Abstract

Interactions between growing axons are considered to play important roles for the establishment of precise neuronal con- nections during the development of the nervous system. Here we used time-lapse imaging techniques to examine the behavior of neocortical and thalamic axons when they encounter each other in vitro. Results indicate that axonal growth cones are able to respond to specific cues expressed on the surface of fibers. Thalamic growth cones often extended along the surface of other thalamic axons and, likewise, cortical growth cones formed fascicles with cortical axons. In contrast, after contacts between cortical and thalamic fibers, in most cases growth cones collapsed and retracted from the axons. Collapse assays using membrane preparations from cortical or thalamic explants demonstrated the existence of cell-type specific collapsing factors whose activity was enhanced by a member of the semaphorin protein family, Sema3A (expressed in the thalamocortical pathway), as it increased the rate of homotypic fasciculations and at the same time amplified the segregation between cortical and thalamic axons. The interaction between axonal surface molecules and environmental cues might mediate the segregation of afferent and efferent fiber tracts in the neocortical white matter.

Introduction

During embryonic development, projection neurons in the cerebral cortex and relay neurons in the thalamus establish reciprocal connections between distinct cortical regions and specific thalamic nuclei. Anatomical studies using carbocyanine dyes revealed that pattern formation of afferent and efferent projections in the thalamocortical system is precisely regulated in space and time (McConnell et al., 1989; Catalano et al., 1991, 1996; De Carlos and O'Leary, 1992; Ghosh and Shatz, 1992; Métin and Godement, 1996). In recent years, families of secreted and membrane-associated proteins, the netrins, semaphorins and ephrins, have been discovered which can act as axonal guidance molecules (Kennedy et al., 1994; Serafini et al., 1994; Cheng et al., 1995; ColaMarino and Tessier-Lavigne, 1995; Drescher et al., 1995; Mark et al., 1997; Bagnard et al., 1998). Members of each of these protein families are expressed along the path of thalamic and cortical fibers at the developmental stages when these axons extend towards their targets. Moreover, such molecules are considered as important factors of the genetic control of regionalization and connectivity of the neocortex (Mackarehtschian et al., 1999; Rubenstein et al., 1999). Previous work provided evidence that semaphorins can act both as attractive and repellent guidance cues during the initial patterning of corticofugal projections (Bagnard et al., 1998, 2000), and that netrin-1 might direct these fibers towards the internal capsule, the gateway between cortex and thalamus (Métin et al., 1997; Richards et al., 1997).

Based on observations on the timing of axonal outgrowth, it has been proposed that molecules on axonal membranes con- tribute to the targeting in the thalamocortical system. These studies revealed that the first axons which extend towards subcortical regions arise from the cortical subplate, a transient structure composed of early generated neurons (Marin-Padilla, 1971; Luskin and Shatz, 1985). Axons of subplate cells might therefore provide a scaffold for the later growing cortical plate neurons (McConnell et al., 1989), analogous to the ‘pioneer axons’ originally described in the invertebrate nervous system (Bastiani et al., 1986). Moreover, because descending cortico- fugal and ascending thalamocortical axons arrive at the internal capsule in synchrony, it has also been suggested that these two fiber populations might fasciculate and thereby guide each other towards their final destination (Molnár and Blakemore, 1991, 1995; Molnár et al., 1998).

While such models are attractive in their simplicity, the forma- tion of axon fascicles among and between cortical and thalamic fibers has not yet been demonstrated. In the present study, we used time-lapse imaging to examine the behavior of thalamic and cortical axons when they encounter each other in vitro. On growth-permissive substrates, fasciculations were only observed between axons of the same type (homotypic fasciculation), whereas in many cases thalamic and cortical axons retracted from each other. In vivo, a repellent axonal guidance molecule, semaphorin 3A (Sema3A), is expressed in the internal capsule when projections between the thalamus and cortex are being formed (Skaliora et al., 1998). We found that addition of Sema3A to the substrate not only increased the rate of homotypic fasciculations, but also enhanced heterotypic growth cone retractions. These results are consistent with in vivo observa- tions which indicate that the thalamic and cortical axons form segregated fiber tracts within the white matter.

Materials and Methods

Preparation of Explants

E16 and E18 embryos were obtained after cesarean section of pregnant Lewis rats receiving lethal dose of anesthetics (the day of sperm detection was designated as E1). Embryonic brains were removed and dissected in cold Gey's balanced salt solution (GBSS). Blocks of neocortical tissue were placed on a MacIlwain tissue chopper and cut into 200 μm3 cubes, whereas thalamic tissue was cut into 200-μm-thick slices. Because of their different shapes, it was always possible to distinguish between cortical and thalamic explants in co-cultures. In most experiments explants were prepared from the whole thalamus and the whole neocortex. In some experiments, however, we dissected only the presumptive visual cortex and the lateral geniculate nucleus (LGN) of the thalamus. Explants of the visual cortex were obtained by cutting the posterior part of the neocortical block of tissue. To identify the LGN of the thalamus, we first visualized this nucleus in separate experiments in fixed E18 brains. For this, DiI crystals were injected in the eye of fixed embryos kept at 37°C in 4% paraformaldehyde. After 10–20 days, dye diffusion was sufficient to stain the termination zone in the dorsolateral part of the posterior diencephalon. This region was then dissected in living embryos for time- lapse experiments. Cultures were maintained in a medium consisting of 50% Eagle's basal medium, 25% Hank's balanced salt solution, 25% horse serum supplemented with 0.1 mM glutamine and 6.5 mg/ml glucose (all from Gibco). In most experiments, cortical and thalamic explants were cultured on glass coverslips covered with laminin (1 μg/ml), poly-l-lysine (1 μg/ml) and postnatal cortical membranes prepared at postnatal day (P3) with an optical density of 0.1 (see below). Additional experiments were done with substrates consisting of laminin alone, or laminin/ poly-l-lysine without postnatal cortical membranes or postnatal cortical membrane enriched with Sema3A-containing membranes.

Membrane Preparations

Postnatal cortical tissues (P3) or E18 embryonic cortex and thalamus were dissected and treated as described previously (Bagnard et al., 1998). Blocks of tissues were transferred in a homogenization buffer [10 mM Tris–HCl, 1.5 mM CaCl2, 1 mM spermidine, 25 μg/ml aprotinine, 25 μg/ml leupeptine and 15 μg/ml 2,3-dihydro-2-desoxy-N-acetyl- neuraminic acid (Sigma)], pH 7.4. Homogenates were centrifuged in sucrose step gradients (10 min, 50 000 g). The membrane fraction was washed twice in PBS without Ca2+ and Mg2+ at 14 000 r.p.m. in an Eppendorf biofuge. The same procedure was used to prepare Sema3A- containing membranes after collection of Sema3A-expressing 293 cells from confluent dishes washed in 1 ml PBS. Untransfected HEK293 cells were used to prepare control membranes. Concentrations of membrane solutions were determined by their optical density (OD) at 220 nm with a spectrophotometer after 15-fold dilution in 2% SDS and then adjusted to 0.1 OD.

Time-lapse Procedure

After 24 h in the incubator, there was sufficient outgrowth to start time-lapse recordings. Cultures were transferred into a cell culture chamber on an inverted microscope (Axiovert 135, Zeiss) equipped with phase-contrast optics. In the culture chamber, temperature (37°C) and CO2 concentration (5%) were kept constant. A CCD camera (Imac) was attached to the microscope and connected to a time-lapse video recorder (Panasonic). Single images were recorded every 3 or 5 min under the control of an image processing system (Argus 10, Hammamatsu). An electronic shutter closed the light path each time after an image has been taken. Co-cultures were arranged in such a way that one thalamic slice was surrounded by 4–6 cortical explants; it was therefore always possible to recognize the identity of fibers by following axons back to the explant. Growth cone contacts were analyzed off-line with an image editing system (AG-A570, Panasonic).

Collapse Assays

Explants were grown on glass coverslips covered with laminin (1 μg/ml) and poly-l-lysine (1 μg/ml). In one set of experiments, embryonic cortical or thalamic membrane solutions were added to the explants cultures. After 2 h, membrane particles adhered to the substrate and it was possible to analyze the morphological features of growth cones reacting to the membranes. As a control, experiments were also performed after heat inactivation of the membrane solutions for 30 min at 65°C. In another set of experiments, aliquots of medium containing recombinant Sema3A were added to the cultures of thalamic explants. Growth cone morphology was examined 4 h after incubation in the culture chamber under live conditions. Neurites without lamellipodia and filopodia were scored as ‘collapsed’ (Bagnard et al., 1998).

Generation of Sema3A-expressing Cell Lines

Human embryonic kidney 293 cells (ATTCC CRL 1573) were transfected with expression vectors for Sema3A (pBKSema3A-AP, pBKFlag- Sema3AP1b) (Adams et al., 1997) by calcium phosphate co-precipitation (Sambrook et al., 1989). Stable cell lines were selected in 1 μg/ml G418 (Gibco-BRL). Expression level of recombinant fusion protein were determined with the PhosphaLight chemiluminescence assay (Tropix) or by Western blot analysis using the anti-Flag M2 antibody (Kodak) as described previously (Püschel et al., 1995). Western blots of membranes obtained from Sema3A-expressing cells confirmed the presence of the recombinant protein in the membranes preparations.

Co-culture Assay

We prepared aggregates of HEK 293 cell stably expressing Sema3A according to the method described previously (Bagnard et al., 1998). Thalamic explants were co-cultured with cell aggregates in 20 μl chicken plasma coagulated with 20 μl thrombin. During coagulation of the clot, explants were disposed around cell aggregates at distances <900 μm. After 48 h, cultures were fixed in 4% paraformaldehyde and axonal growth was analyzed using a grid ocular and a ×20 phase-contrast objective. Mean axonal length and mean number of axons were determined on the side facing the cell aggregates and on the side distal to the aggregates.

Results

In this study, time-lapse video microscopy was used to examine the behavior of cortical and thalamic growth cones when they encounter each other. In a first set of experiments, explants from cortex and thalamus were prepared from E16 rat embryos, the developmental stage at which afferent and efferent cortical projections meet within the internal capsule (De Carlos and O'Leary,1992). The substrate consisted of a mixture of laminin and poly-l-lysine enriched with membranes prepared from post- natal cortex, because previous work indicated that cortical membranes promote fiber outgrowth from cortical and thalamic explants (Götz et al., 1992; Hübener et al., 1995). After 1 day in vitro, numerous axons emerged from the explants, and we then started with time-lapse imaging of two or more growth cones, which were close to each other. The recordings lasted for 2–8 h, and all together >1000 contacts between a growth cone with another growth cone or with an axon have been analyzed. As illustrated in Figure 1, three different types of reactions occurred during such contacts: either axons grew along each other (fasciculation, Fig. 1A); growth cones collapsed and retracted from axons for several minutes before they started to grow again (retraction, Fig. 1B); or fibers continued their growth without changing speed or direction (crossing, Fig. 1C). For the quantitative analysis of growth cone interactions the angle at which growing fibers encountered each other were taken into account. We distinguished between three cases: fibers growing in about the same direction (‘parallel’, contact angle 0–60°); fibers growing in opposite directions (‘anti-parallel’, contact angle 120–180°); and fibers approaching each other at about right angles (‘perpendicular’, contact angle 60–120°).

Crossing was the predominant behavior when growth cones encountered each other almost perpendicularly (angle 60–120°). In this situation, 34/50 cortical growth cones (68%) and 73/87 thalamic growth cones (84%) crossed axons of the same type (homotypic contacts) without alterations of their growth behavior (Fig. 2A). After contacts between cortical and thalamic fibers (heterotypic contacts), in half of the cases axons crossed (27/54 contacts) and in the other half they collapsed. Thus, after ‘perpendicular’ approaches, axonal fasciculation did not occur at all for heterotypic contacts, and for homotypic contacts fasciculation was observed in only two cases with cortical axons (Fig. 2A).

The rate of fasciculation increased strikingly when fibers extending from the same explant encountered each other at a shallow angle (0–60°). After such ‘parallel’ approaches of homotypic fibers, 60/82 cortical growth cones (73%) and 24/100 thalamic growth cones (24%) continued to grow on the surfaces of cortical and thalamic axons respectively (Fig. 2B). Similar growth cone reactions after homotypic contacts were observed when cortical and thalamic axons from different explants were growing towards each other (‘anti-parallel’ approach, contact angle 120–180°; Fig. 2C). There was no statistical difference in the rate of fasciculation between ‘parallel’ and ‘anti-parallel’ approaches (P > 0.05, χ2 test). This suggests that there are no polarity cues for fasciculation on cortical and thalamic axons. Once fasciculation is established, axons are able to extend along their homotypic counterpart in both directions.

In contrast to homotypic contacts, axonal fasciculation was never observed after heterotypic contacts. Obviously, in cortex– thalamus co-cultures heterotypic contacts can only be observed when axons extending from different explants are growing towards each other. When the contact angle was between 120 and 180°, 39/57 growth cones (68%) retracted from the heterotypic axon. In the remaining cases, growth cones crossed, but never fasciculated after axonal contacts. Thus, compared with ‘perpendicular’ approaches, the rate of retractions after heterotypic contacts increased significantly, from 50 to 68%, for ‘anti-parallel’ approaches (P < 0.05, χ2 test). The rate of retraction was higher for cortical growth cones contacting thalamic axons (86% retractions, 14% crossings, n = 37) than for thalamic growth cones contacting cortical fibers (46% retractions, 54% crossings, n = 33; P < 0.05, χ2 test).

Effects of Substrate Properties

There is compelling evidence that the rate of axonal fascicu- lation depends on the properties of the substrate: the less a substrate is permissive for axonal growth, the higher is the tendency of axons to grow on the surfaces of other axons, i.e. to form fascicles (Tessier-Lavigne and Goodman, 1996). The average growth rate of cortical and thalamic axons on postnatal cortical membranes was ~40 μm/h, whereas after fasciculation with homotypic axons the growth rate increased to ~60 μm/h (Table 1). Thus, axons might form fascicles because the surfaces of homotypic axons provide a better substrate for fiber growth than cortical membrane preparations. Likewise, the reason why fasciculation between cortical and thalamic axons did not occur in these in vitro experiments might be that cortical membrane substrates are more permissive for axonal growth than the surfaces of heterotypic axons. To address this issue, we examined interactions between cortical and thalamic axons on different substrates without cortical membranes. On laminin alone, the growth rate was only ~20 μm/h, and on a 1:1 mixture of laminin with poly-l-lysine it was further reduced to ~10 μm/h (Table 1). However, as depicted in Table 2, in 60 contacts examined (contact angle 120–180°), there was not a single case of heterotypic fasciculation.

Effects of Developmental Stages

Surface molecules that mediate or modulate fasciculation of axons are often only expressed during restricted developmental stages. The timing of the expression of such molecules could be disturbed after axons regenerate from explants in vitro. We therefore also prepared co-cultures from E18 embryos and imaged axonal interactions. As illustrated in Table 3, at E18 only 12/36 cortical axons (33%) fasciculated with other cortical axons, significantly less than at E16 (P < 0.001, χ2 test). In contrast, homotypic fasciculation of thalamic axons increased almost threefold, from 24% at E16 to 73% (16/22) at E18 (P < 0.001, χ2 test). However, as was observed for co-cultures prepared at E16, heterotypic fasciculation did also not occur for axons extending from E18 explants (79% retraction, 21% crossing, n = 14). Finally, we prepared heterochronic co- cultures, using E16 cortical and E18 thalamic explants, the time when both fiber populations exhibited a high tendency to form homotypic fascicles in vitro. Again, in the large majority of cases (13/17, 77%), cortical and thalamic axons retracted from each other. In the remaining cases fibers crossed, but they never fasciculated with each other (Table 3).

Lack of Regional Specificity during Axonal Interactions

The cortex is composed of distinct areas which are inter- connected by specific thalamic nuclei. For example, the somatosensory cortex is connected with the ventrobasal nucleus of the thalamus, whereas the visual cortex receives input from the lateral geniculate nucleus (LGN) of the thalamus. To test whether there exists a regional specificity for axonal inter- actions, we co-cultured slices from the occipital part of the neocortex, the presumptive visual cortex, with explants from the LGN. As described in Materials and Methods, the position of the LGN was initially visualized in fixed E18 brains by tracing the termination zone of the optic nerve in the diencephalon with a fluorescent dye. For the culture experiments, we dissected this diencephalic region in living brains and prepared slices. As illustrated in Table 3, after contacts between fibers from visual cortex and LGN, 87% of the growth cones retracted and the remaining 13% crossed (n = 23 contacts analyzed). There was no statistical difference between the results obtained with co-cultures from visual cortex and LGN and those obtained with whole neocortex and thalamus (P > 0.05, χ2 test). Thus, axon fascicles are also not formed between fibers from thalamic and cortical regions which are connected with each other in vivo.

Cortical and Thalamic Neurons Express Membrane-bound Collapse Factors

The results presented so far suggest that repellent signals on the surfaces of cortical and thalamic axons induce cell-type specific retractions of growth cones. To further address this issue, we used membranes from E18 cortex and E18 thalamus in collapse assays with cortical and thalamic explants on laminin/ poly-l-Lysine substrates. As depicted in Table 4, the addition of embryonic cortical membranes induced collapses in 43% of thalamic growth cones, but in only 21% of cortical growth cones. Addition of thalamic membranes elicited the opposite effects, with 52% collapsed cortical growth cones and only 24% collapsed thalamic growth cones. The cell-type-specific growth cone collapse was not observed when cortical and thalamic membranes were heat-inactivated (P > 0.05, χ2 test; Table 4). Thus, compared with inactivated membranes, native embryonic cortical membranes possess a specific collapse activity for thal- amic growth cones, and conversely, native embryonic thalamic membranes induce growth cone collapse specifically in cortical fibers.

Modulation of Axonal Interactions by Sema3A

At the time when thalamic and cortical axons traverse the internal capsule, cells in the ganglionic eminence express Sema3A, a member of the semaphorin gene family (Skaliora et al., 1998). Sema3A induces growth cone collapse and acts as a repellent guidance cue for different populations of axons, including cortical fibers (Bagnard et al., 1998, 2000). In this study, we examined the effects of recombinant Sema3A on thalamic axons. As depicted in Table 5, Sema3A also elicits collapses of thalamic growth cones. Moreover, in co-cultures of thalamic explants with Sema3A-producing cell aggregates, thal- amic fibers growing towards the source of Sema3A were shorter and less numerous than fibers growing in the opposite direction (Figure 3). Such effects were not observed in control experi- ments with aggregates from untransfected cells. Thus, Sema3A acts as a repellent signal for both cortical and thalamic axons.

Previous work suggested that semaphorins expressed in the environment of growing axons can induce fiber fasciculation. When Sema1a in the grasshopper limb bud was blocked by antibodies, axons that were normally tightly fasciculated were instead defasciculated (Kolodkin et al., 1992). Because Sema3A is a repellent signal for both cortical and thalamic axons, its presence along the thalamocortical pathway might create a ‘hostile’ environment that pushes these axons together. To test this possibility, we examined the growth of cortical and thalamic axons on a substrate composed of a 1:1 mixture of postnatal cortical membranes and membranes from Sema3A-expressing HEK C293 cells. In the presence of Sema3A the growth speed of axons was almost three times lower than on control substrates (P < 0.001, Student's t-test, Table 1). Sema3A also significantly increased the rate of axon fasciculation after homotypic contacts. As illustrated in Table 2, 14/18 (78%) cortical growth cones and 8/14 (57%) thalamic growth cones extended on the surfaces of homotypic axons (P < 0.05, χ2 test). In contrast, when thalamic and cortical growth cones encountered heterotypic axons on a Sema3A-containing substrate, in all cases examined (n = 16) they collapsed and retracted from the fibers.

Discussion

In this study, time-lapse imaging techniques served to examine the behavior of individual cortical and thalamic growth cones as they encountered each other in vitro. Results indicate that growth cones are able to distinguish between specific cues expressed on axonal surfaces. After heterotypic contacts, cortical and thalamic growth cones collapsed and retracted from each other. Collapse assays confirmed that cortical and thalamic membranes possess specific collapse activities for thalamic and cortical growth cones respectively. Axon fasciculation was observed when cortical growth cones encountered cortical fibers or when thalamic growth cones contacted thalamic axons. Strikingly, addition of the repulsive guidance molecule Sema3A to the substrate increased the rate of homotypic fascicula- tions and at the same time intensified the collapse sensitivity of cortical and thalamic growth cones during heterotypic contacts.

For cortical axons, the tendency to form fascicles was highest when explants were prepared at E16 and the fasciculation rate decreased significantly with E18 explants. In the rat, cortical explants prepared at E16 contain subplate cells and layer 6 cells (Miller, 1988). Neurons in layer 6 establish the adult connection from the cortex back to the thalamus, and during development this projection is formed after subplate neurons have already sent their axons into the internal capsule (De Carlos and O'Leary, 1992). The high rate of fasciculation between cortical axons emerging from E16 explants observed in the present experi- ments is therefore consistent with the proposal that axons of subplate neurons form a scaffold for descending cortical fibers and ‘pioneer’ the corticofugal pathway (McConnell et al., 1989). The decrease of homotypic fasciculation observed with E18 explants may be due to the presence of layer 5 neurons that project to the contralateral hemisphere via the corpus callosum (Yorke et al., 1975; Wise and Jones, 1976; Lent et al., 1990).

In contrast to cortical fibers, the rate of fasciculation between thalamic fibers was low at E16 but significantly increased at E18. Adhesive properties of developing axons can be tightly regulated in time, and the expression of axonal surface proteins can also be influenced by molecular cues in the environment of growing fibers. For example, subsets of spinal neurons express the cell surface protein TAG-1 only during initial axon extension. After crossing the ventral midline, TAG-1 disappears from these axons and the expression of L1, another surface glycoprotein, is initiated (Rathjen and Schachner, 1984; Dodd et al., 1988). During development of thalamocortical projections, at E18, most thalamic axons have grown past the internal capsule and have reached the upper intermediate zone and the subplate region of the cortex (De Carlos and O'Leary, 1992). During this period, influences from this new environment might trigger the expres- sion of axonal surface molecules that mediate fasciculation between thalamic fibers. Alternatively, the modulation of the fasciculation rate may reflect differences in the survival of subpopulations of thalamic neurons under in vitro conditions. We also observed age-dependent variations in the fasciculation rate for cortical axons. At E14, when explants are composed almost exclusively of subplate neurons, cortical axons rarely fasciculate with each other but often cross the encountered fiber. The fasciculation rate increased at E15 and reached its maximum at E16 when explants contain layer 6 neurons (D. Bagnard and J. Bolz, unpublished observations). The lack of fasciculation observed at an early stage may correspond to the behavior of pioneer fibers that explore the environment to detect the correct pathway. For these axons, fasciculation could represent a disadvantage in case of misrouting of individual fibers. At later stages, however, fasciculation appears as an efficient mechanism to guide the next generation of fibers to the final destination.

When growth cones approached homotypic axons perpen- dicularly, they usually crossed the fibers without delay or any sign of retractions. In most cases, fasciculations were only observed when growth cones encountered homotypic fibers under a shallow angle, i.e. when the surface contact between the growth cone and the fiber was relatively large. Cortical and thalamic growth cones are able to extend along homotypic axons in both directions, either retrogradely towards the distal end of the axons or anterogradely towards the cell bodies, suggesting that there are no polarity cues for fasciculation on cortical and thalamic axons. The axonal surface molecules that mediate homotypic fasciculation of cortical and thalamic axons are not known. From the results of the present experiments it is clear, however, that cortical and thalamic axons must express different surface molecules that promote the selective fascicu- lation of these two fiber populations.

Previous studies already indicated that axons, in addition to surface molecules that make them attractive for particular growth cones, can also express cues that have negative effects upon growth cone elongation (Kapfhammer and Raper, 1987a, b; Moorman and Hume, 1990). The present results demonstrate that after contact, cortical and thalamic growth cones collapse and retract from each other. This collapse is reversible, and after a short pause a new growth cone is formed and the axon advances again. We often observed that this retraction–advance cycle was repeated several times until axons finally changed their growth direction and thereby avoided new contacts. Retractions between cortical and thalamic fibers occurred at all developmental stages examined as well as in heterochronic co-cultures, when both fiber populations exhib- ited a high tendency to form homotypic fascicles. Growth cone collapse also took place when axons from defined regions of the thalamus were confronted with axons from their target area in the neocortex. In addition, cell-type-specific repellent activities were observed in the collapse assays with cortical and thalamic membrane preparations. Taken together, these results indicate that neocortical and thalamic fibers express different collapse factors which prevents the formation of heterotypic axon fascicles.

In a recent study, different carbocyanine dyes were used to label simultaneously thalamocortical and corticofugal axons in the developing rat brain (Molnár et al., 1998). The authors report that in some instances early thalamic fibers run in close association with axons from subplate cells, and they proposed that thalamic axons navigate along the axons of subplate cells towards their cortical target area—the ‘handshake hypothesis’ (Molnár and Blakemore, 1991). With fiber tracings in fixed brains, it is difficult to prove that axons fasciculate and actually guide other fibers to their targets. In the present in vitro experi- ments, we sometimes found that the trajectories of thalamic and neocortical axons were adjacent to each other. The time-lapse movies revealed, however, that also in these cases growth cones first retracted from the fibers after contact, but they then re-established their growth and advanced in close vicinity of the heterotypic fiber. The handshake hypothesis requires a very precise regulation of axon outgrowth in both time and space. We tested the existence of a regional specificity by co-culturing explants from a defined region of the thalamus and explants from their target area in the neocortex. In the absence of cell-type-specific markers it is difficult to prove that thalamic explants consist exclusively of LGN neurons. Neurons from adjacent nuclei, which normally project to other cortical areas, might also be present in these cultures. However, if mutual fasciculation between fibers from specific thalamic nuclei and cortical target regions is a robust effect which is essential for axonal targeting, one would expect to observe at least in a few cases the formation of axon fascicles. Instead, in the present experiments all axons examined from presumptive LGN and visual cortex never fasciculated, they either retracted or at best seemed to ignore each other when they met in culture situations. In contrast, using the same culture techniques as in this study, we previously found that limbic thalamic axons do form fascicles with limbic cortical axons. Antibody perturbation experiments indicated that fasciculation was mediated by the limbic system- associated membrane protein (LAMP), which is expressed on both limbic thalamic and limbic cortical axons (Mann et al., 1998). Thus axon fasciculation might contribute to the estab- lishment of axonal connections between limbic thalamic nuclei and the limbic cortex, whereas the formation of reciprocal pro- jections between non-limbic thalamus and neocortical areas appears to be controlled by other axonal guidance mechanisms.

Recent tract tracing studies in mice with mutations of the transcription factors Gbx-2 or Tbr-1 indicated that these animals lack connections between thalamus and neocortex, and that both thalamocortical and corticothalamic fibers stop their growth in the internal capsule (Miyashita-Lin et al., 1999; Rubenstein et al., 1999). One interpretation of these findings is that direct contacts between thalamic and cortical axons are required for the establishment of connections between thalamus and neocortex. Alternatively, mutations in these transcription factors might lead to changes in the expression patterns of downstream genes that cause abnormalities in different brain regions, including the internal capsule, that prevent thalamic and cortical fibers to traverse this structure. Moreover, in COUP-TF1 mutants thalamic projections to the internal capsule are normal, but very few thalamic fibers are able to project out of the internal capsule to reach the cortex (Zhou et al., 1999). Subplate cells in COUP-TF1 mutants, however, project normally towards the thalamus at the time when thalamic axons fail to exit the internal capsule. It has therefore been concluded that the subplate scaffold alone is not sufficient to guide thalamic axons to their final cortical target (Zhou et al., 1999).

In the developing neocortical system, the pathway of thalamo- cortical fibers is centered in the subplate zone and upper region of the intermediate zone, whereas corticofugal axons grow deep in the intermediate zone. The present results suggest that homotypic axon fasciculation in combination with heterotypic fiber repulsion leads to segregated tracts for thalamic and cortical axons. However, our experiments do not explain the preference of fibers for a specific compartment within the white matter. Previous studies provided evidence for molecular hetero- geneity in the future white matter that, for example, selectively promotes the growth of thalamic axons in the subplate and upper intermediate zone where these fibers grow toward neocortical target areas (Bickenese et al., 1994; Tuttle et al., 1995; Henke-Fahle et al., 1996). On the other hand, Sema3C, which has been shown to attract cortical axons (Bagnard et al., 1998), is expressed deep in the intermediate zone and in the subventricular zone, the pathway of corticofugal axons. Thus, the differential distribution of attractive and repulsive axonal guidance molecules in combination with fiber–fiber interactions might explain the separation of afferent and efferent fiber tracts in different zones in the white matter and how these locations are selected.

Because the present experiments were done under in vitro conditions, an obvious question is whether thalamic and cortical axons would exhibit similar reactions when they meet each other in vivo. As outlined above, we used explants from defined cortical and thalamic regions, prepared cultures at different developmental stages and also examined heterochronic co-cultures. All these different in vitro situations led to the same results, homotypic fasciculation and heterotypic axon repulsion. One major difference between in vitro assays for axonal targeting and the in vivo situation is that axons are exposed to different environments and grow on different substrates. This seems to be particularly important in the present context, because it has been demonstrated recently that extracellular matrix components which promote axon outgrowth can also modify the response of growth cones to axonal guidance cues (Hopker et al., 1999). This raises the possibility that the nature of the substrate might also influence interactions between growing axons. This is why we tested several combinations of substrates which are more or less growth permissive for axonal growth. Results indicated that whatever substrate was used, and despite a >6-fold variation of the growth speed on different substrates, the rate of homotypic fasciculation and heterotypic contact repulsion did not change significantly. However, the presence of the repulsive guidance molecule Sema3A, which is expressed in vivo along the thalamocortical pathway, not only increased the rate of homotypic fasciculations but also amplified the segregation between cortical and thalamic axons. As previously suggested, one possible mechanism underlying Sema3A-induced axon fasciculation is that this molecule makes the substrate less permissive for fiber growth and therefore forces fibers to extend on axonal surfaces (Kolodkin et al., 1992, 1993; Matthes et al., 1995). The present findings reveal, however, that Sema3A can also influence axon repulsion. The higher rate of retractions between cortical and thalamic fibers in the presence of Sema3A might be explained by an increase of the sensitivity of growth cones to different collapse factors expressed on the surfaces of cortical and thalamic axons. It is tempting to speculate that this mechanism is related to the modulation of intracellular levels of cyclic nucleotides (cGMP and cAMP), which have been implicated as key regulators of axonal attraction and repulsion (Song et al., 1999). If collapsing factors on axonal surfaces share the final transduction pathway with Sema3A, a reduction of intracellular cGMP levels might push the internal state of growth cones towards the collapse threshold and thereby might favor the induction of collapses during axonal contacts. Our results therefore provide evidence that a guidance molecule not only defines axonal pathways, but also actively influences the reactivity of growth cones to their environment.

Notes

We thank Marion Lohrum for establishment of semaphorin secreting cell lines. This work was supported by the Human Frontiers Science Program (J.B.) and the DFG (A.W.P.).

Address correspondence to Dr Dominique Bagnard, LNDR—Centre de Neurochimie du CNRS, 5 Rue Blaise Pascal, F-67084 Strasbourg cedex, France. Email:bagnard@neurochem.u-strasbg.fr.

Table 1

Growth cone velocity on different substrates

Substrate Growth rate (μm/h) 
 Cortical axons Thalamic axons 
Cortical and thalamic explants were grown on cortical membranes, laminin, laminin/poly-l-lysin or Sema3A substrates for 24 h. ‘Axonal surface’ corresponds to growth cones fasciculating on axons from the same population. All growth speeds were expressed in μm/h and were compared to those measured on cortical membrane substrates (* P < 0.01, Student's t-test). 
Cortical membranes 37.2 ± 7.4 (n = 46) 40.9 ± 8.4 (n = 26) 
Laminin 18.4 ± 2.4 (n = 8)* 20.7 ± 1.8 (n = 9)* 
Laminin/poly-l-lysine  8.7 ± 4.8 (n = 13)*  9.2 ± 5.1 (n = 12)* 
Axonal surface 59.2 ± 5.2 (n = 21)* 63.5 ± 8.7 (n = 9)* 
Sema3A 11.8 ± 11.9 (n = 24)* 14.2 ± 7.4 (n = 19)* 
Substrate Growth rate (μm/h) 
 Cortical axons Thalamic axons 
Cortical and thalamic explants were grown on cortical membranes, laminin, laminin/poly-l-lysin or Sema3A substrates for 24 h. ‘Axonal surface’ corresponds to growth cones fasciculating on axons from the same population. All growth speeds were expressed in μm/h and were compared to those measured on cortical membrane substrates (* P < 0.01, Student's t-test). 
Cortical membranes 37.2 ± 7.4 (n = 46) 40.9 ± 8.4 (n = 26) 
Laminin 18.4 ± 2.4 (n = 8)* 20.7 ± 1.8 (n = 9)* 
Laminin/poly-l-lysine  8.7 ± 4.8 (n = 13)*  9.2 ± 5.1 (n = 12)* 
Axonal surface 59.2 ± 5.2 (n = 21)* 63.5 ± 8.7 (n = 9)* 
Sema3A 11.8 ± 11.9 (n = 24)* 14.2 ± 7.4 (n = 19)* 
Table 2

Influence of the substrate during axonal interactions

 Fasciculation (%) Retraction (%) Crossing (%) 
Rate of fasciculation, crossing and retraction are presented for cortex–thalamus interactions on cortical membranes, laminin or laminin/poly-l-lysin substrates. Results obtained on laminin or laminin/poly-l-lysin were not different from those observed on cortical membrane substrates (ns, not significant, P > 0.05, χ2 test). Addition of Sema3A resulted in a significant increase in the rate of homotypic fasciculations and heterotypic retractions (*P < 0.05, χ2 test). 
Ctx–thal    
 Cortical membranes (n = 57)  0.0 68.4  31.6 
 Laminin (n = 43)  0.0 65.0ns  35.0ns 
 Laminin/poly-l-lysin (n = 17)  0.0 70.6ns  29.4ns 
Sema3A membranes    
 Ctx–ctx (n = 18) 78.0* 11.0  11.0 
 Thal–thal (n = 14) 57.0* 28.6  14.4 
 Ctx–thal (n = 16)  0.0  0.0 100 
 Fasciculation (%) Retraction (%) Crossing (%) 
Rate of fasciculation, crossing and retraction are presented for cortex–thalamus interactions on cortical membranes, laminin or laminin/poly-l-lysin substrates. Results obtained on laminin or laminin/poly-l-lysin were not different from those observed on cortical membrane substrates (ns, not significant, P > 0.05, χ2 test). Addition of Sema3A resulted in a significant increase in the rate of homotypic fasciculations and heterotypic retractions (*P < 0.05, χ2 test). 
Ctx–thal    
 Cortical membranes (n = 57)  0.0 68.4  31.6 
 Laminin (n = 43)  0.0 65.0ns  35.0ns 
 Laminin/poly-l-lysin (n = 17)  0.0 70.6ns  29.4ns 
Sema3A membranes    
 Ctx–ctx (n = 18) 78.0* 11.0  11.0 
 Thal–thal (n = 14) 57.0* 28.6  14.4 
 Ctx–thal (n = 16)  0.0  0.0 100 
Table 3

Specificity of axonal interactions

 Fasciculation (%) Retraction (%) Crossing (%) 
Fibers from explants prepared from whole cortex (Ctx) and whole thalamus (Thal) at embryonic stage E16 or E18. Ctx vis and thal LGN: fibers from visual cortex and LGN explants prepared at E18. (ns, not significant, *P < 0.05, χ2 test). 
Ctx E16–ctx E16 (n = 82) 73.2 24.3  2.5 
Ctx E18–ctx E18 (n = 36) 33.3*  0.0 66.4 
Thal E16–thal E16 (n = 100) 24.0 30.0 46.0 
Thal E18–thal E18 (n = 22) 72.7*  0.0 27.3 
Ctx E18–thal E18 (n = 14)  0.0 78.6ns 21.4 
Ctx E16-Thal E18 (n = 17)  0.0 76.5ns 23.5 
Ctx vis–thal LGN (n = 23)  0.0 87.0ns 13.0 
 Fasciculation (%) Retraction (%) Crossing (%) 
Fibers from explants prepared from whole cortex (Ctx) and whole thalamus (Thal) at embryonic stage E16 or E18. Ctx vis and thal LGN: fibers from visual cortex and LGN explants prepared at E18. (ns, not significant, *P < 0.05, χ2 test). 
Ctx E16–ctx E16 (n = 82) 73.2 24.3  2.5 
Ctx E18–ctx E18 (n = 36) 33.3*  0.0 66.4 
Thal E16–thal E16 (n = 100) 24.0 30.0 46.0 
Thal E18–thal E18 (n = 22) 72.7*  0.0 27.3 
Ctx E18–thal E18 (n = 14)  0.0 78.6ns 21.4 
Ctx E16-Thal E18 (n = 17)  0.0 76.5ns 23.5 
Ctx vis–thal LGN (n = 23)  0.0 87.0ns 13.0 
Table 4

Collapse assay using membrane preparations

Collapsed growth cones Cortex (%) Thalamus (%) 
Percentage (and absolute numbers) of cortical and thalamic growth cones cultured on laminin/poly-l-lysin substrate with ‘collapsed’ morphology after addition of E18 native cortical membranes, E18 native thalamic membranes, heat-inactivated E18 cortical membranes or heat-inactivated E18 thalamic membranes. *P < 0.05, χ2 test; ns, not significant. 
E18 native cortical membranes 21.2 (n = 250) 43.1 (n = 186)* 
E18 native thalamic membranes 51.5 (n = 163)* 24.4 (n = 74) 
E18 heat-inactivated cortical membranes 19.5 (n = 77) 18.4 (n = 71)ns 
E18 heat-inactivated thalamic membranes 22.3 (n = 54)ns 21.7 (n = 46) 
Collapsed growth cones Cortex (%) Thalamus (%) 
Percentage (and absolute numbers) of cortical and thalamic growth cones cultured on laminin/poly-l-lysin substrate with ‘collapsed’ morphology after addition of E18 native cortical membranes, E18 native thalamic membranes, heat-inactivated E18 cortical membranes or heat-inactivated E18 thalamic membranes. *P < 0.05, χ2 test; ns, not significant. 
E18 native cortical membranes 21.2 (n = 250) 43.1 (n = 186)* 
E18 native thalamic membranes 51.5 (n = 163)* 24.4 (n = 74) 
E18 heat-inactivated cortical membranes 19.5 (n = 77) 18.4 (n = 71)ns 
E18 heat-inactivated thalamic membranes 22.3 (n = 54)ns 21.7 (n = 46) 
Table 5

Collapse assays using conditioned medium

Cultures conditions Normal (%) Collapsed (%) 
Percentage (and absolute numbers) of thalamic growth cones cultured on laminin/poly-l-lysin substrate with ‘normal’ and ‘collapsed’ morphology after addition of conditioned medium from Sema3A-expressing 293 cells or untransfected control cells. **P < 0.001, χ2 test. 
Sema3A 48.7 (n = 74) 51.3 (n = 78)** 
Control 90.6 (n = 135) 9.4 (n = 14) 
Cultures conditions Normal (%) Collapsed (%) 
Percentage (and absolute numbers) of thalamic growth cones cultured on laminin/poly-l-lysin substrate with ‘normal’ and ‘collapsed’ morphology after addition of conditioned medium from Sema3A-expressing 293 cells or untransfected control cells. **P < 0.001, χ2 test. 
Sema3A 48.7 (n = 74) 51.3 (n = 78)** 
Control 90.6 (n = 135) 9.4 (n = 14) 
Figure 1.

Sequence of micrographs showing axonal interactions. Explants of embryonic cortex and thalamus were grown on different substrates and interactions between identified growth cones were recorded by time-lapse videomicroscopy. Images were taken every 3 min for 6–24 h periods. The sequences presented here were recorded over 12 min and illustrate the capacity of axons to fasciculate with each other (A), the induction of retraction and growth cone collapses (B) or the crossing of fibers (C) after axonal contact. observations were done using a ×20 phase-contrast objective (A,C) or a ×40 phase-contrast objective (B).

Figure 1.

Sequence of micrographs showing axonal interactions. Explants of embryonic cortex and thalamus were grown on different substrates and interactions between identified growth cones were recorded by time-lapse videomicroscopy. Images were taken every 3 min for 6–24 h periods. The sequences presented here were recorded over 12 min and illustrate the capacity of axons to fasciculate with each other (A), the induction of retraction and growth cone collapses (B) or the crossing of fibers (C) after axonal contact. observations were done using a ×20 phase-contrast objective (A,C) or a ×40 phase-contrast objective (B).

Figure 2.

Quantification of axonal interactions. In the co-culture assays, axons approached each other under various angles. We defined three classes called (A) ‘perpendicular’ (contact angle 60–120°), (B) ‘parallel’ (contact angle 0–60°) and (C) ‘anti-parallel’ (contact angle 120–180°). For each class, we determined the percentage of fasciculation (gray), crossing (white) and retraction (black) after interactions between two cortical axons (ctx–ctx), two thalamic axons (thal–thal) or one cortical and one thalamic axon (ctx–thal). χ2 test analysis revealed no statistical differences between axonal interactions obtained in (B) and (C) (P > 0.05).

Figure 2.

Quantification of axonal interactions. In the co-culture assays, axons approached each other under various angles. We defined three classes called (A) ‘perpendicular’ (contact angle 60–120°), (B) ‘parallel’ (contact angle 0–60°) and (C) ‘anti-parallel’ (contact angle 120–180°). For each class, we determined the percentage of fasciculation (gray), crossing (white) and retraction (black) after interactions between two cortical axons (ctx–ctx), two thalamic axons (thal–thal) or one cortical and one thalamic axon (ctx–thal). χ2 test analysis revealed no statistical differences between axonal interactions obtained in (B) and (C) (P > 0.05).

Figure 3.

Sema3A act as a repulsive signal for thalamic axons. HEKC293 cell aggregates producing Sema3A were cultured next to thalamic explants in a three-dimensional matrix. In order to quantify the effects of Sema3A, we determined the mean axonal length (A) and mean number of axons (B) on the side towards the aggregates (dark bars) and away from the aggregates (open bars) for 136 thalamic explants (n = number of explants; **P < 0.0001, ns, not significant; Student's t-test).

Figure 3.

Sema3A act as a repulsive signal for thalamic axons. HEKC293 cell aggregates producing Sema3A were cultured next to thalamic explants in a three-dimensional matrix. In order to quantify the effects of Sema3A, we determined the mean axonal length (A) and mean number of axons (B) on the side towards the aggregates (dark bars) and away from the aggregates (open bars) for 136 thalamic explants (n = number of explants; **P < 0.0001, ns, not significant; Student's t-test).

2
Present address: LNDR—Centre de Neurochimie du CNRS, 5 Rue Blaise Pascal, F-67084 Strasbourg, France

References

Adams RH, Lohrum M, Klostermann A, Betz H, Püschel AW (
1997
) The chemorepulsive activity of secreted semaphorins is regulated by furin-dependent proteolytic processing.
EMBO J
 
16
:
6077
–6086.
Bagnard D, Lohrum M, Uziel D, Püschel AW, Bolz J (
1998
) Semaphorins act as attractive and repulsive guidance signals during the develop- ment of cortical projections.
Development
 
125
:
5043
–5053.
Bagnard D, Thomasset N, Lohrum M, Püschel AW, Bolz J (
2000
) Spatial distributions of guidance molecules regulate chemorepulsion and chemoattraction of growth cones.
J Neurosci
 
20
:
1030
–1035.
Bastiani MJ, du Lac S, Goodman CS (
1986
) Guidance of neuronal growth cones in the grasshopper embryo. I. Recognition of a specific axonal pathway by the pCC neuron.
J Neurosci
 
6
:
3518
–3531.
Bicknese AR, Sheppard AM, O'Leary DDM, Pearlman AL (
1994
) Thalamocortical axons extend along a chondroitin sulfate proteoglycan-enriched pathway coincident with the neocortical subplate and distinct from the efferent path.
J Neurosci
 
14
:
3500
–3510.
Catalano SM, Robertson RT, Killackey HP (
1991
) Early ingrowth of thalamocortical afferents to the neocortex of the prenatal rat.
Proc Natl Acad Sci USA
 
88
:
2999
–3003.
Catalano SM, Robertson RT, Killackey HP (
1996
) Individual axon morphology and thalamocortical topography in developing rat somatosensory cortex.
J Comp Neurol
 
367
:
36
–53.
Cheng HJ, Nakamoto M, Bergemann AD, Flanagan JG (
1995
) Comple- mentary gradients in expression and binding of ELF-1 and MEK4 in development of the topographic retinotectal projection map.
Cell
 
82
:
371
–381.
ColaMarino SA, Tessier-Lavigne M (
1995
) The role of the floor plate in axon guidance.
Annu Rev Neurosci
 
18
:
497
–529.
De Carlos JA, O'Leary DDM (
1992
) Growth and targeting of subplate axons and establishment of major cortical pathways.
J Neurosci
 
12
:
1194
–1211.
Dodd J, Morton SB, Karagogeos D, Yamamoto M, Jessell TM (
1988
) Spatial regulation of axonal glycoprotein expression on subsets of embryonic spinal neurons.
Neuron
 
1
:
105
–116.
Drescher U, Kremoser C, Handwerker C, Löschinger J, Noda M, Bonhoeffer F (
1995
) In vitro guidance of retinal ganglion cell axons by RAGS, a 25 kDa tectal protein related to ligands for Eph receptor tyrosine kinases.
Cell
 
82
:
359
–370.
Ghosh A, Shatz CJ (
1992
) Pathfinding and target selection by developing geniculocortical axons.
J Neurosci
 
12
:
39
–55.
Götz M, Novak N, Bastmeyer M, Bolz J (
1992
) Membrane bound molecules in rat cerebral cortex regulate thalamic innervation.
Development
 
116
:
507
–519.
Henke-Fahle S, Mann F, Götz M, Wild K, Bolz J (
1996
) A dual action of a carbohydrate epitope on afferent and efferent axons in cortical development.
J Neurosci
 
16
:
4195
–4206.
Hopker VH, Shewan D, Tessier-Lavigne M, Poo M, Holt C (
1999
) Growth-cone attraction to netrin-1 is converted to repulsion by laminin-1.
Nature
 
401
:
69
–73.
Hübener M, Götz M, Klostermann S, Bolz J (
1995
) Guidance of thalamocortical axons by growth-promoting molecules in developing rat cerebral cortex.
Eur J Neurosci
 
7
:
1963
–1972.
Kapfhammer JP, Raper JA (
1987
) Collapse of growth cone structure on contact with specific neurites in culture.
J Neurosci
 
7
:
201
–212.
Kapfhammer JP, Raper JA (
1987
) Interactions between growth cones and neurites growing from different neural tissues in culture.
J Neurosci
 
7
:
1595
–1600.
Kennedy TE, Serafini T, De la Torre JR, Tessier-Lavigne M (
1994
) Netrins are diffusible chemotropic factors for commissural axons in the embryonic spinal cord.
Cell
 
78
:
425
–435.
Kolodkin AL, Matthes DJ, O'Connor TP, Patel NH, Admon A, Bentley D, Goodman CS (
1992
) Fasciclin IV: sequence, expression, and function during growth cone guidance in the grasshopper embryo.
Neuron
 
9
:
831
–845.
Kolodkin AL, Matthes DJ, Goodman CS (
1993
) The semaphorin genes encode a family of transmembrane and secreted growth cone guidance molecules.
Cell
 
75
:
1389
–1399.
Lent R, Hedin-Pereira C, Menezes JR, Jhaveri S (
1990
) Neurogenesis and development of callosal and intracortical connections in the hamster.
Neuroscience
 
38
:
21
–37.
Luskin MB, Shatz CJ (
1985
) Studies of the earliest generated cells of the cat's visual cortex: cogeneration of subplate and marginal zones.
J Neurosci
 
5
:
1062
–1075.
Mackarehtschian K, Lau CK, Caras I, McConnell SK (
1999
) Regional differences in the developing cerebral cortex revealed by Ephrin-A5 expression.
Cereb Cortex
 
9
:
601
–610.
Mann F, Zhukareva V, Pimenta A, Levitt P, Bolz J (
1998
) Membrane- associated molecules guide limbic and nonlimbic thalamocortical projections.
J Neurosci
 
18
:
9409
–9419.
Marin-Padilla M (
1971
) Early prenatal ontogenesis of the cerebral cortex (neocortex) of the cat (Felis domestica). A Golgi study.
Z Anat Entwickl
 
134
:
117
–145.
Mark MD, Lohrum M, Püschel AW (
1997
) Patterning neuronal connections by chemorepulsion: the semaphorins.
Cell Tissue Res
 
290
:
299
–306.
Matthes DJ, Sink H, Kolodkin AL, Goodman CS (
1995
) Semaphorin II can function as a selective inhibitor of specific synaptic arborizations.
Cell
 
81
:
631
–639.
McConnell SK, Ghosh A, Shatz CJ (
1989
) Subplate neurons pioneer the first axon pathway from the cerebral cortex.
Science
 
245
:
978
–982.
Métin C, Godement P (
1996
) The ganglionic eminence may be an inter- mediate target for corticofugal and thalamocortical axons.
J Neurosci
 
16
:
3219
–3235.
Métin C, Deleglise D, Serafini T, Kennedy TE, Tessier-Lavigne M (
1997
) A role for netrin-1 in the guidance of cortical efferents.
Development
 
124
:
5063
–5074.
Miller MW (1988) Development of projection and local circuit neurons in neocortex. In: Cerebral cortex. Vol. 7. Development and maturation of cerebral cortex (Peters A, Jones EG, eds), pp. 133–175. New York: Plenum Press.
Miyashita-Lin EM, Hevner R, Wassarman KM, Martinez S, Rubenstein JL (
1999
) Early neocortical regionalization in the absence of thalamic innervation.
Science
 
285
:
906
–909.
Molnár Z, Blakemore C (
1991
) Lack of regional specificity for connections formed between thalamus and cortex in coculture.
Nature
 
351
:
475
–477.
Molnár Z, Blakemore C (
1995
) How do thalamic axons find their way to the cortex?
Trends Neurosci
 
18
:
389
–397.
Molnár Z, Adams R, Blakemore C (
1998
) Mechanisms underlying the early establishment of thalamocortical connections in the rat.
J Neurosci
 
18
:
5723
–5745.
Moorman SJ, Hume RI (
1990
) Growth cones of chick sympathetic preganglionic neurons in vitro interact with other neurons in a cell-specific manner.
J Neurosci
 
10
:
3158
–3163.
Püschel AW, Adams RH, Betz H (
1995
) Murine semaphorins D/collapsin is a member of a divers gene family and creates domains inhibitory for axonal extension.
Neuron
 
14
:
941
–948.
Rathjen FG, Schachner M (
1984
) Immunocytological and biochemical characterization of a new neuronal cell surface component (L1 antigen) which is involved in cell adhesion.
EMBO J
 
3
:
1
–10.
Richards LJ, Koester SE, Tuttle R, O'Leary DDM (
1997
) Directed growth of early cortical axons is influenced by a chemoattractant released from an intermediate target.
J Neurosci
 
17
:
2445
–2458.
Rubenstein JL, Anderson S, Shi L, Miyashita-Lin E, Bulfone A, Hevner R (
1999
) Genetic control of cortical regionalization and connectivity.
Cereb Cortex
 
9
:
524
–532.
Sambrook, J, Fritsch EF, Maniatis T (1989) Molecular cloning: a laboratory manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
Serafini T, Kennedy TE, Galko MJ, Mirzayan C, Jessell TM, Tessier-Lavigne M (
1994
) The netrins define a family of axon outgrowth-promoting proteins homologous to C. elegans UNC-6.
Cell
 
78
:
409
–424.
Skaliora I, Singer W, Betz H, Puschel AW (
1998
) Differential patterns of semaphorin expression in the developing rat brain.
Eur J Neurosci
 
10
:
1215
–1229.
Song HJ, Poo MM (
1999
) Signal transduction underlying growth cone guidance by diffusible factors.
Curr Opin Neurobiol
 
9
:
355
–363.
Tessier-Lavigne M, Goodman CS (
1996
) The molecular biology of axon guidance.
Science
 
274
:
1123
–1133.
Tuttle R, Schlaggar BL, Braisted JE, O'Leary DDM (
1995
) Maturation- dependent upregulation of growth-promoting molecules in develop- ing cortical plate controls thalamic and cortical neurite growth.
J Neurosci
 
15
:
3039
–3052.
Wise SP, Jones EG (
1976
) The organization and postnatal development of the commissural projection of the rat somatic sensory cortex.
J Comp Neurol
 
168
:
313
–344.
Yorke CH Jr, Caviness VS Jr (
1975
) Interhemispheric neocortical connections of the corpus callosum in the normal mouse: a study based on anterograde and retrograde methods.
J Comp Neurol
 
164
:
233
–245.
Zhou C, Qiu Y, Pereira FA, Crair MC, Tsai SY, Tsai MJ (
1999
) The nuclear orphan receptor COUP-TFI is required for differentiation of subplate neurons and guidance of thalamocortical axons.
Neuron
 
24
:
847
–859.