The thalamic complex consists of the anteriorly located pre-thalamus and the posterior located thalamus 1 . The prosomeric model would describe these two areas as main dorsal components of the prosomere 3 (P3) and prosomere 2 (P2) respectively 2 . Between these two neural segments there is an intervening ventricular ridge - the zona limitans intrathalamica (ZLI). The anatomical ZLI border zone contains a small cell population, which releases signaling molecules. This signaling center orchestrates thalamus development by controlled release of the morphogen Sonic hedgehog and thus, we termed it the middiencephalic organizer (MDO, formerly known as the ZLI organizer; 3 ). Lack of the Shh-positive MDO leads to gross malformation of the caudal forebrain and loss of the entire thalamus. Local abrogation of Shh signaling in small cell clones blocks acquisition of thalamic neuronal cell fate in vertebrates 4 5 6 . Thus the MDO determines the size of the thalamic complex and orchestrates the neuronal development of the central relay station of the brain.

A further important diffusible, external cue during neural development is Wnt signaling. Patterning of the vertebrate anterior neural tube requires the function of this pathway at multiple stages 7 . Canonical Wnt signaling regulates anteroposterior patterning in the forebrain and midbrain, is required for development of the dorsal telencephalon 8 and the eyes 9 , and allows the establishment of the midbrain-hindbrain boundary (MHB) organizer 10 11 . Several lines of experimental evidence have demonstrated that other signaling pathways counteract Wnt signaling during neural development, independent from direct antagonists, such as sFRPs or Dkk1. Indeed, Shh and Wnt signaling are mutually antagonistic during some events in embryonic development, such as spinal cord patterning 12 . Despite the recognized importance of Wnt signaling for central nervous system (CNS) development, its functional relevance during diencephalon formation and how Wnt signaling and Shh signaling interact there remains unknown. Receptors, ligands and modifiers of the Wnt signaling pathway are expressed during early stages of caudal forebrain regionalization 13 14 . Recently, we showed that Wnt signaling is required for cell adhesion in the thalamus and thalamic neurogenesis 15 . However, the early function of Wnts during MDO establishment has to be determined.

Here we show that blockage of the canonical Wnt signaling pathway leads to malformation of the MDO. By a Morpholino-based knock-down approach we identified Wnt3 and Wnt3a as the responsible ligands and hence are required to maintain the primordium of the MDO. Lack of canonical Wnt signaling per se or knock-down of Wnt3/Wnt3a leads similarly to an increase of apoptosis specifically within the organizer primordium. Consistently, blockage of Tp53-mediated apoptosis is able to rescue the MDO. Furthermore, abrogation of the repressive factors Fezf2 and Irx1b leads to restoration of the organizer. In summary, we propose that canonical Wnt signaling triggered by Wnt3/Wnt3a is necessary to suppress Tp53-mediated apoptosis and thus maintain the organizer tissue during development.

Several members of the Wnt family are localized at or near the roof plate - the dorsal pole of the developing nervous system where they dorsalize the embryonic brain and spinal cord 16 . In the latter, canonical Wnt signaling is opposed by the ventralizing Shh signal 17 . A similar scenario has been described for the telencephalon in which wnt8b expression is restricted by the Shh-dependent transcription factor FoxG1 8 . To test whether canonical Wnt signaling is required for the formation of the Shh-positive MDO, we treated embryos with a specific Wnt signaling antagonist, the tankyrase inhibitor IWR1 18 . We used IWR1 to block Wnt signaling from 10 to 28 hpf, after initial neural plate patterning has taken place. Under these conditions, we observed a strong reduction of Shh in the central part of the MDO, whereas expression at the dorsal tip remained (Figure 1A, B). Initiating treatment at 12 hpf led to a mild decrease whereas treatment after 14 hpf did not affect shh expression in the MDO (Figure 1C, D), suggesting that canonical Wnt signaling is required for MDO induction in a narrow time window of four hours. We used a heat-shock inducible Dkk1-GFP construct as an additional tool to inhibit Wnt signaling 19 . After activation of Dkk1-GFP at 10 hpf, we found a lack of Shh::RFP expression within the MDO at 28 hpf (Figure 1E, F). Consistently, enhancement of Wnt signaling between 10 and 28 hpf using the GSK3ß inhibitor BIO 20 led to a broader expression domain of shh at the MDO (Figure 1G). In agreement, shh expression is expanded in the diencephalon of axin1^-/- mutant embryos 21 , independent of morphological alteration of the telencephalon (Figure 1H). Similarly, grafting Wnt3a loaded beads in the diencephalon at 10 hpf (Figure 1I) resulted in expansion of shh expression at the MDO, compared to embryos implanted with PBS-loaded control beads. Therefore, we hypothesize that canonical Wnt signaling is necessary for proper induction of the MDO between 10 h and 14 hpf.

Prime candidates to mediate local canonical Wnt signaling in the developing thalamus are Wnt3 and Wnt3a. At 15 hpf and 28 hpf, both of these Wnt ligands are expressed in the MDO, whereas wnt3a also marks the roof plate (Figure 2A, B). Wnt3 and Wnt3a are expressed in the mid-diencephalon prior induction of shh in the MDO at 15 hpf (Figure 2C, E). At 28 hpf, shh overlaps with wnt3 expression domain in the ventricular zone of the organizer (Figure 2F). However, expression of wnt3a becomes restricted to the dorsal MDO at 28 hpf and is subsequently localized subjacent to shh (Figure 2D). To address whether these two Wnt ligands mark the primordium of the MDO, we analyzed the expression pattern relative to markers in the thalamic complex. The pre-thalamic marker lhx5 22 abuts the wnt3a expression domain anteriorly (Figure 2G, H). Wnt3a overlaps with the MDO/thalamus marker otx2 (I, J; 23 ). The thalamic marker irx1b is expressed mainly posterior to the expression domain of wnt3a except of the most dorsal area (Figure 2K, L). Thus, wnt3 and wnt3a mark the primordium of the MDO prior to the induction of shh expression.

To elucidate the early function of Wnt3/Wnt3a during development of the thalamic complex, we blocked their function by performing a Morpholino antisense oligomere (MO)-based knock-down approach 24 . Injection of a splice-blocking wnt3 MO leads to a severe down-regulation of Wnt3 protein by Western blot analysis (Figure 3A). Brain morphology in embryos injected with either wnt3 MO or wnt3a MO is not visibly distinguishable from control MO injected embryos (Figure 3B-H). This suggests that Wnt3 and Wnt3a have redundant functions during brain development.

However, combining MOs targeting Wnt3 and Wnt3a (wnt3/wnt3a MO) led to a strong decrease of the expression of the pan-MDO marker shh in the central part of the MDO at 24 and 36 hpf (Figure 4A-D) These results are comparable to global Wnt signaling inhibition between 10 and 28 hpf. (Figure 1A, B). A combinatorial requirement of canonical Wnt signaling during MHB organizer formation has been previously reported 11 . Indeed, the MHB organizer marked by fgf8 expression is lacking in double morphant embryos (Figure 4A-D, asterisks) thus validating the specificity of the wnt3/wnt3a MO approach. In addition to the altered shh expression (also known as shh-a), we find that the expression of the paralogue shh-b is absent in the organizer in compound morphant embryos (Figure 3E, F; black arrow). Down-regulation of shh and shh-b is accompanied with a down-regulation of the Shh-dependent target gene nkx2.2, (here Nkx2.2::GFP) adjacent to the MDO in wnt3/wnt3a morphant embryos (Figure 4G, H).

Next, we analyzed the expression of components of the Wnt signaling pathway, wnt8b and axin2 in wnt3/wnt3a morphant embryos. We observed a reduction of wnt8b expression in the central area of the MDO (Figure 4I, J). Consequently, the Wnt signaling target gene axin2 shows a reduced expression in the caudal forebrain (Figure 4K, L). These results suggest that Wnt3 and Wnt3a are required for proper MDO formation. All analyzed marker genes show a consistent alteration - the central area of the MDO shows a down-regulation, whereas the dorsal tip seems to be less affected by the knock-down and displays a residual robust expression of MDO markers such as shh and wnt8b (Figure 4D, H, J, L). We wondered whether MDO fate is induced within tip cells of compound morphant embryos or if cells from the basal plate migrate dorsally to form the MDO spot. Therefore, we performed a time-lapse analysis using the Shh::RFP transgenic zebrafish line with strong expression of the transgene in the basal plate prior MDO formation 25 . At 27 hpf, we detected Shh::RFP expression in the ventral MDO with a progressively dorsal expansion over the next 12 h (Figure 4M). In Wnt3/Wnt3a-deficient embryos we observed induction of dorsal MDO shh expression independently of basal plate contact (Figure 4N). This is in agreement with the so-called bucket-brigade induction model of the MDO 6 . In this model cells progressively adopt MDO fate from ventral to dorsal without changing their dorso-ventral position. Importantly, activation of the canonical Wnt pathway using the GSK3ß inhibitor BIO between 10 and 28 hpf is sufficient to restore MDO formation in Wnt3/Wnt3a deficient embryos (Figure 4O, P).

From the performed knock-down experiments, we conclude that canonical Wnt signaling between 10 and 14 hpf, by Wnt3/Wnt3a, is required for shh induction at the MDO.

Next we studied the consequences of the knock-down of Wnt3/Wnt3a on thalamic regionalization. Therefore, we analyzed the expression pattern of the prethalamic marker fezf2 and the thalamic marker irx1b at 15 hpf and at 28 hpf (Figure 5A-D, 5G-J). We find that the size of expression domains of fezf2 and irx1b are unaltered in Wnt3/Wnt3a compound morphant embryos compared to the control embryos. In addition, in mouse and fish embryos with reduced Otx function, a similar lack of the MDO and of the thalamus has been noted 23 26 . We find that the expression of the forebrain/midbrain marker otx1l (Figure 5E, F, K, L) and of otx2 (Figure 6D, E') is reduced in the morphant forebrain and midbrain suggesting a similar functional connection between Otx and Shh in the compound morphant embryos.

To elucidate this aspect further, we mapped the expression of markers of the prethalamus relative to markers of the thalamus. Interestingly, we found the lhx5 positive prethalamus abuts the irx1b positive thalamus and the MDO anlage is lacking in compound morphant embryos at 28 hpf (Figure 6A-B'). Likewise, analysis of the MDO and thalamus primordia in these embryos showed that the otx2-positive MDO is severely decreased, whereas the size of the irx1b and otx2-positive thalamus is unaltered (Figure 6D-E'). This suggests that Wnt3/Wnt3a are required for establishment of the MDO primordium. Consistently, blockage of Wnt signaling with IWR1 between 10 and 28 hpf leads to a similar phenotype (Figure 6C, C', F, F'). This suggests that Wnt3/Wnt3a function is required to maintain the anlage of the MDO, but not for maintenance of the primordia of the prethalamus and thalamus.

We next considered the Wnt-mediated process that maintains the MDO anlage. Wnt signaling plays a pivotal role in cell survival 27 providing a possible explanation for the lack of the MDO primordium in Wnt3/Wnt3a deficient embryos. Therefore, we analyzed apoptosis by usage of the fluorescent cationic dye, acridine orange, which permeates dying cells to bind chromatin 28 . Indeed, compound morphant embryos display an accumulation of dying cells within the organizer primordium (arrow, Figure 7A-B'). Consistently, blockage of canonical Wnt signaling by treatment with IWR1 leads to a similar increase of apoptotic cells specifically within the organizer anlage (Figure 7C, C'). Consistently, blockage of Tp53 function is able to rescue shh expression in the organizer in Wnt3/Wnt3a morphant embryos as well as in IWR1 treated embryos (Figure 7D-F). This suggests that a main function of Wnt3/Wnt3a in the caudal forebrain is to ensure the survival of MD organizer cells.

Previous evidence has suggested that members of two transcription factor families, Fez 29 and Irx 30 are involved in induction of the Shh-positive organizer. However, recent observations show that prethalamic Fez limits the anterior border of the MDO 31 , whereas the thalamic Irx genes are able to repress Shh expression at the posterior limit of the MDO 1 23 . Therefore, we asked if the main function of Wnt3/Wnt3a in the developing thalamus is the establishment of a Fez/Irx-free zone in the caudal forebrain to allow formation of the MDO. We first tested whether in this situation a similar reduction of shh is seen in embryos deficient for Wnt3/Wnt3a function (Figure 8A, B). As a broadening of the MDO in embryos deficient for FezF2 31 and Irx1b 23 has been reported previously, we wondered if we would be able to restore the MDO in the individual triple morphant embryos. Indeed, upon simultaneous reduction of Fezf2 or Irx1b function, we observe a rescue of shh expression at the MDO of wnt3/wnt3a morphant embryos (Figure 8C, D). To elucidate the impact of the knock-down on the formation of the thalamic complex, we analyzed the expression of the pre-thalamic marker lhx5 and the thalamic proneural factor neurog1. In wnt3/wnt3a/fezf2 triple morphant embryos, the broadening of the shh-positive MDO is accompanied with down-regulation of the dorsal pre-thalamic lhx5+ domain, suggesting that the MDO expands at the expense of the pre-thalamus (Figure 8G). On the other hand, the shh expression domain at the MDO that is restored in the wnt3/wnt3a/irx1b morphant embryos is accompanied by a reduction of the neurog1 positive thalamus, presumably by expansion of the MDO into the thalamic territory (Figure 8H). We, therefore, conclude that both adjacent areas - prethalamus and thalamus - would be competent to act as MDO. However, Fez and Irx are required to repress organizer formation, to shape the overall domain of the MDO and to maintain the identities of adjacent brain regions (Figure 8I-L).

Here we have described a new aspect of Wnt signaling during caudal forebrain development. Wnt3 and Wnt3a mark the MDO prior to the expression of the principal thalamus organizing signal Shh. Blockade of Wnt signaling leads to the lack of the MDO tissue (Figure 9). We show that cells of the organizer primordium go into Tp53-mediated apoptosis upon loss of the Wnt stimulus. We hypothesize that the main function of Wnt3/Wnt3a-mediated signaling in the organizer is to protect cells from cell death as blockage of Tp53-mediated apoptosis is able to restore the organizer function. Thus, Wnt3/Wnt3a-mediated signaling is the survival factor for organizer cells only, but not for the surrounding thalamic complex.

In summary, we show that canonical Wnt signaling is required for regionalization of the caudal forebrain. Alteration of the canonical Wnt signaling pathway leads to apoptosis of the MDO primordium and subsequently to a mis-specification of the entire thalamic complex (Figure 9). We suggest that Wnt3 and Wnt3a are the crucial Wnt ligands, which are required between 10 h and 14 h to maintain the MDO anlage by protecting the cells from Tp53mediated apoptosis. Thus, by determining MDO fate and thalamic compartition, Wnt3 and Wnt3a control the development of the organizer of the major relay station in the brain - the thalamus.

Cer: Cerebellum; cFb: Caudal forebrain; DMSO: Dimethyl sulfoxide; Eth: Epithlamus; HyTh: Hypothalamus; Mb: Midbrain; MDO: Mid-diencephalic organizer; MHB: Midbrain-hindbrain boundary; PBS: Phosphate-buffered saline; PTh: Pre-thalamus; PTh-p: Pre-thalamus primordium; Tel: Telencephalon; Th: Thalamus; Th-p: Thalamus primordium; ZLI: Zona limitans intrathalamica.

The authors declare that they have no competing interests.

BM, SW and JP carried out the molecular genetic studies displayed in this manuscript. QC carried out the immunoassays. SS, CH and GD conceived the study, and participated in its design and coordination. SS guided the project and wrote the manuscript. All authors read and approved the final manuscript.