INTRODUCTION

Much attention has been paid to molecular mechanismsunderlying the regionalization of the central nervous system.Regionalization of the optic tectum, which differentiates fromthe alar plate of the mesencephalon, has been well analyzed.In the early stage of chick embryos, En1 and Pax2areexpressed in the entire mesencephalon down to the isthmus(Gardner and Barald, 1992; Okafuji et al., 1999; Shamim et al.,1999). Misexpression of either En1or Pax2 in thediencephalon results in the fate change of the presumptivediencephalon to the tectum (Araki and Nakamura, 1999;Okafuji et al., 1999). Mice mutant in either En1or Pax2showdefects in the mesencephalon (Wrust et al., 1994; Favor et al.,1996; Schwarz et al., 1997, 1999; Urbánek et al., 1997). Otx2is expressed in the rostral neural tube down to the isthmus(Simeone et al., 1992; Bally-Cuif et al., 1995; Millet et al.,1996). Otx2 mutant mice lack the prosencephalon andmesencephalon (Acampora et al., 1995). Misexpression ofOtx2 in the metencephalon changes the fate of the alar plate tothe tectum (Broccoli et al., 1999; Katahira et al., 2000).

It is accepted that isthmic region acts as an organizer oftectum formation, and Fgf8 is thought to play a key role(Alvarado-Mallart, 1993; Marin and Pulles, 1994; Crossley etal., 1996; Joyner, 1996). An Fgf8-soaked bead implanted intothe diencephalon causes transformation of the presumptivediencephalon to the tectum (Crossley et al., 1996). It was

recently shown that Fgf8, Pax2/5 and Enform the positivefeedback loop for their expressions (Song et al., 1996; Lun andBrand, 1998; Araki and Nakamura, 1999; Funahashi et al.,1999; Okafuji et al., 1999; Shamim et al., 1999). This feedbackloop may maintain the differentiated state of the tectum andcontribute to its rostrocaudal polarity (Lee et al., 1997; Pickeret al., 1999).

Otx2 is expressed in the prosencephalon andmesencephalon, while Gbx2is expressed in the metencephalon(Nis and Leutz, 1998; Shamim and Mason, 1998; Hidalgo-Sanchez et al., 1999). The caudal limit of Otx2expressioncorresponds to that of the tectum (Bally-Cuif et al., 1995;Millet et al., 1996). Misexpression of Otx2 in themetencephalon resulted in a fate change of the alar plate to thetectum (Broccoli et al., 1999; Katahira et al., 2000). In contrast,misexpression of Gbx2 in the mesencephalon caused rostralshift of the caudal limit of the tectum (Millet et al., 1999;Katahira et al., 2000). Since expression domains ofOtx2 andGbx2 overlap and Otx2 and Gbx2 repress each other’sexpression, it was proposed that repressive interaction betweenOtx2 and Gbx2may determine the caudal limit of the tectum(Broccoli et al., 1999; Millet et al., 1999; Katahira et al., 2000).

Pax6, which is expressed in the prosencephalon, is essentialfor the development of the diencephalon (Walther and Gruss1991; Bally-Cuif et al., 1994; Stoykova et al., 1996, 1997;Grindley et al., 1997; Mastick et al., 1997; Warren et al., 1997).Pax6mutant mice, Sey, show fate change of the pretectum, the

2357Development 127, 2357-2365 (2000)Printed in Great Britain © The Company of Biologists Limited 2000DEV1532

Transcriptional factors and signaling molecules areresponsible for regionalization of the central nervoussystem. In the early stage of neural development, Pax6 isexpressed in the prosencephalon, while En1and Pax2areexpressed in the mesencephalon. Here, we misexpressedPax6 in the mesencephalon to elucidate the mechanism ofthe di-mesencephalic boundary formation. Histologicalanalysis, expression patterns of diencephalic marker genes,and fiber trajectory of the posterior commissure indicatedthat Pax6 misexpression caused a caudal shift of the di-mesencephalic boundary. Pax6 repressed En1, Pax2 andother tectum (mesencephalon)-related genes such as En2,Pax5, Pax7, but induced Tcf4, a diencephalon marker gene.To know how Pax6 represses En1 and Pax2, we ectopically

expressed a dominant-active or negative form of Pax6. Thedominant-active form of Pax6 showed a similar but moresevere phenotype than Pax6, while the dominant-negativeform showed an opposite phenotype, suggesting thatPax6 acts as a transcriptional activator. Thus Pax6 mayrepress tectum-related genes by activating an interveningrepressor. The results of misexpression experiments,together with normal expression patterns of Pax6, En1andPax2, suggest that repressive interaction between Pax6andEn1/Pax2defines the di-mesencephalic boundary.

Key words: Pax6, En1, Pax2, Di-mesencephalic boundary, Chick,Inovoelectroporation

SUMMARY

Pax6 defines the di-mesencephalic boundary by repressing En1 and Pax2

Eiji Matsunaga, Isato Araki ‡ and Harukazu Nakamura*

Department of Molecular Neurobiology, Institute of Development, Aging and Cancer, Tohoku University, Seiryo-machi 4-1, Aoba-ku, Sendai 980-8575, Japan‡Present address: Department of Neurobiology, University of Heidelberg, IM Neuenheimer Feld 364, D-69120 Heidelberg, Germany*Author for correspondence (e-mail nakamura@idac.tohoku.ac.jp)

Accepted 9 March; published on WWW 10 May 2000

2358

caudal part of the diencephalon, to the mesencephalon(Mastick et al., 1997). En1 or Pax2misexpression repressedPax6and caused the fate change of the dorsal diencephalon tothe tectum (Araki and Nakamura 1999; Okafuji et al., 1999).We thus presumed that Pax6 represses En1and Pax2expression, and that repressive interaction between Pax6andEn1/Pax2defines the rostral limit of the tectum. To provethis assumption, we ectopically expressed Pax6in themesencephalon by in ovo electroporation (Muramatsu et al.,1997; Ogino et al., 1998; Funahashi et al., 1999; Momose etal., 1999). We showed that Pax6misexpression repressed En1and Pax2expression and caused the fate change of the rostraltectal swelling to the diencephalon (pretectum). We alsoexamined precise spatial and temporal expression patterns ofPax6, En1 and Pax2 in normal embryos. The results of Pax6misexpression, together with normal expression patterns ofPax6, En1and Pax2, suggest that repressive interaction of Pax6and En1/Pax2defines the di-mesencephalic boundary. Finallywe showed that dominant-active Pax6misexpression caused asimilar but more severe phenotype than Pax6misexpression. Itis indicated that Pax6 acts as a transcriptional activator in theformation of the di-mesencephalic boundary.

MATERIALS AND METHODS

Pax6 expression vectorThe full length chickPax6cDNA, which was kindly provided by DrK. Yasuda, was inserted in pMiwIII, a derivative of pMiwZ (Suemoriet al., 1990), which has Rous sarcoma virus enhancer and chicken β-actin promoter. Pax6 cDNA used in this experiment is the mostprevalent isoform (Cvekl and Piatigorsky, 1996), which containsentire pairedbox, homeobox and S/T/P (serin-, threonine- and proline-enriched activation domain) but does not include a 14 amino acidinsertion (5a) in the paired box.

The Pax6-VP16 construct is a fusion of a Pax6fragment(corresponding to amino acids 1-345) and a VP16fragment encodingthe activation domain (Triezenberg et al., 1998). Pax6-EnR is a fusionof Pax6 (corresponding to amino acids 1-311), an En2 fragment(corresponding to amino acids 1-120, including the repressor domain),and HA-tag. These constructs were also inserted in pMiwIII.

DiI labeling 1,1′-dioctadecyl-3,3,3′,3′-tetramethyindocarbo-cyanine perchlorate(DiI) was saturated in tetraglycol (Sigma) and injected by air pressurearound the di-mesencephalic region.

In ovo electroporationFertilized chicken eggs from a local farm were incubated at 38°C.Pax6 expression vector (3.0 µg/ml) and the green fluorescence(GFP) expression vector (pEGFP-N1, Clontech) (0.5 µg/ml) weretransfected to chick embryos by in ovo electroporation as previouslydescribed (Funahashi et al., 1999). GFP expression vectors were co-transfected to check the efficiency of electroporation. In ovoelectroporation was carried out in the embryos at stage 7-8 or stage10-11 (Hamburger and Hamilton, 1951). There were no differencesin the phenotype between the embryos electroporated at stage 7-8 andstage 10-11. Most embryos were electroporated at stage 10-11 toobtain higher survival rate of the embryos.

In situ hybridizationWhole-mount in situ hybridization was performed as described byBally-Cuif et al. (1995) or Stern (1998). In situ hybridization forsections was carried out as described by Ishii et al. (1997). Briefly,

samples were fixed in 4% paraformaldehyde in phosphate-bufferedsaline (PBS), and immersed in PBS containing 20% sucroseovernight, and embedded in OCT compound (Tissue-Tek) to makecryosections. Probes for En1, Pax2, Pax5and Pax6were describedpreviously (Funahashi et al., 1999; Okafuji et al., 1999; Araki andNakamura, 1999). An approximate 1.4 kb fragment of Lim1, whichcovers the whole open reading frame was obtained by RT-PCR fromE2.5 chick embryonic brains. A partial clone of chick Tcf4 wasisolated from a cDNA library of E3 chick brain (GenBank accessionnumber: AB040438). These fragments were inserted in pBluescriptII SK(−) (Stratagene). After linearization, digoxigenin (DIG)- orfluorescein isothiocyanate (FITC)-labeled antisence RNA wasgenerated by T3 or T7 RNA polymerase (Funahashi et al., 1999).Alkaline phosphatase (ALP)-conjugated anti-DIG or FITC goatpolyclonal antibody (Roche Molecular Biochemicals) was used. Fordouble in situ hybridization, Fast Red TR/Naphthol AS/MX (SigmaFASTTM; Sigma) was used for the detection of the first signal, and4-nitroblue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) were used for the second signal. ALP, usedin the first detection, was inactivated by incubating with 100 mmol/lglycine-HCl (pH 2.2) for about 15 minutes at room temperature. Insome cases, fast red staining was washed out in ethanol. NBTstaining was washed out by incubating in dimethylformamide (DMF)at 55°C

ImmunohistochemistryAnti-Pax6 rabbit polyclonal antibody kindly provided by Dr N. Osumi(Inoue et al., 2000), anti-Pax7 monoclonal antibody (DevelopmentalStudies Hybridoma Bank, DSHB, Kawakami et al., 1997), anti-En1monoclonal antibody, 4G11 (DSHB, Ericson et al., 1997), anti-En2monoclonal antibody, 4D9 (American Type Culture Collection, Patelet al., 1989), and anti-GAP-43 monoclonal antibody, GAP-7B10(Sigma), were used as the primary antibodies. Horseradish peroxidase(HRP)-conjugated anti-mouse IgG antibody (JacksonImmunoResearch Laboratories) was used as the second antibody. Indouble staining for Pax6 and En1 on sections, Alexa-488-conjugatedanti-rabbit IgG antibody (Molecular Probes) and Cy3-conjugated anti-mouse IgG antibody (Jackson ImmunoResearch Laboratories) wereused as the second antibodies.

HistologyEmbryos were fixed in 4% paraformaldehyde in PBS, embedded inTechnovit (Kulzer), serially sectioned at 5 µm, and stained withhematoxylin-eosin. Tiling images were automatically composed byMCID Image analyzer (Imaging Research Inc).

RESULTS

Expression pattern of Pax6 and En1, Pax2Pax6 is repressed when the diencephalon transdifferentiatesinto the tectum by En or Pax2 misexpression (Araki andNakamura., 1999; Okafuji et al., 1999). Pax2/Pax5doubleknockout mice show complete deletion of the mesencephalon,and caudal shift of Pax6expression and the posteriorcommissure (Schwarz et al., 1999). Sey mutant mice show fatechange of the pretectum to the mesencephalon resulting indefects in the di-mesencephalic boundary (Mastick et al.,1997).

In normal development, Pax6expression is first detectableat the 2-somite stage (stage 7+) in the neural plate of thepresumptive prosencephalic region (Li et al., 1994). En1 andPax2 (denoted as En1/Pax2below) are expressed over thewhole of the mesencephalic territory, and have been suggestedto be involved in territory formation (Rowitch et al., 1995;

E. Matsunaga, I. Araki and H. Nakamura

2359Pax6 defines di-mesencephalic boundary

Araki and Nakamura, 1999; Okafuji et al., 1999; Shamim etal., 1999). En1 mRNA is first detectable at the 3-somite stagein the presumptive mesencephalic region (Shamim et al.,1999). Pax2expression is first detectable at the ventral portionof the prospective mes-metencephalic boundary at the 4-somitestage (Okafuji et al., 1999). We thus assumed that repressiveinteraction between Pax6and En1/Pax2played a pivotal rolein boundary formation between the diencephalon andmesencephalon. To address this issue, we first examinednormal expression patterns of Pax6, En1and Pax2, payingspecial attention to their spatial relations.

At the 6- or 7-somite stage (stage 8-9), Pax6transcripts weredetected in the prosencephalon, while En1 and Pax2mRNAswere detected in the mesencephalon. Analysis of whole mountand sections indicate that expression of Pax6and En1/Pax2overlapped around the di-mesencephalic boundary (Fig. 1A-C,E-G). To show more precisely the extent of overlappingexpression, double staining for Pax6 and En1 was carried outon a section of an 8-somite stage embryo (Fig. 1K-M).

Superimposed images clearly show that the cells at theboundary region express both Pax6 and En1 (Fig. 1M, arrows).

At stage 10, Pax6 expression was detected in theprosencephalon. En1 and Pax2 were expressed in themesencephalon and rostral metencephalon in an almostoverlapping manner, although Pax2expression was regressingfrom the rostral mesencephalon. Serial sections of embryos atthe 12-somite stage show that the expression domain of Pax6and En1/Pax2were segregated (Fig. 1H-J). Double staining forPax6 and En1 confirmed that Pax6 and En1 expressions weresegregated completely by the 11-somite stage (Fig. 1N-P).Thereafter expression domains of En1and Pax2regressed, andbecame localized in the isthmic region by E2.5 (data notshown; Gardner and Barald, 1992; Okafuji et al., 1999).

To identify the position of the presumptive di-mesencephalicboundary in the early stage, we labeled a small populationof cells at the boundary region (Fig. 1D, white arrow) andtraced it until stage 12 (Fig. 1D′, white arrow). DiI labelingexperiments clearly showed that the di-mesencephalic

Fig. 1.Spatial and temporalexpression patterns of Pax6, En1and Pax2.(A-C) Whole-mount insitu hybridization of 6-somite stageembryos for Pax6(A), En1(B) andPax2(C). Arrowheads in A-Cindicate the expression borderaround the di-mesencephalicboundary. The area where Pax6andEn1/Pax2expressions overlap isindicated by a bracket (A).(D,D′) Tracing of the di-mesencephalic boundary region byDiI. DiI was put in the di-mesencephalic boundary regionwhere Pax6and En1/Pax2expressions overlap (D, whitearrow). Tracing the same embryoshows that the boundaryretrospectively corresponds to theDiI injection site (D′, white arrow).(E-J) Serial sections of the sameembryo at the 7-somite stage (E-G),and at the 12-somite stage (H-J). Insitu hybridization for Pax6(E,H),En1(F,I) and Pax2(G,J). Expressionof Pax6and that of En1/Pax2overlap at the 7-somite stage aroundthe di-mesencephalic boundary (thearea between arrowheads in E-G).By the 12-somite stage, expressionof Pax6and that of En1/Pax2aresegregated completely (arrowheadsin H-J). (K-P) Double staining forPax6 (green) and En1 (red) at the 8-somite stage (K-M) and at the 11-somite stage (N-P). M and P are thecombined images of the boxed areasin K,L and N,O respectively, athigher magnification. The areabetween arrowheads in K and L shows the area where expression of Pax6and that of En1/Pax2overlap. Arrowheads in (N) and (O) indicateexpression border around the di-mesencephalic boundary. In M, cells that express both Pax6 and En1 are seen (white arrows), whereas in P,such double-positive cells are not seen at all indicating that Pax6 and En1 segregate completely. Samples were sectioned at 10 µm. Scale barsare 250 µm (D,D′,G,J), 100 µm (K,N) and 50 µm (M,P). pro, prosencephalon; di, diencephalon; mes, mesencephalon.

2360

boundary was formed at the site where Pax6 and En1/Pax2expressions overlap (compare Fig. 1D with 1A-C).

Caudal shift of the di-mesencephalic boundary byPax6 misexpression To examine the role of Pax6 in di-mesencephalic boundaryformation, we next misexpressed Pax6 in the presumptive di-mesencephalic region. Pax6expression vector was transfectedonly in the right side of the neural tube from the diencephalonto the metencephalon in this study (Figs 2A, 5B).

Pax6misexpression caused a morphological change in themesencephalon by 24 hours after electroporation. The size ofthe tectum on the experimental side was smaller than on thecontrol side. Size difference was most conspicuous at E3.5(n=22/27) (HH 21; 48 hours after electroporation), and thengradually diminished (data not shown).

At E5.5 (HH27, 96 hours after electroporation), tectalexpansion was somewhat smaller on the experimental sidethan that on the control side. The gross morphological positionof the di-mesencephalic boundary seemed not to be affected

(Fig. 2C). Histological examination, however, revealed thatPax6 misexpression indicated a caudal shift of the di-mesencephalic boundary (n=4/5) (Fig. 2D). At this stage, onthe control side, the rostral part of the tectum consists of athick neuroepithelial layer and two layers outside theneuroepithelium. The diencephalon consists of a thinneuroepithelial layer, a thick mantle layer and a marginal layer.On the control side, the morphological transition from thediencephalon to the tectum corresponds well to the site wheretectal swelling begins (Fig. 2D, arrow). In contrast, on theexperimental side, the most rostral part of the tectal swelling

E. Matsunaga, I. Araki and H. Nakamura

Fig. 2. Caudal shift of the di-mesencephalic boundary by Pax6misexpression. (A-D) Morphology after Pax6misexpression.Transfection was confirmed by GFP at 24 hours after electroporation(A), and the embryos were fixed at E5.5 (HH27) (B-D). Lateral view(B), dorsal view (C). Horizontal section stained with hematoxylin-eosin (D). Transfection occurred only in the right side of the neuraltube from the diencephalon to the metencephalon (A). Pax6misexpression caused caudal extension of the diencephalon (D,arrowheads). On the control side, the diencephalic structure consistsof the thin neuroepithelial layer and the thick mantle layer and therostral part of the tectal swelling consists of a thick neuroepitheliallayer and two thin layers outside the neuroepithelial layer. On theexperimental side, the rostral part of the tectal swelling consists of athin neuroepithelial layer and thick mantle layer (D, arrowheads).(E-M) In situ hybridization for Lim1(E-I) at E4.0 (HH23) andimmunohistochemical staining for Pax7 (L-M) at E6.5 (HH 29). H,Iand M are high power magnification micrographs of the areaindicated in G and L. The area indicated by arrowheads in M iswhere Pax7-positive cells are detected in the mantle layer. (F,J) Viewfrom the experimental side, (E) view from the control side, (K)dorsal view. On the control side, Lim1-positive cells are found in thepretectum, whereas on the experimental side Lim1-positive cells arefound in the rostral part of the tectal swelling (indicated byarrowheads in F and G). (H,I) These cells are in the mantle layer.(L,M) In the pretectum, Pax7 is expressed in the neuroepithelial layerand mantle layer. (M) On the experimental side, the Pax7-positivemantle layer extends more caudally. Arrows indicate the approximatesite of transition from the diencephalon to the mesencephalon, andarrowheads indicate caudal extension of the diencephalon. In allpanels except for E, the experimental side is down and rostral is tothe right. The lines in B, E and J indicate the approximate planes ofD, G and L, respectively. Scale bars are 1.0 mm (B,F,J), 500 µm(D,G,L), 250 µm (M), and 100 µm (I). cont, control side; exp,experimental side; met, metencephalon; mg, marginal layer; ml,mantle layer; ne, neuroepithelial layer; tec, tectum.

Fig. 3. The effect of Pax6misexpression on fiber trajectory of theposterior commissure. The embryo was fixed at E4.0 (HH23) andstained with anti-GAP-43 antibody. (A) Control side, (B) experimentalside, (C) dorsal view. (A) is printed as the mirror image forcomparison with B. Ectopic fibers are detected in the tectal swelling(B, arrowheads). Ectopic fibers in the rostral portion of the tectalswelling cross the roof plate (C, arrow), but fibers in the middleportion of the swelling curve caudally, to avoid the roof plate (C,arrowheads). pc, the posterior commissure. Scale bar, 0.5 mm (C).

2361Pax6 defines di-mesencephalic boundary

consisted of a thin neuroepithelial layer, a thick mantle layerand a marginal layer, indicating that the structure of thediencephalon expanded caudally (Fig. 2D, arrowheads).

Caudal shift of the di-mesencephalic boundary after Pax6misexpression was confirmed by its effect on Lim1and Pax7expression. Both genes are expressed in the pretectum so thatthese genes are good markers for the pretectum (Mastick et al.,1997; Kawakami et al., 1997). In the pretectum of the controlside at E4.0 (HH23), Lim1is expressed in the mantle layer(Fig. 2E,H). At this stage, Lim1 is not expressed in the tectum.On the experimental side, the mantle layer, which expressesLim1, extends to the tectal swelling (Fig. 2F,G,I) (n=6/9),which strongly suggests caudal shift of the di-mesencephalicboundary. Pax7 was expressed in both the tectum andpretectum region in E6.5 (HH 29), but the expression patternis different. In the pretectum, Pax7 was expressed in the mantlelayer (Fig. 2L). On the experimental side, the Pax7-positivemantle layer of the pretectum extended to the rostral part of thetectal swelling (Fig. 2M, arrowheads) (n=3/3).

Next we looked at fiber trajectory of the posteriorcommissure. The posterior commissure is an importantlandmark for the di-mesencephalic boundary (Mastick et al.,1997), and distinguished by immunostaining with anti-GAP-43antibody (Fig. 3A,B; M. Nakafuku, personal communication).On the experimental side, GAP-43-positive fibers werediscerned ectopically on the rostral mesencephalon (Fig. 3Barrowheads) (n=7/10). These ectopic fibers originated from the

ventral diencephalon, curved caudally, andextended to the rostral mesencephalon. Some ofthese fibers crossed the roof plate in the rostralregion of the tectum (Fig. 3C, arrow). Ectopicfibers near the middle region of the tectum couldnot cross the roof plate, and they curved caudallyto keep a required distance from the roof plate(Fig.3C, arrowheads). Fiber trajectory of theposterior commissure also indicates caudalextension of the diencephalon.

Down-regulation of tectum-related genesand up-regulation of Tcf4 by Pax6misexpressionWe have shown that Pax6misexpression caused acaudal shift of the di-mesencephalic boundary.Since we assume that the di-mesencephalicboundary is set through repressive interactionbetween Pax6and En1/Pax2, we further looked atthe effect of Pax6misexpression on En1and Pax2expression. In normal development, expression ofEn1 and Pax2 regress caudally, and come tolocalize in the isthmic region by E2.5. Weelectroporated embryos at stage 7-8 and examinedthe effect before the regression of En1/Pax2expression. Repression of En1 (n=7/8) and Pax2(n=9/11) was clearly detected by 12 hours afterelectroporation (Fig. 4A,B), but not detected at 6hours after electroporation (data not shown). Toexamine repression by Pax6 more precisely, wecarried out double staining for Pax6 and En1.Repression of En1 was not detected at 6 hoursafter electroporation (n=3/3) (Fig. 4D,E), butdetected at 12 hours after electroporation (n=3/3)

(Fig. 4G,H). Double staining for Pax6 and En1 revealedthat En1 was repressed in the cells in which Pax6wasmisexpressed, indicating that repression is cell-autonomous.

Next, we analyzed effects of Pax6misexpression on thetectum-related genes such as En2, Pax5 and Pax7. En2 isexpressed in the tectum in a gradient which is high caudallyand low rostrally. It has been shown that En confers caudalcharacteristics to the tectum (Itasaki et al., 1996; Logan et al.,1996; Shigetani et al., 1997). Pax5is expressed at the isthmus,and is thought to be a factor that maintains the isthmicorganizing activity (Funahashi et al., 1999). En2 (n=9/9), Pax5(n=6/6) and Pax7 (n=7/7) were repressed by Pax6misexpression (Fig. 5C,F,I). Higher magnification micrographsshow no overlapping expression of Pax6and marker genes forthe mesencephalon, suggesting that the repression is in acell-autonomous manner (Fig. 5B′,C′,E′,F′,H′,I′). Pax6misexpression also repressed Wnt1 and EphrinA2(data notshown).

The effects of Pax6misexpression on Tcf4expression wasalso examined. Tcf4 is expressed in the alar plate of thediencephalon in an early developmental phase, and a goodmarker for the dorsal diencephalon (Cho et al., 1998). At stage17, on the control side Tcf4 is specifically expressed in thedorsal diencephalon (Fig. 5J). In the experimental side, Tcf4expression was induced ectopically in the mesencephalon by24 hours after electroporation (n=5/5) (Fig. 5L), although theinduction was not detected at 12 hours after electroporation

Fig. 4.Repression of En1and Pax2by Pax6misexpression. (A,B) In situhybridization for En1and Pax2at 12 hours after electroporation. Repression of En1(A) and Pax2(B) is detected on the experimental side at 12 hours afterelectroporation by whole-mount in situ hybridization. (C-H) A cell-autonomousrepression of En1by Pax6. (C-E) Immunochemical staining for Pax6 (green) andEn1 (red) at 6 hours after electroporation, and (F-H) at 12 hours afterelectroporation. At 6 hours after electroporation (6 h.a.e.), most Pax6-positive cellsare En1-positive (E, yellow cells indicated by arrows). However, at 12 hours afterelectroporation (12 h.a.e.), most cells express either Pax6 or En1 (H). In all panels,the experimental side is down, and the rostral is to the right (A-H). Scale bars are250 µm (A,B), 100 µm (D,G) and 20 µm (E,H).

2362

(data not shown). High power magnification micrographs showthat cells ectopically expressing Tcf4 are located around cellsthat express Pax6strongly (Fig. 5K′, white arrows).

A dominant-active form of Pax6 caused more severeboundary shift It was demonstrated that Pax6 can function as a transcriptionalrepressor as well as an activator (Duncan et al., 1998). To knowwhether Pax6 represses En1and Pax2directly or indirectly byactivating some repressors in our system, we examined theeffect of dominant-active and dominant–negative forms ofPax6(Pax6-VP16, Pax6-EnR, respectively). In these constructs,the paired domain and homeodomain of Pax6 were fused withthe activation domain of VP16 (Triezenberg et al., 1988) andthe repression domain of En2 (eh1 domain) (Logan et al., 1992;Smith and Jaynes, 1996), respectively.

Dominant-active Pax6misexpression induced a similar butmore severe phenotype than Pax6misexpression. At E5.5 (HH27) (96 hours after electroporation), tectal expansion wassmaller on the experimental side (Fig. 6B). The fate changeof the rostral mesencephalon tothe diencephalon was clear; theneuroepithelial layer on theexperimental side in the rostralregion of the tectal swelling wasthinner than that on the controlside (Fig. 6C, arrowheads) (n=4/4).Pax6-VP16 also induced Tcf4expression (data not shown) andrepressed En1 (n=5/5) and Pax2(n=4/4) (Fig. 6D,E).

On the other hand, dominant-negative Pax6 (Pax6-EnR)misexpression induced an oppositephenotype to Pax6misexpression:expansion of the size of the tectum(Fig. 6G,H), rostral shift of thedi-mesencephalic boundary, andreduction of the pretectum area(n=3/3) (Fig. 6H, arrowheads).Pax6-EnR misexpression alsocaused rostral expansion of En1(n=6/12) and Pax2 (n=5/9)expression (Fig. 6I,J).

DISCUSSION

We have shown that, (1) theexpression domains of Pax6and En1/Pax2 overlap aroundthe presumptive di-mesencephalicregion at an early stage, (2) thedi-mesencephalic boundary isformed at the site where Pax6andEn1/Pax2 expression overlap, (3)Pax6 misexpression repressestectum-related genes, and causes afate change of the rostral part of thetectal swelling to the pretectaltissue, and (4) dominant-active

form of Pax6misexpression causes a similar but more severephenotype than wild-type Pax6 misexpression. The possiblerole of Pax6in the formation of the di-mesencephalic boundaryis discussed below.

Caudal extension of the diencephalon by Pax6misexpressionWe showed that Pax6 misexpression repressed the tectum-related genes. Moreover, histological examination suggestedthat the rostral part of the tectal swelling exhibited the structureof the diencephalon resulting from a caudal shift of the di-mesencephalic boundary. Labeling of the di-mesencephalicboundary by DiI after Pax6 misexpression showed that theexpansion of the diencephalic character was due to the fatechange of the anterior part of the mesencephalon, but not dueto the increased proliferation of the diencephalic cells nor tothe decreased proliferation of the mesencephalic cells (ourunpublished data). The caudal shift of the di-mesencephalicboundary was more clearly shown by the change in expressionpatterns of the pretectum-specific molecules, Lim1 and Pax7.

E. Matsunaga, I. Araki and H. Nakamura

Fig. 5.Repression of En2, Pax5and Pax7 and induction of Tcf4by Pax6misexpression. (A-C′) Double staining for Pax6(blue) and En2 (brown), (D-F′) for Pax6(red) and Pax5(blue), (G-I′) for Pax6(blue) and Pax7 (brown), and (J-K′) for Pax6(red) and Tcf4(blue).(B′,C′,E′,F′,H′,I′,K′) are high power magnification of boxed areas of (B,C,E,F,H,I,K), respectively.(B,E,H,K) View from the experimental side; (A,D,G,J) view from the control side. Arrowheads inB′,E′,H′ and C′,F′,I′ indicate the same position, respectively. To show repression clearly, the colorfor Pax6 was washed away in (C,F,I,L). In the Pax6-expressing cells, En2, Pax5and Pax7expression is repressed. (K′) Tcf4expression is induced around the cells that misexpress Pax6strongly. Arrowheads indicate cells that express Pax6 strongly. Arrows indicate Tcf4-expressingcells. Scale bars are 0.5 mm (C,F), 0.25 mm (I,L), 50 µm (B′,E′,H′) and 10 µm (K′).

2363Pax6 defines di-mesencephalic boundary

Lim1, which is expressed in themantle layer of the pretectum atE4.0 (HH23) in normal embryos,extended caudally to the rostral partof the tectal swelling. Pax7 isexpressed both in the tectum andpretectum but the expression patternis different. In the pretectum, Pax7 isexpressed in the neuroepitheliallayer and mantle layer. After Pax6misexpression, the Pax7-positivemantle layer extended caudallyinto the rostral part of the tectalswelling. In addition, the posteriorcommissure, the trajectory of whichis seen in the pretectum, extendedcaudally after Pax6misexpression.All these results indicate that Pax6repressed tectum-related genesresulting in the caudal shift of the di-mesencephalic boundary.

Expression of the tectum-relatedgenes was repressed in the wholeof the tectal region by Pax6misexpression. But the fate changeoccurred only in the rostral part ofthe tectal swelling. A possibleexplanation is that the expressionvector we used assures transientexpression, so that repression oftectum-related genes may betransient. One example to supportthis assumption is that Pax7 wascompletely repressed at 24 hoursafter electroporation, but that itsexpression was almost restored by

48 hours after electroporation (data not shown). Therefore itmay be plausible that expression of the tectum-related genesare re-organized by the organizing signal from the isthmus, sothat only the rostral tectal swelling is transformed.

The di-mesencephalic boundary is defined byinteraction of Pax6 and En1/Pax2In the present study,Pax6misexpression repressed En1/Pax2and caused the fate change of the rostral tectal swelling to thepretectum. Previously we reported that either En or Pax2/5misexpression caused the fate change of the presumptivediencephalon to the tectum by repressing Pax6 expression(Araki and Nakamura, 1999; Funahashi et al., 1999; Okafuji etal., 1999). Araki and Nakamura examined the time course ofthe Pax6expression after En2misexpression. Since Pax6wasrepressed immediately after the translation product of En2appeared, they concluded that En2 repressesPax6directly. InPax2/5 double knockout mice, Pax6expression and theposterior commissure expands caudally (Schwartz et al., 1999).However, the fate of the pretectum is changed to themesencephalon in Pax6 mutant mice (Mastick et al., 1997).Thus, repressive interaction between Pax6and En1/Pax2maydefine the di-mesencephalic boundary. This notion is veryconsistent with the expression patterns of Pax6and En1/Pax2in normal development. Expression of Pax6 and En1/Pax2

Fig. 6. Effects of dominant-active and -negative forms of Pax6. (A,F) Lateral view, (B,G) dorsalview. (C,H) Hematoxylin-eosin staining at E 5.5 (96 hours after electroporation). The lines in A andF are the approximate planes of C and H. (B,C) The dominant-active Pax6 caused a more severephenotype than wild-type Pax6. (C) The rostral region of the tectal swelling is transformed to thepretectum-like tissue (arrowheads). The size of the tectum is markedly reduced. (H) Dominant-negative Pax6misexpression caused reverse effects. The right-hand-side is the experimental side, androstral is up. (D,E) Repression of En1and Pax2by the dominant-active Pax6. Whole-mount in situhybridization for En1(D) and for Pax2(E). (I,J) Induction of En1and Pax2by the dominant-negativePax6. Whole-mount in situ hybridization for En1(I) and for Pax2(J). Rostral limit of En1and Pax2expression domains (arrowheads in I,J) shifted rostrally on the experimental side. Embryos werefixed 12 hours after electroporation (D,E,I,J). The experimental side is down, and rostral is to theright. Scale bars are 1.0 mm (A,F), 500 µm (C,H) and 0.25 mm (D,E,I,J).

Fig. 7. A model of howthe di-mesencephalicboundary is formed.(A) Pax6is induced in theprosencephalon by someunknown factor. (B) En1and Pax2are inducedpresumably by the axialsignal such as Fgf4.Expression domain ofPax6and that ofEn1/Pax2overlap aroundthe di-mesencephalicboundary. (C) Repressiveinteraction of Pax6andEn1/Pax2may determinethe di-mesencephalicboundary. Pax6andEn1/Pax2repressed eachother, but the mechanismis different. Pax6repressed En1/Pax2indirectly through negative regulators. On theother hand, En1 repressed Pax6directly (Araki and Nakamura,1999). En1and Pax2are in a positive feedback loop and maintaintheir expression each other. (D) Expression domains of Pax6andEn1/Pax2are segregated completely, and finally the di-mesencephalic boundary is formed.

2364

overlaps at the di-mesencephalic boundary region in the earlystages, but the overlapping region gradually reduces duringdevelopment, and their expression domains are completelysegregated by the 11-somite stage.

A similar mechanism is likely to work in defining theposition of the mes-metencephalic boundary. Otx2is expressedin the presumptive prosencephalon and mesencephalon, whileGbx2 is expressed in the presumptive metencephalon. Atfirst their expression domains overlap around the mes-metencephalic boundary, and then through mutual repressiveinteraction, the mes-metencephalic boundary is defined(Broccoli et al., 1999; Hidalgo-Sanchez et al., 1999; Millet etal., 1999; Katahira et al., 2000).

Pax6 indirectly represses the fate of the tectum byactivating repressorsPax6 is reported to function as a repressor of the β-crystallinegene in lens fiber cells (Duncan et al., 1998). In the presentstudy, we have shown that expression of the tectum-relatedgenes are repressed by Pax6. It is of great interest whether Pax6represses these genes directly or indirectly via activating otherrepressor molecules. The paired domain and homeodomainwas fused with the activation domain of VP16 (dominant-active Pax6), or the eh1 domain of En2 (dominant-negativePax6). The dominant-active form of Pax6caused a similar butmore severe phenotype than wild-type Pax6: repression ofEn1 and Pax2, fate change of the tectal swelling to thediencephalon, and much more reduction of the tectal swelling.On the other hand, the dominant-negative form of Pax6misexpression caused the opposite phenotype: rostral shift ofEn1 and Pax2expression, fate change of the pretectum to thetectum, and expansion of the tectum. These results imply thatPax6 represses tectum-related genes by activating anintervening repressor.

This possibility that repression of En1and Pax2by Pax6 isindirect is supported by the time course of repression afterelectroporation. In our in ovo electroporation system, ectopicexpression of transfected genes can be detected by 2 hours afterelectroporation. Repression of Pax6by En misexpression couldbe detected at 3 hours after electroporation, suggesting that Enrepresses Pax6 directly (Araki and Nakamura, 1999). Incontrast, repression of En1by Pax6misexpression was notdetected by 6 hours after electroporation. This time lagindicates that repression of En1and Pax2by Pax6 is indirect.Thus, it is reasonable to assume that Pax6 acts as an activatoraround the di-mesencephalic boundary region in the earlystage.

CONCLUSION

We propose the following process for the formation of the di-mesencephalic boundary. First Pax6expression commences atthe 2-somite stage in the prosencephalic region (Fig. 7A). ThenEn1 is expressed at the 3-somite stage, and Pax2expressioncommences at the 4-somite stage (Okafuji et al., 1999). En1and Pax2expression cover the whole of the mesencephalon atthe 5-somite stage. En1expression may be induced by a signal,such as Fgf4, from the axial mesoderm (Shamim et al., 1999).At these steps, Pax6and En1/Pax2expression overlap aroundthe di-mesencephalic boundary (Fig. 7B). Pax6 is strongly

induced in the rostral part of the di-mesencephalic boundary,while En1/Pax2 are strongly induced in the caudal part.Repressive interaction between Pax6and En1/Pax2may definethe di-mesencephalic boundary. Negative regulators inducedby Pax6 may repress mesencephalon-related genes. En issuggested to repress Pax6directly (Araki and Nakamura, 1999)(Fig. 7C). Through the repressive interaction, the di-mesencephalic boundary may be finally determinedcorresponding to the border of Pax6and En1/Pax2expressiondomains (Fig. 7D).

We thank Dr Kunio Yasuda for chick Pax6 cDNA and Dr NorikoOsumi for anti-Pax6 polyclonal antibody. We also thank Drs.Takayoshi Inoue, Masato Nakafuku, Noriko Osumi, YoshioWakamatsu and Yuji Watanabe for helpful discussion, Drs. NorikoOsumi and Yuji Watanabe for critical reading of the manuscript, andMasanori Takahashi and Masaaki Torii for technical advice. This workwas supported by the Ministry of Education, Culture and Sports,Japan, New Energy and Industrial Technology DevelopmentOrganization, and the Agency of Science and Technology, Japan.

REFERENCES

Acampora, D., Mazan, S., Lallemand, Y., Avantaggiato, V., Maury, M.,Simeone, A. and Brulet, P.(1995). Forebrain and midbrain regions aredeleted in Otx2−/− mutants due to a defective rostral neuroectodermspecification during gastrulation. Development121, 3279-3290.

Alvarado-Mallart, R.-M. (1993). Fate and potentialities of the avianmesencephalic/metencephalic neuroepithelium. J. Neurobiol.24, 1341-1355.

Araki, I. and Nakamura, H. (1999). Engraileddefines the position of dorsaldi-mesencephalic boundary by repressing diencephalic fate. Development126, 5127-5135.

Bally-Cuif, L. and Wassef, M. (1994). Ectopic induction and reorganizationof Wnt-1 expression in quail/chick chimeras. Development120, 3379-3394.

Bally-Cuif, L., Cholley, B. and Wassef, M.(1995). Involvement of Wnt-1 inthe formation of the mes/metencephalic boundary. Mech Dev53, 23-34.

Broccoli, V., Boncinelli, E. and Wurst, W. (1999). The caudal limit of Otx2expression positions the isthmic organizer. Nature401, 164-168.

Cho, E. A. and Dressler, G. R.(1998). TCF-4 binds beta-catenin and isexpressed in distinct regions of the embryonic brain and limbs. Mech. Dev.77, 9-18.

Crossley, P. H., Martinez, S. and Martin, G. R. (1996). Midbraindevelopment induced by FGF8 in the chick embryo. Nature380, 66-68.

Cvekl, A. and Piatigorsky, J.(1996). Lens development and crystallin geneexpression: many roles for Pax-6. Bioessays18, 621-630.

Duncan, M. K., Haynes, J. I. II., Cvekl, A. and Piatigorsky, J.(1998). Dualroles for Pax-6: a transcriptional repressor of lens fiber cell- specific beta-crystallin genes. Mol. Cell Biol.18, 5579-5586.

Ericson, J., Rashbass, P., Schedl, A., Brenner-Morton, S., Kawakami, A.,van Heyningen, V., Jessell, T. M. and Briscoe, J.(1997). Pax6 controlsprogenitor cell identity and neuronal fate in response to graded Shhsignaling. Cell 90, 169-180.

Favor, J., Sandulache, R., Neuhauser-Klaus, A., Pretsch, W., Chatterjee,B., Senft, E., Wurst, W., Blanquet, V., Grimes, P., Sporle, R. andSchughart, K. (1996). The mouse Pax2(1Neu) mutation is identical to ahuman PAX2 mutation in a family with renal-coloboma syndrome andresults in developmental defects of the brain, ear, eye, and kidney. Proc.Natl. Acad. Sci. U S A 93, 13870-13875.

Funahashi, J.-i., Okafuji, T., Ohuchi, H., Noji, S., Tanaka, H. andNakamura, H. (1999). Role of Pax-5 in the regulation of a mid-hindbrainorganizer's activity. Dev. Growth. Differ. 41, 59-72.

Gardner, C. A. and Barald, K. F. (1992). Expression patterns of engrailed-like proteins in the chick embryo. Dev. Dyn.193, 370-388.

Grindley, J. C., Hargett, L. K., Hill, R. E., Ross, A. and Hogan, B. L.(1997). Disruption of PAX6 function in mice homozygous for the Pax6Sey-1Neu mutation produces abnormalities in the early development andregionalization of the diencephalon. Mech. Dev.64, 111-126.

E. Matsunaga, I. Araki and H. Nakamura

2365Pax6 defines di-mesencephalic boundary

Hamburger, V. and Hamilton, H, L. (1951). A series of normal stages in thedevelopment of the chick embryo. J. Morph.88, 49-92.

Hidalgo-Sanchez, M., Millet, S., Simeone, A. and Alvarado-Mallart, R.-M. (1999). Comparative analysis of Otx2, Gbx2, Pax2, Fgf8 and Wnt1 geneexpressions during the formation of the chick midbrain/hindbrain domain.Mech. Dev.81, 175-178.

Inoue, T., Nakamura, S. and Osumi, N.(2000). Fate mapping of the mouseprosencephalic neural plate. Dev. Biol. 219, 373-383.

Ishii, Y., Fukuda, K., Saiga, H., Matsushita, S. and Yasugi, S.(1997). Earlyspecification of intestinal epithelium in the chicken embryo: a study on thelocalization and regulation of CdxA expression. Dev. Growth. Differ.39,643-653.

Itasaki, N. and Nakamura, H. (1996). A role for gradient en expression inpositional specification on the optic tectum. Neuron16, 55-62.

Joyner, A. L. (1996). Engrailed, Wnt and Pax genes regulate midbrain-hindbrain development. Trends. Genet.12, 15-20.

Katahira, T., Sato, T., Sugiyama, S., Okafuji, T., Araki, I., Funahashi, J.-i. and Nakamura, H. (2000). Interaction between Otx2 and Gbx2definesthe organizing center for the optic tectum. Mech. Dev.,91, 43-52.

Kawakami, A., Kimura-Kawakami, M., Nomura, T. and Fujisawa, H.(1997). Distributions of PAX6 and PAX7 proteins suggest their involvementin both early and late phases of chick brain development. Mech. Dev.66,119-130.

Lee, S. M., Danielian, P. S., Fritzsch, B. and McMahon, A. P.(1997).Evidence that FGF8 signalling from the midbrain-hindbrain junctionregulates growth and polarity in the developing midbrain. Development124,959-969.

Li, H. S., Yang, J. M., Jacobson, R. D., Pasko, D. and Sundin, O.(1994).Pax-6 is first expressed in a region of ectoderm rostral to the early neuralplate: implications for stepwise determination of the lens. Dev. Biol.162,181-194.

Logan, C., Hanks, M. C., Noble-Topham, S., Nallainathan, D., Provart, N.J. and Joyner, A. L. (1992). Cloning and sequence comparison of themouse, human, and chicken engrailed genes reveal potential functionaldomains and regulatory regions. Dev. Genet.13, 345-358.

Logan, C., Wizenmann, A., Drescher, U., Monschau, B., Bonhoeffer, F. andLumsden, A. (1996). Rostral optic tectum acquires caudal characteristicsfollowing ectopic engrailed expression. Curr. Biol. 6, 1006-1014.

Lun, K. and Brand, M. (1998). A series of no isthmus(noi) alleles of thezebrafish pax2.1 gene reveals multiple signaling events in development ofthe midbrain- hindbrain boundary. Development125, 3049-62.

Marin, F. and Puelles, L.(1994). Patterning of the embryonic avian midbrainafter experimental inversions: a polarizing activity from the isthmus. Dev.Biol. 163, 19-37.

Mastick, G. S., Davis, N. M., Andrew, G. L. and Easter, S. S., Jr.(1997).Pax-6functions in boundary formation and axon guidance in the embryonicmouse forebrain. Development124, 1985-1997.

Millet, S., Bloch-Gallego, E., Simeone, A. and Alvarado-Mallart, R.-M.(1996). The caudal limit of Otx2 gene expression as a marker of themidbrain/hindbrain boundary: a study using in situ hybridisation andchick/quail homotopic grafts. Development122, 3785-3797.

Millet, S., Campbell, K., Epstein, D. J., Losos, K., Harris, E. and Joyner,A. L. (1999). A role for Gbx2in repression of Otx2 and positioning themid/hindbrain organizer. Nature401, 161-164.

Momose, T., Tonegawa, A., Takeuchi, J., Ogawa, H., Umesono, K. andYasuda, K. (1999). Efficient targeting of gene expression in chick embryosby microelectroporation. Dev. Growth. Differ.41, 335-344

Muramatsu, T., Mizutani, Y., Ohmori, Y. and Okumura, J. (1997).Comparison of three nonviral transfection methods for foreign geneexpression in early chicken embryos in ovo. Biochem. Biophys. Res.Commun.230, 376-380.

Niss, K. and Leutz, A. (1998). Expression of the homeobox gene GBX2during chicken development. Mech. Dev.76, 151-155.

Ogino, H. and Yasuda, K. (1998). Induction of lens differentiation byactivation of a bZIP transcription factor, L-Maf. Science280, 115-118.

Okafuji, T., Funahashi, J.-i. and Nakamura, H. (1999). Roles of Pax-2 ininitiation of the chick tectal development. Brain. Res. Dev. Brain. Res.116,41-49.

Patel, N. H., Martin-Blanco, E., Coleman, K.G., Poole, S.J., Ellis,M. C.Kornberg, T. B. and Goodman, C. S.(1989) Expression of engrailedproteins in arthropods, annelids, and chordates. Cell 58, 955-968.

Picker, A., Brennan, C., Reifers, F., Clarke, J. D., Holder, N. and Brand,M. (1999). Requirement for the zebrafish mid-hindbrain boundary inmidbrain polarisation, mapping and confinement of the retinotectalprojection. Development126, 2967-78.

Rowitch, D. H. and McMahon, A. P.(1995). Pax-2expression in the murineneural plate precedes and encompasses the expression domains of Wnt-1andEn-1. Mech. Dev.52, 3-8.

Schwarz, M., Alvarez-Bolado, G., Urbánek, P., Busslinger, M. and Gruss,P. (1997). Conserved biological function between Pax-2and Pax-5inmidbrain and cerebellum development: evidence from targeted mutations.Proc. Natl. Acad. Sci. U S A 94, 14518-14523.

Schwarz, M., Alvarez-Bolado, G., Dressler, G., Pavel, U., Busslinger, M.and Gruss, P.(1999). Pax2/5 and Pax6 subdivide the early neural tube intothree domains. Mech. Dev.82, 29-39.

Shamim, H. and Mason, I. (1998). Expression of Gbx-2 during earlydevelopment of the chick embryo. Mech Dev76, 157-9.

Shamim, H., Mahmood, R., Logan, C., Doherty, P., Lumsden, A. andMason, I. (1999). Sequential roles for Fgf4, En1 and Fgf8 in specificationand regionalisation of the midbrain. Development126, 945-959.

Shigetani, Y., Funahashi, J.-i. and Nakamura, H.(1997). En-2 regulates theexpression of the ligands for Eph type tyrosine kinases in chick embryonictectum. Neurosci. Res.27, 211-217.

Simeone, A., Acampora, D., Gulisano, M., Stornaiuolo, A. and Boncinelli,E. (1992). Nested expression domains of four homeobox genes indeveloping rostral brain. Nature358, 687-690.

Smith, S. T. and Jaynes, J. B.(1996). A conserved region of engrailed, sharedamong all en-, gsc-, Nk1-, Nk2- and msh-class homeoproteins, mediatesactive transcriptional repression in vivo. Development122, 3141-3150.

Song, D. L., Chalepakis, G., Gruss, P. and Joyner, A. L.(1996). Two Pax-binding sites are required for early embryonic brain expression of anEngrailed-2transgene. Development122, 627-635.

Stern, C. D.(1998) Detection of multiple gene products simultaneously by insitu hybridization and immunohistochemistry in whole mounts of avianembryos. In Cellular and Molecular Procedures in Developmental Biology,Current Topics in Developmental Biology36 (ed. F. d. Pablo, A. Ferrús andC. D. Stern), pp. 223-243. San Diego: Academic Press Ltd.

Stoykova, A., Fritsch, R., Walther, C. and Gruss, P.(1996). Forebrainpatterning defects in Small eyemutant mice. Development122, 3453-3465.

Stoykova, A., Götz, M., Gruss, P. and Price, J.(1997). Pax6-dependentregulation of adhesive patterning, R-cadherinexpression and boundaryformation in developing forebrain. Development124, 3765-3777.

Suemori, H., Kadodawa, Y., Goto, K., Araki, I., Kondoh, H. andNakatsuji, N. (1990). A mouse embryonic stem cell line showingpluripotency of differentiation in early embryos and ubiquitous beta-galactosidase expression. Cell. Differ. Dev.29, 181-186.

Triezenberg, S. J., Kingsbury, R. C. and McKnight, S. L.(1988). Functionaldissection of VP16, the trans-activator of herpes simplex virus immediateearly gene expression. Genes. Dev. 2, 718-729.

Urbánek, P., Fetka, I., Meisler, M. H. and Busslinger, M. (1997).Cooperation of Pax2 and Pax5 in midbrain and cerebellum development.Proc. Natl. Acad. Sci. USA94, 5703-5708.

Walther, C. and Gruss, P. (1991). Pax-6, a murine paired box gene, isexpressed in the developing CNS. Development113, 1435-1449.

Warren, N. and Price, D. J.(1997). Roles of Pax-6 in murine diencephalicdevelopment. Development124, 1573-1582.

Wurst, W., Auerbach, A. B. and Joyner, A. L. (1994). Multipledevelopmental defects in Engrailed-1mutant mice: an early mid-hindbraindeletion and patterning defects in forelimbs and sternum. Development120,2065-2075.