Formation of specific afferent connections in organotypic slice cultures from rat visual cortex cocultured with lateral geniculate nucleus.

The development of the cerebral cortex involves the specification of intrinsic circuitry and extrinsic connections, the pattern of inputs and outputs. To investigate the development of a major afferent input to the cortex, we studied the formation of thalamocortical connections in an organotypic culture system. Slices from the lateral thalamus of young rats were cocultured with slices from the visual cortex. Thalamocortical projections in vitro were examined anatomically with fluorescent dyes and physiologically with electrophysiological and optical recording techniques. Axons emerged from thalamic explants radially in all directions. The outgrowth of thalamic fibers and the course of the axonal trajectories were not influenced by the presence of the cocultured cortex. Only those thalamic axons that happened to grow toward the cortical slices invaded their target tissue. Thalamocortical projection cell in vitro had the characteristic morphology of thalamic relay neurons. Cells with the morphology of interneurons were present in thalamic explants, but these neurons did not project to the cocultured cortex. Thalamocortical axons in vitro terminated in their appropriate cortical target layer, formed axonal arbors, and made functional synaptic contacts. Such specific connections between thalamic neurons and their cortical target cells were established regardless of whether thalamocortical axons invaded the cortex from the white matter side or from the pial surface. These results suggest that thalamic projection neurons have an innate mechanism that allows them to recognize their cortical target cells. Thus, intrinsic factors play a significant role in the laminar specification of cortical connections during development.     

Changes in the thalamocortical connectivity during anesthesia-induced transitions in consciousness

Thalamocortical networks play an important role in information integration during consciousness. However, little is known about how the information flows between the thalamus and the cortex are affected by a loss of consciousness. To investigate this issue, we analyzed effective connectivity between the cortex and the thalamus in animals during anesthesia-induced transitions. By recording the electroencephalogram from the primary motor and the primary somatosensory cortex and by recording local field potentials from the ventral lateral and the ventrobasal thalamic nuclei, we evaluated changes in the conditional Granger causality between cortical and thalamic electrical activity as mice gradually lost consciousness from the use of anesthesia (ketamine/xylazine). The point of loss of consciousness was indicated by a moment of loss of movement that was measured using a head-mounted motion sensor. The results showed that 65% of the thalamocortical information flows were changed by anesthesia-induced loss of consciousness. Specifically, the effective connectivity between the cortex and the ventral lateral thalamus was altered such that the primary motor and the primary somatosensory cortex Granger-caused the ventral lateral thalamus before loss of consciousness whereas the ventral lateral thalamus Granger-caused the primary motor cortex and the primary somatosensory cortex after loss of consciousness. In contrast, the primary somatosensory cortex consistently Granger-caused the ventrobasal thalamus, regardless of the loss of consciousness. These results suggest how information flows change across the thalamocortical network during transitions in consciousness.     

In vitro labeling of axonal projections in the mammalian central nervous system

Horseradish peroxidase crystals or HRP-NP40 detergent chips were directly applied to brain slices from mice to label primary afferent fibers and their terminal arbors in the brainstem trigeminal complex, and neurons of the ventrobasal complex of the thalamus, their axons in the internal capsule, and their terminals in the primary somatosensory cortex. Anterograde and retrograde labeling of fibers, as well as retrograde labeling of somata, were observed. In vitro labeling of selected non-trigeminal structures and fiber pathways was also demonstrated. Experimental variables have been dealt with in some detail, as have specific advantages and disadvantages of the technique. This in vitro HRP labeling method for neuronal fiber systems is a useful adjunct to currently employed in vivo labeling techniques in the mammalian central nervous system.     

Overlapping corticostriatal projections from the rodent vibrissal representations in primary and secondary somatosensory cortex

To determine whether the neostriatum receives overlapping projections from two somatosensory cortical areas, the anterograde tracers biotinylated dextran amine (BDA) and fluoro-ruby (FR) were injected into the whisker representations of primary (SI) and secondary (SII) somatosensory cortex. Reconstructions of labeled terminals and their beaded varicosities in the neostriatum and thalamus were analyzed quantitatively to compare the extent of overlapping projections to both subcortical structures. Corticostriatal projections from focal sites in both somatosensory areas exhibited substantial amounts of divergence within the dorsolateral neostriatum. Most of the labeled terminals were concentrated in densely packed arborizations that occupied lamellar-shaped regions along the dorsolateral edge of the neostriatum. Tracer injections in both cortical areas also produced dense anterograde and retrograde labeling in the thalamus, especially in the ventrobasal complex (VB) and in the medial part of the posterior (POm) nucleus. Because these thalamic regions are topographically organized and have reciprocal connections with corresponding representations in both SI and SII, the amount of labeled overlap in the thalamus was used to indicate the degree of somatotopic correspondence at the SI and SII injection sites. We found that the proportion of overlapping projections to the neostriatum was moderately correlated with the amount of overlap observed in the thalamus. This result strongly indicates that specific sites in the dorsolateral neostriatum receive convergent projections from corresponding somatotopic representations in SI and SII, but also suggests that some of the corticostriatal divergence may reflect neostriatal integration of somatosensory information from noncorresponding representations in SI and SII.     

The Course and Termination of Corticothalamic Fibres Arising in the Visual Cortex of the Rat

Corticothalamic axons have been studied in adult Lister hooded rats with single or dual injections of tracers into the visual cortex. Labelled axons leave medial and lateral injection sites in separate or partially overlapping bundles along parallel trajectories in the subcortical white matter. In the internal capsule they converge and both bundles enter roughly the same sector of the thalamic reticular nucleus (TRN). Their reticular terminal fields, however, differ. Axons from a medial injection site innervate more lateral parts of the TRN than do the axons from lateral injection sites. The most medial third of the TRN is not innervated from area 17 but receives a topographically arranged input from peristriate cortex (Crabtree and Killackey, 1989, Eur J. Neurosci., 1, 94-109; Coleman and Mitrofanis, 1996, EWK J. Neurosci., 8, 388-404). The two groups of axons then separate in the dorsal thalamus, axons from medial parts of visual cortex turning caudally into lateral regions of the lateral geniculate nucleus, whereas fibres from more lateral cortex continue into medial parts of the nucleus. Connolly and van Essen (1984, J. Comp. Neurol., 226, 544-564) and Nelson and LeVay (1985, J. Comp. Neurol., 240, 322-330) have shown that in the geniculocortical pathway the two groups of fibres cross over in the subcortical white matter, probably in the region of the subplate. We show that the corticothalamic pathway also has a crossing, but it occurs in, or close to, the diencephalon itself, in the region of the perireticular nucleus. This result suggests that each of these pathways, the geniculocortical and the corticogeniculate, may undergo reorganization within distinct cerebral zones, one diencephalic for the corticothalamic axons and the other telencephalic for the thalamocortical axons.     

Overlapping corticostriatal projections from the rodent vibrissal representations in primary and secondary somatosensory cortex

To determine whether the neostriatum receives overlapping projections from two somatosensory cortical areas, the anterograde tracers biotinylated dextran amine (BDA) and fluoro-ruby (FR) were injected into the whisker representations of primary (SI) and secondary (SII) somatosensory cortex. Reconstructions of labeled terminals and their beaded varicosities in the neostriatum and thalamus were analyzed quantitatively to compare the extent of overlapping projections to both subcortical structures. Corticostriatal projections from focal sites in both somatosensory areas exhibited substantial amounts of divergence within the dorsolateral neostriatum. Most of the labeled terminals were concentrated in densely packed arborizations that occupied lamellar-shaped regions along the dorsolateral edge of the neostriatum. Tracer injections in both cortical areas also produced dense anterograde and retrograde labeling in the thalamus, especially in the ventrobasal complex (VB) and in the medial part of the posterior (POm) nucleus. Because these thalamic regions are topographically organized and have reciprocal connections with corresponding representations in both SI and SII, the amount of labeled overlap in the thalamus was used to indicate the degree of somatotopic correspondence at the SI and SII injection sites. We found that the proportion of overlapping projections to the neostriatum was moderately correlated with the amount of overlap observed in the thalamus. This result strongly indicates that specific sites in the dorsolateral neostriatum receive convergent projections from corresponding somatotopic representations in SI and SII, but also suggests that some of the corticostriatal divergence may reflect neostriatal integration of somatosensory information from noncorresponding representations in SI and SII.     

Thalamocortical connections in newborn mice

Thalamocortical axons reach the developing neocortex and become distributed within the cortical subplate during the third week of gestation. The present study is an analysis of the organization of connections that link thalamus and cortical subplate (corresponding to future layers V and VI) at birth. This age antedates the ascent of thalamic axons to contact cells of the supragranular layers, their principal targets in the adult cortex. At birth thalamic nuclear subdivisions are explicit; field-characteristic cytoarchitectonic features, relating principally to the infragranular layers, delineate the majority of neocortical fields. The projection of principal relay nuclei upon the majority of fields of the cerebral convexity has been mapped by means of retrograde transport of HRP. Nucleus-to-field interrelationships as well as topologic order of the overall thalamic projection prove to be identical to that in the adult animal. The neonatal projection appears to be somewhat more divergent than that of the adult.     

A simplified approach to retrograde/anterograde axonal labeling using combined injections of horseradish peroxidase and ibotenic acid.

Combined injections of ibotenic acid and horseradish peroxidase (HRP) were made into the region of the mouse ventrobasal thalamus that is related to the large mystacial vibrissae. Examination 4 and 5 days later of the corresponding area of the primary somatosensory cortex (i.e., barrel cortex), in thick and in thin sections, showed it to contain numerous corticothalamic projection cells the somata, dendrites and axons of which were densely labeled by the retrograde transport of HRP. Analysis of serial thin sections showed that thalamocortical axon terminals, which had degenerated in response to the injection of ibotenic acid, formed approximately 20% of the asymmetrical synapses in barrel cortex. The fine structure and distribution in cortex of degenerating thalamocortical axon terminals and of intrinsic HRP-labeled corticothalamic axon terminals were identical to those reported in previous studies in which the injection of HRP into the thalamus was combined with the making of electrolytic lesions. This indicates that injecting ibotenic acid is an effective replacement for electrolytic lesioning of the thalamus. The combined injection of ibotenic acid and HRP represents a new and efficient approach for studying reciprocal projection pathways.     

Thalamic connectivity of the primary motor cortex of normal and reeler mutant mice

Reeler, an autosomal recessive mutation in mice, causes cytoarchitectonic abnormalities of the cerebral cortex, which are characterized by malposition of neurons. Retrograde and anterograde transport of horseradish peroxidase (HRP) was employed to examine the reciprocal connectivity between the hindlimb area of the primary motor cortex (MI) and thalamus of normal and reeler mutant mice. In the normal mouse, most of the cells labelled after HRP injection into the hindlimb area of MI were located in the ventrolateral nucleus, the lateral division of the ventrobasal nucleus, the central lateral, paracentral and central intralaminar nuclei, and the medial division of the posterior complex. HRP reaction product anterogradely transported was also observed in the same nuclei and in the thalamic reticular nucleus. In the reeler mutant mouse, retrogradely labelled neurons and anterogradely labelled terminals were again found in the nuclei referred to above, and the distribution pattern and morphology of HRP-filled neurons were also similar to those of normal controls. The present results suggest therefore that the normal reciprocal connectivity between MI (hindlimb representation) and thalamus is preserved in the reeler mouse. That is to say, dislocated cortical neurons appropriately project to their target nuclei of the thalamus, and conversely, thalamic neurons send their axons precisely to their target cortical areas of the radially disorganized cortex.     

Efferent connections of the centromedian and parafascicular thalamic nuclei: an autoradiographic investigation in the cat

The efferent projections of the centromedian and parafascicular (CM-Pf) thalamic nuclear complex were analyzed by the autoradiographic method. Our findings show that the CM-Pf complex projects in a topographic manner to specific regions of the rostral cortex. These fibers distribute primarily to cortical layers I and III; however, the projection to layer I is more extensive. Following an injection into the rostral portion of the CM-Pf complex, label is found within the lateral rostral cortex, particularly within the presylvian, anterior ectosylvian, and anterior lateral sulci, and within the rostral medial cortex where label is present within the cruciate and anterior splenial sulci and anterior cingulate gyrus. An injection into the caudal dorsal portion of the CM-Pf complex results in label within the more ventral portions of the rostral lateral cortex where it is present within the anterior sylvian gyrus, presylvian regions, and gyrus proreus; and within the rostral medial cortex, where it is present within the rostral cingulate gyrus, and within the cruciate sulcus, and an extensive region ventral to the cruciate sulcus which includes the anterior limbic area. Injections into the caudal ventral portion of the CM-Pf complex result in virtually no cortical label, although a few labeled fibers are found in the subcortical white matter. The subcortical projection from the CM-Pf complex terminates within the caudate nucleus, putamen, globus pallidus, subthalamic nucleus, zona incerta, fields of Forel, hypothalamus, thalamic reticular nucleus, and rostral intralaminar nuclei. Prominent silver grain aggregates are also present within the ventral lateral, ventral anterior, ventral medial, and lateral posterior nuclei, and ventrobasal complex. The aggregates in the thalamus appear to be fibers of passage, but whether these are also terminals cannot be determined with the techniques used in the present study.     

Cytoarchitecture and somatic sensory connectivity of thalamic nuclei other than the ventrobasal complex in the cat

This study is a re-examination, using autoradiographic and axonal degeneration methods, of the distribution of spinal, dorsal columnlemniscal and cortico-thalamic fibers within the thalamus of the cat. The emphasis was placed upon making an exact cytoarchitectonic delineation of regions outside the ventrobasal complex which receive spino-thalamic fibers and corticofugal fibers arising in the sensory-motor regions of the cerebral cortex. It is concluded, in agreement with Boivie ('70, '71a,b) that spino-thalamic fibers arising below the level of the lateral cervical nucleus terminate, not in the ventrobasal complex, but in a recognizable portion of the ventrolateral complex adjacent to the ventrobasal complex. This region appears to be also the thalamic relay for Group I muscle afferents. Combined experiments in which, in the same animal, the dorsal column-lemniscal path was labeled autoradiographically and the spino-thalamic path by the Nauta method, indicate virtually no overlap of the two systems in the ventral nuclear complex. Spinal fibers also end in the medial division of the posterior group (Pom), which extends posteriorly as a small-celled zone along the ventromedial aspect of the magnocellular medial geniculate nucleus. Reports of spinal terminations in the magnocellular nucleus proper may result from a failure to recognize the extent of Pom. A third part of the thalamus receiving spinal fibers consists of a posteriorly situated group of large, deeply staining cells belonging to the central lateral nucleus which lie mainly posterior to the internal medullary lamina and which, as Mehler ('69) has mentioned, have previously been confused with the centre médian and parafascicular nuclei. Corticofugal fibers arising in the somatic sensory cortex terminate in both the ventrobasal complex and the spinal part of the ventrolateral complex as well as in the central lateral nucleus and Pom and this confirms the work of Rinvik ('68a). Those arising in the motor cortex terminate only in (a different part of) the ventrolateral complex and in the centre médian nucleus.     

Axon collaterals in the thalamic reticular nucleus from thalamocortical neurons of the rat ventrobasal thalamus

Thalamocortical relay neurons from the rat ventrobasal nucleus were identified physiologically and injected intracellularly with horseradish peroxidase. The axons of these cells were followed through serial sections in order to determine if collaterals were given off within the ventrobasal nucleus or the thalamic reticular nucleus. No local collaterals were seen in the ventrobasal nucleus, thus indicating that interactions between relay cells in this nucleus are minimal. Of axons that could be followed into the internal capsule, 76% gave off visible collaterals in the thalamic reticular nucleus. Half of these axons had collaterals showing extensive branching with the potential of innervating a large number of thalamic reticular neurons. The other half had short, simple branches of restricted extent. No correlations were found between the physiological properties of a cell and the existence or extent of axon collaterals. These results describe the anatomical basis for the initial part of a feedback loop through the thalamic reticular nucleus that provides the major inhibitory influence on rat ventrobasal neurons.     

Somatotopic reciprocal connections between the somatosensory cortex and the thalamic Po nucleus in the rat

Retrogradely labeled neurons are observed in the posterior group of the thalamus (Po) after injection of wheatgerm agglutinin-horseradish peroxidase in the rat somatosensory cortex. These neurons are organized in rods elongated rostrocaudally, defining a clear somatotopic map. Injections of tritiated leucine in the somatosensory cortex indicate that these somatotopically organized connections are reciprocal. Injections of tritiated leucine in the dorsal column nuclei label afferent fibers in a small area dorsal to Po but not in the core of the nucleus. Po does not receive direct projections of ascending somatosensory afferents. It is hypothesized that this thalamic area participates in a thalamo-cortico-thalamic loop.     

Severe neuronal losses with age in the parietal cortex and ventrobasal thalamus of mice transgenic for the human NF-L neurofilament protein

Transgenic mice expressing human light neurofilament protein (NF-L) display early perikaryal accumulations of disarrayed neurofilaments in layers II/III of the parietal cortex and in the ventrobasal complex of thalamus. This cytoskeletal abnormality, reflected by strong NF-L immunoreactivity, is transient in the developing cortex but persists until old age in the thalamus. To investigate whether it leads to neuronal death, the unbiased cell counting method of the dissector was applied to the parietal cortex and the thalamus of normal and transgenic mice at various postnatal (P10, P20, P90) and advanced ages (14-18 months). Similar data were also obtained from the primary visual cortex free of NF-L accumulation. Compared with normal, the total number of neurons in the parietal (but not occipital) cortex of transgenic mice showed little change during the postnatal period, but decreased markedly with old age, particularly in layers II/III. Severe neuronal loss was also documented in the thalamic ventrobasal complex of aged transgenic mice. The delayed neuronal death in the parietal cortex, occurring long after recovery from the NF-L accumulations, was suggestive of a combination of deleterious factors, including the early overproduction of neurofilament protein and subsequent loss of afferent input from the affected somatosensory thalamic nuclei. Furthermore, strong accumulation of lipofuscin in the neurons of aged transgenic mice suggested that oxidative stress partakes in the mechanisms through which NF-L overproduction compromises neuronal viability.     

Development of the pyramidal tract in the hamster. I. A light microscopic study

The development of the pyramidal tract and other projections from the sensorimotor cortex was studied in the postnatal hamster with both (³H) proline and horseradish peroxidase (HRP) as anterograde tracers. In the 1-day-old animal labeled axons extend as far as the pons. Other corticofugal fibers have penetrated into the corpus striatum and the thalamus. By 2 days postnatally, the pyramidal tract has grown to midmedullary levels and there is substantial retrograde (HRP) and anterograde labeling in the thalamus. The pyramidal decussation is formed at 3 days of age and by 4 days the pyramidal tract has descended in the dorsal funiculus as far as midcervical spinal cord. Corticofugal fibers invade the pontine nuclei at 4 days and both the dorsal column nuclei and the superior colliculus at 6 days of age. At 6 days the pyramidal tract can be traced to midthoracic levels of the spinal cord, by 8 days the tract reaches lumbar levels, and by 14 days it has completed its caudal growth to the coccygeal spinal cord. Fibers first penetrate the gray matter of a given spinal cord level approximately 2 days after the tract has grown past that level in the dorsal funiculus. Pyramidal fibers continue their lateral growth into the dorsal horn at all levels of the cord throughout the third postnatal week such that by 21 days of age the pyramidal tract appears similar to that of the adult. The projections from sensorimotor cortex to the pontine nuclei, the superior colliculus, and the dorsal column nuclei appear to have a pattern similar to that of the adult soon after the fibers grow into these structures. There is a consistent delay of 2 to 3 days between the arrival of the pyramidal tract axons in the white matter adjacent to target structures and their innervation of a given terminal field. The pyramidal tract grows more quickly through the dorsal funiculus of the spinal cord than it does along the ventral surface of the medulla. Extensive elongation of pyramidal tract axons is achieved long before the growth and differentiation of the sensorimotor cortical neurons from which they originate. Finally, the pyramidal tract appears to grow as a compact bundle and not by the addition of temporally staggered groups of fibers. The relatively protracted period of innervation of the spinal cord by the pyramidal tract coupled with the immaturity of the cortical neurons at birth may be factors contributing to the significant regrowth of pyramidal tract axons severed early in development.     

Transplantation of embryonic occipital cortex to the tectal region of newborn rats: A light microscopic study of organization and connectivity of the transplants

Occipital cortex was taken from fetal rats and transplanted to the tectal region of newborn rats, where it developed a specific structural identity reflecting in part its cortical origin. The implants showed locally distributing intratransplant connections, and the majority formed connections with defined regions of the host cerebral cortex and the brainstem. A sparse afferent projection from the host had its origin in visual, somatosensory, and cingulate areas of the cortex, pretectum, superior colliculus, central gray, hypothalamus, pontine reticular formation, raphe nuclei, and the locus coeruleus. No input was identified from either the retina or the dorsal thalamus. Efferent fibers were observed in normal fiber preparations as compact bundles running through the host brainstem along two main routes, one group of bundles in a dorsal position and a second group more ventral. Efferent fibers traveling rostrally along the first pathway distributed in the ower part of the stratum griseum superficiale and in intermediate laminae of the superior colliculus, and in some cases they reached the pretectum and the lateral posterior thalamic nucleus. Deep efferent fibers ran rostrally and caudally in the central gray, and in some cases laterally directed fibers were seen to distribute in the midbrain tegmentum and reticular formation, in one case reaching the pontine gray. The finding that most afferent and many efferent connections of cortical transplants are uncharacteristic of normal cortex stands in marked contrast to retina and tectum, which, when transplanted to the same region, make relatively normal connections.     

Is there capacity for anatomical and functional repair in the adult somatosensory thalamus?

The capacity for structural and functional remodeling in damaged adult CNS sensory systems can be studied by replacement of neurons in damaged structures by fetal cells from these anatomical origins. For integration to take place, the replacement paradigm assumes that (a) reconnection of adult host afferent fibers onto developing neurons is possible and (b) that the correct molecular signals exist also in the adult brain for fetal neurons to extend axons and pattern synaptic contacts. We have tried to answer some of these fundamental questions by using neuronal depletion models followed by neuronal replacement in the adult rat CNS (Isacson et al. 1984. Nature (London) 311: 458-460; Isacson et al. 1988. Prog. Brain Res. 78: 13-27; Nothias et al. 1988. Brain Res. 461: 349-354; Peschanski and Isacson. 1988. J. Comp. Neurol. 274: 449-463; Sofroniew et al. 1990. Prog. Brain Res. 82: 313-320). In one such model, kainic acid infusions deplete the ventrobasal complex (VB) of all neurons projecting to the somatosensory cortex, while afferent axons from the lemniscal and monoaminergic systems remain in the area. Direct implantation of fetal neurons (gestation age 15-16) of ventrobasal destination allows reconnection of circuitry to take place at the thalamic level, as studied by anatomical tracers, electron microscopy, and functional 2-deoxyglucose studies, while fetal thalamic VB neurons appear less likely to grow through the internal capsule toward the cortical level.     

Cytology and distribution in normal human cerebral cortex of neurons immunoreactive with antisera against neuropeptide Y

The frontal, parietal, and temporal cortices in normal human brains (Brodmann areas 10, 7a, 7b, and 21) are well endowed with numerous neurons, identifiable by immunoreactivity with antisera against the 36-amino acid brain peptide neuropeptide Y (NPY). These neurons with rare exception are small, intracortical, nonspiny neurons, 12-20 microns in somatic size, with long slender dendrites and exuberant axon plexuses exhibiting finely beaded varicosities. The cells are rarest in layers I and II, are found with frequency in the lower cortical layers (IV-VI) and in significant numbers in the subcortical white matter. Within the cortex the axonal plexuses of these peptide neurons rise straight up into the upper cortical layers or descend deep into the white matter. In layers I and II, numerous other lengthy axons, some possibly of extracortical afferent origin, run along the pial surface at right angles to the axial ones running perpendicular to the cortex. This endowment of peptide neurons and their processes forms a rich network in the cerebral cortex, relating with one another in complex fashion within palisades of terminals as well as with the other cortical neurons not labeled by these methods. It remains to be shown what functions these NPY neurons have individually and in their remarkable networks, and how they are altered in neurological disease.     

Orderly anomalous retinal projections to the medial geniculate,ventrobasal, and lateral posterior nuclei of the hamster

Experiments were performed to determine (1) under what conditions early brain surgery can cause sensory afferents to the thalamus to form connections at abnormal thalamic sites and (2) the extent to which such ectopic projections are receptotopically organized. In newborn Syrian hamsters, two of the retina's principal synaptic targets, the superior colliculus and dorsal lateral geniculate nucleus, were destroyed, respectively, by a direct lesion and by retrograde degeneration following a lesion of the occipital cortex. In the same brains, alternative terminal space for the retinofugal axons was made available in auditory (medial geniculate) or somatosensory (ventrobasal) thalamic nuclei by lesions of ascending auditory or somatosensory pathways, respectively; additional terminal space was made in the lateral posterior nucleus by degeneration of afferents from the superior colliculus. The projections of the contralateral retina were traced in neonatally operated adults by making one or two small peripheral retinal lesions and intraocular injections of ³H-proline 5 days and 1 day, respectively, prior to sacrifice. The neonatal surgery reliably produced anomalous crossed retinal projections to the partially deafferented structures. These projections terminate preferentially at the nuclear surfaces. Computer reconstructions from serial sections demonstrated several signs of spatial order suggestive of receptotopic organization in the anomalous retinothalamic projections. In order of increasing stringency, these signs (which are not mutually exclusive) are: (1) In each nucleus, a restricted retinal sector gives rise to a limited part of the abnormal projection. (2) In each nucleus, different parts of the retina give rise to different parts of the anomalous projection. (3) In each nucleus, there is a more or less consistent polarity of the anomalous connection. Each small retinal sector appears to be represented along a “line of projection” in each of its abnormal thalamic targets, as it normally is in the dorsal and ventral lateral geniculate nuclei and in the superior colliculus. In some brains, some of the abnormal projections produce only a partial representation of the retina. However, in a single animal, a retinal sector not represented in the anomalous projections to one nucleus can contribute to the abnormal connections with another nucleus. In additional experiments, an attempt was made to direct developing auditory and somatosensory fibers normally terminating in the medial geniculate and ventrobasal nuclei, respectively, to anomalous thalamic targets. The axons were deprived of some of their normal thalamic sites of termination and alternative terminal space was made available in another thalamic sensory nucleus. These experiments failed to produce reliable evidence of ectopic auditory or somatosensory thalamic projections. The anomalous retinal projections to nuclei that normally recieve little (lateral posterior) or no (medial geniculate, ventrobasal) optic tract input, show that the preference of retinal axons for their normal targets is relative, not absolute. The orderliness of the ectopic projections opposes the hypothesis that the formation of retinotopic connections depends upon the matching of a set of signals distributed among the retinofugal fibers and a corresponding set of cues unique to the normal terminal fields of optic axons. The results are consistent with the formation of receptotopic connections by interactions among developing axons and suggest the action of additional factors that determine the terminal sites and organization of central neuronal connections.