We have examined the role of Tlx, an orphan nuclear receptor, in dorsal-ventral patterning of the mouse telencephalon. Tlx is expressed broadly in the ventricular zone, with the exception of the dorsomedial and ventromedial regions. The expression spans the pallio-subpallial boundary, which separates the dorsal (i.e. pallium) and ventral (i.e. subpallium) telencephalon. Despite being expressed on both sides of the pallio-subpallial boundary, Tlx homozygous mutants display alterations in the development of this boundary. These alterations include a dorsal shift in the expression limits of certain genes that abut at the pallio-subpallial boundary as well as the abnormal formation of the radial glial palisade that normally marks this boundary. The Tlx mutant phenotype is similar to, but less severe than, that seen in Small eye(i.e. Pax6) mutants. Interestingly, removal of one allele ofPax6 on the homozygous Tlx mutant background significantly worsens the phenotype. Thus Tlx and Pax6 cooperate genetically to regulate the establishment of the pallio-subpallial boundary. The patterning defects in the Tlx mutant telencephalon result in a loss of region-specific gene expression in the ventral-most pallial region. This correlates well with the malformation of the lateral and basolateral amygdala in Tlx mutants, both of which have been suggested to derive from ventral portions of the pallium.
The embryonic telencephalon can be divided into distinct dorsal and ventral domains. The dorsal domain (i.e. the pallium) generates projection neurons of the neocortex, hippocampus and piriform cortex(Bayer and Altman, 1991). The ventral telencephalon (i.e. the subpallium) gives rise to the striatum and globus pallidus of the basal ganglia as well as to the basal forebrain(Smart and Sturrock, 1979). In addition, the subpallium also contributes GABAergic interneurons to pallial-derived structures via tangential migration (for a review, seeMarin and Rubenstein, 2001). Another prominent nuclear complex, the amygdala, has received less attention. It is currently unclear where the different nuclei of the amygdala are generated (Swanson and Petrovich,1998), however, it has been suggested that both pallial and subpallial regions contribute to its formation(Fernandez et al., 1998;Puelles et al., 1999;Puelles et al., 2000).
The dorsal and ventral portions of the telencephalon are separated by a morphologically identifiable radial glial palisade(Stoykova et al., 1997;Götz et al., 1998;Chapouton et al., 1999). The location of this glial palisade marks the pallio-subpallial boundary and coincides with the juxtaposition of cells expressing distinct developmental control genes. Precursor cells on the subpallial side express the homeobox gene Gsh2 and the bHLH gene Mash1 (Ascl1 –Mouse Genome Informatics), whereas cells on pallial side express the paired homeobox gene Pax6 as well as the bHLH genes neurogenin 1(Ngn1) and Ngn2. Several of these developmental control genes play important roles in the establishment and/or maintenance of this boundary (Stoykova et al.,1997; Stoykova et al.,2000; Chapouton et al.,1999; Chapouton et al.,2001; Corbin et al.,2000; Toresson et al.,2000; Yun et al.,2001). For example, Gsh2 is required to repress pallial gene expression in the subpallium and in its absence ventricular zone cells in the lateral ganglionic eminence (LGE) are misspecified, resulting in a truncation of the LGE-derived striatum(Corbin et al., 2000;Toresson et al., 2000;Yun et al., 2001). Conversely,in Small eye (Sey) mutants, which contain a point mutation in the Pax6 gene (Hill et al.,1991), the pallium is misspecified by ectopic ventral gene expression along with restricted loss of pallial gene expression(Stoykova et al., 1996;Stoykova et al., 2000;Toresson et al., 2000;Yun et al., 2001;Muzio et al., 2002).
The above-mentioned genes are unlikely to be the only molecular players involved in the establishment and/or maintenance of the pallio-subpallial boundary. In the present study, we have examined a role for the orphan nuclear receptor Tlx (also known as tailless; Nr2e1 –Mouse Genome Informatics) (Yu et al.,1994; Monaghan et al.,1995) in this process. Homozygous Tlx mutants have been shown to exhibit altered telencephalic morphology as well as abnormally aggressive behavior (Monaghan et al.,1997; Young et al.,2002). However, the underlying mechanisms that lead to these defects have not been elucidated. We show here that Tlx is required for correct dorsal-ventral patterning of the embryonic telencephalon and for the normal differentiation of amygdalar structures in the ventrolateral telencephalon. Moreover, our data demonstrate that Tlx andPax6 cooperate genetically to establish the pallio-subpallial boundary.
MATERIALS AND METHODS
Animals and genotyping
Tlx mice and embryos (Yu et al., 2000) were genotyped by PCR using the following primers:3TlxlacZ – ATT CGC GTC TGG CCT TCC TGT AG, 3Tlxwt – ACC CTG GGG AGT ACC TGG TTT CC, and 5Tlxcommon – CTC TTC CCG TCT TTC AGG CCG. The wild-type allele results in a 177 bp band, whereas the targeted allele gives a 453 bp band. Sey mice and embryos were typed visually based on eye morphology and eye size as described previously(Hill et al., 1991).Dlx5/6-cre-IRES-EGFP mice were genotyped as described in Stenman et al. (Stenman et al., 2003). B6;129-Gtrosa26tm1Sho (gtROSA) mice(Mao et al., 1999) were obtained from Jackson laboratories and genotyped as described in Jackson's Gtrosa genotyping protocol available online(http://www.jax.org). Fate mapping studies were performed on mice that were double-transgenic for the Dlx5/6-cre-IRES-EGFP and gtROSA alleles. For staging of embryos, the morning of the appearance of the vaginal plug was designated as embryonic day 0.5 (E0.5).
Embryos were fixed and sectioned on a cryostat as previously described(Toresson et al., 2000). Adult brains were removed from 3-week- to 8-month-old animals and fixed overnight by immersion in 4% paraformaldehyde at 4°C before sinking in 30% sucrose and sectioning at 30 μm on a cryostat. Immunohistochemistry was performed on slide-mounted sections for the embryonic tissue, whereas the adult brain sections were immunostained under free-floating conditions. The protocol was as previously described (Olsson et al.,1997) with the modification that 0.3% H2O2was used for 10-15 min instead of 3% H2O2. Primary antibodies were used at the following concentrations: rabbit anti-DARPP-32(1:1000, Chemicon); rabbit anti-distalless (i.e. pan DLX) (1:1000, provided by G. Panganiban); rabbit anti-Er81 (1:5000, provided by S. Morton and T. Jessell); rabbit anti-GSH2 (1:5000)(Toresson et al., 2000);rabbit anti-ISL1/2 (1:500, provided by T. Edlund); rabbit anti-MASH1 (1:1000,provided by J. Johnson); rabbit anti-MEIS2 (1:5000, provided by A. Buchberg);rabbit anti-nestin (1:500, provided by R. McKay); rabbit anti-parvalbumin(1:1000, provided by P. Emson); rabbit anti-PAX6 (1:200, Covance) and goat anti-PAX6 (1:250, Santa Cruz). The secondary antibodies used were biotinylated swine anti-rabbit antibodies (DAKO), FITC-conjugated anti-rabbit (Jackson Immunoresearch) and Cy3-conjugated anti-goat antibodies (Jackson Immunoresearch). The ABC kit (Vector labs) was used to visualize the reaction product for the biotinylated antibodies with diaminobenzidine as the final chromogen. Confocal microscopy was performed on a Zeiss LSM510 confocal microscope.
For acetylcholine esterase (AChE) staining, adult brain sections were mounted on slides and dried at 37°C for 1 hour before placing in incubation medium (3 mM cupric sulfate, 10 mM glycine, 15 mM acetic acid, 35 mM sodium acetate, 0.08 mM tetraisopropyl pyrophosophoramide, 1.5 mM acetylthiocholine iodide, adjusted to pH 5.0) at 37°C for 4 hours. They were then developed in a 40 mM sodium sulphide solution (pH 7.5).
Altered patterning of the lateral telencephalon in Tlxmutants
Tlx is expressed in the mouse forebrain as early as the five-somite stage (approximately E8.5)(Monaghan et al., 1995). The telencephalic and diencephalic expression of this gene is restricted to the ventricular zone (Fig. 1). At E12.5, Tlx is highly expressed in the lateral portion of the telencephalon, including both the dorsal and lateral pallium and the lateral ganglionic eminence (LGE) (Fig. 1A). Low but detectable levels are found in the dorsomedial pallium and in the medial ganglionic eminence (MGE)(Fig. 1A). This expression pattern is maintained at both E14.5 (Fig. 1B) and E16.5 (data not shown).
At E12.5, Tlx mutants appear morphologically normal, however, at the molecular level they exhibit defects in the patterning of the lateral telencephalon. The pallio-subpallial boundary is marked by the juxtaposition of GSH2- and PAX6-expressing cells of the subpallial and pallial ventricular zones, respectively (Fig. 2A-C). Confocal micrographs show only a minimal overlap (i.e. 2-3 cell diameters) of cells expressing these two proteins in the wild-type telencephalon (Fig. 2C,D). InTlx mutants, GSH2-expressing cells are found dorsal to their normal limit (i.e. up to and in the LGE-pallium angle,Fig. 2E). In these mutants, the overlap of GSH2 and PAX6 in ventricular zone cells is much broader than in the wild types (Fig. 2E-H). Interestingly, this broader overlap correlates with the altered organization of PAX6-positive cells in the subventricular region. In wild types,PAX6-positive cells are seen to emanate from the region where GSH2 and PAX6 overlap in the ventricular zone and to form a stream from the subventricular zone (SVZ) out to the pial surface (Fig. 2B-D) (Puelles et al.,1999). In Tlx mutants, however, the PAX6-positive cells form a broader domain in the SVZ (Fig. 2H), which corresponds well with the wider overlap of GSH2 and PAX6 expression. The dorsal shift of GSH2 expression in the Tlxmutants is still observed at E14.5 (Fig. 3E), however, high-level PAX6 expression appears to have retracted dorsally by this stage (Fig. 3H). Moreover, in the Tlx mutants, the position of the PAX6-expressing cells in the SVZ region has shifted dorsally and there are 2.2-fold more of these cells (Fig. 3H) than in wild types (Fig. 3D; 514±89 versus 235±90 cells, respectively,P<0.02, n=3).
The bHLH genes Mash1 and Ngn2 also mark cells on the pallial and subpallial sides of the pallio-subpallial boundary, respectively. In addition to the dorsal expansion of GSH2 expression in Tlx mutants(Fig. 2E,Fig. 3E), MASH1-positive cells can also be found in a more dorsal position in the mutant telencephalon(Fig. 3F). Concomitant with this, Ngn2 expression in Tlx mutants is retracted from its normal ventral limit (Fig. 3G),as is the case for PAX6 (Fig. 3H). These alterations in gene expression are seen at both rostal and caudal levels. Taken together, these results demonstrate a dorsal shift in the expression limits of genes, which normally abut at the pallio-subpallial boundary.
Based on gene expression patterns, the pallium has recently been divided into medial, dorsal, lateral and ventral portions(Puelles et al., 1999;Puelles et al., 2000;Yun et al., 2001). The ventral pallium is a small domain located immediately dorsal to the pallio-subpallial boundary. This pallial region is normally marked, at least in part by the expression of Dbx1, a homeobox gene(Yun et al., 2001), andSfrp2 (secreted frizzled related protein 2), which encodes a putative Wnt inhibitor (Kim et al.,2001). In addition to the dorsal shift in the expression limits of GSH2, MASH1, Ngn2 and PAX6 in the lateral telencephalon, the ventral pallial region of the Tlx mutants lacks expression of bothDbx1 (Fig. 4B) andSfrp2 (Fig. 4D). This loss in expression is specific to the ventral pallium as both genes continue to be expressed in the mutant diencephalon(Fig. 4B andFig. 4D).
Unlike the case at E12.5, by E14.5 and onwards the Tlx mutant forebrains do not appear normal in that there is a significant reduction in the size of the LGE (see Figs3,4,5). Despite this reduction, the dorsal limit of the DLX-expressing SVZ appears to have shifted dorsally in accordance with the observed shift of GSH2 and MASH1(Fig. 5B). Notably, this is not the case for the dorsal limit of Islet1 expression(Fig. 5D). However, we observed a dorsal expansion in the LGE SVZ-expression domain of the ETS transcription factor Er81 (Fig. 5F). In the wild type, Er81 is also expressed in the ventricular zone of the pallium,including at least the ventral and lateral pallium(Fig. 5E). Commensurate with the dorsally expanded subpallial SVZ expression of Er81, the pallial ventricular zone expression is greatly reduced in at least the ventral pallium(Fig. 5F). We have recently shown that Er81 and Islet1 mark separate dorsal and ventral progenitor pools in the DLX-expressing SVZ of the LGE, respectively(Stenman et al., 2003). These findings indicate that the patterning defects around the pallio-subpallial boundary in Tlx mutants result in the selective expansion of certain dorsal LGE characteristics (i.e. DLX and Er81 expression) at the expense of those normally marking the ventral pallium. Thus the data presented above show that the loss of Tlx gene function results in a misspecification of the ventral pallium (Fig. 6).
In addition to gene expression patterns, the pallio-subpallial boundary is also marked by a palisade of radial glial fibers originating in the ventricular zone near the LGE-pallium angle and coursing to the pial surface. In the wild-type telencephalon, this glial palisade can be visualized by Nestin staining (Fig. 7A).Tlx mutants show fewer nestin-positive radial glial fibers in the region of the pallio-subpallial boundary. Moreover, these fibers do not appear to fasciculate to form the palisade (Fig. 7B). The calcium-binding protein parvalbumin also marks radial glia in the pallio-subpallial boundary(Fig. 7C). Again, the parvalbumin-positive radial glial fibers in the mutant telencephalon fail to fasciculate (Fig. 7D). Thus, in addition to regulating gene expression at the pallio-subpallial boundary,Tlx gene function is also required for the normal formation of the radial glial palisade. Indeed, the altered patterning, described above, may contribute significantly to the abnormal formation of this radial glial palisade.
Gene dosage of Tlx and Pax6 regulates the patterning of the lateral telencephalon
Several studies have shown that Pax6 is required for the correct patterning at the pallio-subpallial boundary(Stoykova et al., 1996;Stoykova et al., 2000;Toresson et al., 2000;Yun et al., 2001) as well as the formation of the radial glial palisade(Stoykova et al., 1997;Götz et al., 1998;Chapouton et al., 1999). Homozygous Sey mutants show ectopic ventral (i.e. Gsh2,Mash1 and Dlx) gene expression in the pallium and a loss of dorsal (i.e.Ngn1 and Ngn2) gene expression in the corresponding domain. Moreover, the ventral pallium markers Dbx1(Yun et al., 2001) andSfrp2 (Kim et al.,2001; Muzio et al.,2002) are both lost in the Sey/Sey mutants. The fact thatTlx mutants exhibit a similar, but much less severe, phenotype toSey homozygotes could be because of Pax6 directly or indirectly regulating Tlx expression. This does not seem to be the case, however, because Tlx is expressed in Sey/Sey mutants both at E12.5 and E14.5 (data not shown). Furthermore, the data presented above demonstrate that Tlx is not a general regulator ofPax6 expression.
Tlx and Pax6 may, therefore, co-operate genetically to establish the pallio-subpallial boundary. To analyze this we generated a series of compound Sey and Tlx alleles at E14.5 and E16.5. Mice heterozygous for either the Sey or Tlx mutations do not show alterations in gene expression at the pallio-subpallial boundary with respect to any of the markers we have employed here (data not shown). However,in Tlx+/–;Sey/+ compound heterozygotes, a few scattered GSH2- and MASH1-positive cells are seen in the region of the ventral pallium (data not shown). Furthermore, the expression ofSfrp2 is reduced, and at some levels missing, inTlx+/–;Sey/+ compound heterozygotes (data not shown). The double heterozygous phenotype, however, is much less severe than that observed in Tlx homozygous mutants (described above). We also analyzed Tlx–/–;Sey/+ mutants and in all cases the limit of GSH2 (Fig. 8D), MASH1 (Fig. 8E) and DLX (data not shown) expression extends further dorsally into the pallium than in Tlx homozygous mutants(Fig. 8A,B). Furthermore,Ngn2 is downregulated in the pallial domain containing ectopic GSH2 and MASH1 cells (Fig. 8F). These findings demonstrate that removal of one allele of Pax6 on theTlx–/– mutant background results in a significant dorsal shift in the expression limits of GSH2, MASH1, DLX and Ngn2, as compared to the Tlx–/–mutants alone. It is interesting to note that the patterning defects in theSey/Sey mutants (Fig. 8G-I) are more severe than those observed inTlx–/–;Sey/+ mutants. Moreover, the patterning defects observed in Tlx+/–;Sey/Sey (data not shown) andTlx–/–;Sey/Sey mutants(Fig. 8J-L) were not noticeably different from those seen in the Sey/Sey mutants. Thus Pax6appears to be required for correct patterning in broader portions of the pallium (i.e. lateral and dorsal pallium) than Tlx. Both, however,are important for the correct patterning of gene expression around the pallio-subpallial boundary. Moreover, our findings show that correct gene dosages of both Tlx and Pax6 are required to properly establish this boundary (Fig. 9).
Altered amygdalar development in Tlx mutants
Tlx mutants have smaller than normal brains, which exhibit gross morphological defects in numerous telencephalic regions, including the amygdalar region (Monaghan et al.,1997). However, specific defects in the amygdala of these mutants have not, as yet, been described. Previous studies have suggested that the basolateral amygdala derives from the ventral pallium(Fernandez et al., 1998;Puelles et al., 1999;Puelles et al., 2000). Given the molecular misspecification of the ventral pallial region of Tlxmutants (described above), alterations in the development of this amygdalar nucleus would be predicted. We analyzed the amygdalar region in perinatal animals, however, because Tlx mutants are viable, we were also able to analyze this region in mature brains (i.e. three weeks to eight months old).
In wild-type animals, the basolateral amygdala is marked by the expression of Er81 from late-embryonic stages into adulthood(Fig. 10A). Tlxmutant brains stained for Er81 reveal little evidence of a normal basolateral nucleus in either perinatal or mature brains(Fig. 10B). In addition to marking the basolateral nucleus, acteylcholine esterase (AChE) staining also reveals the lateral nucleus and the central nucleus of the amygdala(Fig. 10C). In theTlx mutants, some AChE staining is found in the presumptive region of the basolateral and lateral amygdala (Fig. 10D), however, the amount of staining is drastically reduced as compared to the wild type. Despite this, the size of the central nucleus appears rather similar to that in wild types(Fig. 10C,D). Staining for the phosphoprotein DARRP-32 outlines the lateral and basolateral amygdala in wild types, revealing the `teardrop'-shaped nuclei(Fig. 10E). This morphology is not apparent in Tlx mutants (Fig. 10F). DARPP-32 marks the interstitial nucleus of the amygdala(Fig. 1E), as does MEIS2(Fig. 10G). This nucleus also seems to be present in Tlx mutants and although its morphology is changed, it appears to be somewhat similar in size to that in wild types(Fig. 10G,H). Thus the alterations in the Tlx mutant amygdala seem to be rather specific to the basolateral and lateral amygdala.
In order to strengthen the correlation between the misspecification of the ventral pallium and the alterations in the development of the basolateral and lateral amygdalar nuclei, we have performed fate-mapping studies that, by exclusion, support a pallial origin for these nuclei. Using a Dlx5/6enhancer (Zerucha et al.,2000) to drive cre recombinase in the subpallial SVZ(Stenman et al., 2003) ofgtROSA reporter mice (Mao et al.,1999), we have found that the lateral and basolateral amygdala contain very few cells originating in the subpallium(Fig. 10I). In fact, few, if any, of the subpallium-derived cells in the basolateral amygdala express Er81(Fig. 10J). Interestingly, the normal expression of Er81 in the ventricular zone of the ventral pallium is lost in Tlx mutants (Fig. 5F). Thus, this data, together with the data presented above,indicate that the Er81-expressing cells in the basolateral amygdala are probably derived from the ventral pallium and may represent the glutamatergic projection neurons characteristic of this nucleus(Swanson and Petrovich, 1998). The small population of subpallium-derived cells(Fig. 10J) might represent the GABAergic interneuron population in this nucleus (e.g.Smith et al., 2000). In addition, our data show that the central and medial nuclei of the amygdala are largely derived from the subpallium (Fig. 10I). In summary, the present findings suggest that the patterning defects around the pallio-subpallial boundary in Tlx mutants have severe consequences for the development of basolateral and lateral nuclei in the amygdalar complex.
Tlx is required for patterning of the lateral telencephalon
Our results demonstrate that the orphan nuclear receptor Tlx plays an important role in the patterning of the lateral telencephalon. Specifically, Tlx mutants do not correctly establish the pallio-subpallial boundary. This is evidenced in both altered gene expression at the pallio-subpallial boundary as well as abnormal formation of the radial glial palisade. In Sey mutants, the formation of the pallio-subpallial boundary is also disturbed, which results in an increased migration of subpallial-derived GABAergic neurons into cortical regions(Stoykova et al., 1997;Stoykova et al., 2000;Götz et al., 1998;Chapouton et al., 1999;Toresson et al., 2000;Yun et al., 2001). It will be interesting to determine whether the alterations in the establishment of the pallio-subpalial boundary observed in Tlx mutants also leads to an increased ventral to dorsal migration of subpallial neurons.
A stream of PAX6-positive cells, which emanate from the ventricular zone and course downward towards the pial surface, is found at the pallio-subpallial boundary (Puelles et al., 1999). Interestingly, the point at which these cells emerge from the ventricular zone seems to correlate with a slight overlap of GSH2-and PAX6-expressing cells (see Fig. 2), which is also the case in the embryonic chick telencephalon(von Frowein et al., 2002).Gsh2 appears to be important for the development of this stream of cells because the number of PAX6-positive cells is reduced (but not missing)in Gsh2 mutants (Toresson et al.,2000). Although the PAX6-positive stream of cells has been suggested to belong to the subpallium(Puelles et al., 1999;Puelles et al., 2000), it may,in fact, represent a population of transitional cells between the pallial and subpallial compartments. Unlike the case in Gsh2 mutants, the subventricular domain of PAX6-positive cells appears to be broader inTlx mutants as compared to that in wild types. This correlates well,at least at early stages, with the increased overlap of the GSH2- and PAX6-expression domains in the Tlx mutant ventricular zone.
In Tlx mutants, the dorsal shift in the expression limits of genes that abut at the pallio-subpallial boundary, is accompanied by a loss ofDbx1 and Sfrp2 expression (i.e. ventral pallial identity). It is unclear, however, whether this indicates a direct role for Tlxin the development of the ventral pallium or if the effect is indirect. Because Tlx is not required for the diencephalic expression ofDbx1 and Sfrp2, a direct role for this gene in the regulation of these ventral pallial markers seems unlikely. Alternatively, the dorsal expansion of GSH2 and MASH1 in the mutants might suggest a role forTlx in the repression of these factors within the ventral pallium. The loss of ventral pallial identity in the mutants could therefore be because of the ectopic expression of subpallial genes. In support of this, the ventral pallium marker Dbx1 is up-regulated in the LGE of homozygousGsh2 mutants (Yun et al.,2001), suggesting a role for Gsh2 in the repression ofDbx1. Homozygous Sey mutants (which display ectopicGsh2 gene expression in the pallium) also exhibit a loss of ventral pallial identity (Kim et al.,2001; Yun et al.,2001; Muzio et al.,2002). However, Sey/Sey mutants display more severe patterning defects, including both the lateral and the dorsal pallium as well (Toresson et al.,2000; Yun et al.,2001).
The origins of different amygdalar nuclei has previously been unclear. Although it has been suggested that its nuclear components derive from both the dorsal and ventral halves of the embryonic telencephalon(Swanson and Petrovich, 1998). In particular, the basolateral amygdala and the lateral amygdala have been suggested to derive from the ventral pallium and the lateral pallium,respectively (Fernandez et al.,1998; Puelles et al.,1999; Puelles et al.,2000). However no direct evidence for this has been provided thus far. A recent study (Gorski et al.,2002) has fate mapped the Emx1-expression region of the dorsal telencephalon. This study showed that, in addition to the neocortex and hippocampus, many structures in the ventrolateral cortical region are also derived from the Emx1-expression domain, including both the basolateral and lateral amygdala. A defining feature of the ventral pallium is the lack of Emx1 gene expression, along with the expression ofPax6, Tbr1, Dbx1 and Sfrp2(Puelles et al., 1999;Yun et al., 2001;Kim et al., 2001). Thus theEmx1-expressing pallial regions should not contribute neurons to the basolateral and lateral amygdala. However, it is possible that low levels ofEmx1 are normally expressed in the ventral pallium. This expression might not be easily detected by in situ hybridization but could drive sufficient levels of cre recombinase to mark cells derived from the ventral pallium (i.e. neurons in the basolateral and lateral amygdala). Our data support a ventral pallial origin for both of these amygdalar nuclei because the patterning defects observed in the Tlx mutants appear to be restricted to the ventral pallium. However, it is difficult to determine whether a portion of the lateral pallium is also affected in these mutants because of a lack of specific markers for this pallial region. These amygdalar nuclei are largely generated between E11 and E14 in the mouse(McConnell and Angevine,1983), which correlates well with the timing of the observed patterning defects in Tlx mutants. Moreover, our fate-mapping studies, using the subpallial Dlx5/6 expression domain, demonstrate a largely pallial origin for the basolateral and lateral amygdala because only a few cells are labeled in these nuclei. In contrast, the central and medial amygdalar nuclei do appear to derive from subpallial sources. Taken together,these fate-mapping studies provide convincing evidence for both pallial and subpallial contributions to the amygdalar complex.
Homozygous Tlx mutants can survive after birth and have been reported to display abnormally aggressive behavior(Monaghan et al., 1997;Young et al., 2002). Because the lateral and basolateral amygdala are both thought to be involved in the regulation of fear rather than aggression(Oakes and Coover, 1997;Nader et al., 2001), it is unlikely that the amygdalar defects in these mutants are responsible for the aggressive behavior. It should be noted, however, that other morphological defects are present in the Tlx mutant forebrain(Monaghan et al., 1997), which may contribute more or less to their aggressive behavior. Notably, the LGE inTlx mutants is considerably more reduced in size than other telencephalic structures such as the cortex. This reduction in the size of the LGE is unlikely to contribute significantly to the observed defects at the pallio-subpallial boundary in the Tlx mutants because the opposite phenotype would be predicted, at least with respect to the gene expression. Indeed, it is probable that a smaller LGE will result in a ventral shift of the expression limits of genes that abut at the pallio-subpallial boundary. We are currently investigating the mechanisms that underlie the reduced LGE size in the Tlx mutants, which may include either, or a combination of,cell death, lack of proliferation or a patterning defect.
Tlx and Pax6 interactions
As mentioned above, the pallial phenotype of homozygous Tlx andSey mutants share several similarities, specifically the alteration in gene expression around the pallio-subpallial boundary as well as altered development of the radial glial palisade. This motivated us to further examine the relationship between these two genes in the process of telencephalic dorsal-ventral patterning. Our findings show that the correct gene dosage of both Tlx and Pax6 is crucial for the establishment of the pallio-subpallial boundary. Although the loss of one allele of Pax6on the homozygous Tlx mutant background results in a significant worsening of the pallial phenotype, removal of either one or both of theTlx alleles on the homozygous Sey background does not further exacerbate the phenotype as compared to Sey/Sey mutants alone. Therefore Tlx is required to augment Pax6 gene function in the ventral-most portions of the pallium and thereby to correctly position the pallio-subpallial boundary. This genetic interaction withPax6 provides an explanation for why Tlx, despite its broad expression pattern, is crucial for the establishment of the pallio-subpallial boundary. It seems that Tlx is not the only gene that is expressed across the pallio-subpallial boundary and regulates correct gene expression at this boundary. The zinc finger gene Gli3, which is expressed on both sides of the pallio-subpallial boundary, is also required in this process(Tole et al., 2000;Rallu et al., 2002). As is the case with Tlx, this is not through the direct regulation ofPax6 expression.
The fact that Tlx and Pax6 interact genetically in the establishment of telencephalic dorsal-ventral identity suggests that their protein products might do so through direct molecular interactions. PAX6 is known to physically interact with the HMG box protein SOX2 in the regulation of eye development (Kamachi et al.,2001). Sox2 is expressed in the telencephalic ventricular zone in regions overlapping with Pax6 expression(Zappone et al., 2000),suggesting that similar interactions may be involved in telencephalic patterning. No interacting partners for TLX have, as yet, been identified in vertebrates, not even retinoid X receptors (RXRs), which are known to interact with many orphan nuclear receptors (for a review, seeBlumberg and Evans, 1998). Furthermore, we have not been able to detect a physical interaction between TLX and PAX6 (unpublished data). It seems therefore that these two genes regulate telencephalic patterning through independent but convergent pathways. The convergence of Tlx and Pax6 to pattern the pallio-subpallial boundary could be mediated through the regulation of common gene targets. Pax6 has been implicated in the regulation ofNgn2 expression in the pallium(Stoykova et al., 2000;Toresson et al., 2000;Yun et al., 2001). Recently,an enhancer, which is capable of driving Ngn2 expression in the pallium, including the ventral pallium, has been identified(Scardigli et al., 2001). This enhancer was shown to require Pax6 gene function for its correct expression. Interestingly, this enhancer element contains a putative TLX binding site (J. S., K. C. and F. Guillemot, unpublished data), suggesting that TLX as well as PAX6 may be involved in directly regulating Ngn2gene expression in the ventral pallial region. Such a regulation may explain,at least in part, the Tlx mutant phenotype, becauseNeurogenins have previously been shown to negatively regulate subpallial (e.g. Mash1 and Dlx genes) gene expression(Fode et al., 2000). Interestingly, co-regulation of a common gene by TLX and PAX6 has been shown to occur in eye development. The paired homeobox genes Pax6 andPax2 are expressed in the developing retina and optic stalk,respectively. These factors regulate the development of these two eye regions,in part, through direct mutual repression (i.e. PAX6 represses Pax2gene expression in the retina and vice versa)(Schwarz et al., 2000). Moreover, the Pax2 promoter contains a functional TLX binding site,which, when bound by TLX, results in the repression of promoter activity(Yu et al., 2000). It is interesting to note that Tlx is expressed in both the retina and the optic stalk (Yu et al., 1994;Monaghan et al., 1995;Yu et al., 2000), and yet it is proposed to participate in patterning the retina-optic stalk transition(Yu et al., 2000). This is similar to the data presented here in which Tlx is expressed on both sides of the pallio-subpallial boundary but is involved in the establishment of this boundary. The present results indicate that this function is dependent on a genetic interaction with the pallial-enriched Pax6 gene.
We thank Richard Converse for excellent technical assistance. We also gratefully acknowledge the kind gifts of probes and antibodies from D. Anderson, A. Buchberg, T. Edlund, P. Emson, M. Götz, J. Johnson, T. Jessell, S. Morton, R. McKay and G. Panganiban. This work was supported by funds from the Children's Hospital Research Foundation and the Human Frontiers Science Program (RG160-2000B).