During telencephalic development, a boundary develops that restricts cell movement between the dorsal cortical and basal striatal proliferative zones. In this study, the appearance of this boundary and the mechanism by which cell movement is restricted were examined through a number of approaches. The general pattern of neuronal dispersion was examined both with an early neuronal marker and through the focal application of DiI to telencephalic explants. Both methods revealed that, although tangential neuronal dispersion is present throughout much of the telencephalon, it is restricted within the boundary region separating dorsal and ventral telencephalic proliferative zones. To examine the cellular mechanism underlying this boundary restriction, dissociated cells from the striatum were placed within both areas of the boundary, where dispersion is limited, and areas within the cortex, where significant cellular dispersion occurs. Cells placed within the boundary region remain round and extend only thin processes, whereas progenitors placed onto the cortical ventricular zone away from this boundary are able to migrate extensively. This suggests that the boundary inhibits directly the migration of cells. To examine whether the signal inhibiting dispersion within the boundary region acts as a long- or short-range cue, we apposed explants of boundary and nonboundary regions in vitro. Within these explants we found that migration was neither inhibited in nonboundary regions nor induced in boundary regions. This suggests that the boundary between dorsal and ventral telencephalon isolates these respective environments through either a contact-dependent or a short-range diffusible mechanism.
Within the telencephalon, discrete regions of proliferation and differentiation become evident during development (Bulfone et al., 1993; Puelles and Rubenstein, 1993). Their territories are characterized by their patterns of gene expression (Price et al., 1992; Simeone et al., 1992; Figdor and Stern, 1993;Rubenstein et al., 1994; Bulfone et al., 1995; Shimamura et al., 1995). The most prominent of these proliferative regions are the pallial cortical ventricular zone (CVZ) and the basally positioned lateral ganglionic eminence (LGE, the striatal proliferative zone). They are separated by a boundary that falls along the longitudinal axis and will be referred to henceforth as the L–C boundary (LGE–CVZ). The L–C boundary is demarcated by the transition from the cortical proliferative zone, which resembles an epithelial sheet, to the striatal proliferative zone, which has a pillow-like morphology. In addition, this boundary delimits sharply the expression of a series of regional markers expressed in the cortical proliferative zone (e.g.,Emx1 and Pax6) or the LGE (e.g.,Dlx2 and Gbx2) (Simeone et al., 1992; Bulfone et al., 1993).
These transient developmental structures give rise to markedly different territories in the mature forebrain. Both the gross cytoarchitecture and the single-cell morphology of the cortex and striatum are quite distinct. Whereas the cortex is organized into laminae (Angevine and Sidman, 1961; Boulder, Committee, 1970), in which specific cell types occupy specific layers, the striatum has a nuclear structure and only one predominant neuronal morphology (Smart and Sturrock, 1979). To understand the mechanisms by which these areas maintain such divergent patterns of organization, we focused on examining the boundary region dividing these proliferative territories.
Here we examine restrictions to tangential dispersion within the telencephalic VZ (Walsh and Cepko, 1992; Fishell et al., 1993; Liang and Walsh, 1995; Reid et al., 1995), with particular emphasis on the behavior of cells at the L–C boundary. To study this, we examine the distribution of cells within telencephalic VZ expressing a neuron-specific form of tubulin, identified with the antibody TUJ1 (Lee et al., 1990). Second, to examine the general patterns of cellular dispersion, we apply DiI focally to specific regions in telencephalic explants and examine the patterns of cell dispersion after short-term survival. Both these studies reveal that although considerable dispersion of neurons occurs within the CVZ away from the L–C boundary (Fishell et al., 1993; O’Rourke et al., 1997), little dispersion occurs within the border region. To test the cellular mechanism underlying this restriction, precursor cells from the LGE are transplanted heterotopically onto the CVZ and L–C boundary region of telencephalic explants in vitro. This experiment demonstrates that cells that are competent to migrate will not do so if placed within the L–C boundary. By apposing both boundary and nonboundary regions in vitro, we demonstrate that this restriction occurs through a contact-dependent or short-range diffusible mechanism. Together this work suggests that the integrity of the cortical and striatal proliferative zones is maintained by inhibitory cues within the L–C boundary.
MATERIALS AND METHODS
Preparation of explants. Explants were obtained from embryonic day 15–17 (E15–E17) rat embryos (Charles River, Wilmington, MA; Taconic Laboratories, Germantown, NY). The cerebral hemispheres were opened dorsally via a parasagittal cut. A telencephalic explant containing the medial ganglionic eminence (MGE), LGE, and CVZ was dissected from the rest of the telencephalon and cultured on a small tissue culture insert (Nunc, Naperville, IL, catalog #162243) in serum-free medium (DMEM/F12, N2, B27 supplements, glutamine 2 mm, mito C supplement; Collaborative Research).
Dissociation of cells and preparation of cell aggregates.Cells were obtained from either E16 or E17 rat embryos (Charles River; Sprague Dawley, Indianapolis, IN). The cerebral hemispheres were opened via a parasagittal cut, and an area containing the proliferative zone of the LGE was pinched off. The fragments were incubated for 30 min at 37°C in 1.5 ml of trypsin 0.08%–EDTA 0.02% containing 100 μl of DNase (1 mg/ml). Heat-inactivated FCS (0.5 ml) was added, and fragments were triturated using fire-polished Pasteur pipettes. Dissociated cells were washed, PKH-26-labeled (Sigma, St. Louis, MO; Xynaxis) with a concentration of PKH-26 of 1 μl/ml. The labeling reaction was stopped by washing the cells twice in DMEM containing 10% FBS. The cells were plated at a concentration of 4 × 105 cells per well in serum-free medium.
When aggregates were needed, cells were resuspended in 200 μl of serum-free medium and allowed to reaggregate overnight in uncoated Terassaki wells (Nunc). Aggregates of cells were pooled, washed, and triturated three times in 1 ml of serum-free medium containing DNase. Aggregates or dissociated cells were sprinkled onto explants and allowed to settle for 1 hr, then the extra medium was removed and the explants were incubated at 37°C. Dissociated cell experiments were repeated for 52 explant preparations. For reaggregate experiments, 28 explants were examined.
As a positive control, reaggregates or dissociated PKH-26-labeled-LGE cells were placed either onto the pial (rather than the ventricular) surface of explants or onto poly-d-lysine-coated dishes. In both these cases, almost 100% of cells underwent active migration, as suggested by their elongated migratory profile.
DiI injections into explants. Explants were prepared on tissue culture inserts (Nunc) as described above. For injections, solutions of fixable DiI (Molecular Probes, Eugene, OR) were made up in EtOH (for iontophoresis injections) or in a 1:9 mixture of 11′-dioctadecyl 3,3,3′,3′-tetramethyl indocarbocyanine perchlorate (DiI; Molecular Probes) (5 mg/ml in EtOH) and 0.3 m sucrose (for pressure injections). Iontophoretic injections were made using pulled glass pipettes (20 MΩ resistance in 3 m KCl) using 10 nA current for 10 sec. Pressure injections were performed with an IM6 microinjector (Narishige, Tokyo, Japan) and micromanipulator (Narishige).
Acridine orange staining. Explants were dissected and submerged immediately in a 5 μg/ml solution of acridine orange dissolved in serum-free medium at 37°C for 30 min. After three washes in PBS, the explant was visualized for staining using fluorescent microscopy.
TUJ1 and RC-2 immunohistochemistry. Explants were prepared and cultured onto tissue culture inserts as explained above. Comparison of TUJ1 staining in tissue taken from in vivo preparation and comparably timed explants cultured in vitro demonstrated that the patterns of TUJ1 staining were indistinguishable. They were fixed for 10 min in 4% paraformaldehyde, quenched in methanol 0.3% H2O2, washed in PBS, and blocked for 1 hr in PBS containing 10% NGS and 0.5% Triton X-100. For RC-2 staining, explants were cut into 100 μm sections using a vibratome. Explants were incubated overnight in TUJ1 or RC-2 antibody diluted 1:50, containing PBS, 1% NGS, and 0.5% Triton X-100. Explants were washed three times in PBS and incubated for 1–2 hr with a peroxidase-coupled goat anti-mouse (anti-IgG in the case of TUJ1 and anti-IgM in the case of RC2) antibody (1:100). Explants were washed in PBS, and a DAB reaction was performed.
When a fluorescent secondary antibody was used, fixed explants were incubated for 1 hr in a 10% NGS blocking solution and then transferred to a solution containing 10% NGS and 1:50 TUJ1 antibody. To maintain the PKH labeling in these preparations, detergent was omitted from this procedure. No permeabilization was needed to obtain good TUJ1 staining. FITC goat anti-mouse secondary antibody was used to visualize TUJ1-positive cells.
Explant apposition experiments. Explants were prepared from both boundary and nonboundary regions taken from E17 Sprague Dawley rats. Boundary explants were defined as CVZ territories within 75 μm of the border where the cortical epithelium abruptly thickens and becomes the LGE. Nonboundary explants were defined as CVZ territories >150 μm distant from this same landmark. Taking the explants in this manner excludes the transitional zone where graded numbers of TUJ1-positive cells are seen. In each case, one of the two apposed explants was vitally stained with PKH 26 dye in a manner identical to that described above (except of course that the explants were not dissociated). This labeling procedure resulted in the outer layer of cells in the explants being brightly labeled. Explants were placed in Nunc tissue culture insert and apposed against one another, and the excess media removed. Explants were cultured from 12 to 24 hr, fixed in 4% paraformaldehyde, and stained for TUJ1 immunoreactivity (no Triton was used in these staining procedures). Apposed explants adhered to one another, allowing for easy determination of the apposition point between them. A minimum of 20 explant appositions were examined for each experimental condition discussed.
Microscopy. Both whole-mounted explants and stained sections were visualized using a cooled CCD camera (Princeton Instruments) mounted on an upright microscope (Axioscope, Zeiss). Images were acquired using Metamorph software (Universal Imaging, West Chester, PA). Double-labeling was achieved by digital superposition of pseudocolored black and white images. (See Fig 4D,E for transmitted light images superimposed onto DiI labeling.) The image of the original extent of dye application was taken from images acquired immediately after labeling and transferred digitally to the images of the point of sacrifice.
Marked changes occur in the pattern of tangentially oriented neurons during telencephalic VZ development
We examined the appearance of tangentially oriented neurons within telencephalic VZ regions during the early to midneurogenic period. Migratory cells have been shown to be labeled by TUJ1 (Menezes et al., 1995; O’Rourke et al., 1995), an antibody that recognizes a neuron-specific form of tubulin (Lee et al., 1990). Previous work has demonstrated in individual sections the presence of young migratory neurons within telencephalic VZ regions (Menezes and Luskin, 1994); however, the global distribution of these cells within the entire CVZ has not been determined. To do so, we used flat-mounted explants of telencephalon stained with TUJ1 antibodies. By viewing the ventricular surface, the entire population of tangentially oriented neurons within this region can be visualized simultaneously.
Examination of flat-mounted preparations stained with the TUJ1 antibody at a variety of developmental ages allowed us to infer both the initial appearance of these cells and their distribution in the CVZ. Although small numbers of tangentially oriented neurons within the CVZ were detected as early as E15 in rats, significant numbers of TUJ1 cells were observed only 1 d later (or approximately the midpoint of cortical neurogenesis).
At E16, TUJ1 cells were observed to be distributed widely within the CVZ (Fig.1 A,C). The distribution of these neurons appeared not to be preferentially oriented except within the region of the L–C boundary. Here migrating neurons were oriented perpendicular to the boundary (Fig.1 B,D; see Fig.4 D). Notably, ∼80% of the perpendicularly aligned TUJ1-positive neurons have their leading processes oriented away from the boundary region. In addition, adjacent to the boundary region relatively fewer TUJ1 cells were observed (Fig 1B). It should be noted that we cannot rule out the possibility that some of these perpendicularly oriented neurons are crossing the L–C border from the striatum at this time.
By E17, tangentially orient TUJ1 cells were still present throughout most of the CVZ, except for in the vicinity of the L–C boundary. The distribution of these cells at this time was graded, such that they were most abundant in areas distant from the L–C boundary and dwindled in number in areas adjacent to it (Fig.2 D,J,L). Within 75 μm of the L–C boundary, neurons expressing TUJ1 were almost completely absent (Fig.2 G,H,J–L). Furthermore, whereas neurons situated distant from the L–C boundary have no preferential orientation, those situated within 75 μm of the boundary had their cell body and leading processes preferentially aligned parallel with it (Fig. 2 K). Given that 24 hr previously, neurons within this region were preferentially oriented perpendicular rather than parallel to the L–C boundary (Fig 1B,D), it appears that this period represents a time when the topographic cues within the L–C boundary are changing rapidly.
The presence of TUJ1-positive neurons was not limited to the CVZ. Regions of basal telencephalon, including the LGE and MGE, showed the same patterns of randomly oriented TUJ1 cells. Two differences in the patterns of tangential-oriented neurons within the dorsal versus ventral telencephalon were notable. First, these neuronal populations within the basal telencephalon preceded the appearance of similar cells within the dorsal telencephalon by ∼36 hr (i.e., at ∼E14.5). Second, the boundary region between morphologically defined proliferative zones within ventral telencephalon (i.e., the LGE and the MGE) did not appear to restrict the distribution of tangential-oriented neurons. For instance, TUJ1 cells were observed to span between the LGE and the MGE (Fig. 1 I). Hence, all boundary regions within the telencephalon do not restrict cellular dispersion, making the L–C boundary perhaps unique in this aspect.
Examination of focal DiI labeling of telencephalic VZ
To assess the dynamics of tangential dispersion within the CVZ, we focally labeled E17 CVZ with DiI. To do this, we applied DiI to explants in 10 μm dots to a variety of areas within the CVZ both adjacent and distant from the L–C boundary. The results indicated that the TUJ1-positive cells provide an accurate picture of the general patterns of tangential dispersion in the telencephalon.
DiI spots applied to the CVZ in regions distant from the L–C boundary spread out from their initial distribution of ∼10 μm to encompass a 100 μm area of the CVZ within a 12 hr period (Fig.3 A–C). In addition, individual cells were seen at distances up to 500 μm from the application site. Notably, dispersing cells were often not oriented orthogonal to the spot of dye application (Fig.3 A–C), consistent with our previous work (Fishell et al., 1993), which suggested that dispersing cells often change direction during their movement. When explants were sectioned coronally, such that the migration of cells out into the intermediate zone could be visualized, the expected pattern of radial migration was observed (O’Rourke et al., 1992, 1995).
In explants, the L–C boundary is morphologically apparent using transmitted light to visualized the thinner CVZ versus the thicker LGE (i.e., the CVZ transmits more light and hence appears brighter than the LGE). Focal applications of DiI in areas near the L–C boundary resulted in proportionally fewer labeled dispersed cells (Fig.3 D). Labeling near the L–C border resulted in an elongated patch of labeled cells along the boundary but not across it (Fig.3 D). In cases in which DiI was applied to the TUJ1-negative boundary zone, little dispersion of cells was seen and the focally applied DiI remained localized (Fig. 3 E).
The L–C boundary inhibits tangential dispersion
We have demonstrated previously that LGE cells are able to integrate back into explanted preparations within 4–12 hr of being put in contact with the VZ surface in vitro. Although many of the cells applied in this way migrate radially into the explant (Fishell, 1995), a substantial proportion also disperse tangentially through the VZ. As such, this procedure provides an effective assay for investigating the topographic cues that guide dispersion in the telencephalic VZ. Although dissociated cortical cells on explants behave similarly to LGE cells (C. Neyt, unpublished observations), we used LGE cells, because, unlike cortex, it is possible to isolate ventricular populations away from the underlying intermediate zone (Fishell, 1995).
To examine whether the L–C boundary is able to influence the pattern of dispersion of heterotopically positioned cells, we placed PKH-26-labeled LGE cells randomly onto the ventricular surface of cortical explants. After 12 hr, preparations were analyzed for the morphology and migration of the LGE cells. To show that cells transplanted onto the surface of telencephalic explants were behaving like those endogenous to the host region, we compared the distribution and morphology of TUJ1 cells within host tissue with that of PKH-26-labeled LGE cells transplanted onto the CVZ. Invariably, transplanted and host cells were seen within the same plane of focus, and the morphology and distribution of these two populations looked identical (Fig.4 C,D). This result established that transplanted cells are able to respond to topographic cues that normally guide tangential dispersion in explants.
In both E16 and E17 explants, LGE cells that integrated into cortical areas distant from the L–C boundary extended long leading processes (ranging from 20 to 50 μm) (Fig. 4 A–C). That these cells were migrating was demonstrated by placing reaggregates of LGE neurons onto this region (n = 20). LGE cells from these reaggregates migrated out to over an area of the CVZ of up to 200 μm surrounding the reaggregate (Fig.4 A). Given that preparations were left for 12 hr before fixation after reaggregates were placed on their surface, it implies that the fastest dispersing neurons migrate at a maximal rate of ∼16.5 μm/hr.
Consistent with both the DiI labeling experiments and the patterns seen in TUJ1-stained cells, in E17 explants >60% of cells placed in the CVZ extended leading processes and have the appearance of actively migrating cells. Interestingly, whereas tangentially oriented neurons are still quite prominent in E17 explants, cells with a migratory profile were significantly fewer in number compared with 1 d earlier (t = 1.577; p < 0.05).
Reflecting changes in the border’s inhibitory function, how LGE cells behaved when placed adjacent to the L–C boundary varied with age. At E16, cells placed in this region (like TUJ1 cells endogenous to this region) aligned perpendicular to the L–C boundary (Fig.4 D). The orientation of both transplanted cells and endogenous TUJ1 cells was consistent with these cells undergoing active migration out of the boundary region. Notably, as in the explants stained for TUJ1 immunoreactivity, 80% of the neurons that were perpendicularly aligned in the boundary region have their leading processes oriented away from the boundary region. When these explants were left in culture for periods of 24 hr, their appearance was similar to that seen for the E17 experiments. This result supports the notion that cues in the E16 boundary region actively induce cells to migrate out of this area.
By E17, when TUJ1 cells are only sparsely present within the L–C boundary region, dissociated cells that settled within the L–C boundary region are unable to migrate or assume a migratory profile (Fig. 4 G,H). These cells remain rounded and extend thin axonal-like processes (Fig. 4 H). To test whether these cells were undergoing apoptosis, we stained freshly isolated explants with acridine orange (Fig. 4 I). Although previous authors report that extensive cell death occurs within the VZ (Blaschke et al., 1996), we saw no evidence from our analysis to suggest that this is more prevalent within the L–C boundary.
In cases in which cells or reaggregates were placed adjacent to the boundary region, we observed that cellular dispersion from reaggregates was reduced considerably (Fig. 4 E). When dispersion did occur, the migration of these cells was biased toward moving parallel with the boundary (Fig. 4 F).
Inhibition to tangential dispersion within the L–C boundary mediated by a short-range diffusible signal or a contact-dependent mechanism
To assess whether inhibition to tangential dispersion within the L–C boundary is mediated by a short- or long-range signal, we apposed cortical explants in vitro from both boundary and nonboundary regions. To ensure isolation of boundary regions where little tangential migration is occurring, we confined boundary explants to areas within 75 μm of the LGE, an area where TUJ1 cells are only sparsely present. Similarly, to ensure that nonboundary regions were areas of active migration, we used explants >150 μm from the LGE. In these experiments, we vitally labeled one of the explants with the dye PKH26 to assess whether cells transited from one explant to the other. These explants were maintained in vitro from 12 to 18 hr and then fixed and stained for the presence of TUJ1-positive neurons.
Formally, the lack of tangentially oriented neurons in the boundary region could result from an inhibitory signal in the boundary region or the presence of a signal (which is absent in boundary regions) in the nonboundary region that induces tangential dispersion. As such, it is of interest to note whether nonboundary tissue induces migration in boundary tissue, as well as whether boundary tissue inhibits migration in nonboundary areas.
We found no evidence that a diffusible inhibitory signal exists in boundary regions or that a diffusible migratory stimulatory signal exists in nonboundary regions (Fig. 5). Regardless of the adjacent presence of the complementary territory, numerous TUJ1-positive cells were present in nonboundary regions and were almost completely absent in boundary regions. A sharp line separating the boundary region (where tangentially oriented neurons were scarce) from the nonboundary region (where neurons were abundant) was always observed (Fig. 5 A,B). When PKH-26-labeled cells were placed on explants apposed in this manner, cells that attached to a boundary region remained round and sent out nontapering processes, but did not assume a migratory profile (Fig. 5 E). In contrast, cells placed within nonboundary regions behaved identically to those endogenous to the explant (Fig.5 F).
Development of the radial glial palisade at the L–C boundary
Given the absence of TUJ1-positive cells in the L–C boundary and our previous observations that this boundary acts to restrict tangential dispersion (Fishell et al., 1993), we were interested in assessing whether this region contains a specialized cellular organization. During the period that tangential dispersion is initiated, it has been demonstrated that bundles of radial glia form a palisade that acts to shepherd newborn cortical cells through the intermediate zone and cortical plate to the ventrolateral region of cortex (Alvarez-Buylla et al., 1988; Bayer et al., 1991; Misson et al., 1991). These radial glia are in the right position to potentially act as a specialized border structure. Previous studies suggest that the glial palisade separating the LGE from the CVZ appears in the period in which we observed that tangential dispersion is initiated (Smart and Sturrock, 1979; Misson et al., 1988; Edwards et al., 1990; De Carlos et al., 1996). Using the radial glia-specific antibody RC2, we confirmed that indeed the development of this palisade correlates with the time at which significant tangential dispersion is initiated. At E15, when the first significant number of TUJ1-positive cells are visible within the telencephalon and when minimal dispersion is occurring (Fishell et al., 1993), the radial glia within the CVZ and LGE are distributed evenly (Fig.6 B). Between E15 and E17, the radial glia in the region of the L–C boundary coalesce to form a palisade extending from the L–C boundary to the ventral lateral cortex (Fig. 6 C), with the end feet of these glia being positioned within the L–C boundary.
In this study, we examined the role regional boundaries play in restricting tangential dispersion within telencephalic VZ regions. Previous work suggested that the L–C boundary restricted the movements of cells between the proliferative zones of the dorsal and basal telencephalon (Fishell et al., 1993). Furthermore, heterotopic transplantation of precursor cells across the L–C boundary change their regional phenotype in accordance with their novel position, suggesting that such restriction is necessary to the proper regional specification of the telencephalon (Brustle et al., 1995; Campbell et al., 1995; Fishell, 1995). The present work supports the notion that during normal development, the L–C boundary inhibits the movements of precursors between VZ regions. Recently, examination of theSey (small eye/Pax6 mutant) mutant mice has provided the first genetic evidence that the L–C boundary may play an active role in maintaining regional patterning in telencephalon. InSey mice, a large population ofDlx2-positive cells are ectopically present within the cortex early in development. This suggests that the L–C boundary fails to form normally in these animals (Stoykova et al., 1996). In addition, in these animals, both the LGE and the CVZ are abnormally enlarged, consistent with the idea that failure of formation of the L–C boundary may result in improper telencephalic development.
Two independent studies have suggested that postmitotic striatal cells may migrate into the cerebral cortex during normal development (De Carlos et al., 1996; Anderson et al., 1997). This migration appears to occur not in the ventricular zone, but through migration through the postmitotic striatum into the overlying intermediate zone of the cortex (Fig. 7). Recently, Anderson et al. (1997) have provided evidence that this population represents a specific population of GABA-containing interneurons. This suggests that the production of a cortical cell type actually occurs in the striatum rather than in the cortex. This is consistent with the notion that specific cues exist locally in proliferative zones and that the restriction to migration of cells within the VZ (but not in more differentiated populations) is critical to maintaining appropriate local induction (Lumsden and Gulisano, 1997).
Possible mechanisms of restriction at the L–C border
Three classes of mechanisms could account for the restriction to tangential dispersion at the L–C boundary: (1) the presence of a structural barrier (Snow et al., 1990), 2) the presence of inhibitory cues within the boundary region (Luo, et al., 1993; Keynes and Cook, 1995), and 3) the expression in the boundary of extracellular matrix or homotypic adhesion molecules (Moscona, 1963; Rutishauser and Jessell, 1988; Krushel and van der Kooy, 1993; Gates et al., 1995; Matsunami and Takeichi, 1995; Gotz et al., 1996). We have used in vitroheterotopic transplants to distinguish these mechanisms.
The presence of a glial palisade within the L–C boundary is consistent with the notion that tangential dispersion is constrained physically by this palisade. However, if a structural barrier were responsible for restriction to migration at the L–C boundary, the arrest of tangential dispersion at the boundary should be sharply delineated. However, within the CVZ, the inhibition to tangential dispersion appears graded. This is reflected in the decreasing numbers of tangentially oriented cells within the CVZ near the L–C boundary. Nonetheless, formally it remains possible that decreases in cell dispersion near the L–C border results from higher densities of cells in this region. Presently, we have no data to support or refute this possibility.
The decrease in the number of tangentially oriented neurons in CVZ areas near the L–C boundary is consistent with either diffusible long-range inhibitory cues or local adhesion molecules/short-range diffusible cues inhibiting migration in the L–C boundary region. Our apposition explant suggest the later possibility, that the inhibition to lateral dispersion is mediated by either a short-range diffusible cue or a contact-dependent mechanism.
At present, the molecular nature of the inhibition to tangential migration within the L–C border is unknown. Previous work suggests a number of candidates that should be tested in the future. For instance, work by a number of groups has shown that homotypic adhesion systems are active within the telencephalon throughout much of development (Krushel and van der Kooy, 1993; Gotz et al., 1996). Furthermore, a number of adhesion and extracellular matrix molecules have been reported to have high levels of expression within the L–C boundary. For example, the adhesion molecule tenascin is present in this region (Gates et al., 1995) and has been shown previously to act to inhibit migration (Wehrle-Haller and Chiquet, 1993). Similarly, a recent report has demonstrated that during a similar period in mice development, R-cadherin is present within the L–C border (Matsunami and Takeichi, 1995). In line with our suggestion that the radial glial palisade acts to mediate the inhibitory properties of the L–C boundary, it is possible that they express these or similar molecules. It will be interesting to examine whether targeted mutations of these molecules affect the integrity of the L–C boundary.
The role of tangential dispersion in development
Examination of early neurogenesis in ferret (E29) suggests cell divisions remain coherent during these phases of development (Chenn and McConnell, 1995). Furthermore, in rodents, progenitors within the VZ are joined by gap junctions into cohorts of ≥60 cells during early neurogenesis (LoTurco and Kriegstein, 1991). These observations are consistent with the movements of progenitors being constrained during this period of development. Restriction of tangential dispersion might be important for the establishment of regional identity within the telencephalic neuroectoderm, because prevention of cell dispersion during this period may be necessary to ensure the stability of positional cues (Rakic, 1988; Placzek, et al., 1993; Ericson et al., 1995; Lumsden and Gulisano, 1997).
In contrast, later in telencephalic development, cell dispersion becomes pronounced. The onset of tangential dispersion within proliferative zones correlates with a reduction in the size of the cohorts of cells joined by gap junctions (LoTurco and Kriegstein, 1991). Commencing at E15–E16, our present studies show that tangential dispersion of neurons occurs widely throughout telencephalic proliferative zones, but is restricted at the L–C boundary.
Organization and the role of the L–C border
The L–C boundary, at the time it restricts tangential dispersion, contains a specialized cellular organization. Radial glia within the intermediate zone form a palisade that acts as a preferential pathway for the migration of postmitotic neurons to the lateral cerebral cortex (Austin and Cepko, 1990; Bayer et al., 1991). We show that this radial glial palisade originates in the L–C boundary, raising the possibility that the inhibitory nature of the border results from a special property of these radial glial end feet. We suggest the radial glial palisade serves the dual functions of preventing movement of cells between the LGE and CVZ while also redirecting tangentially dispersing CVZ cells toward the lateral cortical laminae (Fig. 6). Recently, numerous lines of evidence support that striatal cells (albeit a specialized population) are able to migrate into the cerebral cortex through the intermediate zone. As such, it is unlikely that the posited role of the glial pallisade in restricting migration between the VZ regions acts to restrict migration between regional territories in the intermediate zone. Indeed, such migration may be essential for normal patterning.
We suggest the following scenario (Fig. 7). During early development, cells undergo coherent cell division with little cell dispersion. After the earliest born neurons are postmitotic, tangential dispersion commences. Final cell positioning is reached by the competing influences of tangential dispersion, radial migration, and proliferative zone boundary restrictions. Radial glia, by providing a preferential pathway for migration, act to shepherd nascent neurons along specific migratory pathways (Rakic, 1972; Smart and Sturrock, 1979; Alvarez-Buylla et al., 1988; Gasser and Hatten, 1990a,b; Gray and Sanes, 1991; Hatten, 1993; Kornack and Rakic, 1995; Rakic, 1995), whereas tangential movements of cells in the VZ (Fishell et al., 1993;Walsh and Cepko, 1993; O’Rourke et al., 1997), SVZ (Menezes and Luskin, 1994; Doetsch and Alvarez-Buylla, 1996), and intermediate zone (Gadisseux et al., 1989; O’Rourke et al., 1992; De Carlos et al., 1996; Anderson et al., 1997) act to disperse them. We suggest that the L–C boundary plays a critical role during the early specification steps occurring within the proliferative zone, by maintaining the integrity of adjacent telencephalic regions.
Together, this suggests that the L–C boundary serves two functions. First, it isolates neighboring regions, allowing environments to be maintained with distinct local cues. Second, it restricts cellular interactions at the boundary region, permitting newly postmitotic (and perhaps still mitotic) striatal and cortical cells from encountering overlapping sets of instructional cues. Although the present evidence supports only that the L–C boundary acts to restrict cell migration early in the differentiation process, evidence from work in Drosophila and examination of other borders in the CNS (such as the mesencephalic/metencephalic border) argue that borders act as important signaling centers (Bally-Cuif et al., 1992; Vincent and Lawrence, 1994). It will be interesting to test whether in addition to restricting cell movements, the L–C boundary also possesses signaling properties.
This work was supported by Grant NS 32993 from National Institutes of Health. We thank A. Ruiz i Altaba, M. E. Hatten, K. Zimmerman, K. Campbell, A. Schier, P. Rakic, A. Joyner, A. Alvarez-Buylla, W. Talbot, and T. O’Connor for valuable discussions and critical reading of this manuscript; R. Baker for help with the DiI injections; and A. Ruiz i Altaba for help with the summary schematic. We also thank Dr. A. Frankfurter for generously supplying us with TUJ1 antibody and Dr. D. Ellis for advice on the acridine orange staining.
Correspondence should be addressed to Dr. Fishell, Developmental Genetics Program and the Department of Cell Biology, The Skirball Institute of Biomolecular Medicine, New York University Medical Center, 540 First Avenue, New York, NY 10016.