Pax6 controls the expression of Lewis x epitope in the embryonic forebrain by regulating alpha 1,3-fucosyltransferase IX expression.

Pax6 is a transcription factor involved in brain patterning and neurogenesis. Expression of Pax6 is specifically observed in the developing cerebral cortex, where Lewis x epitope that is thought to play important roles in cell interactions is colocalized. Here we examined whether Pax6 regulates localization of Lewis x using Pax6 mutant rat embryos. The Lewis x epitope disappeared in the Pax6 mutant cortex, and activity of alpha1,3-fucosyltransferase, which catalyzed the last step of Lewis x biosynthesis, drastically decreased in the mutant cortex as compared with the wild type. Furthermore, expression of a fucosyltransferase gene, FucT-IX, specifically decreased in the mutant, while no change was seen for expression of another fucosyltransferase gene, FucT-IV. These results strongly suggest that Pax6 controls Lewis x expression in the embryonic brain by regulating FucT-IX gene expression.

The development of the central nervous system requires the specification of distinct regions, which presumably results from spatiotemporally restricted expression of various regulatory genes, including Pax6. Pax6 is expressed in several discrete domains of the developing central nervous system (Refs. 1 and 2, see also review by Osumi (3)). Pax6 protein is a transcription factor containing two DNA-binding motifs (a paired domain and a paired-type homeodomain) (4). Molecular and morphological analyses of the Pax6 mutant mice and rats suggest that Pax6 is crucial for the normal development of the forebrain (5)(6)(7)(8)(9)(10)(11)(12)(13)(14). Pax6 is expressed in the cortex but not in the striatum of the telencephalon (1,2). Pax6 mutant fails to establish the boundary between the cortical and striatal regions (6, 8, 14 -18). It has been proposed that Ca 2ϩ -dependent selective adhesion of the cortical cells segregating from the striatal cells contributes to the formation and maintenance of boundaries between these telencephalic regions (19). In Pax6 mutant, this selective adhesion of cortical cells is lost (16) and neural cell migration from the striatum into the cortex is strongly enhanced (20).
It has been reported that the cortex of the rat telencephalon at embryonic days 12 to 15 (E12-15) 1 distinctively expresses Lewis x epitope (19,21,22). The Lewis x epitope, which is also known as CD15 and SSEA-1 (stage-specific embryonic antigen-1), has been identified as a glycan epitope with a Gal␤1-4(Fuc␣1-3)GlcNAc␤1-structure (23)(24)(25). It has been proposed that Lewis x glycans interact with each other in a Ca 2ϩ -dependent manner (26,27). The expression of the Lewis x epitope is highly regulated during embryogenesis. The epitope appears in the mouse embryo at the morula stage and decreases rapidly after compaction (28), and it has been demonstrated that the multivalent Lewis x epitope induces decompaction of the embryo (29). Several immunohistochemical and biochemical studies have demonstrated that the expression of Lewis x epitope is spatiotemporally regulated in the developing central nervous system (21, 30 -34). However, the mechanism by which Lewis x epitope expression is regulated during embryogenesis remains unknown.
In this study, we examined the relationship between Pax6 and the expression of Lewis x epitope in the developing nervous system. Our preliminary observations led us to hypothesize that Pax6 regulates the expression of Lewis x epitope by controlling ␣3FucT activity. To test this hypothesis, we investigated the expression of Lewis x epitope and ␣3FucT activity in the rat E13.5 telencephalon of the wild type and the Pax6 mutant. By investigating the expression of FucT-IV and FucT-IX in E13.5 telencephalon, we demonstrated that the expression of FucT-IX was decreased in the Pax6 mutant. Based on these results, we proposed that one of the mechanisms that control the expression of Lewis x epitope is through regulating the gene expression of FucT-IX by Pax6.

EXPERIMENTAL PROCEDURES
Animals-Rat Small eye (rSey2) strain was used as Pax6 mutant (50), and Sprague-Dawley strain was used as the wild type. Homozygous embryos were obtained from intercrosses of heterozygous rSey2 rats. The day when the vaginal plug was found was designated as embryonic day 0 (E0). Homozygous embryos could easily be distinguished from those of the wild type from their external features; the former lacked the eye and nose primordia.
Immunoprecipitation and Immunoblotting-Proteins were extracted from the telencephalon dissected from three embryos at E13.5. The specimens were homogenized by sonication for 15 s in 2 volumes of ice-cold homogenization buffer (50 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 0.5% Triton X-100, plus protease inhibitor mixture Complete, Roche Molecular Biochemicals) and centrifuged at 10,000 ϫ g for 3 min. To the resulting supernatant, 5 g of anti-Lewis x antibody (clone 73-30; Seikagaku), 2.5 g of rat anti-mouse IgM (Zymed Laboratories Inc.), and 5 l of protein G-Sepharose beads (Amersham Biosciences Inc.) were added, and the solution was rotated overnight at 4°C. The beads were then washed three times with homogenization buffer. Bound protein was released from beads by incubation in homogenization buffer containing 0.5% SDS and 1% 2-mercaptoethanol for 30 min. To a 10-l aliquot of the immunoprecipitate was added 10 l of chondroitinase ABC protease free (0.1 unit; Seikagaku), endo-␤-galactosidase (10 milliunits; Seikagaku), glycopeptidase A (0.5 milliunits; Seikagaku), or homogenization buffer. After incubating overnight at 37°C, the samples were diluted 1:1 with 2 ϫ reducing sample buffer and subjected to SDS-PAGE and immunoblotting using anti-Lewis x antibody (clone TÜ 9; Quartett) as described previously (54). Cross-linked phosphorylase b (Sigma) was used as the molecular weight standard.
Thin-layer Chromatography and Immunostaining-Lipids were extracted from the telencephalon dissected from six embryos at E13.5. The specimens were homogenized by sonication for 15 s in distilled water. Neutral glycolipids were isolated from the homogenate and subjected to high performance thin-layer chromatography (HPTLC), followed by immunostaining with anti-Lewis x antibody (clone 73-30; Seikagaku), as described previously (54).
Cloning of Rat FucT-IV and FucT-IX-Rat FucT-IV cDNA fragment was obtained by PCR in the presence of 5% dimethyl sulfoxide using rat genomic DNA (CLONTECH) as a template. Based on the previously reported sequence of rat FucT-IV gene (56), oligonucleotides used to amplify the cDNA were as follows: 5Ј-GCCCGCCCCCTCTATG-3Ј and 5Ј-AATAGCCTCATCGCTGGAACC-3Ј. The PCR product was subcloned into the pGEM-T Easy vector (Promega) and the subcloned fragments were sequenced. The sequence of the PCR product revealed some discrepancies (7 bases in nucleotide and the resultant 4 amino acid changes) with the previously reported sequence (GenBank TM accession number U58860). Although the reason for the discrepancies is not clear, we used this sequence as the rat FucT-IV gene in this study. The FucT-IV fragment thus obtained was cloned into pBluescript II SKϪ (Stratagene).
A partial cDNA of rat FucT-IX was obtained by PCR using rat whole brain cDNA as a template. Based on the sequences of mouse and human FucT-IX genes (44,49), oligonucleotides used to amplify the cDNA were as follows: 5Ј-ATGACATCAACATCCAAAGGCATT-3Ј and 5Ј-ACCAA-CAGACTTATATTCTTGATGCC-3Ј. The PCR product was subcloned into the pBluescript SKϪ, and the subcloned fragments were sequenced. To obtain both the 5Ј-and 3Ј-sequence of rat FucT-IX, rapid amplification of cDNA ends was employed using rat cerebellum Marathon Ready cDNA (CLONTECH). The products of PCR using Advantage cDNA Polymerase Mix (CLONTECH) were subcloned into pGEM-T Easy vector and sequenced. While this manuscript was in preparation, the sequence was also reported by others (57). Based on the obtained sequences, the DNA fragment encoding the full-length open reading frame of FucT-IX was amplified by PCR using rat genomic DNA (CLONTECH) as a template. Oligonucleotides used to amplify the cDNA were as follows: 5Ј-CCCAAGCTTCTCTACCGTGAAAAAT-TATG-3Ј containing HindIII site (underlined) and 5Ј-GCTCTAGAGT-GACGATGATGGACATTTTAA-3Ј containing XbaI site (underlined). After restriction digestion, the PCR product was subcloned into pBluescript II SKϪ.
Enzymatic Activity of Recombinant Fucosyltransferases-Rat FucT-IV and FucT-IX fragments were subcloned into pCDM8 vector. COS-7 cells (1 ϫ 10 6 cells) were transfected with one of the expression plasmids using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. After 3 days, the cells were harvested, washed with phosphate-buffered saline, homogenized in 100 l of 0.3% Triton X-100, and used as the enzyme source for the ␣1,3-fucosyltransferase assay.
Quantification of mRNA by Real-time Detection-PCR-Real-time detection-PCR was performed using a set of PCR primers and a probe complementary to the sequence located in nonconserved regions between FucT-IV and FucT-IX. Oligonucleotides used were as follows: FucT-IV, 5Ј-GTTTGAGAACTCACAGCACGT-3Ј, 5Ј-CCACGTTTCGGT-CGAG-3Ј; the TaqMan probe, 5Ј-ACATCACTGAGAAGCTGTGGCG-3Ј; FucT-IX, 5Ј-GGGCAGACCTTTGACCTTACAT-3Ј, 5Ј-ACGGCGATAAG-TTAGAGTCAGGT-3Ј; and the TaqMan probe, 5Ј-TGCCAAGCCATGT-TCAACATCC-3Ј. Each TaqMan probe was attached with the reporter dye FAM to the 5Ј end and the quencher dye TAMRA to the 3Ј end. Standard RNAs were synthesized using FucT-IV or FucT-IX cloned into pBluescript II as a template and T7 RNA polymerase and MEGAscript in vitro transcription kits (Ambion). Synthetic RNA was treated with DNase I and purified using an RNeasy column (Qiagen). The possibility of contamination of the DNA template in the purified RNA was eliminated by PCR without reverse transcription. Synthetic RNA copy numbers were calculated from the quantity and molecular weight of RNA according to conventional methods. Real-time detection-PCR was performed using a GeneAmp Ez rTth RNA PCR kit and an ABI PRISM 7700 Sequence Detector (PE Biosystems) as described previously (58), except that the reaction mixture contained 50 nM forward primer and 300 nM reverse primer.

Expression of Lewis x Epitope on Telencephalic Cortex Disappears in the Pax6
Mutant-Whole-mount in situ staining of the dissected brain revealed Pax6 mRNA expression in the E13.5 rat dorsal telencephalon (Fig. 1F), as was observed in mouse embryos (1,2,6). Immunohistochemistry using anti-Pax6 antibody showed intense staining in the ventricular zone (VZ) of the telencephalic cortex and dorsal thalamus of the wild type (Fig. 1C). We examined the expression of Lewis x epitope in the telencephalon of both wild type and Pax6 mutant embryos. The VZ of the telencephalic cortex was stained with anti-Lewis x antibody in wild type E13.5 embryo (Fig. 1A), as was reported in the E12-15 rat forebrain (19,21,22). The expression of Lewis x epitope was not detected in the dorsal thalamus where Pax6 staining was positive, but was detected in the lateral and medial ganglionic eminences where Pax6 staining was negative (Fig. 1, A and C). Thus the expression patterns of Pax6 and Lewis x epitope partially overlapped in the wild type telencephalon. In the homozygous embryo, however, little staining with anti-Lewis x antibody was observed in the telencephalic cortex (Fig. 1D). In contrast with Lewis x, no differences in the expression of HNK-1 epitope in the outer layer of the cortex and the broad area of the basal ganglia (51) were observed between the wild type and homozygous embryos (Fig. 1, B and E). These results suggest that the expression of Lewis x epitope in the telencephalic cortex is regulated by Pax6.
Lewis x Epitope on Proteoglycan Disappears in the Pax6 Mutant-Lewis x epitope has a Gal␤1-4(Fuc␣1-3)GlcNAc␤1structure at the terminus of the glycan and is present on some glycoconjugates (glycoproteins, glycolipids, proteoglycans). The following two mechanisms may account for our immunohistochemical results of Lewis x (Fig. 1, A and D): 1) the carrier glycoconjugate itself disappears in the homozygote or 2) the glycan on the existing carrier in the homozygote differs from the Lewis x-positive one in the wild type. To determine the molecular basis of the regulation of Lewis x expression by Pax6, we searched for Lewis x glycan-bearing materials that exist in the wild type tissue but not in the homozygous brain.
Proteinaceous materials from E13.5 telencephalon were immunoprecipitated with anti-Lewis x antibody (clone 73-30) and subjected to immunoblotting analysis using anti-Lewis x antibody (clone TÜ 9). A positive band was observed at the position corresponding to ϳ500 kDa in the immunoprecipitate from the wild type but not from the homozygote (Fig. 2A). The mobility of the positive band was increased to the position corresponding to ϳ400 kDa upon treatment with chondroitinase ABC, which catalyzes the removal of chondroitin sulfate side chains from proteoglycans (Fig. 2B, lane 2). The positive band disappeared after treatment with endo-␤-galactosidase which hydrolyzes ␤-galactoside linkage in poly-N-acetyllactosamine (Fig.  2B, lane 3) or peptideglycanase A, which acts on N-linked glycan (Fig. 2B, lane 4). These results indicate that the Lewis x-positive protein is a chondroitin sulfate proteoglycan (CSPG) and that the Lewis x epitope is present on a poly-N-acetyllactosamine structure on the N-linked glycan on the CSPG. While these features are common to phosphacan/receptor-type protein-tyrosine phosphatase-/␤ (RPTP-/␤) (22,59,60), it has been suggested on the basis of immunoblotting and immunohistochemistry using anti-phosphacan antibodies that Lewis x-CSPG is probably distinct from phosphacan/RPTP-/␤. 2,3 The CSPG has yet to be identified.
Neutral glycolipids extracted from the E13.5 telencephalon were subjected to HPTLC and stained with anti-Lewis x antibody (clone . Lewis x-positive bands were observed in material obtained from both the homozygote and the wild type (Fig. 2C). The simplest form of Lewis x-glycolipid was not detected. It is likely that Lewis x-glycolipids in the E13.5 telencephalon have a more complex structure, with the major one probably being Gal␤1-4(Fuc␣1-3)GlcNAc␤1-3Gal␤1-4Glc-NAc␤1-3Gal␤1-4Glc␤1-1ЈCer as described for rat E15 cerebral cortex by Chou et al. (34).
␣1,3-Fucosyltransferase Activity in Telencephalic Cortex Decreases in the Pax6 Mutant-As described above, Lewis x-bearing CSPG was detected in the wild type but not in the homozygote, whereas Lewis x-glycolipid was detected in both the homozygote and the wild type. The most reasonable explanation of the disappearance of Lewis x epitope in our immunohistochemical analysis (Fig. 1B) is that the expression of the core protein of the CSPG in the homozygote is down-regulated. Another possibility is the abnormal expression of a type of glycosyltransferase that is specific to, or preferential for, glycan on the CSPG rather than glycolipid.
Since Lewis x glycan is synthesized from the N-acetyllactosamine (Gal␤1-4GlcNAc) structure at the terminus of the glycan on glycoconjugate to which fucose is added as the final step by ␣1,3-fucosyltransferase (␣3FucT), we assayed ␣3FucT activity in the E13.5 telencephalon. As acceptor substrates, we used a PA-glycan and a glycolipid, both of which have an N-acetyllactosamine structure at the terminus. The telence- phalic cortex, which expresses Lewis x epitope (Fig. 1A), was dissected from the embryo and used as an enzyme source. ␣3FucT activities toward both PA-glycan and glycolipid were detected in E13.5 cortex (Fig. 3).
One of the simplest hypotheses that may account for the Lewis x immunohistochemistry results (Fig. 1, A and D) is that ␣3FucT activity in the homozygote decreases, compared with the wild type. To test this hypothesis, we analyzed ␣3FucT activity in the telencephalic cortex dissected from wild type and homozygous embryos and compared their specific activities. The ␣3FucT activities in the homozygote toward PA-glycan (Fig. 4A) and glycolipid (Fig. 4B) decreased to 16 and 33%, respectively, of those of the wild type. As a control, we also analyzed the activity of ␤1,4-galactosyltransferase (␤4GalT) that catalyzes the formation of the N-acetyllactosamine structure, a precursor of Lewis x glycan. The ␤4GalT activities toward both PA-glycan and glycolipid showed little difference between the wild type and homozygous embryos (Fig. 4, A and  B). These results suggest that Pax6 may regulate ␣3FucT activity in the E13.5 telencephalon. As the telencephalic cortex showed little synthesis of sialyl Lewis x glycan in contrast with Lewis x glycan (data not shown; described for human brain by Mollicone et al. (61)), it is considered that the enzymatic properties of ␣3FucT in the rat cortex are similar to those of FucT-IV and FucT-IX in the other mammalian ␣3FucT family (see Introduction). These data suggest that Pax6 may control the activity of FucT-IV or FucT-IX in the rat embryonic brain, possibly by regulating the gene expression.
Cloning and Expression of Rat FucT-IV and FucT-IX-To examine the expression of FucT-IV and FucT-IX in the embry-onic rat brain, we first cloned rat FucT-IV and FucT-IX cDNAs using a PCR-based approach. The nucleotide sequences obtained in this study were registered in DDBJ/GenBank TM /EBI (accession numbers: AB049938 (FucT-IV) and AB049819 (FucT-IX)).
To determine the enzymatic properties of the two cloned rat ␣3FucTs, we analyzed the ␣3FucT activity of COS-7 cells transfected with FucT-IV or FucT-IX. Both homogenates of these transfectants showed ␣3FucT activities toward both PA-glycan and glycolipid (Table I, see also Figs. 2C and 3C). Cells transfected with mock vector showed little activity (Table I and Fig.  3C). These results indicate that both rat FucT-IV and FucT-IX synthesize Lewis x epitope. Interestingly, the specific activity of FucT-IX-transfectant toward PA-glycan was 4-fold that of FucT-IV-transfectant, while their specific activities toward glycolipid were very similar (Table I). This observation suggests that FucT-IX may act on soluble acceptor substrate more efficiently than FucT-IV.
Expression of FucT-IX Decreases in the Pax6 Mutant-As mentioned above, the expression of FucT-IV and FucT-IX may be regulated by Pax6. Thus, using conventional reverse transcription-PCR, we examined whether FucT-IV and FucT-IX are expressed in the E13.5 rat telencephalon. Both FucT-IV and FucT-IX products were amplified using telencephalic RNA of both the homozygote and the wild type (data not shown). We then investigated the localization and quantification of their transcripts in the telencephalon. In situ hybridization analysis showed a striking contrast between the expression patterns of FucT-IV and FucT-IX in the E13.5 telencephalon (Fig. 5). Weak FucT-IV expression was detected in the broad area of the telencephalon in either the wild type or the homozygote (Fig. 5,  D and H). On the other hand, FucT-IX expression was localized in the VZ of the cortex and dorsal half of the medial wall in the wild type telencephalon (Fig. 5C), where FucT-IX expression was co-localized with Pax6 ( Fig. 5A; see also Fig. 1C). In the dorsal thalamus where Pax6 was expressed, however, little FucT-IX expression was detected. Interestingly, FucT-IX expression was strongly reduced in the telencephalic cortex in the homozygote (Fig. 5G).
To quantify the transcripts of FucT-IX and FucT-IV in the telencephalic cortex, we used a real-time detection-PCR system. The sensitivity and linearity of the assay were examined using synthetic FucT-IX and FucT-IV RNA (data not shown). The quantities of FucT-IX, FucT-IV, and GAPDH (internal standard) transcripts in the telencephalic cortex were determined using three or four each of wild type and homozygous embryos. The relative amounts of FucT-IX and FucT-IV tran- scripts were expressed as copy numbers relative to that of GAPDH transcript, which was taken as 1.0 ϫ 10 5 (Fig. 6). The amount of FucT-IX transcript in the homozygote was about three times less than that of the wild type, while little difference was noted for the values of FucT-IV. This result is consistent with that obtained by in situ hybridization, and both results strongly suggest that Pax6 regulates the gene expression of FucT-IX in the rat embryonic brain.

DISCUSSION
In this study, we demonstrated that the Lewis x epitope, which is localized in E13.5 rat telencephalic cortex, disappears in the Pax6 homozygous mutant. Enzymatic activity of ␣3FucT, which catalyzes the final step of Lewis x biosynthesis, decreased in the homozygote, compared with the wild type. The telencephalic cortex that expressed Pax6 co-expressed FucT-IX, a member of the ␣3FucT family. Moreover, the FucT-IX expression also decreased in the Pax6 mutant. These results suggest that a functional Pax6 regulates, either directly or indirectly, the gene expression of FucT-IX and, as a consequence, the localization of Lewis x epitope. The transcripts of FucT-IX, however, were detected in the homozygote by reverse transcription-PCR as one third of the wild type (Fig. 6A). On the other hand, FucT-IX was not detected by in situ hybridization in the wild type dorsal thalamus where Pax6 is expressed (data not shown). It seems likely that distinct transcription factor(s) may regulate the expression of FucT-IX gene at a basal level and that Pax6 can act synergically to promote FucT-IX expression to a functional level. It has been reported that a Pax6 protein binds to the promoter region of L1 (62, 63), ␣A-crystallin (64), or N-CAM gene (65). The recognition motifs, which are necessary for binding to Pax6 protein, have been reported (63)(64)(65)(66). To elucidate the regulatory mechanism of the FucT-IX gene expression, we are now analyzing the 5Ј-flanking region of the FucT-IX gene to search for the reported Pax6binding sequences and to identify regulatory elements that bind to transcription factors.
The amount of FucT-IX mRNA in the telencephalic cortex of the wild type was 3-fold that of the homozygote (Fig. 6A). FucT-IV, another member of the ␣3FucT family, was also expressed in the telencephalic cortex, but the expression level of its transcript was not affected by the Pax6 mutation (Fig. 6B). ␣3FucT activity toward PA-glycan in the wild type was 6-fold that of the homozygote (Fig. 4A). It is likely that the ␣3FucT activity measured in this study includes both FucT-IX and FucT-IV activities. The difference between the wild type and homozygote in the activities on glycolipid was less than that on PA-glycan (Fig. 4, A and B). This finding is consistent with the result obtained using recombinant enzymes; FucT-IX acts on PA-glycan more efficiently than FucT-IV, compared with their actions on glycolipid (Table I). The net increase of FucT-IX activity of the wild type compared with the homozygote, therefore, must be more than 6-fold. A 3-fold increase in the expression level of the transcript resulted in at least a 6-fold increase in the enzyme activity. This result was probably due to both increases of a translation rate and a turnover time of the transcript. Similarly, it is likely that the increase in the enzyme activity, which was at least 6-fold more, increased the production of the epitope, which resulted in the expression of the Lewis x epitope on the CSPG in the wild type ( Fig. 2A). Thus, the immunohistochemical detection of Lewis x epitope in the wild type (Fig. 1A) may be attributable to the appearance of the epitope on the CSPG.
It has been reported that in the E14.5 mouse forebrain cortex, no obvious change in Lewis x immunoreactivity was observed between the Pax6 mutant and the wild type (16). A possible explanation is that the Pax6 target genes in the cerebral cortex may differ between the mouse and rat and that Pax6 does not regulate the expression of either FucT-IX or Lewis x in the mouse. It is also possible that the differences in the results between this and previous studies may be due to the different antibody specificity of anti-Lewis x antibodies used. In our study, a similar variability in antibody specificity was also observed in Pax6 mutant rat embryos. In the telencephalon of E13.5 Pax6 mutant rat embryos, Lewis x expression was not observed in the immunohistochemistry (Figs. 1D and 5F), whereas Lewis x glycolipids were detected in the HPTLC immunostaining (Fig. 2C). In support of this notion, different results were also obtained in other similar studies using different anti-Lewis x antibodies (e.g. Refs. 21, 33, 67, and 68). The exact specificity of the antibodies is likely to depend on the carrier of the Lewis x epitope; i.e. whether the epitope is on the protein or lipid, and the length, type, and modification of the core glycan. In our immunohistochemical analysis, the anti-Lewis x antibody (clone 73-30) appeared to preferentially detect the epitope on the CSPG rather than that on the glycolipid. However, the anti-Lewis x antibody used in other studies of Sey mice (clone RB11.2 in Ref. 16) may have preferentially detected the epitope on the glycolipid.
We showed that Lewis x-bearing CSPG was present in the telencephalic cortex in the wild type but not in the Pax6 mutant (Fig. 2). The core protein of the CSPG, however, remains to be identified. Therefore, we cannot exclude the possibility that Pax6 also regulates the expression of the core protein of the CSPG. Phosphacan/RPTP-/␤, a CSPG expressed in the VZ of developing brain, binds with high affinity to the neural cell adhesion molecules, L1 and N-CAM, and the extracellular matrix protein tenascin-C (69,70). This binding is mediated by N-linked glycans on the CSPG (71). In the mouse mesencephalon, it is suggested that the interaction between phosphacan and L1 is involved in the neuronal cell migration (72). Since only a subpopulation of phosphacan bears the Lewis x epitope (73), this finding suggests that the expression of Lewis x on phosphacan/RPTP-/␤ may affect cell migration by modulating the association of the CSPG with cell adhesion molecules. Similarly, the Lewis x-CSPG present in the telencephalic cortex may also be implicated in cell migration. In Pax6 mutant rat, defect in radial migration, which was shown to be not cellautonomous, was observed in the later-born cortical precursor cells (9). This observation suggests that Pax6 affects a cortical environment. The disappearance of the Lewis x epitope from the CSPG, therefore, can be considered to be one of the molecular bases for the abnormal cortical environment in the Pax6 mutant.
It has been proposed that cell interactions via Lewis x epitope in pre-implantation embryos are mediated by the interaction of Lewis x glycans (26,27). The interaction between Lewis x and Lewis x glycan is Ca 2ϩ -dependent, and it has been suggested that it may cause autoaggregation of Lewis x-expressing cells (26,27). It has also been demonstrated that cortical cells from the telencephalon aggregate with each other but segregate from striatal cells from the telencephalon (16,19). This selective adhesion is Ca 2ϩ -dependent (19) and is lost in Pax6 mutant (16). Taken together, it is considered that the selective adhesion of cells of the telencephalic cortex may be mediated, at least in part, by homophilic interaction of Lewis x glycan that is present on the CSPG. Lewis x glycan may act synergistically with R-cadherin, which has been shown to be involved in region-specific cell adhesion (74) and has been shown to disappear in the telencephalic cortex of Pax6 mutant (16). In pre-implantation embryos, both Lewis x glycan and E-cadherin are involved in the compaction process (75,76). It has been suggested that glycan-glycan interaction takes place more rapidly than other species of intermolecular interactions and, although highly specific, is weaker than other interactions (77). The Lewis x epitope on the CSPG may act in the initial step of the cortex-specific cell adhesion in the telencephalon. Leukocyte adhesion deficiency type II (LAD II) is a rare inherited disease caused by a metabolic disorder of GDP-fucose that results in hypofucosylation of glycoconjugates (reviewed in Refs. 78 and 79). LAD II is characterized by recurrent infections and leukocytosis and patients are reported to exhibit mental retardation and numerous facial abnormalities. These symptoms suggest that fucose-containing glycoconjugates, including Lewis x-carrying molecules, may play important roles in craniofacial morphogenesis and the development of brain function. It is well known that the Pax6 mutant mice and rats show craniofacial defects, i.e. the small eye phenotype in heterozygotes and the absence of eyes and nose in homozygotes (80,81). It has been reported that in a family with inherited aniridia, only individuals with Pax6 mutation showed abnormalities in frontal lobe function (82). In patients with schizophrenia, it has been reported that the incidence of the high activity variant of the Pax6 promoter is higher in patients with the paranoid subtype than in the control (83). These observations may implicate a relationship between Lewis x epitope and Pax6 during development of the brain and face.
In conclusion, the findings of this study suggest that the expression of Lewis x epitope in the embryonic brain may be regulated by Pax6 via the expression of the FucT-IX gene. Based on these findings, we have now grasped a link between the regulatory gene and the carbohydrate epitope, which is expressed in a spatiotemporally regulated pattern. LGE, lateral ganglionic eminence; MGE, medial ganglionic eminence. Scale bar, 100 m.

FIG. 6. Quantitative analysis of transcripts of FucT-IX (A) and
FucT-IV (B) by the real-time detection-PCR method. The telencephalic cortex RNA of wild type (closed bars) or homozygote (open bars) was used as a template, and reverse transcription-PCR was performed using rTth DNA polymerase as described under "Experimental Procedures." Values are expressed as copy numbers relative to that of GAPDH transcript, which was taken as 1.0 ϫ 10 5 . Each histogram indicates the mean value Ϯ S.E.