The extracellular matrix molecule reelin is a crucial molecule in CNS development, in particular in the cerebellum and cerebral cortex. In the cerebral cortex, reelin is provided by a small number of neurons located in the marginal zone (MZ). These neurons belong to the earliest neurons generated, but little is known about the molecular mechanisms of their specification. Here we describe that reelin-positive cells are strongly increased in the developing cortex of the Pax6 mutant mice Small eye. Shortly after the onset of reelin expression, the number of reelin- and calretinin-positive cells is doubled in the cortex of Pax6 mutants and this increase is further enhanced during development. In contrast, calbindin-positive cells in the MZ do not co-express reelin and are not altered in the Pax6 mutant cortex. The split of the preplate cells was also defective in the Pax6 mutant cortex, suggesting that the amount of reelin is crucial for positioning of the cortical plate between the MZ and subplate. We further show that Pax6 mutant cortical cells isolated in vitro do not develop an increase in reelin-positive cells, while cells isolated from the entire telencephalon do. Consistent with non-cell-autonomous mechanisms contributing to the increase in reelin-positive cells in the Pax6-deficient cortex, tangential migration of diverse cell types from the ventral telencephalon into the cortex is enhanced in the Pax6 mutant mice. Taken together, these experiments further elucidate how patterning of the forebrain by the transcription factor Pax6 regulates the specification of distinct neuronal subtypes in the cortical MZ.
The preplate is an essential structure for formation of the allo- and neocortex throughout vertebrates [reviewed by Supèr et al. (Supèr et al., 1998)]. The preplate contains the earliest generated neurons and is split by the later formed cortical plate (CP) into the marginal zone (MZ, later layer 1), situated above, and the subplate zone (SP), located below the CP (Marin-Padilla, 1978). Recent work has highlighted a great deal of cellular heterogeneity in the MZ. The MZ neurons differ in their morphology, neurotransmitter and Ca-binding protein content [for a review see Meyer et al. (Meyer et al., 1999)]. Most preplate and later MZ cells are GABAergic and co-express the Ca-binding proteins calretinin, calbindin and parvalbumin (Vogt Weissenhorn et al., 1994; Fonseca et al., 1995; Fairén et al., 1998; Meyer and Goffinet, 1998; Meyer et al., 1998, 1999). While the precise combinations of co-localization patterns are not fully understood and exhibit some species-specific differences, reelin has been detected with calretinin or calbindin and GABA in a subset of cells in the MZ of the developing cortex (Meyer et al., 1998; Ringstedt et al., 1998). The cellular diversity of MZ neurons seems to play a crucial role during cortical development, because MZ neurons have been shown to regulate various key developmental features, such as neuronal migration, radial glia morphology, layer formation and axonal pathfinding [for a review see Supèr et al. (Supèr et al., 1998)]. For example, the earliest generated neurons located in the preplate and later in the MZ and SP establish the first efferent axonal projections of the cerebral cortex and have been suggested to serve as pioneer fibers for later axonal projections (McConnell et al., 1989; del Rio et al., 1997; Supèr et al., 1998). Moreover, a key molecule in cortical development is the extracellular matrix molecule reelin (D’Arcangelo et al., 1995; Ogawa et al., 1995), contained in subsets of MZ neurons as described above. In the absence of reelin in the mouse mutant reeler, cortical lamination is severely affected due to defects in the split of the preplate, radial glia morphology and neuronal migration (D’Arcangelo et al., 1995; Magdaleno et al., 2002). Furthermore, these defects influence the laminar-specific properties of cortical neurons (Polleux et al., 1998).
An important question, therefore, is how the cellular diversity of MZ neurons and in particular the number of reelin-secreting cells are specified during development. During recent years it has become increasingly clear that tangential migration of distinct cell types contributes to many cell populations in the developing cortex, including the heterogeneous MZ [for a review see Marin and Rubenstein (Marin and Rubenstein, 2001)]. For example, a stream of Lhx6- and reelin-positive neurons originating in the medial ganglionic eminence (MGE) can be followed into the MZ of the cortex (Lavdas et al., 1999). Meyer and collegues identified a region in the retrobulbar ventricle from which streams of reelin-, calbindin- and calretinin-positive cells seem to emerge and migrate into rodent and human cortex forming the subpial granule layer (SGL) (Meyer et al., 1998, 2000; Meyer and Wahle, 1999). Little is known about the molecular mechanisms regulating the tangential migration of these different neuronal subtypes populating the MZ. Ligands of the TrkB receptors have been implicated in the regulation of tangential migration of reelin-positive cells. An excess of TrkB ligands in cortical development cause heterotopia of reelin- and calretinin-immunoreactive cells in the cortical MZ in vitro and in vivo (Brunstrom et al., 1997; Ringstedt et al., 1998) and time lapse videomicroscopy directly visualized an influence of TrkB-mediated signaling on tangential cell migration in the MZ (Polleux et al., 2002).
While some extracortical origin of MZ neurons is evident, it is not fully understood whether a subpopulation of MZ neurons might also originate locally in the cortex and, if so, which subtype that is. Indirect evidence for local control mechanisms is the relevance of transcription factors specifically expressed in the dorsal telencephalon for the development of reelin-positive cells. The transcription factors Tbr1 and Emx2 are both strongly expressed in the cerebral cortex (including MZ neurons), but they are not or only weakly expressed in the ventral telencephalon (Briata et al., 1996; Gulisano et al., 1996; Hevner et al., 2001). The T-box transcription factor Tbr-1 is expressed in neurons of the cortical MZ, binds together with the guanylate kinase CASK to the reelin promoter region and regulates its expression in cultured neurons (Hsueh et al., 2000). Together with the lack of reelin-positive cells in the cortex of Tbr1 mutant mice (Hevner et al., 2001), these results imply the involvement of Tbr1 in the direct regulation of reelin expression. In addition to Tbr1, reelin-positive cells in the cortical MZ also express Emx2. In the Emx2 mutant cortex, reelin-positive cells first differentiate, but disappear later (Mallamaci et al., 2000a) and in Emx1/2 double-mutant mice reelin-expressing neurons never develop (Shinozaki et al., 2002). Taken together, this evidence suggests essential roles of these transcription factors in the specification of this important neuronal subtype.
Several lines of evidence implicate the transcription factor Pax6 is opposing Emx2 in patterning the dorsal telencephalon and specification of its area subdivision (Bishop et al., 2000; Mallamaci et al., 2000b; Muzio et al., 2002a,b). Furthermore, tangential cell migration from ventral into dorsal telencephalon is enhanced in Pax6 mutant mice (Chapouton et al., 1999), while it is decreased in Emx2 mouse mutants (Shinozaki et al., 2002). We therefore examined here whether Pax6 is also involved in the specification of reelin-positive cells in the cortical MZ, potentially in a manner opposite to the positive role of Emx2. Indeed, we observed a prominent increase specifically in the number of reelin- and calretinin-, but not calbindin-immunoreactive cells in the MZ of homozygous Pax6 mutant mice. We could further determine that the increase in the number of reelin-positive cells seems to involve non-cell-autonomous mechanisms, since Pax6 mutant cortex isolated in vitro at E12.5 prior to the onset of reelin immunoreactivity develops similar numbers of reelin-positive cells as wild-type cortex. Taken together with evidence for an increased ventral to dorsal cell migration, these data suggest that the increased number of reelin-positive cells in the Pax6 mutant cortex is at least partially due to enhanced cell migration from an extracortical origin.
Materials and Methods
The Pax6 mutant mouse used in this work carries the Sey allele on a C56BL6/6J × DBA/2J background. This naturally occurring point mutation in the Pax6 gene leads to the expression of a truncated non-functional protein (Stoykova et al., 1996, 2000). Heterozygous Sey/+ mice, recognized by their eye phenotype, were crossed to obtain homozygous, heterozygous and wild-type embryos. In this study we used only wild-type and homozygous Sey/Sey littermates recognized by the lack of eyes (Hill et al., 1991). The day of vaginal plug was considered as embryonic day 0.5 (E0.5).
Pregnant animals were killed at increasing CO2 concentrations and by cervical dislocation. Embryos were removed by hysterectomy and transferred to Hank’s buffered salt solution (HBSS) (Gibco) with 10 mM HEPES (Gibco). Embryonic brains were removed and fixed for 6 h in 2% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C, washed in PBS and embedded in 3% agarose in PBS. Vibratome sections were cut frontally at 100 μm thickness and processed for immunohistochemistry. As primary antibodies, we used the monoclonal mouse antibodies (mAb) CR50 (1:10) (a kind gift of X. Ogawa) and E4 (1:100) (a kind gift of A. Goffinet) and polyclonal rabbit antibodies directed against calretinin or calbindin (1:500) (Swant), Emx2 (1:4000) (Suda et al., 2001), Tbr1 (1:500) (a kind gift of M. Sheng) or Pax6 (1:500) (Babco). The sections were incubated free floating in primary antibody solution containing 0.5% Triton X-100 and 10% normal goat serum (Boehringer Ingelheim Vector Laboratories) overnight at 4°C. After several washes in PBS or Tris-buffered saline the sections were incubated in the secondary antibody for 45 min at room temperature. Secondary subclass-specific FITC- or TRIC-coupled antisera were used at a dilution of 1:50 (EuroPath Ltd and Boehringer Ingelheim Vector Laboratories) and Cy2- or Cy3-coupled antisera (Dianova Immundiagnostics) at 1:100. After three further washes, the sections were mounted in Aqua Poly/Mount (Polysciences), a glycerol-based mounting medium.
5′-Bromo-2′-deoxyuridine (BrdU) labeling
Mice were injected i.p. with 5–14 mg BrdU in PBS per 100 g body wt and killed 7 days later. Vibratome sections were then prepared as described above. For the immunodetection of BrdU, pretreatment with 2 N HCl for 30 min was required to denature double-stranded DNA. This was followed by two washes with 0.1 M sodium tetraborate buffer (pH 8.5) for 15 min. After three further washes in PBS, staining with the mAb anti-BrdU (1:10) (IgG1; BioScience Products, Switzerland) was performed as described above.
Dissociated cell cultures were prepared and maintained as described previously (Götz et al., 1998). Briefly, cortices were dissected from embryos at embryonic day 12.5 in ice-cold HBSS containing 10 mM HEPES (all from Gibco). After trypsinization for 15 min, cells were mechanically dissociated using a fire-polished Pasteur pipette and washed twice in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum (FCS) and antibiotics (all from Gibco). Cells were plated on poly-d-lysine-coated glass coverslips at 5 × 105/well in 0.5 ml FCS-containing medium in a 24-well plate (Nunc). Next day, 0.5 ml of chemically defined medium (SATO) (see for example Götz et al., 1998) was added and further medium changes each second day were all performed with SATO medium. Cultures were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature after different times in vitro.
Sections were analyzed using a fluorescence microscope (Zeiss Axiophot) or a two-channel confocal fluorescence laser microscope (TCS 4D Leica) at ×10–40 magnification. Single optical section images (thickness 0.5–10 μm) and maximum intensity images (thickness 10–100 μm) were derived.
Quantification of reelin-, calretinin-, calbindin- and BrdU-positive cells was performed by counting the immunoreactive cells in the MZ of one cortical hemisphere in each section from wild-type and Sey/Sey littermates (∼5–10 sections analyzed). Using a ×40 objective, cells in the MZ were analyzed by progressing through adjacent fields of view covering the entire distance of the cortical MZ from the sulcus adjacent to the ganglionic eminence to the sulcus delineating the medial cortex and hippocampal anlage. The same procedure was used to quantify BrdU-labeled cells in the MZ and SP.
Increase of the Marginal Zone and Decrease of the Subplate in the Pax6 Mutant Cortex
In our attempt to characterize the function of Pax6 in the developing cortex, we examined the cells of the MZ in the Pax6 mutant Small eye (Sey). Figure 1A,B depicts an overview of reelin- and calretinin-immunoreactive cells in the lateral telencephalon of wild-type and Sey/Sey littermates at embryonic day (E) 14.5. As described previously for the wild-type telencephalon, reelin-positive cells detected by the CR50 antiserum (Ogawa et al., 1995) are restricted to the MZ in the developing cortex and many of them co-localize with the Ca-binding protein calretinin (Ringstedt et al., 1998). Consistent with previous expression analyses (Alcántara et al., 1998), we also detected reelin-immunoreactive cells accumulating in the differentiating caudate putamen nucleus (see arrow in Fig. 1A) and the olfactory tubercle. These reelin-positive cells do not co-express calretinin, in contrast to most of the reelin-positive cells in the cortex (Fig. 1A). In contrast, in the Pax6 mutant telencephalon reelin-positive, calretinin-negative cells in the caudate putamen were not detectable (broken arrow in Fig. 1B), while the population of reelin- and calretinin-double-labeled cells in the MZ appeared to be enlarged (Fig. 1B). Consistent with previous results (Kawano et al., 1999; Jones et al., 2002), the calretinin-immunopositive thalamo-cortical fibers (filled arrowhead in Fig. 1A) were absent in the Sey/Sey mutant cortex (empty arrowhead in Fig. 1B), while an abnormal vesicle-like structure filled with reelin-positive cells is apparent in a ventrolateral position, most likely a vestigial olfactory bulb (asterisk in Fig. 1B) (Stoykova et al., 1996, 1997, 2000; Jiménez et al., 2000).
Closer inspection at higher magnifications confirmed a prominent increase in reelin- and calretinin-immunoreactive cells in the MZ of Sey/Sey mutant cortex compared to wild-type littermates. As shown in Figure 2A,B, the band of reelin- and calretinin-immunoreactive cells in the E14.5 Sey/Sey cortex is almost double the size of that in wild-type cortex. This phenotype was more pronounced in rostral sections, consistent with the more severe phenotype of Pax6 mutant cortex in the rostral domain, where Pax6 is normally most strongly expressed (Bishop et al., 2000; Muzio et al., 2002a). In the wild-type cortex, the subplate zone formed a distinct band of calretinin-positive, but reelin-negative cells in the wild-type cortex (arrows in Fig. 2A), while very few such cells were present in the Pax6 mutant littermate cortex at corresponding positions (arrow in Fig. 2B). This staining pattern was reminiscent of an incomplete split of the preplate into MZ and SP in the Sey/Sey cortex.
In order to evaluate whether the increase in MZ cells in the Pax6 mutant cortex is simply due to a delay in the split of the preplate [for the delay of cortical development in Sey/Sey CNS see also (Schmahl et al., 1993; Stoykova et al., 1996; Sun et al., 1998)], we examined reelin and calretinin immunoreactivity in wild-type and Sey/Sey littermates at E17.5. As depicted in Figure 2C–F, the band of reelin- and calretinin-positive cells in the MZ of Sey/Sey mice had further increased in width and was now more than two to three times broader than in wild-type cortex. This difference was observed in both medial and lateral cortical areas (Fig. 2C,D and E,F). At this stage, all reelin-positive cells in the MZ also contained calretinin, in both wild-type and Sey/Sey cortex. As described previously (Alcántara et al., 1998), a second band of reelin-immunoreactive cells had now developed at the position of the SP in wild-type cortex, and some of these cells also co-expressed calretinin (arrows in Fig. 2C,E). It appears as if the calretinin-positive cells seen at this position in the E14.5 cortex have now also acquired reelin immunoreactivity, differentiating into a new reelin- and calretinin-double-positive cell population. Similar to the defects observed at E14.5, the SP cells were still very sparse at E17.5 in the Pax6 mutant cortex at medial and lateral positions (Fig. 2D,F). In lateral regions of the Sey/Sey cortex, however, we noted many reelin-positive, but calretinin-negative cells scattered throughout the cortical plate region (Fig. 2F). These cells might correspond to the Lhx6-positive cells that have been reported to express reelin (Lavdas et al., 1999) and we found enhanced in the Pax6 mutant telencephalon, as described below. Taken together, these data show that neither the increase in reelin- and calretinin-immunoreactive cells in the MZ nor the sparse appearance of SP cells were transient in the Sey/Sey cortex. Both phenotypes not only persisted, but also worsened during cortical development.
To further examine the split of preplate cells in the Pax6 mutant cortex, we performed cell division studies by injection of the DNA base analog BrdU at E11.5, the time when neurons of the SP are generated in mouse telencephalon (Bayer and Altmann, 1990; Wood et al., 1992). While precursor cells that continue to divide dilute the label, neurons that became post-mitotic soon after the pulse maintain their label and can be detected at later stages. When BrdU-labeled cells were analyzed in E18 cortex, they were detected in the MZ and SP of the wild-type cortex as expected (Table 1). Moreover, ∼80% of the BrdU-labeled cells in the MZ were double labeled with calretinin antiserum (Table 1), indicating that this is the time at which most calretinin-/reelin-positive cells arise in the MZ. While the overall pattern of BrdU-labeled cells was similar, after quantification three main differences between labeled cells in the cortex of wild-type and Sey/Sey littermates became apparent (Table 1). First, the total number of BrdU-positive cells was reduced in the SP and MZ of the Pax6 mutant cortex compared to wild-type (a total of 53 cells/section/hemisphere in wild-type, 41 SP + 12 MZ; 42 cells in Sey/Sey littermates, 26 SP + 16 MZ). These data are consistent with the reduced neurogenesis in the absence of functional Pax6 (Heins et al., 2002). Second, the proportion of BrdU-positive cells residing at the SP position was reduced in Sey/Sey mice by 40% compared with wild-type cortex (Table 1). Conversely, the number of BrdU-positive cells in the MZ of Sey/Sey cortex was slightly increased compared with the wild-type cortex (Table 1). These data further support the interpretation that the formation of the SP is impaired in the Pax6 mutant cortex, apparently due to an impairment of the distribution of preplate layer (PPL) cells between the MZ and SP. Some SP characteristics, however, also develop in the absence of Pax6 function, such as the specific localization of chondroitin sulfate proteoglycans at this position (data not shown), and the SP is recognizable at later stages at posterior levels of the Pax6 mutant cortex (Schmahl et al., 1993).
Subtype-specific Effects on MZ Cells in the Pax6 Mutant Cortex
Since the MZ also contains a prominent population of calbindin-positive cells (Soria and Fairén, 1999), we next examined whether this population is also affected by the absence of functional Pax6 in cortical development. As depicted in Figure 3A,B, most calbindin-immunoreactive cells of wild-type cortex at E14.5 did not contain detectable levels of reelin protein, suggesting that the calbindin-positive, reelin-negative and calretinin-positive, reelin-positive cells are two distinct populations in the cortical MZ (Huntley and Jones, 1990). In the cortex of Sey/Sey mutant littermates, calbindin also hardly co-localized with reelin and the band of calbindin-positive cells in the MZ of Sey/Sey littermates closely resembled its counterpart in wild-type cortex (Fig. 3A,C). A second stream of tangentially oriented calbindin-positive cells was detected in the intermediate zone of wild-type cortex that was previously suggested to originate in the MGE (Chapouton et al., 1999; Anderson et al., 2001). Surprisingly, this stream appears strongly reduced in the Pax6 mutant compared to wild-type (arrows in Fig. 3A,C).
To ensure the specific effect seen only on calretinin-positive, reelin-positive, but not on calbindin-positive cells in the Sey/Sey cortex, we quantified the number of differently labeled cells in the cortical MZ of wild-type and Pax6 mutant littermates from E13.5 to E15.5 (see Materials and Methods for details; Fig. 3D,E). The number of calbindin-positive cells in the cortical MZ was very similar between wild-type and Pax6 mutant littermates at all of the stages analysed (Fig. 3F). In contrast, the number of calretinin-positive cells in the MZ of the Sey/Sey cortex was already slightly increased at E13.5 (120% calretinin-positive cells in Pax6 mutants normalized to wild-type) and this effect was further augmented during development (154% at E14.5, 179% at E15.5; Fig. 3D). At E13.5 reelin immunoreactivity was too weak to identify individual cells with certainty. At E14.5, however, the number of reelin-positive cells was already prominently increased in the MZ of Sey/Sey cortex as compared to wild-type cortex (172%; Fig. 3E). Note that the number of reelin-positive cells in the wild-type and Pax6 mutant is slightly lower than the number of calretinin-positive cells at E14.5, but not at E15. Indeed, in agreement with Schwartz et al. (Schwartz et al., 1998), we found in calretinin and reelin co-localization analyses that ∼80% of the calretinin-positive cells are also reelin-immunoreactive at E14.5, while this value increases to 100% at E15.5. Thus, the differentiation of reelin immunoreactivity and its co-localization with calretinin is not affected by the absence of Pax6, but the number of reelin- and calretinin-double-positive cells increases progressively in the Pax6 mutant cortex. These results therefore show a specific increase of only one neuronal subpopulation in the MZ of the Pax6 mutant cortex, the reelin- and calretinin positive cells, rather than a general misspecification of the earliest generated neurons residing in the MZ.
MZ cells have also been described as expressing Emx2, Tbr1 and Lhx6 (Mallamaci et al., 1998; Lavdas et al., 1999; Mallamaci et al., 2000a). Consistent with the above analysis we observed an increase in cells containing these transcription factors in the MZ of Pax6 mutant cortex (Figs 4 and 5). While most of the MZ cells at E14.5 in the cerebral cortex contain Emx2 protein (Fig. 4B,C), we also detected a small subset of MZ cells (∼30% at E14.5 and a weak signal still persisting in a subset of MZ cells at E17.5) that were Pax6 immunoreactive (arrows in Fig. 4A). Many more MZ cells, however, were Emx2 immunoreactive and these cells were increased in both the medial and lateral regions of Pax6 mutant cortex compared with wild-type (Fig. 4B–E). Previous evidence indicated that the expression of Emx2 is enhanced and expanded, thus abolishing the normal medial to lateral expression gradient of Emx2 in the absence of functional Pax6 (Bishop et al., 2000; Mallamaci et al., 2000b; Muzio et al., 2002a,b). Our results further support these data suggesting that Emx2 and Pax6 oppose each other genetically not only in the cortical plate, but also in the cortical MZ.
The transcription factor Tbr1 has previously been suggested to directly regulate reelin expression. Consistently, almost all reelin-positive cells in the MZ are Tbr1 immunoreactive in wild-type and Sey/Sey cortex (Fig. 5A,B). In addition, we found that Lhx6-expressing cells were increased in the MZ of the Sey/Sey cortex compared with wild-type (Fig. 5C–F; see also Stoykova et al., 2000). As described previously (Lavdas et al., 1999; Stoykova et al., 2000), Lhx6-expressing cells appear to migrate from the MGE into the cortex in two separate streams, one just underneath the pia and the second between the intermediate zone (IZ) and the subventricular zone (Fig. 5E). Interestingly, while the band of Lhx6-expressing cells in the MZ is increased, the stream of cells in the IZ was not detectable in the Pax6 mutant telencephalon, a phenotype which is most clearly seen at rostral levels (Fig. 5C,D), suggesting that migrating Lhx6-positive cells might combine underneath the pial surface instead of splitting into two pathways when they enter the cortex. Notably, Lhx6 signal was seen in cells of the MZ in wild-type telencephalon at the border between the olfactory bulb and the rostral-most telencephalon from where they seem to spread dorsolaterally. In addition, we also found that the caudal ganglionic eminence and the MZ above it contained many more Lhx6-positive cells in the mutant as compared with wild-type cortex (data not shown). As Lhx6-positive cells have been shown to contain reelin (Lavdas et al., 1999), these data are further consistent with the idea that migration of the enlarged population of Lhx6-positive cells originating at ventral positions might contribute to the increase in the number of reelin-positive cells in the Pax6 mutant MZ.
No Cell-autonomous Function of Pax6 in the Increase in Reelin-positive Cells in the Cortex
The results presented above would be consistent with both, cell-autonomous as well as non-cell-autonomous mechanisms contributing to the increase in reelin-positive cells in the Pax6 mutant MZ. An increase in the number of ventrodorsally migrating cells would be consistent with a continuous increase in the number of reelin-positive cells in the MZ of the Pax6 mutant cortex and act as a non-cell-autonomous mechanism. Conversely, the lack of Pax6 in a subpopulation of MZ cells could lead to the misspecification of these cells by expansion of an Emx2-positive MZ cell fate (Muzio et al., 2002a) acting as a cell-autonomous mechanism. To test the possibility that the lack of Pax6 function might have a direct effect on the misspecification of reelin-positive cells, we isolated cortical cells prior to the appearance of reelin immunoreactivity at E12.5 in vitro.
A subpopulation of neurons acquired reelin immunoreactivity in both cultures derived from wild-type and Sey/Sey littermates (Fig. 6A–D). As in vivo, reelin was co-localized with calretinin such that almost all reelin-immunoreactive cells also contained calretinin (85%, n = 200), while 29% of the calretinin-positive cells contained reelin (n = 200). The proportion of reelin-positive cells amongst the calretinin-positive population is lower in vitro compared with the proportion in the MZ in vivo. This is most likely due to the loss of layer information in vitro and hence calretinin-positive cells from other positions (non-MZ) in the cortex were included in the quantification in the dissociated cell cultures. In contrast to the high proportion of reelin-positive cells that were calretinin-positive in vitro and in vivo (85%), only a few of the reelin-positive cells contained calbindin in vitro (18%, n = 100), in close agreement with the in vivo analysis (Fig. 3A,B).
As in the sections, no difference in the proportion of cells that contain both calretinin and reelin was observed between wild-type and Sey/Sey cortex in vitro (wild-type 29%; Sey/Sey 26%), but in contrast to the in vivo results, no difference in the number of reelin-immunoreactive cells was detectable in wild-type and Sey/Sey cultures. These data demonstrate that the source of the increase is absent in this culture system
We also performed a BrdU labeling analysis to examine which cells were generated in vitro when BrdU was continuously added to the culture medium. Hardly any BrdU/reelin-double-positive neurons could be detected. This could be due to either their generation outside the cortex or their generation prior to E12.5. However, some calretinin- and calbindin-immunoreactive cells were generated in vitro [27% of calretinin-positive cells were BrdU-positive after 7 days in vitro (n = 150) and 33% of calbindin-positive cells were BrdU-positive (n = 150)] and their number was also not increased in cultures isolated from Pax6 mutant cortices [20% of calretinin-positive cells were BrdU-positive (n = 100); 27% of calbindin-positive cells were BrdU-positive (n = 90)]. These data further support the notion that there is no difference in the generation of calretinin-immunoreactive cells from the cortical neuroepithelium itself in the absence of functional Pax6.
An alternative explanation could be, however, that the misspecification of calretinin- and reelin-positive cells cannot occur under the conditions in vitro, but only in the in vivo environment. We therefore isolated cells from the entire dorsal and ventral telencephalon and cultured them in vitro. A clear increase in the number of calretinin- and reelin-positive cells was apparent in cultures from Sey/Sey compared with wild-type telencephalon (Fig. 6E). Taken together, these experiments could not reveal any cell-autonomous contribution to the increase in reelin-positive cells in the cortex of Pax6 mutant mice.
We report in this study on a new phenotype of the mutant Pax6/Small eye brain. We found that during development, the MZ in the Sey/Sey cortex is hypercellular and contains a considerably increased number of cells expressing reelin, an extracellular molecule with a profound effect on cortex formation. While considerable progress has been made on the characterization of signaling pathways initiated by reelin, much less is known about the specification of reelin-synthesizing cells, despite their important function during development. Indeed, this is to our knowledge the first mouse mutant with an increase in reelin-positive MZ cells. Importantly, the analysis of this mutant phenotype showed that distinct mechanisms are involved in the specification of neuronal subtypes composing the cortical MZ, as only the calretinin- and reelin-double-positive cells were affected in the Pax6 mutant cortex, while the number of calbindin-positive cells was not altered. We could further implicate mainly non-cell-autonomous mechanisms depending on Pax6 function for this phenotype, as cells isolated from the Pax6 mutant cortex in vitro prior to the onset of reelin immunoreactivity did not acquire this phenotype. We therefore suggest that enhanced tangential migration from extracortical sources results from the patterning defects in the Pax6 mutant telencephalon and apparently leads to the progressive increase in reelin-positive cells in the cortical MZ.
Heterogeneity of Preplate Cells
The MZ neurons (the future molecular layer of the cortex) are generated as a component of the earliest post-mitotic cells of the PPL. It is widely accepted that the younger neurons of the cortical plate accumulate within the preplate neurons, thus splitting the PPL into a superficial layer I (MZ) and a deep subplate layer with polymorph cells (Raedler and Sievers, 1976; Marin-Padilla, 1978). A variety of types of early neurons have been identified in the PPL of rodents and humans and there is currently a debate on which cells express reelin, partially due to species differences, even between mouse and rat (see Introduction) (Meyer et al., 1999). Here we show that in the developing mouse cortex a subpopulation of MZ cells synthesizes reelin and calretinin and this population constitutes the majority of reelin-positive cells in the cortex during development (80%). In contrast, in the mouse, only a minor population of reelin-positive cells contains calbindin (18%). From our morphological and quantitative analysis we would suggest that these two populations are largely non-overlapping and hence represent two distinct populations of reelin-positive neurons in the MZ of the mouse cortex.
We could further demonstrate that these distinct populations are specified independently. In our analysis of the Pax6 mutant cortex only the calretinin-positive, reelin-positive cells are influenced by the lack of Pax6 while the calbindin-positive cells develop normally even in the absence of functional Pax6. Indeed, we have recently shown that Pax6 is essential only for a subtype of cortical plate neurons (Heins et al., 2002) (for further discussion see below). Analysis of additional marker molecules detected in subtypes of MZ neurons revealed an increase in Emx2- and Tbr1-positive cells. These data further support the previous concept that the expression of reelin in MZ cells depends on Tbr1 gene activity (Hsueh et al., 2000; Hevner et al., 2001). Our data are also consistent with the previous suggestion that Emx1 and Emx2 function is required for the maintenance and possibly specification of many, if not all, reelin-positive cells in the MZ (Mallamaci et al., 2000a; Shinozaki et al., 2002). Emx2 immunoreactivity is detected in almost all reelin-positive cells in wild-type and Sey/Sey cortex, further supporting a positive regulatory influence of Emx transcription factors on the expression of reelin in MZ neurons. In contrast to the situation seen in the Pax6 mutant cortex with an increase in reelin-positive cells, reelin-positive cells fail to develop in Emx1/2 double-mutants, possibly due to an up-regulation of Pax6 in these mice, as Pax6 and Emx2 cross-regulate each other in the developing cortex (Muzio et al., 2002a,b). Thus, Pax6 seems to exert a negative while Emx2 exerts a positive regulatory effect on the number of MZ cells that express reelin in the developing cortex. Since these transcription factors are mostly expressed in the dorsal region of the telencephalon, one might assume that their presence in MZ neurons indicates a dorsal origin of these cells. However, while there is still no direct evidence that this is the case, it has been shown that at least a subpopulation of MZ cells originates ventrally (Lavdas et al., 1999). These data therefore leave open several possibilities: either the transcription factors Tbr1 and Emx1/2 act at late differentiation stages and become up-regulated only when reelin-positive cells have entered the cortex or they label a subpopulation of reelin-positive MZ cells that are born dorsally. Our data would argue in favor of the first possibility, since Tbr1 is contained in all and Emx2 in most reelin-positive cells in our double stainings. A third possibility is that Emx2 and Pax6 genetically interact in an extracortical, but dorsal, region, such as the olfactory bulb anlage. Since we detected the Pax6 protein in a subpopulation of reelin-positive cells in the MZ, this is supposedly a distinct population from the reelin-positive, Emx2-positive cells in the MZ. In the Pax6 mutant cortex, the number of Pax6-negative, but Emx2- and reelin-positive neurons is increased. It is therefore possible that Pax6 might be involved in the specification of a small population of the MZ cells through a repressive effect on Emx2 activity, as suggested previously for the cortical plate (Muzio et al., 2002a). If indeed the specification of a sub-population of reelin-positive cells in the MZ depends on genetic interactions between Pax6 and Emx2, this effect might be achieved either within the MZ itself or, more probably in our view, in some extracortical regions, for instance the olfactory bulb-like structure (OBLS) in the mutant brain (Jiménez et al., 2000), where these genes are also expressed. This idea would also be consistent with our in vitro data, where the whole telencephalon preparation would include the remnant olfactory bulb (OBLS) in the Pax6 mutant telencephalon (for further discussion see below). Clearly, further experiments are required to elucidate precisely the origin of reelin-positive cells by a fate map analysis of the developing telecephalon.
Impaired Split of the Preplate in Pax6 Mutant Cortex
One surprising result of this analysis was the defect observed in the split of the preplate cells, despite the excess of reelin in the MZ. Calretinin and reelin immunostaining as well as BrdU cell division studies showed that the PPL does not split correctly in the Sey/Sey cortex and the number of SP cells generated at E11.5 is reduced in the Sey/Sey compared with wild-type cortex. The expression levels of Dab1 have been shown to correlate inversely with reelin signaling, with increased expression levels in reeler mutant mice and decreased levels in transgenic mice over-expressing reelin (Magdaleno et al., 2002). In this regard, it is worth mentioning our preliminary observations that Dab1 expression seems decreased in the ventricular zone of the Sey/Sey cortex, while it is comparable with wild-type levels in the cortical plate (data not shown). These results indicate rather an increase than a decrease in reelin signaling, but further studies are needed to elucidate reelin signaling to Pax6 mutant cells. Notably, the lower number of SP cells is not a transient effect, since characteristic calretinin-positive cells at the SP position are still reduced in the Pax6 mutant cortex at E17.5. Consistent with an impaired SP formation in the Pax6 mutant cortex, the neurocan immunoreactivity in the SP in the Small eye rat is much fainter as compared with the immureactivity detected in the MZ (Kawano et al., 1999). However, it is important to note that we detected no increase in calretinin-positive, reelin-negative cells in the MZ of Sey/Sey cortex, as would be expected from the incorporation of properly specified SP neurons into the MZ. It is therefore possible that SP cells are simply not specified in the Pax6 mutant cortex.
Tangential Migration of Reelin-positive Cells
Several extracortical sources for distinct populations of reelin-positive MZ neurons have been suggested, one from the MGE (Lavdas et al., 1999) and another from the retrobulbar germinal neuroepithelium at the junction between the olfactory vesicle and the most anterior telencephalon (Meyer et al., 1998), where later the anterior olfactory nucleus develops (Gadisseux et al., 1992). Our observations suggest that both sources as well as the aberrant OBLS might all contribute to the increase in reelin-positive cells in the cortex MZ in Sey/Sey mice. We have previously shown an increase in the Lhx6 expression domain in the septum and the retrobulbar neuroepithelium of the Sey/Sey mutant telencephalon, from where more Lhx6-positive cells, as compared to the wild-type brain, appear to populate the subpial space (Stoykova et al., 2000). In addition, here we observed an increase in the Lhx6-positive stream of cells, possibly originating from the caudal ganglionic eminence (CGE), invading the cortical MZ posteriorly. Thus, misrouted Lhx6- and reelin-positive cells seem to invade the MZ from the retrobulbar region, septum, MGE and CGE and might thereby contribute to the increase in these cells in the Pax6 mutant cortex. A previous report suggested that the Lhx6- and reelin-positive cells in the cortical MZ do not contain calretinin (Lavdas et al., 1999). In the Pax6 mutant cortex, however, the number of reelin-positive cells that also contain calretinin was increased, while we could not detect an increase in reelin-positive, but calretinin-negative cells. It is therefore conceivable that the Lhx6-positive cells could be misspecified and start to express calretinin in the Pax6-deficient telencephalon. Alternatively, we cannot rule out the possibility that the Lhx6-positive cells are a population distinct from the reelin-positive cells that we find to be increased in the Sey/Sey cortical MZ.
A good candidate for an alternative or potentially additional source of subpially migrating reelin-positive cells is the aberrant olfactory bulb, a vesicle-like structure forming at early stages in the basolateral telencephalon of Pax6 mutant telencephalon (designated the OBLS, or olfactory bulb-like structure, by Jiménez et al. (Jiménez et al., 2000)] (Schmahl et al., 1993; Stoykova et al., 1996, 1997). The cells in the abnormal vesicle abundantly express reelin and R-cadherin (Stoykova et al., 1997, 2000), both of which are normally expressed in the mitral cells of the olfactory bulb. The OBLS further contains many reelin- and calretinin-immunoreactive cells that persist abnormally in the Pax6 mutant telencephalon and extend in a band underneath the pial surface (Jiménez et al., 2000) (Fig. 1B). This abnormal OBLS would be included in our cultures derived from whole telencephalon, but not in those where we isolated cortex only. This structure might therefore contribute to the increase in reelin-positive, calretinin-positive cells in the Pax6 mutant whole telencephalon cultures. Taken together, subpial migration of reelin- and calretinin-positive cells from the OBLS might well be the most important contribution to the increase in these cells in the cortical MZ of the Sey/Sey telencephalon in vivo and in vitro.
Continued migration of reelin-positive cells into the MZ from several potential sources would also be consistent with the progressive increase in this phenotype during development. An obvious candidate molecule to affect patterning of the Small eye telencephalon is Sonic Hedgehog, which is expressed in a broader domain in the Pax6-deficient telencephalon compared with its normal expression pattern (Stoykova et al., 2000). Indeed, Sonic Hedgehog might directly regulate the expansion of genes expressed in the MGE, such as Nkx and Lhx6. Thus, patterning changes in several extracortical regions might contribute to misspecify reelin-positive, calretinin-positive cells in the Pax6 mutant telencephalon. These data also imply that the calbindin-positive cells are specified at a different position and/or are not affected by Sonic Hedgehog or other patterning defects in the Pax6 mutant cortex. Taken together, our results demonstrate not only distinct molecular mechanisms specifying distinct neuronal subtypes in the MZ, but also imply telencephalic dorso-ventral patterning mediated by Pax6 in the regulation of the correct number of reelin-positive cells.
|BrdU injection was performed at E11.5 and the number of BrdU-positive cells in the SP and MZ was analyzed at E18.5 (one cortical hemisphere).|
|BrdU+ cells||41 ± 3.4||26 ± 2.5||12 ± 2||16 ± 1.9|
|Calretinin+BrdU+ cells (%)||15 ± 1.8||7 ± 3.5||79 ± 5||66 ± 7.4|
|BrdU injection was performed at E11.5 and the number of BrdU-positive cells in the SP and MZ was analyzed at E18.5 (one cortical hemisphere).|
|BrdU+ cells||41 ± 3.4||26 ± 2.5||12 ± 2||16 ± 1.9|
|Calretinin+BrdU+ cells (%)||15 ± 1.8||7 ± 3.5||79 ± 5||66 ± 7.4|
We are very grateful to Silke Eckert and Mücella Öcalan for excellent technical assistence and to Nicole Haubst for helpful comments on the manuscript. This work was supported by grants from the EU (QLG3-CT-2000-00158 to A.S. and P.G.; QLK3-CT-1999-00894 and QLG3-CT-2000-01471 to M.G.), the Human Frontiers Science Program (RG160/2000B to M.G.), the German Research Foundation (Go 640/3-2 to M.G.) and the Max-Planck Society.