Inductive signal and tissue responsiveness defining the tectum and the cerebellum

Molecular mechanisms involved in the regionalisation of the vertebrate central nervous system (CNS) have been a focus of many studies. A combination of transcription factors and inducing signals from the regional boundary may determine the fate of the region (reviewed by Nakamura, 2000). Loss-of-function and gain-of-function studies of Otx2, En1/2, Pax2/5 suggest that these genes are indispensable transcription factors for mesencephalic development. Transplantation experiments have indicated that the mes/metencephalic boundary (isthmus) has an organizing activity for the mesencephalon and metencephalon (Martinez et al., 1991; Martinez et al., 1995). When the isthmic region was transplanted to the diencephalon, the transplant induced En expression, and surrounding tissue differentiated into the tectum (Martinez et al., 1991). When the isthmus was transplanted in the rhombencephalon, the transplant also induced En but the surrounding tissue differentiated into the cerebellum (Martinez et al., 1995). The tectum and the cerebellum, which differentiate from the mesencephalic and the metencephalic alar plate respectively, are exposed to the isthmic organizing signal, but their mode of differentiation is quite different (Jacobson, 1991; LaVail and Cowan, 1971). Consequently, histological architecture of the tectum and the cerebellum is different.

The molecular nature of the organizing signal and the responsiveness of the tissue to the signal are of great interest. It is accepted that Fgf8 provides the organising activity for the mesencephalon since Fgf8-soaked beads implanted into the diencephalon mimics the isthmic organizer (Crossley et al., 1996; Martinez et al, 1999; Shamim et al., 1999). It was reported that Fgf8 expression is induced at the interface of Otx2 and Gbx2 expression domains overlapping with the Gbx2 domain (Hidalgo-Sanchez et al., 1999; Katahira et al., 2000; Simeone, 2000), where the cerebellum differentiates. We hypothesised that strong Fgf8 signal induces cerebellum. There are 8 Fgf8 isoforms identified so far (Crossley and Martin, 1995; MacArthur et al., 1995a), and it has been reported that Fgf8b has stronger transforming activity than Fgf8a (MacArthur et al., 1995b). Recently, it was reported that Fgf8b could induce a metencephalic phenotype in the diencephalon or in the mesencephalon (Martinez et al., 1999; Liu et al., 1999). We compared the effect of Fgf8a and Fgf8b by in ovo electroporation, a strong gene transfer system (Funahashi et al., 1999). We first examined Fgf8a and Fgf8b expression at the isthmic region by reverse transcription polymerase chain reaction (RT-PCR), then compared the organizing activity of Fgf8a and Fgf8b by misexpression using in ovo electroporation (Funahashi et al., 1999). We found that Fgf8b completely transformed the mesencephalic alar plate to differentiate as cerebellum. We then carried out quantitative experiments to ascertain if the type-difference in the effect of Fgf8 could be ascribed to the difference in the strength of the signal. We also paid special attention to the expression of Otx2, Gbx2 and c-Irx2 as these genes may determine the responsiveness to the organizing signal.

Fgf8a, Fgf8b cDNAs were cloned by PCR from mRNA of embryonic day (E) 3 chick embryos. Probes for En1, Pax6, Pax2, Pax5 and Gbx2 were isolated as described previously (Araki and Nakamura, 1999; Itasaki and Nakamura, 1996; Katahira et al., 2000; Okafuji et al., 1999) Otx2 and Irx2 probes are kind gifts from Drs Kitamura and Ogura, respectively. Fgf8a, Fgf8b and Otx2 cDNAs were inserted in the expression vector, pMiwSV, which has a chick b-actin promoter and RSV enhancer (Suemori et al., 1990; Wakamatsu et al., 1997)

In ovo electroporation was carried out as described previously (Funahashi et al., 1999; Nakamura et al., 2000). Briefly, fertilized chicken embryos were incubated in humid condition at 38°C for 36 hours to reach stage 10 (Hamburger and Hamilton, 1951). DNA solution was injected into the lumen of the neural tube. The electrodes (Unique Medical Imada, Natori, Japan) were placed on the vitelline membrane 4 mm apart, then a rectangular pulse of 25 V, 50 mseconds was given 4 times by electroporation (CUY21, Tokiwa Science, Fukuoka, Japan). To monitor the ectopic expression, a GFP expression vector was added to the DNA solution (0.35 mg/ml). Only the anode side of the neural tube is transfected, as DNA is negatively charged. The other side was used as a control. Since transfection on younger embryos (stage 7-9) gave similar results to embryos at stage 10, the latter embryos were used throughout because of better survival rates.

Poly(A)⁺ RNA was isolated from the isthmic region of stage 9-10 embryos with QuickPrep^“ mRNA Purification kit (Amersham Pharmacia Biotech). Reverse transcription was performed with SuperScript™II RT (Gibco BRL^“), and cDNA was amplified by PCR with primers that can distinguish the Fgf8 subtypes. The primer set used in the present study was: 5′-TTCATGCACTTGTTCGTCC-3′ and 5′-TCTCGACGATGAGCTTGG-3′. For the quantitative analysis, PCR product was harvested every 4 cycles, and the amount of amplified DNA was analysed by NIH Image.

Whole-mount in situ hybridization was carried out according to the method of Bally-Cuif et al. (Bally-Cuif et al., 1995). For double-colored in situ hybridization, one probe was labeled with fluorescein isothiocyanate (FITC), and the other was labeled with digoxigenin (DIG) according to the manufacturer’s protocol (Promega). Alkaline phosphatase (ALP)-conjugated anti-FITC and anti-DIG antibodies (Roche Diagnostics KK) were colored with Fast Red/Naphthol AS/MX (red), and nitroblue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) (purple), respectively. In some cases, the red color was removed with ethanol to show the effect clearly.

Monoclonal antibodies used were: anti-En2 antibody, 4D9 (American Type Culture Collection), anti-neurofilament antibody, 3A10, anti-Islet1 antibody, 40.3A4 (Developmental Studies Hybridoma bank), and anti-calbindin antibody (Sigma). As a second antibody, horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Jackson), or Cy3-conjugated anti-mouse IgG (Jackson) was used. DAB (3,3′-diaminobenzidine) was adopted as the chromogen for HRP. Biotinylated anti-mouse IgG was also used as the second antibody, and immunoreactivity was detected using the ABC-Elite system. Immunostaining was carried out after in situ hybridization.

Embryos were fixed with 4% paraformaldehyde, and embedded in Historesin (Leica). Serial 5 mm sections were stained with Hematoxylin and Eosin.

In mice, there are at least 6 exons in the Fgf8 gene, which are alternatively spliced to give potentially eight different Fgf8 isoforms (Crossley and Martin, 1995; MacArthur et al., 1995a). To identify the isoforms of Fgf8 expressed in the isthmus of chick embryos, we designed primer sets (Fig. 1A) that could distinguish Fgf8 isoforms by PCR, and carried out RT-PCR on mRNA isolated from the isthmic region of E1.5 (embryonic day 1.5) chick embryos. Both Fgf8a and Fgf8b were detected (Fig. 1B). Since PCR was performed using the same primer pairs, differences in the quantity of PCR product before saturation may reflect quantitative difference in the amount of mRNA expressed in the isthmus. By analysis of Fig. 1B by NIH Image, it was shown that both DNA was amplified logarithmically and with equal efficiency until 40 cycles, and saturated around 44 cycles. Quantitative analysis shows that the level of Fgf8b expression was much higher than that of Fgf8a (Fig. 1B) Other isoforms were not detected.

We misexpressed Fgf8a and Fgf8b on the right side of the brain vesicles by in ovo electroporation at stage 10. The left side served as the control. At E6.5-7.5 (stage 30-32) tectal swelling on the experimental side extended anteriorly and became larger than that on the control side (8/8, Fig. 2A). Histologically, the alar plate of the prospective diencephalon appeared to have differentiated into the tectum as a result of Fgf8a misexpression (Fig. 2B). Anterior extension of the tectal swelling was already discernible at E3.5 (stage 21, Fig. 2C). A trunk of the oculomotor nerve arose from the basal plate of the mesencephalon on the control side (Fig. 2D). On the transfected side, additional nerve trunks running similarly to the oculomotor nerve were apparent (3/3, Fig. 2E), suggesting that the basal plate of the diencephalon was transformed into that of mesencephalon. Consistent with this idea, expression of Islet1, which is expressed in the oculomotor nucleus on the control side, extended more anteriorly in the prospective diencephalic territory (Fig. 2F). The trochlear nucleus was not affected (Fig. 2F), indicating that Fgf8a did not affect isthmic development.

Misexpression of Fgf8b produced drastic effects on the mesencephalon. The tectal swelling of the transfected side of E13.5-E15.5 (stage 39-42) embryos was smaller than that of the control side, and had fissures that are typical for the cerebellum (12/12, Fig. 3A-C). Histologically, a well differentiated external granular layer was present (Fig. 3D-F), and beneath it was a layer with large cell bodies, which resembled the Purkinje cell layer of the normal cerebellum (Fig. 3E,F). Immunoreactivity to anti-calbindin antibody, a marker for Purkinje cells in the cerebellum, confirmed that these large cells were Purkinje cells (Fig. 3G,H). Thus, we concluded that Fgf8b transformed the alar plate of the prospective mesencephalon to the cerebellum instead of differentiating into the tectum. In E3.5 embryos (stage 21), in which morphological alteration was already obvious (5/5, Fig. 3I), the oculomotor nerve trunk was missing in the ventral mesencephalon on the transfected side (3/3, Fig. 3J,K). The oculomotor, or trochlear, nucleus did not exist at its proper site, whereas one nucleus was identified at the diencephalic level on the transfected side (Fig. 3L). These data suggest that both the alar plate and the basal plate of the prospective mesencephalon were transformed into metencephalon by Fgf8b misexpression.

Next, we examined the effects of Fgf8a and Fgf8b misexpression on the mes/metencephalon-related molecules. First, we examined the effects on the expression of Otx2 and Gbx2. It has been shown that repressive interaction between Otx2 and Gbx2 determines the location of the mes/metencephalic boundary (Broccoli et al., 1999; Millet et al., 1999; Katahira et al., 2000). Misexpression of Otx2 in the metencephalon changed its fate to mesencephalon (Broccoli et al., 1999; Katahira et al., 2000), which indicates that Otx2 plays a crucial role in the development of the tectum. Fgf8a did not affect Otx2 (4/4) or Gbx2 (9/9) expression at 24 hours after electroporation (stage 17, Fig. 4A-D). In contrast, Fgf8b repressed Otx2 expression (5/5, Fig. 5A-C), and induced Gbx2 expression in the mesencephalon and in the caudal diencephalon at 24 hours after electroporation (5/5, Fig. 5D-F).

We then examined the expression of c-Irx2, one of chick homologues of Drosophila Iroquois expressed in the metencephalon (Goriely et al., 1999). c-Irx2 expression was not affected by Fgf8a (3/3, Fig. 4E,F), but was induced by Fgf8b in the mesencephalon and in the caudal diencephalon (6/6, Fig. 5G-I). Pax2 and Pax5 have been shown to play essential roles in isthmus formation (Krauss et al., 1992; Okafuji et al., 1999; Pfeffer et al., 1998; Funahashi et al., 1999). Misexpression of Fgf8a did not affect Pax2 (3/3, Fig. 4G,H) or Pax5 (5/5, Fig. 4I,J) expression. In contrast, Fgf8b downregulated the expression of Pax2 in the isthmus, but induced its expression in the mesencephalon and caudal diencephalon (3/3, Fig. 5J-L).

En1 and En2 are required for regionalisation of the mesencephalon and the metencephalon, and En2 plays an important role in rostrocaudal polarity formation of the tectum (Hanks et al., 1995; Itasaki and Nakamura, 1992; Itasaki and Nakamura, 1996; Wurst et al., 1994). Both Fgf8a and Fgf8b induced En2 in the caudal diencephalon, mesencephalon and metencephalon, with Fgf8b being the more effective (Figs 4K,L, 5M,N). En1 was induced by Fgf8a in the anterior mesencephalon and the caudal diencephalon (4/4, Fig. 4M). The posterior limit of the Pax6 expression domain was shifted anteriorly by Fgf8a (7/7, Fig. 4N,O).

It has been reported that Fgf8b has stronger transforming activity than Fgf8a (MacAuthur et al., 1995b). We, therefore, tested if weaker Fgf8b signal would exert similar effects as Fgf8a. The expression level of a transgene resulting from in ovo electroporation was shown to depend on the concentration of the DNA solution injected (Momose et al., 1999). We therefore misexpressed Fgf8b at concentrations of 1.0, 0.1, 0.01, and 0.001 mg/ml of Fgf8b expression vector (Fig. 6). As expected, morphological changes at 0.01 mg/ml were similar to those caused by Fgf8a, that is, the diencephalon changed its fate to tectum, and as a result the tectum enlarged (4/4, Fig. 6E,F). The effects on the expressions of molecular markers were also similar to those of Fgf8a (data not shown). These data may justify our working hypothesis that the difference in organizing activity between Fgf8a and Fgf8b is attributable to the difference in the strength of the signal.

Competence, i.e., tissue responsiveness to the signaling molecule, may be determined by the combination of the transcription factors. It has been shown that Otx2 and En1 are indispensable for the tectum development (Araki and Nakamura, 1999; Wurst et al., 1994). Since En1 is expressed in the metencephalon and in the mesencephalon, we thought that Otx2 may be a key molecule for the mesencephalic competence. To verify this idea, co-transfection of Otx2 and Fgf8b was carried out (Fig. 7). Misexpression of Fgf8b using a 0.1 mg/ml expression vector resulted in complete substitution of the tectum with the cerebellum (4/4, Fig. 6C,D). However, co-transfection of Fgf8b (0.1 mg/ml) and Otx2 (1.0 mg/ml) expression vectors recovered the tectal swelling (6/6, Fig. 7C,D). Moreover, diencephalon and metencephalon changed their fate to tectum. These results confirm that Fgf8 signal activates the gene expression cascade for the optic tectum in the neural tube only where Otx2 is expressed. Therefore, Otx2 is a key molecule in mesencephalic determination.

We have shown here by misexpression experiments that (1) Fgf8a induced En2 expression in the diencephalon and changed its fate to the mesencephalon, (2) Fgf8b upregulated Gbx2, Pax2 and Irx2 expression, repressed Otx2 expression, and induced cerebellum in place of the tectum and the caudal diencephalon, and (3) weak Fgf8b signals exerted Fgf8a-type effects. We have confirmed that Otx2 is involved in the competence of mesencephalon.

Classical transplantation experiments showed that presumptive diencephalon could change its fate to differentiate into tectum when transplanted into the mesencephalon near the isthmus (Nakamura et al., 1986; Nakamura and Itasaki, 1992). Then it was shown that isthmus has an organizing activity for the tectum and the cerebellum (Martinez et al., 1991; Martinez et al., 1995). Since an Fgf8-soaked bead implanted into the diencephalon induced expression of En2 and Wnt1, and since the neighboring prospective diencephalon differentiated into the tectum, it was assumed that Fgf8 is the isthmus organizing molecule (Crossley et al., 1996). Subsequent gain-of-function studies of Fgf8 in chick and mice, and analyses in mutant zebrafish and mice have all suggested that Fgf8 is a signaling molecule emanated from the isthmus (Brand et al., 1996; Lee et al., 1997; Liu et al., 1999; Martinez et al., 1999; Meyers et al., 1998; Reifers et al., 1998; Shamim et al., 1999).

In the present study, Fgf8a did not affect Otx2, Gbx2 or Pax2/5 expression. Fgf8a induced En1 and En2 expression in the diencephalon, and as a result, En1 and En2 expression extended anteriorly to the presumptive diencephalic region. Corresponding to the anterior extension of the En expression domain, tectal swelling extended anteriorly. The posterior limit of the Pax6 expression domain shifted anteriorly. As indicated by Matsunaga et al. (Matsunaga et al., 2000), the interface of Pax6 and En1 expression domains corresponded to the boundary between the diencephalon and the tectum on the experimental side. Since the oculomotor nucleus extended anteriorly, and since additional oculomotor nerve trunks were observed after Fgf8a misexpression, it was concluded that Fgf8a converted the fate of the alar plate and the basal plate from diencephalon to mesencephalon.

After Fgf8b misexpression, tectal tissue was completely replaced by cerebellar tissue. In this ectopic cerebellum, the external granular layer and Purkinje cell layer differentiated. Purkinje cell differentiation in the ectopic tectum was confirmed by immunocytochemistry with anti-calbindin antibody on E15.5 embryos. After Fgf8b misexpression Gbx2, Pax2 and Irx2 expression extended to the diencephalic level, and Otx2 expression disappeared from the mesencephalon and the posterior diencephalon. This phenomenon suggests that the isthmus may have moved to the prospective diencephalic level. Thus the motor nucleus that differentiated at the level of original anterior diencephalon (see Fig. 3L) may be the trochlear nucleus. The effect of Fgf8b both morphologically and on downstream gene expression, indicate that the alar plate of the presumptive mesencephalon and the posterior diencephalon changed their fate to the cerebellum.

It is a very interesting phenomenon that the quite different architecture of tectum and cerebellum differentiate as a result of exposure to the isthmus organizing signal. Their mode of differentiation is also quite different. The difference in sequence between Fgf8a and Fgf8b is subtle (see Fig. 1A; Crossley et al., 1995; MacArthur et al., 1995a), but the effect of misexpression of Fgf8a and Fgf8b is very different as shown in the present study and in transgenic mice in which Fgf8a or Fgf8b was misexpressed under Wnt1 regulation (Lee et al., 1997; Liu et al., 1999). Fgf8a-misexpressing mice show overgrowth of the di-mesencephalic region (Lee et al., 1997, Liu et al., 1999). However, Fgf8b-misexpressing mice show transformation of the presumptive diencephalon and mesencephalon to metencephalon as judged by marker gene expression (Liu et al., 1999). Fgf8b transgenic mice showed severe exencephaly, so that the final fate of the presumptive mesencephalon and diencephalon was not reported by Liu et al. (Liu et al., 1999). The present study is the first to show that Fgf8b changed the fate of the entire mesencephalic alar plate to differentiate as cerebellum, while Fgf8a exerted little effect on the mesencephalon. It has been shown that the expression level of a transfected gene by in ovo electroporation is dependent on the concentration of the DNA solution injected (Momose et al., 1999). Since electroporation with 1% Fgf8b expression vector exerted a similar effect as Fgf8a, the type difference between Fgf8a and Fgf8b as an isthmic organizing signal can be attributed to the intensity of the signal. Liu et al. (Liu et al., 1999) also interpreted the phenotype difference between Fgf8a and Fgf8b transgenic mice as quantitative, the effect of Fgf8a being milder than that of Fgf8b. Implantation of beads soaked in Fgf8b in chick embryos also support this hypothesis (Martinez et al., 1999). Martinez et al. implanted the Fgf8b-soaked bead in the diencephalon, and showed that the cells closest to the bead acquired the cerebellar characteristics, and that the tectum was induced surrounding the cerebellar islet. In this context, Fgf8 signal might have been weak in the Fgf8b bead implantation experiment of Crossley et al. (Crossley et al., 1996) in which midbrain structure was induced in the diencephalon. The results of Shamim et al. (Shamim et al., 1999) in which Fgf8a- and Fgf8b-soaked beads implanted in the diencephalic region induced similar changes are consistent with our studies, in that weak Fgf8b signal exerted similar effects to Fgf8a.

Responsiveness of the tissue to the signaling molecule, may be determined by the combination of the transcription factors expressed in the responding area. It has been suggested that Otx2 and En1 are indispensable for the tectum development (Araki and Nakamura, 1999; Wurst et al., 1994). Otx2 misexpression in the metencephalon resulted in ectopic differentiation of the tectum or inferior colliculus in the metencephalic region (Broccoli et al., 1999; Katahira et al., 2000). Misexpression of En2 or En1 in the diencephalon also changed the fate of the diencephalic alar plate to that of mesencephalon (Araki and Nakamura, 1999). Since En1 is expressed commonly in the metencephalon and in the mesencephalon, we thought that Otx2 may be a key molecule for mesencephalic competence. As discussed above, electroporation with a lower concentration of Fgf8b expression vector exerted the Fgf8a-type effects. Co-transfection of Otx2 turned the Fgf8b-type to Fgf8a-type effects at higher concentration of Fgf8b than electroporation with Fgf8b alone. The results indicate that Otx2 is crucial for the mesencephalic competence.

How do these results relate to the isthmic organizing signal and differentiation of the tectum and the cerebellum in normal development? It has been suggested that Fgf8 expression is settled at the interface of Otx2 and Gbx2 expression, overlapping with Gbx2 expression (Broccoli et al., 1999; Hidalgo-Sanchez, 1999; Millet et al., 1999; Irving and Mason; 2000; Katahira et al., 2000). The region that is exposed to the strong Fgf8 signal, and expresses Gbx2 and c-Irx2 may acquire characteristics of rhombomere 1 (R1) (Fig. 8). Indeed, the results of our RT-PCR show that Fgf8b is predominantly expressed in the isthmic region (see Fig. 1). In the R1 region, Hox genes are not expressed, and cerebellum differentiates from the dorsal part of R1, the rhombic lip (Irving and Mason, 2000; Martinez and Alvarado-Mallart, 1989; Wingate et al., 1999). Gbx2 and Irx2 are expressed in R2, but the fate is different. It may be because Hoxa2 is expressed in R2 (Irving and Mason, 2000; Wingate et al., 1999). Since Otx2 represses Fgf8 expression (Broccoli et al., 1999; Katahira et al., 2000), Fgf8 transcripts do not localize in the mesencephalon (Irving and Mason, 2000; Martinez et al., 1999). Although we do not know the exact range of Fgf8 signaling, diffusion of Fgf8 protein in the mesencephalon may be limited. Thus, the mesencephalic region may receive weak Fgf8 signal, which does not repress Otx2 expression. Consequently, mesencephalic alar plate may differentiate into the tectum. Since En, which confers posterior characteristics to the tectum (Itasaki and Nakamura, 1996), is so sensitive to upregulation by Fgf8, Fgf8 may play a crucial role in rostrocaudal polarity formation of the tectum, as proposed by Shamim et al. (Shamim et al., 1999). The idea that the region that receives strong Fgf8 signal acquires the R1 characteristics may also be supported by the result of Fgf8b-bead implantation experiment in which cerebellar structure was induced just near the bead, and the tectum at a distance from the bead (Martinez et al., 1999). Recently, Fgf17 and Fgf18 were shown to be expressed at the mes/metencephalic boundary (Maruoka et al., 1998; Xu et al., 2000). It was also reported that expression patterns and the affinity to the receptors of Fgf8, Fgf17 and Fgf18 in vitro are similar (Xu et al., 1999). Fgf18 has been found to induce Fgf8 expression in the caudal diencephalon, and transform the caudal diencephalon to the midbrain (Ohuchi et al., 2000). Fgf17 knock-out mice exhibited a tissue loss in the caudal part of the mesencephalon, inferior colliculus and the vermis cerebellum (Xu et al., 2000). These results indicate functional redundancy of these genes. Further analysis is required to assess the differential role of these genes.

En1 and En2 are so sensitive to upregulation by Fgf8 signal that both are induced by Fgf8a in the diencephalon (see Fig. 4K-M). This may well explain why Fgf8a transforms the presumptive diencephalon into the mesencephalon.

We thank Dr Y. Wakamatsu for critical reading of the manuscript, Dr T. Ogura for c-Irx2 probe and for discussion and Dr K. Kitamura for Otx2 probe. This work was supported by the grants from the Ministry of Education, Culture and Sports and from the Agency of Science and Technology, Japan.