Abstract
We labeled axonal projections using carbocyanine dyes in the developing rat brain to study cellular interactions that might underlie the establishment of thalamocortical connectivity. By embryonic day 14 (E14), groups of neurons in the ventral diencephalon and the primitive internal capsule have established projections to the dorsal thalamus, and thalamic fibers pass in topographic order among them. Simultaneously, axons from the early-born cells in both subplate and marginal zone (i.e., the original cortical preplate) establish an ordered array that fills the intermediate zone. Thalamic axons and preplate fibers meet in the lateral part of the internal capsule (at E15 for occipital cortex and dorsolateral thalamus). Subsequently, selective labeling of corresponding thalamic and early corticofugal projections reveals thalamic fibers growing in association with early corticofugal axons, right up to the cortical subplate. A small carbocyanine crystal implanted at any point in the cortex shortly after the arrival of thalamic axons (E16 for the occipital cortex) labels a single, tight bundle containing both descending and ascending fibers, rather than two separate tracts, providing further evidence for intimate topographic association of the two axon systems. Crystals placed in a row, parasagittally or coronally along the hemisphere, reveal separate, topographically distributed, discrete fiber bundles throughout the pathway, leading to spatially ordered groups of back-labeled thalamic cells. These results indicate that the topography of thalamic axons is maintained throughout the pathway and that they reach the cortex by associating with the projections of a number of preexisting cells, including the preplate scaffold.
- cortex
- thalamus
- rat
- pioneer axons
- development
- subplate
- perireticular nucleus
- preplate
- “handshake” hypothesis
- internal capsule
- thalamic reticular nucleus
- axon guidance
The adult mammalian neocortex consists of numerous areas, differing in connectivity and functional properties (Peters and Jones, 1985), many of which have a distinctive cytoarchitectural appearance (Brodmann, 1909). Thalamic axons, the carriers of most afferent information to the cortex, arrive before the majority of cortical neurons are born (Lund and Mustari, 1977; Rakic, 1977; Wise and Jones, 1978; Shatz and Luskin, 1986). These early afferent projections, if they are topographically distributed and distinctive for each particular area, could play a part in initiating regional differentiation (O’Leary, 1989). In that case, unraveling the mechanisms by which thalamic axons are directed toward their target regions is important in understanding how cortical organization is determined.
The representation of the thalamus on the cortex and the topographic distribution of thalamic axons have been studied extensively in postnatal and adult rodents (Caviness and Frost, 1980; Frost and Caviness, 1980; Crandall and Caviness, 1984; Bernardo and Woolsey, 1987; Caviness, 1988; Agmon and Connors, 1991; Agmon et al., 1993,1995; Mitrofanis and Guillery, 1993). All areas of the rodent neocortex receive a thalamic input. In general, neighborhood relationships are preserved in the pattern of projection, but, in some cases, contiguous neocortical fields receive input from nonadjacent thalamic nuclei. However, studies of the mature projection do not directly illuminate its initial topography or the means by which connections are established. The spatial relationships of thalamic cell groups might change substantially, and thalamic axons themselves might undergo local rearrangement, after their projection to the cortex.
The capacity to trace axons in fixed embryonic tissue offered by the carbocyanine dye technique has stimulated many recent studies of early axonal outgrowth between cortex and thalamus for several species and various parts of the projection system (McConnell et al., 1989;Blakemore and Molnár, 1990; Molnár and Blakemore, 1990;Catalano et al., 1991, 1996; De Carlos and O’Leary, 1992; Erzurumlu and Jhaveri, 1992; Ghosh and Shatz, 1992; Miller et al., 1993; Clasca et al., 1995; Métin and Godement, 1996). However, no coherent picture has emerged of the relationships between growing thalamic axons and other fiber systems or of the mechanisms by which thalamic axons navigate to their appropriate target regions.
McConnell et al. (1989) drew attention to the fact that axons from early postmitotic neurons of the cortical subplate (Marin-Padilla, 1971; Luskin and Shatz, 1985a,b) pioneer the pathway from the cortex toward subcortical targets, before neurons of layers 5 and 6 have migrated into position. They suggested that these early corticofugal axons might play a part in guiding subsequent descending and ascending projections.
Here we characterize the entire period of development of early corticofugal and thalamocortical projections in the rat and the degree of order that both sets of fibers maintain. We describe the cellular elements that thalamic axons encounter as they grow out from the diencephalon, and specifically we present evidence that they associate closely with axons emanating from early cortical neurons.
MATERIALS AND METHODS
Hooded Lister rats, whose gestation period is 21–22 d, were time-mated at the University Laboratory of Physiology (Oxford, UK). Assessment of the embryonic age of the fetuses was based on the plug date, defined as embryonic day 0 (E0). Fetuses were removed by cesarean section under pentobarbital anesthesia (100 mg/kg, i.p., to the pregnant female) at various stages, from the 13th to the 21st or 22nd postconceptual day. They were immediately chilled on ice and transcardially perfused with 4% paraformaldehyde in 0.1 mphosphate buffer, pH 7.4, delivered through a glass micropipette under a dissecting microscope. Postnatal animals of various ages were anesthetized by chilling [up to postnatal day 4 (P4)], or by overdose of intraperitoneal sodium pentobarbital (if older than P4), before perfusion.
The brains were removed, and, after overnight post-fixation (in 4% buffered paraformaldehyde), single crystals (0.1–0.3 mm diameter) of fluorescent carbocyanine dyes [1,1′-dioctadecyl 3,3,3′3′-tetramethylindocarbocyanine perchlorate (DiI), 4-[4-(dihexadecylamino)styryl]-N-methylpyridinium iodide (DiA), 4-[4-(didecylamino)styryl]-N-methylpyridinium iodide (DiAsp), and 3,3′-dipentyloxacarbocyanine iodide (DiO); Molecular Probes, Eugene, OR] were inserted into different parts of the cortex or diencephalon under an operating microscope (Godement et al., 1987).
For each cortical placement, a single carbocyanine crystal was inserted into a tiny slit made in the cortex with the tip of a pair of fine forceps or, in embryonic brains, with a thin stainless steel wire. The depth to which the crystal was inserted into the cortex varied with age, from as little as 0.1 mm at E13 to 0.5 mm in the oldest animals. In half of these brains, we made deposits of a number of different dyes, at points across the cortex, usually spaced 1–2 mm apart in either a parasagittal or a coronal row in one hemisphere. For most of these experiments, we used alternating placements of the two dyes that are best transported, DiI and DiA, which are clearly distinguishable under different wavelengths of fluorescent illumination.
For crystal placements in the thalamus [usually in the vicinity of the putative lateral geniculate nucleus (LGN)], we first transected the brainstem coronally, rostral to the superior colliculus, to expose the posterior, dorsal thalamus. For insertions into deeper or more rostral thalamic nuclei (e.g., the ventrobasal complex), we bisected the brain sagittally and cut away the medial part of the diencephalon to expose the relevant part of the thalamus. Then a single carbocyanine crystal was inserted into a small slit made in the tissue with the tip of a fine pair of forceps under an operating microscope.
In other specimens, the diencephalon was removed from the hemisphere by means of a parasagittal cut through the internal capsule, and a small crystal of carbocyanine dye was inserted into the exposed telencephalic face of the internal capsule, to stain descending connections reaching that region from other parts of the hemisphere, as well as any ascending axons that had already passed through the internal capsule. Table 1 lists the number of brains (both hemispheres) used for the various studies at each developmental stage (E13 to P8).
After crystal insertion, the brains were stored in fixative or in PBS containing 0.1% Na azide, to prevent contamination, at room temperature (22°C) or at 37°C. They were incubated for periods ranging from 1 to 10 weeks, depending on age and temperature (Table2), to allow diffusion of the dye. We came to prefer using room temperature because of the slightly lower background labeling. At the end of the incubation period the brains were embedded in 5% agar, and coronal or horizontal sections, 75–250 μm thick, were cut on a vibratome (Oxford Instruments). All sections were counterstained with bisbenzimide (10 min in 2.5 μg/ml solution in PBS; Riedel-De Haèn AG, Seelze-Hannover, Germany) or acridine orange (10 μg/ml in PBS; Molecular Probes). The sections were coverslipped under Hydromount (National Diagnostics) or glycerol–phenylenediamine solution (0.1%p-phenylenediamine, 10% PBS, and 90% glycerol, made up and stored according to the method of Johnson et al., 1982).
For the three-dimensional serial reconstructions, we coverslipped the sections under PBS, sealed the coverslips with Paraseal (Raymond A. Lamb, London, UK), and imaged the sections within 24 hr to prevent differential shrinking and distortion of the individual members of the series. From the serial sections, double- or triple-exposure color micrographs were taken (Kodak Ektachrome 400 ASA color slide film), and camera lucida drawings were made under epifluorescent illumination, using different filters to reveal the various dyes.
In early experiments, some sections from each brain (and from other control brains not used for dye tracing) were subsequently stained with cresyl violet to help with identification of the major divisions of the thalamus and to reveal the full thickness and lamination pattern of the cortex. Counterstaining with chromatin stains (bisbenzimide and acridine orange) helped us interpret the sections by revealing the outlines and boundaries of obvious cell groups (e.g., cortical layer I and layers in the hippocampus). Observation of nuclear chromatin staining with high-power objectives and/or obtaining fine optical sections with confocal microscopy enabled us to confirm whether labeled profiles were cell bodies.
Some sections, selected in conventional fluorescence microscopy, were subsequently examined and photographed with a laser-scanning confocal microscope (CLSM-Fluovert; Leica, Heidelberg, Germany). Digitized fine optical sections (up to 64 sections) were stacked to generate high-resolution extended focus images (see Fig. 10F). Stereo pairs (with ±7° disparity) could be constructed from these three-dimensional data sets (see Fig. 10G).
RESULTS
This study is based on a total of 285 axon-tracing experiments, covering the period from E13, before the arrival of the first cells in the cerebral cortex, to P8, when the organization and innervation of the cortex appear virtually fully mature (see Table 1). For the sake of clarity we present the results in five parts:
(1) We describe the timing of outgrowth of thalamocortical fibers, their pattern of organization and initial topography, their arrival at the cortex, and invasion of the cortical plate, based on anterograde labeling from one or more dye crystals implanted in the dorsal thalamus.
(2) The cells of origin of the earliest corticofugal projections were labeled and examined after placement of crystals in the primitive internal capsule before thalamic fibers approach the cortex.
(3) The timing of outgrowth and topography of these early corticofugal projections is then described, on the basis of anterograde tracing from dye crystals placed in the cortex, before the arrival of thalamic axons.
(4) We then describe the relationship between early corticofugal and thalamocortical fibers—at their meeting point in the primitive internal capsule, within the intermediate zone, and as the thalamic axons approach the cortical subplate (from E14.5 to E16 for different parts of the cortex). This is based partly on double labeling (with different dyes placed in corresponding regions of thalamus and cortex) and partly on the labeling of both sets of axons and cortical cell bodies from crystals placed in the internal capsule.
(5) Finally, single- and multiple-crystal placements in the cortex, shortly after the arrival of thalamic axons, provided further information about the relationship between afferent and efferent axons and about the global topography of the pathway established between thalamus and cortex.
Development of thalamocortical projections
Thalamocortical fibers grow out in topographic order among cell groups that project to the thalamus
Implantation of a small crystal of carbocyanine dye into the dorsolateral part of the thalamus (including the presumptive LGN) on E14 reveals anterogradely labeled thalamic fibers that have already grown down through the diencephalon and are passing through the primitive internal capsule. The example in Figure1 is representative of all 14 experiments performed at this age. The thalamic axons run under and through the anlage of the developing corpus striatum to reach the intermediate zone of the ventrolateral telencephalon at approximately E15.5 (see Fig.3).
DiI implantation in the dorsal thalamus, even early on E14, also labels numerous cell bodies in the ventral diencephalon (presumably in the thalamic reticular nucleus; Mitrofanis and Baker, 1993) and a few within the primitive internal capsule, lying below both medial and lateral ganglionic eminences (Fig. 1E,F), in the region that Mitrofanis (1992) calls the perireticular nucleus. Even where they are close to the tips of growing thalamocortical axons, the back-labeled perireticular cells can be convincingly distinguished from the surrounding growth cones by their chromatin content, revealed by matching fine confocal optical sections collected with different filters to distinguish the carbocyanine dye and the bisbenzimide and/or acridine orange counterstaining (Fig. 1G,H,L). The fact that the cells of the thalamic reticular and perireticular nuclei are densely stained with carbocyanine dye (even after the shortest incubation periods) and are intermingled with entirely unlabeled cells makes us confident that they were labeled retrogradely along axons already projecting into the dorsal thalamus at E14, rather than by transcellular diffusion of dye from labeled thalamic axons.
We obtained some information about the topography of the initial outgrowth of thalamic fibers by tracing the axons labeled by two or more crystals of different dyes, placed in the dorsal thalamus. The entire thalamus is so small at this stage that these placements necessarily had to be very close together. However, such experiments reveal that fiber bundles originating from different thalamic regions remain separate as they descend in the ventral thalamus, enter the internal capsule, and grow through and under the developing striatum (Fig. 2). The location of the back-labeled thalamic reticular and perireticular cells was also clearly dependent on the location of the crystal placement within the dorsal thalamus, indicating at least some degree of topographic order in the projection of those two cell groups to the thalamus at very early stages.
The examples in Figure 2 show serial coronal sections (Fig.2A–D) and horizontal (Fig.2E–H) sections from two brains in which a pair of different carbocyanine dye crystals were placed into the dorsal thalamus, only ∼200 μm apart in the coronal plane, at E15. The diffuse halos of dye around the individual crystals are large and overlapping, but it is likely that uptake of dye was largely restricted to the dense cores (Fig. 2A,E). In coronal sections, two differently colored bundles of axons are seen descending side-by-side through the ventral diencephalon (Fig.2A–C). In the most rostral coronal section (Fig.2D) the advancing wave front of thalamic fibers is passing through the primitive internal capsule, beneath the anlage of the striatum. Here, the two axon bundles overlap each other in the coronal plane, but the series of horizontally cut sections from a similarly labeled brain (Fig. 2E–H) shows that they are still quite separate in that plane. Fibers from the more medial thalamic site run rostral to those from the more lateral implantation. This suggests that the array of fibers from this region of the dorsal thalamus rotates by ∼90° but maintains its topographic order as it leaves the diencephalon. Additional evidence for both the topography of the axon array and its rotation comes from retrograde labeling of the same axons from the cortex ∼1 d later (see below).
Growth through the primitive internal capsule and intermediate zone
Between E15 and E16, thalamocortical fibers pass through the primitive internal capsule (Fig. 3). The homogeneous, topographic array of axons leaving the diencephalon (Fig.4C) parcellates into distinct fascicles, which open up in a fan-shaped pattern (Fig.4D), as they pass within the anlage of the corpus striatum. The fibers do not appear to cross each other extensively at any point along their course, although some individual axons and small axon bundles can be seen switching from one fascicle to another within the striatal anlage (Fig. 4B). Viewed in coronal sections, the ordering of fibers is such that the inferior lateral fascicles are destined for the ventral part of the cortex, whereas the superior medial fascicles turn upward and head toward more dorsal cortical areas (Fig. 3G,K).
At the border between the corpus striatum and the intermediate zone the thalamic fibers defasciculate and form a fairly uniform array of individual fibers (Fig. 4E). The trajectories of the axons remain more or less parallel as they turn up into the intermediate zone. Fibers from the dorsolateral thalamus (primordial LGN) are advancing through the intermediate zone beneath the ventrolateral cortex at approximately E15.5, and they reach the subplate layer of the occipital cortex itself at E16 (Figs. 3-5 are examples representative of all experiments; see Table 1). As with the fascicles from which they emerge, the ordering of the individual fibers is such that the more inferior and lateral ones are destined for the more ventral cortical segment, whereas the more medial axons, situated deeper in the white matter, head toward more dorsal cortical segments (Figs. 3,5). The entire, ordered array of thalamic fibers fills the full depth of the intermediate zone, from the cortical subplate down to the subventricular zone (Fig.3G,H).
Accumulation of thalamic fibers below the cortex
Although axons from the region of the LGN reach the occipital cortex at approximately E16, they do not substantially invade the cortical plate for some time. Most of the labeled fibers accumulate in and remain restricted to the intermediate and subplate zones until E19. During this period, thalamic fibers develop increasing numbers of local branches along their length within the subplate layer. These branches are nearly all directed upward toward the cortical plate, but most are short and restricted to the subplate region, at least in the occipital cortex. A few penetrate the lowest part of the cortical plate (Fig. 5, E19), and very rarely some are seen reaching close to the lower boundary of the dense cortical plate—the tightly packed, recently arrived, immature neurons that form the upper part of the cortical plate.
The generation of cortical plate neurons in the ventricular and subventricular zones is at its peak when thalamic axons arrive (Miller, 1988). Migrating cells therefore have to pass through the mass of fibers accumulating in the subplate layer, as well as between the subplate cells themselves.
Invasion of the cortex
Before approximately E19–E20, very few thalamic axons appear to have grown into the cortical plate itself in the occipital region (Fig.5). Invasion on a massive scale suddenly begins at approximately E19–E20, just a couple of days before birth. The initial course taken by these invading fibers is distinctive and highly ordered; most make virtually a 90° turn from their trajectory in the white matter, and the vast majority initially grow radially, straight up into the cortex. A small proportion take more erratic initial routes, running somewhat obliquely upwards (Fig. 6). The ingrowth could partly originate from the side branches of thalamic axons, some of which penetrate the cortical plate at earlier stages.
In favorable preparations, at all stages from E15 onward, individual thalamic axons can be seen forming an impressively ordered, parallel array running up through the intermediate zone and into the subplate layer below the cortex (Figs. 3G, 5, 6A). Fibers heading for more ventral and lateral cortex run superficial to those destined for more dorsal segments. Because the earlier axons enter ventral cortical areas first, and these areas develop relatively early, the ordering of thalamic fibers mirrors the temporal order of maturation of their cortical target areas.
Establishment of laminar termination patterns
By birth (P0, equivalent to E21–E22) and afterward, the trajectories of thalamic fibers within the cortical plate appear far less regular and parallel than in the white matter (Figs. 5,6A). Some individual thalamic axons have branches in layer 6 of the cortical plate and may terminate there; but most of them branch and appear to terminate above in putative layer 5 and especially in layer 4. By approximately P2, most axons have bifurcated at least once near the top of layer 5 and have arborized quite widely in the lamina that is differentiating into layer 4, directly below the dense cortical plate, which is itself reducing in thickness as the upper layers mature. Many axons now appear to have lost their growth cones. A few fibers continue to grow all the way to the marginal zone, but the vast majority terminate within a fairly narrow laminar range and avoid the uppermost sector of the cortex.
Individual labeled axons were followed in 200-μm-thick sections and drawn via a camera lucida (Fig. 6). At P2 their arborizations in presumptive layer 4 typically extend over an area ∼300 μm wide. This wide lateral spread of the terminal arbors of thalamic axons contrasts with the remarkable order with which they are arranged at the junction between white and gray matter (Figs. 5, 6A). Between P2 and approximately P8 (when the cortex has achieved its mature lamination), layer 4 and the arborizations of thalamic afferents within it appear to be progressively displaced downward as the supragranular layers thicken.
Tracing from the internal capsule
Through the entire period of outgrowth of thalamocortical fibers, their accumulation below the cortex and the initial invasion of the cortical plate, small crystal placements in the thalamus itself back-labeled very few cortical cells, even within the subplate, and none was ever labeled from crystals restricted to the dorsal thalamus (Figs. 3-6). Clasca et al. (1995) described similar findings in the ferret; they suggested that early corticofugal projections from the subplate and layer 6 are delayed for several weeks in the cerebral white matter, and that axons from layer 5 are the first to pass through the internal capsule into the diencephalon. We used small crystal placements within the internal capsule itself to determine when corticofugal projections reach the internal capsule in the rat and which cortical neurons project through it at early stages.
At E14 a small crystal implanted into the primitive internal capsule back-labels numerous cell bodies and their processes within the presumptive perirhinal and lateral cerebral cortex. It also labels a segment of the striatal radial glia above and below the implantation site and some adjacent neurons within the internal capsule itself. Figure 7 demonstrates two representative examples from six experiments. Many of the back-labeled cells within the putative perirhinal cortex have a pyramidal shape and differ from the typical polymorphous, polygonal morphology of the labeled preplate cells in the lateral and dorsal cortex (Fig. 7, compare D, E, respectively). Even by E14–E14.5, in the more mature lateral segment of the cortex this population of first-generated cells is already being split into marginal zone and subplate by newly arrived migrating neurons of the emerging cortical plate (Fig. 7E). Interestingly, back-labeled cells are seen in both subplate and marginal zone of the lateral cortex at E14.5, indicating that some neurons from throughout the depth of the original preplate have axons in the primitive internal capsule at this age. Occasionally we observed a few back-labeled cells within the cortical plate itself, at least some of them with pyramidal morphology, but the vast majority are in the preplate or its derivatives (Fig. 7).
At E14.5, crystals placed in the internal capsule back-label cells only in the most mature, lateral neocortical segments. At later ages the territory of back-labeled cortical cells expands dorsally, indicating that axons from the preplate of more dorsal areas reach the internal capsule progressively later than those from more ventral regions.
Carbocyanine crystals implanted into the internal capsule after E16 back-label cells in the subplate but, interestingly, very few if any in the marginal zone, implying that the cells of layer 1 that originally projected to the internal capsule have either died or withdrawn their axons. At this stage cortical lamination is more advanced. Back-labeled pyramidal neurons are seen in both layers 5 and 6, showing that corticofugal projection neurons in all these layers have now reached the internal capsule. Because cells in the cortex can be labeled before birth from the internal capsule but rarely if at all from the dorsal thalamus, we suspect that axons of subplate cells and even those from true corticothalamic neurons of layer 6 slow down or stop somewhere in the ventrolateral part of the diencephalon, medial to the primitive internal capsule, and do not enter the body of the dorsal thalamus prenatally.
Anterograde labeling of the earliest corticofugal projections
E14–E15: establishment of the preplate “scaffold”
The exact appearance of the cerebral wall (and the state of axon outgrowth, as revealed by crystal placements in the cortical surface) varies with position across the hemisphere, at any particular age, in line with the rostrolateral-to-caudomedial gradient of cortical development (Bayer and Altman, 1991). The rostrolateral cortex is ∼1.5 d ahead of the medial occipital cortex in its state of maturity until several days after birth. Unless otherwise stated, the following descriptions apply to crystal placements in the occipital cortex (putative area 17).
Crystal implantation in the surface of the occipital cortex early on E14 results in staining (to varying degrees) of the pial surface of the cortex and of a population of radial processes extending from the ventricular surface to the pial surface, with labeled bipolar, nucleated cell bodies in the lower half of the cerebral wall (Fig.8B). The radial processes are presumably those of radial glia, and the somata probably include both progenitor cells of the ventricular zone and postmitotic, immature neurons, migrating toward the pial surface. In the most dorsal areas of cortex the preplate has apparently not yet begun to form; no cell bodies are present close to the pial surface. Nor are any outgrowing axons seen at this early stage.
During the 14th embryonic day, early-born cells begin to accumulate below the surface of the occipital cerebral wall to form the preplate. Crystal placement in the cortex at successively later times on E14 produces progressively more intense superficial staining, presumably concentrated in the membranes of these preplate cells. The first axons are seen growing out from these cells at mid-E14 (Fig.8C), and they appear, virtually without exception, to be directed ventrally in the intermediate zone, heading toward the basal telencephalon and the primitive internal capsule, which they begin to enter during the second half of E14 (Fig. 8D). Wherever the crystal is placed, a substantial population of axons is labeled, and they form a fairly tight bundle of fibers running within and then directly beneath the subplate zone.
All the axons at the advancing front of the stained bundle end in expanded structures, which are indubitably growth cones rather than cell bodies, because chromatin staining reveals no nuclei within them. These growth cones exhibit complex filopodial morphology and are exceptionally large (∼10 μm across) as they descend in the nascent intermediate zone (Fig. 8F). The bundle of axons originating at any particular point in the cerebral wall maintains its coherence quite tightly and follows a direct, smoothly curving path toward the primitive internal capsule. As they approach the perireticular cells below the lateral ganglionic eminence, the growing tips of the axons are somewhat more dispersed than the diameter of the bundle behind them, perhaps implying that they are entering territory with different properties.
Crystal placements in the occipital cortex on E14 never resulted in the back-labeling of thalamic cells, indicating that no thalamic fibers have reached the site of crystal placement at that age, in agreement with the results of anterograde labeling from the thalamus described above.
Topography of the scaffold
Examination of brains after multiple dye placements, in parasagittal or coronal rows across the cortical hemisphere, clearly demonstrated the temporal order of descending projections from the preplate, matching the gradient of maturation of cortical lamination (Bayer and Altman, 1991). Along each row, the more anterior and lateral the site, the more advanced the fiber outgrowth. These multiple injections also revealed the quite precise topographical organization of these early corticofugal projections; each individual crystal labeled a group of axons, forming a discrete, fairly tight bundle, and the spatial arrangement of the separate labeled bundles was maintained throughout their path.
Along the rostrocaudal axis, at these early stages, axons follow straight or slightly curved trajectories toward the primitive internal capsule and, hence, form a fan-shaped array converging on that region, maintaining topography correlated with the spatial separation of their sites of origin.
Topography is also preserved along the coronal axis (Fig.9), despite the fact that all the fibers from any mediolateral strip of cortex share the same narrow portion of the intermediate zone. Figure 9, B and C, shows the separate, approximately parallel trajectories of the bundles of descending axons labeled by a coronal row of three crystals. At any point in the intermediate zone, the descending axons derived from more dorsal (later-maturing) parts of the hemisphere lie deep to those from more lateral (earlier-maturing) regions; the full array appears to extend through the entire thickness of the intermediate zone. Fiber order is, then, correlated with the time of initial outgrowth. This implies that, as new fibers grow down from progressively more dorsal regions, they take their place immediately adjacent and deep to the array of preexisting fibers, and that this order is preserved throughout the pathway. Thus the array of preplate axons approaching the ventral telencephalon forms a three-dimensional representation of the entire surface of the hemisphere.
Evidence that thalamic fibers grow over the scaffold of preplate axons within the internal capsule and intermediate zone
Combined double labeling of cortical and thalamic axons
By late on E14, both advancing axon fronts, from occipital preplate and dorsolateral thalamus, reach a region in the basal telencephalon, at the lateral edge of the primitive internal capsule. The outgrowths of the two corresponding sets of axons, corticopetal and corticofugal, are remarkably synchronized and coordinated in their timing and spatial patterns.
Following the suggestion of McConnell et al. (1989) that the preplate pioneer fibers might play a part in subsequent axon guidance, we proposed that, after thalamic axons reach the internal capsule, they complete their navigation to the appropriate cortical region by growing over fibers that have descended from their target area—the “handshake” hypothesis (Blakemore and Molnár, 1990;Molnár and Blakemore, 1995). To examine this notion, we attempted to label corresponding regions of dorsal thalamus and occipital cortex with different dyes to see whether the two fiber systems do literally contact each other, intermingle, and appear topographicaly aligned within the intermediate zone. Results from such experiments performed at mid-E16 (Table 1) were rejected when we discovered that cell bodies in the thalamus were retrogradely labeled with the dye placed in the cortex. This was to be expected from the results of the thalamic tracing experiments described above, which showed that thalamic axons reach the occipital cortex at that age (Fig. 3). At even later stages, the picture becomes further complicated by the outgrowth of fibers from layers 6 and 5 of the true cortical plate (De Carlos and O’Leary, 1992; Miller et al., 1993). Our analysis with this double-labeling technique was therefore restricted to the period from E14.5 to very early E16.
The fundamental problem with this approach is that the exact correspondence between developing thalamic and cortical loci cannot be established empirically until thalamic fibers have actually reached their target areas in the cortex, but after that point it is obviously very difficult reliably to label fibers arising from cells in that exact target region separately from the incoming thalamic axons. The handshake hypothesis demands that thalamic axons should intermingle only with axons from the corresponding region of the cortex; therefore, any misalignment of dye placements in the two structures would be expected to reveal spatially separate groups of labeled fibers. Indeed, in 11 of 26 (42%) attempts to label matched cortical and thalamic sites with small crystal placements, from E14.5 to E15.5, the bundles of differently labeled descending and ascending axons lay close to each other within the intermediate zone but did not overlap extensively, just as in the results of Miller et al. (1993). However, because of the inherent difficulty in defining correspondence, negative results in such experiments are to be expected; even a small fraction of positive results would be definitive in demonstrating that the two axon systems grow through the same compartment within the intermediate zone.
In fact, in the majority of cases (5 of 8 at E14.5, 4 of 10 at E15, and 6 of 8 at E15.5) there was definite intermixing of a substantial fraction of the bundles of labeled preplate and thalamic axons in the ventral telencephalon and intermediate zone (Fig.10 shows several representative positive examples). This was also true in one case at early E16 in which the labeled thalamic axons had not quite reached their target area, and no cells back-labeled by the cortical dye were seen in the thalamus. In several of these positive cases, the overlap of thalamic and corticofugal axons was extensive.
In these 16 successful double-labeling experiments, conventional fluorescence microscopy showed the two fiber arrays to be mixed together within individual 100-μm-thick sections, with differently colored fibers lying side-by-side in the same plane of focus. This was also the case in the thin optical sectioning afforded by confocal microscopy. For example, individual red and green fibers were seen close together in many of the 32 individual 1-μm-thick sections that were stacked to create the extended focus image of Figure10F. The stereo pair derived from this data set (Fig.10G) clearly shows corticofugal and thalamic fibers co-resident in single, bidirectional fascicles.
Cross-sectioning of the mixed bundles, in the intermediate zone lateral to the striatal anlage, confirmed the very close association of the two populations, with individual descending and ascending axons often lying side-by-side, literally in contact. In Figure 10C, for instance, there is a considerable area of overlap, with some red (thalamic) and green (corticofugal) axons so close together that they appear optically fused as yellow.
Labeling of cortical neurons and thalamic axons as they approach the subplate
Between E15 and E16, a dye crystal in the primitive internal capsule labels not only early corticofugal axons and their cell bodies (retrogradely) but also thalamic axons growing toward the cortex (anterogradely). In 42 such cases we examined the relationship between ascending and descending axons that pass through the same region of the internal capsule, especially in brains that were cut in a plane (near-coronal) chosen to be parallel to the axon arrays close to the labeled region of cortex.
Figure 11 shows results from an E15.5 brain in which a crystal of DiI had been placed in the medial part of the primitive internal capsule. At the rostrocaudal level of the crystal site, an ordered array of intensely stained, thick thalamic axons, identified by the growth cones at their tips, is advancing into the basal part of the intermediate zone. A number of thinner axons, intermingled with this mass of thalamic fibers, extend all the way back to the cortex, some of them unequivocally identified as corticofugal by being followed back, in serial confocal sections, to labeled cell bodies in the subplate (Fig. 11C). Further rostral, where the entire pattern of growth is more advanced, a dense array of thin axons without growth cones leads back to a population of back-labeled cells in the cortical subplate and some in the lowest part of the cortical plate. Many of these fine, labeled axons can be traced back individually to their cells of origin. Among this ordered descending array of corticofugal axons, thicker axons, presumed to be thalamic, because many of them end in definite growth cones within the section, are seen growing up toward the cortex, running parallel to and in intimate association with corticofugal fibers (Fig.11D). There is no tendency for the two types of axons to separate from each other, even as the thalamic axons enter the subplate layer.
Figure 12 is an enlarged view of the region of the confocal image of Figure 11D, in which the outlines of several identified corticofugal axons have been filled in black, together with their cells of origin, whereas the outlines of some of the definite thalamic fibers, identified by their growth cones, are shaded white. The two types of identified axons run parallel, in close association, filling the same compartment of the intermediate zone, as the thalamic fibers approach within 100 μm of the cortex.
Shared topography of early corticofugal and thalamocortical projections
Although exactly corresponding thalamic and preplate axons cannot reliably be separately labeled after thalamic fibers have reached the subplate layer, further support for the handshake hypothesis can be drawn from experiments involving crystal placement in the cortex alone at slightly later ages. Fibers from the putative LGN reach the subplate of the occipital pole of the cortex at E16 (Fig.3G is representative for all experiments; n= 8). In accordance with this, a single dye crystal in the occipital cortex at E16 leads to the labeling not only of fibers, extending through the telencephalon and diencephalon, linking the two structures (Fig. 11C), but also of cell bodies in the LGN itself (especially clear after relatively long incubation periods). Therefore the fibers labeled from any such single site in the cortex must be amixed population of anterogradely labeled, corticofugal axons and retrogradely labeled, thalamocortical fibers. And, by definition, they must be spatially corresponding. If these two sets of axons were to take very different routes through the telencephalon, any single crystal should stain two or more distinctly separate bundles.
We therefore examined the axon arrays labeled by small dye placements at various points on the cortex, at ages from E16 onward, following the axons in serial coronal or horizontal sections with conventional fluorescence and confocal microscopy. As well as viewing these serial sections, we made three-dimensional reconstructions from thin confocal optical sections at various magnifications to reveal more of the fine and gross topography.
Unfortunately, even very small crystal placements lead to a wide halo of label around the uptake site itself, especially after the relatively long incubation periods that we used to obtain good retrograde labeling of thalamic cell bodies. This halo consists partly of intrinsic cell and fiber staining in the cortex itself, partly of direct labeling of radial processes and cells in the subadjacent intermediate and subventricular and ventricular zones, and partly of background dye that has presumably diffused radially around the crystal. The fact that the total bundle of fibers emerging from such a halo and running down toward the internal capsule is always much smaller than the diameter of the halo implies that the effective uptake site for fiber labeling is very much smaller than the halo, as with most axonal tracers.
Although the labeled fibers could not be traced reliably within the dye halo itself, very close to the uptake site, they could be clearly resolved over the remainder of the pathway, throughout most of the intermediate zone, the internal capsule, and right into the thalamus. The mixed descending and ascending axons labeled by any small crystal placement in the cortex at E16 form a discrete bundle rather than two separate fiber tracts. This single, tight bundle is maintained throughout its trajectory within the intermediate zone and primitive internal capsule, over the region containing both sets of axons. The stained bundle narrows as it approaches and runs through the internal capsule, where, for a small crystal site, it can be as tiny as 50 μm in diameter (Fig. 13B). A fine ribbon of stained fibers extends into the diencephalon and follows a smoothly curving trajectory to the appropriate region of the thalamus, ending in a cluster of back-labeled cells.
From serial confocal microscopic optical sections, we made high-resolution reconstructions to reveal the fine topography of stained fibers within the entire bundle between thalamus and cortex. This technique did not quite allow us to follow individual axons all the way from the implantation site to the thalamus, but it did show that the bundle consists of an approximately parallel set of axons throughout its course, with very little crossing over within the array (just as in the images of separately labeled thalamic and preplate axons seen at earlier stages; Fig. 10F).
Because this method does not allow discrimination of the two classes of fibers, it is impossible to be sure whether the segment of the labeled bundle within the diencephalon contains corticofugal axons. However, the bundle usually appears thinner and yet less densely packed within the diencephalon (Fig. 13C), suggesting that it contains fewer fibers. And the fact that we never succeeded in back-labeling subplate cell bodies with small dye placements restricted to the dorsal diencephalon (see above) also implies that their axons do not reach the relay nuclei of the thalamus itself. Therefore, the fine bundle of stained axons running within the thalamus, toward the group of back-labeled thalamic neurons, probably consists entirely of the retrogradely stained axons of those cells.
It is important to note that the topographic arrangement of the fiber bundles in the diencephalon and the cell clusters in the thalamus itself, after cortical placement at E16, is entirely compatible with the picture derived from the implantation of crystals in the thalamus ∼1 d earlier, shortly after axon outgrowth (compare Figs. 2,13D). This strengthens the conclusion that outgrowing thalamic axons form a highly ordered array, even within the diencephalon itself.
Taken together with the double-labeling experiments and those in which both descending and ascending axons were labeled from the internal capsule, these results suggest that (1) the topography of the scaffold of early corticofugal axons within the intermediate zone and basal telencephalon is extremely similar in form to that of the thalamocortical fibers; (2) each set of fibers appears to be distributed, in order, within the entire depth of the intermediate zone; and (3) ascending and descending axons share the same compartment and are intermingled, through an extensive region of the ventral telencephalon, to within a very short distance of the target area for the thalamic fibers.
It might be argued that some mixing of the two sets of axons is inevitable if they share a single extracellular growth compartment, especially as they approach the constriction of the internal capsule. However, compelling evidence for global, topographically ordered intermeshing of the two axon arrays comes from the experiments involving multiple cortical dye placements (Figs. 13A,B,D,14), in which the individual bundle of fibers labeled by each crystal can be followed all the way to the thalamus. At E16, each distinct bundle is unitary and clearly separate from the others, even as they converge and pass through the primitive internal capsule. This is most clearly shown in Figure 14,C and D, which is a three-dimensional reconstruction of the entire trajectory of two closely spaced fiber bundles running between frontal cortex and anterior thalamus.
Within the resolution of this method, the apparently single bundle labeled from any one crystal could conceivably represent closely adjacent but nonoverlapping strands of thalamocortical and corticofugal axons. But obviously, close proximity of corresponding parts of the two axon arrays without overlap of any part is incompatible with the overall topography and closely packed fiber ordering seen within each array. Ascending thalamic axons appear, then, to follow the same topographically ordered paths through the telencephalon as do their counterparts descending from the subplate. Either the two arrays are systematically very slightly misaligned from each other, or, as demanded by the handshake hypothesis, they are spatially corresponding.
Retrograde labeling of thalamic neurons: global topography of the projection
Although it is impossible to resolve the individual components of the mixed fiber population labeled from a cortical implantation site, important inferences about their interrelationships and topography can be drawn from the patterns of retrograde labeling of thalamic cell bodies. In horizontal sections, the thalamic cells back-labeled from any single crystal in the cortex tend to form an anteroposteriorly elongated slab. Different crystals placed in a parasagittal line along the hemisphere at E16 revealed that more anterior cortical regions are innervated by more medial slabs within the developing thalamus, whereas more posterior cortical regions receive from more lateral thalamic slabs (Fig. 14A). In concordance with this, any two adjacent cortical implantation sites arranged along a coronal line produced a pair of elongated groups of back-labeled thalamic cells aligned to form a more-or-less continuous anteroposterior slab, the rostral portion of which projects more ventrally in the cortex (Fig.14B).
Other corticofugal fiber systems
De Carlos and O’Leary (1992) showed that pyramidal neurons of the lowest part of the true cortical plate (presumably layer 6) start to extend axons into the white matter as early as E15 in the rat. Interestingly, at E16, axon bundles labeled from the cortex were always unitary throughout their visible course, although the telencephalic portion should have contained at least some axons of layer 6 cells as well as those of preplate cells and thalamocortical fibers. This surely means that all these axon systems run close to each other. And the separate, ordered arrangement of bundles resulting from multiple dye placements implies not only that all the component axon types share the same growth compartment but also that they are distributed according to the same overall topography.
For the portion of the pathway running through the intermediate zone, this general pattern persists even after E18–E19, when there is also significant extension of axons from layer 5 of the cortical plate, ultimately directed toward structures below the diencephalon (De Carlos and O’Leary, 1992; Mitrofanis and Guillery, 1993). Again this implies topographically ordered mixing of all the fiber systems within the telencephalon. However, the capacity of the single-crystal technique to demonstrate divergence of axon populations when they do exist is established by the fact that, at these later ages, each single dye placement in the cortex does indeed reveal a fiber bundle (presumably consisting of layer 5 cell axons) that diverges from the main labeled tract into the posterior extension of the primitive internal capsule to form the putative cerebral peduncle (Fig.15). At this stage, dye placed in the posterior part of the primitive internal capsule exclusively labels cell bodies in layer 5 of the cortical plate itself, presumably the cells of origin of corticotectal, corticopontine, and corticospinal projections. Thus, it seems that these nonpioneering, later-descending fibers from layer 5 share the same gross topography as the earlier ascending and descending projections within the telencephalon (although possibly systematically slightly displaced from them), but that the layer 5 axons follow radically different trajectories in the ventral telencephalon, presumably responding to guidance cues in that region.
DISCUSSION
Topographic order of thalamic axons
The arrangement of the bundles of thalamic axons labeled by crystal placements in the thalamus, shortly after their outgrowth (Fig.2), or in the cortex, immediately after their arrival (Figs.13D, 14C,D), suggests that they maintain ordered topography, running as an approximately parallel array along their entire route, with a single 90° rotation of the array as it leaves the ventral diencephalon (Fig. 2). Fibers from parasagittal slabs of thalamic cells innervate loci lying along approximately coronal lines across the cortex; more ventral cortex is innervated by cells lying more rostral in the thalamic slab, and the more anterior the coronal strip, the more medial (and slightly ventral) is the corresponding thalamic slab (Fig. 14). The global topographic relationship between thalamus and cortex appears very simple at this stage, before the clear segregation of thalamic nuclei. The entire volume of the thalamus maps continuously onto the cortical sheet.
The initial pattern of thalamocortical innervation could subsequently be altered through differential growth in thalamus or cortex, irregular ingrowth of thalamic fibers, selective growth and retraction of their side branches (Naegele et al., 1988; Ghosh and Shatz, 1992), or activity-dependent redistribution of thalamic afferents in the cortex (see Shatz, 1990; Molnár, 1998).
Initial guidance of thalamic outgrowth
The ordered outgrowth from the thalamus might depend simply on intrinsic interactions between neighboring thalamic fibers. However, a number of other neural systems in this region could be players in the process of guidance. Like Mitrofanis and Baker (1993), we saw cells of the thalamic reticular nucleus back-labeled from the dorsal thalamus at E14 and also small numbers of back-labeled neurons within the primitive internal capsule, presumably belonging to the “perireticular” cell group of Mitrofanis (1992)but projecting into the thalamus earlier than previously described (Mitrofanis and Baker, 1993; Mitrofanis and Guillery, 1993). Thus, at least two populations of fibers have reached the thalamus when thalamic axons are emerging, and they could conceivably play a part in guiding that initial outgrowth.
The “handshake” hypothesis: possible guidance by preplate axons
Axons from preplate cells pioneer a pathway through the intermediate zone (McConnell et al., 1989; Blakemore and Molnár, 1990; De Carlos and O’Leary, 1992; Erzurumlu and Jhaveri, 1992; Miller et al., 1993; Molnár et al., 1998). Rows of dye crystals placed in the cortex as early as E15 show that axons form a highly ordered array, reflecting the spatiotemporal wave of maturation that sweeps across the developing cortex. Moreover, this array fills the intermediate zone, through which thalamic axons subsequently grow in the opposite direction as a similar topographic array. Back-labeling from the internal capsule at this stage (Fig. 7) shows that cells in both divisions of the original preplate—marginal zone as well as subplate—contribute to the early corticofugal axon array.
Corresponding thalamic and cortical axons enter the lateral part of the internal capsule approximately simultaneously, from opposite sides (McConnell et al., 1989; Molnár and Blakemore, 1990; Bicknese and Pearlman, 1992; De Carlos and O’Leary 1992; Erzurumlu and Jhaveri, 1992; Molnár et al., 1998). Preplate axons could constitute a scaffold (McConnell et al., 1989) over which thalamocortical fibers grow toward the appropriate cortical area—the handshake hypothesis (Molnár and Blakemore, 1995).
The most direct test of the handshake hypothesis is to use double labeling to see whether axons from corresponding cortical and thalamic regions do meet and intermingle. There is a narrow window of opportunity, when the two axon systems can indubitably be separately labeled from the cortex and the thalamus (E14.5 to early E16 for the geniculocortical pathway). But, paradoxically, it is impossible to be certain of the precise spatial correspondence between thalamic and cortical regions until thalamic axons actually reach their target areas.
Miller et al. (1993) did not see obviously intermixed thalamic and cortical axons in any of their nine double-labeling experiments in the hamster. But the handshake hypothesis predicts that even slight offset in double labeling should result in nonoverlapping fiber bundles, so negative results are not definitive. We too saw little or no overlap in 42% of 26 such experiments between E14.5 and E15.5 in the rat. However, in 58% of cases, substantial fractions of the axon bundles were clearly mixed, with thalamic and preplate fibers juxtaposed, following the same trajectories, even in thin confocal optical sections and in stereo reconstructions (Fig. 10). Ascending and descending axon systems, at least for the dorsolateral segment of the cortex, share the same compartment within the internal capsule and the intermediate zone.Bicknese and Pearlman (1992) and Erzurumlu and Jhaveri (1992), using double labeling in the mouse and the hamster, also reported that thalamic and early corticofugal fibers interdigitate in the internal capsule.
The full handshake hypothesis proposes that thalamic fibers navigate all the way to the cortex along the axons of subplate cells. However,Bicknese et al. (1994) described thalamic axons in the rat extending within the subplate layer, whereas fibers of subplate and cortical plate cells ran below the subplate. On the basis of these observations, they suggested that ascending and descending axons run in separate, adjacent compartments close to the target area. On the other hand, labeling with small crystals in the internal capsule between E15 and E16 clearly shows a broad array of identifiable thalamic axons running in close association with corticofugal fibers, right up to the subplate (Figs. 11, 12). These findings, together with the positive results of double labeling, strongly suggest that many thalamic fibers follow preplate axons right to the target area. Although some thalamic axons may continue to grow further within the subplate layer, the two projection systems appear to share the same growth compartment and to run along similar trajectories within the intermediate zone, at least toward the dorsolateral segment of the cortex.
Further powerful evidence that descending and ascending arrays are at least approximately locked in correspondence comes from the arrangement of individual, discrete bundles of mixed afferent and efferent axons seen after multiple crystal placements in the cortex at E16 (Fig.14D). The results of similar experiments at later ages, when the axons of cells of layers 6 and 5 are also labeled, suggest that they follow the same basic global topography as the earlier projections, until the point at which layer 5 axons peel off at the medial edge of the internal capsule to form the cerebral peduncle (Fig. 15). Thus the pioneering preplate scaffold may play some part in guiding not only afferent thalamic axons but also other descending corticofugal projections, as suggested by McConnell et al. (1989).
In the adult rat, corticofugal axons from the dorsal occipital (visual) cortex run some 150 μm deep to the corresponding thalamocortical axons in the white matter (Woodward and Coull, 1984; Woodward et al., 1990). Perhaps postnatal loss of subplate axons leads to the appearance of this distinct misalignment of afferent and efferent fibers in the adult.
Of course, even close association of corresponding preplate and thalamic axons does not prove that the former actually guidethe latter. But, at the very least, the early corticofugal scaffold and the subplate layer appear to provide attractive compartments within which thalamic axons grow. The subplate is rich in chondroitin sulfate proteoglycans (Bicknese et al., 1994), especially neurocan (Fukuda et al., 1997), and it is possible that proteoglycan expression in the intermediate zone, although much weaker (Bicknese and Pearlman, 1992;Fukuda et al., 1997), is associated with the axons of subplate cells. Indeed, Emerling and Lander (1994) found that dissociated embryonic thalamic neurons selectively adhere to and extend neurites on the intermediate zone as well as the subplate of embryonic cortical slices (E15–E16), and this behavior is eliminated by the enzymatic removal of chondroitin sulfate (Emerling and Lander, 1996).
Is there a “waiting period” in rodents?
Although the accumulation of thalamic fibers below the cortical plate, which is such a striking feature of development in carnivores (Shatz and Luskin, 1986; Ghosh and Shatz, 1992) and primates (Rakic, 1977, Kostovic and Rakic, 1990), was first described by Lund and Mustari (1977) in the rat, the existence of a waiting period in rodents has recently been questioned. Catalano et al. (1991, 1996) and Kageyama and Robertson (1993) observed thalamic fibers simply advancing steadily into the cortex as it matures. We saw some axons entering the lower layers of the occipital cortex as early as E16 (Fig. 5), but they were rare and trivial in density compared with the massive invasion that starts at E19–E20, 2–3 d after their arrival (Figs. 3, 5). This is compatible with the results of co-culture experiments, which suggest that the occipital cortex does not express membrane-bound growth-permissive properties until approximately E19–E20 (Götz et al., 1992; Molnár and Blakemore, 1995).
Chemospecificity and chronotopy
Molecular signals, perhaps expressed in gradients across the developing cortex, might play a part in guiding, or at least refining, the distribution of thalamic fibers. Experiments involving co-culture of thalamic and cortical explants (Yamamoto et al., 1992) suggest that the cortex produces a diffusible factor that stimulates the outgrowth of thalamic neurites from as early as E14 (Blakemore and Molnár, 1990; Molnár and Blakemore, 1991, 1995; Rennie et al., 1994;Lotto and Price 1995; Molnár, 1998). This remote influence may play a part in initiating outgrowth of thalamic axons in vivo, but it seems inconceivable that it could impose the topographic order that they exhibit throughout their course.
Bolz and Götz (1992) have shown that a thalamic explant extends neurites somewhat more profusely over a surface coated with membranes prepared from the corresponding region of cortex than over a membrane preparation from the opposite end of the hemisphere. However, the cortex does not express membrane-bound growth-permissive properties until after thalamic axons have already arrived, in topographic order, below the cortical plate; it therefore seems unlikely that such properties control their overall distribution. However, chemical cues expressed at the growing tips of preplate axons might play a role in establishing their relationships with thalamic axons during the handshake.
In the rat, like all mammalian species so far examined, there is a pronounced gradient of maturation across the cerebral hemispheres (Bayer and Altman, 1991). There is also a temporal pattern of maturation within the thalamus (Jones, 1985), the first thalamic axons to reach the cortex being from the ventrobasal complex (Braisted and O’Leary, 1995). If outgrowing axons deposit themselves on the carpet of axons that have already extended, and if axons tend to maintain neighborly order and not to cross each other, the timing of fiber outgrowth could potentially determine one axis of the initial mapping of thalamus to cortex (Molnár and Blakemore, 1995).
Footnotes
This study was supported by grants from the Medical Research Council (MRC), the Wellcome Trust, the Soros-Hungarian Academy of Sciences Foundation, the Human Frontier Science Program, and Merton College (Oxford, UK). It forms part of the work of the Oxford McDonnell-Pew Centre and the MRC Interdisciplinary Research Centre for Cognitive Neuroscience, for which R.A. is a Research Scientist. Z.M. held an MRC Training Fellowship. We are indebted to Laurence Waters, Lorraine Chapell, and William Hinkes for help with photography and to Pat Owens and Duncan Fleming for excellent animal care.
Correspondence should be addressed to Zoltán Molnár, University Laboratory of Physiology, Parks Road, Oxford OX1 3PT, UK.