Differences in the spatiotemporal expression and epistatic gene regulation of the mesodiencephalic dopaminergic precursor marker PITX3 during chicken and mouse development

Neurons producing the neurotransmitter/modulator dopamine (DA) are present in the central nervous system of all chordates and in most other invertebrate phyla (Barron et al., 2010; Yamamoto and Vernier, 2011). The most prominent DA-synthesizing neuronal population, however, has evolved in tetrapods (amphibians, reptiles, birds and mammals) (Marín et al., 1998; Yamamoto and Vernier, 2011). These neurons emerge from ventral mesencephalon and diencephalon [prosomeres 1-3 (p1-p3)], and have been termed mesodiencephalic dopaminergic (mdDA) neurons (Puelles et al., 2013; Smidt and Burbach, 2007). The implication of mdDA neuron degeneration and/or dysfunction in several human neurological and psychiatric diseases, such as Parkinson's disease, addictive disorders, schizophrenia, attention deficit/hyperactivity disorders and depression (Baik, 2013; Dauer and Przedborski, 2003; Howes and Kapur, 2009; Hyman et al., 2006; Nieoullon, 2002), has led to a strong clinical interest not only in deciphering the normal physiology and pathophysiology of these neurons, but also the cues responsible for their generation and survival during development and in adulthood.

In mouse, the intersection of three secreted morphogens, Fgf8, Wnt1 and Shh, controls the induction of proliferating mdDA progenitors in the ventral mesodiencephalic area (Hegarty et al., 2013; Smidt and Burbach, 2007). The mid-hindbrain boundary (MHB) releases Fgf8 and Wnt1, although Wnt1, however, additionally shows a localized expression in ventral mes- and diencephalon (Hegarty et al., 2013; Smidt and Burbach, 2007; Wurst and Prakash, 2014). Shh secreted from the floor plate (FP) is required for the establishment of progenitor domains in ventral mesencephalon (Blaess et al., 2006; Perez-Balaguer et al., 2009). However, to generate mdDA neurons Wnt/β-catenin signaling has to suppress expression of Shh in this region (Joksimovic et al., 2009; Joksimovic and Awatramani, 2014; Wurst and Prakash, 2014). Wnt1/β-catenin signaling also sequentially induces the expression of several transcription factors (TFs) in postmitotic mdDA precursors, such as Lmx1a and Pitx3, both of which are necessary for proper mdDA neuron development (Blaess and Ang, 2015; Hegarty et al., 2013; Veenvliet and Smidt, 2014; Wurst and Prakash, 2014). Pitx3 regulates the transcription of several genes necessary to ensure the correct differentiation, function and survival of the substantia nigra pars compacta (SNc) mdDA neurons (Jacobs et al., 2009, 2007; Luk et al., 2013; Maxwell et al., 2005; Peng et al., 2011; Veenvliet et al., 2013). Pitx3 in this area has been found to cooperate with the nuclear receptor Nr4a2 (also known as Nurr1) and with Aldh1a1 (also known as Raldh1 and Ahd2) (Jacobs et al., 2009, 2007). Further genes implicated in mouse mdDA neuron development include the TFs Lmx1b, Engrailed (En1/2), and Neurog2 (also known as Ngn2) (Blaess and Ang, 2015; Hegarty et al., 2013; Veenvliet and Smidt, 2014).

In chick, the first neurons expressing the rate-limiting enzyme for DA synthesis, tyrosine hydroxylase (TH), appear in ventrocaudal diencephalon on embryonic day (E) 5.5-6, and only at E8 in the midbrain tegmentum (Puelles and Medina, 1994; Smeets and González, 2000). Previous studies in chicken embryos suggested a time-course similar to that in the mouse embryo as well as a SHH-mediated regulation of mdDA neuron generation (Agarwala et al., 2005; Andersson et al., 2006b; Bayly et al., 2007; Watanabe and Nakamura, 2000). However, little if anything is known about the early steps of mdDA neuron development in chick embryos. We, therefore, embarked on a detailed description of the early spatiotemporal expression pattern of the chicken ortholog of mouse Pitx3 and an analysis of the epistatic relationships between this TF and WNT and SHH signaling pathways in the mesodiencephalic region. We show significant differences between the spatiotemporal expression patterns of selected mouse and chicken mdDA marker genes, and the epistatic regulation of Pitx3 transcription. These results might reflect the evolutionary divergence of the mdDA neuronal population in birds and mammals.

We first determined the expression patterns of the chick orthologs to the mouse mdDA marker genes Pitx3, Nr4a2 and Aldh1a1 in relation to LMX1A and LMX1B, both markers for mdDA precursors in chick (Andersson et al., 2006b), and NGN2, a generic marker for mdDA neurogenesis (Andersson et al., 2006a; Kele et al., 2006). At Hamburger–Hamilton stage (HH) 16/17 (E2.5) PITX3 was expressed in a small domain within the ventral diencephalon (Fig. 1B,B′), whereas neither NR4A2 nor ALDH1A1 were detected in the mesodiencephalic region (Fig. 1C-D′). LMX1B was present along the entire mesodiencephalic area, whereas LMX1A and NGN2 were only found rostrally in this region (Fig. 1E-G′). Coronal sections showed that PITX3, LMX1A/B and NGN2 demarcate an overlapping area along the dorsoventral axis of the diencephalic domain (Fig. 1B′-G′). At HH 21 (E3.5) the postmitotic mdDA marker NR4A2 initiated expression in the caudally expanded PITX3⁺ diencephalic area (Fig. 1H-J). At this time, PITX3 was mostly confined to ventral diencephalic p1/2 domain, which is delimited by the SHH⁺ zona limitans intrathalamica (ZLI) and the PAX6 expression at the di-/mesencephalic boundary (Kiecker and Lumsden, 2005; Puelles and Rubenstein, 2003) (Fig. 1K,L). As development progressed, TH mRNA appeared at HH 29 (E6.5) within the EN1⁺ ventral midbrain (Fig. 1M-N′; Puelles and Medina, 1994), and the expression of PITX3 abutted the FGF8⁺ domain at the MHB (Fig. 1N″). Coronal sections revealed that the TH⁺ territory overlapped only laterally with that of PITX3 and NR4A2 (Fig. 1N′-O′,O″,P″).

At E9 (HH 35), PITX3 expression overlapped with TH and NR4A2 in the mesodiencephalic A9 (SNc) and A10 (ventral tegmental) areas, and a subset of the pretectal and hypothalamic A12/A14 DA cell groups in the chick (Fig. S1A-J). Notably, PITX3 and NR4A2 were not expressed in the dorsal diencephalic A11 DA cells (Fig. S1C,D) and PITX3 is not exclusively expressed in NR4A2- or TH-positive cells (Fig. S1G-J). Strikingly, no expression of ALDH1A1 or its two paralogs, ALDH1A2 and ALDH1A3 (ALDH6), was detected in the mesodiencephalic area during the stages analyzed (Fig. 1D,D′; data not shown). We concluded that PITX3 together with LMX1A/B and NGN2 are the first genes associated with mdDA production to be expressed in these regions of chick neural tube. PITX3 is initially confined to the diencephalic p1-3 domain in the rostral cephalic flexure but extends later into the ventral mesencephalon. Our analysis also revealed that, in contrast to mouse, TH transcription initiates only 4 days after PITX3 in PITX3⁺ and NR4A2⁺ cells. Later in development, TH is co-expressed with PITX3 and NR4A2 in the mesodiencephalic (A9/10), pretectal and hypothalamic (A12/14) DA groups, suggesting that PITX3⁺ and NR4A2⁺ precursors in chick ventral di-mesencephalon generate the mdDA neurons, among others. An overexpression of siRNA (pRNAT-siPitx3-GFP) against PITX3 in midbrain resulted in a reduction of PITX3, NR4A2, LMX1B and TH expression in this region (Fig. S2; n=2), suggesting that PITX3 does indeed play a role for the development of TH⁺ DA neurons in chick mesencephalon.

Our data suggested that, in contrast to mouse, in which Pitx3 expression is confined to postmitotic mdDA precursors and neurons (Maxwell et al., 2005; Smidt et al., 1997; Zhao et al., 2004), in chick PITX3 is expressed earlier in the proliferating neural progenitors. To verify whether this is indeed the case, we pulse-labeled E3.5 chicken embryos for 2 h with 5′-bromo-2-deoxyuridine (BrdU) and determined the number of BrdU⁺/PITX3⁺ double-labeled cells versus those labeled with BrdU alone in the diencephalic domain at E3.5 (Fig. 2A,B). We found that ∼20% of all BrdU-labeled cells were also expressing PITX3 at E3.5 (Fig. 2C) [PITX3⁺/BrdU⁺ cells (mean±s.d.): chick #1, 19.07±8.04% (n=6 sections); chick #2, 21.19±6.65% (n=7); chick #3, 18.62±6.07% (n=5); chick #4, 23.68±4.97% (n=5)]. Thus, at E3.5 a fraction of the diencephalic PITX3⁺ cells were neural progenitors in S phase. Double-labeling for PITX3 and HuC/D, which labels post-mitotic neurons, confirmed that PITX3 was expressed in the ventricular zone (VZ) and intermediate zone (IZ), and not in the HuC/D⁺ mantle zone (MZ) of rostral diencephalon at HH 26 (E5; Fig. 2D,E). However, at E5 in ventral midbrain, PITX3 expression was located exclusively in the HuC/D⁺ MZ (Fig. 2F,G). Hence, chick PITX3 shows an unexpected dichotomy in its expression. In diencephalic cells it is present in proliferating neural progenitors and early postmitotic precursors, whereas in the mesencephalon it appears in late postmitotic precursors and differentiating neurons.

These results suggested that the spatiotemporal expression pattern of PITX3 in this region of the developing brain of mouse and chicken might have diverged during evolution. Thus prompted, we re-examined the spatiotemporal expression pattern of Pitx3 in the developing mouse embryo. Pitx3 is first detected at E11.5 in mouse brain (Smidt et al., 1997), and we confirmed restricted Pitx3 expression in ventral midline of the mouse mesencephalon between E11.5 and E14.5 (Fig. S3A,A′,C,C′,E,E′; data not shown; n=3 per stage; Hoekstra et al., 2013). However, our expression analysis further revealed two additional Pitx3⁺ domains in the di- and mesencephalon. We found that Pitx3 expression was also located at the dorsolateral caudal mesencephalon close to the MHB between E12.5 to E18.5 (Fig. S3B,B′,D,D′,F,F′; n=3 per stage). From E14.5 to E18.5, the initially ventral mesencephalic expression domain of Pitx3 extended as a narrow stripe in the ventrolateral caudal diencephalon (Fig. S3F-H′). The Pitx3⁺ domain in dorsolateral caudal midbrain did not overlap with Th expression (Fig. S1B′,G-H‴; data not shown), whereas the Pitx3⁺ domain in the ventrolateral caudal diencephalon clearly overlapped with Th expression (Fig. S3G-H‴). Thus, in mouse, Pitx3 expression is initially confined to the mesencephalic mdDA domain and only later extends into a diencephalic mdDA domain. This is the exact opposite of the expression pattern of PITX3 in chick, which initiates in the diencephalon and then expands into the midbrain. In mouse, Pitx2, which is expressed in diencephalon might play a similar role as chick PITX3 in this brain region (Martin et al., 2002; Mucchielli et al., 1996). We also discovered a previously unknown non-dopaminergic Pitx3 expression domain in the developing mouse mesencephalon.

Murine Pitx3 is an indirect target of the WNT1/β-catenin signaling pathway during mouse mdDA development (Chung et al., 2009). To investigate whether PITX3 in chick is similarly induced by a member of the WNT/β-catenin family, we conducted a detailed literature and radioactive in situ hybridization (RISH) screen of chick WNT gene expression patterns at E2.5. Our research identified three patterns of WNT gene expression in this region. The first is exemplified by WNT11 and WNT8C, which are not expressed in anterior neural tube (see Fig. 3K for WNT11 and Hume and Dodd, 1993 for WNT8a). The second is displayed by WNT1, WNT3A, WNT4, WNT6, WNT7B and WNT8B, all of which are either restricted to the dorsal midline or spared the ventral midline of the di- and mesencephalon (Fig. 3F-J; for WNT7B and WNT8B see Garda et al., 2002). Thus, their expression did not overlap with the PITX3⁺ mdDA domain in E2.5 brains. The third pattern is that shown by WNT5A, WNT5B, WNT7A and WNT9A, which do show expression in the ventral mesodiencephalic region (Fig. 3A-E′). Of these, WNT9A overlapped closest with PITX3 expression in the cephalic flexure of the E2.5, E3.5 and E5 chick brains (Fig. 3L-R), and thus its expression resembled that of mouse Wnt1. Interestingly, murine Wnt9a expression was mostly restricted to the dorsal midline of the anterior neural tube (fore-, mid- and hindbrain) in midgestational mouse embryos (Fig. S4A-F′), resembling WNT1 expression pattern in chick brain. In E5 chick brain, WNT9A expression became confined to the VZ of the cephalic flexure, whereas PITX3 expression was restricted to the mesencephalic MZ (Fig. 3R; Fig. 2D). Taken together, WNT9A and PITX3 are transcribed in an overlapping pattern in the cephalic flexure of the early chicken brain (E2.5-E3.5; Fig. 3S). Later in development (E5), WNT9A is expressed in di-/mesencephalic neural progenitors in the VZ generating the PITX3⁺ postmitotic precursors in the MZ. Thus, it would seem that WNT9A performs a similar function in chicken as Wnt1 in mouse.

To establish a possible role for WNT9A for the induction of PITX3 expression and the formation of mdDA precursors in the chicken brain, we electroporated full-length and bicistronic WNT9A/GFP (pMES-WNT9A-IRES-eGFP) into the right ventrolateral and/or lateral half of the mesencephalon at E1.5 (HH 10-12). After 1 day post-electroporation (dpe) neither SHH (Fig. 4D), nor PITX3, NR4A2 or LMX1B were induced ectopically (data not shown); only LMX1A (Fig. 4A-C) was detected in the WNT9A electroporated side. At 2 dpe (n=2), SHH and PITX3 expression expanded ectopically towards the dorsal mesencephalon in the electroporated side (Fig. 4E,G,H). WNT9A-induced LMX1A expression had ceased by 2 dpe (Fig. 4E,F,I,J). At 3 dpe (n=5), ectopic PITX3, NR4A2 and NGN2 were still present within the WNT9A electroporated domain (Fig. 4K,L). WNT9A overexpression also resulted in ectopic NGN2 expression in ventrolateral mesencephalon (compare bracketed areas in left and right midbrain in Fig. 4L), and ectopic PITX3 expression.

Importantly, we show that ectopic LMX1A was only found in WNT9A-expressing cells (Fig. 4C). By contrast, cells ectopically expressing SHH, PITX3 and NR4A2 were both those electroporated with WNT9A as well as adjacent non-electroporated, WNT9A-negative cells (e.g. compare Fig. 4E with Fig. 4G,H,L). This suggests a non-cell-autonomous induction of the latter genes. Notably, the expansion of ectopic SHH, PITX3 and NR4A2 respected a ventrodorsal border in the midbrain, with no expression occurring in dorsal mesencepalon even though ectopic WNT9A⁺ cells were present dorsally (compare Fig. 4E with Fig. 4G,H, and Fig. 4I with Fig. 4K,L). This might be due to the fact that there are SHH and WNT inhibitors expressed in the dorsal midbrain (Ladher et al., 2000; Li et al., 2007; Paxton et al., 2010; Quinlan et al., 2009).

To determine whether the ectopic induction of LMX1A, and subsequently of PITX3, after overexpression of WNT9A was specific to WNT9A or if it could also be mediated by other WNTs expressed in the ventral midline of the chicken di- and mesencephalon, we electroporated full-length WNT1, WNT5A and WNT7A into ventrolateral di-mesencephalon (Fig. S5). Interestingly, at 3 dpe we did not detect any ectopic expression of PITX3, LMX1A/B or SHH after overexpression of any of those WNTs.

Our findings suggest that the expression of LMX1A is directly (cell-autonomously) activated by WNT9A-mediated signaling, whereas the ectopic induction of SHH, PITX3 and NR4A2 might be mediated by an intermediate signaling pathway or an indirect (non-cell-autonomous) mechanism in the developing chick mesencephalon.

To determine whether LMX1A is sufficient to induce the subsequent expression of SHH, PITX3, NR4A2 and LMX1B in chick mesencephalon, we electroporated mouse Lmx1a (Andersson et al., 2006b) together with eGFP as marker gene (pCAX-EGFP; Chen et al., 2004) into the right ventrolateral di-mesencephalon. One day after Lmx1a/GFP electroporation (n=5), WNT9A, SHH and PITX3 expression remained restricted to their endogenous domains in the right ventral diencephalon (Fig. 5A-D). At 2 dpe of Lmx1a/GFP (n=4), ectopic expression of SHH and PITX3 was detected in the electroporated ventrolateral mesencephalon in a similar continuous and dorsally extending pattern as observed after WNT9A electroporation, whereas ectopic Lmx1a expression was patchy (compare Fig. 5E with 5G,H). WNT9A expression was only weakly induced in mesencephalon at 3 dpe of Lmx1a/GFP (n=8) (Fig. 5F,I,J). At 3 dpe, there was also strong ectopic, continuous and dorsally extending expression of NR4A2, LMX1B and NGN2, and additionally SHH and PITX3 expression appeared in the transfected ventrolateral mesencephalon. Again, the extension of ectopic expression appeared to obey a similar ventrodorsal limit as observed after WNT9A electroporation (Fig. 5K-O). Thus, Lmx1a can induce ectopic expression of mdDA precursor markers with a similar temporal delay of 2 dpe as WNT9A in chick mesencephalon (Fig. 5P). Taken together, our findings strongly indicate that ectopic induction of mdDA precursor markers by WNT9A and its putative direct target gene LMX1A is in fact mediated by an intermediate signaling pathway (non-cell-autonomous) in the developing chicken mesencephalon. In addition, the indirect activation of WNT9A transcription by Lmx1a suggests a feedback mechanism between these two genes in the chicken midbrain.

Because SHH expression expanded after ectopic WNT9A or Lmx1a electroporation, we next investigated whether SHH was sufficient to induce expression of PITX3 and other mdDA marker genes in the mesencephalon. Overexpression of SHH in ventrolateral di-mesencephalon led to an ectopic induction of PITX3 in the transfected tissue after 2 dpe (n=3) but not after 1 dpe (n=2) (Fig. 6A,B,D,E). Ectopic PITX3 persisted at 3 dpe (n=5) and was present in non-transfected (GFP⁻) cells (compare Fig. 6F,G), with a similar ventrodorsal midbrain boundary to that observed after WNT9A and Lmx1a overexpression (see Figs 4 and 5). SHH did not induce any ectopic expression of LMX1A and WNT9A even after 3 dpe. This indicated that both genes act upstream of SHH (Fig. 6C,H-J). We concluded that WNT9A-mediated signaling directly activates its putative target gene LMX1A and that LMX1A in turn induces the expression of SHH, which is sufficient for an induction of PITX3 expression in the chicken mesencephalon.

Our results suggested that the SHH signaling pathway is the most likely candidate for any indirect (non-cell-autonomous) activation of ectopic PITX3 transcription by WNT9A in chicken mesencephalon. To determine whether SHH signaling is necessary for ectopic induction of PITX3, we treated WNT9A- or LMX1A-electroporated neural tubes with cyclopamine, a potent inhibitor of SHH signaling pathway (Chen et al., 2002; Incardona et al., 1998), or with DMSO, as control. Two days after electroporation of WNT9A and treatment with DMSO (n=2-5 per treatment), PITX3 expression was induced and expanded dorsally into the ectopic WNT9A area (Fig. 7A,B). The expression of PTCH1, the SHH receptor (Chen and Struhl, 1998), appeared also stronger within the electroporated site (Fig. 7C). Embryos at 3 dpe (n≥2 per treatment) of LMX1A treated with DMSO showed weakly ectopic WNT9A and strong ectopic, dorsally expanded expression of SHH, PITX3 and NR4A2 in the mesencephalon (Fig. 7G-K). Thus, as already previously described (Figs 4 and 5), both ectopic WNT9A and LMX1A induced ectopic SHH and PITX3. However, in the cyclopamine-treated embryos no ectopic PITX3 transcription was induced at 2 dpe (Fig. 7D,E,L-O). Endogenous PITX3 and PTCH1 expressions in the electroporated side are both slightly repressed, despite a strong ectopic induction and dorsal expansion of SHH transcription (Fig. 7D,E,L-O; data not shown). Interestingly, cyclopamine treatment did not affect endogenous expression of PITX3, SHH, WNT9A and PTCH1 in the untransfected mesencephalon (Fig. 7D-F,M-P). These results indicate that SHH signaling is required for the ectopic activation but not for the endogenous maintenance of PITX3 transcription.

Interestingly, the endogenous ventrolateral expression domain of PTCH1 corresponds to the ventrodorsal extent of ectopic gene induction observed after WNT9A electroporation (Fig. 7B,C) and is reduced after cyclopamine treatment. Thus, the reception of the SHH signal via PTCH1 might be essential for ectopic activation of PITX3 in this region. Blocking the SHH signal transduction pathway by cyclopamine mainly abolished ectopic transcription of PITX3 without affecting endogenous and/or ectopic transcription of WNT9A and SHH (Fig. 7H,I). We conclude that signaling mediated by the ectopically induced expression of SHH after electroporation of WNT9A or LMX1A is necessary and sufficient to activate the ectopic transcription of PITX3 in chick mesencephalon.

In this study, we have shown a number of key differences between chicken and mice with respect to the spatiotemporal expression pattern of selected mdDA marker genes, such as PITX3 and ALDH1A1, and their regulation by the WNT/β-catenin and SHH signaling pathways. These differences point to evolutionary divergences in the genetic mechanisms that control the generation of proliferating mdDA progenitors and precursors between birds and mammals.

In mice, Pitx3 is crucial for the correct differentiation of mdDA neurons, especially of the SNc DA neurons, and their survival during development and adulthood (Blaess and Ang, 2015; Hegarty et al., 2013; Veenvliet and Smidt, 2014). Pitx3 expression initiates in the ventral mesencephalon and later extends into the caudal ventral diencephalon and it is restricted to postmitotic mdDA precursors and neurons located in the MZ of mesencephalic tegmentum (this work; Hoekstra et al., 2013). The expression of this gene initiates only shortly before or just after Th expression at E11.5 in mouse (Maxwell et al., 2005; Smidt et al., 1997; Zhao et al., 2004). We have shown here that the murine Pitx3 expression pattern is only partly conserved in chick embryos: PITX3 transcription initiates in the ventral chick diencephalon (p1/2) and only later extends into the mesencephalon. In fact, PITX3 together with LMX1A/B and NGN2 are the first mdDA marker genes expressed in the cephalic flexure of chick brain even before NR4A2 transcription is initiated. In the chick diencephalon, PITX3 is expressed in proliferating neural progenitors and early postmitotic precursors. However, in the chick mesencephalon PITX3 expression is restricted to postmitotic precursors and neurons within the tegmentum. We also detect a notable time gap of 4 days between the initiation of PITX3 transcription in chick mesodiencephalic region (at E2.5) and the first expression of TH (at E6.5; Puelles and Medina, 1994). This time gap contrasts with that seen during mouse development but resembles the 4-day delay in TH expression during the histogenesis of chick DA amacrine neurons (Gardino et al., 1993; reviewed by Smeets and González, 2000). TH transcription in the mesencephalon of the E6.5 chick embryo is restricted to a lateral tegmental domain that appears to derive from a PITX3⁺ and NR4A2⁺ neural (mdDA) precursor area. At E9, expression of TH overlaps with that of PITX3 and NR4A2 in the mesodiencephalic A9/10, pretectal and hypothalamic A12/14 DA cell groups, suggesting that these neurons do indeed derive from PITX3⁺ and NR4A2⁺ precursors.

We also discovered another notable difference between chicken and mice and that is the lack of ALDH1A1 expression in the chick ventral mesencephalon. In mouse, Aldh1a1 expression initiates around E9.5 in proliferating mdDA progenitors and is later confined to a rostrolateral mdDA neuron subset (Jacobs et al., 2007; Smits et al., 2013; Stuebner et al., 2010; Wallén et al., 1999). The complete absence of expression of ALDH1A1 and its two paralogs, ALDH1A2 and ALDH1A3 (ALDH6), in the chicken mesencephalon is similar to the lack of ALDH1A expression in quail mesencephalon during development (Reijntjes et al., 2005). Although we cannot exclude the expression of these enzymes in the chicken VM later in development, our finding strongly suggests that the retinoic acid-synthesizing and DA-metabolizing ALDH1 family members are not involved in early stages of chicken mdDA neuron development and survival.

By contrast, we found that the spatiotemporal expression patterns of LMX1A/B, NR4A2 and NGN2 were conserved between chick and mice (this report; Andersson et al., 2006b), suggesting that these TFs are under similar transcriptional control and most likely direct the same developmental pathways (Andersson et al., 2006b). The transcription of endogenous and ectopic NR4A2 always initiated one day after endogenous and ectopic PITX3 expression in chick, indicating that the onset of NR4A2 transcription after PITX3 was preserved after ectopic overexpression of upstream inducing factors.

The divergent spatiotemporal expression pattern of PITX3 in chick and mouse suggests different transcriptional regulation between the species. In mice, WNT1-mediated β-catenin signaling controls the correct differentiation of mdDA progenitors/precursors into mature mdDA neurons through the direct activation of Lmx1a (Chung et al., 2009; Joksimovic et al., 2009; Prakash et al., 2006; Tang et al., 2009, 2010; Joksimovic and Awatramani, 2014; Wurst and Prakash, 2014). Murine LMX1A, in turn, binds to and activates the transcription of the Wnt1, Pitx3 and Nr4a2 promoters (Chung et al., 2009; Wurst and Prakash, 2014; our own unpublished data). In contrast to the chick (Andersson et al., 2006b), in the murine brain transcription of Lmx1a, Pitx3 or Nr4a2 is not regulated by the SHH signaling pathway (Chung et al., 2009). Hence mature mouse TH⁺ mdDA neurons do not derive from SHH-responsive cells (Mesman et al., 2014). Furthermore, ablation of SHH signaling components in midgestational mdDA precursors does not affect the generation of mdDA neurons in the mouse embryo (Blaess et al., 2006; Zervas et al., 2004). Instead, the expression of Shh has to be antagonized by active WNT/β-catenin signaling to enable mdDA neurogenesis in mouse mesencephalon (Joksimovic et al., 2009). Thus, the SHH signaling pathway seems to play a minor role during murine mdDA neuron development (Fig. 8B).

We have also investigated the contribution of the WNT/β-catenin pathway to the generation of mdDA precursors in the chick and the epistatic relationships between SHH and WNT signaling pathways and PITX3 in this species. We confirmed that WNT1 is not expressed in the chick ventral mesodiencephalic region (Hollyday et al., 1995), but rather found that WNT9A exhibits the closest overlap with PITX3 expression in this region. Although WNT1 and WNT9A expression in mouse and chick is reversed, both WNT proteins signal via the WNT/β-catenin (‘canonical’) pathway (Guo et al., 2004; Megason and McMahon, 2002), and in keeping with this we found that ectopic expression of WNT9A activated ectopic transcription of LMX1A in a cell-autonomous manner. Thus, the epistatic relationships and gene regulatory interactions between the WNT signaling pathway and Lmx1a seem to be conserved between chicken and mice (Fig. 8A,B). Ectopic and non-cell-autonomous induction of SHH, PITX3, NR4A2 and NGN2 expression with a temporal delay of 2-3 days after electroporation of WNT9A or LMX1A into the mesodiencephalic area of the chick suggested that these are indirect targets of WNT9A and LMX1A.

In contrast to the mouse, WNT9A and LMX1A appear to act upstream of the SHH signaling pathway in the chick. We showed that SHH signaling was both necessary and sufficient to induce ectopic and non-cell-autonomous expression of PITX3, but not of WNT9A and LMX1A in the chicken mesodiencephalic area (Fig. 8A). Our results agree with those of Watanabe and Nakamura (2000) who showed that ectopic expression of SHH in the chicken mesencephalon induced non-cell-autonomously ectopic TH⁺ DA neurons, and thus strongly supports a crucial role of the SHH pathway for the generation of mdDA neurons in chick but not for the maintenance of PITX3 transcription. In agreement with Andersson et al. (2006b), we noted that despite a strong expression of transgenes in dorsal mesencephalon no ectopic induction of mdDA marker genes in dorsal midbrain ever occurred. This dorsoventral expression boundary seems to coincide with expression of PTCH1, which is restricted to the ventral midbrain. Thus, only the ventrolateral mesodiencephalic region of the chick embryo appears to be competent to respond to the SHH inductive signal for mdDA neuron development in this region. The limited ectopic induction of SHH and PTCH1 to ventral midbrain might be due to the expression of several proteins present in dorsal midbrain, such as RAB23 and GLI3, that suppress SHH signaling dorsally (Li et al., 2007; Litingtung and Chiang, 2000; Persson et al., 2002). Our observations also further confirm the presence of a dorsoventral signaling boundary in the chick midbrain (Li et al., 2005).

The clear differences in spatiotemporal appearance of the PITX3⁺ mdDA progenitors/precursors and in PITX3 transcriptional regulation by the WNT/β-catenin and SHH signaling pathways point not only to ontogenetic but also phylogenetic differences in mdDA development between birds and mammals. In fact, the transcriptional regulation of PITX3 in the early chick embryo exhibits more similarities to zebrafish than to mouse. Zebrafish pitx3 is expressed in proliferating diencephalic progenitors and postmitotic precursors, but not in Th⁺ DA neurons (Filippi et al., 2007). Moreover, zebrafish pitx3 is a target of the Nodal (a TGF family member) and Hh pathways in neural tissues (Zilinski et al., 2005). We and others showed that both pathways are implicated in chicken mdDA development (Agarwala et al., 2005; Andersson et al., 2006b; Bayly et al., 2007; Farkas et al., 2003; Watanabe and Nakamura, 2000). Sequence comparisons show the chick PITX3 gene being more closely related to Xenopus and zebrafish pitx3 than to the mammalian Pitx3 genes (Zilinski et al., 2005). The cues directing the expression (and possibly also function) of Pitx3 during mdDA development thus appear to have diverged considerably between the avian and teleost lineages and the mammalian lineage, and it remains to be investigated whether this is also the case for the amphibian and reptile lineages.

Fertilized White Leghorn chicken eggs (Brüterei Hölzl, Moosburg, Germany) were incubated at 38°C until the desired HH stage (Hamburger and Hamilton, 1951). Outbred CD-1 mouse embryos were collected from timed-pregnant females (Charles River, Kisslegg, Germany); noon of the day of vaginal plug detection was designated as E0.5. Pregnant dams were killed by CO₂ asphyxiation. All expression studies are based on a minimum of three embryos per stage. This study was carried out in strict accordance with the recommendations in the EU Directive 2010/63/EU and the Guide for the Care and Use of Laboratory Animals of the Federal Republic of Germany (TierSchG). The protocol was approved by the Institutional Animal Care and Use Committee (ATV) of the Helmholtz Zentrum München. All efforts were made to minimize suffering.

Paraffin sections (8 µm) were processed for radioactive ISH (RISH) as described by Fischer et al. (2007). Whole-mount in situ hybridization (WISH) using digoxigenin- or fluorescein-labeled riboprobes was performed according to Henrique et al. (1995). Riboprobes used were: chicken PITX3 (ChEST246m15), ALDH1A1 (ChEST396f5), ALDH1A2 (ChEST650k18), EN1 (ChEST92p12), TH (ChEST1010e8), WNT3A (ChEST1005M7) obtained from UK chicken EST Consortium (Boardman et al., 2002); ALDH6 (ALDH1A3; RefSeq NM_204669), LMX1A (XM_001236605) and PTCH1 (NM_204960) (R. Klafke, Helmholtz Zentrum München, Neuherberg, Germany); LMX1B (Matsunaga et al., 2002); NGN2 (NEUROG2) (D. Henrique, Instituto de Medicina Molecular and Instituto de Histologia e Biologia do Desenvolvimento, Lisboa, Portugal); SHH (Nohno et al., 1995); PAX6 (Goulding et al., 1993); FGF8 (Crossley et al., 1996); WNT1 (Bally-Cuif and Wassef, 1994); WNT4, WNT5A and WNT5B (Hartmann and Tabin, 2000); WNT6 (NM_001007594) and WNT7A (NM_204292) (A. Wizenmann, Institute of Clinical Anatomy and Cell Analysis, Tübingen, Germany); WNT9A (Hartmann and Tabin, 2001); WNT11 (C. Hartmann, Institute of Experimental Musculoskeletal Medicine, Muenster, Germany); mouse Pitx3, Th and Nr4a2 (Brodski et al., 2003); mouse Wnt9a (NM_139298) (J. Zhang, Helmholtz Zentrum München, Neuherberg, Germany); and GFP (R. Koester, Cellular and Molecular Neurobiology, Braunschweig, Germany).

Whole-mount IHC was performed as described by Li et al. (2005) after WISH for PITX3 using mouse anti-HuC/D, a marker for early postmitotic and differentiating neurons (1:600; A-21271, Molecular Probes). Post-fixed [4% paraformaldehyde (PFA)], gelatin-albumin-embedded embryos were vibratome sectioned (40 µm).

BrdU (250 µg/ml in 0.9% saline; Sigma) was injected into anterior neural tube of HH 21 (E3.5) embryos (Belecky-Adams et al., 1996), incubated for 2 h, fixed in 4% PFA, processed for PITX3 WISH, post-fixed in 4% PFA, transferred into 20% sucrose/PBS and cryosectioned (16 µm) and processed with a BrdU Labeling and Detection Kit II (Roche). PITX3⁺/BrdU⁺ double-labeled and BrdU⁺ single-labeled cells were counted (five to seven sections, four different BrdU-treated embryos), and the proportion of double- versus single-labeled BrdU⁺ cells was calculated for each embryo.

Images were taken with a LSM 510 META confocal laser scanning microscope, Axioplan2 microscope or StemiSV6 stereomicroscope (Zeiss) and processed with Adobe Photoshop CS3 software (Adobe Systems).

Full-length chicken WNT9A (pMES-WNT9A-IRES-eGFP), WNT5A (pMES-WNT5A-IRES-eGFP), WNT7A (pMES-WNT7A-IRES-eGFP), SHH (pMES-SHH-IRES-eGFP), and mouse Lmx1a (pMES-LMX1A-IRES-eGFP) and Wnt1 (pMES-WNT1-IRES-eGFP) cDNAs were inserted into the pMES expression vector, which contains an internal ribosomal entry site (IRES) followed by an enhanced green fluorescent protein (eGFP) (Swartz et al., 2001). The pECE-mLmx1a and pCAX-EGFP vectors were described by Andersson et al. (2006b) and Chen et al. (2004). For further information on generation and testing of vector constructs, see supplementary Materials and Methods.

Chick neural tubes were injected with vector DNA [1-3 µg/µl, 0.005% Fast Green (Sigma)] and electroporated into ventrolateral and/or lateral di-mesencephalon (Huber et al., 2013). After 1-3 days (dpe), embryos were removed from eggs, staged, and fixed in 4% PFA at 4°C.

Cyclopamine [100 µM in DMSO (Biomol)] or DMSO (as control) were injected into the neural tube 3 and 24 h after electroporation. Embryos were incubated for another 24-48 h before fixation (4% PFA).

We thank S. Badeke for excellent technical assistance; J. Ericson for the pECE-Lmx1a construct; and J. Guildford, C. Hartmann, D. Henrique, R. Koester and J. Zhang for plasmids and riboprobes. To S. Blaess and A. Graham we are indebted for advice, helpful discussions and for critical reading of the manuscript.