN cells showed a minimal response. Varying the orientation alignment of the inner and outer components of the stimulus in a randomised, interleaved fashion showed that bringing the stimulus into alignment resulted in a 24.28% increase in the surround antagonism of the centre response. Blocking cortical activity showed this effect of alignment to be strongly dependent on corticofugal feedback. This effect of orientation alignment appears to apply for any absolute orientation of the alignment condition and supports the view that an entire subset of cortical orientation columns generate the feedback influencing any given dLGN cell. This mechanism makes dLGN cells sensitive to the orientation domain discontinuities in elongated contours moving across their receptive field.

Response of retinal terminals to loss of postsynaptic target neurons in the dorsal lateral geniculate nucleus of the adult cat

We have used the neurotoxin kainic acid to produce rapid degeneration of neurons in the dorsal lateral geniculate nucleus (dLGN) of the adult cat. This degeneration mimics the rapid loss of geniculate neurons seen after visual cortex ablation in the neonate. Subsequent anterograde transport of horseradish peroxidase injected into the eye was used to reveal the projection patterns of retinal ganglion cell axons at different survival periods after the kainic acid injection. The density of retinal projections to the degenerated regions of the geniculate was reduced considerably at 4 and 6 months survival, but at 2 months was not significantly different from normal. The laminar pattern of projections to degenerated regions of the geniculate did not change in any animals studied, even when an adjacent lamina contained surviving cells. Electron microscopic examination of degenerated dLGN revealed intact retinal (RLP) and RSD terminals at all survival times, although the density of terminals appeared much reduced when compared to controls. Some RLP terminals exhibited the "dark reaction" of degeneration and these degenerating terminals were most numerous at 2 months survival. These findings demonstrate that, in response to degeneration of their usual target cells, mature retinal ganglion cells with withdraw their axon terminals from these regions of degeneration. We conclude that mature retinal ganglion cells continue to be dependent on target integrity for the maintenance of a normal axonal arborization.

A quantitative study of synaptic contacts on interneurons and relay cells of the cat lateral geniculate nucleus

The relative proportions of synapses made by retinal and extraretinal terminals on interneurons and relay cells in lamina A of the dorsal lateral geniculate nucleus (LGN) of the cat were estimated quantitatively in a sample of 4003 synapses. Processes of interneurons or relay cells were identified by presence or absence of GABA immunoreactivity, respectively, in thin sections treated with post-embedding anti-GABA immunogold. On the basis of ultrastructural features, synaptic terminals were interpreted as belonging to retinal axons, cortical axons or axon collaterals of relay cells. GABAergic terminals were positively identified by being immunoreactive. GABA(-) terminals with heterogeneous and poorly defined characteristics, which could not be identified in the above classes, were grouped together in an "undetermined" category. Among the total synaptic inputs to interneurons, the following relative percentages of synapses from different terminals were obtained: retinal 25%, cortical 37%, GABAergic 26%, axon collaterals 2%, undetermined 6%. The vast majority of retinal terminals synapse on dendritic appendages of interneurons rather than on their dendritic trunks (about 20:1). By contrast, the majority of cortical terminals synapse on dendrites rather than on dendritic appendages (about 5:1). Virtually all axon-collaterals synapses were established on dendritic appendages. 17% of the dendritic profiles of interneurons contain synaptic vesicles; many of these profiles were seen in postsynaptic relation to cortical axons and in presynaptic relation with relay dendrites. Given the extensive electrotonic lengths of these cells observed by others, and the expected high electric resistance of the slender stalks that are known to connect the dendritic appendages to interneurons, these results suggest that microcircuits involving the interneuronal dendritic appendages with dendrites of relay cells are under predominantly retinal control. The microcircuits established by presynaptic dendritic trunks with relay dendrites, are under predominantly cortical control. The axonal (spiking) output of interneurons would be under control of the few retinal synapses on proximal dendrites of these cells. Among the total synaptic inputs to relay cells, the following relative percentages of different synapses were obtained: retinal 12%, cortical 58%, GABAergic 24%, axon collaterals 0.3%, undetermined 5%. Relay cells receive twice the number of cortical synapses than interneurons, suggesting that direct cortical excitatory influences on relay cells are more preponderant than cortico-interneuron mediated inhibition on these cells. The observed proportions of dendritic profiles of relay cells and interneurons (80% and 20%, respectively) in the geniculate neuropil are similar to the known proportions of somata of both types of cells in the A-laminae.(ABSTRACT TRUNCATED AT 400 WORDS)

Connectional studies of the primate lateral geniculate nucleus: distribution of axons arising from the thalamic reticular nucleus of Galago crassicaudatus

Anterograde and retrograde transport methods have been used to explore the interconnections between the thalamic reticular nucleus (TRN) and the dorsal lateral geniculate nucleus of Galago crassicaudatus. We first defined the region of the TRN, which is connected to the lateral geniculate nucleus, by examining the distribution of geniculo-TRN axons, cortico-TRN axons arising from area 17, and the location of TRN-geniculate neurons. Following an intraocular injection of 3H-proline/3 H-leucine, trans-synaptically transported protein is present bilaterally within the lateral portion of the caudal TRN. This same caudal and lateral region is also targeted by cortico-TRN axons and contains neurons which project upon the lateral geniculate nucleus. Light microscopic anterograde transport methods were used to analyze the distribution of TRN-geniculate axons. Our data reveal that all layers and interlaminar zones of the dorsal lateral geniculate nucleus contain TRN axons. Electron microscopic-autoradiographic data support and extend our light microscopic findings by revealing labeled TRN terminals within all geniculate layers. These TRN profiles are the same size throughout the geniculate and exhibit morphological characteristics similar to F1 terminals described by others. That is, they possess predominantly pleomorphic vesicles, a dark cytoplasmic matrix, dark mitochondria, and symmetrical synaptic contacts. Two additional features of TRN terminals have been observed in some profiles. These include dense-core vesicles and a dense, punctate cytoplasmic matrix, which is sometimes associated with the postsynaptic specialization. In addition to their morphology and size, the postsynaptic targets of TRN terminals are similar within the three sets (parvi-, magno-, and koniocellular) of geniculate layers. TRN profiles terminate upon dendrites of all sizes and somata. These findings suggest that the TRN modulates the retino-geniculocortical pathway and that this modulation is occurring in all three streams.

Electron-microscopic analysis of synaptic input from the perigeniculate nucleus to the A-laminae of the lateral geniculate nucleus in cats

The perigeniculate nucleus of carnivores is thought to be a part of the thalamic reticular nucleus related to visual centers of the thalamus. Physiological studies show that perigeniculate neurons, which are primarily GABAergic, provide feedback inhibition onto neurons in the lateral geniculate nucleus. However, little is known about the anatomical organization of this feedback pathway. To address this, we used two complementary tracing methods to label perigeniculate axons for electron microscopic study in the geniculate A-laminae: intracellular injection of horseradish peroxidase (HRP) to fill an individual perigeniculate cell and its axon; and anterograde transport of Phaseolus vulgaris leucoagglutinin to label a population of perigeniculate axons. Labeled perigeniculate terminals display features of F1 terminals in the geniculate neuropil: they are small, contain dark mitochondria, and form symmetric synaptic contacts. We found that most of the perigeniculate terminals (greater than 90%) contact geniculate cell dendrites in regions that also receive a rich innervation from terminals deriving from visual cortex (e.g., "cortico-recipient" dendrites). The remainder of the perigeniculate synapses (10%) contacted dendrites in regions that also received direct retinal input (e.g., "retino-recipient" dendrites). Serial reconstruction of segments of dendrites postsynaptic to perigeniculate terminals suggests that these terminals contact both classes of relay cell in the A-laminae (X and Y), although our preliminary conclusion is that an individual perigeniculate cell contacts only one class. Finally, our quantitative comparison between labeled perigeniculate terminals and unlabeled F1 terminals indicates that these perigeniculate terminals form a distinct subset of F1 terminals. We quantitatively compared the labeled perigeniculate terminals to unlabeled F1 terminals. Although the parameters of the perigeniculate terminals fell entirely within the range of those for the unlabeled F1 terminals, as populations, we found consistent differences between these two groups. We thus conclude that, as populations, other sources of F1 terminals are morphologically distinct from perigeniculate terminals and innervate different targets.

The parabigeminogeniculate projection: connectional studies in eight mammals

Anterograde and retrograde tracing methods have been used to analyze the origin and distribution of parabigeminogeniculate axons in the gray squirrel, the gopher, the rat, the opossum, the cat, the greater bushbaby, the squirrel monkey and the macaque monkey. Our findings reveal that parabigeminogeniculate axons most heavily innervate regions of the lateral geniculate that are also targeted by axons arising from the superior colliculus (tectogeniculate). These geniculate layers and zones of parabigeminal and tectal overlap contain small cells, and in several species are associated with the small W cell retino-geniculocortical pathway. In addition to the dense input to small-celled layers and zones, parabigeminal axons in several species also innervate regions of the lateral geniculate nucleus that are relatively free of tectogeniculate axons and that are associated with the medium (X) and large (Y) cell streams. Finally, our data reveal that the laterality of parabigeminogeniculate pathways varies across mammals, being primarily crossed in the gray squirrel, the gopher, the rat, and the opossum, bilateral in the cat, and primarily ipsilateral in the three primates.

Phaseolus vulgaris leucoagglutinin (PHA-L): a neuroanatomical tracer for electron microscopic analysis of synaptic circuitry in the cat's dorsal lateral geniculate nucleus

Phaseolus vulgaris leucoagglutinin (PHA-L) is a plant lectin that is anterogradely transported by neurons in the central nervous system. PHA-L is selectively taken up by cells at iontophoretic injection sites and, when immunohistochemically demonstrated, labels individual neurons completely, including their dendrites, axons, and terminal boutons. PHA-L is generally not taken up by fibers passing through the injection site and, because it produces a Golgi-like staining of even very fine axons over long distances, it is sometimes possible to light microscopically reconstruct individual neurons and their entire axon terminal arbors. When prepared for electron microscopy, the PHA-L-labeled terminals are densely and completely stained, allowing their synaptic relationships to be defined. These properties make PHA-L advantageous for studying the patterns of projection and the modes of termination of select groups of neurons in their target nuclei. We used PHA-L to study the extraretinal innervation of the cat's dorsal lateral geniculate nucleus, a thalamic visual center. Although much is known about the retinal contribution to geniculate synaptic circuitry, relatively little is known about other sources of innervation, even though these provide the majority of synaptic terminals in the nucleus (Guillery: Z. Zellforsch., 96:1-38, 39-48, 1969; Wilson et al.: Proc. R. Soc. Lond. [Biol.], 221:441-436, 1984). We used both light and electron microscopy to describe synaptic circuitry from three extraretinal sources of projections to the lateral geniculate nucleus: the visual cortex, the perigeniculate nucleus, and the parabrachial region of the brainstem. Cortical terminals labeled with PHA-L were small and formed asymmetrical synaptic contacts onto small-caliber dendrites of geniculate neurons. Perigeniculate terminals formed symmetrical synaptic contacts primarily onto small-caliber dendrites, but some synapses were also formed onto the proximal, retinorecipient portions of geniculate dendrites. Parabrachial terminals synaptically contacted the retinorecipient portions of dendritic appendages and shafts, small-caliber dendrites, and the specialized dendritic (F2) terminals of geniculate interneurons. The symmetry of the parabrachial synaptic contacts was variable and was related to the postsynaptic target. Contacts onto dendritic appendages were asymmetrical while those onto dendritic shafts and F2 terminals were symmetrical. Our data suggest that in unlabeled material these brainstem terminals would be difficult to distinguish from cortical or perigeniculate profiles. The positioning of the parabrachial input onto the retinorecipient portions of geniculate dendrites indicates that this projection is well situated to control primary retinal transmission through the nucleus, while the location of most cortical and perigeniculate innervations implicates them in secondary feedback interactions or other aspects of geniculate function.

Quantitative immunogold analysis reveals high glutamate levels in synaptic terminals of retino-geniculate, cortico-geniculate, and geniculo-cortical axons in the cat

A postembedding immunogold procedure was used to estimate quantitatively, at the electron-microscopical level, the intensity of glutamate (GLU) immunoreactivity in different identifiable profiles of the lateral geniculate nucleus (LGN) and perigeniculate nucleus (PGN) of the cat. Synaptic terminals of retinal and cortical origins in the LGN, and of axon collaterals of geniculo-cortical relay cells in the PGN, were identified by previously determined ultrastructural features. Processes of interneurons or relay cells were identified by being immunoreactive or non-immunoreactive, respectively, in serial thin section reacted with a GABA antibody. The results showed that synaptic terminals of geniculo-cortical relay cells in the PGN have significantly higher levels of GLU immunoreactivity than their parent somata or dendrites in the LGN; this suggests transmitter storage of this amino acid in these terminals. By contrast, synaptic terminals of interneurons did not show enrichment of GLU relative to their parent somata. This argues against the possibility that the relative enrichment of GLU in relay cells terminals is due to factors other than presynaptic storage. In addition, axon collateral terminals of relay cells in the pGN, as well as retinal and cortical terminals in the LGN, showed significantly higher GLU immunoreactivity than GABAergic terminals. These immunocytochemical results suggest that GLU is a neurotransmitter in the retino-geniculate, cortico-geniculate, and geniculo-cortical pathways in the cat.

Synaptic connections between corticogeniculate axons and interneurons in the dorsal lateral geniculate nucleus of the cat

Anatomical evidence is provided for direct synaptic connections by axons from visual cortex with interneurons in lamina A of the cat's dorsal lateral geniculate nucleus. Corticogeniculate axon terminals were labeled selectively with 3H-proline and identified by means of electron microscopic autoradiography. Interneurons in the lateral geniculate nucleus were stained with antibodies that had been raised against gamma aminobutyric acid (GABA). We found that corticogeniculate terminals synapsed with dendrites stained positively for GABA about three times as often as with unstained dendrites. Of the corticogeniculate terminals that contacted GABA-positive dendrites, 97% made synaptic connections with dendritic shafts. Only 3% synapsed with F profiles, the vesicle-filled dendritic appendages characteristic of lateral geniculate interneurons. These results suggest that the corticogeniculate pathway in the cat is directed primarily at interneurons and is organized synaptically to influence the integrated output of these cells, rather than the local interactions in which their dendritic specializations participate.

Organization of cholinergic synapses in the cat's dorsal lateral geniculate and perigeniculate nuclei

In the preceding article, we showed that cholinergic fibers originating from the brainstem reticular formation provide a dense innervation of the lateral geniculate nucleus. In this report we describe the ultrastructure of these fibers and their relations with other elements in the neuropil of the lateral geniculate nucleus. Cholinergic fibers were labeled with an antibody to choline acetyltransferase (ChAT), the synthesizing enzyme for acetylcholine (ACh). In the A-laminae of the lateral geniculate nucleus, ChAT + profiles are small and contain tightly packed, mostly round vesicles. Some end in encapsulated synaptic zones where they form asymmetrical synaptic contacts with processes of both projection cells and interneurons. Others form synapses upon the shafts of dendrites. Of the four classical types of vesicle-containing profiles identified by Guillery (Z. Zellforsch. Mikrosk. 96:1-38, '69; Vision Res. [Suppl.] 3:211-227, '71), ChAT + profiles most closely resemble RSD profiles (Round vesicles, Small profile, Dark mitochondria). However, as a population, ChAT + profiles can be distinguished from the unlabeled population of RSD profiles because they are larger in size, contain more mitochondria, and make synapses with smaller postsynaptic membrane specializations. Each of these differences is statistically significant and together they indicate that ChAT + profiles are a distinct morphological type of synaptic profile. ChAT + profiles in the perigeniculate nucleus resemble those found in the lateral geniculate nucleus; they also make synapses with obvious postsynaptic thickenings.

Quantitative immunogold analysis reveals high glutamate levels in retinal and cortical synaptic terminals in the lateral geniculate nucleus of the macaque

An immunogold procedure has been used on ultrathin sections of the parvo- and magnocellular layers of the dorsal lateral geniculate of the rhesus monkey to estimate quantitatively at the electron microscopic level the intensity of immunoreactivity to an antibody against glutamate over profiles of retinal, cortical, GABAergic synaptic terminals and glial cells. GABAergic terminals were identified directly by immunogold reactivity to a GABA antibody or by ultrastructural features. The results showed that in both of the main subdivisions of the geniculate the densities of immunogold particles over cortical and retinal terminals were about two- to three-fold higher than those over GABAergic terminals or glial profiles. In addition, cortical and retinal terminals showed higher positive correlations of glutamate immunogold particle densities to synaptic vesicle densities than did GABAergic terminals. These differences suggest higher and lower concentrations of glutamate corresponding to transmitter and metabolic pools of this amino acid in axon terminals of retinal and cortical origins versus GABAergic terminals, respectively, in the dorsal lateral geniculate nucleus of the macaque.

Fine structure of synaptogenesis in the vertebrate central nervous system

This article reviews studies of the formation of synaptic junctions in the vertebrate central nervous system. It is focused on electron microscopic investigations of synaptogenesis, although insights from other disciplines are interwoven where appropriate, as are findings from developing peripheral and invertebrate nervous systems. The first part of the review is concerned with the morphological maturation of synapses as described from both qualitative and quantitative perspectives. Next, epigenetic influences on synaptogenesis are examined, and later in the article the concept of epigenesis is integrated with that of hierarchy. It is suggested that the formation of synaptic junctions may take place as an ordered progression of epigenetically modulated events wherein each level of cellular affinity becomes subordinate to the one that follows. The ultimate determination of whether a synapse is maintained, modified or dissolved would be made by the changing molecular fabric of its junctional membranes. In closing, a hypothetical model of synaptogenesis is proposed, and an hierarchial order of events is associated with a speculative synaptogenic sequence. Key elements of this hypothesis are 1) epigenetic factors that facilitate generally appropriate interactions between neurites; 2) independent expression of surface specializations that contain sufficient information for establishing threshold recognition between interacting neurites; 3) exchange of molecular information that biases the course of subsequent junctional differentiation and ultimately results in 4) the stabilization of synaptic junctions into functional connectivity patterns.