Mammalian mDN axonal projections course rostrally towards forebrain structures that are expanded in both the dorsoventral and mediolateral axes. The most medial and ventral mDN projections, within the VTA, regulate reward-associated behavior and novelty learning and project rostrally on a ventromedial course to the nucleus accumbens and to extrastriatal structures including the pallidum and subthalamus. In contrast, more lateral mDN axons originate within the SN and take a relatively dorsolateral course on their rostral traverse (Figure 1A). Although several factors, including Slit-1 and -2 [3], Netrins [4], Ephrins [5] and Semaphorins [6], have been implicated in mDN pathfinding, the molecular basis of the diversity of mDN axonal projections is unknown.

To investigate a role for Shh as a chemoattractant for mDN axons, we established an explant culture model in which E11.5 midbrain tissue was maintained next to an aggregate of HEK293 cells transfected with full-length Shh cDNA or control vector for 3 days (Figure 2A). Whereas control vector-transfected HEK293 had no effect, cells expressing Shh promoted unidirectional outgrowth toward the Shh source of TH-positive dopaminergic axons (Figure 2B). If Shh is a chemoattractant for TH-positive neurons, we would predict that ventral midline tissue rostral to the explant, which expresses Shh (Figure 1A, 1C), should also attract mDN axons. Consistent with this, ventral (but not dorsal) tissue was effective at attracting mDN axons (Figure 2C). Cyclopamine, a specific antagonist of Smo function, partially blocked the chemoattractant activity of ventral tissue, indicating a role for Shh as a ventral guidance cue for mDN axons. The remaining attractant activity of the ventral tissue is likely due to an additional, unidentified chemoattractant expressed in the rostral ventral areas. Shh did not alter the overall outgrowth of TH-positive axons from the explant, but rather promoted unidirectional outgrowth towards the chemoattractant (Figure 2B). As residual endogenous Shh may be present in the midbrain explants, the effects we observe may underestimate the role of Shh. E11.5 midbrain explants harbor mature mDNs, and consistent with this, Shh did not alter the number or specification of mDNs in the explants (data not shown).

To investigate the function of Shh in mDN axonal pathfinding in the intact CNS, we took advantage of conditional mutant mice in which the Shh receptor Smo was inactivated by a transgenic Cre recombinase regulated by the Nestin promoter (Nestin-Smo cko mice). In these mice Cre-mediated recombination leads to complete inactivation of the Smo allele in the central nervous system by E11.5 [11], [14]. Analysis of lateral dopaminergic fibers that project rostrally to the striatum in the Nestin-Smo cko mice at E13.5 revealed normal projections (Figures 3B–3C). In contrast, the medial dopaminergic projections that normally course ventromedially were defective in these mice (Figure 3D, 3E). This was particularly evident in the medial fibers that course most ventrally to extrastriatal targets including the pallidum and the subthalamus, as confirmed by immunostaining for the pallidum precursor marker Nkx2.1 (Figure 4A). Similar to the defects observed at E13.5 in Nestin-Smo cko mice, the most medial projections remained defective at a later developmental time point, E15.5 (data not shown). These findings suggest a role for Shh signaling in the ventral targeting of medial dopaminergic axonal projections.

Prior analyses and our expression studies argue strongly against the possibility that the defective mDN axonal targeting in Nestin-Smo cko mice is an indirect effect of altered neuron specification of mDNs or target tissue. Patterning and specification of the embryonic ventral midbrain, including mDNs [11], as well as embryonic telencephalic target tissue such as the striatum [14], appear unaltered in Nestin-Smo cko mice at the time period (E12.5–E16.5) during which the mDN axonal projections are formed (Figure 3A and Figure S2). The total density of mDN axonal projections appears unaltered in the mutant CNS (data not shown). Genetic fate mapping and conditional knockout analyses have demonstrated that canonical Shh signaling is no longer required in mDNs after E10 [11], [13]. Consistent with this, gene expression of several markers for hypothalamic and forebrain target tissue, including FoxA2, Islet 1, and Shh, appear normal in E12.5 Nestin-Smo cko mice mice (Figure 4B). Expression of Netrin1, a chemoattractant present in the ventral midline, is similarly unaltered in the Nestin-Smo cko mice (Figure 4C).

We hypothesized that lateral mDN axons, such as those that ascend to the striatum, are intrinsically less sensitive to Shh than the medial projections, and are thus less affected in the Nestin-Smo cko mice. To test this, midbrain explants were divided into lateral and medial halves and then tested for chemoattraction to a Shh source (as described above). Axonal projections from medial (but not lateral) midbrain explants displayed unidirectional outgrowth towards the Shh point source (Figure 5, Figure S3). The molecular basis of this differential sensitivity remains to be determined; since the Shh receptor component Smo appears to be expressed evenly in both, medial and lateral mDN (data not shown).

Shh expression starts around E7.5 in the notochord and prechordal mesoderm that underlies the medial neural plate and subsequently induces Shh expression in the ventral midline along the entire neuroaxis of the neural tube. In the midbrain, Shh signaling from the ventral midline is essential for ventral patterning, including the induction of mDNs, but only prior to E10 [10], [11]. Shh expression is maintained in the ventral-medial midbrain and hypothalamus for the subsequent days of embryonic development (Figure 1 and S.B. unpublished data), after mDN induction and differentiation are completed and during the period of mDN axonal outgrowth. Shh in the ventral hypothalamus can therefore act as a local guidance cue for medially projecting dopaminergic axons (Figure 1A).

Several factors have been implicated in mDN axonal pathfinding, and thus we have sought to address the possibility that Shh functions indirectly through these factors to alter mDN pathfinding. Prior studies have presented evidence of a role for Slit-1 and Slit-2 in the chemorepulsion of mDN axons at the midline. Slit2 has been shown to be a chemorepellant for mDNs in explant cultures, and Robo1 and 2, Slit receptors appear to be expressed in mDNs [4]. Furthermore, Shh expression is essential for the expression of Slit1 in the CNS [15]. However, it is unlikely that the phenotype we observe in the absence of Shh signaling is a consequence of deficient Slit expression, as, in fact, mice deficient in Slit1 and Slit2 display the opposite phenotype: ventrally displaced ascending dopaminergic fibers, with a relative expansion of the medial fibers that course ventrally [3]. We therefore hypothesized that Slit signaling may serve to antagonize Shh at the ventral midline and thus refine the axonal targeting of mDN axonal projections. Consistent with this model, Slit2 expression along with Shh in HEK293 effectively suppressed the chemoattractive activity of Shh towards mDN axons in explant cultures (Figure S1).

In vitro studies additionally suggest a role for Netrin signaling in mDN axonal chemoattraction [4]. It is unlikely that Shh functions upstream of Netrin signaling to alter mDN axonal pathfinding, as Netrin expression appears unaltered in the conditional Smo mutant mice (Figure 4C). We hypothesize that Netrin signaling may serve to refine the synaptic targeting of dopaminergic axons at forebrain regions.

Our data indicate that Shh functions through the Smo receptor as a ventral midline cue for ascending dopaminergic fibers. Furthermore, we find that the other Shh receptor component, Ptc is expressed in the ventral midbrain at this point in development. However, other canonical Shh signaling components, such as Gli1, Gli2, and Gli3, are absent, suggesting a non-canonical pathway. Interestingly, a recent paper demonstrated that Shh guides commissural axons in the spinal cord by stimulating the activity of Src family kinases in a Smo-dependent manner [16]. While this study would indicate a primary role for Smo, prior studies have indicated an additional role for the novel Shh receptors, Boc and Hip, in Shh mediated axonal chemoattraction [17], [9]. It will be of interest to investigate a possible role for these kinases and receptors in mDN axonal targeting.

Further studies are also needed to identify the molecular basis for the differential sensitivity of medial and lateral mDN populations to Shh chemoattraction. We hypothesize that unidentified noncanonical signaling components may be differentially expressed in these cell populations. Alternatively, chemorepulsive signaling pathway components that antagonize the Shh signal, such as Slit pathway components, may be differentially expressed on these cell populations. One limitation to determining differential expression of these signaling components in lateral and medial mDN populations during embryogenesis is the lack of specific markers for embryonic mDN subpopulations.

Figure S1

Figure S2

Figure S3