Specialized subsets of early-generated neurons provide the cellular cues that are necessary for the establishment of characteristic cell and fiber interactions in each brain region. During the development of the mammalian cerebral cortex, the early-generated cells line up in the most superficial part of the telencephalic pallium forming the preplate. It has been generally thought that the preplate derivatives are exclusively located in the cortical region and govern the early histogenetic phase of cortical development. However, we here disclose an unexpected evidence that a subset of early-generated neurons of the piriform preplate migrate inward into and disperse within the subcortical structure striatum during the embryonic stage. Their migratory route is unique and its direction is opposite to the ordinary migration of neuronal precursors directed outward from the periventricular germinal zone. After immigrating into the developing striatum, these early-generated cells are closely associated with the intrastriatal fascicules of axons. The majority of these cells are eliminated by apoptotic cell death during the early postnatal stage. Based on these findings, we propose a new concept: the preplate neurons may not only direct cortical histogenesis but also change their location to play a role in the development of subcortical structures.
The development of mammalian brain is a multifactorial process in which specific cellular and molecular cues are necessary for the establishment of characteristic cell and fiber interactions in each brain region (Jessell, 1991). Among the cellular cues are specialized subsets of early-generated neurons (Ghosh et al., 1990; Ogawa et al., 1995; Sretavan et al., 1995; Del Rio et al., 1997; Sato et al., 1998). During the development of the mammalian neocortex, the earliest postmitotic neurons line up in the most superficial part of the telencephalic pallium, forming what has been called the preplate (Rickmann et al., 1977; Stewart and Pearlman, 1987; De Carlos and O'Leary, 1992; Supèr et al., 1998), primordial plexiform layer (DeDiego et al., 1994; Marin-Padilla, 1998), or early marginal zone (Rickmann et al., 1977; Raedler and Raedler, 1978). The dual origin theory of the mammalian neocortex (Marin-Padilla, 1998) proposes that the preplate represents a primitive cortical organization that is phylogenetically and ontogenetically derived from a premammalian cortex. With the development advanced, the preplate is split in two layers, where the specialized subsets of early-generated neurons arise; Cajal-Retzius cells in the marginal zone and subplate neurons in the subplate (Marin-Padilla, 1971, 1998; Kostovic and Molliver, 1974; Luskin and Shatz, 1985; Valverde et al., 1989; Kostovic and Rakic, 1990). Cajal-Retzius cells secrete a molecule, Reelin, that is essential for orderly lamination of cortical neurons (D'Arcangelo et al., 1995, 1997; Ogawa et al., 1995; Schiffmann et al., 1997). Cajal-Retzius cells are also found in the developing hippocampus, where Reelin directs hippocampal afferents to their final destination (Del Rio et al., 1997). Subplate neurons form the transient connections with the cortical afferents (Kostovic and Rakic, 1990) and are involved in accurate target selection by thalamocortical inputs (Ghosh et al., 1990). Most of these cortical early-generated neurons arguably die; their life is transient during corticogenesis (Bradford et al., 1977; Kostovic and Rakic, 1980, 1990; Luskin and Shatz, 1985; Derer and Derer, 1990; Wood et al., 1992; Del Rio et al., 1995, 1996).
The paleocortex of mammals develops in the ventrolateral aspect of the telencephalon. This phylogenetically ancient cortex consists of only three layers, similar to the cerebral cortex of reptiles. In the embryonic stage, the piriform preplate in the paleocortical area contains a large cluster of early-generated neurons (Valverde and Santacana, 1994; De Carlos et al., 1996; Meyer et al., 1998). Cajal-Retzius cells develop there just as seen in neocortical regions (Del Rio et al., 1995; Valverde et al., 1995). In contrast, the subplate does not form a clearly separated cell layer (Bayer and Altman, 1991; Valverde and Santacana, 1994). Another subset of early-generated neurons, named lot cells, originates from the neocortical neuroepithelium, tangentially migrates, and constitutes a cellular array in the piriform area as guidepost cells for olfactory bulb axons (Sato et al., 1998; Tomioka et al., 2000). It was suggested that some neocortical early-generated neurons are derived from the retrobulbar compartment and are temporarily located in the piriform area en route to their destination (Meyer et al., 1998). These findings may imply that the piriform preplate is composed of various cell populations.
We now show that a subset of early-generated neurons in the piriform preplate migrates inward into and dispersed within the subcortical structure striatum during embryonic stage. These cells do not express Reelin or lot1, a marker for lot cells, indicating that they may belong to a distinct cell population. We here name these cells ‘SPEG (subcortical preplate-derived early-generated) neurons’. We suggest that the preplate derivatives may have a wider influence on the development of telencephalon.
Materials and Methods
Pregnant Wistar/albino rat dams were obtained from a commercial vendor (SLC, Japan). The morning on which the vaginal plug was detected was recorded as embryo day 0 (E0); the day of birth was recorded as postnatal day 0 (P0). Delivery usually occurred on the twenty-first day of gestation. 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.
Organotypic Slice Cultures
Wistar/albino rats timed-pregnant to E14 or E18 were placed under deep anesthesia with fluothane gas. After a midline laparotomy, fetal brains were removed and sliced (300–350 μm) coronally using a tissue chopper. Tissue slices were kept in ice-cold dissecting medium [100% DMEM/F-12 (Life Technologies, Gaithersburg, MD) with 3.85 mg/ml glucose, pH 7.35]. The telencephalon from E14 and the striatum, neocortical intermediate zone (IZ) from E18 were excised. The dissected tissues were then transferred onto a collagen-coated membrane (Transwell-COL 3491, Coster, Japan), and incubated in culture medium [15.6 mg/ml D-MEM/ F-12 medium containing 10% fetal bovine serum, 1% N2 supplement (Gibco-BRL, Japan), 1.2 mg/ml NaHCO3 and 10 mM 2-mercaptoethanol and 3.85 mg/ml glucose, pH 7.20]. Penicillin (100 U/ml) and streptomycin (100 mg/ml) were also added. The culture plates were placed in an incubator at 37°C under a 5% CO2-enriched moist atmosphere. The medium was changed twice a week.
Collagen was obtained from rat tail tendon (Ebendal and Jacobson, 1977). The piriform preplate was dissected from E14 coronal slices of 300 μm and cut into pieces 100–300 μm in diameter. These explants were embedded in the collagen gel or in the 1:1 mixture of collagen gel and Matrigel (Becton Dickinson, Bedford, MA), and were then cultured with the same medium in the same condition as described above.
To trace cells and axons, the fluorescent lipophilic carbocyanine dye 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) or 3,3-dioctadecyloxacarbocyanine perchlorate (DiO) crystals (Molecular Probes, Eugene, OR) were implanted into the living explants (Honig and Hume, 1986). After the incubation period appropriate for each experiment, the tissues were fixed with 4% paraformaldehyde. Fluorescent images were obtained using a Fluoview confocal laser-scanning microscope (Olympus, Japan). We also used the horseradish peroxidase-conjugated cholera toxin B subunit (HRP–CTBS) (RBI, Natick, MA) (Trojanowski et al., 1981) instead of the fluorescent dyes. For double-immunolabeling study for DiO or HRP–CTBS, cryostat sections (10 μm) were prepared from the fixed tissues, mounted on MAS-coated slides (Iwaki Glass, Japan), and then processed for calretinin or 5-bromo-2-deoxiuridine (BrdU) immunostaining.
For the in vivo study, rats aged E14, E16, P18, P0, P3, P6 and P14 were used. Rat embryos or neonates received an intraperitoneal injection of a lethal dose of pentobarbital and were perfused transcardially with 0.9% saline in 0.01 M phosphate buffer, pH 7.4 (PBS), followed by ice-cold 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 (PB). The brains were removed and postfixed with the same fixative at 4°C overnight, and were then kept in 0.1 M PB containing 30% sucrose at 4°C overnight for cryoprotection. The brains were embedded in OCT compound (Sakura Finetechnical, Japan) and frozen in dry-ice/acetone. Cryostat sections were cut and kept in PBS until use. For organotypic slice cultures, the tissues were fixed with 4% paraformaldehyde in 0.1 M PB at 4°C overnight, and then processed as described above. For gel cultures, the whole mount tissues were fixed by immersion in 4% paraformaldehyde in 0.1 M PB at 4°C overnight.
Timed pregnant dams were injected intraperitoneally at several stages of pregnancy with a single pulse (30 mg/kg body weight) of BrdU (10 mg/ml dissolved in saline; Nacalai Tesque, Japan). Cryostat sections (10 μm) were prepared from the neonate brains and the cultured tissues derived from E14 fetus brains, and were mounted on MAS-coated slides. They were first incubated in 2 N HCl for 90 min and in PBS, pH 8.5, for 30 s to neutralize the HCl, and then immunostained for BrdU.
Rabbit polyclonal antibody against calretinin (Chemicon, CA), goat polyclonal antibody against HRP (Biomeda, CA), mouse monoclonal antibody against neurofilament (Dako, Glostrup, Denmark), mouse monoclonal antibody against β-tubulin isotype III (TuJ1; Sigma, MO), mouse monoclonal antibody against microtubule-associated protein (MAP)-2 (Sigma), mouse monoclonal antibody against BrdU (Sigma), mouse monoclonal antibody against Reelin [CR-50; from Dr M. Ogawa, The Institute of Physical and Chemical Research (RIKEN), Japan] (Nishikawa et al., 1999), mouse monoclonal antibody against vimentin (Dako), and rat monoclonal antibody against lot1 (from Dr T. Hirata, National Institute of Genetics, Japan) were used as primary antibodies. Immunostaining was performed using the free-floating technique with gentle agitation. The sections were blocked with 3% bovine serum albumin (BSA)–PBS for 1 h and then incubated overnight at 4°C in 3% BSA–PBS containing primary antibodies. Immunoreactivity was detected by FITC- (Vector, CA), Texas Red- (Vector) or Cy3- (Amersham, Bucks, UK) conjugated secondary antibodies. For calretinin-immunostaining of whole mount tissues embedded in gels, PBS containing 0.5% Triton X-100 was used in place of PBS during these procedures. The immunostaining procedure for lot1 was in accordance with the method of Sato et al. (Sato et al., 1998). The fluorescence activities were observed and recorded under a confocal laser-scanning microscope (Fluoview, Olympus, Japan). The obtained images were printed using Pictrography 3000 (Fuji film, Japan). Cross-reactivity between the individual immunoreagents was tested by cross-fluorescence controls.
Terminal dUTP Nick End Labeling (TUNEL) Method with Calretinin Immunostaining
TUNEL labeling was carried out using the in situ Apoptosis Detection Kit MK500 (Takara, Japan). The labeled sections were also immunostained for calretinin.
Cell Counting Method
For in vivo study on postnatal striatum, sagittal sections (50 μm) at the mid-level of the striatum were prepared from P0 (n = 8), P6 (n = 9) and P14 (n = 13) neonate brains, and were processed for double- immunostaining for calretinin and neurofilament. All sections were analyzed by laser scanning. The total numbers of cell bodies positive for calretinin per one recorded field of 1.0 mm2 with a thickness of 50 μm were counted at P0, P6 and P14. Then the numbers at P6 and P14 were corrected by the developmental change in the striatal volume at each period: our study on Nissl-stained sagittal and coronal sections showed that striatal volumes at P6 and P14 were 1.52 and 3.58 times that at P0, respectively. For the preplate-ablation cultures from E14 telencephalon, the fixed tissues were immunolabeled for calretinin or MAP-2, and then the positive cells per one recorded striatal field of 1.0 mm2 with a thickness of 30 μm were counted in each experiment. For DiO-tracing cultures after 4 days in vitro, the area around the DiO injection site was divided into 1/3 sectors (cortical, striatal and basal in Fig. 3J). The numbers of DiO-labeled cells located in these sectors were counted and analyzed. The data were evaluated on a double-blind basis by three observers from our laboratory. Statistical analysis of the data was performed by means of the unpaired t-test and values of P < 0.05 were taken as significant.
Characterization of Calretinin-immunoreactive Cells in E14 Piriform Preplate
To label early-generated cell populations, we performed calretinin immunostaining, since this neuron-specific calcium- binding protein (Rogers, 1987) is expressed in the earliest cell populations of the rat preplate (Vogt Weisenhorn et al., 1994; Fonseca et al., 1995; Meyer et al., 1998) and other brain regions (Jiang and Swann, 1997; Frassoni et al., 1998), and is also a reliable marker for Cajal-Retzius cells (Vogt Weisenhorn et al., 1994; Alcantara et al., 1998; Meyer et al., 1998).
At E13, when striatal neurons just start to be generated (Fentress et al, 1981; Marchand and Lajoie, 1986), calretinin- positive cells were already distributed in the surface area of the telencephalic pallium. These cells formed a thin layer, which corresponded to emerging preplate (data not shown) (Vogt Weisenhorn et al., 1994).
At E14, just lateral to the developing striatum, the piriform preplate enlarged to form a dense cluster of calretinin-positive cells (arrows in Fig. 1A,B). To confirm their early generation, 12- or 12.5-day-old rat embryos (Valverde and Santacana, 1994; Meyer et al., 1998), which may correspond to E10.5–11.5 in mice (Fentress et al., 1981), were administered BrdU. The results demonstrated that a substantial number of calretinin-positive cells in the piriform preplate were labeled for BrdU in their nuclei (Fig. 1C–E). This was not the case when the injection occurred after E13 (data not shown). The piriform preplate cells positive for calretinin were simultaneously immunolabeled with anti-β-tubulin isotype III or anti-MAP-2 antibodies (Fig. 1F–K), which mark postmitotic young neurons or differentiated neurons, respectively (Ishii et al., 2000). Thus, the calretinin- positive early-generated neurons in the piriform preplate appear to exhibit neuronal phenotype at E14.
Transitional Patterns of Calretinin-immunoreactive E-G Cells in the Piriform Preplate and Striatum during the Embryonic Stage
At E16, the calretinin-positive cell cluster in the piriform preplate was dispersed toward the developing striatum (Fig. 2A); some cells could already be seen in the ventrolateral region of the nucleus (arrows in Fig. 2B). Their leading processes were oriented toward the striatum and frequently terminated in a growth cone-like structure (arrow in Fig. 2C). The dorsal part of the striatum near the ventricular or subventricular germinal zones was almost devoid of calretinin-positive cells (Fig. 2A). At E18, the calretinin-positive cells in the piriform preplate became sparse (Fig. 2D) (Meyer et al., 1998). The positive cells were diffusely distributed between the preplate and the striatum. Coincidentally, the number of calretinin-positive cells in the striatum drastically increased (Fig. 2D).
Based on these results, we hypothesized that calretinin- positive early-generated neurons in the piriform preplate migrate inward into the developing striatum. To address this issue, we performed in vitro experiments described in the following sections.
Inward Migration of Piriform Preplate Cells into the Developing Striatum In Vitro
To test our hypothesis that piriform preplate neurons migrate inward, we performed cell-tracing experiments using slice culture preparations. Twelve hours after the injection of DiO into the piriform preplate of E14 forebrain slices (inset in Fig. 3A), we observed some labeled cells located away from the injection site (arrows in Fig. 3A). The cells exhibited a leading process that extended toward the striatum (arrows in Fig. 3A). After 2 days in culture, the labeled cells traveled further and formed a migratory stream (Fig. 3B). By day 4, the labeled cells had entered the striatum where they were dispersed and oriented diversely (Fig. 3C). The time course of this cell displacement in vitro was in good agreement with data we obtained in vivo.
To characterize further the DiO-labeled migrating cells, we performed immunostaining of telencephalon slices after 4 days in culture. Our results confirmed that the DiO-labeled cells in the striatal area were immunopositive for calretinin (Fig. 3D–F). In addition, when BrdU had been injected at E12 or E12.5, their nuclei were BrdU-immunolabeled (Fig. 3G–I). Thus, at least a part of the DiO-labeled cells migrating from the piriform pre- plate into the striatum were calretinin-positive early-generated neurons.
It was also noted that a small number of cells labeled with DiO tangentially migrated (arrowheads in Fig. 3A). To evaluate the directional migration of the piriform preplate cells, we counted the numbers of DiO-labeled cells found in three defined sectors (Fig. 3J). The cell counts in cortical, striatal and basal sectors (Fig. 3J) after 4 days in vitro (n = 22) were 78 (13.6%), 480 (83.6%) and 16 (2.8%), respectively (Fig. 3K). This result suggests that most of the cells migrating out of E14 piriform preplate are directed toward the developing striatum.
Migratory Potential of Piriform Preplate Cells
To examine migratory potential of piriform preplate cells, we made explants from E14 piriform preplate, which were cocultured with explants from E18 subcortical IZ (Fig. 4A, left and upper right). In the beginning of coculture experiments, we injected DiO in the preplate explant to prelabel the intrinsic cells. After 6 days in vitro (n = 45), all the DiO-labeled cells we observed formed streams in the IZ (Fig. 4B) just as seen in Fig. 3A–C. Their leading processes were oriented away from the preplate explant. Some labeled cells displayed granular labeling, indicating that these cells were migrating (Honig and Hume, 1986). Furthermore, we cultured explants from E14 piriform preplate embedded in collagen gel or in a 1:1 mixture of collagen gel and Matrigel, a three-dimensional extracellular matrix gel of collagen IV, laminin, heparin sulfate proteoglycans and entactin–nidogen (Kleinman et al., 1982) (Fig. 4A, left and lower right). After 24 h in culture, many calretinin-labeled cells were migrating out from the preplate explants in collagen gel either with or without Matrigel (Fig. 4C,D). It was also noted that long leading processes (>200 μm) were more prominently observed in the presence of Matrigel (open arrows in Fig. 4D). The extension of leading processes is an important prerequisite for cell migration (Yee et al., 1999). These data indicated that piriform preplate cells preserve their migrating activity in the explant cultures and that their ability to extend the leading processes seemed to be influenced by extracellular matrix components.
Most of the Calretinin-Positive Cells in the Neonate Striatum Are Derived from the Piriform Preplate
To confirm that the calretinin-positive cells in the developing striatum originate from the piriform preplate, we prepared E14 lateral telencephalic slices, in which the preplate was included (controls) or ablated (preplate-ablated group) (Fig. 5A,B, top). After 7 days in culture, when the age is estimated to correspond to the neonate stage, there was no apparent difference between these two groups with regard to slice size (Fig. 5A,B, bottom) and numerical density of the cells expressing MAP-2 (Figs 5C,D, 6A) [5724.8 ± 347.9 and 5810.4 ± 311.1 cells per recorded fields in control (n = 15) and preplate-ablated (n = 21) cultures, respectively], suggesting that the migration and differentiation of striatal neurons derived from periventricular zones were not affected by the preplate ablation. In striking contrast, the striatum in preplate-ablated slices contained only a few calretinin-expressing cells (3.6 ± 5.0 cells per recorded field; n = 21) as compared with the control slices (105.0 ± 23.9 cells per recorded field; n = 15) (Figs 5E,F, 6B). Thus, the vast majority of calretinin-positive cells in the neonate striatum appear to be derived from the piriform preplate: it is unlikely that they may arise directly from the ventricular or subventricular germinal zones. Based on these findings, we also suggest that calretinin can be used as a marker for preplate-derived cells in the developing striatum.
Novel Features of Calretinin-positive Cells in the Neonate Striatum
To determine whether cells comparable to the preplate-derived striatal cells observed in vitro are present in the striatum in vivo, we performed calretinin-immunostaining of neonate striatum (P0). Calretinin-expressing cells were distributed throughout the striatum (Fig. 7A) (Schlosser et al., 1999). Double-labeling study disclosed that most calretinin-positive cells (>95%) were attached or located close to the intrastriatal fascicles of axons, the pencil fibers (Fig. 7A–G). In addition, the processes of some calretinin-immunoreactive cells were found to extend into axon fascicles (arrows in Fig. 7F). Cell-birthdating study also revealed that a significant number of these cells were labeled with BrdU after the BrdU injection at E12 or E12.5 (Fig. 7H–J). Thus, in vivo, the calretinin-positive early-generated neurons were localized in close proximity to the pencil fibers in the neonate striatum. Furthermore, the presence of calretinin- positive early-generated neuron in the postnatal striatum (Fig. 7A–J) confirms their inward migration in vivo, because these cells are distributed exclusively in the surface area of the telencephalon at E14 (Fig. 1A–E). In contrast, the calretinin- positive cells in the piriform area or around the lateral olfactory tract (LOT) had almost completely disappeared except for a few Cajal-Retzius cells (arrowheads in Fig. 7K).
To examine the possible relationships with other types of early-generated neurons, we performed double-immunostaining studies. Our results show that the calretinin-positive cells in P0 striatum were negative for Reelin recognized by CR-50 antibody (Fig. 7L,M) (Ogawa et al., 1995; Nishikawa et al., 1999) nor lot1, a marker for lot cells (Fig. 7N,O) (Sato et al., 1998).
We have identified a novel subset of early-generated preplate cells (SPEG neurons) that migrate inward into and dispersed within the striatum. They are marked by calretinin staining but not by Reelin or lot1 staining, indicating that they may belong to a distinct cell population among the early-generated neurons.
Close Association of SPEG Neurons with Striatal Growing Axons In Vitro
To examine the relationship between SPEG neurons and striatal growing axons within the perinatal striatum, embryonic (E18) striatal explants (n = 134) were cultured for 6 days. When a small crystal of DiI was implanted into the striatal explant on the fifth day of culture (inset in Fig. 8A), a distinct subset of Dil-labeled cells were found close to DiI-labeled axon fascicles at a long distance (~500 μm) from the DiI injection site (Fig. 8A). To determine their neurochemical nature, we used HRP–CTBS instead of DiI as a tracer since the cell morphology demonstrated by DiI becomes indistinct in the course of immunolabeling. Double-immunolabeling disclosed that all traced cells we examined were indeed calretinin-positive (Fig. 8C–E). They were preferentially labeled with BrdU (Fig. 8F–H) when BrdU was administered at E12 or E12.5. Thus, the association of SPEG neurons with striatal axons was reproduced in vitro. Furthermore, because their trailing processes were short and never extended into a tracer-injection site, we posited that these axon-associated cells might not be labeled retrogradely from the injection site but rather that they might change their locations along axons (O'Rourke et al., 1997).
Postnatal Change in the Distribution of SPEG Neurons
Since most preplate neurons are arguably transient and disappear by cell death, we performed calretinin-immunostaining to elucidate the postnatal changes in the distribution of SPEG neurons in the striatum. Sagittal sections of the striatum at P0 (Fig. 9A), P6 (Fig. 9B) and P14 (Fig. 9C) were examined. With advancing development, SPEG neurons recognized by the calretinin antibody were distributed more sparsely (Fig. 9A–C). Quantitative study disclosed that the numerical cell density of SPEG neurons counted per one recorded field at P0, P6 and P14 was 394 ± 36 (n = 8), 108 ± 17 (n = 9) and 39 ± 9 (n = 13), respectively (white columns in Fig. 9D). Considering their developmental dilution (Marin-Padilla, 1990; Vogt Weisenhorn et al., 1994; Del Rio et al., 1995, 1996; Spreafico et al., 1995), these numbers were corrected by the developmental change in the striatal volume: our study showed that the striatal volumes at P6 and P14 were 1.52 and 3.58 times the striatal volume at P0. The corrected numbers at P6 and P14 were 164 ± 26 and 139 ± 31, respectively, when the number at P0 was taken as the standard (black columns in Fig. 9D). Thus, our quantitative study clearly disclosed a significant diminution of SPEG neurons in the early postnatal stage; between P0 and P6 the decrease was pronounced (Fig. 9D).
To test the hypothesis that apoptotic cell death accounts for the postnatal diminution of SPEG neurons (Del Rio et al., 1995, 1996), we used TUNEL to detect dying cells directly (Spreafico et al., 1995) in the striatum at P3. The nuclei of a significant number of calretinin-positive cells were TUNEL labeled; the cytoplasm appeared shrunken and there was swelling in segments of dendrites (Fig. 9E–G) (Del Rio et al., 1995, 1996; Supèr et al., 2000), indicating that these cells underwent apoptotic death. As is the case in cortical Cajal-Retzius cells (Del Rio et al., 1995, 1996), programmed cell death seems to account for the postnatal diminution of SPEG neurons. This suggests that the life of a large part of SPEG neurons is transient during striatogenesis.
The Presence of Inward Cell Migration in Developing Telencephalon
In the present study, we identify a new set of migrating cells in the developing telencephalon. By local injection of DiO into the piriform preplate, we clearly demonstrated that the preplate cells in this region migrate inward into the striatum. This migratory pathway was further confirmed by the preplate-ablated slice culture experiment, where the calretinin-positive cells almost completely disappeared in the striatal area. BrdU labeling in vivo has shown both the exclusive distribution of calretinin-positive early-generated neurons in the E14 preplate and the appearance of this neuronal population in the neonatal striatum. This also supports the shift of these cells to the striatum since BrdU permanently labels neuronal cohorts that underwent the last cell division at the administration (Miller and Nowakowski, 1988). So far, three novel kinds of neuronal migration have been established: radial migration in which precursor cells climb on the shafts of radial glia outward to the brain surface (Rakic, 1990; Komuro and Rakic, 1995), tangential migration in which cells move parallel to the ventricular surface and perpendicular to radial glial fibers (O'Rourke et al., 1992; DeDiego et al, 1994; Anderson et al., 1997; Tamamaki et al., 1997; Zhu et al., 1999), and chain migration in which neuronal precursors move rostrally from the SVZ to the olfactory bulb as chain formations (Lois et al., 1996). In the developing striatum (Halliday and Cepko, 1992; De Carlos et al., 1996) or piriform cortex (Bayer et al., 1991; De Carlos et al., 1996), most neurons arise from the periventricular germinal zone, migrate radially outward and differentiate in the superficial positions. Thus, we suggest that the inward migration of SPEG neurons is novel because its direction is opposite to that of ordinary cell migration in the basal telencephalon and is also different from those of well- defined neuronal migrations in the other site of the developing telencephalon.
To get an insight into the possible substrate for the inward migration of preplate cells, we performed the double-immunostaining for calretinin and vimentin, a marker for radial glial fibers (Schnitzer et al., 1981). Our result shows that the migration pathway does not correspond with the distribution of radial glial fibers (data not shown). Furthermore, using the gel culture system, we tested the possibility that diffusible cues derived from the destination may regulate the directional migration of preplate cells, but failed to get an evidence that embryonic striatal explant makes an influence on the migratory profile of the preplate cells. Thus, the mechanism(s) underlying the inward migration of preplate cells is currently undetermined and under investigation in our laboratory.
Possible Origin of SPEG Neurons
Recently, it has been implied that Cajal-Retzius cells arise in other sites than the neocortical VZ: lateral ganglionic eminence (LGE) (De Carlos et al., 1996), paraolfactory VZ (Meyer et al., 1998) or medial ganglionic eminence (MGE) (Lavdas et al., 1999). We identified SPEG neurons initially in the E14 piriform preplate as calretinin-positive cells born at E12 or E12.5. Since we used BrdU-labeling technique to clarify their birthdate in the present study, we could not determine the birthplace of SPEG neurons. De Carlos et al. performed tracing studies with DiI injection and argued that cells born at E12 in the VZ of the LGE migrate outward to the piriform area of the developing telencephalon (De Carlos et al., 1996). Meyer et al. suggested that subpial granular layer (SGL) cells originate in paraolfactory VZ and subsequently form a large cluster in the E14 piriform area after their presumptive tangential migration in the subpial layer (Meyer et al., 1998). Slender fusiform shape, positivity for calretinin, and birthdate of SPEG neurons are similar to those of SGL cells that Meyer et al. described in piriform area on E14 (Meyer et al., 1998). Tomioka et al. recently reported a novel migration pathway from the neocortical to paleocortical area: the authors argue the neocortical origin of lot cells (Tomioka et al., 2000). Although SPEG neurons are negative for lot1, we cannot exclude the possibility that SPEG neurons are derived from the neocortical area before they are located in E14 piriform preplate. Thus, we would infer that there are three possible birthplaces of SPEG neurons: LGE, paraolfactory VZ or neocortical VZ. It is interesting to note that SPEG neurons and other types of early-generated neurons may have the common precursor and share the same site of origin.
Fate of SPEG Neurons
We show that striatal calretinin-positive cells are eliminated during first two postnatal weeks. TUNEL study suggests that the diminution is caused, at least in part, by apoptotic cell death. It was also noted that a small part of striatal calretinin-positive cell population remains over the early postnatal stage (Fig. 9C,D). This finding may raise the possibility that some preplate- derived cells may differentiate into striatal calretinin-positive interneurons. We performed the birthdate study; however calretinin-positive cells were never labeled for BrdU in the adult striatum after the injection at E12 or E12.5. Alternatively, recent report suggests that most of striatal interneurons are derived from the MGE (Marin et al., 2000). Taken together, we propose that almost all the striatal preplate-derived cells disappear during the early postnatal stage, while calretinin-positive interneurons, which account for a very small population in the adult striatum (Yamada et al., 1995), are supplied via the tangential cell migration from the MGE (Marin et al., 2000). Thus, SPEG neurons exist transiently during striatogenesis, suggesting that the function of SPEG neurons is probably limited to the primary implementation of striatal structures. We found that SPEG neurons are closely associated with the pencil fibers perinatally with a peak at P0 (Fig. 7A–G). Furthermore, SPEG neurons often extend their processes into the fascicules of axons (Fig. 7F,G). We suppose that SPEG neurons may play a role in the growth of the pencil fibers, as is the case for transient early-generated neurons required for the development of a particular brain region (Ghosh et al., 1990; Ogawa et al., 1995; Sretavan et al., 1995; Del Rio et al., 1997; Sato et al., 1998). In this context, the preplate derivatives may play a role not only in the development of the cortical area but also in the histogenesis of subcortical nuclei. Proper function or differentiation of subcortical preplate derivatives in brain development remain an intriguing issue and should be clarified in future. They may give rise to new insights into cellular mechanisms underlying brain development.
We thank Dr M. Ogawa of The Institute of Physical and Chemical Research (RIKEN), Japan for providing a monoclonal antibody CR-50, and Dr T. Hirata of the National Institute of Genetics, Japan for providing a monoclonal antibody lot1. This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan.
Address correspondence to Satoshi Goto, Department of Neurosurgery, Kumamoto University Medical School, Honjo 1-1-1, Kumamoto 860-8556, Japan. E-mail: firstname.lastname@example.org.