For staging, the day when the vaginal plug was detected was considered as embryonic day 0.5 (E0.5). All mouse manipulation and experimental procedures were performed according to the directives of the Spanish and European Union governments and the protocols were approved by the Universidad Miguel Hernández OIR Committee (2016/VSC/PEA/00190). The Wnt1^–/– transgenic mouse line generation and genotype were previously described in McMahon et al. (1992).

Pregnant mice were intraperitoneally injected 5 μl/gr of body weight with Bromodeoxyuridine (BrdU; 5 mg/ml) at the desired gestational stage (E11.5-E13.5) and sacrificed 2 h later. Embryos were fixed overnight in 4% Paraformaldehide (PFA) and washed in Phosphate Buffered Saline (PBS).

Brains were fixed 1 h in 4% PFA and embedded in 4% agarose in PBS. Afterward, brains were sectioned in coronal plane with a vibratome until reaching the habenular nucleus. At this point, the fluorescent crystals were placed in both habenulae: DiI (1,1′-dioctadecil 3,3,3′,3′-tetra-metilindocarbo-cianina perchlorate; Molecular Probes) on one side and DiD (1′-dioctadecyl-3,3,3′,3′- tetramethylindodicarbocyanine, 4-chlorobenzenesulfonate salt; Molecular Probes) on the contralateral side. Samples were left at 37°C in 4% PFA until crystal had diffused along the retroflexus tract. Labeled brains were sectioned in the vibratome and IHC processed without detergent to avoid the signal loss.

Mouse embryo brains were fixed overnight in PFA in PBS. Samples were agarose-embedded and sectioned at 100 μm in coronal or sagittal planes by vibratome or wax embedded and sectioned at 10 μm by microtome.

For IHC, it was performed as previously described (Murcia-Ramón et al., 2020). The primary antibodies used were: αTH (1:1,000; 208020234/Inst. J. BOY), αNTN-1 (1:500; MAB1109/RD Systems), αDCC (1:100; sc-6535/Santa Cruz), αCNTN2 (1:500; AF4439/RD Systems), αNFEM (1:500; AB1987/Chemicon), αBrdU (1:200; M0744/Dako), αCALB (1:1,000; CB-38/Swant), αROBO3 (1:300; AF3076/RD systems), αSOX2 (1:500; ab97959/Abcam) and NFEM (1:1,000; Ab7794/Abcam).

For ISH, embryonic brains were washed three times for 20 min with detergent mix (1% IGEPAL, 1% SDS, 0.5% Deoxycholate, 50 mM Tris pH 8, 1 mM EDTA, 150 mM NaCl). Afterward, brains were post fixed in 4% PFA and rinsed in PBS. The next step was the pre-hybridization with hybridization buffer (50% Deionized Formamide, 5x Salt sodium citrate (SSC) pH 5.3, 50 mg/ml Heparin, 0.1% Tween 20). Hybridization chamber was prepared with the pre-hybridization mix (50% formamide, 5x SSC, 50 μg/ml heparin, 50 μg/ml tRNA, 50 μg/ml ssDNA, 0.1% Tween-20) for 1 h at 65°C. Then, the Digoxigenin-labeled RNA probe (Fgf8, Addgene plasmid #22090) was denaturalized at 80°C and the tissue was incubated in hybridization buffer with the probe overnight at 65°C. Samples were washed 4 times for 30 min at 65°C with 2x SSC pH 5.3, 50% formamide, 1% SDS. Three additional washes of 5 min and one of 30 min were carried out with 1x MABT (MAB solution (500 mM Maleic acid, 750 mM NaCl, 0.95M NaOH pH 7.5) and 0.1% Tween-20). Next step was to incubate embryos with 2% RBR (Roche Blocking Reagent) and MABT for 1 h. Then, brains were blocked with 2% RBR, and 20% Sheep Serum in MABT for 1 h. Then, samples were incubated in a wet chamber overnight at 4°C with an alkaline phosphatase-coupled anti-digoxigenin antibody (11207733910/Roche) prepared in the blocking solution at 1:2,000. The following day, samples were extensively washed with MABT. Then, before the color reaction, samples were washed with NTMT (100 mM NaCl, 100 mM Tris-HCl pH 9.5, 50 mM MgCl₂, 1% Tween20). The NBT/BCIP (Boehringer, Mannheim) was used as a chromogenic substrate to detect the probes. The alkaline phosphatase reacts with this substrate producing a solid precipitate.

E18.5 fixed brains were boiled in 0.01M sodium citrate at 80°C for 8 min and dehydrated with 90 min methanol washes at increasing concentration (20, 40, 60, 80, 100%). Brains were left overnight in 100%methanol at room temperature (RT). Next day, brains were incubated overnight at 4°C in a solution consisting of 1/3 100% methanol and 2/3 DCM (dichloromethane). The following day, brains were washed twice with methanol 100% for 1 h. At this point, brains were incubated in a solution of 5% hydrogen peroxide prepared in 100% methanol overnight at 4°C. The next day, brains were rehydrated in descending methanol concentrations. After that, brains were washed twice for 1 h at RT with PTx2 (10x PBS, 2% TritonX-100) and were incubated overnight at 37°C with permeabilization solution (0.3 M glycine, 20% DMSO in PTx.2). After this time, brains were incubated overnight with blocking solution (6% donkey serum, 10% DMSO in PTx.2) at 37°C. Then, samples were incubated for 10 days at 37°C with the primary antibody solution: αTH (1:1,000; N° 268020234/Institute Jacques Boy Cat), αROBO3 (1:300; AF3076/RD systems). Brains were extensively washed with PTwH (0.2% Tween 20, 0.01% heparin in 1x PBS). Afterward, samples were incubated overnight at 37°C with the secondary antibody solution prepared at 1:500 (5% DMSO and 3% Donkey serum in PTwH). This solution was previously filtered with a 0.2 μm filter. The following day they were washed with PTwH and dehydrated with methanol. Once dehydrated, brains were incubated with a solution of 2/3 DCM and 1/3 100% methanol. Then, samples were immersed in DCM twice for 30 min and were stored in BDE (Benzyl ether) until being processed. In all the steps, the brains are kept in movement to improve the processes.

Sections were immersed in Cresyl Violet for 3–4 min and were submerged twice in distilled water for 2–3 min. Then, samples were washed with ethanol at increasing concentration (70, 96, 100, and 100%). Finally, samples were immersed in xylol twice and covered with Eukitt Clasic Mounting Medium.

Fluorescent images were acquired with confocal microscope Leica SPE II whereas bright field images were acquired with a camera (Leica DFC500) associated with a stereomicroscope (Leica MZ16FA). 3D imaging was performed with a light-sheet fluorescence microscope (Ultramicroscope II, LaVision BioTec) and processed with Imaris software X64 9.3.0. Figures were composed with Adobe System software (CS6).

The proliferation assay was quantified by the area occupied by the red fluorophore using the measure tool of the Fiji software (ImageJ 1.53C). To do this, the image was binarized and the percentage of area occupied by white points was measured. The habenular volumes were measured with the surface tool of the Imaris software.

Statistical analysis was performed with unpaired t-test using GraphPad Prism 8.4.3 software.

In order to have a first overview of the possible phenotype displayed by our mouse transgenic model, we analyzed E18.5 Wnt1^–/– and wt brains in sagittal sections. In the wt, by cresyl violet staining, we observed the general histology of the brain. The Hb was located in the most dorsal aspect of the thalamic prosomere (Figure 1A). From its ventral tip, we detected the fr extending toward the diencephalic tegmentum before bending caudally toward its main hindbrain target, the Ip nuclei (Figure 1A; see Figure 1 in Moreno-Bravo et al., 2016).

With the aim to discern between the mHb and lHb axons, we studied the distribution of specific markers for each axonal projection. CNTN2 protein is specifically located in mHb axons (Figure 1B) and NFEM protein in lHb axons (Figure 1C). Note that NFEM distribution is slightly wider than CNTN2, this is due to the fact that mHb axons occupied the core of the fascicle meanwhile the lHb axons are located in the shell (compare Figures 1B,C). In Wnt1^–/– brain, the Hb territory appeared wider in its anteroposterior axis and the thalamic territory displayed a clear size reduction (Figure 1D). The fr showed a slightly abnormal appearance (Figure 1D). Core and shell fr specific markers allowed us to distinguish a wider distribution of the mHb axons (arrow in Figure 1E) as well as in the case of the lHb axons (arrow in Figure 1F). Other diencephalic tracts such as the posterior commissure and the mammillothalamic and mammillotegmental tracts did not display any apparent alteration (Figures 1E,F).

To confirm our first results, we analyzed ROBO3 distribution (mHb marker) by iDISCO in E18.5 brains with the aim to calculate the mHb volume in mutants compared to wt embryos. In a lateral view of the scanned brain, we were able to observe the full extension of the mHb as well as the initial fr portion (Figure 2A). In the mutant brain, the mHb displayed an abnormal and enlarged shape (Figure 2B). Also, the initial fr portion displayed a wider appearance (Figure 2B). The iDISCO scanned brains allowed us to quantitatively measure the volume occupied by both mHb. The Wnt1^–/– Hb volume was larger than the wt, being the difference among them statistically significant (Figure 2C; Unpaired t-test, p < 0.0001).

Once we demonstrated the enlargement of the habenular territory, we aimed to analyze the main subdivisions distribution in the Wnt1^–/– Hb compared to the wt. We grouped the different subnuclei in four divisions of the habenular complex. The mHb was formed by a medial division (mHbm) and a lateral division (mHbl) and the lHb was subdivided as well in a medial division (lHbm) and a lateral division (lHbl). Then, we analyzed several markers for the different territories. The NFEM protein distribution strongly labeled the sm in both samples (Figures 3A,B). It was absent from the mHb divisions meanwhile in the lHb divisions was weakly present, no obvious difference was found between the mutant and the wt (Figures 3A,B). CALB presented a complex pattern, it was strongly located in the medial aspect of the lHbm. Meanwhile in the mHbl presented a gradient distribution pattern from dorsal to ventral. Again, we did not find any obvious alterations between mutant and wt (Figures 3C,D). DCC, marker of the mHb, was homogeneously found in the mHb divisions and a weaker distribution in the lHb divisions. Again, no differences were found between samples (Figures 3E,F). CNTN2 protein, a mHb marker, strongly labeled the mHbm but also presented a scattered distribution in the mHbl. In the lHb divisions we observed some mHb axons fasciculating the initial fr. Once again, no significant difference was observed (Figures 3G,H). Finally, we analyzed SOX2 distribution pattern which specifically labeled the periventricular layer of the mHbm, and it specifically labeled in the ventrolateral subnuclei in the mHbl division. Once again, this labeling was observed in both samples (Figures 3I,J).

Summarizing, no defects were observed in the cellular organization of both mHb and lHb. Nevertheless, it must be pointed out that the mutant Hb displayed a general abnormal flattened shape when compared with the wt Hb.

With the aim to unveil the reason for the abnormal growth of the habenular territory, we did proliferation assays analysis. From E11.5 to E13.5 (time window for the habenula differentiation; Funato et al., 2000; Belle et al., 2014), we collected embryos 2 h after a BrdU pulse was injected intraperitoneally to the pregnant females on the given day. Thereafter, we compared the proliferation between wt and mutant embryos.

At E11.5, when we compared the wt (Figures 4A,B) with the Wnt1^–/– (Figures 4C,D), the proliferation was slightly higher in the mutant (Figure 4E). However, at E12.5, the proliferation in the wt (Figures 4F,G) in contrast to the Wnt1^–/– (Figures 4H,I) displayed the strongest difference (Figure 4J). Finally, at E13.5 the correlation between the control (Figures 4K,L) and the mutant (Figures 4M,N) also showed a significant variation (Figure 4O). Summarizing, the Wnt1^–/– mice displayed a higher proliferation rate in the habenular territory in all the analyzed periods showing the highest difference at E12.5.

Having described the Hb phenotype, and the affected mechanism that underlined it, we focus our attention on the fr trajectory, main efferent tract of the Hb complex. First, we compared in sagittal sections the mHb fr trajectory (by CNTN2 labeling) with the SNc localization (by TH labeling) at E15.5. In the wt, the mHb reached the basal thalamic territory in close contact with the SNc (Figure 5A). However, in the Wnt1^–/–, the SNc was strongly reduced and a thicker mHb fr mainly followed the normal trajectory (Figure 5B). Nevertheless, a bunch of fibers adopted abnormal directions (arrows in Figure 5B). In order to study the fasciculation and final trajectory of the full tract (mHb and lHb), we placed lipophilic dyes on E15.5 normal and mutant Hbs (DiI on the right Hb and DiD on the left Hb). In a diencephalic frontal section, we observed in the wt two compact fr, each one labeled with the specific dye (Figure 5C). In the Wnt1^–/–, the fascicles occupied a broader domain and a slight defasciculation was detected (arrow in Figure 5D). It is known that in the last caudal portion of the fr, the fibers crossed the mes-rombencephalic boundary and the mHb axons criss-crossed the floor plate to innervate its target, the IP nucleus. Thus, in the wt, once the fibers crossed the dopaminergic territory, we nicely observed the characteristic eight shaped pattern of the mHb axons innervating the Ip nucleus (Figure 5E). Meanwhile, we confirmed in the Wnt1^–/– mice that the dopaminergic neurons showed a dramatic reduction and an abnormal trajectory of the Hb fibers (Figure 5F). The Hb axons were able to go through the mid-hindbrain boundary, but once located in the rhombencephalon not all of them were able to cross the midline but only just one time (arrow in Figure 5F). Finally, we studied the distribution of mHb (labeled by CNTN2) and lHb axons (labeled by NFEM) in the fr. In the wt, the mHb axons occupied the core of the fascicle, whereas the lHb are distributed in the sheath (Figure 5G). In the Wnt1^–/–, the mHb axons were still located in the core of the fascicle but the lHb axons presented a less compacted distribution and we found some intermingled fibers (Figure 5H).

Taking the technical advantage of observing the specimen in a 3D way we analyzed the full trajectory of the fr with iDISCO protocol. We processed E18.5 mutant and wt brains for whole mount immunohistochemistry with ROBO3 for mHb and its projections and TH for the dopaminergic populations. In the wt, in a lateral 3D vision we observed the complete trajectory, first in its dorso-ventral path and in its final rostro-caudal portion and how it is strongly related with the mes-diencephalic dopaminergic neurons (Figures 6A,B and Supplementary Video 1). In a frontal view, we were able to show how the fibers reached the pial surface at both sides of the floor plate before they bend caudally and criss-cross the floor plate (Figures 6C,D and Supplementary Video 1). In the Wnt1^–/– mutant embryo, in a lateral view we confirmed the reduction in dopaminergic neurons and the thicker mHb fr that displayed an abnormal final trajectory (Figures 6E,F and Supplementary Video 2). Consequently, in a frontal view, we clearly detected the aberrant trajectory that the axons followed once they reached the isthmic territory and the abnormal floor plate cross of the fibers (Figures 6G,H and Supplementary Video 2).

The specification of the mesencephalic and rostral rhombencephalon is controlled by the isthmic organizer (IsO). This secondary organizer expresses a combination of morphogens that include Wnt1 and Fgf8, among others. The alteration of any of these genes produces a strong malfunction of this vital organizer. The fr misdirection in the Wnt1^–/– could be due to alterations in the specification of the territories that it must navigate. In order to proof this, we first studied the Fgf8 expression pattern. In the wt at E10.5, Fgf8 was expressed in the isthmic constriction and in the anterior neural ridge (Figure 7A). In the Wnt1^–/– mutant, the Fgf8 expression is strongly reduced but it was not absent (arrow in Figure 7B). Rostrally, the IsO induces the specification of the dopaminergic populations. As recently demonstrated (Company et al., 2021), the Ntn1 expression by these populations are necessary for the correct fr trajectory. In the wt, in a midbrain frontal section we observed the correlation between the dopaminergic populations and the NTN1 distribution (Figure 7C). In the Wnt1^–/– mutant, we confirmed the dramatic reduction of dopaminergic neurons and how this alteration produced a NTN1 deprivation in the territory (arrow in Figure 7D). Finally, we studied the Ip nucleus, final target of the mHb axons. We detected the Ip location by cresyl violet staining (arrow in Figure 7E) and the mHb axons terminals labeled by CNTN2 allowed us to confirm the Ip identification (arrow in Figure 7F). In the Wnt1^–/– mutant embryo, the Ip area did not show any apparent alteration (arrow in Figure 7G). However, the CNTN2 labeling confirmed us the Ip localization (arrow in Figure 7H). The Ip nucleus was indeed specified but its neuronal distribution was slightly abnormal.

Morphogenes are crucial for the proper development of neuronal populations at least in two aspects. On one hand, they specify the genetic cascades that direct the specific differentiation programs of the neuronal populations (Brafman and Willert, 2017). On the other hand, they regulate the proliferation rate of the neuroblast that will give rise to the different neurons (Cayuso and Martí, 2005).

Our results indicate that Wnt1 is not directly related with the specification of the habenular neurons as we were able to identify all the different subnuclei that compose this complex. This result is in contrast to the differentiation changes observed in the dorsal Hb (equivalent to mammal mHb) of Zebrafish in the lack of Wnt1 or the absence of ventral Hb (equivalent to mammal lHb) and mis-specification of the dorsal Hb in the complete blockage of the Wnt signaling in the wls mutant (Beretta et al., 2013; Hüsken and Carl, 2013; Kuan et al., 2015; Guglielmi et al., 2020). The strong differences in habenular organization and also the strong asymmetry found in Zebrafish may explain the phenotype variances found between these two vertebrate models. The alterations observed in the zebrafish wls mutant suggest that other Wnt members must be involved in the specification of the habenular territory. Nevertheless, the increased habenular volume in the mutant indicated that Wnt1 is involved in the control of the number of habenular neurons. The phenotype observed can be due to two different phenomena. The Wnt1 lack of function can induce a change in the differentiation program of the thalamic neuroblast and modify their destiny into habenular neurons (Martinez-Ferre et al., 2013). Another possibility is that this absence can modify the proliferation rate of the habenular progenitors and give rise by increment to a larger habenular complex (Cayuso and Martí, 2005).

Nevertheless, the fact that the mHb subpopulations did not change their size proportions among them prompted us to discard this first hypothesis. However, the strong increment in the proliferation rate of the habenular neuroblasts that we have shown supports the second hypothesis. The decrease in the thalamic territory could also be explained by a reduction in its proliferation rather than a change in their identity. Therefore, Wnt1 is needed to determine the proliferation rate of the habenular precursors. It must be highlighted that the strongest effect was found in the mHb. The fact that the highest proliferation increase in the mutant was found at E12.5, time window when the mHb is generated (Funato et al., 2000; Belle et al., 2014). It is also striking that “single cell” RNAseq experiments demonstrated that the mHb neurons express several Lef/Tcf downstream genes while only Tcf7l1, a repressor of Wnt1 target genes (Mao and Byers, 2011), is expressed in both mHb and lHb. Therefore, it is not surprising that the Wnt1 absence has a deeper effect in the mHb compared with the lHb (Hashikawa et al., 2020; Wallace et al., 2020).

The defects found in the fr trajectory in the Wnt1^–/– mouse display two different aspects. First, we detected a clear increment in the size of the tract and a slight disorganization of the fibers inside the fascicle. Obviously, the increment in size is due to the increased number of habenular neurons in the mutant. The lack of Wnt1 did not produce changes in the differentiation of the habenular subnuclei but the increment in size and surface extension may account for the slight disorganization that we have observed in the sheath of the mutant fascicle. Second, we found a clear misdirection in the final fr trajectory. This phenotype can also be due to alterations in the surface molecules of the fr axons that alter their response to the surrounding navigation signals or to alterations in the specification of the territories that the axons must cross and therefore of the navigating signals (Company et al., 2021). The fact that all the different Hb subnuclei are specified suggest that the Hb axons contain the correct combination of receptors. However, it is well known that Wnt1 lack of function produces severe alterations in the specification of the IsO (McMahon et al., 1992). As we have shown, the mutant IsO displayed a strong reduction of Fgf8 expression, its main morphogen. Increasing reductions of FGF8 amount in the IsO resulted in escalating severe phenotypes at both sides of the organizer (Garda et al., 2001; Chi et al., 2003; Basson et al., 2008). Thus, a malfunction of this organizer produces strong differentiation impairments in the midbrain and rostral hindbrain. The effect on the surrounding territory is not symmetrical, the midbrain is more sensible than the hindbrain (Basson et al., 2008). In fact, the SNc and the VTA are strongly reduced in our non-functional IsO, due to the Fgf8 reduction and Wnt1 loss (Prakash et al., 2006; Smidt and Burbach, 2007; dos Santos and Smidt, 2011), meanwhile the Ip generated in rhombomere 1 is almost not affected (Basson et al., 2008). Recently, it has been proved that SNc and VTA are involved, via Netrin1, in the correct navigation of the fr axons through this territory (Company et al., 2021). The absence of a Netrin1 correct signal must be responsible for the misdirection of a part of the fr axons. Some of them are still able to reach the Ip probably due to the fact that not all the dopaminergic neurons are lost (Wurst and Prakash, 2014). Thus, the Wnt1 lack of function altered the IsO functionality which produced a strong reduction in SNc and VTA neurons. The miss-specification of this intermediate target of the fr axons must be responsible for their disrupted trajectory in our mutant model.

In summary, we can conclude that Wnt1 is not directly related in the specification of the habenular neurons but it is responsible for the proliferation rate regulation of the habenular neuroblasts. Other morphogenes expressed in the diencephalic roof plate (like BMPs or other Wnt family members, must account for the determination of this neuronal population. This is supported by the results obtained with the wls mutant (protein involved in the WNT secretion) in zebrafish. The wls habenular phenotype is dramatic compared with the Wnt1 mutant indicating that other Wnt members must be involved in the Hb specification (Kuan et al., 2015). The Wnt1 lack of function produced and altered IsO, this alteration produced a miss differentiation of the territory crossed by the fr resulting in an erroneous direction of the fascicle.