Wnt5a Regulates Ventral Midbrain Morphogenesis and the Development of A9–A10 Dopaminergic Cells In Vivo

Wnt5a is a morphogen that activates the Wnt/planar cell polarity (PCP) pathway and serves multiple functions during development. PCP signaling controls the orientation of cells within an epithelial plane as well as convergent extension (CE) movements. Wnt5a was previously reported to promote differentiation of A9–10 dopaminergic (DA) precursors in vitro. However, the signaling mechanism in DA cells and the function of Wnt5a during midbrain development in vivo remains unclear. We hereby report that Wnt5a activated the GTPase Rac1 in DA cells and that Rac1 inhibitors blocked the Wnt5a-induced DA neuron differentiation of ventral midbrain (VM) precursor cultures, linking Wnt5a-induced differentiation with a known effector of Wnt/PCP signaling. In vivo, Wnt5a was expressed throughout the VM at embryonic day (E)9.5, and was restricted to the VM floor and basal plate by E11.5–E13.5. Analysis of Wnt5a−/− mice revealed a transient increase in progenitor proliferation at E11.5, and a precociously induced NR4A2+ (Nurr1) precursor pool at E12.5. The excess NR4A2+ precursors remained undifferentiated until E14.5, when a transient 25% increase in DA neurons was detected. Wnt5a−/− mice also displayed a defect in (mid)brain morphogenesis, including an impairment in midbrain elongation and a rounded ventricular cavity. Interestingly, these alterations affected mostly cells in the DA lineage. The ventral Sonic hedgehog-expressing domain was broadened and flattened, a typical CE phenotype, and the domains occupied by Ngn2+ DA progenitors, NR4A2+ DA precursors and TH+ DA neurons were rostrocaudally reduced and laterally expanded. In summary, we hereby describe a Wnt5a regulation of Wnt/PCP signaling in the DA lineage and provide evidence for multiple functions of Wnt5a in the VM in vivo, including the regulation of VM morphogenesis, DA progenitor cell division, and differentiation of NR4A2+ DA precursors.


Introduction
Wnts comprise a family of 19 lipid-modified secreted glycoproteins that signal via different pathways and regulate multiple aspects of development [1,2]. These pathways include the canonical Wnt/b-catenin, noncanonical Wnt/Ca 2+ and noncanonical Wnt/planar cell polarity (PCP) pathways.
Mutations in PCP genes produce specific and distinctive phenotypes. These include the general convergent extension (CE) defects seen in the overall shortened and broadened morphology of mutants and in Keller explants of the Xenopus dorsal marginal zone [23,24], the Drosophila wing with misdirected bristles and disorganization of the compound eye [2,25], the murine cochlea with misdirected hair cells [4,26], and the murine neural tube with a broadened Shh-expressing floor plate (FP) and neural tube closure defects [9,27,28]. PCP is defined as the organization of cells within a single layered sheet of cells. CE however, is a morphological process regulated by PCP signaling and involving the coordinated movement of cells within a 3dimensional structure, leading to an overall elongation and narrowing of the structure [24]. Moreover, a role for PCP genes in the development of region-specific neuronal cell types outside of an epithelial plane is becoming increasingly apparent. For instance, PCP genes Celsr3 and Frizzled3 are involved in axon growth and guidance, while Celsr1 and Dvl1 also regulate dendritic arborization (for review see [29]).
In zebrafish, the Wnt5 and Wnt11 mutants, pipetail and silberblick respectively, display CE defects but their roles have been suggested to be more permissive than instructive [30,31]. However, Wnt5a has been found to be required for stereocilia orientation in the cochlea, indicating that Wnt5a can play an instructive role in PCP during development [4]. We hypothesized that Wnt5a may be able to activate PCP signaling and contribute to patterning and subsequent neural differentiation in specific brain regions.
We have previously reported that Wnt5a promotes DA differentiation of NR4A2 precursors in primary mesencephalic cultures [32]. However, the function of Wnt5a in the developing VM in vivo is unknown. A9-A10 dopaminergic neurons of the substantia nigra (SN) and ventral tegmental area (VTA) respectively, are born between E10.5 and E13.5. We therefore set out to investigate whether deletion of Wnt5a results in a Wnt/PCP phenotype and/or deficits in NR4A2 precursor differentiation in vivo at these and later stages up to E18.5. We provide evidence that Wnt5a is required for adequate morphogenesis of the midbrain by controlling the proper polarity and proliferation of VM progenitors, and for the differentiation of postmitotic NR4A2+ precursors into DA neurons.

Results
Wnt5a is expressed in a restricted temporal and spatial pattern during midbrain development We have previously reported that Wnt5a is highly expressed in the VM at the time of birth of DA neurons [32]. To examine the spatial and temporal expression of Wnt5a more closely we performed in situ hybridization on E9.5-E13.5, E18.5 and P56 CD1 mice. Interestingly, in situ hybridization showed a gradient of Wnt5a expression, with higher expression of Wnt5a at E9.5 and 10.5 in rostral levels of VM as compared to caudal levels ( Figure 1A). From E11.5 to E13.5, the expression of Wnt5a became progressively restricted to the FP and basal plate (BP). Until E11.5 Wnt5a was only expressed in the ventricular zone (VZ)/ subventricular zone (SVZ), but from E12.5, and mainly in caudal levels, Wnt5a was found in the intermediate and marginal zones, extending laterally ( Figure 1A). We next examined the expression of Wnt5a in relation to tyrosine hydroxylase (TH -the rate-limiting enzyme in the synthesis of dopamine and a marker of DA neurons) expression. Double immunohistochemistry for TH and NR4A2 (a marker of DA neurons and precursors) on sections probed for Wnt5a revealed extensive overlap of expression, particularly in the caudal midbrain at E12.5 ( Figure 1B and data not shown for other stages). From E13.5 on, the expression of Wnt5a in the ventricular zone (VZ) was reduced, and at E18.5, it became mostly restricted to DA neurons of the ventral tegmental area (VTA) ( Figure 1C). In the postnatal VM, expression of Wnt5a was not detected in substantia nigra compacta (SNc) neurons, was faintly detectable in VTA neurons, and was mainly confined to the red nucleus ( Figure 1C).

Wnt5a activates Rac1, and inhibition of Rac1 blocks Wnt5a-induced DA neurogenesis
We have previously shown that Wnt5a promotes the differentiation of dopaminergic (DA) neurons in primary midbrain cultures [32] and that Wnt5a signals via Dishevelled and Casein Kinase 1 in a dopaminergic cell line [17,33]. Wnt5a is known to activate the PCP pathway and to signal via small GTPases in different systems [34], but it is unknown which of the small GTPases transduces the Wnt5a signal in cells of the DA lineage. We therefore first investigated whether Wnt5a could activate Rac1, RhoA or Cdc42 in a DA cell line, SN4741, a validated model for studying Wnt signaling [33,[35][36][37]. We found that treatment with recombinant mouse Wnt5a induced the activation of Rac1 (Figure 2A), while RhoA and cdc42 activity were unchanged. In order to verify whether Rac1 mediates the pro-differentiation effects of Wnt5a, we used an in vitro assay in which Wnt5a induces the differentiation of primary DA precursors into DA neurons [32]. A dose-response curve for NSC 23766, a Rac inhibitor, showed that 10 mM had no effect on TH+ cell number. At higher doses, from 50 mM and up, cell death was seen (data not shown). Interestingly, treatment of these cultures with 10 mM of the Rac1 inhibitor NSC 23766 blocked the increase in the number of TH+ neurons otherwise induced by Wnt5a after 3 days in vitro (Figure 2 B, C), suggesting that the pro-differentiation effects of Wnt5a are mediated by Rac1. Note that the morphology of the TH+ cells is unchanged, and that the cells appear healthy in all conditions ( Figure 2C).
In the next part of our study we analyzed the VM phenotype of Wnt5a2/2 mice generated previously [38]. We have previously reported that Wnt5a does not activate canonical signaling in a dopaminergic neuron cell line [33]. Data in the literature suggests that Wnt/b-catenin signaling could be decreased in Wnt5a2/2 mice, since Wnt5a has been reported as capable of activating Wnt/b-catenin signaling [3], or increased since non-canonical signaling has also been reported to inhibit Wnt/b-catenin signaling [39]. To address possible regulation of canonical signaling in the Wnt5a2/2 VM, we first examined the expression of Wnt1, a Wnt expressed in the VM that activates the Wnt/bcatenin pathway and is required for VM DA neuron development [40]. In situ hybridization did not show any increase in signal, but rather a wider spacing of the two ventral stripes of Wnt1 expression in the FP ( Figure S1A). This was confirmed by QPCR analysis of E12.5 WT and Wnt5a2/2 VMs tissue ( Figure S1B), a finding that argues against a significant imbalance between Wnt1 and Wnt5a. To examine canonical signaling at the effector level, we investigated the levels of active b-catenin by Western blot in the VM of E10.5 and E12.5 Wnt5a2/2 mice, where no difference was detected ( Figure S1C and data not shown). In sum, Wnt5a activated the small GTPase Rac1 and promoted DA differentiation via Rac1, but loss of Wnt5a had no effect on Wnt1 expression or canonical Wnt-signaling via b-catenin, suggesting that Wnt5a may regulate Wnt/PCP signaling in the ventral midbrain in vivo.

Differentiation of dopaminergic neurons is altered in
Wnt5a2/2 mice To assess the role of Wnt5a in vivo, we examined the A9-A10 DA neuron populations in Wnt5a knockout mice at E11.5, E12.5, E14.5, E17.5 and E18. 5. Surprisingly, at E11.5 and E12.5, the number of TH+ DA neurons in the Wnt5a2/2 mice was not statistically different from wild-type (WT) littermate controls ( Figure 3A, D). This was confirmed by quantitative PCR (QPCR) for TH and for Pitx3 ( Figure S 2A,B), a homeobox transcription factor required for DA neuron survival [41][42][43]. A transient 25% increase in the number of Th+ DA neurons was detected at E14.5  Figure 3C). Although the number of DA neurons normal at all ages except E14.5, the morphology of the midbrain and distribution of the DA neurons was altered in Wnt5a2/2 mice from E10.5, note the broader lateral and dorsoventral distribution of TH+ cells at E12.5, E14.5 and E17.5 ( Figure 4A, B). We also examined the number of cleaved caspase 3+ cells (a marker of cells in apoptosis), but found no difference at E14.5 or E17.5 (data not shown).
Increased proliferation and DA precursor pool in Wnt5a2/2 mice In vitro, Wnt5a promotes DA precursor differentiation, reducing the number of proliferating (in S-phase) bromodeoxyuridine positive (BrdU+) cells [32]. To assess the role of Wnt5a in proliferation in vivo and the reason for the transient 25% increase in Th+ cells in the Wnt5a2/2 embryos, the density of BrdU+ cells/area in the FP and the alar plate (AP) of E11.5 Wnt5a2/2 and WT embryos were compared (black squares in Figure 4A). While the density of BrdU+ cells in the AP of WT and Wnt5a2/2 embryos was not different (after a 2 hour pulse), a 35% increase in BrdU+ cells/area was detected in the FP of Wnt5a2/2 mice ( Figure 4A, B). Furthermore, BrdU+ cells accumulated in the ventral midline of the Wnt5a2/2 midbrain at E11.5 ( Figure 4A). The increase of proliferating cells within the FP was first evident at E11.5. This was confirmed by a 45% increase in Ki67+ cells (a marker of cells in the cell cycle) detected in Wnt5a2/2 mice ( Figure 4A, B). Similarly, the number of cells that are dividing (mitotic cells, arrowheads, Figure 4A) increased by 25% in Wnt5a2/2 mice at E11.5 ( Figure 4A, C). Interestingly, the increase of cells in S-phase and M-phase was maintained along the entire antero-posterior axis of the VM.
Neurogenesis in the VM is controlled by a code of homeodomain and proneural bHLH transcriptional regulators [44]. Expression of the members of the bHLH superfamily Ngn1, Ngn2 and Mash1 can be subdivided into three distinct domains within the VZ of the mouse VM [45]. Ngn2 is essential for proper DA neurogenesis, and Mash1 is required for the generation of VM GABAergic neurons [45][46][47]. The expression of Ngn2 (and the low levels of Mash1) in the midbrain FP therefore define the DA progenitor domain, whereas the high expression levels of Ngn2 and Ngn1 in the adjacent basal plate (BP) demarcates the oculomotor (OM) and red nucleus (RN) progenitor domains [45,46,48]. In the Wnt5a2/2 embryos, the Ngn1-positive domain within the BP and the Ngn2-positive domain comprising the FP and BP were broadened in the mutant midbrain at E11.5 ( Figure 5A), whereas the expression domains of these two proneural factors remained unchanged in the dorsal midbrain. We next examined whether the increase in cell division and broadened neurogenic domain had generated more postmitotic cells, and found a 47% increase in the number of NR4A2+ postmitotic cells compared to control at E12.5, suggesting that the neurogenic process is indeed enhanced in the Wnt5a2/2 mice ( Figure 5B, C). Furthermore, the 47% increase in NR4A2+ cells is consistent with a 45% increase in proliferating cells. However, a 25% excess of TH+ at E14.5 indicates that not all of the excess NR4A2+ cells differentiate into Th+ neurons at this point. Moreover, we found that the domain occupied by NR4A2+ cells was wider in Wnt5a2/2 than in wild type mice ( Figure 5B). An increase in the amount of NR4A2 mRNA was also detected at E12.5 by QPCR ( Figure S2C).
We have previously reported that one of the functions of Wnt5a is to promote the differentiation of NR4A2+ DA precursors into DA neurons in vitro. This has been shown in diverse in vitro preparations, including rat E14.5 VM primary cultures [32], VM neurospheres derived from E10.5 mice or E12.5 rat [49] and mouse ES cells (unpublished observation). In order to investigate whether Wnt5a also promotes this differentiation step in vivo, we examined the proportion of TH+ cells out of NR4A2+ cells in the VM of Wnt5a2/2 mice at E12.5. In agreement with our previous in vitro results, we found that the proportion of NR4A2+ cells differentiating into Th+ cells was reduced by 23% at this stage ( Figure 5C). However since the number of TH+ cells increased at E14.5, and subsequently normalized, our results suggest that differentiation is rescued by an alternative mechanism. We thus conclude that Wnt5a is only partially and transiently required for the differentiation of NR4A2+ precursors into DA neurons in vivo.  Mediolateral/PCP and apicobasal morphogenetic defects in the VM of Wnt5a2/2 mice Several of the defects described in previous sections are reminiscent of PCP developmental defects, including the accumulation of BrdU+ cells in the midline, the lateral expansion of the Ki67+, Ngn2+, NR4A2+ and TH+ domains in the VM as well as a flattening of the midbrain ventricle (Figures 3, 4, 5). Previous reports of animals lacking PCP components have shown that the invagination of the VZ in the ventral midline is flattened and that the Shh domain, that defines the FP and BP in the midbrain, is broadened [50][51][52]. In the Wnt5a2/2 mice, a lateral expansion of the Shh and Foxa2 expression domains was first detected at E11.5 ( Figure 6A, Figure S2D,E). This was associated with a lateral expansion of the Lmx1a+ DA progenitor domain ( Figure 6A). Expression of the Shh-target genes (Ptch1 and Gli1) was not expanded in the midbrain AP ( Figure S3). The expression of class I (Dbx1) and class II (Nkx2-2, Nkx6-1) genes was not changed in the midbrain of the Wnt5a2/2 embryos compared to WT although the aberrant morphology of the Wnt5a2/2 midbrain noted previously led to a wider separation of the Nkx6-1-positive domains in the BP of the mutant midbrain ( Figure S3 and data not shown). To characterize this defect in more detail, the analysis of the angle formed between the invagination of the ventricular epithelia and the midline in the VM revealed that this angle appeared greater in Wnt5a2/2 mice (59.4u62.8u), than in WT mice (49.1u64.3u) ( Figure 6B). This resulted in the midbrain ventricle adopting a ''U''-shape in the Wnt5a2/2 mice instead of the typical ''V''shape in controls ( Figure 6A, B). A similar phenotype, including a broadened Shh domain, has been observed in the neural tube of the PCP mutants Scribble and Vangl2 [50][51][52].
It has previously been shown that cadherins regulate intercellular adhesion in neural progenitors and that a disruption of the complex formed with aand b-catenin leads to increased Shh signaling and proliferation [53]. We therefore examined the levels of N-cadherin in Wnt5a2/2 mice by Western blot and found a reduction at E9.5, prior to the expansion of the FP ( Figure 6C).
Moreover, since apico-basal polarity depends on PCP and adhesion, we examined apico-basal polarity of cells in the ventricular zone. Interestingly, the change in the general morphology of the VM neuroepithelia was accompanied by an alteration in the orientation of the cells, as shown by the nonuniform orientation of propidium iodide stained nuclei in the Wnt5a2/2 mice, compared to WT at E12.5 ( Figure 7A). To assess this quantitatively, the angle formed by the longest axis of each nuclei with the ventral midline was measured for 10 cells (starting from the midline and counting laterally) at different anteroposterior levels of the VM. This angle changed from very acute to less acute as the cells were positioned further away from the midline in the wild type. This angle reached almost 50u in wild type, but did not exceed 30u degrees in Wnt5a2/2 mice ( Figure 7B). These results reflect the fact that, in Wnt5a2/2 mice, the nuclei of apical FP cells are oriented ventrally, while the nuclei of control mice are oriented more ventro-laterally ( Figure 7A, B). Furthermore, when the frequency of nuclei aberrantly oriented towards the contralateral side was examined, Wnt5a2/2 mice showed a 7-fold increase compared to WT ( Figure 7A, C). Note that while cells in the midline of both WT and Wnt5a2/2 mice exhibited nuclei oriented contralaterally, only Wnt5a2/2 mice showed nuclei with contralateral orientation in lateral positions of the FP ( Figure 7A).
Several neuronal populations in the VM are redistributed following the PCP defect A more detailed analysis of the distribution of A9-A10 DA neurons, in the very same animals that did not show any change in total DA cell number at E12.5, revealed changes in the mediolateral, rostro-caudal, and dorsoventral axis ( Figure 8A). Sections at regular intervals throughout the A9-A10 nucleus were examined. The three levels analyzed in WT and Wnt5a2/2 mice (rostral, intermediate and caudal) are shown at E12.5 in ( Figure  S4). In the mediolateral axis, TH+ cells extend more laterally in the Wnt5a2/2 mice than in WT mice, especially in the rostral portion of the A9-A10 nuclei ( Figure 8A-D). This phenotype persisted until E17.5, and followed the earlier morphological defects in the distribution of progenitor (Shh, Foxa2, Lmx1a, Ngn2) and precursor (NR4A2) markers, all of which were expanded already at E11.5.
In the rostro-caudal axis, the distribution of TH+ cells was unchanged at E11.5 ( Figure 8E). However, at E12.5 and E14.5, the A9-A10 nuclei were shortened in Wnt5a2/2 mice, seemingly at the expense of the most anterior portion of the nuclei (Figure 8F, G). Finally, by E17.5 the distribution of VM DA neurons in the anteroposterior axis tended to normalize but remained broad in the dorsoventral axis ( Figure 3B). This shortening of the anterior neural tube at E12.5 was also apparent when the length of the En1 expression domain was measured in sagittal sections through the VM of Wnt5a2/2 mice, and compared to WT (390 mm compared to 540 mm, Figure S5).
To distinguish between a deregulation of TH and a true misplacement of the DA neurons, we examined the spatio-temporal expression pattern of other DA neuron markers such as Pitx3 and Slc6a3/Dat. Both marker genes were expressed within the marginal zone (MZ) of the Wnt5a2/2 ventral midbrain at E12.5 ( Figure 8H). A mediolateral expansion of the Th, Pitx3 and Slc6a3/DAT expression domains was detected in the Wnt5a2/2 mice, confirming the redistribution of DA neurons within the VM ( Figure 8H).
Since Wnt5a regulates PCP in several structures and the Shh, Foxa2 and Ngn1 domains were also broadened in the BP, we examined whether deletion of Wnt5a altered the distribution of other mature ventral neuronal populations in coronal midbrain sections at E18.5, when no difference in number of DA neurons was detected (data not shown). Interestingly, the area occupied by Th2, Islet12 or Brn3a2 expressing cells was increased in Wnt5a2/2 mutants (30-40%, Table 1). These results indicate that while the absolute numbers of midbrain neuron populations such as DA neurons are unchanged, the cells are re-distributed and positioned in a laterally and dorsoventrally enlarged domain in the Wnt5a2/2 VM (Table 1 and Figure 3).

Discussion
The impact of deletions of Wnt/PCP signaling components [2,29], including ligands such as Wnt5a [4], have been studied in very specific structures that have become standard Wnt/PCP functional assays. These include studying the orientation of bristles on the Drosophila wing, and in mammals the orientation of hair cells in the inner ear, convergent extension movements during embryo elongation, or neural tube closure. However, Wnt/PCP signaling components and ligands are expressed in very diverse tissues and at multiple developmental stages, where no previous PCP phenotypes have been described. In our study we investigated the function of Wnt5a in one such structure and developmental time, the VM during neurogenesis. We report that Wnt5a regulates VM morphogenesis, limits DA progenitor proliferation and enhances DA precursor differentiation.

Regulation of DA precursor differentiation by Wnt5a
Despite the clear effects of Wnt5a on DA differentiation in both gain and loss of function experiments in vitro [32,54], Wnt5a2/2 mice exhibited only a mild and transient DA differentiation phenotype in vivo. Moreover, in addition to a decrease in the proportion of NR4A2+ precursors differentiating into TH+ DA neurons (25% decrease in the Wnt5a mutants compared to WT), we found that NR4A2+ DA precursors were generated in excess by E12.5 (47% in the Wnt5a mutants compared to 39% in WT). These results suggest that both progenitor proliferation and the differentiation of NR4A2+ precursors were affected in the Wnt5a mutants. These alterations were transient and were not detected at E14.5, when the number of TH+ cells increased by 25%, then returned to control level at E17.5 and E18.5. Thus, our results show that Wnt5a is only transiently required for the differentiation of endogenous midbrain NR4A2+ precursors in vivo, and is sufficient for their differentiation in vitro, via Rac1 activation. Interestingly, the surprising alteration in the kinetics of DA neurogenesis in the Wnt5a2/2 mice suggests that other phenomena such as increased neurogenesis (increased number of DA postmitotic precursors) and a compensatory mechanism at E14.5, may prevent a stronger DA differentiation phenotype. Indeed, several non-canonical Wnts are expressed in the VM [37], and a functional redundancy between wnt5 and wnt11 has been described in double zebrafish pipetail (ppt)/wnt5 and silberblick (slb)/ wnt11 mutants [30], suggesting that another non-canonical Wnt may compensate for the loss of Wnt5a.

A role for Wnt5a in midbrain morphogenesis
The involvement of non-canonical Wnt-signaling in CE movements and neural tube morphogenesis has been widely documented [29,55,56]. Moreover, Wnt5a has recently been clearly implicated in the regulation of CE movements and PCP during cochlear development and neurulation in the mouse [4]. Interestingly, the mesencephalon undergoes unique morphogenic movements to form the cephalic flexure proper, a process that is coupled to CE movements in the ventral domain and outgrowth in the dorsal domain of the mesencephalon [57]. Moreover, Wnt5a is expressed exclusively in the VM and not dorsal midbrain throughout mouse embryonic development. Our analysis of the Wnt5a2/2 mice has revealed a large variety of subtle but clear alterations in the morphogenesis and the cytoarchitecture of the VM. Overall the VM was broader and shorter as demonstrated by: a lateral expansion of the Shh-expressing FP/BP at E11.5, accompanied by broader lateral expression of Foxa2, Lmx1a, NR4A2, Th, Pitx3 and DAT. Other markers examined were also more laterally placed such as Wnt1 and Nkx6.1. The shortened midbrain phenotype was manifested in the number of levels with TH+ cells and their distribution, and suggested by a reduced length of En1 expression in sagittal sections. In coronal sections, the mesencephalon did not acquire its distinct heartlike morphology and the ventricle in the ventral midline adopted a wide ''U'' shape instead of the typical ''V'' shape at E10.5. Thus, our findings suggest diminished CE and an involvement of Wnt/PCP signaling in the VM, as described for other structures [4,22,24,28,58]. Interestingly, we also found a change in the apico-basal orientation of midbrain VZ cells in Wnt5a2/2 mice compared to WT. The axis of individual WT FP apical ventricular cells and the midline formed an average 36u angle and did not cross the midline ventrally. However, cells in the Wnt5a2/2 mice formed an average 20u angle with the midline and in some instances individual cells showed a negative angle (i.e. their axis pointed contralaterally). A possible interpretation of these results is that the increase in proliferation allows nuclei to be oriented aberrantly in Wnt5a2/2 mice during mitosis. However, misorientation was never seen in WT mice in lateral positions, regardless of cell cycle phase, but was seen in Wnt5a2/2 mice. This could also reflect a consequence of the flattened VM morphology, allowing greater deviations towards the contralateral side. A more intriguing possibility is that the alteration in PCP directly regulates or causes secondary changes in the geometry of the cells and in apicobasal polarity. PCP and apicobasal polarity require cell attachment, a process that both pathways also regulate [59,60]. In line with this, we found a decrease in the levels of N-cadherin in the Wnt5a2/2 mice at E9.5, which could contribute to decreased attachment and altered polarity. These results suggest PCP and apico-basal polarity may be coordinately regulated in certain structures by adhesion proteins. Triple slb;ppt;wnt4-morphant zebrafish embryos phenocopy the neurulation defect of trilobite (tri)/Van Gogh-like 2 (Vangl2) PCP mutants [61]. Interestingly, the defects in the Wnt5a2/2 VM resemble the tri neurulation defects with an increase in proliferation, a broadening of the Shh domain/expansion of the FP, and an ectopic accumulation of neural progenitors within the midline [61]. These findings strongly suggest that Wnt5a also regulates cell intercalation after cell division and VM morphogenesis. Our analysis of the VM phenotype of Wnt5a2/2 mice suggests a model in which Wnt5a regulates morphogenesis by a mechanism involving cell adhesion and altered PCP signaling. This is accompanied by an alteration in apico-basal polarity, and a broader distribution of VM cell populations, but no alteration in the final number of neurons, all of which is compatible with a PCP phenotype.

Proliferation: a link between differentiation and morphogenesis
Our results indicate that the loss of Wnt5a directly or indirectly regulates cell division in VZ progenitors. Indeed, we found that deletion of Wnt5a increased the number of both BrdU+ and Ki67+ cells in the apical VZ, and the number of NR4A2+ postmitotic cells in the IZ, which suggested a regulation of both proliferative and neurogenic divisions. Interestingly, Wnt/PCP signaling has been previously reported to regulate oriented cell division during gastrulation, allowing axis elongation [61,62], and to regulate asymmetric cell division [63], which is the predominant mode of division used by VM progenitors during DA neurogenesis. Another mechanism that could also result in the regulation of proliferation and neurogenesis is cell adhesion. Our results indicate that Wnt5a may regulate cell adhesion in the neuroepithelia by regulating the levels of N-cadherin as early as E9.5. Interestingly, it has been described that a reduction in a-catenin binding to N-cadherin leads to decreased cell adhesion, increased Shh signaling and proliferation [53]. We therefore suggest that the increase in Shh signaling (first detected at E11.5 by the regulation of Foxa2 and Lmx1a), may be induced by a decrease in the levels of N-cadherin protein and cell adhesion at earlier stages, when no other phenotype is apparent. In agreement with a role for Shh in regulating proliferation, we found an increase in the number of proliferating apical VZ progenitors and a lateral expansion of the Shh, Foxa2 and Lmx1a domains in the Wnt5a2/2 mice. Thus, taken together, these data suggest that Wnt5a can affect both the extent and the type of cell division in VZ progenitor cells by different mechanisms.
In sum, our results show that CE defects and lowered cadherin expression were the first detected phenotype at E9.5 in the Wnt5a2/2 mice. This was followed by an alteration in morphogenesis and DA differentiation, and at E12.5 by defects in cell polarization and adhesion, a shortening and a broadening of the A9-A10 nucleus and other ventral domains of the mesencephalon. We suggest that the broadening of the Shh domain may lead to an increase in Shh-signaling, increased proliferation in the FP/BP, accumulation of NR4A2+ precursors and a transient expansion of the Th+ cell population at E14.5. These defects were most pronounced in the FP, thus affecting the morphology and development of cells in the DA lineage in the Wnt5a2/2 mice.
In conclusion, we show for the first time that, in the VM, Wnt5a regulates CE movements required for axial elongation, polar growth and morphogenesis as well as proliferation, neurogenesis, some aspects of differentiation, and the actual positioning of neurons in the VM.

Animals, Immunohistochemistry and in situ Hybridization (ISH)
Wnt5a tm1Amc (referred to as Wnt5a2/2 in this article) transgenic mice [38] and CD1 mice (Charles River) were housed, bred and treated in accordance with the approval of the local ethics committee (Stockholms Norra Djurförsöketiska Nä mnd). Wnt5a+/ 2 mice were kept on a C57BL/6 background. In all experiments, the Wnt5a2/2 embryos were compared to their wild-type (Wnt5a+/+ and Wnt5a+/2, heterozygotes have wild-type phenotype) littermates, n$4 for each genotype, if not otherwise stated in the text. Mice of the relevant genotype were mated overnight and noon of day of plug was taken as E0.5. Embryos were dissected in ice-cold PBS, fixed in 4%PFA 4 hrs to overnight, cryoprotected in 20% sucrose and frozen in OCT compound on dry ice. Serial coronal 7, 14 or 16 mm sections of the brain were obtained on a cryostat. Immunohistochemistry (IHC) and in situ hybridization (ISH) were carried out as previously described [32,64]. IHC and ISH for Wnt5a were visualized with a Zeiss HBO100 microscope; images were collected with a C4742-95 Hamamatsu camera and processed with OpenLab software and/or ImageJ.

BrdU treatments
Pregnant females were injected intra-peritoneally with 5-bromo-2-deoxyuridine (BrdU, 10 mg/g body weight) two hours before they were sacrificed and processed as described [71]. The number of BrdU-positive cells in a square area (2500 mm2) was counted on at least four serial sections from the midbrain of stage-matched littermates (n = 4) using the Neurolucida 6 software (MBF Bioscience, Williston, VT/USA).

TH+ Cell Counts and Distribution
For counts of total number or distribution of TH+ cells in the ventral midbrain, alternate 14 mm sections from coronal series of the entire A9-A10 population were counted and plotted versus position at E11.5 and E12.5 for distribution analysis. At E14.5 and E17.5, every sixth section was counted and plotted in the distribution analysis. The lateral spread of TH+ cells was measured in coronal sections through the VM, in three consecutive levels, where level 1 corresponds to the rostral midbrain-P1 boundary, level 2 corresponds to the intermediate midbrain and level 3 corresponds to the caudal midbrain (see Figure S 5). Pictures were taken with identical acquisition settings and magnification and the lateral spread was measured by drawing a horizontal line between the outermost TH+ cells. This distance was measured in pixels using ImageJ and then normalized to littermate controls.
The volumes of the Th2, Brn3a2 and Isl1-expression domains at E18.5 were measured by the Cavalieri method using the Stereo Investigator 5.05.4 software (MBF Bioscience) as described [65]. Cell numbers were averaged for each genotype and subjected to a Student's t-test for the estimation of statistical significance. Calculation of slope of ventricular invagination and analysis of nuclei orientation E10.5 and E12.5 mice were stained with Hoechst or propidium iodide respectively. The slope of the wall of the VM at E10.5 was measured by drawing a cross bisecting the FP, where the origin coincided with the ventricular invagination. A tangent to the ventricular wall was drawn and the angle between the tangent and the midline was measured in 4 embryos of each genotype. To measure the nucleus orientation, confocal pictures of serial sections of E12.5 midbrains were taken with an optical slice of 1 mm. Ten nuclei along the ventricle were analyzed per image from the midline outwards, 3 images per mouse (rostro-caudally distributed), and 3 mice per genotype were analyzed. The nucleus was outlined in Image J and the angle of the longest axis was measured compared to the vertical parallel to the midline. This gave an angle that was plotted against its position from the midline (nucleus number position). Certain nuclei were abnormally oriented towards the contralateral side, these gave a negative angle. The number of nuclei with abnormal orientation (negative angle), was compared to the total number of cells and this ratio was used for analysis of contralateral orientation.

Primary ventral mesencephalic cultures and immunocytochemistry
Ventral midbrains of E11.5 CD1 mice were dissected out in icecold PBS supplemented with 0.2%glucose, mechanically dissociated in serum-free N2 through flame-narrowed Pasteur pipettes and plated at a final density of 125,000 cells/well in poly-D-lysinecoated 48-well plates as described previously [32]. Cultures were treated with recombinant mouse Wnt5a (200 ng/ml, RnD), 0.05% CHAPS (Wnt5a control, Sigma), and/or 10 mM Rac1 inhibitor (NSC 23766, Calbiochem). Treatment of cultures was initiated 1 hour after plating and cultures were incubated for 3 days in N2 at 37uC in 5%CO 2 . Cells were then fixed for 15 minutes with 4% paraformaldehyde, washed in PBS, and used for immunocytochemical analysis.
Real-time Quantitative PCR cDNA was generated as described previously [32]. In brief, RNA from E8.5-P0 VMs of CD1 (for developmental stages) or wildtype and Wnt5a2/2 mice was extracted using RNeasy Mini Kit (Qiagen). 1 mg of RNA was reverse-transcribed using SuperScript II Reverse Transcriptase (Invitrogen) and random primers (Invitrogen). Quantitative PCR and the primers used in this study have been described previously [32] with the exception of NR4A2 primers; Forward: 59CAGCTCCGATTTCTTA ACTCCAG39 and Reverse: 59GGTGAGGTCCATGC-TAAACTTGA39.

Statistics
Data were analysed using Graphpad Prism version 4.01 for Windows, GraphPad Software, San Diego California USA, www. graphpad.com. For analyses comparing WT and Wnt5a2/2 mice, t-tests were used. A paired t-test was used if 1WT/Wnt5a littermate pair was used from individual litters. If several pairs were taken from the same litter then an unpaired t-test was used. At least 3 litters were used for each analysis. Each graph value is the mean6SEM. For analyses of distribution of Rac inhibition of dopaminergic neuron differentiation and for dopaminergic neuron distribution in vivo, One-way and Two-way ANOVA were used, with Bonferroni's post-hoc test.