Loss of Emx2 function leads to ectopic expression of Wnt1 in the developing telencephalon and cortical dysplasia

During development of the elegantly structured mammalian cerebral cortex,generation and subsequent migration of specialized neurons and glia within the brain follows a characteristic sequence. The earliest differentiated neurons emigrate from the telencephalic ventricular zone to form a superficial primordial layer called the preplate(Allendoerfer and Shatz, 1994). In mice, the marginal zone (MZ, layer I) and subplate (SP) derive from splitting of the preplate at ∼11-12.5 days postcoitum (dpc) as successive waves of neurons from the underlying ventricular zone undergo radial migration into the cortical plate proper (layers II-VI). The MZ and SP comprise a heterogeneous population of cells with neuronal properties, including reelin-expressing Cajal-Retzius (CR) cells, subpial granule cells(glutamate-positive) and subpallial-derived interneurons (GABA-positive)(Fairen et al., 2002; Meyer et al., 1998; Parnavelas, 2000; Zecevic et al., 1999). Important roles for MZ and SP cells during tangential and radial migration of cortical neurons have been previously described(Parnavelas, 2000; Xie et al., 2002). For example, proper formation of the MZ and SP has been shown to be necessary for tangential emigration of GABA-expressing interneurons from ventral forebrain regions (e.g. the ganglionic eminence) into the cortical plate(Lavdas et al., 1999). Second,cells in the MZ provide repulsive or stop cues to radially migrating cortical neurons (Rice and Curran,1999). Indeed, the MZ includes the foot processes of radial glia that form the subpial glia limitans, a structure that ultimately limits the radial migration of neural cells and defines the external boundary of the central nervous system (CNS) (Super et al., 2000).

The term cortical dysplasia is used to describe a heterogeneous group of CNS malformations in which the organization of the cortical plate is disrupted, typically owing to abnormal cellular migration(Friede, 1989; Gleeson and Walsh, 2000). Marginal zone heterotopias (MZH) or leptomeningeal glioneuronal heterotopias(LGH) are one form of dysplasia in which ectopic nests of glial and neuronal cells are observed in the cortical MZ or overlying leptomeninges, respectively(Greenfield et al., 2002). These lesions are frequently seen in association with other features of dysplasia as well as other CNS malformations and congenital syndromes such as Trisomy 13 or type II lissencephalies (Walker-Warburg syndrome)(Friede, 1989; Greenfield et al., 2002; Hirano et al., 1992). They are among the most common neuropathological malformations found in human brains with cortical dysplasia (Mischel et al.,1995). Although their clinical significance is not fully understood, they are presumed foci of seizure activity because of their cortical location, neuronal composition and high incidence in individuals with epilepsy (Mischel et al.,1995).

Insights into the genetic and molecular causes of cortical dysplasia are beginning to emerge from general investigation of cortical development in mice and humans (Gleeson and Walsh,2000). In particular, certain transcription factors have been described with important roles in MZ and SP development. Mutation of Lmx1a, a LIM-homeodomain factor that is expressed in the dorsomedial telencephalon, results in diffuse and focal abnormalities of the MZ(Costa et al., 2001). Second,function of the homeodomain protein EMX2 is required for several aspects of cortical development including regulation of neural precursor cell proliferation (Galli et al.,2002; Heins et al.,2001; Tole et al.,2000) and lamination of the cortical plate(Galli et al., 2002; Heins et al., 2001; Mallamaci et al., 2000a; Tole et al., 2000), as well as formation of preplate derivatives. Loss-of Emx2 function or Emx1/2 function is associated with a decrease or absence,respectively, of Cajal-Retzius and SP cells within regions of the cortex(Bishop et al., 2003; Mallamaci et al., 2000a; Shinozaki et al., 2002). Additionally, defects in tangential neuronal migration and axon pathfinding from the ventral forebrain to the cortex of Emx-null animals have been described (Bishop et al.,2003; Shinozaki et al.,2002).

Despite its pivotal roles during brain development, downstream transcriptional targets of Emx2 have yet to be identified. Emx2 is expressed in the cortical ventricular zone and cortical hem, which is located adjacent to the dorsal midline of the embryonic telencephalon. The cortical hem is generally considered to be an organizing center and source of secreted molecules with important roles in the specification and proliferation of forebrain cell types (Furuta et al.,1997; Grove et al.,1998). Numerous Wnt genes are expressed in overlapping patterns within the cortical hem (Grove et al.,1998), and Wnt3a has been shown to be essential for proliferation and expansion of hippocampal precursor cells in this region(Lee et al., 2000). Although expression of Wnt1 is observed throughout most of the dorsal midline of the CNS it is excluded from the telencephalon and cortical hem. We have previously proposed that this aspect of Wnt1 regulation relies on transcriptional repression, because deletion of certain Wnt1cis-acting regulatory sequences results in ectopic reporter gene expression in the telencephalon of transgenic mice(Rowitch et al., 1997). Mutation of a single homeodomain core-binding site (HBS1) within the 5.5 kb Wnt1 enhancer results in ectopic dorsomedial telencephalic reporter gene expression (Iler et al.,1995; Rowitch et al.,1998) in a pattern reminiscent of Emx2(Boncinelli et al., 1993). EMX2 as well other homeodomain proteins have been shown to bind HBS1 in vitro(Iler et al., 1995), and in a previous analysis of Emx2-null embryos, weak ectopic expression of Wnt1 in the 10.5 dpc telencephalon was observed(Yoshida et al., 1997). Taken together, these experiments are consistent with a model in which EMX2 functions to repress Wnt1 expression in the developing telencephalic organizer regions (Rowitch et al.,1997). However, further experiments to demonstrate functional consequences of ectopic Wnt1 expression in the cortex and functional or phenotypic correlation between this and Emx2-null animals are required to assess this putative interaction and establish its possible significance for normal development.

We report the presence of marginal zone and leptomeningeal heterotopias in Emx2^–/– mice. Our data suggest that this lesion and other abnormalities of the superficial cortical plate (layer I) in Emx2^–/– mice result from a failure to repress Wnt1. First, we observed striking ectopic expression of Wnt1in the cortical hem and medial cortical ventricular zone of Emx2^–/– mice but not heterozygous littermates. Second, Wnt1 gain-of-function analysis produced heterotopias identical to those in Emx2^–/– mice. Finally,we observed diffuse abnormalities of the early-born Calretinin- and reelin-positive cells and also the plexiform network of cellular processes derived from these cells within the marginal zone and subplate, similar to those described for Emx-null animals. These findings indicate that EMX2 acts to repress Wnt1 in the developing telencephalon. Furthermore, they suggest that ectopic activation of Wnt1 signaling is sufficient to cause anomalous formation of the marginal zone and subplate,as well as defects in neural migration and heterotopias in Emx2^–/– mice.

Construction of the lacZ reporter (Wnt-lacZ-ΔHBS)or Wnt1 misexpression transgenes (Wnt1-mutHBS1 and Wnt1ΔHBS) was carried out as follows. Full-length lacZor Wnt1 genomic sequences, including 1.25 kb of upstream DNA(promoter) and all Wnt1 exons, introns and poly A sequence(Danielian and McMahon, 1996),were cloned upstream of the 1.1 kb Wnt1 regulatory elements that were modified to have mutation of HBS1 (Iler et al., 1995), or deletion of HBS1 and 300 bp of surrounding sequence(construct #16), (Rowitch et al.,1998). Confirmation of plasmid identity was by restriction endonuclease digestion and DNA sequencing. Prior to pronuclear injection,transgenes were purified from the plasmid backbone after digestion with the restriction endonuclease, SalI (Wnt-lacZ-ΔHBS) or AatII (Wnt1-mutHBS1 and Wnt1ΔHBS).

Four independent founder lines of transgenic mice that expressed the Wnt-lacZ-ΔHBS reporter construct were generated by pronuclear injection of one-cell BL6CBAF1/J embryos as described(Rowitch et al., 1998). Genotyping was by Southern blot with a probe directed against Wnt1(details supplied upon request). Of 12 independent Wnt1ΔHBS transgenic founders generated, one was analyzed at 13.5 dpc, and three at 18.5 dpc-PN1. One stable Wnt1ΔHBS transgenic line was followed and could be maintained in a hemizygous state. Homozygous offspring (n=4)were generated at 12.5 dpc for analysis. This line was subsequently lost because of infection with MHV. Extent Emx2 mutant mice (generously provided by Prof. Peter Gruss, Goettingen, Germany) on a mixed C129Sv/J-C57Bl6 genetic background were generated and genotyped as previously described(Pellegrini et al., 1996). Matings of transgenic mice were timed and embryos harvested such that noon the day of vaginal plug discovery was considered to be 0.5 dpc.

Surgical biopsy and autopsy material containing leptomeningeal glioneuronal heterotopias was identified by review of the case records and archival material in the Neuropathology Division of Children's Hospital, Boston. Four appropriate cases were identified and analyzed by Hematoxylin and Eosin analysis in addition to immunohistochemistry for glial (GFAP) and neuronal markers (NeuN, synaptophysin).

Mouse embryos and brains were fixed in freshly made 4% paraformaldehyde/PBS overnight at 4°C. Tissue for frozen sections was then impregnated in 50%Sucrose/PBS overnight at 4°C, frozen in OCT, and then sectioned at 16μm. Tissue for paraffin wax-embedded sections was dehydrated, embedded and sectioned at 5 μm according to standard protocols. Human neuropathological specimens were fixed in 10% buffered formalin, according to standard clinical protocols, and then embedded in paraffin wax for sectioning. Sections were stained with Hematoxylin and Eosin and photomicrographs were taken digitally using a Zeiss Axioskop and Axiocam imaging system.

Paraffin wax embedded and frozen sections were subjected to heat antigen retrieval at 99°C in 10 mM sodium citrate buffer for 20 minutes for all antibodies. Staining was performed using the HRP/DAB based Envision+ staining system (DAKO) according to the manufacturers specifications with the only modification being increased incubation time for primary antibodies(overnight, 4°C) and secondary antibodies (1 hour, room temperature). Primary antibodies were Calretinin (Zymed #18-0211), GFAP (DAKO #Z0334),Nestin (BD #611658), NeuN (Chemicon #MAB377) and laminin (DAKO). Staining forβ-galactosidase activity in transgenic embryos was performed as previously described (Whiting et al.,1991).

Frozen and paraffin wax-embedded section RNA in situ hybridization was performed as previously described (Lu et al., 2001). Protocol modifications specific to use with paraffin wax-embedded sections included pretreatment of sections with 0.2 N HCl for 10 minutes at room temperature followed by incubation in 2× SSC for 10 minutes at 70°C. Sections were then treated with Proteinase K (50μg/ml) for 10 minutes at 37°C. After hybridization, slides were treated with RNaseA (20 μg/ml) for 15 minutes at 37°C, then washed and developed according to standard protocols. The probes used were for the genes Wnt1 (Parr et al.,1993), Emx2 and Wnt2b(Tole et al., 2000), Bmp4 (Lee et al.,2000), reelin (Alcantara et al., 1998) and Gfap (modified construct from A. Ruiz i Altaba).

Loss of Emx2 function in mice has been shown to result in generalized cortical dysplasia with defects in cortical lamination and hippocampal growth, as well as abnormal formation of the marginal zone (MZ)and subplate (SP) (Boncinelli et al.,1993; Mallamaci et al.,2000a; Mallamaci et al.,2000b; Shinozaki et al.,2002; Yoshida et al.,1997). We performed further neuropathological analysis of the cortex of Emx2^–/– mice at 18.5 dpc, which revealed abnormal features of the marginal zone that had not previously been described (Fig. 1). The MZ of Emx2^–/– mice exhibited diffusely increased cellularity and disorganization with an indistinct boundary between the marginal zone and the developing cortical plate proper (layers II-VI)(Fig. 1D-F). By contrast,wild-type and Emx2 heterozygous mice showed no such defects(Fig. 1A-C). Focal lesions in the Emx2^–/– MZ consisted of heterotopic eruptions of cells from the cortical plate into the overlying marginal zone,the subpial space or even into the overlying leptomeninges(Fig. 1D-F). Serial sectioning of the lesions showed them to consistently have a polypoid shape with a stalk and oftentimes they were associated with a central penetrating blood vessel. In the area of the lesions the disorganization of the migrating neural cells was noted to extend through the full thickness of the cortical plate and into the subplate (Fig. 1E).

The phenotype was highly penetrant, as heterotopias were observed in all Emx2^–/– mice examined (n=6) whereas no lesions were detected in heterozygous littermates (n=3) or wild-type controls. Between two and six lesions were detected in each Emx2^–/– brain on serial sectioning at 18 μm intervals. The lesions ranged from ∼30-200 μm in size. The lesions were preferentially located in the dorsal frontoparietal neocortex. Heterotopias were not found in ventral or medial cortex or within the diencephalon of wild-type or mutant mice. The distribution of the lesions overlapped that of the normal expression domain of Emx2 to a significant degree(Boncinelli et al., 1993).

The distribution and morphological appearance of the marginal zone heterotopias were strikingly similar to human heterotopias we identified in surgical and autopsy tissue from individuals with cortical dysplasia(Fig. 1G-I). The lesions most closely resembled those referred to in humans as leptomeningeal glioneuronal heterotopias (nodular heterotopias or `brain warts')(Friede, 1989; Greenfield et al., 2002). Comparison of the histological features of representative lesions from a 36-week-old infant and lesions from an 18.5 dpc Emx2^–/– mouse, demonstrated the shared findings of frontoparietal regional restriction, extension of cells into the overlying meninges, frequent association with a central penetrating blood vessel and a polypoid shape (Fig. 1F,G) (Ellison,1998; Friede,1989; Greenfield et al.,2002; Hirano et al.,1992; Iida et al.,1994). Serial sectioning of the lesions confirmed that nearly all the lesions involved the subarachnoid space or leptomeninges. The more differentiated appearance of the human lesions was probably due to the older developmental age of the individual relative to the mouse. Both human and mouse lesions contained NeuN-positive neurons as well as GFAP-expressing glia by immunohistochemical and in situ hybridization analysis, confirming shared features within the heterotopic tissue(Fig. 1H-K). We conclude that formation of marginal zone and leptomeningeal glioneuronal heterotopias comprises a novel neuropathological feature in Emx2-null mice, and that such lesions are comparable to human LGH.

The findings above and previous observations indicate critical roles for EMX2 during cortical development and highlight its possible relevance in the understanding of human LGH. In order to investigate mechanisms underlying Emx2 function, we focused on possible transcriptional targets of EMX2 during brain development. As we have previously proposed, one candidate locus for repression by EMX2 is Wnt1(Iler et al., 1995; Rowitch et al., 1997; Rowitch et al., 1998). However, it was unclear whether ectopic expression of Wnt1 occurred in Emx2^–/– embryos at times associated with formation of preplate derivatives (∼11-12.5 dpc)(Monuki and Walsh, 2001). To investigate this, we performed whole-mount in situ hybridization on wild-type and Emx2-null embryos at 12.5 dpc. Wnt1 is normally expressed along the dorsal midline of the CNS with a rostral limit in the rostral diencephalon (Fig. 2A). In contrast to the wild-type pattern, Emx2^–/–embryos also showed strong ectopic expression of Wnt1 in the dorsomedial telencephalon and the rostral diencephalon that peaked at 12.5 dpc(Fig. 2B). Ectopic expression was barely detectable at 14.5 dpc and was absent at 18.5 dpc (data not shown).

A dissected coronal view of a stained Emx2-null telencephalon indicated that ectopic Wnt1 expression was localized both to the cortical hem ventricular zone, which is abnormally small in Emx2-null mice (Muzio et al., 2002), as well as the adjacent hippocampal and neocortical VZ(Fig. 2C). These findings confirmed that robust ectopic expression of Wnt1 occurs in neuroepithelial precursor cells of the cortical hem and medial cortical plate of Emx2^–/– mice in a pattern reminiscent of endogenous Emx2.

The findings above suggested that EMX2 activity was necessary for repression of Wnt1 expression in the telencephalon during the time of formation of the preplate and its derivatives. Indeed, previous work has suggested that EMX2 and other homeodomain proteins (e.g. DLX2, MSX1) can directly bind the Wnt1 enhancer via a core-binding site (HBS1)(Iler et al., 1995) that is conserved at the pufferfish wnt1 locus(Rowitch et al., 1998). Additionally, mutation or deletion of HBS1 resulted in ectopic expression of a reporter transgene in the dorsomedial telencephalon(Iler et al., 1995; Rowitch et al., 1998). To test whether ectopic expression of Wnt1 was sufficient to recapitulate aspects of the Emx2 cortical phenotype, we first over expressed Wnt1 using a strong 1.1 kb enhancer fragment that carried a mutation of HBS1 (Fig. 3A)(Iler et al., 1995). However,preliminary analysis indicated that this resulted in defects of neural tube closure in the midbrain region (data not shown), most probably because of the proliferative effects of Wnt1 and excessive neural overgrowth, as previously observed (Danielian and McMahon,1996; Dickinson et al.,1994). To address this problem, we characterized the expression of additional reporter transgene constructs with various deletions within the 1.1 kb Wnt1 enhancer (Rowitch et al.,1998). An enhancer fragment with deletion of HBS1 and 300 bp of surrounding sequence (Fig. 3B,referred to hereafter as ΔHBS) drove strong ectopic expression of the lacZ reporter gene in the telencephalon(Fig. 4A-D). Yet, in comparison to previous results with the full-length 1.1 kb enhancer(Iler et al., 1995; Rowitch et al., 1998),β-galactosidase activity was relatively weak in regions of endogenous Wnt1 expression, such as the roofplate and midbrain-hindbrain region(Fig. 4B). Ectopic telencephalic expression of the ΔHBS reporter transgene was first detectable at ∼10-somites (8.75-9 dpc, Fig. 4A) and was maintained at least until 16.5 dpc (Fig. 4C). These studies suggested that ΔHBS DNA regulatory sequences were suitable to drive transgene expression within the dorsomedial telencephalon while avoiding adverse consequences of Wnt1 overexpression in its endogenous domain (e.g. neural tube defects).

As shown (Fig. 3C), the Wnt1 locus was cloned upstream of the ΔHBS enhancer, and independent founder transgenic mice were generated for analysis at embryonic(13.5 dpc, n=1), fetal (18.5 dpc, n=2) and neonatal (P1, n=1) stages (hereafter referred to as Wnt1-Tg). Although 10-40% of litters genotyped antenatally carried the transgene, less than 10%of postnatal animals were transgenic. Only one stable line proved viable past one month of age and this was used primarily to generate embryos for analysis at 12.5 dpc (n=4). Although these data indicate that expression of Wnt1-ΔHBS-transgene caused lethality after birth, no transgenic animals showed neural tube defects.

Analysis of Wnt1 mRNA transcripts in situ showed that the expression pattern of the Wnt1-ΔHBS-transgene in the dorsomedial telencephalon of animals derived from the stable line closely resembled that of Emx2 (compare Fig. 4E with Fig. 4G,H). At 12.5 dpc, the transgene was most strongly expressed in the ventricular zone neuroepithelium of the cortical hem region adjacent to the choroid plexus primordium (cpp, Fig. 4G). Moreover, we observed an apparent gradient of transgene expression within the adjacent cortical ventricular zone that was stronger in dorsomedial regions than dorsolateral regions of the telencephalon (Fig. 4G,H) similar to the expression gradient of endogenous Emx2 (Fig. 4E). Ectopic expression was reduced but maintained within the medial most cortical regions within the ventricular zone and subplate at 18.5 dpc (data not shown). Even so, Wnt1 expression was not seen within the marginal zone at early or late stages in the transgenic mice.

During forebrain development, organizing centers are important determinants of regional specialization. One such organizing center, termed the cortical hem, is markedly reduced in size in Emx2-null mice(Muzio et al., 2002). The finding of ectopic expression of Wnt1 in the cortical hem of Wnt1-Tg mice initially suggested the possibility of early patterning abnormalities. However, histopathological analysis of Wnt1-Tg and littermate controls at 12.5 dpc showed only subtle differences in the delineation of the choroid plexus primordium from the adjacent thalamic eminence and cortical hem in transgenic animals, and no reduction in size of the cortical hem (Fig. 5E,F). The most medial part of the cortical primordium (hippocampal anlagen) was also examined just before birth with a range of hippocampal field-specific gene expression markers as previously described(Tole et al., 2000). Although this region was sometimes significantly larger than normal in Wnt1-Tg mice, hippocampal field specification was comparable with that in littermate controls (data not shown).

Analysis of cortical hem markers at 12.5 dpc revealed normal Emx2expression within the cortical hem and adjacent cortical ventricular zone in Wnt1-Tg animals, including preservation of the characteristic medial to lateral gradient of cortical expression (data not shown). Indeed,expression of numerous markers of the roofplate organizer and cortical hem showed no significant alterations in pattern (e.g. Fig. 5C,D,G,H and data not shown). We conclude that patterning of the telencephalic roofplate organizer and dorsomedial telencephalon is normal despite ectopic expression of Wnt1.

To investigate phenotypic similarities between Emx2 null and Wnt1-Tg mice further, we focused on development of the marginal zone and subplate, which show a marked reduction in the number of reelin-positive CR cells and Calretinin-positive cells within the MZ and SP of Emx2-null mice (Fig. 6A,B) (Mallamaci et al.,2000a; Shinozaki et al.,2002; Yoshida et al.,1997). As shown (Fig. 6C), analysis of the anterolateral cortex of Wnt1-Tg mice at 18.5 dpc revealed patchy areas containing markedly reduced numbers of reelin-positive CR cells. Scattered ectopic reelin-positive cells were seen in the cortical plate, marginal zone and subplate of both Emx2^–/– and Wnt1- Tg mice. These cells lacked normal Cajal-Retzius morphology and were extremely small,possibly representing reelin-positive cortical interneurons or abnormally differentiated Cajal-Retzius-like cells(Fig. 6B,C, arrowheads).

Calretinin is a Ca²⁺-binding protein that is strongly expressed in the cell bodies and processes of Cajal-Retzius cells and other neurons within the preplate, marginal zone and subplate. As shown(Fig. 6E,F),immunohistochemical analysis revealed a marked reduction in the number and complexity of Calretinin-positive CR cells and neuronal processes within the marginal zone and subplate of the Emx2^–/– and Wnt1-Tg mice at 18.5 dpc (Fig. 6D-F). Although abnormal reelin and calretinin expression was observed diffusely throughout the anterior to mid-cortex of both Emx2^–/– and Wnt1-Tg mice, the changes appeared more severe and widespread in Emx2^–/–mice than in Wnt1-Tg mice at 18.5 dpc. Medial and posterior regions were relatively less affected in both mice. Together, these findings suggest that ectopic expression of Wnt1 in the telencephalon under control of the ΔHBS1 regulatory element produces abnormalities in preplate derived structures (MZ and SP) that are remarkably similar to those seen in Emx2^–/– mice(Mallamaci et al., 2000a; Shinozaki et al., 2002).

Having observed abnormal formation of preplate derivatives, we next examined Wnt1-Tg mice for other features of cortical dysplasia. Histological analysis at 18.5 dpc revealed that Wnt1-Tg founder mice had well-formed LGH (Fig. 7A-C). Approximately two to four lesions/brain were observed in two out of three independent founder animals examined at 18.5dpc-PN1. Moreover, the lesions occurred in an anterior and dorsal distribution similar to that seen in Emx2^–/– mice. Immunohistochemistry confirmed the presence of numerous NeuN-positive neurons(Fig. 7D), as well as calretinin-positive neurons and processes(Fig. 7E,F) within the lesions from both mice.

The presence of cells within the leptomeninges suggested that the mechanism of heterotopia formation might involve radial migration of cells through a focal defect of the glia limitans and along misplaced radial glial foot processes. Analysis of Emx2^–/– and Wnt1-Tg mice at 18.5 dpc showed that the lesions in both cases were associated with well-defined focal gap defects in the nestin positive radial glial foot processes forming the glia limitans(Fig. 7G-I). In addition,Nestin positive radial glial processes were noted to extend through the SP and MZ and into the substance of the heterotopias in an abnormal fashion and without formation of organized foot processes at the surface of the lesions(Fig. 7H,I, white arrowheads). The glia limitans in regions away from the lesions in both Wnt1-Tg and Emx2^–/– mice had a normal appearance(Fig. 7, red arrowheads). This novel finding suggests that radial glia may provide the substrate for abnormal continued radial migration of cells through outer layers of the marginal zone and into the leptomeninges.

Ectopic Wnt1 expression in the telencephalon of Emx2^–/– mice showed a peak at 12.5 dpc, and was undetectable by 18.5 dpc (Fig. 2 and data not shown). This suggested that heterotopias might form contemporaneously with splitting of the preplate and the onset of radial neuronal migration. To assess timing further, we analyzed effects of ectopic Wnt1 expression in an additional transgenic founder at 13.5 dpc. Similar to findings at 18.5 dpc, we observed a severe reduction in the number of calretinin- and reelin-positive cells and processes within the anterior to mid-cortical fields in the marginal zone and subplate of 13.5 dpc Wnt1-Tg animals (Fig. 8A,B). Posterior and medial cortical regions showed a relatively normal number and distribution of calretinin and reelin-positive cells within the marginal zone although the processes of cells within the subplate were variably reduced within these areas (Fig. 8B, parts i-l).FIG9 

In addition, small MZH (Fig. 8C, part a, arrowheads) and LGH (Fig. 8C, part b, arrowhead) were readily discernable on Hematoxylin and Eosin staining in a distribution that correlated well with the abnormal areas of calretinin and reelin staining (Fig. 8C,parts a,b) in the most anterolateral portions of the telencephalon. The morphology at this early stage was strikingly similar to that seen in 18.5 dpc Emx2^–/– and Wnt1- Tg mice (Fig. 8C, parts a,b). Marginal zone heterotopias were not exclusively associated with an absence of calretinin-positive cells as several such cells(Fig. 8C, part c, lower arrowheads) were detected by immunohistochemistry immediately underlying a MZH(Fig. 8C, part c, upper arrowhead). In addition, no abnormalities were detected within the laminin-containing pial basement membrane(Fig. 8C, part d) by IHC or Nestin-positive processes of radial glia (data not shown) either in association with or distant from marginal zone heterotopias. These observations suggest that ectopic Wnt1 expression adversely affects development of preplate-derived calretinin-positive cell types at∼12.5-13.5 dpc at the onset of cortical plate formation and are further confirmation of phenotypic similarities between Emx2^–/– and Wnt1-Tg mice.

Loss of Emx2 function results in cortical dysplasia including defects in cortical lamination and tangential cell migration in the mouse(Mallamaci et al., 2000a; Shinozaki et al., 2002; Tole et al., 2000). We have found that an additional feature of cortical dysplasia in Emx2^–/– mice is the frequent occurrence of MZH and LGH. We detected MZH/LGH in six out of six Emx2^–/– mice analyzed, but not Emx2^+/– littermates or wild-type controls. These heterotopias are a focal type of dysplasia of cortical layer I and are probably a more severe manifestation of the previously described marginal zone abnormalities in Emx2^–/– mice(Mallamaci et al., 2000a; Shinozaki et al., 2002). The finding of extension of the heterotopias into the leptomeninges might have gone unnoticed in previous analyses of Emx2-null mice because only two to six of these relatively subtle neuropathological lesions are detectable per brain. Moreover, varied penetrance of the phenotype might occur on different genetic backgrounds.

Because the heterotopias in Emx2^–/– mice at late stages are histopathologically identical to MZH and LGH found in humans with cortical dysplasia, we propose that Emx2^–/– mice may be useful as a model to better understand this disorder. Our analysis indicates that both human and mouse lesions have a morphology suggestive of overmigration of radially migrating cells into the marginal zone and leptomeninges. In addition, the cortical plate below the lesion is commonly disrupted into deeper layers,sometimes through to the subplate or subcortical regions. Although we were not able to obtain early human lesions to confirm a similar time of onset to the mouse (i.e. prior to 13.5 dpc), mature large heterotopias have been histologically identified in human fetal autopsy specimens as early as 18 and 20 weeks gestation, implying that their onset could coincide with that of the mouse upon evaluation of earlier stage human embryos(Iida et al., 1994).

As MZH and LGH are seen in association with a wide range of CNS disorders and functional mutations in mice and humans, the genetic basis of these lesions is likely to be complex (Costa et al., 2001; Halfter et al.,2002; Moore et al.,2002). Although EMX2 and WNT1 might play a direct role in human heterotopia formation or subplate abnormalities, no such role has been described previously in the literature. Alterations of Wnt pathway signaling have been described within human cortical resections containing focal intra-cortical dysplasias, but the presence of heterotopias was not addressed in this study (Cotter et al., 1999). Loss-of-function mutations in EMX2 have been described in individuals with the radiological diagnosis of schizencephaly, a rare cortical dysplasia syndrome (Brunelli et al., 1996; Faiella et al.,1997; Granata et al.,1997). However, further work will be required to determine whether EMX2 loss in these individuals is associated with heterotopias or MZ/SP abnormalities as the neuroimaging techniques used to identify the individuals lacked the resolution to detect such microscopic lesions and the neuropathological findings in these cases have not been described.

Several lines of evidence support the proposal that EMX2 is a direct repressor of Wnt1 expression during normal development of the telencephalon. We have previously shown that EMX2 binds the highly conserved HBS1-binding site in the Wnt1 enhancer, and that mutation or deletion of HBS1 results in ectopic reporter transgene expression in the dorsomedial telencephalon (Iler et al.,1995; Rowitch et al.,1998). Second, Yoshida et al.(Yoshida et al., 1997) have previously noted weak ectopic telencephalic expression of Wnt1 in Emx2^–/– mice at 10.5 dpc. By contrast, our results establish robust expression of Wnt1 in the cortical hem organizer region and dorsal cortical ventricular zone at 12.5 dpc, extending the observations of Yoshida et al.(Yoshida et al., 1997) and raising the possibility of phenotypic consequences of ectopic Wnt1expression at a critical stage in telencephalic development. Indeed, we confirmed by histological and expression analysis that the cortex of Wnt1-Tg mice was a partial phenocopy of Emx2^–/– cortical dysplasia, sharing profound abnormalities of the preplate and their derivative cell types. Finally, we identified LGH in the cortex of Emx2^–/– mice and verified that precise ectopic expression of Wnt1 within the Emx2 domain was sufficient for LGH formation. Together, these findings strongly support a model in which EMX2 directly binds the Wnt1 enhancer and represses expression of Wnt1 in the developing telencephalon (Fig. 8).

Previous work has shown that EMX2 regulates several distinct aspects of cortical development. First, EMX2 activity regulates proliferation of neural precursors of the developing telencephalon and neural stem cells(Galli et al., 2002; Heins et al., 2001), which may account in part for the small size of the cortex and cortical hem in the Emx2-null mice. Additionally, Emx2 function is necessary for proper development, but not specification, of the hippocampus(Tole et al., 2000). A similar phenotype is observed in Wnt3a mutants(Lee et al., 2000). Thus, one possibility is that EMX2 serves as a `competence factor' that is necessary for neural precursors to respond to a growth signal, while Wnt3a and other Wnt proteins may act principally as mitogens within the developing telencephalon. We noted that the hippocampus and cortical hem are normally formed in the majority in Wnt1-Tg animals, indicating that hypoplasia of these structures in Emx2-null mice is unrelated to ectopic expression of Wnt1. Indeed, the overall size of Wnt1-Tg brain was typically larger than normal (megalencephalic), consistent with the known mitogenic effects of Wnt1 and β-catenin signaling on a wild-type genetic background (Chenn and Walsh,2002; Danielian and McMahon,1996; Dickinson et al.,1994; Megason and McMahon,2002). Ectopic expression of Wnt1 alone also did not reproduce the diffuse architectural abnormalities of cortical plate lamination and radial glial morphology present in the cortex of Emx2^–/– mice, most probably because of variability of gene dosage and/or non-uniform (variegated) expression commonly observed in the analysis of founder transgenic mice. Nevertheless, we observed LGH and abnormal formation of subplate and marginal zone populations in a total of 3/4 (75%) independent transgenic founders. In summary, our results suggest that repression of Wnt1 by EMX2 is important in its capacity as a determinant of cortical layer I and SP structure and function, but not for its other roles in the regulation of neural precursor proliferation,cortical lamination in layers II-VI or development of the hippocampus.

The preplate (PP) forms as the initial wave of neural precursors leaves the ventricular zone and migrates radially to the margin of the developing telencephalon. The PP, MZ and SP then serve as a framework for the subsequent influx of neurons from the ventricular zone, cortical hem and ventral forebrain, potentially along tangential and radial processes of cells that reside in these layers (Parnavelas,2000). Because ectopic Wnt1 expression was never detected in the PP or MZ of Emx2-null or Wnt1-Tg mice, it is likely that it acts principally on the early ventricular zone progenitors of the dorsomedial cortex. It is now known that cells from this region, especially the cortical hem, appear much more migratory than originally assumed. Several reports suggest that neurons from the hem migrate extensively into the adjacent cortical primordium (Meyer et al., 2002; Monuki et al.,2001). As a result, ectopic Wnt1 is expressed early on in the same progenitor cells that will later migrate into and throughout the neocortex thereby creating the potential for very long-range functional effects.

Although ΔHBS Wnt1 regulatory sequences drove ectopic expression in the dorsomedial telencephalon commencing at ∼10-somites(8.75 dpc), we first observed heterotopias in Wnt1-Tg mice at 13.5 dpc, contemporaneous with PP maturation and the earliest formation of the MZ and SP. This finding is consistent with a model in which heterotopias result from defects in the developing MZ and SP or pre-existing defects in the PP. We observed striking and diffuse deficiencies of reelin-expressing CR cells and calretinin-expressing neurons as early as 13.5 dpc in the telencephalon of Wnt1-Tg mice. Evidently, Wnt1 signaling inhibits development or initial migration of preplate cells and their processes.

Another feature common to heterotopias of Emx2^–/– and Wnt1-Tg mice is the novel finding of abnormally located processes of radial glia outside the glia limitans and limiting basement membrane of the leptomeninges. Such ectopically placed processes imply the existence of focal defects in the glia limitans and laminin 1-containing subpial basement membrane during the period of radial migration of cells. In fact, mice with an unstable basement membrane, owing to a mutation in laminin-g-1, develop defects in the basement membrane and glia limitans at early stages (10-12 dpc) associated with numerous LGH(Halfter et al., 2002). Although we found defects in the glia limitans/basement membrane complex associated with heterotopias in Emx2^–/– and Wnt1-Tg mice using Nestin immunostaining at late stages (18.5 dpc),we were unable to detect primary defects in the limiting plate in regions overlying MZ heterotopias of Wnt1-Tg mice at early stages of radial migration (13.5 dpc) by laminin 1, Nestin or TuJ1 immunostaining (Nestin and TuJ1, data not shown). Together, these data suggest that positioning of the radial glial foot process and formation of the pial basement membrane may be linked to early events in formation of the marginal zone and subsequent leptomeningeal heterotopia formation, but that such defects are not required for the formation of heterotopias within the marginal zone.

Regulatory interactions between Wnt signaling and Emx2 expression in the telencephalon have recently been described(Theil et al., 2002). However,we did not observe abnormal expression of Emx2 in the telencephalon of Wnt1-Tg mice (K.L., E.G. and D.R., unpublished), nor would one expect regulation of Emx2 directed by Wnt1 given their non-overlapping patterns of expression during normal development. Because Emx2 expression is unaltered in the ΔHBS Wnt1transgenic mice in precisely the same regions where we see ectopic Wnt1 expression, it is evident that EMX2 cannot repress expression from the ΔHBS Wnt1 enhancer construct. Similar observations apply to analysis of lacZ reporter constructs with mutated or deleted HBS sequences (Iler et al.,1995; Rowitch et al.,1998). Although we do not have evidence for direct EMX2 interaction/repression of Wnt1 at the HBS sequences in vivo, previous work suggests this to be the case. Within the entire 5.5 kb Wnt1enhancer, which has been shown to be necessary and sufficient for regulation of Wnt1 in vivo (Danielian et al., 1997; Echelard et al.,1994), homeodomain proteins have been found to bind only the HBS1 and HBS2 sites located ∼5 kb downstream of the Wnt1 polyA(Iler et al., 1995; Shang et al., 1994). Together,these observations suggest that regulation of Wnt1 by EMX2 is via direct binding to HBS cis-acting regulatory sequences.

Because multiple Wnt genes are expressed in the developing forebrain (e.g. Wnt2b, Wnt3a, Wnt5a, Wnt 7a, Wnt7b, Wnt 8a and Wnt8b), it is not obvious why there should be a mechanism for repression of Wnt1. It might be important to regulate the total dosage of Wnt protein during development; alternatively, Wnt1 could have specific signaling effects that do not overlap with other Wnt proteins. In this regard it is interesting to note that ectopic expression of Wnt1 driven byΔHBS1 regulatory sequences overlaps that of Wnt7a in the cortical ventricular zone (Lee et al.,2000). Wnt family members can be divided into functional classes based on downstream signaling via β-catenin: Wnt1 and Wnt3a signal via β-catenin, in contrast to Wnt5a and Wnt7a (see http://www.stanford.edu/~rnusse/wntwindow.html). Thus, it is possible that ectopic activation of β-catenin signaling per se affects the radial migration of neurons emigrating from the ventricular zone and/or positioning of radial glia. Shinozaki et al. have recently shown requirements for Emx1/2 function during tangential migration of MGE precursors into the neocortex (Shinozaki et al., 2002). In particular, wild-type MGE precursors failed to migrate into the Emx-null cortex, indicating a non-cell-autonomous defect. Bishop et al. have suggested that Emx activity is critical for afferent and efferent axonal projections in the neocortex(Bishop et al., 2003). Further work will be required to establish whether aberrant expression ofβ-catenin or WNT1 protein is directly related to abnormal patterns of neural migration, axon pathfinding and preplate development in Emx mutant mice.

The authors thank Dong-In Yuk for expert technical assistance, Dr P. Gruss for Emx2 mutant mice, Drs Q. Ma and C. Walsh for comments and suggestions, and members of the Rowitch laboratory for helpful advice. We are also grateful to Dr S. Tole for initial analysis of Wnt1-Tg mice. K.L.L. acknowledges the NIH (T32HD007466-08) for support. This work was funded by grants from the NIH [RO1MH059962 (to E.A.G.), RO1HD030249 (to A.P.M.) and RO1NS040511 (to D.H.R.)], March of Dimes (to E.A.G.) and James S. McDonnell Foundation (to D.H.R.). D.H.R. is a Harry Weaver Neuroscience Scholar of the National Multiple Sclerosis Society.