Secreted factors FGF8 and WNT1 are essential either for the inductive activity of the isthmus organizer or for the regionalization of the midbrain-hindbrain boundary (MHB). However, transcriptional regulation of these secreted factors during development remains to be elucidated. Here we show that the LIM homeobox gene Lmx1b is expressed in the anterior embryo as early as E7.5 and its expression becomes progressively restricted to the isthmus at E9.0. Analysis of gene expression in the MHB of the mutant embryos showed that many genes were lost by E9.5. In the MHB of Lmx1b-/- embryos, the expression of Fgf8, which normally occurs at the 4-somite stage, was completely absent, whereas Wnt1 was downregulated before the 4-somite stage. Moreover,transcription factors En1 and Pax2 were also downregulated prior to the 4-somite stage, whereas Gbx2 downregulation occurred at the 4-somite stage. By contrast, Otx2 and Pax6 expression was not affected in Lmx1b-/- embryos. The requirement of specific Lmx1b expression in the MHB was further confirmed by Wnt1-Cre-mediated region-specific conditional knockout of Lmx1b. As a result of these molecular defects, the development of the tectum and cerebellum was severely impaired in Lmx1b-/-mice. Taken together, our results indicate that Lmx1b plays an essential role in the development of the tectum and cerebellum by regulating expression of Fgf8, Wnt1 and several isthmus-related transcription factors in the MHB, and is a crucial component of a cross-regulatory network required for the induction activity of the isthmic organizer in the MHB.
The development of the midbrain and cerebellum are controlled by the isthmic organizer in the isthmus, a constriction at the midbrain-hindbrain boundary (MHB). The isthmic organizer serves as one of the best models to study the molecular mechanism of the regionalization of the central nervous system (CNS) along the anteroposterior axis. The isthmic organizer is thought to secrete planar signals for regulating the development of the mid/hindbrain(Liu and Joyner, 2001b; Nakamura et al., 2005; Wurst and Bally-Cuif,2001).
Two secreted factors, FGF8 and WNT1, emanated from the isthmic organizer are required for the development of the midbrain and cerebellum(Liu and Joyner, 2001b; Nakamura et al., 2005; Wurst and Bally-Cuif, 2001). Beads coated with FGF8 mimic the effects of MHB grafts in chick embryos regarding the induction of ectopic tectum (dorsal part of the midbrain) and cerebellum (Crossley and Martin,1995; Crossley et al.,1996; Martinez et al.,1999; Nakamura et al.,1986). Conversely, inactivation of Fgf8 in the MHB leads to the loss of the tectum and cerebellum(Chi et al., 2003; Meyers et al., 1998; Reifers et al., 1998). The domain of Wnt1 expression is located anterior to that of Fgf8 in the MHB. Unlike Fgf8, Wnt1 does not have induction activity but is essential for mid/hindbrain development(McMahon et al., 1992; McMahon and Bradley, 1990). Therefore, FGF8 is the key molecule for mediating the induction activity of the isthmic organizer.
In addition to Fgf8 and Wnt1, transcription factors that are expressed in the MHB are also involved in mid/hindbrain development. Otx2 and Gbx2 are among the earliest genes expressed in the CNS and mark the anterior and posterior epiblast, respectively(Joyner, 1996; Joyner et al., 2000; Liu and Joyner, 2001b; Nakamura et al., 2005; Simeone, 2000; Simeone et al., 2002; Wurst and Bally-Cuif, 2001). Within the MHB, Otx2 and Gbx2 act antagonistically to set up the position of the isthmic organizer, but are not required for the induction of Fgf8 or Wnt1 expression(Broccoli et al., 1999; Li and Joyner, 2001; Millet et al., 1999; Rhinn et al., 1998; Rhinn et al., 1999). The expression of En1, Pax2 and Wnt1 is earlier, whereas that of En2 and Pax5 is later than Fgf8 expression in the MHB (Crossley and Martin, 1995; Liu and Joyner, 2001b). The expression of the En1/En2 genes is required for the maintenance but not for the initiation of Fgf8 or Wnt1 expression(Liu and Joyner, 2001a; Thomas and Capecchi, 1990; Wurst et al., 1994). By contrast, Pax2 is found to be necessary and sufficient for the induction of Fgf8 in the MHB (Ye et al., 2001). However, whether there are other genes involved in transcriptional control of Fgf8 expression in the isthmic organizer is unclear.
The LIM homeodomain transcription factor Lmx1b is a mouse orthologue of chicken Lmx1, which is required for the limb bud development (Chen et al.,1998). Its mutation in human causes a dominantly inherited disease called nail-patella syndrome (Chen et al.,1998; Morello et al.,2001). During CNS development, Lmx1b is essential for the differentiation of midbrain dopaminergic neurons, hindbrain serotonergic neurons, and medullary and spinal dorsal horn neurons(Asbreuk et al., 2002; Cheng et al., 2003; Ding et al., 2003; Ding et al., 2004; Smidt et al., 2000). In addition to its well known function in neuronal differentiation, Lmx1b is expressed in the developing MHB in chick and zebrafish, and misexpression of Lmx1b in the MHB of the chick embryo causes expansion of the tectum and cerebellum and induces Fgf8 and Wnt1 expression (Adams et al.,2000; Matsunaga et al.,2002), whereas knockdown of lmx1b in the zebrafish results in the loss of isthmic and cerebellar structures and the loss of fgf8 and wnt1 expression as well(O'Hara et al., 2005).
In the present study, we used a loss-of-function approach to analyze the role of Lmx1b in mid/hindbrain development in mice. We focused on the function of Lmx1b in regulating the expression of Fgf8, the essential factor for the induction activity of the isthmic organizer. In addition, we examined the effect of Lmx1b expression on other isthmic organizer-related genes. Our results demonstrate that Lmx1bexpression is necessary for the initiation of Fgf8 expression and for the maintenance of several other genes, including Wnt1, En1, En2,Pax2 and Gbx2. Thus, Lmx1b is essential for the initiation and maintenance of the induction activity of the isthmic organizer during mid/hindbrain development.
MATERIALS AND METHODS
Genotyping and maintenance of animals
Lmx1b mutant, Lmx1bflox/flox, and Wnt1-Cre mice were generated and genotyped as previously described(Chen et al., 1998; Danielian et al., 1998; Kimmel et al., 2000; Zhao et al., 2007). All mouse lines were maintained on a mixed genetic background. To inactivate Lmx1b in the MHB, we crossed heterozygous Lmx1bmice with Wnt1-Cre mice, and their Wnt1-cre,Lmx1b+/- offspring were mated with Lmx1bflox/flox mice. The desired Wnt1-cre,Lmx1b-/flox (hereafter referred to as Lmx1bw CKO) progeny were obtained, and the littermates(e.g. Wnt1-cre/Lmx1b+/flox, or Lmx1b-/flox or Lmx1b+/flox) developed normally; they were used as control embryos. The age of embryos was determined according to the plug date [considered to be embryonic day (E) 0.5]. Because the mating time among the mice may differ, the actual developmental stage of the embryo was further ascertained according to criteria described previously(Kaufman, 1998). In each set of experiments, at least three mutant embryos and an equal number of control embryos were used.
Histological analysis, immunohistochemistry and in situ hybridization assays
Nissl staining and immunocytochemistry were performed as described(Ding et al., 2004). E18.5 embryos and postnatal day (P) 0 pups were perfused with 4% paraformaldehyde(PFA) in 0.01 M phosphate buffered saline (PBS; pH 7.4). For Nissl staining with Cresyl Violet, the brains were embedded in paraffin wax and sectioned at 12 μm in the sagittal plane. For immunostaning, the PFA-fixed brains were directly sectioned on a cryostat after cryoprotection with 15% sucrose in PBS. For double staining of calbindin (Sigma) with BRN3b (Santa Cruz), Cy3-labeled donkey antirabbit IgG (Jackson) and FITC-labeled donkey anti-goat IgG(Jackson) were used. Rabbit anti-Ki67 antibodies (Novocastra) were employed to label proliferating cells.
Whole-mount in situ hybridization was performed on embryos fixed in 4% PFA in PBS, and digoxigenin-labeled ribroprobes were used to detect expression of Lmx1b (Ding et al.,2004), En1, En2 (Li and Joyner, 2001), Fgf8(Wassarman et al., 1997), Gbx2 (Li and Joyner,2001), Hoxa2 (Chi et al., 2003), Otx2(Martinez-Barbera et al.,2001; Suda et al.,2001), Pax2 (Schwarz et al., 1999) and Wnt1(Danielian and McMahon, 1996). For double in situ hybridization of Lmx1b with Wnt1 or with Fgf8,the Lmx1b in situ probe was labeled with fluorescein and visualized in red with Fast Red TR/Napthol AS/MX (Sigma), whereas the other probes were labeled with digoxigenin and visualized in dark blue with substrates of BCIP and NBT.
Whole-mount TUNEL was performed as described previously(Chi et al., 2003) with some modifications. Embryos were fixed in 4% PFA and stored in methanol, then rehydrated in PBS containing 0.1% Tween 20 (PBT) and treated with proteinase K(20 μg/ml in PBT) for 2-8 minutes. The embryos were fixed again with 4% PFA for 1 hour and permeabilized in a solution of 0.1% sodium citrate and 0.1%Triton X-100 for 5 minutes on ice. The embryos were incubated with 1×TDT buffer (Promega) containing 40 μM digoxigenin-dUTP (Roche) at 37°C for 1 hour, then with sheep alkaline phosphatase-labeled anti-digoxigenin antibodies (Roche), and finally the signals were visualized in alkaline phosphatase buffer containing NBT and BCIP.
Lmx1b is expressed in the isthmic organizer
To characterize dynamic expression of Lmx1b in the MHB,whole-mount in situ hybridization was performed at different stages of embryonic development. The mRNA transcript of Lmx1b was first detected in the anterior embryo in the early head-fold stage (E7.5; Fig. 1A). By E8.5, Lmx1b expression was observed in the presumptive MHB(Fig. 1B), and became progressively restricted to the isthmus between E9.0 and E10.5(Fig. 1C,G and data not shown). To further define the location of Lmx1b expression in the MHB, we compared the Lmx1b expression domain with those of Wnt1 and Fgf8 at the 4-somite stage. The expression of Wnt1 and Fgf8 showed an anteroposterior non-overlapping expression pattern,whereas Lmx1b expression overlapped with both Fgf8 and Wnt1, covering the caudal half of Wnt1+ and most of the territory of the Fgf8+ domain in the MHB at the 4-somite stage (Fig. 1D-F). This overlapping expression characteristic was also observed at E9.0 by double whole-mount in situ hybridization for Lmx1b with either Wnt1or Fgf8. The Lmx1b+ domain overlapped but also extended caudally beyond the Wnt1+ domain (red arrows in Fig. 1G,H), and the caudal Lmx1b+ domain overlapped with the anterior part of the Fgf8+ domain (Fig. 1I). This confirmed that Lmx1b expression is localized to the isthmic organizer and overlaps with the expression of Wnt1 and Fgf8, suggesting that Lmx1b may be important for the induction activity of the isthmic organizer.
Lmx1b gene deletion leads to defective tectum and cerebellum development
To test the hypotheses that Lmx1b in the MHB regulates the induction activity of the isthmic organizer, we systematically analyzed the development of the tectum and cerebellum in the Lmx1b-/-mutant. At E18.5 and P0, in the absence of Lmx1b it was apparent from macroscopic examination alone that the tectum and cerebellum were significantly reduced in size compared with the wild-type(Fig. 2A,B). This was corroborated by examination of sagittal brain sections. In the Lmx1b-/- mutant, the medial portion of the cerebellum was not visible and the caudal tectum (inferior colliculus) was missing(Fig. 2B,D). The reduced cerebellum appeared to fuse with the much reduced superior colliculus(Fig. 2D), which was further shown by double immunostaining of BRN3b (POU4F2 - Mouse Genome Informatics)(Xiang et al., 1996) and calbindin, markers for the tectum and cerebellum, respectively(Fig. 2E,F). The cerebellum,which is located caudal to the inferior colliculus in the control brain(Fig. 2C,E), became directly connected with the superior colliculus in the mutant(Fig. 2F). In addition, the cerebral cortex and pretetcal region in Lmx1b-/- mutant mice seemed to extend more caudally compared with the wild-type control(Fig. 2C,D), presumably as a result of the reduced tectum (see below).
To define the earliest stage at which histological changes is detectable in the Lmx1b-/- mutant, Nissl staining was performed at different embryonic stages. Morphological defects were first observed at E10.5. We found that the normal constriction between the midbrain and hindbrain was not apparent and rhombomere 1 was smaller in the mutant embryo(Fig. 3A,B). Loss of this constriction was clearly observed in the Lmx1b-/- embryo at E12.5 (Fig. 3F), and Math1 (Atoh1 - Mouse Genome Informatics) expression in the presumptive cerebellum was dramatically reduced compared with that in the control embryo at this stage (Fig. 3G,H). Despite these severe abnormalities in the tectum and cerebellum, the Otx2 (a marker for the midbrain; see below)expression domain did not show a caudal extension in the mutant embryo as compared with the wild-type control (arrows in Fig. 3I,J). In the ventral tegmentum, expression of Nurr1 (Nr4a2 - Mouse Genome Informatics) and tyrosine hydroxylase (markers for dopaminergic neurons in the midbrain) were downregulated at E12.5 and were lost at E15.5(Fig. 3L,R and data not shown),which is consistent with previous findings(Smidt et al., 2000). However,expression of Brn3a (Pou4f1 - Mouse Genome Informatics), a marker for the red nucleus in the tegmentum, remained unchanged in the mutant embryo at E15.5 (Fig. 3P),although the cerebellar vermis was not seen dorsally(Fig. 3N). Taken together,these findings indicate that deletion of Lmx1b causes severe defects in the development of the tectum and cerebellum.
Lmx1b is required for the initiation of Fgf8expression
The secreted factor FGF8, which is emanated from the isthmic organizer, is essential for the development of the midbrain and hindbrain(Liu and Joyner, 2001b; Nakamura et al., 2005; Wurst and Bally-Cuif, 2001). Therefore, we sought to dissect the possible relationship between the expression of Lmx1b and that of Fgf8 during mid/hindbrain development at stages prior to E10.5, the stage at which morphological defects in the MHB were first observed in the Lmx1b-/- embryo.
In the wild-type embryo, Fgf8 expression was first observed in the presumptive MHB at the 4-somite stage (Fig. 4A). Strikingly, Fgf8 expression was absent in the Lmx1b-/- mutant at this stage(Fig. 4B). By contrast, Fgf8 expression in other regions of the mutant embryo was normal(arrowhead in Fig. 4B),suggesting specific regulation of Fgf8 expression by Lmx1bin the MHB. In the control embryo, Fgf8 expression in the MHB disappeared around E12.5. No Fgf8 expression was detected from the 4-somite stage to E12.5 in the Lmx1b-/- embryo(Fig. 4D and data not shown). Thus, we conclude that Lmx1b expressed in the MHB is required for the initiation of Fgf8 expression in the isthmic organizer.
Lmx1b is required for the maintenance of Wnt1, En1, En2and Pax2 expression
We next examined whether the expression of Wnt1 and other isthmic organizer-related genes (Joyner,1996; Joyner et al.,2000; Simeone et al.,2002) was also affected by the deletion of Lmx1b. Since Lmx1b is required for the initiation of Fgf8 expression, and inactivation of Fgf8 in the MHB causes downregulation and eventual loss of expression of other genes at the MHB region(Chi et al., 2003), it is possible that any change in the gene expression profile in the Lmx1b-/- embryo may be due to the loss of Fgf8expression. We first focused on those genes that are expressed in the MHB prior to Fgf8. In Lmx1b-/- embryos, the expression of Wnt1, En1 and Pax2 was drastically reduced in the MHB at the 3-somite stage (Fig. 4F,J,N), whereas they were highly expressed in the wild-type embryo (Fig. 4E,I,M). The expression was lost altogether by E9.5(Fig. 4H,L,P). Because the reduction in Wnt1, En1 and Pax2 expression occurred before the absence of Fgf8 in the Lmx1b-/- embryo, the reduction should have resulted from the inactivation of Lmx1b, rather than of Fgf8. Furthermore, we found that En2 expression,which is normally expressed later than Fgf8, was also downregulated but maintained at a low level until at least E9.5 (data not shown). Thus, the expression of Wnt1, En1 and Pax2 did not appear to be causally related to Fgf8 expression, whereas En2 expression may depend on Fgf8 expression. Nevertheless, it is clear that Lmx1b expression, although not required for the initiation of Wnt1, En1, En2 and Pax2 expression, is necessary for their maintenance in the MHB.
Effects of Lmx1b deletion on Otx2 and Gbx2expression
In the normal embryo, Otx2 and Gbx2 are expressed in the midbrain and anterior hindbrain, respectively, and both are required for proper positioning of the isthmic organizer(Joyner et al., 2000; Simeone, 2000; Simeone et al., 2002; Wurst and Bally-Cuif, 2001). We found that Gbx2 expression was normal in Lmx1b-/- embryos before the 4-somite stage(Fig. 5B). However, unlike the wild-type embryo, in which the Gbx2 transcript was found at the rostral end of the hindbrain (arrow in Fig. 5C), Gbx2 expression was barely detectable in Lmx1b-/- embryos at the 4-somite stage(Fig. 5D). By contrast, the same level of Gbx2 expression was detected in the caudal hindbrain of both wild-type and Lmx1b-/- embryos (arrowheads in Fig. 5C,D), suggesting specific Lmx1b-dependent expression of Gbx2 in the rostral hindbrain. Since Gbx2 and Otx2 are required for correct positioning of the isthmic organizer, the loss of Gbx2 in the Lmx1b-/- embryo may cause a caudal shift of the midbrain(Simeone, 2000; Simeone et al., 2002; Wassarman et al., 1997). However, we found no change in either the level or the location of Otx2 expression in the Lmx1b-/- embryos at the 7-somite stage and E9.5 (Fig. 5H,J) as compared with the wild-type(Fig. 5G,I). The distance between the two ends of the midbrain (double-arrowed line in Fig. 5J) remained unchanged,and the expression pattern of the pretectal marker Pax6(Walther and Gruss, 1991) was not affected in the mutant embryo (Fig. 5P), suggesting that the position of the MHB is not changed in Lmx1b-/- embryos. To further confirm this, we examined expression of Otx2 and Hoxa2 (a marker for rhombomere 3)(Chi et al., 2003) in the mutant embryos. We found no changes in the distance between the two expression domains up until the 13-somite stage (Fig. 5K-N), at which point increased cell death was detected in the MHB(see below). Taken together, Otx2 expression is not affected by the deletion of Lmx1b, and loss of Gbx2 in Lmx1b-/- embryos does not result in a caudal expansion of the midbrain.
Abnormal cell death in Lmx1b-/- embryos
The above results showed that in Lmx1b-/- embryos, the expression of Fgf8 was absent and the expression of many other genes involved in midbrain/cerebellum development was downregulated. To determine whether this was caused by abnormal cell death in Lmx1b-/-embryos, we performed whole-mount TUNEL staining between the 3- and 13-somite stages, when Lmx1b deletion-induced gene downregulation was prominent. A small but similar number of TUNEL+ cells were found in the MHB of both the mutant and wild-type embryos before the 7-somite stage;this number increased slightly at the 7-somite stage(Fig. 5R). However, at the 13-somite stage, TUNEL+ cells were markedly increased in the MHB of Lmx1b-/- embryos, but not in wild-type embryos(Fig. 5S,T). Thus, the Lmx1b deletion-induced changes in gene expression cannot be attributed simply to increased cell death, as this occurred after these changes in gene expression. On the other hand, because the increased cell death in Lmx1b-/- embryos occurred after the loss of Fgf8 expression and the downregulation of several further developmental genes, it is likely that the abnormal cell death is a consequence of these changes in gene expression.
In addition to increased cell death, we found that the number of Ki67-labelled proliferating cells in the tectum and rhombomere 1 was lower in the Lmx1b-/- embryos at E10.5, as compared with that in wild-type embryos (Fig. 3C,D),suggesting that Lmx1b deletion also reduced cell proliferation. Therefore, it is likely that the abnormal cell death and lower cell proliferation in the tectum and rhombomere 1 in the Lmx1b-/- embryo leads to the reduction in the size of the tectum and cerebellum.
Wnt1-Cre-mediated deletion of Lmx1b in the MHB results in a similar phenotype
As described above, we found that Lmx1b expression in the MHB is essential for the development of the isthmic organizer. To further confirm whether specific expression of Lmx1b in the isthmic organizer is essential, we performed region-specific gene inactivation using a Wnt1-Cre conditional knockout method. In this Lmx1bw CKO embryo, there was a slight reduction in Lmx1b expression in the MHB at the 3-somite stage, but a nearly complete absence of Lmx1b expression at the 6-somite stage(Fig. 6A,B). The incomplete deletion of Lmx1b before the 6-somite stage may be due to the existence of non-overlapping expression domains of Lmx1b and Wnt1 in the MHB (Fig. 1). The nearly complete deletion of Lmx1b at the 6-somite stage may be ascribed to the downregulation of expression of the isthmus-related genes caused by the inactivation of Lmx1b, and to the mutual dependence of their expression in the MHB (see Discussion). The expression of Fgf8 in the MHB was greatly reduced at the 4-somite stage (data not shown) and was lost altogether at E9.5(Fig. 6C,D). If Lmx1bcan regulate Fgf8 expression in a dosage-dependent manner, the remaining Fgf8 expression at the 4-somite stage may be attributed to the incomplete deletion of Lmx1b in the Lmx1bwCKO embryo, whereas nearly complete deletion of Lmx1b after the 6-somite stage results in the complete absence of Fgf8expression.
We also examined the expression of other developmental genes in the Lmx1bw CKO embryos. At E9.5, expression of Wnt1(Fig. 6E,F), Pax2(Fig. 6G,H), En1(Fig. 6I,J) and Gbx2(Fig. 6M,N) was completely lost, and that of En2 was downregulated(Fig. 6K,L), whereas that of Otx2 (Fig. 6O,P) was unchanged. All these patterns are similar to those found in conventional Lmx1b-/- embryos as described above. Consistent with this, Lmx1bw CKO mice also exhibited a drastic reduction in the size of the tectum and cerebellum (Fig. 7). We thus conclude that the developmental defects in the tectum and cerebellum are due to the specific action of Lmx1b in the MHB.
The present study revealed that Lmx1b is expressed in the anterior embryo as early as E7.5, and becomes progressively restricted to the isthmus at E9.0. Among many genes that are expressed in the early embryo, we found that Lmx1b is the earliest gene, besides Otx2, to be essential for mid/hindbrain development. In Lmx1b-/- mice,the tectum and cerebellum were drastically reduced in size, and the conditional inactivation of Lmx1b by Wnt1-Cre in the MHB resulted in a similar phenotype. The function of Lmx1b in the MHB was further exemplified by the absence of the expression of Fgf8, as well as by the downregulation of several other isthmic organizer-related genes in the mutant mice. In fact, Lmx1b mutant brains display a phenotype that is characteristic of the loss of isthmic organizer activity (e.g. the loss of the inferior colliculus and great reduction in the cerebellum). Lmx1b is therefore essential for mid/hindbrain development,controlling the development of the isthmic organizer through regulation of the expression of Fgf8, Wnt1 and other developmentally important transcription factors in the MHB.
Lmx1b is required for the initiation of Fgf8 in the isthmic organizer
It is well established that there is an isthmic organizer in the MHB responsible for patterning of the midbrain and cerebellum(Liu and Joyner, 2001b; Nakamura et al., 2005; Wurst and Bally-Cuif, 2001). The molecular basis of the isthmic organizer patterning is only partially characterized. The growth factor FGF8 has been shown to be responsible for the induction activity of the isthmic organizer. In the present study, we showed that expression of Lmx1b precedes that of Fgf8(Fig. 1), and Fgf8expression was absent in the MHB of Lmx1b-/- embryos(Fig. 4). Importantly, a previous study has shown that misexpression of Lmx1b in chick embryos induces ectopic expression of Fgf8 and causes expansion of the tectum and cerebellum (Matsunaga et al.,2002), suggesting that Lmx1b is sufficient for the initiation of Fgf8 expression. Conversely, knockdown of lmx1b in zebrafish results in the loss of fgf8 expression(O'Hara et al., 2005). Our results provide the first evidence that Lmx1b is required for the initiation of Fgf8 expression in the isthmic organizer of mouse embryos. Together with previous studies, we conclude that Lmx1b is necessary and sufficient for the initiation of Fgf8 expression in the MHB during isthmic organizer development, although the intrinsic molecular machinery may be different among the different species. Furthermore, it is interesting to note that misexpression of Fgf8 in the chick embryo(Matsunaga et al., 2002), or addition of FGF8-coated beads to mouse midbrain slice culture(Liu and Joyner, 2001a), is able to induce ectopic Lmx1b expression. Thus, Lmx1b and Fgf8 appear to cross-regulate each other during the development of the isthmic organizer. Thus, such cross-regulation between Lmx1b and Fgf8 may come into play only at the 4-somite stage when Fgf8expression initiates. Because Lmx1b expression precedes Fgf8,Lmx1b is likely to regulate expression of Fgf8, either directly or indirectly, and a positive feedback regulation between Fgf8 and Lmx1b may exist afterwards to ensure the proper activity of the isthmic organizer.
Lmx1b is required for maintaining Wnt1 expression
The secreted factor Wnt1 is necessary for the proper development of the isthmic organizer and mice lacking Wnt1 show abnormal development of the midbrain and hindbrain(McMahon and Bradley, 1990; McMahon et al., 1992). In the Lmx1b-/- embryos, Wnt1 was downregulated in the MHB and prematurely lost by E9.5 (Fig. 4). This result indicates that Lmx1b is required for the maintenance of Wnt1 expression in the MHB. This is consistent with the previous findings that misexpression of Lmx1b in chick embryos induces Wnt1 expression ectopically(Adams et al., 2000; Matsunaga et al., 2002), and that knockdown of lmx1b in zebrafish leads to the loss wnt1expression (O'Hara et al.,2005). Whether Wnt1 is directly regulated by Lmx1b is unclear. Because Fgf8 is required for maintaining Wnt1 expression in the MHB (Chi et al., 2003; Reifers et al.,1998), and Fgf8 is absent in the MHB of the Lmx1b-/- embryo, it is possible that the Fgf8activity mediates the maintenance of Wnt1 expression by Lmx1b in the MHB. However, as downregulation of Wnt1occurred prior to the 4-somite stage, our data revealed an Fgf8-independent but Lmx1b-dependent Wnt1expression profile in the development of the MHB prior to the 4-somite stage. Furthermore, although ectoptically expressed Wnt1 can induce Fgf8 expression in the chick embryo(Ye et al., 2001), it is unlikely that the absence of Fgf8 expression in the Lmx1b-/- embryo is due to downregulation of Wnt1,because the initiation of Fgf8 expression occurs independently of Wnt1 (Danielian and McMahon,1996; McMahon et al.,1992). Taken together, Lmx1b is essential for the initiation of Fgf8 expression and for the maintenance of Wnt1 expression during development of the isthmic organizer.
IsLmx1b required for the positioning of the MHB?
It is well established that Otx2 and Gbx2 are required for the correct positioning of the isthmic organizer, and are not required for the initiation of Fgf8 or Wnt1 expression(Joyner et al., 2000; Martinez-Barbera et al., 2001; Rhinn et al., 1998; Simeone, 2000; Simeone et al., 2002; Wurst and Bally-Cuif, 2001). Expression of Lmx1b precedes Gbx2, but after Otx2(Fig. 1). In Lmx1b-/- embryos, Gbx2 expression was normal before the 4-somite stage, but became barely detectable at the 4-somite stage and absent thereafter (Fig. 4). By contrast, Otx2 expression in Lmxb-/- embryos was similar to that in the wild-type control (Figs 3, 4 and 5). This is interesting because Lmx1b appears to differentially regulate the expression of Gbx2 and Otx2, which are known to antagonize each other in setting the position of the MHB. These results also argue for the specificity of Lmx1b-dependent regulation of gene expression in the MHB. Moreover, Gbx2-/- mice lack the anterior hindbrain and display abnormal caudal expansion of the midbrain, paralleled by posteriorized expression of Otx2, Wnt1 and Fgf8(Broccoli et al., 1999; Li and Joyner, 2001; Wassarman et al., 1997). However, we did not find such a caudal shift of the midbrain when Gbx2 expression was lost in the Lmx1b-/- embryo,as indicated by the normal expression of Otx2 and Pax6 and unchanged distance between the Otx2+ and Hoxa2+ domains in the mutant embryo(Fig. 5). The reduction in this distance in the mutant embryo at the 13-somite stage is most likely to be due to the increased cell death in the MHB, rather than to caudal extension of the midbrain (Fig. 5). This raises the possibility that Lmx1b deletion in the MHB may activate an unknown compensatory mechanism that prevents caudal shift of the midbrain in the absence of Gbx2 activity. It is also possible that Lmx1bmay be required cell-autonomously in the putative expanded domain for Otx2 expression in the MHB.
Lmx1b, a crucial component in a positive maintenance loop required for the development of the isthmic organizer
The isthmic organizer-related genes encoding transcription factors PAX2,EN1 and EN2 are expressed in the MHB (Liu and Joyner, 2001b; Nakamura et al., 2005; Wurst and Bally-Cuif, 2001; Ye et al.,2001). We have shown that the expression of Lmx1bprecedes that of Pax2 and En1(Fig. 1), and that Lmx1b deletion downregulated their expression(Fig. 4). Because the downregulation of these transcription factors occurred at a time prior to the initiation of Fgf8 expression in the normal embryo, it is unlikely that this downregulation is caused by the loss of FGF8 activity in the mutant embryo. This does not exclude the possibility that Fgf8 normally may contribute to the expression of these transcription factors at later stages,because Fgf8 is required for their maintenance(Chi et al., 2003).
A positive maintenance loop involving Fgf8, Wnt1 and Enand Pax genes during the development of the isthmic organizer has been proposed (Joyner et al.,2000; Nakamura et al.,2005; Wurst and Bally-Cuif,2001). This is based on the evidence that the loss of En1,Pax2, Fgf8 or Wnt1 function affects the maintenance but not the initiation of expression of these genes, with the exception of Pax2that is required for the initiation of Fgf8 expression(Chi et al., 2003; Joyner et al., 2000; McMahon et al., 1992; Meyers et al., 1998; Wurst et al., 1994; Ye et al., 2001). Our study indicates that Lmx1b is a crucial component of the positive maintenance loop for isthmic organizer development. Lmx1b is required not only for maintaining Wnt1 and Pax2 expression, but also for the initiation of Fgf8 expression. Because the downregulation of Wnt1 and Pax2 may not be sufficient to explain the loss of Fgf8, Lmx1b may control the initiation of Fgf8 either directly or via an unknown pathway, which may in turn regulate expression of Lmx1b. The complexity of this positive maintenance loop remains to be fully elucidated. For example, we have found that Lmx1b expression is lost in En1-/- embryos at E9.5 (C.G. and Y.-Q.D.,unpublished). Since Lmx1b is normally expressed earlier than En1, the immediate maintenance of Lmx1b may require possible feedback regulation by En1.
In summary, there are two steps in the regionalization of the developing mid/hindbrain (Adams et al.,2000; Beddington and Robertson,1998). The first step is the positioning of the future MHB in the early embryo, for which Otx2 and Gbx2 are two major players;this process is Lmx1b-independent because in the Lmx1b-/- embryo there was no caudal shift of the Otx2+ domain, which marks the MHB. The second step of the regionalization begins when the isthmic organizer is established in the MHB. We propose that, as a key component of the genetic pathway underlying the isthmic organizer activity, Lmx1b is responsible for the initiation of Fgf8 expression and the maintenance of other key isthmic organizer genes.
We thank Dr A. P. McMahon for providing the Wnt1-Cre mice; Drs A. Simeone,S. Aizawa, G. R. Martin and P. Gruss for providing in situ probes; Drs A. Groves and W. L. Ye for help with whole-mount TUNEL staining; and Mrs Y. F. Zhang for technical assistance. This project was supported by grants from the National Natural Science Foundation of China (30525014, 30628016), the `973'Program and Key State Research Program from the Ministry of Science and Technology of China (2006CB806600 and 2006CB943900), and the `Pujing Project'of Shanghai, China (06PJ14116).