The rate-limiting enzyme, tyrosine hydroxylase (TH), is involved in the synthesis of L-dopa from tyrosine, which represents the first step in the synthesis of catecholamines (dopamine, noradrenaline, and adrenaline). All the known catecholaminergic regions in mid- and forebrain are dopaminergic, corresponding to the classic A8 (retrorubral area), A9 (substantia nigra), A10 (ventral tegmental area), A11 to A15 (diencephalic and hypothalamic groups), A16 (periglomerular cells in the olfactory bulb), and A17 (interplexiform cells in the retina) areas. According to ontogenetic studies in different vertebrates TH expression is an early marker for the development of these nuclei (Specht et al., 1981a, b; di Porzio et al., 1990; Medina et al., 1994; Puelles and Medina, 1994; Puelles and Verney, 1998; Vitalis et al., 2000).
Thus far, TH-expressing cells during embryonic development have been documented using immunohistochemical techniques. We have decided to perform a study on TH mRNA expression. This kind of approach can provide new information on the expression of a given molecule, considering that posttranscriptional regulation can lead to null or low protein levels, undetectable by immunohistochemistry. In fact, TH expression is known to be regulated at the posttranscriptional level by different mechanisms (Vyas et al., 1990; Kumer and Vrana, 1996).
The study of TH mRNA expression could be particularly important in relation to research on mechanisms regulating dopaminergic phenotype. Several factors such as Shh, FGF8, Pax6, Nurr1, or Ptx3 are involved in dopaminergic differentiation (review by Lin and Rosenthal, 2003). The study of these and other molecules as possible inducers of TH expression in the developing brain, and other model systems would benefit from an accurate description of mRNA transcription profile.
Therefore, we have mapped TH mRNA regions in the developing mid- and forebrain of mouse by in situ hybridization (ISH). We have applied the neuromeric model in our characterization. According to histochemical and gene expression data as well as studies on tract formation and cell lineages, this model subdivides the brain in longitudinally ordered segments (neuromeres), which are additionally regionalized in the dorsoventral axis. It has been successfully applied to the study of hindbrain, midbrain, and forebrain (Vaage, 1969; Puelles et al., 1987, 2004; Lumsden and Krumlauf, 1996; Puelles and Rubenstein, 2003). Although currently subjected to revisions on given interneuromeric boundaries (Larsen et al., 2001; Puelles and Rubenstein, 2003), it provides a reliable Cartesian framework to correlate spatiotemporally the expression of different molecules involved in brain development. Concerning mid- and forebrain, we consider them as formed by five neuromeres: midbrain, prosomeres (p) 1 to 3, and secondary prosencephalon (sP) (Puelles and Rubenstein, 2003). The p1 to p3 correspond, respectively, to pretectum, thalamus (formerly dorsal thalamus), and prethalamus (formerly ventral thalamus) in the alar plate. These three prosomeres form the diencephalon. The sP is formed by telencephalon (dorsal) and hypothalamus (ventral). These territories can be delimited according to visible morphological landmarks (i.e., posterior commissure in caudal p1, retroflex fascicle in the p1/p2 limit, zona limitans in the p2/p3 limit, mammillary body in caudoventral sP). We have also included data on parabrachial and parabigeminal nuclei (belonging, respectively, to rhombomere 1 and isthmus [Is]) given their TH expression and closeness to midbrain-positive groups.
Early Stages (10–12 days post coitum): TH Expression Covering Two Ample Regions
In 10 and 10.5 days post coitum (dpc) embryos (Fig. 1A–F), two positive regions were observed. The first occupied the lateral wall of the forebrain (alar plate) with strong signal in p3. It extended rostrally into the secondary prosencephalon (sP). The labeling appeared in cells of the mantle layer. The expression in rostral sP appeared in ventral telencephalon, that is, the primordium of the future preoptic region and ganglionic eminences (arrows in Fig. 1B,D–F), whereas the cortical region and the ventral hypothalamus remained negative. A second caudal patch of signal occupied the floor plate (fp) of the rostralmost midbrain and diencephalon (p1–p3) up to the mammillary recess (m; fp in Fig. 1A,C).
At 11–12 dpc, the two same domains were observed (Fig. 1G–Q). The rostral alar domain was now more intense and appeared as a continuous positive region from alar p2, p3, and extending into sP to reach the primordia medial and lateral ganglionic eminences (mge, lge; arrows in Fig. 1G–J,N–Q). This domain showed a ventralward extension into the mammillary and tuberomammillary regions (basal plate of sP; arrowheads in Fig. 1I,J). The floor-related domain (midbrain to p3) expanded laterally into the mantle layer of the mes–diencephalic basal plate; this process occurs first rostrally (fp in Fig. 1G,H,K–M) and leads to closeness with the dorsal/rostral forebrain domain (arrows in Fig. 1G,H). Expression reaching the ventricular surface was still visible at 11 dpc (fp in Fig. 1G,H). In 12 dpc embryos, the labeling of the floor plate had disappeared from the proliferative zone and moved to the mantle layer (compare Fig. 1C and 1K). This ventral labeling extended caudalward in relation to previous stages, reaching down the isthmic region (compare Fig. 1K,L with 1A,C). At these stages, a third group of TH-expressing cells was observed to form a rostrocaudal gradient of positive cells in the tectum (Mb in Fig. 1P).
Late Stages (12.5–13.5 dpc): TH Expression Organized in the Anlagen of Individual Dopaminergic Nuclei
Rostral forebrain groups
In 12.5–13.5 dpc embryos, the earlier TH mRNA-labeled groups became progressively subdivided into discrete formations corresponding to classic catecholaminergic groups. The alar forebrain domain has given rise to group A13 in the prethalamus (p3), group A14 in the posterior entopeduncular (pep) and posterior anterior hypothalamus (ahp) areas (into sP) and group A15 distributed into the anterior preoptic region (poa) and anterior endopeduncular area (aep). Additional labeling corresponded to the septum (se) and the medial and lateral ganglionic eminences (mge, lge; Figs. 2B,E,G,I,K,M, 3D–F). In these two structures, mRNA expression appeared close to the subventricular zone (sv), as well as in a thin external layer (el; Fig. 2G,I). Comparison with Mash1 mRNA expression (marker for proliferative cells in ventricular and subventricular zones of the basal ganglia; Yun et al., 2002) showed that Mash1- and TH-positive cells were separate populations (Fig. 4A–D). TH mRNA expression in the ganglionic eminences appears thus in postmitotic cells. The expression in the external layer corresponds probably to the developing claustrum, a stream of tangentially migrated cells from the cortex (Puelles et al., 2004). Although all of these mentioned positive groups still keep contact between them, there is a significant disaggregation from the previously observed region (compare Fig. 2H–O with 1J).
A13-related expression appeared as classically described in the zona incerta (zi); however, TH mRNA was also expressed dorsally with a lower intensity in the reticular thalamic region (tr), covering thus a great portion of alar p3 (Figs. 2L–O, 3D,E). The expression in zi and tr were also differentiated because zi expression appeared focused in the subventricular zone, whereas tr expression covered homogenously the full mantle layer (Fig. 2L).
In the basal plate of sP, expression was observed in a ventral portion of A14 in the dorsomedial hypothalamus area, located in the subventricular zone (A14-dm in Fig. 2O,P). Another positive region appeared in the rostral sP basal plate, corresponding to the anlagen of group A12 or arcuate nucleus (Fig. 2Q). These basal plate nuclei could either have an in situ origin or alternatively derive from a ventralward migration of the expanded alar forebrain domain (as shown in Fig. 1J, arrows).
Midbrain and caudal forebrain groups
The ventral group of the midbrain and p1–3 neuromeres showed an incipient mediolateral subdivision into the retrorubral nucleus (A8, in caudal midbrain), ventral tegmental area (A10, from the fovea isthmi to the mammillary recess), and the substantia nigra (A9, more laterally form midbrain to p3). According to the placement of the isthmic fovea, which belongs to the isthmus, this region extends caudalward from midbrain into this segment. As described for earlier stages, a progressive extension occurs into caudal territories of the A8-9-10 anlagen (Fig. 3A,B).
A11 was represented by positive groups in a middle dorsoventral position extending longitudinally from midbrain to p2 (Fig. 2F). Additionally, in 12.5–13 dpc embryos, we found a transversally oriented group following the retroflex fascicle (fr), at the p1/p2 boundary zone, that we call here dorsal A11 (A11d in Fig. 2C and data not shown).
The cell labeling in the tectum appeared now in its caudalmost end (inferior colliculus; co in Fig. 3E,F). Comparison with earlier stages (Fig. 1P) points to a rostral to caudal expression wave, possibly following the rostrocaudal tectal differentiation pattern. Cells of the oculomotor and trochlear nucleus also express weak signal (III and IV in Fig. 3C). A novel positive group was present in the isthmic mantle plate, formed by the anlagen of parabrachial (pb) and parabigeminal (pg) nuclei (Fig. 3C–F).
Our results on ISH with a TH-RNA probe show that the catecholaminergic nuclei of the midbrain and forebrain have a multineuromeric origin, as previously described immunohistochemically in the chick, human, and mouse embryos (Puelles and Medina, 1994; Puelles and Verney, 1998; Vitalis et al., 2000). As commented in these papers, this pattern suggests the possibility that segmental subdivisions of each previously described nucleus (substantia nigra, ventral tegmental area, and so on) may reveal functional or histochemical peculiarities. In fact, the earliest TH-expressing regions are ample, continuous zones across several segments, which become restricted at later stages as the definitive catecholaminergic groups emerge.
The expression pattern in specific longitudinal regions (portions of the floor and basal plates, longitudinal band in the alar forebrain, alar midbrain, and isthmus) suggests multiple induction of TH expression caused by diverse morphogenetic molecules with profiles of expression or biologic action comparable to the observed topographic diversity. For example, it has been described that ventral Shh and isthmic Fgf8 are involved in the induction of dopaminergic groups (Ye et al., 1998). However, we report the onset of ventral midbrain TH expression at rostral levels (Fig. 1A,C) that is, far away from the isthmic Fgf8 expression. This pattern would agree with recent reports on zebrafish that discard an involvement of isthmic Fgf8 in the induction of dopaminergic phenotype (Holzschuh et al., 2003).On the other hand, the novel region we describe in the ganglionic eminences is close to the telencephalic expression of Shh and the Fgf8 expression in the anterior neural ridge (Shimamura and Rubenstein, 1997). The TH expression pattern we describe in alar forebrain (A11, A13, A14, and A15 groups plus the ganglionic eminences) coincides with the expression of Dlx1/2 (Bulfone et al., 1993). Accordingly, these genes are involved in the induction of dopaminergic phenotype, at least in A13 (Andrews et al., 2003).
The TH-mRNA expression profile as described here emphasizes the importance of previously less visible TH-expressing regions, such as the ganglionic eminences, the tectal midbrain, the strong and broad expression in A13 in p3, the dorsalward extension of A11, and the early expression in the midbrain/diencephalon floor plate. Other novel positive sites include the floor plate of given rhombomeres and the spinal cord motor column (data not shown). The novelty of these results in relation to previous studies could be explained by a lack of sensitivity of existing antibodies. In fact, improved techniques (using ameliorated fixation procedures and antibodies) yield TH expression at earlier stages than those previously described (di Porzio et al., 1990; Puelles and Verney, 1998). Nevertheless, these improved immunohistochemical analyses do not identify novel “ectopic” positive regions as our ISH study does. The absence of detectable TH protein in these regions could be more likely explained by posttranscriptional regulation of RNA stability or translation. These mechanisms are known to be involved in the regulation of TH expression (Vyas et al., 1990; Kumer and Vrana, 1996).
In addition to immunohistochemistry and ISH to reveal TH localization, another approach that has provided information has been the use of transgenic mice carrying reporter genes coupled to 5′ TH promoter region (Son et al., 1996; Schimmel et al., 1999). Of interest, these authors also observed broad expression areas in early embryos, the pattern closely resembling our observations, including strong expression in the ganglionic eminences, the tectal midbrain, and the pretectum (our dorsal A11; i.e., Fig. 1D of Son et al., 1996, and Fig. 3B of Schimmel et al., 1999).
Most of these novel positive regions are transitory, as we have observed with the floor plate (expressed at 10–11 dpc stages), the pretectal A11d (at 12.5–13 dpc), and the aforementioned spinal cord motor column (at 10–10.5 dpc). Transient expression of TH protein in given nuclei through development is a known phenomenon (Vitalis et al., 2000; Trigueiros-Cunha et al., 2003; and references therein). Concerning other ectopic sites as ganglionic eminences, colliculi, and parabrachial/parabigeminal nuclei, TH mRNA expression is present until 14.5 dpc stage, while analysis of 17.5 dpc brains yielded no expression in the later ectopic regions (data not shown). However, it remains to be assessed whether they express functional TH enzyme leading to L-dopa synthesis and whether this synthetic machinery has a developmental role. In this respect, it is interesting to note that specific TH knockout in dopaminergic neurons, in addition to alterations in movement and feeding, leads to developmental abnormalities such as reduced brain size and neurochemical changes in the striatum (Zhou and Palmiter, 1995). Additionally, recent works show a developmental role for dopamine as regulator of precursor cell proliferation (Ohtani et al., 2003; Hoglinger et al., 2004). It would remain to be tested whether dopamine is also involved in early neural regionalization as a possible morphogen or growth factor, as has been proposed for other neurotransmitters (Weiss et al., 1998).
Of interest, embryonic ganglionic eminences express mRNA for another molecule involved in the dopaminergic presynaptic phenotype, the vesicular monoamine transporter 2 (VMAT2; Smits et al., 2003). Putative early expression of other dopaminergic markers in these structures could be addressed in further studies.
Our study could have evolutionary implications, because this novel TH mRNA expression pattern may be present in other species. This analysis could be applied thus to comparative studies of forebrain regionalization (Brox et al., 2004). An interesting point would be to consider whether TH protein-positive regions of ventral telencephalon reported in other phylla (i.e., in amphibia, Beltramo et al., 1998; or in teleostea, McLean and Fetcho, 2004) are in fact homologous structures of mammalian ganglionic eminences.
Finally, we wish to point out the importance of in vivo TH mRNA expression by the embryonic ganglionic eminences in relation to experimental therapeutic approaches for Parkinson disease (Mallet, 1996). Cultured cells from embryonic and adult striatum can express TH mRNA and protein in given conditions (Daadi and Weiss, 1999). Here, we show an in vivo correlation for this capacity, that is, TH mRNA expression by the lateral ganglionic eminence, which is the anlagen of the striatum, indicating that this structure as an eventual source for potentially dopaminergic cells.
All experiments were done in accordance with the European Communities Council Directive (86/609/EEC) and the ethical guidelines on animal experiments of our institutions. Pregnant C57BL/6J mice were killed by cervical dislocation at 10 to 14.5 dpc. Embryos were staged according to Theiler (1989). The whole embryos or the head region were fixed for 2–3 days in phosphate buffered 4% paraformaldehyde at 4°C. They were dehydrated in methanol series (25–100%) and included in paraffin blocks. These paraffin blocks were cut serially into 10-μm sections in horizontal, transversal, or in sagittal planes.
The sections were rehydrated in 100–25% ethanol series. They were treated with proteinase K (20 μg/ml) in phosphate buffered saline for 5 min at ambient temperature and then post-fixed in 4% paraformaldehyde. This was followed by an acetylating step with acetic anhydride in triethanolamine buffer, dehydrated through an ethanol series, and dried by evaporation.
Rat TH RNA antisense probe (Blanchard et al., 1994) was prepared by linearizing 1.5 kb of TH cDNA (in pSPT18 plasmid) with HindIII. The cDNA was in vitro transcribed (Promega kit) using T7 polymerase and radioactively labeled) with 35S-UTP. The insert size was reduced by Na+ carbonate treatment, purified by phenol:chloroform treatment, and precipitated with ethanol. RNA probe containing 100 × 103 cpm/μl was used for all the experiments.
Hybridization was performed at 50°C for 14 hr, with the probe diluted in hybridization buffer containing 57% formamide, 340 mM NaCl, 20 mM Tris HCl pH 7.4, 5 mM ethylenediaminetetraacetic acid, 11% dextran sulfate, Denhardt ×1, 500 μg/ml yeast RNA, and 10 mM NaH2PO4 pH 8.0. Posthybridization steps included successive washings in formamide/standard saline citrate (SSC) 2× (1:1) at 50°C for 20 min, followed by RNAse A (100 μg/ml in SSC ×2) treatment at 37°C for 30 min and finally in SSC ×2 at room temperature for 20 hr. The sections were then dehydrated in ethanol series containing 300 mM NH4 acetate. They were brought to 100% ethanol and subsequently dried by evaporation.
For microscopic analysis the slides were soaked in Kodak NTB2 photographic emulsion (diluted 1:1 in H2O), dried by evaporation, and exposed for 30 days at 4°C. After development of the emulsion and fixation, the sections were lightly stained with cresyl violet, dehydrated, and mounted using Eukitt. The emulsion-coated sections were analyzed in darkfield for in situ TH labeling and in brightfield for cresyl violet staining.
Double ISH for detection of Mash1 and TH mRNA expression was performed using Vibratome 50-μm sections from 12.5 dpc embryos, as described in Nieto et al. (1996). Mash1 cDNA was obtained from the I.M.A.G.E. consortium.
We thank M.A. Nieto for critical reading of this manuscript and lab facilities.