A Wnt1-regulated genetic network controls the identity and fate of midbrain-dopaminergic progenitors in vivo

Neurons synthesizing the neurotransmitter dopamine are found in different areas of the mammalian brain. The most prominent dopaminergic (DA) cell group is that of the ventral midbrain, comprising the substantia nigra (SN), the ventral tegmental area (VTA) and the retrorubral field (RrF)(Dahlstrom and Fuxe, 1964). Midbrain dopaminergic (mDA) neurons modulate a broad range of processes in the brain, including movement, cognition and reward. Thus, degeneration or dysfunction of mDA neurons in the human brain leads to severe neurological and psychiatric disorders, among them Parkinson's disease (PD)(Cooper et al., 2001; Lang and Lozano, 1998).

Understanding the development and maintenance of mDA neurons is of high clinical interest as replacement of this cell population in the diseased brain is considered to be one of the most promising therapeutic approaches for PD(Lindvall et al., 2004). Despite considerable advances made in recent years, the factors and steps controlling the development and maintenance of mDA neurons are far from being fully identified and understood. mDA neurons develop in the ventral midbrain in close vicinity to two important signaling centers of the embryonic neural tube, the floor plate (FP) and the mid-hindbrain boundary (MHB). The FP comprises specialized cells that secrete the glycoprotein sonic hedgehog(Shh). In turn, Shh regulates the expression of a variety of transcription factors, whose expression code confers the cellular identity along the dorsoventral axis of the neural tube(Jessell, 2000). One of these Shh-responsive genes is Nkx2-2, which encodes a type II homeodomain transcriptional regulator required for the specification of ventral cell populations in the hindbrain and spinal cord(Briscoe et al., 1999; Pattyn et al., 2003a; Pattyn et al., 2003b). The MHB is established at the expression border of two transcriptional repressors,Otx2 in the fore- and midbrain and Gbx2 in the hindbrain(Liu and Joyner, 2001a; Prakash and Wurst, 2004; Wurst and Bally-Cuif, 2001). In turn, transcription of the secreted proteins Wnt1 and fibroblast growth factor 8 (Fgf8) is initiated at the MHB, and transcription factors belonging to the engrailed (En) and Pax families are expressed across the MHB. Using explant cultures, Ye et al. (Ye et al.,1998) demonstrated that both Shh and Fgf8 are together required for induction of mDA neurons at ectopic locations, thus suggesting that the signals coming from the FP and the MHB play an important role in the development of these neurons. We have recently shown that the position of the MHB indeed controls the location and size of the mDA neuronal population(Brodski et al., 2003). Other factors implicated in the terminal differentiation and maintenance of mDA neurons are the LIM-homeodomain factor Lmx1b, transforming growth factors(TGFs) α and β, the En proteins, the orphan nuclear receptor Nr4a2(Nurr1), and the mDA-specific paired-like homeodomain transcription factor Pitx3 (Alberi et al., 2004; Blum, 1998; Farkas et al., 2003; Hwang et al., 2003; Maxwell et al., 2005; Nunes et al., 2003; Simon et al., 2001; Smidt et al., 2000; Smidt et al., 2004; van den Munckhof et al., 2003; Zetterstrom et al., 1997). However, the precise mechanism that links early inductive signals to the molecular network regulating the differentiation and maintenance of mDA neurons still remains elusive.

Wnts are secreted palmitoylated glycoproteins involved in the control of cell proliferation, differentiation, polarity, migration and death(Baek et al., 2003; Hirabayashi et al., 2004; Megason and McMahon, 2002; Willert et al., 2003). Wnt1 is expressed in a ring encircling the neural tube at the MHB, in the dorsal midline (roof plate) of the midbrain, and in the ventral midline [FP and basal plate (BP)] of the cephalic flexure. The latter Wnt1 expression domain coincides with the region where mDA progenitors first arise at mouse embryonic day 9.5 (E9.5) (see Fig. S1 in the supplementary material) and mDA neurons later develop at E10.5-12.5. Previous in vitro studies indicated that Wnt5a,another member of the Wnt family (Parr et al., 1993), promotes mDA neuron differentiation, whereas Wnt1 predominantly enhances mDA progenitor proliferation(Castelo-Branco et al., 2003). In order to assess the role of Wnt1 in the generation of mDA neurons in vivo,we used different transgenic mouse lines as well as explant cultures. Here, we provide evidence for the first time that Wnt1 is indeed required for the generation of mDA neurons in vivo by controlling a molecular cascade that leads to the establishment of the mDA progenitor domain in early neural development and to the acquisition of the full mDA phenotype at later developmental stages.

Generation and genotyping of En1^+/Wnt1 and En1^+/Otx2 knock-in, of conditional En1^+/Cre; Otx2^flox/flox, and of Nkx2-2^-/- and Wnt1^+/-(int1^+/-) single mutant mice is described elsewhere(Broccoli et al., 1999; McMahon and Bradley, 1990; McMahon et al., 1992; Panhuysen et al., 2004; Puelles et al., 2004; Sussel et al., 1998; Thomas and Capecchi, 1990). Viable En1^+/Cre; Otx2^+/flox;Nkx2-2^+/- triple heterozygous and Nkx2-2^+/-;Otx2^+/flox double heterozygous mice were selected and intercrossed to generate compound En1^+/Cre;Otx2^flox/flox; Nkx2-2^-/- triple mutants.

Mouse embryos were fixed and processed for radioactive in situ hybridization as described elsewhere(Brodski et al., 2003). Probes used for in situ hybridization are as described previously(Brodski et al., 2003) and(Puelles et al., 2004).

Immunohistochemistry was performed as previously reported(Castelo-Branco et al., 2003; Puelles et al., 2004). The rabbit antibodies were directed against Otx2 (1:3500), Shh (1:200; Santa Cruz Biotechnology), Th (1:125; Pel-Freez) and Pitx3 (1:100; kindly provided by P. Burbach); the mouse antibodies were directed against Nkx2-2 (1:100; Hybridoma Bank), Th (1:300; Chemicon) and 5HT (1:100; Chemicon).

Injections of pregnant females with BrdU and immunochemical processing of the embryos was performed as described previously(Panhuysen et al., 2004).

Explant cultures of anterior neural plates of embryos derived from heterozygote Wnt1^+/- intercrosses were essentially prepared as reported previously(Echevarria et al., 2001).

Heparin-acrylic beads (Sigma) were soaked in 1 μg/μl recombinant mouse Fgf8b (R&D Systems) as described in(Echevarria et al., 2001). BSA(0.1% in PBS, Sigma)-coated beads were used as controls.

Explants were fixed and whole-mount in situ hybridization was carried out using standard procedures.

Explant cultures were prepared and treated as described above. Total RNA from pooled explants was reverse-transcribed using random hexamers and the Advantage RT-for-PCR Kit (BD Biosciences Clontech). 4 μl each of 1:5 diluted single-stranded cDNA was amplified with primer pairs specific for Th, Pitx3, Nr4a2, Aldh1a1, Wnt1 and GAPD. Primers and conditions are available upon request. cDNA from E12.5 CD1 mouse embryo heads was used as positive control. All gene-specific primer pairs except of Nr4a2 were intron spanning.

First, we analyzed the expression pattern of different genes associated with the MHB in transgenic mice in which one En1 allele was replaced by the Wnt1 cDNA [En1^+/Wnt1 knock-in mice(Panhuysen et al., 2004)]. In these mutant mice, the Wnt1 expression domain is expanded both anteriorly into the rostral midbrain and posteriorly into the rostral hindbrain (rostral third of rhombomere 1). Previous examination of these mice by means of whole-mount in situ hybridization revealed no alteration of the position and gene expression patterns at the MHB(Panhuysen et al., 2004). Most importantly, ectopic expression of Wnt1 in the mid-hindbrain region of the En1^+/Wnt1 mice resulted only in an increase in size of the inferior colliculi (a caudodorsal midbrain derivative) without repatterning of this region. It thus had been argued that Wnt1 has no patterning activity on its own. It was therefore unexpected when a more thorough analysis of the En1^+/Wnt1 transgenic mice using radioactive in situ hybridization showed a caudal expansion of the Otx2 domain but only within the FP of the rostral hindbrain(Fig. 1A). The caudal shift of Otx2 expression was detected from E9.5 onwards, 24 hours after activation of the En1 promoter (data not shown), and overlapped with the ectopic Wnt1 expression in the rostral hindbrain FP albeit not to the same caudal extent (Fig. 1A). As in the wild type, Fgf8 was not expressed within the FP of the rostral hindbrain in En1^+/Wnt1 embryos(Fig. 1A). Expression of other genes within the midbrain/rostral hindbrain FP such as Wnt5a and Lmx1b, was not changed in the En1^+/Wnt1 mutants(Fig. 1C and data not shown).

Concomitant with the ectopic induction of Otx2 within the rostral hindbrain FP of En1^+/Wnt1 mice, a caudal shift of the aldehyde dehydrogenase 1 family member A1 (Aldh1a1, also known as Raldh1) expression domain was detected in the FP of these mutants(Fig. 1A). Aldh1a1 is so far the only known marker for proliferating mDA progenitors and starts to be expressed at E9.5 within the cephalic flexure rostral to the MHB,overlapping with the Wnt1 expression domain(Wallen et al., 1999) (see Fig. S1 in the supplementary material). At E12.5, the ectopic mDA progenitors had indeed developed into fully differentiated mDA neurons in the En1^+/Wnt1 mutants, as judged by the expression of tyrosine hydroxylase (Th), Nr4a2 and Pitx3(Fig. 1B). At E15.5, the ectopic mDA neurons robustly expressed the dopamine transporter(Slc6a3), and they persisted into adulthood in En1^+/Wnt1 mice (see Fig. S2A,B in the supplementary material). The ectopic induction of Aldh1a1 correlated with a repression of Gbx2 in the same domain within the rostral hindbrain FP of En1^+/Wnt1 mutants(Fig. 1C), suggesting that Gbx2 was repressed by the ectopic Otx2 expression in this territory. Notably, no obvious increase in proliferating cells was detected in the FP of the En1^+/Wnt1 mutant midbrain and rostral hindbrain, as assessed by BrdU incorporation at E11.5(Fig. 1D), indicating that the appearance of the ectopic mDA neurons is not due to an overproliferation of their progenitor pool. Given the tight link between the ectopic expression of Wnt1 and Otx2 in the rostral hindbrain FP of En1^+/Wnt1 mutants and the concomitant induction of ectopic mDA neurons in this region, we hypothesized that Wnt1 and Otx2 together are required for the generation of ectopic mDA neurons. Therefore, we analyzed Wnt1 expression in the En1^+/Otx2 mouse mutant,which shows a caudal repositioning of the posterior border of the Otx2 expression domain and consequently of the MHB(Broccoli et al., 1999). In the En1^+/Otx2 transgenic mouse, additional mDA neurons are generated within the ectopic Otx2 domain(Brodski et al., 2003). At E11.5, the Wnt1 expression domain in the FP of En1^+/Otx2 embryos was enlarged caudally to the same degree as the ectopic mDA neurons (see Fig. S3 in the supplementary material). We thus concluded that a positive regulatory feedback loop controls ectopic expression of Wnt1 and Otx2 in the ventral midline (FP) of the hindbrain in these mutants, and that this regulatory network may be relevant for the subsequent ectopic generation of mDA neurons. We next asked for the downstream effectors of this regulatory feedback loop by analyzing another mouse mutant in which Otx2 was conditionally inactivated in the mid/hindbrain region, including the ventral midline[En1^+/Cre; Otx2^flox/flox mice(Puelles et al., 2004)].

Previous analyses of the conditional En1^+/Cre; Otx2^flox/flox mouse mutant(Puelles et al., 2004) had shown that the loss of Otx2 within the midbrain FP and BP leads to a ventral expansion of the Nkx2-2 domain (Fig. 2A). In addition, the Shh domain was expanded dorsally in this conditional mutant (Fig. 2A). At the same time, the mDA neuronal population (marked by Th expression) was drastically reduced in the ventral midbrain and ectopic serotonergic (5HT)neurons were generated instead in this region(Fig. 2A). Most notably, Wnt1 expression was lost only in the ventral midbrain (FP and BP) but left intact in the dorsal midline (roof plate, RP) of the midbrain of En1^+/Cre; Otx2^flox/flox conditional mutants at E12.5 (Fig. 2A). This finding would be consistent with a positive feedback between Otx2 and Wnt1 that is also maintaining their expression in the ventral midbrain. The almost complete loss of mDA neurons in the En1^+/Cre; Otx2^flox/flox mice could therefore be due to: (1) a specific and cell-autonomous requirement of Otx2 for the specification of mDA neurons; (2) the repression of the mDA neuronal fate by Nkx2-2 alone or in combination with Shh; and/or (3) a requirement of Wnt1 for the generation of mDA neurons. To address these possibilities we first investigated whether any alterations in the mDA and rostral 5HT cell populations could be detected in the Nkx2-2^-/- single mutants. As had been published before(Briscoe et al., 1999), neither the mDA nor the rostral 5HT neuronal populations were affected in size or location in the Nkx2-2^-/- single mutant embryos(Fig. 2A). Furthermore,expression of Otx2, Shh and Wnt1 was not affected in the ventral midbrain of Nkx2-2^-/- mice(Fig. 2A). Then we asked whether the ectopic expression of Nkx2-2 in the ventral midbrain of the conditional En1^+/Cre; Otx2^flox/floxmice was responsible for the loss of mDA neurons and the ectopic induction of 5HT neurons in this region. To resolve this issue, compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutant mice were generated.

The mDA neuronal population was rescued in the ventral midbrain of the compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutants compared with the En1^+/Cre; Otx2^flox/flox mutant embryos(Fig. 2A). Furthermore, no ectopic 5HT neurons were detected in the ventral midbrain of the triple mutants (Fig. 2A). The rescue of mDA neurons occurred in the triple mutant, even though Otx2 was completely lost in the midbrain FP and BP from the earliest time point studied (E9.5)(Puelles et al., 2004)(Fig. 2A). Importantly, the ventral Wnt1 expression domain was also rescued in the midbrain of the compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutants(Fig. 2A), although it had a somewhat `fuzzy' appearance in the compound triple mutant. The reason for this is unclear at present. The rescue of the ventral Wnt1 expression domain suggested that ectopic expression of Nkx2-2 in the ventral midbrain of conditional En1^+/Cre; Otx2^flox/floxmutant embryos also has a direct or indirect repressive effect on the transcription of Wnt1 in this region. However, it also showed that Otx2 is not necessary for maintaining Wnt1 expression in the ventral midbrain of compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutants. Furthermore, the comparison of the phenotypes observed in the conditional En1^+/Cre; Otx2^flox/floxmouse and the compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutant suggested that Otx2 is required for the repression of Nkx2-2 in the midbrain FP and BP, thereby establishing an Nkx2-2-negative territory from which mDA progenitors can develop (Fig. 2B).

To confirm this hypothesis, we reinvestigated the En1^+/Wnt1 mutants, in which ectopic mDA neurons were generated within the Otx2-positive hindbrain FP. Coronal sections of E12.5 En1^+/Wnt1 embryos taken at the rostral hindbrain level revealed that the ectopic Th-expressing cells were located in the mantle zone of the ectopic Otx2-positive neuroepithelium within the Shh and Wnt1 expressing FP of the mutant hindbrain(Fig. 3A). The ventral Nkx2-2 expression domain of the hindbrain, which extended ventrally towards, but did not overlap with, the FP in wild-type embryos(Fig. 3B), appeared to be repressed in its most ventral part overlapping with the region where the ectopic Otx2 was induced and the ectopic mDA neurons were generated in the En1^+/Wnt1 mutant (brackets in Fig. 3B). Changes of gene expression patterns in the hindbrain of En1^+/Wnt1 mutants were much more subtle. Thus, the tissue surrounding the ectopic mDA neurons retained its hindbrain identity as demonstrated by the presence of rostral hindbrain 5HT neurons, which were somewhat displaced but otherwise unaffected(Fig. 3A). Furthermore,expression of other marker genes, such as Lmx1b, Wnt5a and Lhx1 was not changed in the rostral hindbrain tissue of En1^+/Wnt1 mutants surrounding the ectopic mDA neurons(data not shown). Other cell populations of the ventral mid- and hindbrain(such as motoneurons of the III and IV brain nuclei, and the red nucleus) also appeared not to be changed in En1^+/Wnt1 mice in regard of their location and size (data not shown). We thus concluded that the ectopic induction of mDA neurons in the hindbrain FP of En1^+/Wnt1mutants is specific for this cell population, and is based on the ectopic induction of Otx2, which in turn represses the most ventral Nkx2-2 expression domain (Fig. 3C). Thereby, an ectopic mDA progenitor domain is established in the hindbrain FP of the En1^+/Wnt1 mutants.

Our data support the conclusion that a Wnt1-controlled genetic network leads to the establishment of the mDA progenitor domain in the ventral midbrain during early stages of mouse neural development (i.e. between E9.5 and E12.5) by repressing Nkx2-2 within this domain. The loss of Wnt1 expression in the ventral midbrain of conditional En1^+/Cre; Otx2^flox/flox mutants (which also lack mDA neurons) and the rescue of Wnt1 expression in the same region of compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutants (in which mDA neurons were also rescued) suggested that Wnt1 in addition plays a role in mDA cell fate specification. To clarify this point, a thorough analysis of the mDA domain was performed in Wnt1^-/- null mutant mice(McMahon and Bradley, 1990; Thomas and Capecchi, 1990). Based on the expression of En1 and on the radial glial marker RC2, a residual ventral mid-hindbrain domain could still be detected in Wnt1^-/- mice between E9.5 and E11.5(Fig. 4A,B). This is in line with a previous report (McMahon et al.,1992). Within this residual En1-positive domain, cells expressing Nr4a2 and Th were detected between E10.5 and E12.5(Fig. 4C; data not shown). The number of Th-expressing neurons, however, was strongly reduced in the Wnt1^-/- null mutant at E11.5 when compared with the wild type (Fig. 4C). Remarkably, the Th-positive cells did not express other mDA neuron-specific marker genes, such as Pitx3 or Slc6a3, at the time points analyzed in the Wnt1^-/- mutants (Fig. 4C; data not shown). Furthermore, no Aldh1a1-expressing cells could be detected between E9.5 and E10.5 in the Wnt1^-/- embryos (Fig. 4A), indicating a loss of proliferating mDA progenitors. To rule out the possibility that these Th-positive cells are noradrenergic neurons of the rostral hindbrain, expression of the enzyme dopamine-β-hydroxylase(Dbh, required for the synthesis of norepinephrine) was analysed in Wnt1^-/- mutants at E11.5 and E12.5. However, no Dbh signal was detected in these cells (data not shown). In addition, Otx2 was normally expressed and Nkx2-2 was repressed in the region where the Th-positive cells arose (data not shown), suggesting that the mDA progenitor domain was still established in the Wnt1^-/-embryos. This observation may be explained by the partially redundant action of other members of the Wnt family, in particular that of Wnt5a and Wnt7a,which are expressed in the same region(Castelo-Branco et al., 2003; McMahon et al., 1992; Parr et al., 1993). Therefore,we concluded that a DA precursor still develops in the remnants of the ventral midbrain in the Wnt1^-/- mutant, but this precursor does not properly proliferate and differentiate into an mDA neuron in the absence of Wnt1 as judged by the lack of Pitx3 and Slc6a3 expression(Fig. 4D). Even though the DA precursors in the Wnt1^-/- knockout start to express Th,these cells were probably dying as they did not initiate their correct differentiation program. In agreement with this possibility, active caspase 3 staining was detected in the ventral midbrain of Wnt1^-/-embryos at E11.5 (data not shown) and Th-positive cells were not detected in these embryos at stages later than E12.5, concomitant with the complete loss of the mid-hindbrain region (data not shown)(Chi et al., 2003).

To further support the possibility that Wnt1 is indeed required for the generation of mDA neurons, in addition to other factors such as Fgf8 and Shh,we repeated the explant experiments performed by others(Ye et al., 1998) but using the whole anterior neural plate from Wnt1^-/- mutant and wild-type (Wnt1^+/+ and Wnt1^+/-) mouse embryos at E8.0-8.5. It had previously been shown that ectopic mDA neurons were induced in forebrain explants only in the presence of both Shh and Fgf8(Ye et al., 1998). We used whole anterior neural plate explants (comprising the presumptive fore-, mid-and hindbrain) for our experiments because the presumptive mid-hindbrain region will be eventually deleted in Wnt1^-/- explants, and any negative result could therefore also be due to the loss of this tissue. Additional experiments have shown that Wnt1 expression in the presumptive midbrain is maintained by Fgf8b-coated beads implanted in the presumptive ventral forebrain (Fig. 5B). Indeed, 6 days after implantation of an Fgf8b-coated bead close to the ventral midline of the presumptive forebrain, Th-expressing cells were detected at a distance from the bead in wild-type explants (Fig. 5A). By contrast, no Th-positive cells were detected in Wnt1^-/- explants under the same conditions(Fig. 5A). The mDA identity of the induced Th-positive cells was further confirmed by RT-PCR for the specific marker Pitx3. After 6 days of culture in the presence of an Fgf8b-coated bead, expression of Pitx3 was only detected in wild-type but not in Wnt1^-/- explants(Fig. 5C). These results confirmed that, even in the presence of Fgf8 and Shh, no ectopic mDA neurons can be induced in the absence of Wnt1, indicating that Wnt1 is necessary for the generation of mDA neurons.

We provide evidence that the secreted glycoprotein Wnt1 plays a crucial role in the generation of mDA neurons in vivo by exerting two different activities.

First, Wnt1 is involved in the establishment of the mDA progenitor domain by maintaining Otx2 expression in the ventral midbrain, which in turn is required for the repression of Nkx2-2 in this territory(Fig. 6A). Failure to repress Nkx2-2 in the midbrain FP and BP leads to the generation of 5HT instead of mDA neurons. The function of Wnt1 in this early developmental context may be similar to the role of Fgf8 in the maintenance of Gbx2expression within the rostral hindbrain(Chi et al., 2003; Liu and Joyner, 2001b; Liu et al., 1999; Sato et al., 2001). As Fgf8 can induce or maintain Wnt1 expression non-cell-autonomously(Chi et al., 2003; Liu and Joyner, 2001b), and Wnt1 is required for maintaining Fgf8 expression in the rostral hindbrain (Lee et al.,1997), this early activity of Wnt1 may be part of an integrated regulatory network controlling the maintenance of the MHB and consequently the establishment of the distinct progenitor domains in the ventral mid- and hindbrain. It should be noted, however, that this Wnt1 activity appears to be restricted to the ventral neural tube, as the overexpression of Wnt1in the dorsal neural tube leads only to an enhanced proliferation of dorsal neural precursors (Panhuysen et al.,2004). Our results also revealed a so far underestimated plasticity of the ventral mid-and hindbrain neural precursors, in the sense that these precursors are able to generate both mDA and rostral 5HT neurons,depending on the genetic program and the environmental cues acting on these cells. Furthermore, the specification of rostral hindbrain 5HT neurons appears to depend in part on the same signals and factors required for the development of the more-caudal 5HT neuronal populations in the hindbrain and spinal cord,namely Shh and its downstream effectors of the Nkx family(Briscoe et al., 1999; Pattyn et al., 2003a). This finding is in line with a recent report on a different conditional Otx2 mouse mutant (Vernay et al.,2005). Given the particular expression pattern of Nkx2-2 in the ventral midbrain, a repressive effect of this transcription factor on the specification of the mDA neuronal fate would not be detected in the Nkx2-2^-/- single mutant. The normal appearance of the rostral 5HT population in the Nkx2-2^-/- mouse, however,indicates that the loss of Nkx2-2 in the rostral hindbrain must be compensated by yet another factor, probably the related Nkx2-9 transcriptional regulator showing a similar expression pattern as Nkx2-2(Briscoe et al., 1999; Pattyn et al., 2003a).

Second, Wnt1 has a later specifying activity required for the proper differentiation of mDA precursors into mature mDA neurons(Fig. 6B). In the absence of Wnt1, no Aldh1a1 expression was detected at E9.5 and E10.5,and less than 1% of Th-positive cells were born within a residual En1-positive ventral mid-hindbrain domain. Moreover, these cells did not express Pitx3 and Slc6a3. It is possible that for the generation of the few Th-positive cells in the Wnt1^-/-mutants, other members of the Wnt family, such as Wnt5a and Wnt7a may have in part compensated for the loss of Wnt1, as they are expressed in this region and Wnt5a exhibits partially overlapping activities in vitro(Castelo-Branco et al., 2003; McMahon et al., 1992; Parr et al., 1993). Another possibility is that the remaining Th-expressing cells in the Wnt1^-/- embryos are generated independently of a Wnt signal and may directly derive from the medial FP, similar to the few Th-positive neurons in the conditional En1^+/Cre; Otx2^flox/flox mutants. Previous studies of mDA neuron development have suggested that two independent pathways are active during terminal differentiation of these cells: a Nr4a2-controlled pathway required for initiation of Th and Slc6a3 transcription(Kim et al., 2003; Saucedo-Cardenas et al., 1998; Smits et al., 2003; Zetterstrom et al., 1997); and an Lmx1b-controlled pathway necessary for Pitx3 expression(Smidt et al., 2000). As Wnt1 is induced and/or maintained by Lmx1b, but not vice versa(Adams et al., 2000; Matsunaga et al., 2002), and as we could not detect any changes of Lmx1b expression patterns in our mutant mice, it is very likely that Wnt1 is acting downstream of Lmx1b in the regulation of Pitx3 expression. Pitx3 is a homeobox transcription factor uniquely expressed in all mDA neurons(Smidt et al., 2004; Smidt et al., 1997; Zhao et al., 2004), and loss-of-function experiments have shown that Pitx3 is required for the terminal differentiation and survival of a subset of mDA neurons, mainly those of the SN (Hwang et al.,2003; Maxwell et al.,2005; Nunes et al.,2003; Smidt et al.,2004; van den Munckhof et al.,2003). Notably, we did not detect Slc6a3 expression in the Wnt1^-/- mutant at the time points analysed. As Slc6a3 expression is found in the remaining Th-positive cells of the Pitx3mutant (Hwang et al., 2003; Smidt et al., 2004) but not in the Nr4a2^-/- (Smits et al., 2003) or Wnt1^-/- embryos, it could be that transcription of the Slc6a3 gene is controlled by both Wnt1 and Nr4a2. However, it is not known whether activation of the Slc6a3 gene is controlled directly or indirectly by Wnt1. The ability of Wnts to promote the acquisition of a mDA phenotype was recently demonstrated in vitro using ventral midbrain precursor cultures(Castelo-Branco et al., 2003). In this study, Wnt1 had a predominantly proliferative effect on these neural precursors, whereas Wnt5a, another member of the Wnt family expressed throughout the FP of the mid- and hindbrain(Parr et al., 1993), promoted the differentiation of ventral midbrain precursors into mDA neurons without affecting their proliferation in vitro. However, in contrast to the in vitro studies, our present findings show that Wnt1 alone could control the specification of the mDA neuronal fate in vivo. The drastically reduced number of Th-positive cells in the Wnt1^-/- mouse, however,suggests that Wnt1 in addition controls the proliferation of mDA progenitors in vivo. Importantly, the appearance of ectopic mDA neurons in the hindbrain of En1^+/Wnt1 mice and in wild-type but not in Wnt1^-/- explants together with the severe loss of DA neurons in the Wnt1^-/- mutants indicate that Wnt1 has a crucial role in the development of mDA neurons. One reason for the discrepancy with the in vitro results could be that cultured ventral midbrain precursors lose domain-specific positional information and do not keep the regulatory network normally controlling their differentiation into mDA neurons in vivo. It is not yet clear how the Wnt1-signal is transduced in mDA progenitors. However, recent in vitro data have shown that activation of the canonical Wnt-pathway in ventral midbrain precursor cultures enhances the differentiation of DA neurons in these cultures(Castelo-Branco et al., 2004),suggesting that this pathway may as well participate in mDA cell differentiation in vivo.

Our data also indicate that in the absence of Wnt1, both Shh and Fgf8 are not sufficient for ectopic induction of mDA neurons. The activity of these two secreted factors in mDA fate specification has recently been investigated in vivo. First, a dorsal expansion of the ventral midbrain Shh territory does not always correlate with an ectopic induction or increase of the mDA population. The ventral Shh domain is dorsally expanded in conditional Otx1^+/Cre; Otx2^-/flox mutant embryos,in which Otx2 is inactivated in the lateral midbrain but left intact in the midbrain FP and BP (Puelles et al.,2003). In these conditional mutants, the number of mDA neurons is remarkably enlarged owing to an expansion of the mDA progenitor domain that is positive for Shh and to an increased proliferation of their precursors (Puelles et al.,2003). A similar dorsal expansion of the Shh domain is seen in the conditional En1^+/Cre; Otx2^flox/floxmutants, but in these mice, the mDA phenotype is almost completely absent(Puelles et al., 2003; Puelles et al., 2004). This indicates that Shh may rather act as a mitogen on mDA precursors, thus expanding their progenitor pool during development, but cannot compensate on its own for the loss of the mDA fate in the conditional En1^+/Cre; Otx2^flox/flox mutant. Second, the position and extent of the Fgf8 expression domain in the ventral hindbrain is not affected in En1^+/Cre; Otx2^flox/flox mice(Puelles et al., 2004),indicating that Fgf8 alone or together with the dorsally enlarged Shh domain is also not able to rescue the mDA cell population in these mutants. Notably,the different phenotypes of the two conditional mouse mutants regarding the mDA neuronal population can be explained by the differences in Otx2,Nkx2-2 and Wnt1 expression between them. Expression of Otx2 in the midbrain FP and BP is unaffected in the Otx1^+/Cre; Otx2^-/flox embryos and, as a consequence, Nkx2-2is still repressed and mDA neurons are present in this region of the mutant midbrain (Puelles et al.,2003). However, loss of Otx2 in the midbrain FP and BP of En1^+/Cre; Otx2^flox/flox embryos leads to a ventral expansion of the midbrain Nkx2-2 domain and to the loss of mDA neurons [this work and that of Puelles et al.(Puelles et al., 2004)]. Most importantly, the specification of the mDA neuronal fate strongly correlates with the expression of Wnt1 in the ventral midbrain or hindbrain of all mutant mouse lines analysed. Thus, the Wnt1 domain and mDA neurons are unaffected in the ventral midbrain of conditional Otx1^+/Cre; Otx2^-/flox, of Nkx2-2^-/- single and of compound En1^+/Cre; Otx2^flox/flox; Nkx2-2^-/- triple mutants [this work and that Puelles et al. (Puelles et al., 2003)]. By contrast, Wnt1 expression is lost and mDA neurons are miss-specified in the ventral midbrain of conditional En1^+/Cre; Otx2^flox/flox and Wnt1^-/- embryos. Finally, ectopic expression of Wnt1 together with Otx2 in the rostral hindbrain FP of En1^+/Wnt1 and En1^+/Otx2 transgenic mice is sufficient to induce ectopic mDA neurons in this region of the mutant hindbrain. Our results therefore indicate that Shh and Fgf8 act in concert with a Wnt1-controlled regulatory network (including Otx2 and Nkx2-2), leading to the establishment of the mDA progenitor domain in the ventral midbrain and the subsequent terminal differentiation of mDA neurons. Furthermore, our present data suggest the Wnt1-controlled signaling pathway as a promising target in the treatment of diseases affecting mDA neurons, including PD.

The authors thank R. Koester for critical reading of the manuscript; M. Capecchi for providing the Wnt1 (Int1) mutant mice; and S. Laass, A. M. Lucchese, S. Pirrung, M. Lozada, K. Karim, M. Oesterheld and A. Osterrieder for excellent technical assistance. This work was supported by the SSF, VR, KI, EU, Swedish Royal Academy of Sciences and Wallenberg Foundation grants to E.A.; by the MRC, Wellcome Trust, Italian Association for Cancer Research (AIRC), FIRB Genotpo, FIRB Neuroscienze and the Fondation Bettencourt-Schueller to A.S.; by the Federal Ministry of Education and Research (BMBF) in the framework of the National Genome Research Network(NGFN), Förderkennzeichen 01GS0476 (to W.W.); by the BMBF Förderkennzeichen 01GN0512 (to W.W.); by the Deutsche Forschungsgemeinschaft (DFG) WU 164/3-1 and WU 164/3-2 (to W.W. and N.P.); and by European Union (Eumorphia) grants to W.W. A.H. was supported by HFSP. The authors are responsible for the contents of this publication. The authors declare that they have no competing financial interests.