The mes-metencephalic boundary (isthmus) has been suggested to act as an organizer in the development of the optic tectum. Pax-5 was cloned as a candidate for regulator of the organizing center. Isthmus-specific expression of Pax-5 and analogy with the genetic cascade in Drosophila suggest that Pax-5 may be at a higher hierarchical position in the gene regulation cascade of tectum development. To examine this possibility, a gain-of-function experiment on Pax-5 was carried out. In ovo electroporation on E2 chick brain with the eucaryotic expression vector that encodes chick Pax-5 cDNA was used. Not only was a considerable amount of Pax-5 expressed ectopically in the transfected brain, but irregular bulging of the neuroepithelium was induced in the diencephalon and mesencephalon. At Pax-5 misexpressing sites, uptake of BrdU was increased. Histological examination of E7 transfected brain revealed that Pax-5 caused transdifferentiation of diencephalon into the tectum-like structure. In the bulges of the E7 mesencephalon, differentiation of laminar structure was repressed when compared to the normal side. In transfected embryos, En-2, Wnt-1 and Fgf8 were up-regulated ectopically, and Otx2 was down-regulated in the diencephalon to mesencephalon. Moreover, Ephrin-A2, which is expressed specifically in the tectum with a gradient highest at the caudal end, is suggested to be involved in pathfinding of the retinal fibers, and was induced in the bulges. When the mouse Fgf8 expression vector was electroporated, Pax-5 and chick Fgf8 were also induced ectopically. These results suggest that Pax-5, together with Fgf8, hold a higher position in the genetic hierarchy of the isthmus organizing center and regulate its activity.
Revealing the molecular mechanisms required to define certain compartments and to establish positional specificity in the developing central nervous system (CNS) are growing issues of investigation. In the development of vertebrate CNS, the junction of the mesencephalon and metencephalon (isthmus) acts as an organizing center (for review, Bally Cuif & Wassef 1995; Joyner 1996). When the mes-metencephalic boundary of chick is transplanted to the caudal prosencephalon, it keeps its developmental fate and the surrounding host tissue is induced to differentiate into the tectum (Alvarado-Mallart 1993; Marin & Puelles 1994). These observations suggest that instructive influence is exerted from the isthmus to give positional specificity to surrounding tissues.
Large efforts have been made to uncover the molecular basis of the isthmus’ influence. Many genes are expressed in a spatially restricted manner at the isthmus, including transcription factors En-1, En-2, Pax-2, Pax-5 and Pax-8, and diffusible factors Fgf8 and Wnt-1. The roles of these genes in the development of CNS have been analyzed by gene-knockout studies. En, Wnt-1 and Pax-5 knockout mice are affected in isthmus formation. The most extreme deletion is caused by Wnt-1 mutation where most of the midbrain and cerebellum are lost (McMahon et al. 1992). En-1 knockout phenotype is a little milder than Wnt-1 with truncation of colliculi and a trace of cerebellar tissue (Wurst et al. 1994). The Pax-5 knockout mouse has a partial deletion of the colliculi and a slightly enlarged third lobe of the cerebellum (Urbánek et al. 1994). In the Pax-2 knockout mice, exencephaly (resulting from failure of the neural folds to close at the midbrain region) was observed, while the expression patterns of Wnt-1, Pax-5 and En-1 appear normal at E9.5 (Torres et al. 1996). Zebrafish Pax[b] is thought to be an ortholog of Pax-2 and its mutant (noi) shows loss of the tectum (corresponding to colliculi in mice) and the cerebellum (Brand et al. 1996). The En-2 homozygous mutant displays a reduction in the size of the cerebellum and abnormal foliation (Millen et al. 1994).
It is difficult to explore the genetic network in the isthmus from mutant phenotypes because of redundancy and because the isthmus itself is lost in these homozygous mutants. To overcome this problem, the gain-of-function approach is beneficial. One such line of experiments was carried out by Crossley et al. (1996a). They showed that FGF8 protein can mimic the organizing activity of the isthmus by implanting FGF8 soaked bead to caudal prosencephalon. En-2 and Wnt-1 were induced by FGF8 and a tectum of a mirror image was formed ectopically. Although the results suggest that FGF8 might be a key factor in the organizer’s activity, an overview of the genetic network is not clear enough.
As En has been shown to play an important role in positional specificity of the tectum (Itasaki & Nakamura 1996), we attempted to clone upstream genes of En. For cDNA cloning in chick, we constructed subtracted cDNA libraries between the rostral (di-mesencephalic) and caudal (mes-metencephalic) area according to Wang and Brown (1991). One of the clones isolated as caudal-specific is Pax-5. Pax-5 has a paired box that is shared by Drosophila pair-rule gene, paired, which is thought to be one of the upstream genes of engrailed and wingless. If there is a homologous genetic network in vertebrates, Pax-5 can regulate En and Wnt-1 expressions. In accordance with this assumption, it is reported that there are two Pax-2, -5, -8 binding sites in the upstream region of the En-2 gene (Song et al. 1996). To explore this possibility in chick brain, we first examined the expression pattern of Pax-5 in normal development. Next, we transfected chick brain with Pax-5 expression vector by in ovo electroporation, and analyzed the expression patterns of En-2, Wnt-1, Fgf8, Otx2 and Ephrin-A2 (Elf-1) in those embryos.
Materials and Methods
Preparation and screening of subtracted cDNA libraries
Total RNA was isolated from the tissues of the di-mesencephalic border (rostral) and met-mesencephalic border (caudal) that had been isolated from 7000 chick embryos of HH stages 9–13 (Hamburger & Hamilton 1951) by the single-step method (Chomczynski & Sacchi 1991). Total RNA of 500 μg for rostral or 350 μg for caudal was applied to a POLY(A) Quick™ mRNA purification kit (Stratagene, La Jolla, CA, USA), and poly(A) + RNA was selected. Oligo(dT) was used to prime the first strand cDNA synthesis from 5 μg of poly(A) + RNA. Double-stranded cDNA was synthesized using an Amersham kit, and divided into two aliquots. One portion was used for the construction of a conventional cDNA library. The other portion was used for subtraction.
Subtraction between rostral and caudal cDNA
Subtraction was performed as described by Wang and Brown (1991) except that 50 μg of biotinylated driver and 2 μg of non-biotinylated tracer cDNA were used for subtractive hybridization. Subtracted cDNA were inserted in plasmid vector pBluescriptIIKS(+) (Stratagene) at the EcoRI site, and used for the transformation of bacteria.
The transformants were selected at random, and plasmids were recovered. Sense and antisense RNA probes of each clone were synthesized using DIG RNA labeling mix (Boehringer Mannheim, Tokyo, Japan), as there are no criteria to indicate which strand corresponds to the antisense at this point. Cross-hybridization tests were carried out between clones before screening, because there might be redundancy caused by polymerase chain reaction (PCR) amplification. The non-redundant clones were screened by whole-mount in situ hybridization for E2 (HH stage 10) chick embryos.
Cloning and reconstruction of full-length cDNA of Pax-5
The Oligo(dT) primed conventional cDNA library constructed on λZAPII (Stratagene) was screened with the probe synthesized by a random primer DNA labeling kit (Takara, Tokyo, Japan) using clone 49C (Fig. 1B) as a template. The size of the Pax-5 clone obtained was 5.5 kb (clone 49C-a) and 3.2 kb (clone 49C-b). Although the former has the initiation codon at a reasonable position compared with the sequence in other species, the 15 residues downstream from the first methionine codon of the deduced peptide sequence were quite different from others (Fig. 1A). As this might mean that there is a different initiation site or alternative splicing, other cDNA having a different 5′ sequence were searched by 5′-AmpliFINDER RACE kit (Clontech, Tokyo, Japan). Of at least four types of sequences identified, one clone that had a highly conserved sequence between species was used to reconstruct Pax-5 cDNA (Fig. 1A), which contains a region that fits well with the translational initiation consensus (GCC[A/G]CCAUGG; Kozak 1991).
Construction of the expression vectors
Prior to subcloning of Pax-5 cDNA, HA tag (YPYDVPDYAS) was added to the 3′ end of the coding region using PCR. The tagged cDNA was cut out and inserted in pMiwSV, a derivative of pMiwZ (Suemori et al. 1990) that has chick β-actin promoter and RSV enhancer.
Mouse Fgf8 cDNA was cut from pSc17 (Tanaka et al. 1992) by HindIII and XbaI and subcloned on the expression vector pRc/CMV (Invitrogen, Carlsbad, CA, USA) at HindIII and XbaI sites.
In ovo electroporation
In ovo electroporation was adopted to transfect the embryos (Muramatsu et al. 1997). The chick eggs were incubated at 38°C in high humidity for 2 days. After removing 3–4 mL of albumen, a small window was opened on the shell, and five-times diluted drawing ink (Rotring, Hamburg, Germany) was injected beneath the embryo. Stages were determined according to the number of somites, and to Hamburger and Hamilton (1951).
For plasmid injection, the mesencephalon and metencephalon of the embryo were exposed by cutting the vitelline membrane with a microscalpel made from a sewing needle. Approximately 0.1–0.2 μL of plasmid solution in Tris-EDTA buffer was injected into the canal from metencephalon to mesencephalon by micropipette (Fig. 2D). The micropipettes were made from glass capillary tubes (GD-1, Narishige, Tokyo, Japan) by micropipette processor (Narishige).
Ectopic expression of the transfected gene was always observed only on one side of the neural tube. This is because the electric field applied is one-way and the DNA moves in one direction. When we put the anode on the right side of an embryo, ectopic expression was induced on the right side of the neural tube, but no ectopic expression was observed on the other side. Efficiency of the transfection by in ovo electroporation was evaluated using pMiwZ, which contains the lacZ reporter gene. LacZ expression was already discernible by 2 h after the electroporation (Fig. 3A), became stronger thereafter (Fig. 3B), and very strong expression was observed 24–48 h after electroporation (Fig. 3D,E). Expression of the transfected gene is transient, but there was still strong expression of lacZ 72 h after electroporation (Fig. 3F). If we use the green fluorescent protein (GFP) vector (pEGFP-N1, Clontech), we can monitor the state of transfection under a dissection fluorescence microscope. The GFP was first detected after 3 h of transfection under a fluorescence microscope, became stronger after 5 h, and reached a plateau after 9 h (Fig. 3C). Electrochemical damage was restricted to the area contacted on the electrodes (Fig. 2G). The body of the embryo was little damaged (Fig. 2F), and showed no malformation after transfection with pMiwZ or pEGFP (Fig. 3), which indicates that overexpression of the non-functional protein does not induce morphological changes, and that this type of transfection serves as a control.
Viability of embryos 24–48 h after electroporation is about 50% (Table 1), and the embryos can survive longer (Table 1). Transfection efficiency is very high; all embryos fixed 24–48 h after electroporation showed lacZ expression (Table 1), and more than half of the cells at the transfection zone showed lacZ expression (Fig. 3E).
Table 1. . Efficiency of in ovo electroporation
Whole-mount in situ hybridization
Whole-mount in situ hybridization was performed as described by Bally-Cuif et al. (1992) except that the hybridization and wash were carried out at 65°C instead of 70°C. For screening, sense and antisense probes were tested. For two-color in situ hybridization, fluorescein isothiocyanate (FITC) labeled probe and digoxigenin (DIG) labeled probe were used. Alkaline phosphatase (ALP) conjugated anti-FITC (Boehringer Mannheim) was used for the first detection with Fast Red TR/Naphthol AS/MX (Sigma FAST™; Sigma Chemical Co., St Louis, MO, USA), and ALP conjugated anti-DIG (Boehringer) was used for the second detection with 4-nitroblue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP). The ALP for the first step was inactivated by incubating in 100 mmol/L glycine-HCl pH 2.2 for 15 min at room temperature. The template for the Pax-5 probe was made by removing the BamHI fragment from the clone 49C-a-Hd2, a subclone of 49C-a. The templates for chick Fgf8 and Wnt-1 probes were cloned by PCR (495 and 403 bp, respectively), and inserted in pBluescriptSK- (Stratagene). The template for the Otx2 probe was the kind gift of Dr K. Kitamura.
Immunostaining for En-2 was always carried out after in situ hybridization according to Bally-Cuif et al. (1995) with some modification. After alkaline phosphatase activity had been revealed, the embryos were washed sufficiently in phosphate buffered saline containing 0.1% Tween 20 (PBT, 137 mmol/L NaCl, 2.6 mmol/L KCl, 8.1 mmol/L Na2HPO4, 1.5 mmol/L KH2PO4, 0.1% Tween 20), and incubated overnight at 4°C in 4D9 antibody (Gardner et al. 1988) diluted by half with PBT containing 2.5% heat-inactivated normal goat serum. Subsequently the embryos were washed in PBT three times for more than 1 h each, and the antibody was revealed using horseradish peroxidase conjugated antimouse immunoglobulin (Jackson, West Grove, PA, USA) with 3,3′-diaminobenzidine as a substrate.
Immunostaining for Ephrin-A2 was performed directly after fixation. Embryos were treated in five-times diluted 30% H2O2 in MeOH for 5 h at room temperature. Rehydrated embryos were incubated overnight at 4°C with anti-Ephrin-A2 monoclonal antibody. Subsequent treatments were the same as described earlier.
5-bromo-2′deoxyuridine (BrdU) incorporation was examined by the in situ cell proliferation kit, AP (Boeringer Mannheim). The BrdU was injected into the neural tube of the mesencephalon 24 h after electroporation, and the embryos were fixed 15 and 30 min after the injection of BrdU. After whole-mount in situ hybridization with the Pax-5 probe, cryosections cut at 10 μm were stained immunohistochemically with anti-BrdU antibody.
Some specimens were processed for paraffin embedding after whole-mount in situ hybridization or whole-mount immunohistochemistry, cut serially at 7 μm, and observed with a Nomarski interference microscope. Some embryos were incubated to E7 (HH stage 30), fixed in 4% paraformaldehyde, and embedded in paraffin or in Historesin (Leica, Tokyo, Japan). Serial horizontal sections cut at 5 or 7 μm were stained by hematoxylin and eosin.
Cloning of full-length cDNA of chick Pax-5 and its expression in the developing brain
We cloned chick Pax-5 by subtracting cDNA prepared from the mes-metencephalic and di-mesencephalic junction. The deduced peptide sequence of the reconstructed cDNA showed striking homology to Pax-5 sequences in other species (Fig. 1B); 94% identical to mouse and human. Within the paired, octapeptide and homeobox homology domains, the peptide sequence was completely identical to that of mouse and human. Ser/Thr-rich and Pro/Ser/Thr-rich domains were also conserved.
Although Pax-5 (BSAP) has been cloned in many vertebrates (mouse, rat, human, zebrafish) and their expression patterns analyzed (Krauss et al. 1991a; Adams et al. 1992; Asano & Gruss 1992; Rowitch & McMahon 1995), this is the first report on chick Pax-5. The spatial and temporal expression pattern of chick Pax-5 during normal development was analyzed by whole-mount in situ hybridization. We first used the clone 49C as a probe, which covers the homeobox homology domain, and has 64% homology to Pax-2 and lower homology (less than 50%) to Pax-8 of mouse and human. In later experiments, we used the 5′ upstream region without the homeobox homology domain but with octapeptide and paired domains for the probe (Fig. 1B). This probe has 64 and 65% homology to Pax-2 and Pax-8, respectively, and gives a stronger signal than 49C.
We could not detect the Pax-5 signal before the 8-somite stage (HH stage 9 +). At the 8-somite stage, no signal or a very low signal of Pax-5 can be detected in the neural tube (Fig. 4A) while mouse Pax-5 is reported to begin its expression from the 3- or 5-somite stage (Urbánek et al. 1994; Rowitch & McMahon 1995). At the mes-metencephalic boundary, Pax-5 expression is first detectable at the 10-somite stage (HH stage 10; Fig. 4B). The Pax-5 expression band remains wider at the 16-somite stage (Fig. 4C), and narrows at later stages (Fig. 4D). Another region of Pax-5 expression in the neural tube is the anterior-most end of the prosencephalon, where expression begins at around the 10-somite stage (HH stage 10).
Misexpression of Pax-5 changes the morphology of the neural tube
To examine precisely the Pax-5 function for the isthmus organizer’s activity, we took the gain-of-function approach. Pax-5 was ectopically expressed by in ovo electroporation. Whole-mount in situ hybridization revealed a high level of ectopic Pax-5 expression at the side facing the anode in 27 embryos out of 45 examined at E3–E4 (HH stages 18–24, Fig. 5A). When transfected embryos were examined from the dorsal side, bulging of the neuroepithelium was observed in 22 embryos (Fig. 5B). The mes-metencephalic and the di-mesencephalic constrictions were diminished on the affected side, and the size of the mesencephalic swelling was also reduced in size (Fig. 5B). As the tube had many bulges, the surface area of the transfected zone was rather increased (Fig. 5C). No such bulging was detected in 20 control embryos that were transfected with the vector alone or with pMiwZ (Fig. 3D). Histologic observations showed that the Pax-5 expressing cells were located within the bulges (Fig. 5C). The positions did not necessarily match the swelling or hollow; in some cases they were at the apex of the bulge and in other cases at the base. The wall of the neural tube at the untransfected area was thicker than that at the bulge as the marginal zone was differentiating at the untransfected area.
Some transfected embryos were grown to E7 (HH stage 30) and analyzed histologically. As in younger embryos, a vast irregularity was observed especially at the anterior tectum in 13 out of 14 embryos, in a manner as if the swellings had grown extensively (Fig. 5D,E). Histological examination confirmed large swelling from the anterior tectum to the posterior diencephalon. It is striking that transfected tissue of the diencephalon formed the tectum-like structure. On the normal side, the wall of the diencephalon was thin (Fig. 5F), while at the transfected side, the wall was thick and comparable to the posterior tectum (compare Fig. 5G and 5J). At the swelling in the mesencephalon, neuronal differentiation seemed to have been suppressed, the matrix layer was thick while the marginal layer was poorly differentiated. On the normal (upper side in Fig. 5E,H) side of the tectum, laminar differentiation was faster in the rostral portion (Fig. 5I) than in the caudal portion (Fig. 5J; LaVail & Cowan 1971). In the Pax-5 misexpressing area even at the anterior tectum (Fig. 5K), the marginal layer was more poorly differentiated than at the most caudal part of the normal tectum. It seemed that Pax-5 kept cells at the proliferation phase.
In order to check cell proliferation, BrdU incorporation was examined 24 h after the electroporation. Three embryos were fixed 15 and 30 min after BrdU injection and stained with anti-BrdU antibody. It was notable that many cells which incorporated BrdU were located near the central canal at the Pax-5 misexpressing site even 15 min after BrdU injection (Fig. 5L–N). Because it seemed that more cells incorporated BrdU at the Pax-5-transfected side, we compared the difference quantitatively by counting the BrdU-incorporated cells at seven different levels of the brain. The number of BrdU-labeled cells in the rectangular area whose long side is 200 μm was counted (just as the area shown in Fig. 5M,N). At 15 min after BrdU injection, 51.14 ± 12.31 (mean ± SD, n = 7) cells at the transfected side and 38.00 ± 7.32 (mean ± SD, n = 7) cells at the control side incorporated BrdU. At 30 min after BrdU injection, 59.4 ± 13.9 (mean ± SD, n = 7) cells at the transfected side and 44.2 ± 12.3 (mean ± SD, n = 7) cells incorporated the BrdU. Both differences are statistically significant (5% level by Student’s t-test; Snedecor & Cochran 1967).
To prove the possibility that Pax-5 misexpression has changed its fate to caudal tectum, we analyzed the expression pattern of Ephrin-A2 in Pax-5 misexpressing embryos. Because the expression of Ephrin-A2 in normal chick embryos is restricted in the tectum after E4 (around HH 24; Cheng et al. 1995) and shows a rostro-caudal gradient, highest at the caudal end, we can confirm the fate change if a high level of expression was induced at the swelling site. The transfected embryos were incubated to E5 (HH stages 25–26) and fixed for whole-mount immunohistochemistry. As shown in 5Fig. 5O,Q, ectopic Ephrin-A2 expressions were observed in rostral mesencephalon and diencephalon in all of the five Pax-5 transfected brains. Interestingly, no or very low expressing areas were seen in the center of the bulges.
Ectopically expressed Pax-5 induces Fgf8, En-2, and Wnt-1
Earlier observations suggest that ectopic Pax-5 expression induced a ‘microorganizing center’ and it could have caused transdifferentiation of the diencephalon into a tectum-like structure. If this is the case, genes normally expressed in the isthmus organizing center must be expressed in the region. Checking Fgf8 expression is indispensable, because this gene has been suggested to mimic the organizer’s activity (Crossley et al. 1996). Two-color whole-mount in situ hybridization for Pax-5 together with the Fgf8 or Wnt-1 probe showed that ectopic Fgf8 or Wnt-1 was expressed in the same pattern as ectopic Pax-5 from the diencephalon to metencephalon in all of the embryos examined (five embryos for Fgf8 and six embryos for Wnt-1, Fig. 6A,B,E,F). Higher-magnification micrographs showed that Fgf8 was induced in an area slightly narrower than the ectopic Pax-5 expressing area (Fig. 6C,D), while Wnt-1 was induced almost in the same area as ectopic Pax-5 (Fig. 6G,H). The 10 control embryos that were transfected with pMiwZ or pMiwSV displayed no ectopic expression of these genes.
Immunohistochemistry with 4D9 after Pax-5 whole-mount in situ hybridization revealed that En-2 protein was induced not only in the Pax-5 expressing cells but also in the surrounding cells in 10 out of the 11 embryos examined (Fig. 6I,J). Even in an area where ectopic Pax-5 expression was relatively low, ectopic En-2 expression was induced (Fig. 6K,L).
Otx2 expression is repressed in the diencephalon and mesencephalon
Otx2 is a homeobox-containing gene that is expressed in the prosencephalon and mesencephalon (Simeone et al. 1992; Bally-Cuif et al. 1995; Millet et al. 1996). The caudal limit of its expression is shared by that of Wnt-1 (Bally-Cuif et al. 1995; Bally-Cuif & Wassef 1995) and adjacent to that of Fgf8. In other words, rostral isthmus expresses Otx2 whereas caudal isthmus does not. If the cell fate is changed to form an ectopic organizing center by Pax-5 expression, Otx2 expression could be repressed to some extent in the diencephalon and mesencephalon. To examine this assumption, Otx2 expression in Pax-5 transfected embryos was analyzed by double whole-mount in situ hybridization. In three out of five embryos examined, Otx2 expression was repressed within the area where a high level of Pax-5 expression was gained (Fig. 6M–P). The area of repression was always narrower than the Pax-5 misexpressing area.
Ectopic Fgf8 expression induced Pax-5
As the transcription factor Pax-5 can induce Fgf8, we assumed that Pax-5 and Fgf8 are included in the positive feedback loop, that is, Fgf8 could induce Pax-5 in turn. To check this possibility, we transfected chick mesencephalon with mouse Fgf8 (mFgf8) expression vector by in ovo electroporation and tested whether Pax-5 and chick Fgf8 (cFgf8) were induced. Double whole-mount in situ hybridization for Pax-5 and cFgf8 probe revealed that both genes were induced in the mesencephalon and caudal diencephalon of all eight embryos examined. Induction of ectopic Pax-5 is spread over almost all of the hemisphere of the mesencephalon (Fig. 7A). In contrast cFgf8 was induced only in patches of the cells that were expressing a high amount of ectopic Pax-5 (Fig. 7A,B, arrowheads). This suggests that cFgf8 induction is mediated by Pax-5. Interestingly, the morphology of the mesencephalon was altered (Fig. 7C,D). Di-mesencephalic constriction was almost lost, while mes- metencephalic constriction was loosened but still remained.
We cloned chick Pax-5 by subtraction between the cDNA of the di-mesencephalic and mes- metencephalic boundary (isthmus). By analogy with the genetic cascade in Drosophila and its expression pattern, Pax-5 is a candidate gene that regulates the isthmus organizing center. Results obtained by ectopic expression of Pax-5 are that Pax-5: (i) regulates En-2, Wnt-1, Fgf8 and Otx2 expression; and (ii) causes bulging of the mesencephalic neuroepithelium and changes the fate of the diencephalon into a tectum-like structure.
Pax-5 as a key regulator of the isthmus organizing center
Transplantation experiments suggest that the mes- metencephalic boundary (isthmus) functions as an organizing center for the development of the tectum. When the diencephalon is transplanted to the posterior part of the mesencephalon adjacent to the isthmus, the transplant differentiates into the tectum (Nakamura et al. 1986, 1988), and the transplantation of the isthmus induces the tectal structure in the diencephalon (Alvarado-Mallart 1993; Marin & Puelles 1994). In the isthmus region, several genes such as Fgf8, Pax-2, -5, -8, Wnt-1, En-1 and En-2 are expressed specifically. Among them, FGF8 is shown to mimic the organizing activity of the isthmus (Crossley et al. 1996). When FGF8 soaked beads are implanted into the diencephalon, En-2 and Wnt-1 expression is induced and the affected tissue differentiates into the optic tectum. Thus, diencephalon is a region suitable for examination of the morphogenetic process and genetic cascade for tectum development after experimental stimulation as discussed in Sugiyama et al. (1998).
The present study has shown that Pax-5 can induce Fgf8, Wnt-1 and En-2 expression. In turn, FGF8 induced Pax-5 expression, suggesting that Pax-5 and FGF8 are in the positive feedback loop. Morphologically, Pax-5 misexpression caused swelling of the anterior tectum, and changed the diencephalon into a tectum-like structure. These results suggest that Pax-5, in co- operation with FGF8, takes part in the isthmus organizer’s activity.
Moreover, Pax-5 misexpression down-regulated Otx2 expression, and it is notable that the area of repression is narrower than that of Pax-5 misexpression. In other words, Pax-5 and Otx-2 expression overlaps at the periphery of the Pax-5 misexpressing area. Consistently in normal development, Otx2 is expressed at the rostral neural tube down to the mes-metencephalic boundary (Bally-Cuif et al. 1995), and its expression overlaps with Pax-5 at the posterior mesencephalon at an early stage of development. The Pax-5 expressing band regresses posteriorly, and the overlap of Otx2 and Pax-5 expression becomes very narrow around E3 (cf. Figs 5A,6N). It has been shown that Otx2 is indispensable for development of the rostral brain in Otx2 knock-out mice, which have deletion of the prosencephalon and mesencephalon (Acampora et al. 1995; Matsuo et al. 1995; Ang et al. 1996). In addition, it was shown that the posterior boundary of Otx2 expression tightly links to the posterior boundary of the mesencephalon (Millet et al. 1996). Taken together, it is suggested that the repressive interaction between Pax-5 and Otx2 mediates the establishment of a posterior limit for Otx2 to make a sharp mes-metencephalic boundary. Gain-of-function experiments on Otx2 may provide an answer to this issue.
The Pax-2, -5, and -8 subfamily is thought to share similar binding sites (Song et al. 1996). We cannot eliminate the possibility that Pax-5 misexpression activates pathways normally involved with Pax-2 or Pax-8 to some degree. Pax-2 expression starts earlier than Pax-5. Although Pax-2 and Pax-8 are initially expressed more widely in the neural tube, they are restricted as development proceeds (Asano & Gruss 1992; Rowitch & McMahon 1995). In Pax-2 mutant mice, the normal brain phenotype is observed at a high frequency (Schwarz et al. 1997), which suggests that the effects are weakened after the onset of Pax-5 expression, and that they have a similar biological function and share (in part) the same targets. It is likely that these genes work co-operatively. Further study is required to evaluate the individual roles of Pax-2, -5, and -8 in mes- metencephalic development.
Cause of bulging in Pax-5 misexpressing embryos
Ectopic expression of Pax-5 caused an increase in the number of cells that incorporate BrdU, especially those near the central canal, and bulging of the neuroepithelium where differentiation may have been repressed. What is the cue responsible for extensive cell growth and repression of cell differentiation at the Pax-5 misexpressing site? Pax-5 itself is one of the candidates because it is suggested that Pax-5 can repress p53 function by binding to its regulatory element (Stuart et al. 1995). As the p53 tumor suppressor gene is thought to control the cell cycle by arresting cells in the G1 phase (for review, Levine 1997), cell growth may be accelerated when Pax-5 inhibits p53 transcription. Increase in the number of cells that incorporate BrdU, especially those near the central canal, may support this hypothesis. As the duration of the G2 phase in the E3 chick optic tectum is 1.5 h (Wilson 1973, 1974), it may be reasonable to think that BrdU-labeled cells near the central canal 15 min after its injection are those that entered the S-phase before their arrival at the pia surface. In addition, when Pax-5 is knocked out in mice, cell growth in the precursor of inferior colliculus (caudal mesencephalon) is reduced (Urbánek et al. 1994).
Other candidates for the cue to growth and repression of differentiation are FGF8 and Wnt-1, which are induced by Pax-5 misexpression. The phenotype of Fgf8 overexpression in transgenic mice (Lee et al. 1997) is strikingly similar to that of Pax-5 misexpression. When ectopic Fgf8 is expressed under the control of Wnt-1 enhancer, neural precursors continue to proliferate and neurogenesis is repressed. Wnt-1 is also suggested to function as a mitogen in the neural tube (Dickinson et al. 1994). When ectopic Wnt-1 expression is induced under the control of the Hoxb-4 enhancer, dramatic increase in the number of cells undergoing mitosis in the ventricular zone is observed. Thus Wnt-1 is also a candidate for the cue. In this scenario, ectopic Pax-5 increases Wnt-1 expression to cause overgrowth in the neuroepithelium. We cannot exclude any of these possibilities, rather every candidate would work simultaneously to control growth and differentiation of the neuroepithelium.
At E7, bulging caused by Pax-5 misexpression is conspicuous at the diencephalon and at the anterior tectum but not at the posterior tectum. Although misexpression with pMiwSV is transient, Pax-5 misexpression may have kept the anterior part of the mesencephalon in the proliferation phase, and made bulges. Induction of Ephrin-A2 at the anterior part of the tectum by Pax-5 may indicate the acquisition of posterior characteristics of the transfected tissue. The fact that the morphological change was not obvious at E7 at the posterior part of the transfected mesencephalon may be the consequence of regulation after the effect of Pax-5 misexpression weakens.
Possible genetic cascade in tectum development
Induction of Fgf8 and Wnt-1 by Pax-5 misexpression restricted to cells expressing ectopic Pax-5 suggests that Pax-5 activates Fgf8 and Wnt-1 expression cell-autonomously (Fig. 8, pathway a,b). It is also possible that ectopic Wnt-1 is induced by Fgf8 activated by Pax-5 misexpression. This can be supported by the results that FGF8-beads implantation induced Wnt-1 expression (Crossley et al. 1996). Additionally, the expressing area of induced Fgf8 is slightly narrower than that of Pax-5 misexpression. This may indicate that Fgf8 needs more Pax-5 protein than Wnt-1 to be induced, and reflects the normal expression pattern in the isthmus, where the Fgf8 expressing ring is caudal to the Wnt-1 ring.
In contrast to Fgf8 and Wnt-1, En-2 expression was induced not only on the Pax-5 misexpressing cells but also on adjacent cells, which indicate diffusible intermediate(s) in addition to the direct effect of Pax-5 on En-2. As shown by Song et al. (1996), when Pax-2, -5, -8 binding sites in the upstream regulatory region of the En-2 gene were removed or mutated, the mes- metencephalon specific expression pattern was lost in transgenic mice having the lacZ reporter gene. Thus, direct interaction on the binding sites is indispensable for the En-2 gene to be expressed at the mes-metencephalic boundary. For diffusible factors, FGF8 and Wnt-1 are candidates. Bead implantation experiments showed that FGF8 induced En-2 expression in the diencephalon (Crossley et al. 1996a). In the present study, Fgf8 was induced by ectopic Pax-5 expression in the mesencephalon and diencephalon. Taken together, induction of ectopic En-2 expression after Pax-5 misexpression may be intermediated by the FGF8 signal. We can also surmise that Wnt-1 is the mediator for En-2 expression. As discussed earlier, Pax-5 can regulate Wnt-1. In addition, En-1, paralog of En-2, is suggested to be a target of Wnt-1 (Danielian & McMahon 1996). Moreover, we have previously shown that the ectopic expression of mouse Wnt-1 in chick brain caused misexpression of En-2 (Sugiyama et al. 1998). Hence, the pathway from Pax-5 to En-2 through Wnt-1 is also plausible (Fig. 8E). Song et al. (1996) showed that there is a regulatory element other than Pax-2, -5, -8 binding sites that works co- operatively and can specify the expression of En-2 to the mes-metencephalic domain. This regulatory element can be involved in the indirect pathway mediated by FGF8 or Wnt-1.
Summarizing our findings and those of others, we propose a model for the genetic network that regulates the organizing activity of the isthmus and tectum development (Fig. 8). Interaction between Pax-5 and Otx2 may contribute to set the precise position of the isthmus. Once the expression of Pax-5 is started, positive feedback between Pax-5 and Fgf8 regulate En and Wnt-1 expression, and these genes regulate downstream genes such as Ephrin-A5 and Ephrin-A2 that give positional specificity to the mesencephalon and metencephalon. In fact, En-1 and En-2 are shown to give the posterior characteristics to the tectum by inducing ligands for Eph type receptor tyrosine kinase, Ephrin-A2 and Ephrin-A5 (Logan et al. 1996; Shigetani et al. 1997). Wnt-1 and En-2 are interacting with each other in the chick CNS as with Drosophila (Sugiyama et al. 1998).
We thank Dr A. Kanamori for providing us with a detailed protocol for cDNA subtraction. We also thank Dr K. Kitamura for providing the chick Otx2 probe, and Dr N. Itasaki for critical reading of the manuscript. This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture, by Special Coordination Funds for Promoting Science and Technology, by the Uruma Trust Fund, and by the Naito Foundation. J. F. was a recipient of a Postdoctoral Fellowship from the Japan Society for the Promotion of Science.
Author to whom all correspondence should be addressed at: 4-1 Seiryomachi, Aoba-ku, Sendai 980-8575, Japan.