Transcriptional control of midbrain dopaminergic neuron development

The midbrain dopaminergic (mDA) neurons (neurons that use dopamine as a neurotransmitter) constitute about 75% of dopaminergic neurons in the adult brain (Wallen and Perlmann,2003). The midbrain dopamine system has been intensively studied because of its implication in diverse psychiatric and neurological disorders(Hornykiewicz, 1978; Carlsson, 2001). In addition,midbrain dopaminergic neurons play key roles in the generation of pleasure,and in the development of addictive behaviours such as drug abuse(Chao and Nestler, 2004).

Our current understanding of the location of dopaminergic (DA) neurons in the central nervous system (CNS) comes from immunohistochemical studies that determine the expression pattern of the enzyme tyrosine hydroxylase (Th),which is the rate-limiting enzyme of dopamine synthesis. Th⁺ mDA neurons (historically called A8-A10) originate from the ventral part of a domain of the brain that extends rostrally to the ventral thalamus/hypothalamus border and caudally to the midbrain/hindbrain border(Fig. 1)(Bjorklund and Lindvall, 1984; Dahlstrom and Fuxe, 1964; Marin et al., 2005). These neurons can be anatomically divided into three main subgroups. The lateral A9 group corresponds to neurons of the substantia nigra pars compacta (SNpc),which mainly project into the dorsal striatum and form the nigro-striatal pathway. This group of neurons is mainly involved in the control of movement,as revealed by the resting tremor, rigidity, bradykinesia (abnormal slowing of voluntary movement) and gait disturbance that is seen in individuals with Parkinson's disease. These symptoms are due to the specific degeneration of the SNpc neurons (Lang and Lozano,1998). The medial A10 and lateral A8 neuron groups define the ventral tegmental area (VTA) and retrorubal field of the midbrain,respectively. A10 and A8 groups project to the ventromedial striatum, as part of the mesocortical limbic system that is involved in emotional behaviour and mechanisms of reward (Dahlstrom and Fuxe,1964; Tzschentke and Schmidt,2000). Dysregulation of DA transmission in the limbic system has been linked to the development of drug addition(Kelley and Berridge, 2002; Wightman and Robinson, 2002)and depression (Dailly et al.,2004), and is thought to contribute to the psychotic symptoms of schizoprenia (Sesack and Carr,2002).

As for all neurons, the generation of mDA neurons from a neural progenitor cell can be divided into distinct steps. At least three steps have been recognized based on the expression of molecular markers: (1) regional specification, (2) early and (3) late differentiation. Until recently, mDA progenitors were poorly defined and not easily distinguishable from other CNS progenitors. The recent discovery that certain transcription factors,specifically Otx2, Lmx1a and Lmx1b (Lmx1a/b), Engrailed 1 and Engrailed 2(En1/2), Msx1 and Msx2 (Msx1/2), Neurogenin 2 (Ngn2) and Mash1, are all expressed in mDA progenitors has allowed these progenitors to be identified for the first time by a combinatorial transcription factor code(Simon et al., 2001; Puelles et al., 2003; Puelles et al., 2004; Andersson et al., 2006a; Andersson et al., 2006b; Kele et al., 2006). These new transcription factors provide additional markers with which to address the issue of heterogeneity among mDA progenitors, and are candidate regulators for promoting the specification and differentiation of progenitors into mDA neurons. Important insights into the role of these transcription factors in regulating the development of mDA neurons, which have arisen from loss- and gain-of-function experiments in chick and mice, will be discussed in this review. These new findings will underpin exciting research both in the stem cell and developmental biology fields during the next few years.

Molecular markers allow the distinction of three sequential cell populations in the mDA lineage: progenitors, immature neurons and mature neurons (see Fig. 2A,B). These populations are generated during successive developmental steps, which are called regional specification, early and late differentiation (see below). Until recently, the only molecule known to mark specifically DA progenitors in the midbrain was retinaldehyde dehydrogenase Aldh1a1 (Raldh1). Raldh1 catalyzes the oxidation of retinaldehyde into retinoic acid; however, the role of Raldh1 in mDA progentiors remains to be discovered. Remarkably, during the past two years several groups have shown that other transcription factors,such as Otx2, Lmx1a/b, En1/2, Msx1/2, Ngn2 and Mash1, are also expressed in these progenitors (Simon et al.,2001; Puelles et al.,2003; Puelles et al.,2004; Andersson et al.,2006a; Kele et al.,2006; Andersson et al.,2006b).

Despite this historical lack of information on molecular determinants of mDA progenitors, several key transcription factors have been identified and described that mark and regulate the development of postmitotic mDA neurons. For example, the orphan nuclear receptor Nurr1 was found to be required not for the generation, but for the maintenance of mDA neurons(Zetterstrom et al., 1997;Saucedo-Cadenas et al., 1998). It is also required for the expression of Th,vesicular monoamine transporter 2 (Vmat2), dopamine transporter (DAT) and Ret in mDA neurons (Wallen et al.,1999; Wallen et al.,2001; Smits et al.,2003). By contrast, En1/2 homeodomain proteins are required, in part, for the generation of mDA neurons and their survival(Simon et al., 2001). En1/2 proteins have also been shown in vitro to be required cell autonomously in mDA neurons for neuronal survival through their regulation of apoptosis(Alberi et al., 2004). Pitx3 is a paired homeodomain transcription factor that is required for the expression of Th in a subset of mDA neurons (Maxwell et al., 2005), and for the survival of primarily SNpc, but also VTA, neurons (Hwang et al.,2003; Nunes et al.,2003; Smidt et al.,2004; van den Munckhof et al.,2003).

Presently, it is still unknown whether mDA progenitors represent a homogenous or a heterogeneous pool of progenitors. The specific vulnerability of SNpc DA neurons to toxic insults when compared with VTA DA neurons, as is evident in Parkinson's disease, indicates that these neurons are heterogeneous. The earliest molecular marker that distinguishes subpopulations of mDA neurons in mice is Pitx3. By using GFP to trace the lineage of Pitx3-expressing cells in Pitx3 knock-in GFP mouse embryos, two partially overlapping groups of mDA neurons can be distinguished by their temporal profile of expression of Th and Pitx3(Maxwell et al., 2005). Ventrolateral mDA neurons express Pitx3 prior to expressing Th, whereas the dorsomedial mDA neurons express Th ahead of Pitx3 at E12.5. By E13.5, all Th⁺ cells express Pitx3. Although this study indicates that mDA neurons are a heterogenous population transiently at E12.5, it remains to be determined whether their heterogenous connectivity patterns and resistance to toxicity are already specified at the progenitor stage, or, alternatively,only later in postmitotic neurons.

Studies using rat and chick neural plate explants have demonstrated that the floor plate-derived signal sonic hedgehog (Shh) and fibroblast growth factor 8 (Fgf8) from the mid-hindbrain boundary are required for the induction of DA neurons before E9.5 (Ye et al.,1998). In addition, Wnt1(Prakash et al., 2006) and transforming growth factor-β (Tgfβ)(Farkas et al., 2003) probably also play a role in the patterning of midbrain progenitors. These signals specify anteroposterior (AP) and dorsoventral (DV) identity, and result in the activation of a combination of transcription factors, including Otx2, Lmx1a/b,En1/2, Msx1/2, Ngn2 and Mash1, in a temporal sequence(Fig. 2C). The expression of Otx2, Lmx1b and En1/2 genes is already initiated by E9.0(Ang et al., 1994; Smidt et al., 2000). Subsequently, Lmx1a and Msx1/2 expression turn on around E9.5, and Ngn2 and Mash1 expression are not initiated until E10.75(Andersson et al., 2006b). The molecular mechanisms leading to the sequential activation of these genes is not understood. Shh can induce Lmx1a and Msx1/2 expression in neural tube explants of chick embryos (Andersson et al., 2006b). However, these transcription factors are activated endogenously in mouse embryos one day later than the initiation of Shh expression (Echelard et al.,1993). These results suggest that Shh signalling induces another signal or factor that is required for the expression of Lmx1a and Msx1/2. Based on their ventral midline position, mDA progenitors are presumed to be derived from Shh⁺ glial-like floor plate cells. Thus, another step in the specification of mDA progenitors requires the acquistion of neuronal potential by floor plate progenitors, a step that is likely to involve the activity of the proneural genes Ngn2 and Mash1(Andersson et al., 2006a, Andersson et al., 2006b; Kele et al., 2006).

Birthdating studies using tritiated thymidine incorporation demonstrate that mDA progenitors exit the cell cycle and generate postmitotic immature mDA neurons between E9.5 and E13.5 in mice(Bayer et al., 1995). Immature mDA neurons initiate Nurr1 expression(Zetterstrom et al., 1997) and En1/2 expression (Simon et al.,2001; Alberi et al.,2004) during this differentiation step(Fig. 2B). In addition, mDA progenitors, like many other CNS progenitors, downregulate Sox2 expression while initiating the expression of general neuronal markers, such asβIII-tubulin, in postmitotic immature and mature mDA neurons(Kele et al., 2006).

From E11.0 onwards, immature mDA neurons continue to migrate radially on radial glial fibres into the mantle zone, further differentiating into mDA neurons (Kawano et al., 1995). These neurons express Pitx3, Th and aromatic amino acid decarboxylase (Aadc),in addition to the earlier markers expressed in immature mDA neurons(Fig. 2B). Ngn2, however, is not expressed in mature mDA neurons. Aadc converts DOPA into dopamine. Aadc mRNA transcripts are thought to be expressed already in immature mDA neurons (Smidt et al.,2004).

The roles of transcription factors, such as Nurr1, En1/2 and Ptx3, acting during the late differentiation step of the mDA lineage (described briefly above, see Fig. 2B and Table 1) have been extensively reviewed elsewhere (Goridis and Rohrer,2002; Riddle and Pollock,2003; Wallen and Perlmann,2003; Simeone,2005; Smits et al.,2006; Prakash and Wurst,2006). However, the roles of the transcription factors that govern the specification and early differentiation of mDA progenitors have only recently started to emerge during the past few years and will be the focus of this review (Table 1).

Otx2 is a bicoid class homeodomain transcription factor that is widely expressed before gastrulation, but its expression becomes progressively restricted to the anterior third of the mouse embryo after E7.75(Simeone et al., 1993; Ang et al., 1994). Within the nervous system, Otx2 expression is restricted to the forebrain and midbrain between E8.5 and E12.5. In addition to these anterior brain regions,expression is also detected in the rhombencephalon from E12.5 onwards(Mallamaci et al., 1996). Otx2 is required for the formation of the forebrain and midbrain as a result of its role in the anterior visceral endoderm, where it functions to restrict posterior fates (Mallamaci et al., 1996; Perea-Gomez et al.,2001) (reviewed by Simeone and Acampora, 2001). Subsequently, Otx2 is also required for positioning the expression of Wnt1 and Fgf8 at the midbrain boundary(Brodski et al., 2003), and for limiting the dorsal extent of Shh expression in the ventral midbrain(Puelles et al., 2003).

Otx2 also has additional roles in mDA progenitor specification and differentiation based on findings obtained from the phenotypical analyses of Otx2 mutant mice. Several different conditional Otx2 mouse mutants have been generated in an attempt to delete Otx2 specifically at different developmental stages. For example, Otx2 was deleted in the midbrain of En1cre;Otx2^flox/flox embryos from E9.5 onwards (Puelles et al.,2004). In these En1cre;Otx2^flox/flox mutant embryos, midbrain expression of Shh expands dorsally, whereas Fgf8 expression, which is normally restricted to the anterior hindbrain, shifts anteriorly into the midbrain(Puelles et al., 2004). Despite these changes in AP and DV patterning molecules, a small domain of midbrain tissue develops normally. Within this domain, expression of the homeodomain protein Nkx2.2 expands ventrally into presumptive DA progenitors around E9.5, indicating that Otx2 is required for the repression of Nkx2.2 in these progenitors(Prakash et al., 2006). Interestingly, serotonergic neurons are generated ectopically in these Otx2 conditional mutants at the expense Th⁺ mDA neurons. Loss of mDA neurons is directly linked to the abnormal expression of Nkx2.2 in mDA progenitors, as the presence of ectopic serotonergic neurons and the reduction in the number of mDA neurons are rescued in En1cre;Otx2^flox/flox;Nkx2.2^-/- embryos(Prakash et al., 2006). The ectopic expression of Nkx2.2 in En1cre;Otx2^flox/floxembryos is also associated with a loss of expression of Wnt1, which indicates that Otx2 might regulate Wnt1 expression indirectly via its repressive effects on Nkx2.2 in mDA progenitors at E12.5. Accordingly, Wnt1 expression is recovered in En1cre;Otx2^flox/flox;Nkx2.2^-/- embryos that lack both Otx2 and Nkx2.2. It is noteworthy that Wnt1 may also be required upstream of Otx2, as ectopic expression of Wnt1 under the control of the En1 promoter leads to an upregulation of Otx2 in the floor plate of the rostral hindbrain of En1^Wnt1/+ embryos(Prakash et al., 2006).

A different role for Otx2 in mDA progenitors was identified from studies of Nestin-Cre;Otx2^flox/flox embryos(Vernay et al., 2005). In these conditional mutants, loss of Otx2 protein from E10.5 onwards results in loss of expression of the proneural genes Ngn2 and Mash1 in ventral mDA progenitors. Subsequently, mDA neurons are missing at the ventral midline of the midbrain. In addition, Nkx2.2 expression expands ventrally into the ventricular zone adjacent to the red nucleus neurons that are reduced in size in Nestin-Cre;Otx2^flox/flox embryos(Vernay et al., 2005). Red nucleus neurons normally lie dorsal to mDA neurons, and are implicated in the control of locomotion (reviewed by Kennedy, 1990). This loss of red nucleus neurons is fully rescued in Nestin-Cre;Otx2^flox/flox;Nkx2.2^-/- embryos, but the loss of mDA neurons is not. Taken together, these results indicate that Otx2, presumably via regulating the expression of Ngn2 and Mash1 (see below), is also required for the generation of mDA neurons. A later role for Otx2 in regulating neurogenesis in mDA progenitors seems contradictory to the observation that mDA neuronal development is recovered in En1cre;Otx2^flox/flox;Nkx2.2^-/- mutant embryos,despite the loss of Otx2 in mDA progenitors prior to neuronal differentiation in these triple mutants at E9.5. In order to reconcile these two findings, I propose that the requirement for Otx2 in regulating Ngn2 expression in mDA progenitors may be bypassed by changes in Shh expression in the ventral midbrain of En1cre;Otx2^flox/flox;Nkx2.2^-/- mutant embryos(Prakash et al., 2006). This is because Shh can influence the expression of Ngn2, possibly via its effect on Lmx1a (see below). Alternatively, or in addition, the recovered Wnt1 expression in En1cre;Otx2^flox/flox;Nkx2.2^-/-mutants is not completely normal and this may also affect Ngn2 expression in these mutants (Prakash et al.,2006).

Lmx1a and Lmx1b, members of the family of LIM homeodomain transcription factors, play important roles in the developing brain. Lmx1b is expressed in the midbrain from E8.0 onwards (Smidt et al., 2000), but this expression becomes restricted by E9.5 to the roof plate, the mid-hindbrain boundary and the ventral midbrain, including the floor plate. By contrast, Lmx1a expression begins at E9.5 in the ventral midbrain and then progressively expands dorsally(Andersson et al., 2006b). Double antibody labelling studies have revealed that at E9.5, Lmx1b expression encompasses more cells in the ventral midbrain than does Lmx1a, but that by E10.5 the expression domains of the two genes largely coincide. Because the expression of Lmx1a in progenitors directly overlies a region where Th⁺ neurons develop at E11.5, Lmx1a expression has been proposed to mark the dorsal boundary of mDA progenitors(Andersson et al., 2006b),whereas the initial expression of Lmx1b is believed to include progenitors for other cell types as well. This interpretation seems likely but awaits confirmation by lineage- and fate-mapping studies.

Loss-of-function studies have shown that Lmx1b is required for the maintenance of Th⁺ mDA neurons. Th⁺ neurons are found in Lmx1b^-/- mutants up to E16.0, although they fail to express the paired homeodomain transcription factor Pitx3(Smidt et al., 2000). This differentiation defect results in the eventual loss of DA neurons in Lmx1b^-/- embryos. Although this phenotype has been interpreted as demonstrating a role for Lmx1b in the maintenance of DA neurons, Lmx1b is initially broadly expressed in the midbrain; and earlier defects in patterning of the mid-hindbrain region may contribute to the loss of Th⁺ neurons at later stages. In addition, Lmx1b.1 and Lmx1b.2 genes are required for maintenance of the mid-hindbrain organizer and for the expression of Wnt1, Wnt3a, Wnt10b,Pax8 and Fgf8 in zebrafish embryos(O'Hara et al., 2005). Therefore, whether the mDA phenotype of Lmx1b mutant mouse embryos is due to an intrinsic role in the mDA lineage or to earlier patterning functions in the midbrain remains to be clarified. A detailed analysis of Lmx1bfunctions in mDA progenitors and neurons will be important to elucidate its roles at different developmental stages.

Positional cloning has identified Lmx1a as being the gene affected in the spontaneous mouse mutant dreher(Millonig et al., 2000). The point mutation in Lmx1a in dreher mice results in the loss of the caudal roof plate. In addition, overexpression studies have shown that Lmx1a can induce the expression of roof plate markers and of components of roof plate signalling(Chizhikov and Millen, 2004b),thus demonstrating that Lmx1a is both required and sufficient to induce roof plate formation.

A recent breakthrough study by the groups of Ericson and Perlmann has identified Lmx1a as a crucial determinant of mDA neuron fate development (Andersson et al.,2006b). In this study, these researchers showed that the overexpression of Lmx1a in the ventral midbrain of chick embryos promoted the generation of DA neurons over that of other neuronal subtypes. The induction of DA neurons in chick embryos was preceded by a re-specification of progenitor cells, as indicated by the activation of Msx1 and the repression of Nkx6.1 expression. Importantly, the activity of Lmx1a is context dependent, as Lmx1a can only induce ectopic DA neurons in ventral, and not in dorsal, midbrain cells. This context dependence suggests a role for additional factors that are specifically expressed in ventral midbrain cells. Alternatively, Lmx1a might not be able to inhibit dorsal differentiation programs and thus cannot convert dorsal midbrain cells to a more ventral fate. It is noteworthy that in this context, the overexpression of Foxa2 and Gli in transgenic mice results in the generation of ectopic Th⁺ neurons in the dorsal midbrain (Sasaki and Hogan,1994; Hynes et al.,1997). In these transgenic embryos, ectopic DA neurons were found adjacent to dorsal sites of Shh expression. These results indicate that some dorsal midbrain progenitors are competent to acquire a mDA fate in mouse embryos at developmental stages equivalent to those of the chick experiments discussed above (Sasaki and Hogan, 1994; Hynes et al.,1997). Taken together, these results favour the hypothesis that Lmx1a alone is not sufficient to induce mDA neurons, and that it functions cooperatively with ventral factors induced by the Shh pathway(Fig. 3).

Additional support for cooperative interactions between Shh and Lmx1a has come from studies using the differentiation of embryonic stem (ES) cells. Mouse ES cells transfected with Lmx1a differentiate into DA neurons in the presence, but not in the absence, of Shh(Andersson et al., 2006b). Impressively, Lmx1a promotes Th expression in up to 80% ofβIII-tubulin⁺ neurons that were derived from ES cells. Importantly, these neurons also expressed other mDA neuron markers, such as Pitx3, Lmx1b, En1/2 and DAT. By contrast, mock-transfected ES cells produced Th⁺ neurons that were mostly GABA⁺ and Pitx3^-.

A complementary loss-of-function study by RNA interference also supports a role for Lmx1a in the generation of DA neurons in chick embryos(Andersson et al., 2006b). In these experiments, Lmx1a knockdown by siRNA electroporation resulted in a loss of DA neurons, which was not compensated for by unperturbed expression of Lmx1b. This result was surprising, as studies in mice have demonstrated a role for Lmx1b in the development of mDA neurons(Smidt et al., 2000); Lmx1b can also partially rescue roof plate formation in dreher mice (Chizhikov and Millen, 2004a). One way to reconcile these results is to propose that the two genes have overlapping roles in mDA development, with Lmx1a being perhaps more efficient at the specification step, while Lmx1b is required for later differentiation events in the DA lineage. This hypothesis is consistent with the observation that Lmx1b is much less efficient than Lmx1a at promoting mDA neuron differentiation in ES cells (Andersson et al.,2006b). It is also possible that the requirements for Lmx1a and Lmx1b genes are different for the two species. Whether Lmx1a and Lmx1b have unique and/or redundant roles in the development of mDA neurons in mice awaits further studies of Lmx1a,Lmx1a;Lmx1b double mutant embryos and of Lmx1b conditional mutants.

The mouse Msx genes, Msx1, Msx2 and Msx3, encode homeodomain transcription factors that share 98% homology in the homeodomain and function as transcriptional repressors (reviewed by Ramos and Robert, 2005). Msx1 and Msx2 are expressed in DA progenitors in the ventral midbrain (Andersson et al.,2006b), in addition to in the roof plate and adjacent cells in the dorsal neural tube and neural crest, as well as in many other sites where epithelial-mesenchymal inductive interactions occur, such as in the limbs and tooth buds, heart, branchial arches and in craniofacial processes. Msx3, by contrast, is expressed exclusively in the dorsal aspect of the neural tube in the mouse, caudal to the mid-hindbrain boundary(Shimeld et al., 1996; Wang et al., 1996).

Msx1^-/- embryos exhibit a 40% reduction in the normal number of mDA neurons, probably as a result of the downregulation of Ngn2 expression (Andersson et al.,2006b). This partial reduction of mDA neurons in Msx1mutants may be due to compensation by Msx2, a possibility that remains to be addressed by the analysis of Msx1;Msx2 double mutants. In addition, Msx1 is able, and is required, to repress Nkx6.1 expression in ventral midbrain progenitors(Andersson et al., 2006b)(Fig. 3). Premature expression of Msx1 in the midbrain in transgenic mice also leads to the precocious expression of Ngn2 and Nurr1, and to the downregulation of Shh in the floor plate, indicating that Msx1 sets the timing of mDA neuron generation possibly by inducing Ngn2 expression in ventral midbrain progenitors (Andersson et al.,2006b). Given that Msx genes normally function as repressors, Msx1 may regulate the activity of a repressor of Ngn2 in mDA progenitors.

Proneural basic helix-loop-helix genes are crucial regulators of neurogenesis and of subtype specification in many areas of the nervous system. In the ventral midbrain, the proneural genes Mash1, Ngn2 and Ngn1 show an intricate pattern of expression. Ngn2 and Mash1 are expressed in mDA progenitors, whereas Ngn1, Ngn2and Mash1 are co-localized in the ventricular zone more dorsally(Kele et al., 2006). Ngn2 is also expressed in newly born postmitotic immature mDA neurons immediately adjacent to the ventricular zone at the ventral midline. Ngn2 is required for the generation of Nurr1⁺ immature mDA neurons, and probably also for their subsequent differentiation into Th⁺ mature mDA neurons(Andersson et al., 2006a; Kele et al., 2006). Although Mash1 by itself is not required for mDA neuron development, the loss of both Mash1 and Ngn2 in Mash1;Ngn2 double mutant mouse embryos leads to a greater loss of mDA neurons than occurs in Ngn2 single mutants (Fig. 4), suggesting that Mash1 can partially compensate for the loss of Ngn2 function in mDA progenitors. Accordingly, this results in a further rescue of Th⁺ neurons(Fig. 4) in Ngn2^{KIMash1/Mash1} embryos that express Mash1under the control of the Ngn2 promoter(Kele et al., 2006).

Ngn2 has a role in regulating generic neuronal, as well as subtype-specific, differentiation programs in other parts of the CNS (reviewed by Bertrand et al., 2002). The reduced number of Th⁺ neurons in Ngn2 mutants is due to a failure in neuronal production, as suggested by the loss ofβIII-tubulin⁺ neurons, which lie directly beneath mDA progenitors (Kele et al.,2006). Moreover, the loss of two markers of proneural activity, Dll1, a Notch ligand, and Hes5, an effector of Notch signalling, in the ventral midbrain is consistent with a role for Ngn2 in regulating the generic aspect of neurogenesis(Kele et al., 2006). In other parts of the CNS, the role of Ngn2 in subtype specification has been demonstrated by the inability of other classes of proneural genes to compensate for Ngn2 activity (reviewed by Bertrand et al., 2002). Mash1 is able to compensate partially for Ngn2 function, as 60% of the normal number of mDA neurons are generated in Ngn2^{KIMash1/KIMash1} embryos. This partial compensation suggests some unique role for Ngn2 in specification of the mDA neuronal subtype. In addition, the expression of Ngn2, but not Mash1, in postmitotic DA neurons is consistent with an additional and unique role for Ngn2 in regulating later differentiation steps in immature mDA neurons. Complementary gain-of-function studies performed by electroporating Ngn2 in the dorsal midbrain of mouse embryos have demonstrated a role for Ngn2 in promoting the migration of newborn neurons from the ventricular zone to the mantle zone and in inducing expression of the general neuronal marker βIII-tubulin. However, these studies show that Ngn2 alone is insufficient to promote the ectopic expression of DA neuron markers and the generation of ectopic DA neurons(Kele et al., 2006). This finding is not surprising because Ngn2 is known to function in other parts of the CNS in a context-dependent manner, working cooperatively with other transcription factors (Bertrand et al., 2002). Further insights into the function of Ngn2 in mDA neuron development will come from the identification of its transcriptional targets and cofactors in both mDA progenitors and immature mDA neurons.

In summary, the studies described in this section have identified new transcription factors that not only better define mDA progenitors and neurons,but also regulate their specification and differentiation. In light of recent studies on the development of other DA groups in the CNS, a brief comparison of the genetic programs that govern the development of DA neurons is discussed below.

Besides the neurons in the A8 (retrorubral area), A9 (SNpc) and A10 (VTA)areas of the midbrain discussed so far, some catecholaminergic neurons in the forebrain are also dopaminergic. They are found in areas A11 to A15(diencephalic and hypothalamic groups), A16 (periglomerular cells in the olfactory bulb), and A17 (interplexiform cells in the retina; see Fig. 1). Very little is known about the specification of the DA fate in these groups, except for the A13 and A16 groups. The A13 DA neurons reside in the alar plate of neuromere segment p3. Recent studies demonstrate that A13 progenitors express homeodomain transcription factors Dlx1/2 and Pax6, and differentiate into Pax6⁺/Islet⁺/Th⁺ neurons(Andrews et al., 2003; Mastick and Andrews, 2001). In Dlx1/Dlx2 double mutant mice, neurons generated by A13 progenitors fail to express Pax6, Islet1 and Th. These results thus identify a role for Dlx1/2 in the specification of A13 DA neurons. Pax6,however, is not required for A13 progenitors to differentiate into Th⁺ neurons. By contrast, Pax6 is a molecular determinant of A16 progenitors that is required for the neurogenesis and subtype determination of olfactory bulb periglomerular A16 DA neurons. These programs also differ from the one that promotes mDA differentiation, indicating that distinct programs are involved in specifying DA differentiation at different CNS positions. In support of this hypothesis, lateral tuberal hypothalamic DA neurons in the forebrain of chick embryos express Nkx2.1 and Msx1 homeodomain proteins, and have been shown to depend on the expression of the homeodomain transcription factor Six3 in progenitors for their development(Ohyama et al., 2005). By contrast, Six3 is not expressed in A13 and mDA progenitors.

The distinct differentiation programs for CNS DA neurons highlight the importance of identifying a DA neuron by its developmental history in addition to its neurotransmitter phenotype. Crucial parameters affecting the success of a transplantation therapy for the treatment of Parkinson's disease include good graft integration and functional reinnervation (reviewed by Bjorklund and Isacson, 2002; Lindvall and Bjorklund, 2004; Snyder and Olanow, 2005). Recent findings indicate that grafted SNpc and VTA DA neurons differ in their axon projection patterns in the DA-denervated forebrain of adult mice,suggesting that mDA neuronal subtypes display distinct responses to axon guidance cues and target recognition mechanisms regulating reinnervation in the forebrain (Thompson et al.,2005). These and earlier studies(Hudson et al., 1994; Zuddas et al., 1991) indicate that the success of transplantation therapies will be strongly influenced by the type of DA neurons used in these procedures. Therefore, elucidating the molecular determinants that regulate mDA neuron differentiation from CNS progenitor cells in vivo should facilitate the generation of specific populations of DA neurons from stem cells that will be useful for transplantation therapies.

Although some of the key molecular players required in specifying neural progenitors towards a mDA differentiation program are now known, their precise functions remain to be deciphered. Gain-of-function studies suggest that the major determinants, such as Otx2, Lmx1a/b, Msx1/2, Ngn2 and Mash1, are likely to act in a combinatorial manner to promote the mDA fate, as the ectopic expression of some of these genes individually fails to promote the differentiation of dorsal midbrain progenitors into DA neurons(Kele et al., 2006; Andersson et al., 2006b). Future work will shed light on the specific combinatorial interactions of transcription factors that govern the differentiation of midbrain progenitors into mDA neurons, and will also lead to the discovery of downstream components of these regulatory networks. Besides these transcription factors, members of the forkhead/winged helix transcription factor family, Foxa1 and Foxa2, are also expressed in mDA progenitors (Puelles et al., 2003), and their role in the development of mDA neurons is currently being investigated in our laboratory. In addition, the molecules responsible for the generation of SNpc versus VTA DA neurons remain unknown. SNpc and VTA DA neurons settle in different positions along the mediolateral axis of the midbrain and have distinct axon projection patterns. A major challenge for the future is to discover the molecules that are responsible for these distinct migratory and axon growth behaviours. The ability to generate DA neurons from neural progenitor/stem cells of the SNpc subtype and to purify them based on a specific set of molecular markers will lead to significant progress in stem cell-based therapies for Parkinson's disease.

I am extremely grateful to Stella Uwabor for help with references, to Anna Ferri and Wai Han Yau for help with illustrations, and to members of the lab for critical reading of this manuscript. Work in my laboratory is funded by Parkinson's Disease Society, UK and the Medical Research Council.