The mammalian cerebral cortex is organized into horizontal and vertical arrays of neurons and their fiber connections that form anatomically and physiologically distinct laminar and columnar compartments. However, the developmental mechanism(s) underlying this dichotomous pattern remains a mystery. We provide anatomical evidence suggesting that reelin, a diffusible protein produced and secreted by Cajal–Retzius cells, is involved in the developmental formation of the vertical cell structures in the mouse presubicular cortex, the unique site where the vertical columnar arrays of cortical plate neurons and their dendritic branches are clearly identified during the early postnatal period. Our results also suggest that reelin plays a role in the formation of these vertical structures by acting as an inhibitory or stop signal for cortical plate neurons and their dendritic extensions. In addition to having perturbed horizontal laminae, reeler mutant mice, lacking reelin, display disruption of these vertical structures. Based on the present findings, we hypothesize that reelin and Cajal–Retzius cells regulate the developmental formation of not only horizontal laminations, but also vertical columnar structures in the cerebral cortex.
Neuronal migration and positioning are essential steps in the genesis of the central nervous system (CNS). The mammalian cerebral cortex is a highly ordered brain structure and its fundamental framework is similar in all mammals (Jacobson, 1993). The cerebral cortex is organized into clear laminae: the different classes of neurons reside in an organized radial array of horizontal cellular layers (Eccles, 1984; Mountcastle, 1997). There also are vertical columnar arrays of neurons running orthogonal to the horizontal laminae (Eccles, 1984; Mountcastle, 1997): they are particularly conspicuous in the temporal cortex of humans and other primates (Jones, 2000). This dichotomous pattern of cortical cytoarchitecture appears to be the resultant of several cellular and molecular processes that occur in spatial and temporal sequence during development. However, to date, this intriguing issue remains mysterious.
Due to anomalous migration and positioning of postmitotic neurons during development, the reeler mouse, an autosomal recessive mouse mutant (Falconer, 1951), manifests abnormal laminar organization of the cerebral cortex (Caviness and Rakic, 1978; Goffinet, 1984; Rakic and Caviness, 1995). This pheno-type is a result of the disruption of the reeler gene (Relnrl, formerly rl) encoding a diffusible protein, reelin, that has several structural characteristics of extracellular matrix proteins (D’Arcangelo et al., 1995; Hirotsune et al., 1995). During corticogenesis in normal mice, reelin is produced by preplate Cajal–Retzius cells (D’Arcangelo et al., 1995; Hirotsune et al., 1995; Ogawa et al., 1995), which are located just below the pial surface (Bayer and Altman, 1991; Jacobson, 1993). The preplate then splits into two components, the marginal zone and the subplate layer, and the young postmitotic neurons that have migrated along radial glial fibers from the germinal ventricular zone (Rakic, 1995) form the cortical plate between these components (Marin-Padilla, 1998). The vertical position of cortical plate neurons is determined by the time of their origination and then clear horizontal layering is formed in the cortex (Rakic, 1995). In reeler mutant mice, Cajal–Retzius cells do not produce reelin and the preplate does not split. As a consequence, the cortical plate ectopically locates underneath the subplate neurons and the characteristic inside-out layering is perturbed. This is evidence for the role of reelin in the vertical positioning of cortical plate neurons that is essential for the formation of horizontal layering. In the development of the CNS, reelin has been identified as a crucial molecule that defines architectonic patterns by controlling neuronal migration (D’Arcangelo et al., 1995; Ogawa et al., 1995; Miyata et al., 1997; Nakajima et al., 1997; Dulabon et al., 2000; Yip et al., 2000; Ohshima et al., 2001; Magdaleno et al., 2002) and axon growth and synaptic connectivity (Del Río et al., 1997; Ghosh, 1997; Borrell et al., 1999; Rice et al., 2001). Recently, reelin receptors and other molecules involved in the reelin signaling cascades also have been identified (Rice and Curran, 2001).
We now report that during the early postnatal stage, vertical columnar arrays of cortical neurons and their dendritic processes are conspicuous in the mouse presubiculum, a multi-layered and periallocortical region (O’Mara et al., 2001). They provide a good model for investigating the cellular and molecular cues that direct both the vertical and horizontal positioning of postmitotic cortical neurons. Results from our present study suggest that reelin secreted by Cajal–Retzius cells may control the developmental formation of the vertical structures in the presubicular cortex by acting as an inhibitory or stop signal for cortical plate neurons and their dendritic extensions.
Materials and Methods
The reeler mouse colony was originally derived from heterozygous B6C3Fe-a/a-rl adults (The Jackson Laboratory). Timed pregnancies were induced by mating homozygous (rl−/−) males to heterozygous (rl+/−) females, or by bleeding heterozygotes of both sexes. The day of birth was recorded as postnatal day 0 (P0). Embryos and neonates of known age were kept in an in-house breeding colony. Care of the animals was in accordance with regulations promulgated by the Center for Animal Resources and Development of Kumamoto University.
Newborn mice (P0) and mice aged P1, P2, P3, P4, P6, P9, P14 and P28, as well as adult mice were used. They received an i.p. injection of a lethal dose of pentobarbital and were perfused transcardially with 0.9% (w/v) saline in 0.01 M phosphate buffer, pH 7.4 (PBS), followed by ice-cold 4% (w/v) paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 (PB). The brains were removed, postfixed with the same fixative at 4°C overnight and then kept overnight at 4°C in 0.1 M PB containing 30% (w/v) sucrose for cryoprotection. The brains were subsequently embedded in OCT compound (Sakura Finetechnical, Japan) and frozen in dry-ice/acetone. Cryostat sections were cut and kept in PBS until use.
CR-50, a mouse monoclonal antibody that recognizes an epitope in the N-terminal region of reelin (D’Arcangelo et al., 1997), was a gift from Dr M. Ogawa, The Institute of Physical and Chemical Research (RIKEN), Japan (Ogawa et al., 1995; Nishikawa et al., 1999; Hamasaki et al., 2001a). A mouse monoclonal antibody to microtubule-associated protein 2 (MAP2; Sigma, MO) was also used as primary antibody (Hamasaki et al., 2001a). The sections were blocked with 3% bovine serum albumin (BSA)–PBS for 1 h and then incubated overnight at 4°C in 3% (w/v) BSA–PBS containing primary antibody. Immunoreactivity was detected by FITC (Vector, CA) or Texas Red (Vector) conjugated secondary antibodies. Propidium iodide (PI) staining was also used (Hamasaki et al., 2001b). The fluorescence activities were observed and recorded under a confocal laser-scanning microscope (Fluoview, Olympus, Japan). The images obtained were printed using Pictrography 3000 (Fuji Film, Japan).
Scoring of Apical Dendrite Orientation
Apical dendrites of cortical pyramidal neurons from the presubicular cortex of heterozygous (rl+/−) or homozygous (rl−/−) mice aged P14 were visualized with anti-MAP2 antibody. According to the previous report (Polleux et al., 2000), they were scored as being directed towards the pia (45°–135°), the white matter (225°–315°), or as horizontal (135°–225° or 45°–315°).
Vertical columnar arrays of neuronal cell bodies in the presubicular cortex
At age P3 and P4, vertical columnar arrays of neuronal cell bodies running orthogonal to the horizontal laminae were clearly present in the mouse presubicular cortex. Vertical columns were visible even to the untutored eye in PI (Fig. 1A) or Nissl (Fig. 1B,C) stained preparations. They were conspicuous in the upper half of the cortical plate that corresponds to the supragranular layer (i.e. layers II and III) and had a periodicity of ~55 μm (55.2 ± 12.3 mm) in the horizontal extent. Each vertical column was composed of a cluster of young postmitotic neurons; between columns, a narrow inter-columnar space poor in neurons could be seen (Fig. 1B,C).
Relationship between Reelin and Vertical Columnar Arrays of Cortical Plate Neurons
At birth (P0), neurons in layers IV–VI have arrived at their final position, while neurons in layers II and III are still migrating toward their final destinations (Jacobson, 1993). At this stage, the superficial marginal zone appeared to contain Cajal–Retzius cells stained for reelin with CR-50 (Fig. 2A,B). Young postmitotic neurons formed the cortical plate beneath the marginal zone enriched in reelin. No definite vertical structures were found in either the cortical plate or the marginal zone at P0.
At P2–4, late-born cortical plate neurons have migrated from the germinal ventricular zone and are approaching their final position to form the supragranular layer (Jacobson, 1993). At P2, PI staining showed an indication of vertical columnar arrays of neurons in the upper half of the cortical plate (i.e. the future supragranular layer; Fig. 2C). In the marginal zone (Fig. 2D), CR-50 labeling was found in both the neuronal perikarya and neuropils of Cajal–Retzius cells, and in the surrounding extracellular area, probably due to the diffusible nature of reelin (D’Arcangelo et al., 1997). Cajal–Retzius cells appeared to be proceeding to form clusters and CR-50 labeling was inhomogeneously distributed in the marginal zone (Fig. 2D–F). It seemed that cortical plate neurons preferentially accumulated underneath the marginal zone area poor in Cajal–Retzius cells to form the supragranular layer (Fig. 2E–H). At P3–4, vertical columns composed of cortical plate neurons were clearly present in the supragranular layer (Fig. 2I–K), as described above (also see Fig. 1). Interestingly, clusters of Cajal–Retzius cells formed well-delineated territories with a specific topography and CR-50 labeling showed a periodic modulation in the horizontal dimension tangential to the cortical lamination (Fig. 2J,K). Reelin-rich zones appeared to descend into the inter-columnar spaces in the cortical plate (Fig. 2J,K); they contained the perpendicularly oriented processes of Cajal–Retzius cells. Sections cut parallel to the cortical surface revealed a mosaic-like pattern of regularly spaced spots with low CR-50 immunoreactivity in the marginal zone (Fig. 2L). Comparison with adjacent PI-stained sections at the border between the marginal zone and the supragranular layer (Fig. 2M–O) showed that the distributions of cortical plate neurons and CR-50 labeling were almost complementary; regions with poor CR-50 immunoreactivity contained clusters of cortical plate neurons. These data suggest that cortical plate neurons arrange in a columnar fashion due to being stopped by reelin and that their horizontal positioning is stereotypically defined by the specific territorial distribution of reelin secreted by Cajal–Retzius cells in the marginal zone.
Relationship between Reelin and Dendritic Clusters in the Marginal Zone
By P14, all cortical plate neurons have settled in their final positions and are engaged in the formation of fiber connections (Jacobson, 1993). The apical dendrites of cortical pyramidal cells of layers II, III and V form clusters that ascend through the cortical plate and send their terminal arrays to the marginal zone (Fleischhauer et al., 1972; Peters and Walsh, 1972) — also see Fig. 5A. At this stage, the columnar arrays of neuronal cell bodies were no longer discernible in the cortical plate (Fig. 3A,C), although CR-50 labeling continued to show periodic modulation in the marginal zone (Fig. 3B,C). Interestingly, reelin-poor zones contained columnar tufts of dendritic processes of cortical pyramidal neurons positive for MAP2 (Fig. 3D,E). Sections cut parallel to the cortical surface revealed a mosaic-like pattern of irregularly spaced spots poor in CR-50 labeling (Fig. 3F) or enriched in MAP2 labeling (Fig. 3G) in the marginal zone. Comparison with adjacent sections in the marginal zone showed that the distributions of CR-50 labeling and MAP2-immuno-reactive dendrites were almost complementary: regions with poor CR-50 immunoreactivity exhibited clusters of dendrites of cortical pyramidal neurons (Fig. 3H,I). Thus, the dendritic branches of cortical pyramidal neurons cluster by avoiding reelin-rich zones in the marginal zone, suggesting that reelin may act as a barrier to their extensions.
In P28 and adult mice, Cajal–Retzius cells labeled with CR-50 were scarcely seen in the marginal zone (Fig. 3J). At this stage, columnar tufts of dendritic branches were no longer apparent in the marginal zone (Fig. 3K). Also, vertical columnar arrays of neurons were not clearly identifiable in the cortical plate (Fig. 3L).
Disruption of the Vertical Structures in reeler Mutant Mice
Compared to wild-type (rl+/+) or heterozygous (rl+/−) mice (see Fig. 1), in homozygous (rl−/−) P3 mice there were no detectable vertical columnar arrays or horizontal laminae of neurons in the cortical plate (Fig. 4). At P14, clusters of dendritic branches of cortical pyramidal neurons in the marginal zone were clearly evident in wild-type and heterozygous littermates (Fig. 5A), but not in homozygous mice (Fig. 5B). Furthermore, the orientation and positioning of cortical plate neurons varied in the homo-zygous mice (Caviness, 1976; Landrieu and Goffinet, 1981). Although the apical dendrites of cortical pyramidal neurons were oriented almost orthogonal to the pial surface in wild-type or heterozygous mice (Fig. 5A), many appeared to be oblique, to run horizontally, or to be inverted in homozygous mice (Fig. 5B). Our semi-quantitative study (Fig. 5C) showed that most apical dendrites were directed towards the pial surface in the heterozygous (77.5 ± 3.5%; n = 200), but not in the homozygous (28.5 ± 2.6%; n = 200) mice. These findings indicate that reelin function is required for the developmental formation of both the horizontal layering and the vertically orientated architectures in the presubicular cortex.
In addition to the horizontal layering, the microcolumn (or minicolumn) that consists of vertical arrays of neurons and their fiber connections, functions as a fundamental anatomical processing unit of the cerebral cortex (Eccles, 1984; Mountcastle, 1997). While the microcolumns are thought to be produced by the iterative division of a small cluster of progenitor cells and to form a series of repeating units across the horizontal extent of the cortex (Mountcastle, 1997; Jones, 2000), vertical chains of cells are morphologically conspicuous only in the primate temporal cortex but not in other areas, nor in the cortex of non-primates.
In this study we showed the presence of two transient vertical structures, identified with simple anatomical techniques in the developing mouse presubicular cortex (for reference, see Fig. 6). One structure consists of the vertical columnar arrays of young cortical plate neurons in the supragranular layer. These occur most conspicuously at P3–P4, when the late-born cortical neurons ascend in linear arrays along a scaffold of radial fibers of astroglia or are proceeding to their final positions (Jacobson, 1993). The other structure is comprised of columnar tufts of dendritic processes of cortical pyramidal neurons in the marginal zone that are most remarkable at P14 when dendritic deployment proceeds and the fiber connections are being established (Jacobson, 1993): it seems to be a prototype of the dendritic clusters that group cortical neurons into modules of microcolumnar size (Mountcastle, 1997). In P28 as well as adult mice, these structures can no longer be clearly discerned because they are veiled by tightly packed cell bodies and dendritic branches of cortical neurons. As previously suggested (Vogt Weisenhorn et al., 1994; Del Río et al., 1995, 1996; Spreafico et al., 1995; Marin-Padilla, 1998), at this stage there is only a small population of Cajal–Retzius cells in the marginal zone. Our present results lead us to posit that the structures represent micro-anatomical units that underlie the development of the territorial organization that results in the positioning of cortical neurons and their formation of fiber connections in the horizontal dimension. According to the radial unit hypothesis (Rakic, 1988), the horizontal coordinates of cortical neurons are determined by the relative positions of their precursor cells in the germinal ventricular zone. It has been suggested that the germinal ventricular zone contains a mosaic of ontogenetic units that are composed of the neuronal precursors for a cortical column; alternatively, the mosaicism of the germinal ventricular zone is reproducible in the cortical plate. It is presently unknown whether or how the cellular microcolumns shown here are anatomically and functionally related to the microcolumns originating from the clonal modality. We are currently investigating the possible role of these microcolumns as functional units in cortical activities related to intracortical fiber connections.
It has been suggested that reelin functions as an inhibitory or stop signal for neuronal migration that defines architectonic patterning in the CNS (Ogawa et al., 1995; D’Arcangelo et al., 1997; Curran and D’Arcangelo, 1998; Frotscher, 1998; Pearlman et al., 1998; Dulabon et al., 2000; Walsh and Goffinet, 2000; Yip et al., 2000). This hypothesis is strongly supported by our present results. We found that at the early postnatal period, CR-50 labeling in the marginal zone appears to be organized in a mosaic pattern with periodic modulation in the horizontal dimension. This mosaicism of reelin statement resulting from the strategic location of Cajal–Retzius cells is almost complementary to that of vertical columnar arrays of cortical plate neurons or their dendritic tufts, which are both preferentially located by avoiding reelin-rich zones. Thus, we suggest that reelin may play a role in the formation of the vertical columns shown here by acting as an inhibitory or stop signal for cortical plate neurons and their dendritic extensions.
Finally, we hypothesize that reelin and Cajal–Retzius cells in the marginal zone may play a role in the formation of not only horizontal laminations, but also vertical columnar structures in the developing cerebral cortex. This is supported by the present finding that the vertical structures shown here are totally disorganized in reeler mutant mice. Our hypothesis coincides with suggestions that Cajal–Retzius cells may coordinate positional information essential for the early areal and columnar specification of the underlying cortex (Galuske and Singer, 1996; Schmidt et al., 1996; Schwartz et al., 1998; Soria and Fairén, 2000; Hevner et al., 2001; Zecevic and Rakic, 2001). However, this hypothesis is based on developmental and anatomical evidence found in a specialized region of the mouse cortex, i.e. the presubicular cortex. Further studies are needed to determine whether our hypothesis applies to other cortical areas in rodents or different species as a fundamental mechanism that underlies the developmental formation of the cerebral cortex.
This work was supported by a grant-in-aid for scientific research from the Ministry of Education, Culture, Sports Science and Technology of Japan. We thank Dr Ogawa of the RIKEN in Japan for providing monoclonal antibody CR-50 and for his technical advice.