Cerebral Cortex, Vol. 9, No. 1, 50-64,
January 1999
© 1999 Oxford University Press
Morphogenesis of Callosal Arbors in the Parietal Cortex of Hamsters
Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, Brazil and , 1 Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Boston, MA, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The morphogenesis of callosal axons originating in the parietal cortex was studied by anterograde labeling with Phaseolus lectin or biocytin injected in postnatal (P) hamsters aged 7–25 days. Some labeled fibers were serially reconstructed. At P7, some callosal fibers extended as far as the contralateral rhinal fissure, with simple arbors located in the homotopic region of the opposite cortical gray matter, and two or three unbranched sprouts along their trajectory. From P7 to P13, the homotopic arbors became more complex, with branches focused predominantly, but not exclusively, in the supra- and infragranular layers of the homotopic region. Simultaneously, the lateral extension of the trunk axon in the white matter became shorter, finally disappearing by P25. Arbors in the gray matter were either bilaminar (layers 2/3 and 5) or supragranular. A heterotopic projection to the lateral cortex was consistently seen at all ages; the heterotopic arbors follow a similar sequence of events to that seen in undergo regressive tangential remodeling during the first postnatal month, as the lateral extension of the trunk fiber gets eliminated. Radially, however, significant arborization occurs in layer-specific locations. The protracted period of morphogenesis suggests a correspondingly long plastic period for this system of cortical fibers.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The relatively late maturation of interhemispheric connections, and the easy accessibility of the cortex and the corpus callosum to surgical manipulation, have resulted in an increased use of callosal projections as a model system for the study of neural development. Such reports have included examination of maturational processes such as axonal pathfinding, targeting of afferents to particular cell groups, establishment of regional and laminar specificity, mapping of topographic representations, and maturation of axon arbors. Over the last two decades, these studies have led to the formulation of important principles of cortical development. We now know that as interhemispheric connections form, a widespread, homogeneous distribution of callosal neurons transforms into a discontinuous pattern that is typical of the adult mammal. Such a transient exuberance has been reported for callosal neurons in many species (for reviews, see Innocenti, 1986
, 1995) and also for intracortically projecting cells located in several different regions of the hemisphere (Price and Blakemore, 1985; Clarke and Innocenti, 1986; Lent et al., 1990; Price et al., 1994); however, the generality of these obser- vations remains under debate (Dehay et al., 1988; Schwartz and Goldman-Rakic, 1991; Kennedy et al., 1994; Barone et al., 1996). With maturation of the cortex, definitive callosal projections are sculpted out by the elimination of inappropriately directed fibers (Innocenti, 1981; Ivy and Killackey, 1981; O'Leary et al., 1981). This process leads to an overlap in the remaining zones of callosally connected cells and the terminal fields of callosal axons that originate in the opposite hemisphere.
Despite the exuberant early distribution of callosally project- ing neurons, anterograde labeling reveals that the axons of these cells penetrate into the gray matter of only those cortical regions that will be their targets in the mature animal (Wise and Jones, 1976; Innocenti, 1981; Ivy and Killackey, 1981; Innocenti and Clarke, 1984; Floeter and Jones, 1985; Olavarria and Van Sluyters, 1985; Lent et al., 1990; Schwartz and Goldman-Rakic, 1991; Norris and Kalil, 1992). Direct confirmation of this conclu- sion was presented recently by Aggoun-Zouaoui and Innocenti (1994), who serially reconstructed individual, biocytin-labeled callosal axons originating in area 17 of the cat cortex. They showed that, with the exception of cells located at the border between areas 17 and 18, few cortical neurons with axons in the contralateral white matter invade the overlying cortical plate to a significant degree.
While the work of Innocenti's group has provided some insight as to the innervation process of the cat visual cortex by individual axons, such a level of analysis is not available for any other cortical field, either in cats or in other species. For rodents, the earlier studies on callosal connections were performed with the use of tracers that did not show the morphology of individual fibers. Thus, they revealed general patterns of termination for populations rather than the morphology of single fibers (Wise and Jones, 1976; Innocenti, 1981; Ivy and Killackey, 1981; Floeter and Jones, 1985; Olavarria and Van Sluyters, 1985; Lent et al., 1990). Some information on interhemispheric projections of individual, developing axons is available for visual (Hogan and Berman, 1990; Fish et al., 1991; Elberger, 1994) and sensori- motor cortical fibers (Norris and Kalil, 1992). However, of these studies only Fish et al. (1991) used a tracer that transports exclusively anterogradely — a technical feature that is important for differentiating the afferent axon from profiles that might arise from the labeling of recurrent axons and dendritic processes. And none of these studies attempted to reconstruct axon trajectories over long distances through serial sections. Thus, there is little information on the innervation strategy of individual callosal fibers, on later stages of callosal arbor formation and on the types of arbors present in the mature rodent cortex. The importance of understanding the normal morphological sequence is especially important for the rodent brain: powerful in vitro techniques and genetic tools are now available which can help us perturb the system and further dissect the molecular mechanisms that might be involved in axon guidance and arbor formation.
In this study we present an analysis of the connectional patterns of immature callosal axons labeled with exclusively anterograde tracers. Some of these axons were serially recon- structed, and their trajectories were traced through multiple sections, enabling us to evaluate their developmental fate in the tangential as well as radial dimensions. We document that, in contrast to earlier reports on the hamster visual cortex, callosal projections from parietal cortex exhibit arbors that are relatively complex. While there is an extensive overshooting of the target regions by the parent fibers, their collaterals penetrate the cortical plate only in appropriate regions of the cortex, and arborize in the correct laminae. Heterotopic projections derive from axons that are distinct from those giving rise to the homotopic projection.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All surgical procedures involving the use of live animals were approved by the Committee on Animal Care at MIT.
Thirty Syrian hamsters, aged P (postnatal day) 7, P10, P13, P16, P21 and P25 were used in this study (P1 = day of birth). The oldest age included was chosen on the basis of previous studies on the hamster which show that a mature callosal cell distribution is present after the second postnatal week (Lent et al., 1990), and that myelination of callosal axons is well under way by P15 (Lent and Jhaveri, 1991). Until they were 2 weeks of age, animals were anesthetized i.p. using Chloropent (Fort Dodge Labs, 0.35ml/100g body wt); older hamsters were anesthetized with a mixture of Nembutal (0.3ml/100g body wt) and Valium (0.1ml/100g body wt), supplemented with Chloropent when necessary.
Each pup received cortical injections of biocytin (Sigma, St Louis, MO; Molecular Probes, Eugene, OR) or Phaseolus vulgaris leucoagglutinin (PHA-L, Vector Labs, Burlingham, CA). Biocytin was used in younger animals due to its shorter (1–2 days) transport time, as compared to the 4–5 days necessary for complete filling of axons by PHA-L. Two numbers are provided for the age of each animal: the day of injection and that of perfusion (e.g. P25–29).
Biocytin Injection, Perfusion and Tissue Processing
The procedure we used was similar to that of King et al. (1989). Glass micropipettes with extruded filaments were pulled to an external tip diameter of 15–20 µm, and were backfilled with a 5% solution of biocytin in phosphate buffer (pH 8.0, 0.05 M). The parietal cortex was exposed and injections were placed between the two major branches of the middle cerebral artery at a mediolateral position which located them on the curvature of the cortex. In some cases, additional applications of biocytin were made more medially to increase the probability that at least a few callosal axons would be labeled in case one of the injections happened to fall on an acallosal region. With the aid of a micropressure injector (Picospritzer, World Precision Instruments, Sarasota, FL), small deposits of tracer were made along the depth of the cortex at 100 µm intervals, up to a maximum of 600 µm below the pial surface. The skin over the skull opening was sutured and the animals were allowed to recover in a warm, humidified chamber, or were returned to the nest.
Hamsters were killed 24–48 h later with an overdose of Nembutal, and were perfused transcardially with phosphate-buffered saline (PBS; 0.1 M, pH 7.4) followed by buffered 4% paraformaldehyde, and finally by buffered 10% sucrose. Brains were removed, cryoprotected with buffered 30% sucrose, frozen, and cut on a sliding microtome, in the coronal plane at a thickness of 70–80 µm. All subsequent steps were performed under gentle agitation. The tissue was washed in Tris (0.05 M, pH 7.4) buffered saline (TBS), and was incubated for a total of 60 min in three changes of TBS containing 0.1–0.5% Triton-X100. Sections were immersed in a solution containing avidin and biotin complexed with horseradish peroxidase (HRP) (Vectastain Standard ABC Kit, Vector Labs), washed in PBS, transferred to acetate buffer (0.1 M, pH 6.0) and incubated for 30 min in 2.5% nickel ammonium sulfate and 0.05% diaminobenzidine (DAB) in acetate buffer (pH 6.0). B-D-glucose (200 mg) and ammonium chloride (40 mg) were added to 100 ml of this solution and the reaction for visualizing the HRP was initiated by adding 0.5–1.0 mg of glucose oxidase. Sections were rinsed, mounted onto gelatinized slides, air dried, dehydrated and coverslipped.
PHA-L Injection, Perfusion and Tissue Processing
Micropipettes (tip diameter 5–15 µm) containing a 5% solution of PHA-L (Vector) in PBS (0.1 M, pH 8.0) were positioned as described above. The tracer was delivered iontophoretically into the dorsolateral cortex, using a high-voltage current source (Midgard Electronics, positive current, 4–8 µA, 7 s on, 7 s off) for 5–15 min. After 4 or 5 days, animals were perfused transcardially with a rinse of phosphate buffer, then with fixative containing buffered 4% paraformaldehyde and 0.5% glutaraldehyde, and finally with a wash of 0.34% L-lysine (Sigma) and 0.05% sodium- M-periodate (Sigma) in 0.1 M phosphate buffer. Cryoprotected brains were sectioned as above and the PHA-L was localized using the method of Gerfen and Sawchenko (1984) as modified by Wouterlood and Gronewegen (1985). Briefly, sections were incubated in 2% normal rabbit serum (NRS) to block nonspecific immunolabeling, then in goat anti-PHA-L antibody (Vector Labs, 1:2000) for 48 h at 4°C, rinsed, and immersed in biotinylated rabbit anti-goat antibody (Vector Labs, 1:200). Antibody binding was localized by reaction with avidin–biotin– peroxidase complex (Vectastain Standard Kit, Vector Labs), followed by visualization of the peroxidase using DAB as a chromogen. Intensification of the reaction product was achieved by adding 2% cobalt chloride or 2.5% nickel ammonium sulfate (in Tris buffer) to the DAB treatment.
Analysis
Typical injection sites ranged from 100 to 500 µm in mediolateral extent (Fig. 1A,B) and ~1 mm in the rostrocaudal dimension. Axons were selected for reconstruction when the labeling was judged to be complete, as indicated by the immunostaining of growth cones or terminal boutons at their distal ends (Fig. 1C).
|
Axons listed in Table 1
were derived from 13 (out of a total of 30) animals and were serially reconstructed using 50x and 100x objectives, on a Nikon Optiphot microscope connected to a computer. The computer was equipped with Neurotrace, a serial reconstruction system capable of encoding x, y and z coordinates along the axon (Passera et al., 1988). Neurotrace can compute the total length of the axon, the length of each axon segment, the number of branch points, and various aspects of axon topology (see Antonini and Stryker, 1993). It permits rotation of the reconstructed fiber in three dimensions, and has a `z slicing' function that generates a virtual section of the reconstructed axon along an arbitrarily defined z axis, thus allowing a display of specific sectors of the axon arbor. This function was used to determine, for example, the parts of the axon arbor which were restricted to the superficial regions of the cortex and those which were found only in deeper layers. A computerized drawing program (DiaCad) was used to generate the final drawings.
|
Single axons (n = 25) were reconstructed through several serial sections, and their trajectories were followed retrogradely, from the arbors to the parent fiber in the white matter, all the way back to the midline. This permitted a complete view of their course within the contralateral cortex. Ten axons from P7–9 and P10–12 pups were additionally reconstructed manually from serial sections, with the aid of a drawing tube. These fibers are not included in Table 1
, but some of them appear in Figure 2. Charts of selected axon projection patterns were also generated for different ages, using a drawing tube attachment on the microscope. Sections were not counterstained, but the areal and laminar location of individual axons was determined under phase contrast, or by estimation from a parallel set of cresyl violet-stained reference sections from age-matched normal animals.
|
Axons reconstructed with the aid of the computerized system were submitted to quantitative analysis. The following measures were made for each of these axons: (i) arbor surface area, obtained with a digitizing tablet by measuring the area (AutoCad) within an equidistant outline around the arbor as viewed from the pia; (ii) total axon length, obtained by summing the lengths of all axonal segments after the first bifurcation point of the fiber; (iii) number of branch points, obtained by counting the number of bifurcations in each axon; and (iv) average segment (Antonini and Stryker, 1993
), represented by the ratio between the total length and the number of branch points. Terminology
Collaterals and collateral branches designate sprouts arising from the trunk axons. Branches refer to ramifications present in the gray matter.
Trunk axon or main axon or parent axon all name the portion of the axon that is in the white matter and from which collaterals sprout. It is named thus even after the first bifurcation arises in the white matter. Arbor designates the densely branched part of the axon in the gray matter.
Since the serial reconstructions are very time-consuming, it was not possible to harvest large numbers of axons. Thus the quantification of these parameters as presented in the Results were obtained with a limited sample of axons. Statistical comparisons, when applicable, were under- taken using the nonparametric Mann–Whitney test.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Injections of either biocytin or PHA-L in the parietal cortex of hamsters resulted in labeled axons, but no labeled neurons, in the contralateral hemisphere. Retrogradely filled cells were occasionally detected in the ipsilateral thalamus, but only in cases for which some necrosis was visible at the injection site. Thus, our data are not confounded by images of retrogradely labeled neurons or their processes. All reconstructed axons illustrated here were considered to be completely filled by the tracer, as assessed by observation of fine morphological details such as filopodial sprouts, varicosities, growth cones (Fig. 1C
) and terminal swellings. General Aspects of Labeling in the Younger Animals
Axons radiated from the injection site, some coursing medially or laterally within the gray matter to form intracortical projections (Fig. 1A,B), while others reached the white matter directly under the injection site where they became segregated into two distinct strata: axons that coursed laterally away from the injection site were assembled more superficially in the white matter, whereas those directed toward the midline turned sharply, and were bundled in deeper portions of the white matter (Fig. 1B). In some cases, tracer injections resulted in excellent labeling of axons in the ipsilateral cortex, but gave rise to only a few, beaded callosal axons that crossed the midline and reached the other side. These axons could not be followed into the cortical gray matter. For the purposes of this paper, they were discarded, as ones in which the transport of biocytin had failed. However, it is conceivable that the injection sites in these cases were targeted on acallosal patches of the parietal cortex (Wise and Jones, 1976; Akers and Killackey, 1978; Ivy et al., 1979; Ivy and Killackey, 1981; Olavarria et al., 1984), as suggested by the results of Aggoun-Zouaoui and Innocenti (1994), who associated such a beaded morphology with that of dying axons that originate from acallosal regions of the cat visual cortex.
The overall distribution of labeled axons in P7–9 pups comprised a dense plexus in the deep layers of the contralateral cortex, with sparsely branched collaterals that extended more superficially, often as far as layer 1 (Fig. 1A and inset in 2B). This plexus was found predominantly in the region homotopic to the injection site, although a few collaterals reached the cortical plate in other regions, medially and also laterally to the homotopic site in the presumptive secondary somatosensory (SII) cortex. Axons extended as far as the rhinal sulcus and some collateral sprouts were present in the perirhinal region (Fig. 2A). By P10–12, preterminal and terminal ramifications were visible both in infragranular and supragranular layers (Fig. 2C, inset) of the homotopic cortex, along with a consistent, heterotopic projection to the presumptive area SII (see below). In addition to a dense arborization in the homotopic site, there were three heterotopic sites in which callosal innervation was detected after the second postnatal week: medial to the main termination site (in the sensorimotor cortex), lateral to the homotopic site (presumptive SII), and the striatum, a noncortical site (Fig. 3A and B, top). This general pattern persisted until after the fourth postnatal week, except that the projection to the rhinal cortex was eliminated by this time (Fig. 3B, top).
|
Morphogenesis of Callosal Arbors in the Cortex
For P7–9 animals, four axons were serially reconstructed with the Neurotrace software; another five had their terminal arborizations serially reconstructed with the aid of a drawing tube. Observations on numerous other fibers, and on the general projection patterns of labeled axons, served to support con- clusions drawn from the serially reconstructed axons. Three examples of P7–9 callosal axons that terminated in the gray matter homotopic to the injection site are illustrated in Figure 2B. Most of the axons at this age had a simple morphology and were poorly ramified, having only a few short branches tipped with growth cones in the infragranular layers of the developing cortical plate. Some simple axons reached superficial cortical laminae without further branching (Fig. 2B, inset). Growth cones were also present in the white matter, at the distal ends of the parent axons (Fig. 1C). Axons tipped by growth cones were consistently observed at the midline (indicating continued addition of callosal fibers even at this age) and in the white matter all along the mediolateral extent of the cortex.
Parent fibers proceeded laterally in the white matter past the homotopic cortex, often as far as the rhinal fissure, and emitted two or three short sprouts along their length (Fig. 4A). These sprouts were usually unbranched, and did not penetrate deeply into the gray matter. They were present not only under the homotopic cortex, but also under other regions of the cortex. For many axons, the distal growth cone was far from the main collateral branch point (arrowhead in Fig. 4B). For others, it was close to where the collaterals entered the homotopic cortical plate (arrowhead in Fig. 4C), suggesting that collateral arbor- ization may have been initiated by bifurcation of the axon tip while the parent axon was proceeding laterally in the white matter, past the overlying homotopic target. Yet other callosal axons could be traced in the white matter all the way from the midline to the homotopic cortical target, where they turned at right-angles into the gray matter (not shown). Whether these axons later sprouted a lateral branch, or had already lost one, could not be determined from the present material. Finally, we also observed some fibers which extended a collateral to the homotopic cortex, and whose distal axon continued on in the white matter to enter the striatum contralateral to the injection site (Figs 2A and 7C).
|
|
In hamsters aged P10–12, the terminals of callosal axons formed a bilaminar pattern, with plexuses in supragranular and infragranular layers (Fig. 2C
, inset); nevertheless, the morphology of individual arbors was still simple, as illustrated in Figure 2C. Reconstructions of individual axons frequently revealed two or more closely spaced collaterals, emerging off parent axons in the white matter and ramifying into the homotopic gray matter (arrowheads in Fig. 5). This feature was preserved through adulthood, supporting the hypothesis that invasion of the cortical gray matter occurred at least in part by interstitial budding of branches from a long trunk axon in the white matter (multiple, penetrating collaterals were not detected within the homotopic region at earlier time points). In pial view, the arbors of these collateral branches appeared focused from the beginning: the extent of their arbors did not exceed a radius of 250 µm (Figs 4C and 5). The few branches that did extend further out rarely bifurcated (double arrowheads in Figs 4C and 5).
|
By the end of the second postnatal week (P13–17), a striking increase was seen in the number of axonal branches that contributed to the infra- (mostly layer 5) and supragranular (including layer 1) plexuses (Table 2
). Sparse branches were consistently seen in other laminae as well (Figs 3A, 6 and 9A). The lateral extension of the parent axon, beyond the homotopic arborization site, was still visible for some labeled fibers (Fig. 6A,C), but it was of variable length, and in some cases beaded and thinner than the medial parent axon. Growth cones were no longer detected at the tips of the parent axons at this age, but were still present in the major arborization sites. At this age, however, the first terminal boutons were detected. They had a grapelike distribution and were located at the tips of the axon branches. En passant boutons were present in all branches with no preferential distribution. It should be noted that even axonal branches that extended far from the major termination sites (in the tangential dimension) and those branches located in layers that received little or no callosal innervation exhibited en passant boutons.
|
|
|
When more than one collateral emerged from the trunk axon to penetrate the gray matter, they usually entered the same region of homotopic cortex, and their arbors were at least partially overlapping. The arbor had a larger areal extent than before, but it was still restricted to a radius of 250–300 µm. Within this area, however, some arbors now exhibited a patchy distribution (as seen in pial view), defining regions rich in terminal branches intercalated by regions devoid of arborization (Fig. 6C
).
When injections were restricted to the primary somato- sensory cortex (SI), a heterotopic callosal projection was consistently seen in the presumptive SII cortex (Figs 3 and 7). In our material, arborization at this site was first detected during the second postnatal week (Fig. 3B, top), whereas at earlier time points only a few, simple collaterals were present in the region. Parent axons of the heterotopic projection also extended laterally beyond SII (thin arrow in Fig. 7A); some of these fibers exhibited more than one collateral which formed arbors that overlapped at the same cortical site (Fig. 7A,B). Serial recon- structions of the trunk axons in the white matter revealed that the heterotopic projection was exclusive, and did not arise from axons that also had homotopic projections. Thus, by the end of the second postnatal week, the lateral extension of the main axon of homotopically projecting fibers was beginning to regress. The separate heterotopic projection to SII persisted into adulthood.
By the fourth week of postnatal life, homotopically projecting callosal fibers could be classified into two types according to the radial distribution of their arbors: axons with a bilaminar terminal arbor (all axons except one in layers 1–3 and 5; Figs 8B and 9) and those with ramifications confined to supragranular layers (Fig. 9B, last axon to the right). None of the axons that were serially reconstructed had arbors restricted exclusively to the infragranular layers, suggesting that the infragranular plexus derived from the bilaminar axon type. Moreover, the infra- granular plexus was distinctly less dense than the supragranular (Figs 8 and 9; see also Fig. 3B). One axon with sparse arbor- ization in layers 4 and 1 was observed (Fig. 9B, third axon from left). By this age, the lateral extensions of parent axons within the white matter were completely eliminated (Figs 8B and 9; but see fourth axon from left in Fig. 9B) and, correspondingly, the inverted `Ts' typical of collateralizing axons, seen in younger animals, were rarely detected in the white matter below the homotopic cortex.
|
Pial views of reconstructed axons at this age revealed that the arbors were still contained within an area of 250–300 µm radius, although terminal ramifications were not always homogeneously distributed therein. These overall dimensions are compatible with cortical column sizes as reported for axons in the parietal cortex of adult rodents (Jensen and Killackey, 1987
; Purves et al., 1992). Although the breakup of terminal ramifications into tangential patches becomes more distinct at this age, generally only two patches are seen per axon. In order to determine whether each patch arises from arborizations in specific laminae (i.e. whether the infragranular arbors of an axon contribute to one patch, and the supragranular arbors to another), we utilized the `z slicing' function of the Neurotrace program (see Materials and Methods). It was clear from such analyses that single patches as seen in pial view may be comprised of arbors formed in both infra- and supragranular laminae. For example, the axon illus- trated in Figure 8B had the infragranular portion of its arbor in radial register with part of its supragranular arbor (they together contributed to one patch), whereas only the supragranular branches contributed to the second patch.
When the quantitative parameters were analyzed (Table 2) it was observed that during the second postnatal week, the total length of the axons increased ~2-fold, a trend which was sustained through the third and fourth weeks, when the cumulative length contributing to the entire axon was maximal. A significant decrease (P < 0.05) in total axon length was noted for fibers harvested from P25–29 brains. The number of branch points also increased significantly during the first two weeks, and appeared to decrease towards the fourth postnatal week, but that difference was not found to be significant. The average segment length, however, revealed a decrement between the second and fourth postnatal weeks indicative of an increase in complexity by these later axons. The measurements of arbor surface area showed a nonsignificant decrease from the earlier towards later ages (Figure 10).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our results show that callosal axons undergo a developmentally regulated remodeling process that involves both progressive and regressive events. In the tangential dimension, the lateral extension of the parent axon, along with its associated sprouts, overshoots the homotopic cortical target and is subsequently eliminated. Contemporaneous with this regression, collateral branches emitted by the parent fiber penetrate the appropriate (homotopic) region of gray matter, and begin to elaborate an arbor therein. A minor retrieval of sprouts found in inappro- priate laminar locations is also noted (see below) (Fig. 9
). Technical Considerations
The introduction of tracers such as DiI, PHA-L, biocytin, cholera toxin and biotinylated dextran amines has permitted the detailed visualization of anterogradely labeled axons (Gerfen and Sawchenko, 1984; Honig and Hume, 1989; King et al., 1989; Ling et al., 1997) in both adult and developing tissues. Most of these tracers are associated with both retrograde and orthograde labeling of axons. However, PHA-L and biocytin can provide exclusively anterograde labeling when used with specific protocols, thus ruling out the contribution of retrogradely filled neural processes (such as dendritic shafts and recurrent axonal branches, or of labeled radial glial fibers — Floeter and Jones, 1985; Lent et al., 1990; Kageyama and Robertson, 1993), which can obscure the identification of axons and their arbors. In our material, labeled cells were occasionally detected in the ipsi- lateral thalamus, but only in cases where signs of necrosis were seen at the injection site (these cases were not included in the present analysis). Use of the Neurotrace program to reconstruct axons over multiple sections, often back towards the midline, further ensured that the processes we had drawn were indeed axons of callosally projecting cortical neurons.
Biocytin was preferentially used in younger animals because of its shorter transport time, whereas PHA-L was used for the older animals. We did not undertake experiments that could explicitly refute the possibility that differential labeling is obtained with the two tracers. However, the fact that the alterations occur gradually over time militates against this possibility, as does the observation that the lateral extension of axons in the white matter becomes thinner prior to disappearing completely. Only cases in which the axons were judged to be fully filled (see above), either with PHA-L, or with biocytin, are included in the reconstruction and quantitative analyses. There is no indication in the literature that either biocytin or PHA-L results in differential fills of some processes and not others. In a recent report (Ling et al.,1997) it has been shown that tectal terminals to the lateral posterior thalamic nucleus which are labeled with biocytin, biotinylated dextran amines or PHA-L essentially have identical morphologies. Thus, while we cannot completely rule out the possibility that, for example, some of the regressive events noted in the older animals result merely from selective filling with biocytin versus PHA-L, the above arguments indicate that this is highly unlikely.
Pathfinding and Target Selection Strategies
During the initial stages of morphogenesis, callosal axons over- shoot cortical regions that are homotopic to the injection site, an observation made possible by use of the serial reconstruction technique. This finding is contrary to the expected specificity in target selection, as suggested by earlier experiments with HRP, WGA-HRP and DiI labeling (Wise and Jones, 1976; Innocenti, 1981; Ivy and Killackey, 1981; Floeter and Jones, 1985; Olavarria and Van Sluyters, 1985; Lent et al., 1990; Norris and Kalil, 1992). Dorsally and ventrally directed sprouts that emerge from the trunk axon at several points along its length perhaps reflect a strategy used by afferents for the sampling of prospective target regions. This kind of initial overshooting of the target has also been described for other axon systems that are in the elongation mode of growth: thus, retinal axons elongating in the optic tract (Bhide and Frost, 1991; Jhaveri et al., 1991; see also Frost, 1984), corticofugal fibers (Distel and Hollander, 1980; O'Leary and Terashima, 1988), thalamocortical afferents (Ghosh and Shatz, 1992; but see Agmon et al., 1993, 1995) and corticospinal afferents (Meissirel et al., 1993; Kuang and Kalil, 1994) all overshoot their prospective targets. In some systems, as the arborization stage is triggered, one or more collaterals emerge from the parent axons, a process that has been referred to as `interstitial budding' (O'Leary and Terashima, 1988). The present study cannot address this possibility for callosal projections, but unpublished data (C. Hedin-Pereira and S. Jhaveri, 1991) indicate that interstitial budding may be one of several strategies used by callosal axons to collateralize in their homotopic target zones. The long lateral branch of the parent axon persists in the white matter even after the second postnatal week, when complex arbors that invade the gray matter can be detected on many fibers (see below). It is important to point out that the use of methods which do not involve full reconstruction of callosal axons would miss this overshooting since it would be obscured by the heterotopic projection. Our results document that at late stages of development, the distal portion of the parent axon becomes thin and beaded, suggesting an ongoing regressive process in anticipation of its final elimination.
It is possible that the axons that overshoot the homotopic projection zone comprise the subpopulation of callosal fibers which are ephemeral during development. This possibility is refuted by our findings that even axons that support a complex terminal arbor in more mature animals still occasionally display such a lateral extension. Furthermore, the single-axon study of Aggoun-Zouaoui and Innocenti (1994) documents that axons from regions of the cortex that are fated to be acallosal in the adult cat appear not to develop arbors in the cortical plate.
It has been hypothesized that `waiting periods', i.e. the time lags between afferent axon arrival in the vicinity of the target and the time at which the afferent axon actually innervates the target cells, have a role in the formation of cortical connectivity. For example, interactions between afferent axons and subplate cells are reportedly critical for the formation of thalamocortical and other connections (for review, see Allendoerfer and Shatz, 1994). O'Leary and Terashima (1988) proposed a reinter- pretation of waiting periods. Based on the development of corticopontine projections, they suggested that the axon elongation past the target and the subsequent interstitial branching along its trunk would in fact correspond to what was termed the waiting period. In the rodent cortex, it has been proposed that there are no waiting periods, or that if they exist, they are very short (Hogan and Berman, 1990; Catalano and Killackey, 1991; Norris and Kalil, 1992), perhaps due to the temporal compression of developmental events. Nevertheless, the short time lag between the arrival of the callosal axon under the homotopic targets and their eventual innervation may correspond to the axonal elongation phase past the target region and the collateralization process, similar to what was described for the corticopontine system. Similar conclusions have been reached concerning the development of cat callosal axons (Aggoun-Zouaoui and Innocenti, 1994).
Collateral Formation and Arborization within the Cortical Gray Matter
Earlier studies in which axon populations were labeled (Wise and Jones, 1976; Innocenti, 1981; Ivy and Killackey, 1981; Floeter and Jones, 1985; Olavarria and Van Sluyters, 1985; Lent et al., 1990; Norris and Kalil, 1992) led to the conclusion that callosal invasion of the gray matter is topographically specific from the outset, or that it occurs only in areas which are targets of callosal axons in mature animals. These investigators reported that very little tangential reorganization is involved in attaining the adult pattern of callosal projections. Our results, using more sensitive tracers and visualization of single axons as they course over long distances, reveal that callosally projecting axons have short collaterals under regions of cortex that will not be innervated by them in the mature animal. However, these collaterals neither advance significantly into the cortical plate nor arborize therein, but are restricted to the presumptive subplate and infragranular layers (see Aggoun-Zouaoui and Innocenti, 1994), disappearing later on. Collaterals that inner- vate the homotopic target region, on the other hand, advance into the gray matter, in some cases as far as layer 1 before elaborating arbors (see also Norris and Kalil, 1992; Elberger, 1993, 1994). Thus, the process of arborization displays consider- ably more regional specificity than seen during axon elongation. Little is known about the signals that trigger the arborization growth mode in intracortical systems, or about mechanisms involved in triggering innervation of specific cortical fields, but diffusible signals (cf. Bolz et al., 1990; Heffner et al., 1990; Sato et al., 1994; Lotto and Price, 1995) or membrane-bound molecules (cf. Yamamoto et al., 1989; Molnár and Blakemore, 1991; Bolz et al., 1992; Götz et al., 1992; Yamamoto et al., 1992; Henke-Fahle et al., 1996) have been implicated in other corticofugal or corticopetal systems. Recently, an Eph receptor ligand and its receptors were proposed to be related to lamina- specific arborization of intracortical neurons (Castellani et al., 1998).
We cannot rule out the possibility that some growing fibers may innervate the gray matter by making a right-angled bend from their tangential trajectories to grow radially in the cortical plate, leaving no distal process in the white matter. Axons were also seen with collaterals very close to the leading growth cone, suggesting the possibility of bifurcation in the white matter followed by simultaneous growth of both branches in the radial and in the tangential direction. These diverse behaviors could reflect differential cues in the environment traversed by axons during the 7–10 days over which callosal axons continue to cross the midline (cf. Lent et al., 1990; Norris and Kalil, 1990).
Typically, more than one collateral sprout emerges from the main axon under the homotopic cortex, and together they contribute to the terminal arbor formed by a single fiber within the cortical layers, with arbors of single collaterals overlapping in the same general target area. Such a pattern of arborization is different from that reported in the cat, where nonoverlapping (divergent) arbors have been related to the synchronous activation of distinct sets of cortical neurons (Houzel et al., 1994; Innocenti, 1994). In the rodent, on the other hand, multiple afferentation of the same targets by single axons could be the basis for the summation of inputs, capable of increasing the safety factor for the excitation of postsynaptic neurons. However, there are a few cases in which the divergent pattern observed in the cat is present. The most divergent of these is the doubly heterotopic axon which arborizes both in the striatum and in the medial cortex.
For many afferent systems, arbor formation within the target territory also involves the sprouting of extraneous branches which are later eliminated. Thus, for instance, retinal axons within the superficial layers of the superior colliculus have short sprouts along their length during early development; a single collateral then emerges and is elaborated in the topographically appropriate position (Jhaveri et al., 1991; see also Naegele et al., 1988 for axons in cortex). Similarly, retinal axons in the dorsal lateral geniculate nucleus of the cat have short, transient sprouts in inappropriate laminae within the nucleus; these sprouts get eliminated as the terminal arbor develops (Sretavan and Shatz, 1984, 1987). Concerning the callosal system, most axons, once within the cortical plate, appear to form layer-specific arbors from the outset; however, some may exhibit ephemeral sprouts in inappropriate laminae. This is consistent with the reports of Katz (1991) who document laminar specificity for arbors of developing intracortical axons in the kitten visual cortex. It has been documented for the cat visual callosal axons that presumptive synaptic boutons on terminal ramifications of callosal axons that penetrate the correct regions of the cortex appear primarily in the appropriate laminae (Aggoun-Zouaiou et al., 1996). In our material, the functional significance of en passant boutons or terminal boutons could not be evaluated. En passant swellings were found along all branches at all ages examined. Grape-like boutons appeared only after the second postnatal week, when maximal branch numbers were found. However, in agreement with the findings of the above authors in the cat, these types of boutons were present simultaneously with growth cone-like structures. From the second week onwards, growth cones became rare and terminal bouton-like structures increased.
Arbor analysis in the tangential dimension reveals that in older animals, the densest areas of arborization rarely extend beyond a radius of 300 µm. However, in younger animals a few sprouts do reach farther out from the focus of arbor formation, but never show any evidence of ramification, similar to what was found by Callaway and Katz (1990) for the horizontal projections of layer 2/3 cells in the visual cortex . The complexity of the terminal arborization continues to increase through the end of the first postnatal month; thus, although the gross topography and patterning of callosal connections is reportedly adult-like by the end of the second week of postnatal life (Lent et al., 1990), the process of axon elaboration is still ongoing.
Analysis of total axon length shows a pronounced increase from the first to the second postnatal week, reaching a maximum during the second/third postnatal week. During this period of increasing axon length, there is a sharp rise in arborization at the target region and concomitant elimination of inappropriate collaterals along the trunk axon. The decrease observed there- after might in part reflect the elimination of the lateral segment of the parent axon that we have observed qualitatively. Although the number of branch points on callosal arbors appears to decrease from the second to the fourth postnatal week, the average segment number reveals an increase in complexity between these ages.
Patterns of Heterotopic Arborization
Two types of heterotopic axon arbors, single and double, were observed following tracer injections in the parietal cortex. Single heterotopic fibers have one arbor that terminates lateral to the homotopic cortical region, forming a projection to the presumptive cortical area SII. The double heterotopic axon, on the other hand, forms two heterotopic arbors, one in the medial cortex and the other in the striatum. The point of divergence between the two main branches of the arbor is always situated below the (homotopic) parietal cortex, indicating that nerve impulses would be conducted to that region of the parietal cortex before being distributed to the two arbors. It is conceivable that this arrangement may lead to the coactivation of both arbor branches, enabling the synchronization of activity in a noncortical (striatal) and a cortical region. Only callosal projections from the sensorimotor cortex to the contralateral striatum have been previously described (McGeorge and Faull, 1987). The existence of axons that project exclusively to SII confirms the heterotopic projection pattern described in studies with large HRP injections into the primary parietal cortex (Akers and Killackey, 1978). The reconstructions in this study show that axons with heterotopic projections do not have a homotopic arbor, and conversely, axons with homotopic arbors do not display a heterotopic arborization. These data further confirm that those fibers which span the entire mediolateral extent of cortex during development must eliminate collaterals either to the homotopic or to the heterotopic regions.
Arbors of Mature Callosal Axons
In contrast to earlier reports on the hamster visual cortex (Fish et al., 1991), callosal projections from parietal cortex exhibit arbors that are comparably much more complex. This may be due to the partial reconstruction of axons undertaken in that study.
The size of the callosal arbors in the parietal cortex of hamsters is similar to that of thalamocortical arbors terminating in the somatosensory cortex of rats (Jensen and Killackey, 1987), and in the occipital cortex of hamsters (Naegele et al., 1988). The point-to-point tangential mapping of thalamic projections in the cortex would suggest that the terminal ramifications of thalamic axons were more restricted in tangential spread than those of callosal axons. Nevertheless, it should be pointed out that arbor size is only one of the factors that contribute to topography (Salin et al., 1989; Kennedy et al., 1994).
In conclusion, we have shown that the morphogenesis of callosal axons in the hamster is protracted and extends through at least the first month of postnatal life. There is an extensive overshooting of the target, suggesting an initial lack of specificity in the growth of callosal axons. Primitive sprouts along the length of the fibers indicate the beginnings of target selection, a phase followed by the elaboration of branches in selective targets and the elimination of sprouts and axon extensions in inappropriate positions. Evidence was presented for tangential remodeling, via regressive and progressive mechanisms. The extended time over which this occurs suggests a prolonged plastic period for this axon system.
|
|
Notes |
|---|
We are grateful to Professor Gerald Schneider for letting us use his Neurotrace setup, for long hours spent in helping us with the axon reconstructions, and for many discussions. We also thank Elizabeth Moraes and Julie Wu for excellent technical assistance. This work was funded by NIH grant NS26278 (S.J.) and Pronex grant 052/97 (R.L.); C.H.-P. was supported by a scholarship from CNPq (National Council for Scientific and Technological Development, Brazil).
Address correspondence to Dr Cecilia Hedin-Pereira, Departamento de Anatomia, ICB-UFRJ, Centro de Ciências da Saúde, Bl. F, Rio de Janeiro 21941–590, Brazil. Email: hedin{at}ibccf.biof.ufrj.br.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Agmon A, Yang LT, O'Dowd DK, Jones EG (1993) Organized growth of thalamocortical axons from the deep tier of terminations into layer IV of developing mouse barrel cortex. J Neurosci 13:5365–5382.[Abstract]
Agmon A, Yang LT, Jones EG, Dowd DK (1995) Topological precision in the thalamic projection to neonatal barrel cortex. J Neurosci 15:549–561.[Abstract]
Aggoun-Zouaoui D, Innocenti GM (1994) Juvenile visual callosal axons in kittens display origin and fate-related morphology and distribution of arbors. Eur J Neurosci 6:1846–1863.[ISI][Medline]
Aggoun-Zouaoui D, Kiper DC, Innocenti GM (1996) Growth of callosal terminal arbors in primary visual areas of the cat. Eur J Neurosci 8:1132–1148.[ISI][Medline]
Akers R, Killackey HP (1978) Organization of corticocortical connections in the parietal cortex of the rat. J Comp Neurol 181:513–538.[ISI][Medline]
Antonini A, Stryker MP (1993) Development of individual geniculo- cortical arbors in cat striate cortex and effects of binocular impulse blockade. J Neurosci 13:3549–3573.[Abstract]
Barone P, Dehay C, Berland M, Kennedy H (1996) Role of directed growth and target selection in the formation of cortical pathways: prenatal development of the projection of area V2 to area V4 in the monkey. J Comp Neurol 374:1–20.[ISI][Medline]
Bhide PG, Frost DO (1991) Stages of growth of hamster retinofugal axons: Implications for developing axonal pathways with multiple targets. J Neurosci 11:485–504.[Abstract]
Bolz J, Gilbert CD (1990) The role of horizontal connections in generating long receptive fields in the cat visual cortex. Eur J Neurosci 1:263–268.
Bolz J, Novak N, Götz M, Bonhoeffer T (1990) Formation of target-specific neuronal projections in organotypic slice cultures from rat visual cortex. Nature 346:359–362.[Medline]
Bolz J, Novak N, Staiger V (1992) Formation of specific afferent connections in organotypic slice cultures from rat visual cortex cocultured with lateral geniculate nucleus. J Neurosci 12:3054–3070.[Abstract]
Callaway EM, Katz LC (1990) Emergence and refinement of clustered horizontal connections in cat striate cotex. J Neurosci 10:1134–1153.[Abstract]
Castellani V, Yue Y, Gao PP, Zhou R, Bolz, J (1998) Dual action of a ligand for Eph receptor tyrosine kinases on specific populations of axons during the development of cortical circuits. J Neurosci 18:4663–4672.
Clarke S, Innocenti GM (1986) Organization of immature intra- hemispheric connections. J Comp Neurol 251:1–22.[ISI][Medline]
Dehay C, Kennedy H, Bullier J, Berland M (1988) Absence of inter- hemispheric connections of area 17 during development in the monkey. Nature 331:348–350.[Medline]
Distel H, Hollander H. (1980) Autoradiographic tracing of developing subcortical projections of occipital regions in fetal rabbits. J Comp Neurol 192:505–518.[ISI][Medline]
Elberger AJ (1993) Distribution of transitory corpus callosum axons projecting to developing cat visual cortex revealed by DiI. J Comp Neurol 333:326–342.[ISI][Medline]
Elberger AJ (1994) Transitory corpus callosum axons projecting throughout developing rat visual cortex revealed by DiI. Cereb Cortex 4:270–299.
Fish SE, Rhoades RW, Bennett-Clarke CA, Figley B, Mooney RD (1991) Organization, development and enucleation-induced alterations in the visual callosal projection of the hamster: single axon tracing with Phaseolus vulgaris leucoagglutinin and DiI. Eur J Neurosci 3:1255–1270.