Retinotopically inappropriate synapses of subnormal density formed by surgically misdirected optic fibers in goldfish tectum

Computational Predictions on the Receptive Fields and Organization of V2 for Shape Processing

It has been more than 40 years since the first studies of the secondary visual cortex (V2) were published. However, no concrete hypothesis on how the receptive field of V2 neurons supports general shape processing has been proposed to date. Using a computational model that follows the principle of self-organization, we advance two hypotheses in this letter: (1) typical V2 orientation-selective receptive field contains a primary orientation and a secondary orientation component, forming a corner, a junction, or a cross; and (2) V2 columns with the same primary orientation form contiguous domains, divided into subdomains that prefer different secondary orientations. The first hypothesis is consistent with existing experimental evidence, and both hypotheses can be tested with current techniques in animals. In this manner, computational modeling can be used to provide verifiable predictions that eventually allow us to understand the role of V2 in visual processing.

Growth dynamics and morphology of regenerating optic fibers in tectum are altered by injury conditions: an in vivo imaging study in goldfish

The dynamic behavior of axons in systems that normally regenerate may provide clues for promoting regeneration in humans. When the optic nerve is severed in adult goldfish, all axons regenerate back to the tectum to reestablish accurate connections. In adult mammals, regeneration can be induced in optic and other axons but typically few fibers regrow and only for short distances. These conditions were mimicked in the adult goldfish by surgically deflecting 10-20% of optic fibers from one tectum into the opposite tectum which was denervated of all other optic fibers by removing its corresponding eye. At 21-63 days, DiI was microinjected into retina to label a few fibers and the fibers were visualized in the living fish for up to 5-7 h. The dynamic behavior and morphology of these regenerating deflected fibers were analyzed and compared to those regenerating following optic nerve crush. At 3-4 weeks, deflected fibers were found to form more branches and to maintain many more branches than crushed fibers. Although both deflected and crushed fibers exhibited stochastic growth and retraction, deflected fibers spent more time growing but grew for less distance. At 2 months, both deflected and crushed fibers became much more stable. These results show that the morphology and behavior of fibers regenerating into the same target tissue can be substantially altered by the injury conditions, that is, they show state-dependent plasticity. The morphology and behavior of the deflected fibers suggest they were impaired in their capacity to grow to their correct targets.

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.

Cerebellar guidance of premotor network development and sensorimotor learning

Single unit and imaging studies have shown that the cerebellum is especially active during the acquisition phase of certain motor and cognitive tasks. These data are consistent with the hypothesis that particular sensorimotor procedures are acquired and stored in the cerebellar cortex and that this knowledge can then be exported to the cerebral cortex and premotor networks for more efficient execution. In this article we present a model to illustrate how the cerebellar cortex might guide the development of cortical-cerebellar network connections and how a similar mechanism operating in the adult could mediate the exportation of sensorimotor knowledge from the cerebellum to the motor cortex. The model consists of a three-layered recurrent network representing the cerebello-thalamocortical-ponto-cerebellar limb premotor network. The cerebellar cortex is not explicitly modeled. Our simulations show that Hebbian learning combined with weight normalization allows the emergence of reciprocal and modular structure in the limb premotor network. Reciprocal connections allow activity to reverberate around specific loops. Modularity organizes the connections into specific channels. Furthermore, we show that cerebellar learning can be exported to motor cortex through these modular and reciprocal premotor circuits. In particular, we simulate developmental alignment of visuomotor relations and their realignment as a consequence of prism exposure. The exportation of sensorimotor knowledge from the cerebellum to the motor cortex may allow faster and more efficient execution of learned motor responses.

Normal activity-dependent refinement in a compressed retinotectal projection in goldfish

When the optic nerve in a goldfish is crushed, regenerating fibers can reform a normal retinotopic projection. Two processes are thought to generate this retinotopic order. One is an activity-independent process, presumed to be some form of substrate-directed growth, which generates rough retinotopy as seen in the early formed projection. The other is an activity-dependent process that generates fine retinotopy during a protracted period of refinement. This projection also displays two other behaviors. One is retinotopic plasticity, in which optic fibers can compensate for retinal or tectal ablations by expanding or compressing into the available tectal space while preserving retinotopic order. These plasticities can dramatically alter the scale of the projection. The other behavior is the formation of fixed synaptic sites in tectum. Optic fibers make a characteristic number of synaptic connections in tectum, which is not changed by increasing the number of invading optic fibers. This has been interpreted to mean that fibers compete for limited synaptic sites. How the two processes that generate order, substrate-directed growth, and activity-dependent refinement might each be affected by the expression of retinotopic plasticity and altered synaptic competition is largely unknown. In particular, it is not known how fine retinotopic order (activity-dependent refinement) might be affected by altering the scale of the projection. Would optic fibers from neighboring ganglion cells converge into the same-sized area of tectum, or would they expand or compress in proportion to the altered scale of the overall map? To explore this issue, the posterior half of tectum of goldfish was removed, and the optic nerve was crushed, thereby forcing regenerating fibers to form a compressed retinotopic projection onto the anterior half of tectum. Under these conditions, optic fibers are also forced to compete for half the normal number of synaptic sites. The effect on retinotopy was monitored at various times during regeneration by making a small spot injection of wheat germ agglutinin-horseradish peroxidase (WGA-HRP) into nasal retina corresponding to fibers that would normally terminate in the missing posterior half of tectum. To distinguish between activity-dependent and activity-independent processes, retinal impulse activity was blocked in some animals by repeated intraocular injections of tetrodotoxin. The initial projection was found to be unaffected by impulse activity. Regardless of activity, nasal fibers failed initially to grow to the most posterior available regions, but instead were dispersed across much of the "incorrect" anterior half of tectum at 30 days.(ABSTRACT TRUNCATED AT 400 WORDS)

The Role of Constraints in Hebbian Learning

Models of unsupervised, correlation-based (Hebbian) synaptic plasticity are typically unstable: either all synapses grow until each reaches the maximum allowed strength, or all synapses decay to zero strength. A common method of avoiding these outcomes is to use a constraint that conserves or limits the total synaptic strength over a cell. We study the dynamic effects of such constraints.

Two methods of enforcing a constraint are distinguished, multiplicative and subtractive. For otherwise linear learning rules, multiplicative enforcement of a constraint results in dynamics that converge to the principal eigenvector of the operator determining unconstrained synaptic development. Subtractive enforcement, in contrast, typically leads to a final state in which almost all synaptic strengths reach either the maximum or minimum allowed value. This final state is often dominated by weight configurations other than the principal eigenvector of the unconstrained operator. Multiplicative enforcement yields a “graded” receptive field in which most mutually correlated inputs are represented, whereas subtractive enforcement yields a receptive field that is “sharpened” to a subset of maximally correlated inputs. If two equivalent input populations (e.g., two eyes) innervate a common target, multiplicative enforcement prevents their segregation (ocular dominance segregation) when the two populations are weakly correlated; whereas subtractive enforcement allows segregation under these circumstances.

These results may be used to understand constraints both over output cells and over input cells. A variety of rules that can implement constrained dynamics are discussed.

Regenerated synapses persist in the superior colliculus after the regrowth of retinal ganglion cell axons

Synapse formation by retinal ganglion cell axons was sought in the superior colliculus of four adult rats 16-18 months after the optic nerve was transected and replaced by a peripheral nerve graft that guided regenerating RGC axons from the eye to the superior colliculus. The terminals of retinal ganglion cell axons were labelled by intravitreal injections of tritiated amino acids and studied by light and electron microscopic autoradiography. We found that (i) retinal ganglion cell axons had extended from the tips of the peripheral nerve grafts into the superior colliculus for approximately 350 microns; (ii) within the superior colliculus, some regenerated retinal ganglion cell axons became ensheathed by CNS myelin; (iii) retinal ganglion cell terminals formed asymmetric synapses with dendrites of neurons in the superficial layers of the superior colliculus, mainly the stratum griseum superficialis. Regenerated (n = 418) and normal retinal ganglion cell terminals (n = 1775) in the superior colliculus were compared in terms of their size (area, perimeter, and maximum diameter), contacts per terminal, contacts per 10 microns terminal perimeter, and post-synaptic structure contacted (dendritic spine, shaft, or soma). No statistically significant differences in the ultrastructural characteristics of the pre-synaptic profiles were apparent between the two groups. The post-synaptic structures contacted by axon terminals were similar in regenerated and control animals, although there were quantitative differences in the distributions of these contacts among dendritic spines and shafts. These results suggest that the regeneration of retinal ganglion cell axons in adult rats can lead to the formation of ultrastructurally normal synapses in the appropriate layers of the superior colliculus. The re-formed connections appear to persist for the life-span of these animals.

Light microscopic study of degenerating cobalt-filled optic axons in goldfish: role of microglia and radial glia in debris removal

Labeling severed axons with cobaltous-lysine ultimately leads to the degeneration of their distal segments. The present study was designed to determine whether microglia and radial glia have comparable roles in the elimination of degenerating axons. Another purpose was to determine whether the cobalt could escape from degenerating axons and enter intact neuronal cells. Optic axons were filled with cobaltous-lysine for 1 day and the retinal projections were examined from 1 to 106 days later. Optimal filling was obtained 1 day postlabeling. The number of filled axons in the optic tract was significantly reduced at the 2-day time point, indicating that many axons had disintegrated. Many axons contained large swellings that resembled cells. However, transneuronally labeled neuronal cells were never observed. Labeled, rounded microglia appeared among the degenerating axons at the 3-day time point, and the microglia changed shape at 5 days. They became elongated and manifested many processes. In addition, the microglia began to move toward, and entered, the ventricles and vasculature. Virtually all the labeled debris was removed between 17 and 28 days following the application of cobalt. The rapidity with which the axons were removed suggests that the cobalt accelerates the degenerative process either directly, or indirectly by accelerating the arrival of phagocytic cells. Radial glia appeared to play a smaller role in debris elimination. They took up labeled debris to a lesser degree than microglia and were briefly labeled. Interestingly, radial glia did not take up cobalt when it was injected intracranially and diffused through the brain. A previous claim of an axosomatic retinotectal projection to cells deep in the stratum periventriculare of a teleost fish is reinterpreted to represent cobalt within radial glia.

Optic synapse number but not density is constrained during regeneration onto surgically halved tectum in goldfish: HRP-EM evidence that optic fibers compete for fixed numbers of postsynaptic sites on the tectum

The number of optic synapses in the half tectum of goldfish was counted by using an improved HRP-labeling protocol and a columnar sampling method that spanned the entire optic innervation layer, S-SO-SFGS. It was previously found by using this procedure in intact tectum that the normal number of optic synapses was regenerated by 30 days and maintained thereafter even in the absence of impulse activity. This suggested that the number of synapses in this system was intrinsically fixed. In order to examine whether this limit was imposed by optic fibers or by target cells, optic synapses were counted in surgically halved tecta which received compressed optic projections consisting of regenerating optic fibers from the entire retina. We reasoned that if synapse number is a function of the number of afferents, then there should be twice the normal number of optic synapses per column; on the other hand, if their number is fixed by target, then their number per column should be normal. We found that the number of optic (labeled) synapses was normal in sample columns from fish at 70 days and 160 days after optic nerve crush. Thus, retinal ganglion cells, on average, formed half as many synapses on the half tectum compared to intact tectum, indicating the number of optic synapses was limited by the tectum. The number of nonoptic (unlabeled) synapses was also found to be normal. By contrast, the S-SO-SFGS was found to be 88-103% thicker compared to normal fish, apparently because of a 20-fold increase in the number of optic fibers. As a result, the density of synapses was about half normal in half tecta, and so, in contrast to synapse number, synaptic density is not constrained during regeneration. We infer from these data that optic fibers compete for limited numbers of postsynaptic sites during regeneration and suggest that this competition promotes neural map refinement and the various plasticities described for this projection.