The isthmus is the organizing center for the tectum and cerebellum. Fgf8 and Wnt1 are secreted molecules expressed around the isthmus. The function of Fgf8 has been well analyzed, and now accepted as the most important organizing signal. Involvement of Wnt1 in the isthmic organizing activity was suggested by analysis of Wnt1 knockout mice. But its role in isthmic organizing activity is still obscure. Recently, it has been shown that Lmx1b is expressed in the isthmic region and that it may occupy higher hierarchical position in the gene expression cascade in the isthmus. We have carried out misexpression experiment of Lmx1b and Wnt1, and considered their role in the isthmic organizing activity. Lmx1b or Wnt1misexpression caused expansion of the tectum and cerebellum. Fgf8 was repressed in a cells that misexpress Lmx1b, but Fgf8expression was induced around Lmx1b-misexpressing cells. As Lmx1b induced Wnt1 and Wnt1 induced Fgf8 expression in turn, Wnt1 may be involved in non cell-autonomous induction of Fgf8 expression by Lmx1b. Wnt1 could not induce Lmx1b expression so thatLmx1b may be put at the higher hierarchical position thanWnt1 in gene expression cascade in the isthmus. We have examined the relationship among isthmus related genes, and discuss the mechanism of the formation and maintenance of isthmic organizing activity.
INTRODUCTION
The isthmus works as an organizing center for the tectum and cerebellum(Martinez et al., 1991;Alvarado-Mallart, 1993;Marin and Puelles, 1994;Martinez et al., 1995).Fgf8 is one of the secreted molecules expressed in the isthmus. As Fgf8-soaked beads transplanted into the diencephalon induced an ectopic tectum or cerebellum (Martinez et al.,1995; Crossley et al.,1996; Martinez et al.,1999), Fgf8 is thought to be the organizing molecule. This notion has further been confirmed by Fgf8 misexpression in mice and chick in which Fgf8 caused complete fate change of the diencephalon and the mesencephalon to cerebellum (Liu et al., 1999; Sato et al.,2001).
It was shown that Otx2 and Gbx2 repress each other's expression to make mes-metencephalic boundary(Broccoli et al., 1999;Millet et al., 1999;Katahira et al., 2000). At theOtx2 and Gbx2 expression boundary, Fgf8 is induced overlapping with Gbx2 expression domain, which was shown in transplantation or misexpression experiments(Hidalgo-Sanchez et al., 1999;Katahira et al., 2000). In combination culture of mesencephalic and metencephalic tissue, Fgf8expression was induced at the boundary(Irving and Mason, 1999). As Otx2 and Gbx2 are transcription factors, involvement of secreted factor(s) or cell surface molecule(s) in Fgf8 induction is assumed.
Wnt1 is a secreted molecule and is expressed in the isthmus.Wnt1 mutant mice show deletion in the mesencephalon and the metencephalon (McMahon et al.,1992). Lmx1b is one of LIM homeodomain proteins and is expressed in connection with Wnt1. Misexpression by the retrovirus vector showed that Lmx1b could induce Wnt1 expression(Adams et al., 2000). Expression patterns of Lmx1b and Wnt1 are well correlated with Fgf8 expression in the isthmus region. In normal development,expression domain of Lmx1b and Wnt1 and that ofFgf8 overlaps broadly around the isthmic region in the early stage,while their expression domains become segregated and located side by side by E2.5. Therefore, we hypothesized that both Lmx1b and Wnt1 were involved in the formation and maintenance of the isthmus organizer. To explore the function ofLmx1b and Wnt1 in the isthmus organizer, we carried out misexpression of Lmx1b and Wnt1 by in ovo electroporation.Lmx1b misexpression induced Wnt1, Otx2 and Grg4,but repressed Fgf8 cell-autonomously. On the one hand, Wnt1misexpression induced Fgf8 expression non cell-autonomously. Hence,Lmx1b represses Fgf8 expression cell-autonomously provably via Grg4 and induced non cell-autonomously via Wnt1. On the other hand, Fgf8misimpression induced Lmx1b expression non cell-autonomously. Otx2 induced Lmx1b expression, while Gbx2 represses Lmx1bexpression. Thus, cell-autonomous and non cell-autonomous regulation among Otx2, Gbx2, Fgf8, Lmx1b and Wnt1 are deeply involved in formation and maintenance of the isthmus organizer activity.
MATERIALS AND METHODS
Expression vectors
First chick Lmx1b cDNA was isolated as two fragments by PCR from E3 chick brain cDNA as a template. Primers for N- and C-terminal fragments are 5′-CCCATATGGACATCGCCTC-3′, 5′-AGGTCTCCTTGGGTCCTTCC-3′and 5′-GCTGAGAAAAGGGGATGAGT-3′,5′-TTCATGAGGCGAAATAGGAG-3′, respectively. Primers for the N-terminal deletion (LIM domain deletion) of Lmx1b (Lmx1b-C)are 5′-GCATGAGCGATGATGAAGATGGAGA-3′ and 5′-CGAAATAGGAGCTCTGCATA-3′ (the start codon is attached in N-terminal primer). Obtained fragments were fused at SacI site to make a full length of Lmx1b. The Lmx1b-EnR is a fusion ofLmx1b with En2 repressor domain and HA-tag(Matsunaga et al., 2000). The full-length chick Wnt1 cDNA was isolated from E2 chick brain cDNA library. These fragments were inserted in pMiwIII, a derivative of pMiwSV and designated as pMiw-Lmx1b, pMiw-Wnt1, etc.(Suemori et al., 1990;Wakamatsu, 1997), which has Rous sarcoma virus enhancer and chickenβ-actin promoter. Otx2, Gbx2 and Fgf8b expression vectors have been described previously(Katahira et al., 2000;Sato et al., 2001).
In ovo electroporation
Fertilized chicken eggs from a local farm were incubated at 38°C. For transfection, in ovo electroporation on stage 10 chick embryos(Hamburger and Hamilton, 1951)was adopted as previously described(Funahashi et al., 1999). Green fluorescence protein (GFP) expression vector (pEGFP-N1,Clontech) was co-electroporated to check the efficiency.
In situ hybridization
In situ hybridization for whole mount and for sections was performed as described (Bally-Cuif et al.,1995; Ishii et al., 1999). Probes for Fgf8, Otx2, Gbx2, Wnt1,Pax2, Grg4 and Cash1 have been described previously(Jasoni et al., 1994;Araki and Nakamura, 1999;Okafuji et al., 1999;Funahashi et al., 1999;Katahira et al., 2000;Sugiyama et al., 2000). ForLmx1b probe, the full length of Lmx1b was used. Digoxigenin(DIG)- or fluorescein isothiocyanate (FITC)-labeled antisense RNA was generated by T3 or T7 RNA polymerase(Funahashi et al., 1999). Alkaline phosphatase (ALP)-conjugated anti-DIG or anti-FITC sheep-polyclonal antibody (Roche Molecular Biochemicals) was used for detection. For double in situ hybridization, Fast Red TR/Naphthol AS/MX (Sigma FAST™; Sigma) was used for detection of the first signal, and 4-nitroblue tetrazolium chloride(NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) were used for detection of the second signal. ALP for the first detection was inactivated by incubating with 100 mM glycine-HCl (pH 2.2) for about 15 minutes at room temperature. In some cases, Fast Red staining was washed out in ethanol, and NBT staining was washed out by incubating in dimethylformamide (DMF) at 55°C.
BrdU incorporation
BrdU (Bromodeoxyuridine) solution (10 mM, Sigma) was injected into the yolk vein 48 and 72 hours after electroporation. Thirty minutes after BrdU injection, the embryos were fixed in 4% paraformaldehyde in PBS. Incorporated BrdU was detected by the addition of monoclonal anti-BrdU antibody (Roche),followed by incubation with Alexa 594-conjugated anti-mouse secondary antibody(Molecular Probes). For the quantitative analysis, BrdU-positive area was measured by Aqua Cosmos image analyzer (Hamamatsu Photonics), and corresponding area of the experimental and control side on the same section was compared.
Immunohistochemistry
Rat monoclonal anti-HA antibody (Roche Molecular Biochemicals) was used as a primary antibody. Horseradish peroxidase (HRP)-conjugated anti-rat IgG antibody (Iwaikougaku-yakuhin) was used as the second antibody.
Histology
Embryos embedded in Technovite 7100 (Kulter) were serially sectioned at 5μm, and stained with Hematoxylin-Eosin, as previously described(Matsunaga et al., 2001).
RESULTS
Expression pattern of Lmx1b, Wnt1 and Fgf8
We first examined spatial and temporal expression patterns of Lmx1b,Wnt1 and Fgf8. As reported before(Yuan and Schoenwolf, 1999;Adams et al., 2000),Lmx1b is expressed from the diencephalon to the metencephalon at stage 9. At stage 10, Wnt1 expression covers whole mesencephalon and the isthmus. The expression domain of Lmx1b is completely included in that of Wnt1, but a little bit narrower; Lmx1b is not expressed in the anterior part of the mesencephalon. Fgf8 is expressed in the metencephalon and isthmus so that the expression domain ofLmx1b, Wnt1 and Fgf8 overlaps in the isthmic region(Fig. 1A,C,E). Overlapping region becomes gradually reduced. At stage 12, Lmx1b is expressed strongly in the mesencephalon, but weakly in the metencephalon, whereFgf8 is expressed (Fig. 1B,D). Expression domain of Wnt1 is almost segregated from that of Fgf8 (Fig. 1F). By E2.5 (HH17), the expression domain of Lmx1b andWnt1 completely overlaps at the posterior margin of the mesencephalon, and just posterior to it expression domain of Fgf8 is located so that the expression domain of Lmx1b and Wnt1becomes side by side to that of Fgf8 at the mesmetencephalic boundary(Fig. 1G,H).
Morphology after Lmx1b or Wnt1 misexpression
From the spatial and temporal expression pattern of Lmx1b, Wnt1and Fgf8, we suspected that they may regulate each other's expression in the isthmic region and may play a role in organizing activity. It has already been reported that Lmx1b and Wnt1 play important roles in maintenance of the isthmic organizing activity by misexpression with retrovirus vectors (Adams et al.,2000). We adopted in ovo electroporation for misexpression,because in ovo electroporation assures more rapid and stronger misexpression.
First, we carried out Lmx1b misexpression experiment. By E7.5 (6 days after electroporation of pMiw-Lmx1b), the size of the tectum was expanded(n=8/8) (Fig. 2A-C). Torus semicircularis was also expanded (n=2/2)(Fig. 2C). The rhombic lip,which is a primordium of the cerebellum, expanded to the posterior(Fig. 2B,C). These results indicate that both mesencephalon and metencephalon are enlarged byLmx1b misexpression.
It has been reported that Wnt1 misexpression with retrovirus vectors had not affected tectum development(Adams et al., 2000). As transfection by electroporation exerts more drastic effects than retrovirus system (Nakamura and Funahashi,2001), we carried out electroporation with pMiw-Wnt1.Wnt1 misexpression resulted in expansion of the mesencephalon by 48 hours after electroporation (Fig. 2D,E). At E14.5 (13 days after electroporation), the telencephalon, the tectum and the cerebellum were all enlarged. In the cerebellum some extra folia were formed (n=3/5)(Fig. 2F-H).
We wondered whether expansion of the mesencephalon was caused by an increase of cell proliferation, and examined BrdU incorporation after 48 and 72 hours of electroporation of pMiw-Wnt1. Anti-BrdU staining revealed that BrdU incorporation was actually increased at the experimental side compared with the control at 48 hours after electroporation(Fig. 3). For the quantitative analysis, the BrdU-positive area between the corresponding site of the experimental and control side on the same section was compared as a pair. Six pairs from two embryos of 48 hours after electroporation showed that BrdU incorporation was significantly greater at the experimental side than at the control side (Table 1,P<0.05, Student's t-test). Difference in BrdU incorporation between the experimental and control side was not recognized 72 hours after electroporation (data not shown).
Number of pairs . | Area/10000 μm2 . | . | . | |
---|---|---|---|---|
. | Experimental . | Control . | Difference . | |
1 | 431.6 | 59.9 | 371.7 | |
2 | 452.7 | 110.6 | 342.1 | |
3 | 1764.5 | 189.7 | 1574.8 | |
4 | 2191.4 | 388.5 | 1802.9 | |
5 | 929.3 | 711.5 | 217.8 | |
6 | 969.9 | 262.0 | 707.9 | |
Mean±s.e.m. | 1123.2±291.1 | 376.8±96.7 | 836.2±279.2 |
Number of pairs . | Area/10000 μm2 . | . | . | |
---|---|---|---|---|
. | Experimental . | Control . | Difference . | |
1 | 431.6 | 59.9 | 371.7 | |
2 | 452.7 | 110.6 | 342.1 | |
3 | 1764.5 | 189.7 | 1574.8 | |
4 | 2191.4 | 388.5 | 1802.9 | |
5 | 929.3 | 711.5 | 217.8 | |
6 | 969.9 | 262.0 | 707.9 | |
Mean±s.e.m. | 1123.2±291.1 | 376.8±96.7 | 836.2±279.2 |
BrdU-positive areas from corresponding site of the experimental and control sides on the same section were extracted by the Image Analyzer (Aqua Cosmos,Hamamatsu Photonics) and compared as a pair. Six pairs from two embryos were analyzed. BrdU incorporation was significantly greater on the experimental side than on the control side (P<0.05, Student'st-test).
As Wnt1 enhanced cell proliferation, it is of great interest if Wnt1 represses neuronal differentiation. So, we looked at effects on a neurogenesis marker, Cash1. Cash1 forces cells to get into differentiation phase from proliferation phase(Jasoni et al., 1994).Wnt1 misexpression repressed Cash1 expression(n=3/4) (see Fig. 6A)in the dorsal mesencephalon. Limx1b also exerted similar effects(n=3/3) (Fig. 4B).
Regulation of Fgf8 by Lmx1b
As Lmx1b or Wnt1 misexpression affected development of the mesencephalon and metencephalon, we looked at effects on isthmus-related genes.
Effects of Lmx1b on Fgf8 expression are not simple. Fgf8expression was repressed in Lmx1b-expressing cells(Fig. 5C). Repression ofFgf8 by Lmx1b was already detectable at 12 hours after electroporation (n=4/4) (Fig. 5A-C). At 24 hours after electroporation, Fgf8 was still repressed in the Lmx1b-expressing cells, but around themFgf8 expression was induced (n=3/7)(Fig. 5D-F). The results suggest that Lmx1b repressed Fgf8 expression in a cell-autonomous manner, but induced Fgf8 expression in non cell-autonomous manner.
It has been reported that Fgf8 is induced at the border ofOtx2 and Gbx2 expression domain, overlapping with Gbx2 expression (Broccoli et al.,1999; Millet et al.,1999; Katahira et al.,2000; Li and Joyner,2001; Ye et al.,2001). It has also been reported that Otx2 and Fgf8 repress each other's expression. Thus, a possibility remains that Otx2 is involved in cell-autonomous repression of Fgf8 by Lmx1b, that is, Lmx1b at first induces Otx2 expression then Otx2 represses Fgf8 in turn. To check this possibility, we looked at effects of Lmx1b on Otx2expression. We then carried out Otx2 misexpression, and looked at the time course of Fgf8 repression by Otx2.
At 12 hours after Lmx1b misexpression, Otx2 expression was induced ectopically in the metencephalon (n=3/3)(Fig. 6A-C′).Otx2 was induced in the Lmx1b-expressing cells, suggesting that induction is cell-autonomous (Fig. 6B′,C′). At 24 hours after electroporation, ectopicOtx2 expression became weak in the isthmic region, but strong in the caudal metencephalon (n=3/4) (Fig. 6D-F).
At 12 hours after electroporation of pMiw-Otx2, Fgf8 expression was not affected (n=8/8) (Fig. 6G-I), which contrasts the result that repression of Fgf8expression by Lmx1b was detected by 12 hours after electroporation(Fig. 5C). These results indicate that Otx2 is not involved in cell-autonomous repression ofFgf8 by Lmx1b.
Dominant-negative Lmx1b induced ectopic Fgf8 expression in the mesencephalon
Lmx1b is a LIM-homeodomain protein, and is composed of two LIM domains, a homeodomain and a C-terminal transcription activation domain(Johnson et al., 1997). It has been suggested that LIM domain could work as dominant negative, and that deletion of LIM domain could work as constitutional activation of the target gene (Curtiss and Heiling, 1998). In case of Lmx1b, it was shown that deletion of LIM domain resulted in increase of transcription activity in insulin enhancer in vitro (German and Wang,1994; Johnson et al.,1997).
As Lmx1b repressed Fgf8 expression in cell-autonomous manner, we wondered if Lmx1b functions as transcriptional repressor or activator. To answer this question, we misexpressed N-terminal deletion construct ofLmx1b (Lmx1b-C), in which LIM domain is not contained. Lmx1b-C misexpression exerted weak but similar effects as Lmx1b misexpression. In the metencephalon, Lmx1b-C induced Fgf8 expression around the cells where Lmx1b-C was misexpressed at 24 hours after electroporation (n=8/13).
We tried misexpression of LIM domain in order to repress Lmx1bfunction, but it did not work. So, we constructed an expression vector that encodes the fusion protein of Lmx1b and En2 repressor domain.Lmx1b-EnR misexpression induced ectopic expression of Fgf8in the caudal mesencephalon in a cell-autonomous manner (n=7/7)(Fig. 6J-M). Co-transfection of wild type Lmx1b and Lmx1b-EnR canceled the effect of Lmx1b-EnR (n=8/8) (Fig. 6N-P), which indicates that Lmx1b-EnR specifically repressed function of Lmx1b. The results suggest that Lmx1b acts as a transcriptional activator in the mes-metencephalic region.
Candidate repressor of Fgf8
As Lmx1b acted as a transcriptional activator, some repressor(s) should intervene in repression of Fgf8 by Lmx1b. It was indicated that Grg4 interacts with the octapeptide domain of Pax2/5 (Eberhand et al., 2000) to convert it to transcriptional repressor. Grg4 is expressed in the mesencephalon but not in the isthmus in normal development,(Fig. 6Q)(Koop et al., 1996;Sugiyama et al., 2000;Ye et al., 2001), andGrg4 misexpression resulted in repression of Fgf8 expression in the isthmus (Sugiyama et al.,2000). Therefore, we examined the effects of Lmx1b onGrg4 expression. At 6 hours after electroporation of pMiw-Lmx1b,Grg4 expression was induced in the metencephalon, (n=5/7)(Fig. 6R,S). As repression ofFgf8 by Lmx1b was not observed before 6 hours after electroporation(n=4/4) and induction of Grg4 occurred before Fgf8repression, it is plausible that Lmx1b first induced Grg4 and then Grg4 repressed Fgf8.
Wnt1 misexpression affects Fgf8 expression
As Lmx1b induced Fgf8 non cell-autonomously, secreted molecules may be involved in this process. Wnt1 is a secreted molecule expressed in an overlapping manner with Lmx1b. So, we suspected that Lmx1b induced Fgf8 via Wnt1 induction. By 9 hours after electroporation of Lmx1b, Wnt1 was induced in cell-autonomous manner(n=1/3). By 24 hours after electroporation, Wnt1 was induced broadly (n=6/6) (Fig. 7A-E). Then, we examined if Fgf8 could be induced by Wnt1 to assess our idea. As expected, Fgf8 was induced by Wnt1 in the metencephalic region by 12 hours after electroporation (n=2/4). At 24 hours after electroporation of pMiw-Wnt1, Fgf8 expression was expanded in the metencephalic region (n=6/6)(Fig. 7G). But Fgf8was not induced in the mesencephalon and in the caudal metencephalon, though misexpression was seen from the mesencephalon to the metencephalon(Fig. 7G,H).
Next, we checked effects of Wnt1 on Lmx1b expression.Lmx1b expression was not affected by 24 hours after pMiw-Wnt1 electroporation. (n=6/6) (Fig. 7I-K). Wnt1 did not affect Otx2 (n=7/7) orGbx2 (n=7/7) expression at 24 hours after electroporation(Fig. 7L-Q). As Wnt1was induced by Lmx1b, Lmx1b may occupy higher hierarchical position in gene expression cascade in the isthmic region.
Further analysis in gene expression cascade among Lmx1b, Otx2,Gbx2 and Fgf8
We have shown that Lmx1b may be put at the higher hierarchical position in gene expression cascade in the isthmus and may play important roles in mes/mesencephalic development, so we further analyzed their relationship. We have already shown that Lmx1b could induce Otx2. Then we examined if Otx2 could induce Lmx1b. At 24 hours after electroporation of pMiw-Otx2, Lmx1b expression was induced in the mesencephalon and the metencephalon (n=4/6)(Fig. 8A-C). The result indicates that Otx2 and Lmx1b could induce each other's expression.
Next, we looked at the effects of Fgf8 on Lmx1b expression.Lmx1b was broadly induced in the diencephalon and mesencephalon byFgf8b misexpression at 24 hours after electroporation(n=4/4) (Fig. 8D-F). However, by 36 hours after electroporation, Lmx1b expression disappeared from most part of the mesencephalon and ring-like expression in the diencephalon remained, although misexpression was seen broadly when checked by GFP, (n=4/4) (Fig. 8G,H). Moreover, endogenous Lmx1b expression in the isthmus was lost (Fig. 8H).
Fgf8b misexpression induced Gbx2 and Irx2expression widely in the mesencephalon [seefigure 7D,E by Sato et al.(Sato et al., 2001)], and changed the fate of the mesencephalic alar plate to differentiate into the cerebellum (Sato et al.,2001). As Fgf8 induces Gbx2 expression, we wondered ifLmx1b was repressed by Gbx2. To examine this possibility, we misexpressed Gbx2 and looked at Lmx1b expression.Gbx2 repressed Lmx1b expression at 24 hours after electroporation (n=4/4) (Fig. 8I-K).
DISCUSSION
In the present study, we have shown that: (1) expression domains ofLmx1b and Wntl were gradually segregated at the isthmus from that of Fgf8; (2) both Lmx1b and Wnt1 misexpression resulted in enlargement of the tectum and the cerebellum; (3) Fgf8expression was repressed in Lmx1b misexpressing cells, butFgf8 was induced around the Lmx1b misexpressing cells; (4)Lmx1b induced Wnt1, Otx2 and Grg4 cell-autonomously; (5)Wnt1 induced Fgf8 expression non cell-autonomously; and (6)Lmx1b is induced by Otx2 and Fgf8, but repressed by Gbx2. The possible role of Lmx1b in the formation and maintenance of isthmus organizer is discussed below.
Role of Lmx1b and Wnt1 in the isthmus organizer formation
Transplantation experiments showed that isthmic region has the organizing activity for the tectum and cerebellum. As Fgf8 beads mimic the isthmus organizing activity, and misexpression of Fgf8b changed the fate of the mesencephalic alar plate to differentiate into the cerebellum, it has been accepted that Fgf8 is the most important organizing molecule(Crossley et al., 1996;Liu et al., 1999;Martinez et al., 1999;Shamim et al., 1999;Sato et al., 2001). En1/2,Pax2/5 and Fgf8 could induce each other's expression, and this positive feedback loop of En1/2, Pax2/5 and Fgf8 may play an important role for the maintenance of the organizing activity. If one of these molecules is misexpressed in the diencephalon, this feedback loop is turned on, and the ectopic tectum is induced in the diencephalon.
Lmx1b and Wnt1 are expressed in the whole mesencephalon at first, and localized to the isthmus. As Wnt1 knock-out mice show deletion in the midbrain and hindbrain, it was suggested that Wnt1 is necessary for mid-hindbrain development(McMahon and Bradley, 1990;Thomas and Capecchi, 1991; McMahon et al.,1992). But misexpression of Wnt1 by retrovirus vector or by transplanting the Wnt1-producing cells did not exert significant effects on mid-hindbrain development (Sugiyama et al., 1998; Adams et al.,2000), so its role in this system is still obscure. Recently,Lmx1b was misexpressed by retrovirus vector, and it was suggested that Lmx1b should be given higher hierarchical position thanWnt1 in the gene expression cascade in the isthmus(Adams et al., 2000). In the present study, we have shown by misexpression by in ovo electroporation thatFgf8 expression was repressed in Lmx1b misexpressing cells,but Fgf8 expression was induced around Lmx1b misexpressing cells. This complicated phenomenon may be explained as follows. First,Wnt1 may be induced by Lmx1b in cell-autonomous manner, then Wnt1 may induce Fgf8 expression in turn. As Wnt1 is a secreted molecule, Wnt1 may be involved in non cell-autonomous induction of Fgf8 by Lmx1b.
Several molecules are abutting at the mid-hindbrain boundary(Fig. 9). At the midbrain side,Otx2, Lmx1b and Wnt1 are expressed, and Gbx2, Fgf8and Irx2 are expressed at the hindbrain side. Otx2 and Gbx2, Fgf8 and Otx2, and Lmx1b and Gbx2, repress each other's expression. However, Lmx1b cell-autonomously represses Fgf8, but induces Fgf8 in the neighboring cells. Otx2 could not induce Fgf8 expression. By these complicated gene regulation mechanisms gene expression pattern in the isthmic region may be set, in a sense automatically, once initial switch of some of them is turned on. Fgf8b misexpression may be one of such cases of autoregulation. After Fgf8b misexpression, Lmx1b was induced widely in the mesencephalon at first, but later Lmx1b expression was restricted in the diencephalic region just as ring, which reminds us of ring like expression in the isthmus in normal embryos. This self regulation may be explained if we consider that Gbx2 is also induced by Fgf8b(Sato et al., 2001). As Gbx2 represses Lmx1b expression (seeFig. 7I-K), Lmx1bexpression, which was induced broadly in the mesencephalon at first, may be repressed by Gbx2. Lmx1b expression may remain just outside of the Gbx2 area to result in ring-like expression in the diencephalic region. Another example is that Fgf8 was induced when R1 and midbrain was juxtaposed(Hidalgo-Sanchez et al.,1999). This phenomenon may be explained by that Otx2 andLmx1b are expressed in the mesencephalic region, and thatWnt1 may be induced by Lmx1b. As Wnt1 is a secreted molecule, it may induce Fgf8 expression in non cell-autonomous manner, that is, in the R1 region. The intimate relationship of these molecules may participate in setting the site of organizer, and mid-hindbrain boundary.
In the isthmic region many molecules are expressed, and they are in the complicated network of regulation. Fgf8, Pax2/5 and En1 are in the positive feedback loop for their expression. Pax2 expression covers whole the mesencephalon and comes to be localized in the isthmic region(Okafuji et al., 1999). As it could induce Fgf8 in the diencephalon, Pax2 has also been suggested to be involved in Fgf8 induction(Okafuji et al., 1999;Ye et al., 2001). It was further shown that in Pax2-/- mice Fgf8expression in the isthmus was abolished though its expression in the cardiac mesoderm was not affected (Ye et al.,2001).
In normal development, Otx2 and Gbx2 are expressed from very early stage of development. At first their expression domains are overlapping, but are completely segregated around stage 10 in chick embryos. It was shown that Otx2 and Gbx2 repress each other's expression so that their expression domains become segregated. By the repressive interaction between Otx2 and Gbx2, mid-hindbrain boundary may be set. Independently, Pax2expression may be induced by the vertical signal, and Fgf8 may be induced in the isthmic region. Considering appearance of Lmx1bexpression and its induction by Otx2, Lmx1b may be induced by Otx2 in the midbrain region, though there is a possibility that vertical signal contribute to induction of Lmx1b. Wnt1 may be induced in turn. However, Fgf8 expression was not affected at first inWnt1-/- mice, but was later disrupted(Lee et al., 1997). This result together with the misexpression experiments including the present study that Lmx1b or Wnt1 did not exert severe morphological effects, indicates that Lmx1-Wnt1 system may work to maintain Fgf8 expression rather than initiation of its expression. Fgf8 expression may be kept just caudal to the Wnt1 and Lmx1b expression ring(Fig. 9).
Growth accelerating activity of Wnt1
Both Lmx1b and Wnt1 misexpression caused enlargement of cerebellum or rhombic lip and the tectum. Extra folia were developed in the cerebellum. As Lmx1b induces Wnt1 expression, both Lmx1 and Wnt1 may have exerted similar effect. It has been reported that Wnt1transgenic mice show overgrowth of neural tube(Dickinson et al., 1994).Wnt1 was misexpressed under the control of Hoxb4 enhancer,which resulted in dramatic increase in the number of mitosis in the ventricular layer and expansion of it. Very recently, Megason and McMahon(Megason and McMahon, 2002)reported in a very sophisticated manner that Wnt protein is distributed in a dorsal to ventral gradient in the spinal cord. They suggested that Wnt-β-catenin/TCF signaling pathway positively regulates cell cycle progression and negatively regulates cell cycle exit in the spinal cord through transcriptional regulation of cyclin D1 and cyclin D2. In the present study, Cash1 was repressed by both Lmx1b and Wnt1, which may indicates that ventricular cells in the tectum are also prevented from getting into differentiation phase. In the tectum anlagen, more cells incorporated BrdU at the Wnt1-transfection site than at the control side at 48 hours after electroporation (see Fig. 3). Difference in BrdU incorporation was not discerned at 72 hours after electroporation. The results indicate that Wnt1 actually enhanced cell proliferation, but the effect on BrdU incorporation was transient so that the size difference between the Wnt1-transfected and the control tecta may have been subtle.
Acknowledgements
We thank Dr T. Reh for Cash1 probe and Dr I. Araki forWnt1 cDNA. This work was supported by the Ministry of Education,Culture, and Sports, Science and Technology, Japan, and by the Mitsubishi Foundation.