beta-Catenin Regulates Intercellular Signalling Networks and Cell-Type Specific Transcription in the Developing Mouse Midbrain-Rhombomere 1 Region

β-catenin is a multifunctional protein involved in both signalling by secreted factors of Wnt family and regulation of the cellular architecture. We show that β-catenin stabilization in mouse midbrain-rhombomere1 region leads to robust up-regulation of several Wnt signalling target genes, including Fgf8. Suggestive of direct transcriptional regulation of the Fgf8 gene, β-catenin stabilization resulted in Fgf8 up-regulation also in other tissues, specifically in the ventral limb ectoderm. Interestingly, stabilization of β-catenin rapidly caused down-regulation of the expression of Wnt1 itself, suggesting a negative feedback loop. The changes in signal molecule expression were concomitant with deregulation of anterior-posterior and dorso-ventral patterning. The transcriptional regulatory functions of β-catenin were confirmed by β-catenin loss-of-function experiments. Temporally controlled inactivation of β-catenin revealed a cell-autonomous role for β-catenin in the maintenance of cell-type specific gene expression in the progenitors of midbrain dopaminergic neurons. These results highlight the role of β-catenin in establishment of neuroectodermal signalling centers, promoting region-specific gene expression and regulation of cell fate determination.


Introduction
Wnts are a family of secreted lipoproteins, which play a crucial role during embryogenesis via the regulation of patterning, cell fate decision and cell polarity. Wnts mediate their intracellular effects by inducing stabilization and nuclear translocation of bcatenin. In the absence of Wnt ligands, cytoplasmic b-catenin is phosphorylated by glycogen synthase kinase 3b (GSK3b). Phosphorylated b-catenin is targeted for ubiquitination and proteasomal degradation. Binding of Wnt molecules to their cell surface receptors releases b-catenin from the destruction complex followed by accumulation and nuclear translocation of b-catenin. Nuclear b-catenin complexes with the TCF/LEF family of transcription factors to regulate Wnt target gene expression (reviewed in [1]).
During development of CNS Wnt signalling acts as important posteriorizing factor and correct anterior-posterior (AP) patterning requires anterior inhibition of Wnt pathway [2,3]. At the cellular level, canonical Wnt signalling endorses mitogenic pathway. While b-catenin maintains proliferation, [4], inactivation of b-catenin accelerates expression of neurogenic genes [5] and causes premature neuronal differentiation [6]. On the other hand, overexpression of stabilized b-catenin in cortical precursors leads to increased cell cycle re-entry and subsequent overproduction of neurons [7].
In the midbrain, early Wnt activity is responsible for establishment of a local organizing centre, the isthmic organizer [8]. Inactivation of Wnt signalling via Wnt1 or b-catenin gene ablation results in the deletion of posterior midbrain and part of cerebellum [9,10,11]. Wnt family members also play multiple roles in generation of midbrain dopaminergic neurons in vivo and in vitro [12,13,14].
In this work we studied the role of b-catenin in neuronal development in the midbrain. For this, we applied spatially and temporally controlled stabilization and inactivation of b-catenin in mouse embryos. Our findings underscore b-catenin as important transcriptional co-factor regulating midbrain gene expression and patterning.

Generation and genotyping of mice and embryos
Generation and genotyping of an Engrailed 1 (En1) allele carrying Cre-recombinase knock-in [15], Rosa26 locus carrying tamoxifen inducible R26cre-ert [16], conditional b-catenin loss-offunction allele [11], conditional b-catenin loss-of-exon3 allele [17] and transgenic mice expressing LacZ gene under control of bcatenin/TCF responsive elements [18] were described elsewhere. For staging, the day of vaginal plug was counted as embryonic day 0.5 (E0.5). To induce Cre-recombinase in R26cre-ert mice, pregnant females were given intraperitoneal injection of tamoxifen (Sigma) (8 mg/40 g body weight). All animal work has been conducted according to relevant national and international guidelines. Approval has been obtained from the Finnish Committee of Experimental Animal Research.

In situ mRNA hybridization
Whole-mount mRNA in situ analysis was performed by a modified protocol [19] using a digoxigenin-labeled antisense probes. Radioactive mRNA in situ hybridization on paraffin sections was performed as described previously [20] using 35

Immunofluorescence
Immunfluorescent staining on paraffin sections was performed as described previously [25]. Primary antibodies used were mouse monoclonal against b-catenin, (BD South San Francisco, CA) and rabbit polyclonal against Lmx1a (a gift from Michael German), tyrosine hydroxylase (Chemicon) and Aldh1a1 (Abcam, Cambridge, UK).

Microscopy
Whole-mount staining was visualized with a Leica MZFLIII microscope and photographed using an Olympus DP50-CU camera. Staining on paraffin sections were visualized with an Olympus AX70 microscope and photographed using an Olympus DP70 camera. Images were processed and assembled using Adobe Photoshop software. Confocal images were acquired using the Leica TCS SP5 confocal system and LAS-AF software. Confocal stacks and images were processed and deconvoluted using Imaris 6.1 (Bitplane) and AutoQuantX (AutoQuant) software.
To analyse effect of b-catenin stabilization on target gene expression, we carried out whole mount in situ hybridization using probes against direct targets of b-catenin including Tcf1, Dkk1, Axin2 and Drapc1. At E8.5 Drapc1 expression was confined to the midbrain and diencepalon (Fig. 1c,c9). A day later at E9.5, Drapc1 was drastically up-regulated in the dorsal midbrain and also expressed ectopically in the dorsal hindbrain of b-catenin stab embryos (Fig. 1d,d9). Axin2, which was absent from rhombomere1 in wt (wt) embryos, was strongly induced in dorsal midbrain and rhombomere1 in bcatenin stab embryos (Fig. 1e,e9). Profound increase in Tcf1 and Dkk1 mRNA levels was also observed in dorsal midbrain-rhombomere 1 region ( Fig. 1f-g9). Taken together, these results indicate that transcriptional targets of b-catenin are intensely up-regulated upon b-catenin stabilization in rhombomere1 and midbrain.
Effect of b-catenin stabilization on Fgf8 and Wnt1 expression at the isthmic organizer Using tissue-electroporation of chick embryos, induction of Fgf8 by Wnt1 was demonstrated in the anterior hindbrain [26,27]. However, in mouse embryos ectopically expressing Wnt1, only subtle caudal expansion of the Fgf8 expression domain in the hindbrain was reported [28]. Suggesting regulation of Fgf8 expression by the canonical Wnt pathway, we observed strong up-regulation and caudal spreading of Fgf8 expression in the hindbrain of b-catenin stab embryos already at E8.5 (Fig. 2d,d9,f, f9). At E9.5-E10.5, the Fgf8 expression encompassed the whole rhombomere1 and was highly up-regulated (Fig. 2h,h9, j). Interestingly, Fgf8 expression in the limb bud of was expanded to cover the ventral limb ectoderm where En1 Cre also drives recombination of the b-catenin flox(ex3) allele and stabilization of the b-catenin protein in the b-catenin stab embryos (Fig. 2j9). This cell context independence suggests a rather direct regulation of Fgf8 gene expression by b-catenin.
In turn, Fgf8 is important to maintain expression of Wnt1 in the early mouse midbrain [29] and Fgf8 beads induce expression of Wnt1 in chick embryos [30]. Therefore, we studied how overexpression of Fgf8 in b-catenin stab embryos affects Wnt1 expression. Surprisingly, Wnt1 mRNA was undetectable in the midbrain of bcatenin stab embryos already at E8.5 (Fig. 2b,b9). Thus, in addition to the positive feedback-loop through Fgf8, Wnt1 also appears to directly negatively regulate its own expression through b-catenin.

b-catenin stabilization results in anterio-posterior patterning defects
In the b-catenin stab embryos, strong b-catenin signalling and upregulation of Fgf8 can be expected to cause defects in anteroposterior patterning. Indeed, Gbx2 expression is enhanced and extended into anterior midbrain in the b-catenin stab embryos (Fig. 3a,b). Despite expansion of the Gbx2 expression, Otx2 is not down-regulated in the midbrain but it instead is ectopically expressed in the rhombomere1 (Fig. 3c,d). Taken together, these results indicate that b-catenin regulates anterior-posterior patterning directly by promoting midbrain fate and indirectly by activating Fgf8. Activation of both b-catenin target genes and Fgf8 pathway allows co-expression of Gbx2 and Otx2 and disrupts midbrain-hindbrain boundary formation in b-catenin stab embryos.
b-catenin stabilization affects dorso-ventral patterning and neurogenesis in the midbrain Wnt family members are expressed in the ventral midbrain as well as dorsal roof plate. Thus, transcriptional regulation by b-catenin may affect dorso-ventral patterning either directly or through other signalling molecules. In b-catenin stab embryos, the ventralizing morphogen Shh, normally expressed in the midbrain floor plate and basal plate, expanded from the ventral midbrain towards the dorsal midbrain (Fig. 3e,f). This shift was most prominent in the anterior midbrain negative for ectopic Gbx2 activation (see above). Consistent with the spreading of the Shh, expression of Gli1, a target and a mediator of Shh signalling, was enhanced in the dorsal midbrain of b-catenin stab embryos (Fig. 3g,h). Despite spreading of the Shh expression, the dorsal marker Pax3, shown to be repressed by Shh [31] was only slightly down-regulated (Fig. 3i,j). Although in bcatenin stab embryos b-catenin is activated both in the midbrain and rhombomere 1, the dorsal spreading of Shh and Gli1 expression was observed only in the midbrain-like tissue.
Similar to dorsal expansion of Shh and Gli1, we observed early up-regulation and dorsal expansion of several other ventrally expressed genes. However, the ventral up-regulation was often transient and was followed by apparently complete inactivation of expression in the ventral tissue. These genes included Ngn2, which was up-regulatedin the midbrain of b-catenin stab embryos at E9.0 (Fig. 4a,a9). In contrast, high Ngn2 levels in the dorsal midbrain were paralleled by a dramatic decline of Ngn2 expression in ventral rhombomere1 and midbrain of b-catenin stab embryos at E9.5-10.5 ( Fig. 4b-c9). Also expression of another proneural gene Mash1 was down-regulated in the ventral midbrain (Fig. 4l, l9).
Similar changes were seen in markers of specific ventral neuronal populations. In the progenitors of the dopaminergic neurons developing in the most ventral midbrain, we observed up-regulation of Lmx1a in b-catenin stab embryos at E9.0 (Fig. 4d,d9). However, this increase was temporary and reversed to Lmx1a down-regulation 12 hours later (Fig. 4e,e9). Similar to Lmx1a, expression of Aldh1a1, a specific early marker for dopaminergic neuron progenitors, was up-regulated in ventral domain of b-catenin stab embryos at E9.0 (Fig. 4g,g9,h,h9) but decreased by E10.5 (Fig. 4i,i9). In contrast to the dopaminergic neuron progenitor domain, Nkx2.2 expressing progenitors were markedly reduced in the ventral midbrain of bcatenin stab embryos already at E9.0 (Fig. 4j,j9,k,k9). Homeobox transcription factor Phox2a normally expressed in ventral motoneuron precursors of the oculomotor nucleus (III) in the midbrain and trochlear (IV) nucleus in the rhombomere 1. Phox2a expression was expanded to dorsal midbrain of b-catenin stab embryos (Fig. 4f,f9). The continued expression of Phox2a at E9.5-E10.5 may be due to its expression in the post-mitotic precursors as opposed to the proliferative progenitors. Interestingly, the dorsal spreading of Phox2a expression and putative motor neuron population was seen in the midbrain but not in the rhombomere 1. This parallels the changes observed in the Shh expression.
Thus, the effect of b-catenin stabilization on dorso-ventral gene expression largely depends on the developmental stage. Initially, the expression of ventral markers is increased and expanded. Later, several of the ventrally expressed genes are down-regulated in their normal expression domain but can be ectopically expressed in the dorsal regions.

b-catenin inactivation leads to the loss of transcriptional targets of Wnt signalling
To validate results of b-catenin stabilization in b-catenin stab embryos, we used conditional inactivation of b-catenin. The Cre recombinase was activated in R26 cre-ert/+ ; b-catenin flox/flox embryos at E8.5 or E9.5 and the effects of gene expression were analysed 1 day later at E9.5 or E10.5. The in situ hybridization, using probe against wild type b-catenin demonstrated significantly reduced bcatenin level throughout the neuroectoderm in b-catenin lof embryos. However, some residual b-catenin mRNA persisted (Fig. 5a,a9). As expected, Fgf8 mRNA was down-regulated (Fig. 5c,c9), Wnt1 mRNA was drastically up-regulated and ventrally expanded (Fig. 5d,d9) but Otx2 expression was not significantly changed (Fig. 5b,b9) in the b-catenin lof embryos. Ventral expression of Ngn2 was abolished at E10.5 (Fig. 5e,e9). Taken together, these data and findings obtained using b-catenin stabilization corroborate the role of b-catenin as a regulator of cell identities in the midbrain. Residual expression of some of the target genes may be due to incomplete abrogation of b-catenin expression (see below).

b-catenin inactivation distorts cell fate determination
We have shown previously that, consistent with incomplete inactivation of b-catenin mRNA in b-catenin lof mutants, the bcatenin negative cells were scattered within b-catenin positive tissue in a mosaic pattern [32]. We exploited the mosaic bcatenin lof embryos to study whether b-catenin cell-autonomously regulates transcription in the ventral dopaminergic neuron progenitors. Lmx1a protein level drastically decreased in bcatenin lof embryos treated with tamoxifen at E8.5 and analysed at E10.5 (Fig. 6b9,b0). Importantly, Lmx1a protein was reduced predominantly in b-catenin protein-null cells while b-catenin positive cells of b-catenin lof embryos mostly maintain normal level of Lmx1a. Lmx1a protein also decreased in large proportion of bcatenin knockout cells of b-catenin lof embryos treated with tamoxifen at E9.5 (analysis at E11.5) and at E10.5 (analysis at E12.5) (Fig. 6d9,d0,f9,f0). However, a few b-catenin protein-null cells still maintained Lmx1a expression (Fig. 6f9,f0) in E12.5 bcatenin lof embryos. This indicates that cell autonomous regulation of Lmx1a by b-catenin is stage dependent and as the neuronal progenitors progress towards differentiation, b-catenin -mediated control of Lmx1a is lost. In conclusion, our results suggest that bcatenin directly regulates expression of a cell-fate determining transcription factor in the proliferative dopaminergic neuron progenitors. Consistent with expression of Lmx1a in E12.5 b- catenin lof embryos the dopominergic marker tyrosine hydroxylase (TH) is detected in few b-catenin protein-null cells (Fig. 6g). However, majority of b-catenin protein-null cells lack TH.
Stabilization of b-catenin in the midbrain of E12.5 b-catenin stab embryos completely abolishes TH expression (data not shown). This is likely due to general mispatterning of the ventral midbrain.

Discussion
We have analyzed the role of b-catenin in development of the midbrain-rhombomere 1 region using spatially and temporally controlled stabilization and inactivation of the b-catenin locus in mice. We have shown that b-catenin regulates anterior-posterior and dorso-ventral patterning in the midbrain-rhombomere 1 region at least partly by activating expression of Fgf8 and Shh, two other important patterning signals. In addition, our results suggest direct cell-autonomous effects of b-catenin on the expression in the progenitor cells of dopaminergic neurons in the ventral midbrain.

Cross-talk between Wnt and Fgf8 signalling
Accumulation of the degradation resistant b-catenin in the midbrain and rhombomere1 leads to drastic up-regulation of the expression of Wnt target genes, including Drapc1, Axin2 and Tcf1. High level of Wnt/b-catenin signalling in b-catenin stab embryos also coincides with increase and caudal expansion of Fgf8 expression. In the wild-type embryos, Fgf8 is induced broadly in the rhombomere 1 but gets restricted to the most anterior rhombomere 1 next to the Wnt expressing posterior midbrain by E9.0. In the b-catenin stab embryos, Fgf8 is expressed in the entire rhombomere 1 still at E10.5, but is not induced in the midbrain despite robust activation of the b-catenin targets. This strongly suggests that the canonical Wnt signalling pathway, involving transcriptional regulation by b-catenin, maintains Fgf8 expression in the anterior rhombomere 1. Regulation of Fgf8 expression by b-catenin does not appear to be restricted to the rhombomere 1. b-catenin stabilization is sufficient to turn on Fgf8 expression in the ectoderm of ventral limb bud. Furthermore, analyses of the bcatenin lof mutants indicated that in addition to the rhombomere 1, also many other domains of Fgf8 expression are dependent on bcatenin activity. These observations together suggest a rather direct role for b-catenin in Fgf8 regulation.
On the other hand, inactivation of Fgf8 leads to loss of Wnt1 in midbrain-hindbrain border (reviewed in [33]. Surprisingly, Fgf8 up-regulation does not lead to increase in Wnt1 expression in bcatenin stab mutants. On the contrary, we observe complete downregulation of Wnt1. Consistently, Wnt1 was up-regulated in bcatenin lof mutants. Wnt1 may by directly negatively regulated by b-catenin/TCF complex or other intermediate players regulate its expression. The b-catenin stab embryos demonstrate overlapping expression of Otx2 and Gbx2 and consequent loss of midbrainhindbrain border. Correspondingly, anteriorization of Gbx2 expression could exert inhibitory effect on mesencepahlic gene expression and cause Wnt1 down-regulation [34].
b-catenin as transcriptional co-activator regulates patterning of the midbrain In addition to its role in anterior-posterior patterning, Wnt/bcatenin activation has previously been shown to promote dorsal gene expression and inhibit ventral gene expression in the telencephalon and spinal cord [35,36]. At E9.5-E10.5, this was also observed in the midbrain of b-catenin stab embryos. In contrast, at E9.0, a transient elevation of several of the ventrally expressed genes, including Ngn2, Aldh1a1 and Lmx1a was observed in the midbrain. This may reflect the fact that, unlike the spinal cord and telencephalon, the ventral midbrain has a strong endogenous Wnt signalling activity, which is used for ventral patterning [12,37]. The De-regulation of dorso-ventral patterning in b-catenin stab embryos is paralleled by changes in expression of Shh and Gli1 in the midbrain. Shh/Gli pathway plays a major role in ventral patterning of neural tube. In the spinal cord, the Shh pathway is antagonized by dorsally expressed Wnts by activation of the transcriptional repressor Gli3 [35]. Our results reveal somewhat different interactions of the two signalling pathways, as both Shh and Gli1 expression expanded dorsally in the midbrain of bcatenin stab mutants. However, we observe drastic reduction of Wnt1 also in the midbrain roof plate of b-catenin stab embryos, indicating that dorsal expansion of Shh maybe due to the lack of the roof plate specific gene expression. Interestingly, the dorsal expansion of Shh expression was not observed in the rhombomere 1, despite stabilization of b-catenin also in this region. This may reflect the fact that Shh expression is restricted to the floor plate in the rhombomere 1 but it is more broadly expressed in the basal plate in the midbrain. Thus, midbrain neuroepithelium may be more competent to activate Shh expression.
Activation of Shh signalling in the dorsal midbrain may also be the reason for ectopic dorsal Ngn2 expression and production of Phox2a positive precursors also in the dorsal midbrain.

b-catenin and cell fate determination
Ventral midbrain is the location where the dopaminergic neurons (DA) of substantia nigra and ventral tegmental area are generated. Lmx1a is an early and crucial activator of differentiation of ventral midbrain progenitors into DA neurons [38]. Lmx1a has been shown to be controlled by Shh. We show that Lmx1a is down-regulated in b-catenin lof embryos and that bcatenin is cell autonomously required for Lmx1a expression. The loss of Lmx1a in a subpopulation of ventral midbrain progenitors suggests that Lmx1a negative cells, which were otherwise destined to become DA neurons would adopt different fate. Specification of serotonergic neurons is also controlled by a combinatorial effect of Fgf8 and Shh and it would be tempting to speculate that serotonergic neurons are generated at the expense of DA neurons. However, we could not indentify any ectopic 5'HT staining (data not shown). Our results are consistent with a recently published work showing that b-catenin signalling is required for dopaminergic neuron development in the ventral midbrain [37,39]. Thus, in addition to regulation of other signalling pathways, b-catenin directly regulates cell fate specification of the midbrain dopaminergic neurons.
Loss of b-catenin in neuroepithelium has been shown to cause the defects in adherent junctions [39]. We have addressed the role of b-catenin in adherence in a parallel study, where we demonstrated that b-catenin maintains cell polarity and microtubules network in midbrain neural progenitor cells [32].
Taken together our results emphasize the role of b-catenin as a transcriptional co-regulator involved in anterior-posterior and dorsoventral patterning in developing midbrain and rostral hindbrain. In part, this is achieved by interactions with other signalling pathways. We further identify b-catenin as a cell-autonomous regulator of cell fate specification in the ventral mouse midbrain.