The molecular mechanisms controlling neuronal migration have many similarities with those described for axon guidance. For instance, migrating neurons and growing axons are instructed toward their final destination by the same guidance molecules and are able to adapt their response to those cues by modulating the expression of guidance receptors. Transcriptional regulation is thought to be a key determinant in this later process, although we are just beginning to identify the contribution of these mechanisms in neuronal migration. In this review, we will describe recent progress made in understanding the contribution of transcription factors in controlling neuronal migration in the developing mouse brain, with a special focus on the developing telencephalon.
In the developing brain, newborn neurons migrate away from the germinal zones to populate specific locations within the neural tube (Hatten 1999). Migrating neurons can either move along the radial axis, interacting with the radial glial scaffold, or disperse tangentially in a plane orthogonal to radial glial cells along the rostrocaudal or mediolateral axis (Hatten 2002). Migration of neurons to their corresponding target territories is subjected to several levels of regulation, ranging from the selective action of environmental guidance cues to the establishment of neuronal interactions with specific substrates (Marín and Rubenstein 2003). The molecular mechanisms controlling neuronal migration have multiple similarities with those described for axon guidance. For instance, migrating neurons and growing axons are instructed toward their final destination by the same guidance molecules (Brose and Tessier-Lavigne 2000; Bagri and Tessier-Lavigne 2002). Similarly, axons and migrating neurons are conferred with the ability to respond to distinct cues by the differential expression of specific guidance receptors (Dickson 2002). Transcriptional regulation is thought to be a key determinant in this process, as it ultimately represents one of the fundamental mechanisms controlling the repertoire of receptors expressed by each neuron. Whereas much work has been done to decipher the contribution of transcriptional regulation in axon guidance (Butler and Tear 2007; Polleux et al. 2007), we are just beginning to identify the contribution of equivalent mechanisms in regulating neuronal migration.
Transcriptional Control of Axon Guidance: A Model for Neuronal Migration?
During development, axons are instructed to navigate to their corresponding target areas by simultaneously integrating multiple extracellular signals along the pathway (Tessier-Lavigne and Goodman 1996). The precise guidance of each neuronal population is achieved by the selective perception of environmental cues; migrating axons continuously adapt their response by modulating the expression of guidance receptors and their intracellular signaling cascades. A key mechanism regulating selectivity in axon guidance decisions is transcriptional regulation in postmitotic neurons (Shirasaki and Pfaff 2002; Butler and Tear 2007; Polleux et al. 2007). In the vertebrate spinal cord, for example, the combinatorial activities of different LIM-homeodomain (HD) transcription factors confer motor neuron subtypes with the ability to select distinct axonal pathways and final targets (Sharma et al. 1998, 2000; Kania et al. 2000; Thaler et al. 2004; Shirasaki et al. 2006). Studies over the past few years have begun to identify possible downstream guidance effectors regulated by these transcription factors. In spinal motor neurons, Lim1 determines the trajectory of motor axons emerging from the lateral aspect of the lateral motor column (LMCl) by promoting the expression of the EphA4 receptor in these cells. This allows LMCl neurons to select a dorsal trajectory avoiding the ventral limb, which expresses the chemorepellent factor EphrinA5 (Kania and Jessell 2003). Similar mechanisms appear to control the expression of guidance receptors in other axonal tracts, such as the thalamocortical and retinal projections (Shirasaki and Pfaff 2002; Butler and Tear 2007; Polleux et al. 2007). In the retina, for instance, the zinc finger transcription factor Zic2 determines the ipsilateral projection of ventrotemporal retinal ganglion cells by regulating the expression of EphB1, the receptor for the chemorepulsive cue EphrinB2 expressed in the optic chiasm (Williams et al. 2004; García-Frigola et al. 2008; Lee et al. 2008). Although it remains to be elucidated whether transcription factors directly or indirectly regulate the expression of guidance receptors in all these processes, recent evidence in spinal commissural neurons strongly suggests that direct transcriptional regulation operates during axon guidance (Wilson et al. 2008).
Transcriptional Control of Neuronal Migration: The Journey Starts in the Hindbrain
The hindbrain is a suitable system to study the molecular mechanisms controlling tangential migration due to the extensive and stereotyped translocations performed by specific classes of motor neurons. For instance, branchiomotor neurons of the facial nerve (FBM) are born in rhombomere 4 (r4) and migrate tangentially caudal and parallel to the floor plate through r5 and into r6. In r6, FBM neurons migrate dorsolaterally and then radially toward the pial surface to form the facial motor nucleus (nVII) (Chandrasekhar 2004) (Fig. 1A). Several environmental cues and cell-intrinsic mechanisms control the sequential migratory steps of FBM neurons (Garel et al. 2000; Studer 2001; Müller et al. 2003; Schwarz et al. 2004; Rossel et al. 2005). Analysis of different mutant mice over the past few years has shed light into the role of specific transcription factors in controlling the selective responsiveness of FBM neurons to external cues. For example, the T-box transcription factor Tbx20, expressed by postmitotic branchiomotor and visceromotor neurons, encodes a migratory program used by several classes of motor neurons in the hindbrain (Song et al. 2006). Conditional deletion of Tbx20 in FBM neurons disrupts their migration out of r4. In the absence of this transcription factor, FBM neurons fail to express multiple components of the noncanonical wingless (Wnt)/planar cell polarity signaling pathway, which is also implicated in controlling the migration of nVII neurons in zebra fish (Jessen et al. 2002; Carreira-Barbosa et al. 2003; Song et al. 2006) (Fig. 1A). The HD Nkx6-1 and the basic helix-loop-helix (bHLH) Ebf1 transcription factors also control the tangential migration of FBM neurons. Although these transcription factors are expressed both by progenitors and FBM migratory neurons, they seem to be exclusively required for the caudal translocation of these neurons (Garel et al. 2000; Müller et al. 2003). In Nkx6-1−/− embryos, FBM neurons do not migrate caudally out of r4 and prematurely express the Netrin-1 receptor Unc5h3 and the glial cell line-derived neurotrophic factor (GDNF) receptor Ret, normally present in r5/r6 FBM neurons (Garel et al. 2000; Müller et al. 2003) (Fig. 1A). Similarly, a subset of FBM neurons fails to migrate caudally into r6 in the absence of Ebf1 function, undergoing a premature dorsolateral migration while still in r5. This migratory defect correlates with a premature downregulation of the cell-adhesion molecule TAG-1 and with the ectopic expression of the GDNF receptor Ret and the cell-adhesion molecule cadherin8 in FBM neurons deficient for Ebf1 (Garel et al. 2000) (Fig. 1A). Altogether, these results suggest that Nkx6-1 and Ebf1 regulate the timely expression of cell-intrinsic effectors required for the tangential migration of FBM neurons. However, it is also possible that Nkx6-1 and Ebf1 might just regulate the onset of migration in FBM neurons, which in the absence of these genes would fail to leave r4 but could still progress through the differentiation program required for later migratory steps. Furthermore, because other neuronal populations in the hindbrain also express these transcription factors, it is conceivable that non–cell-autonomous mechanisms could also contribute to the migratory abnormalities described for FBM neurons. Additional work is required to elucidate the involvement of these transcription factors in the cell-autonomous regulation of FBM neurons’ migratory behavior.
Hoxa2 and Hoxb2 HD transcription factors, well known for their function in patterning the developing hindbrain (Lumsden and Krumlauf 1996), also seem to play a role in regulating neuronal migration in this region. Pontine neurons (PN) derived from r6–r8 rhombic lip progenitors migrate rostrally through several rhombomeric domains before turning ventrally to reach a stereotypic position in the brain stem (Fig. 1B). In Hoxa2−/− and Hoxb2−/− mice, subsets of PN prematurely migrate ventrally and settle at ectopic caudal positions (Geisen et al. 2008). Some of these migratory defects were phenocopied in compound Robo1;Robo2, Slit1;Slit2, and Robo2;Slit2 mutants, suggesting that Robo receptors could act downstream of Hox genes to control PN migration. Consistently, expression of Robo2 in migrating PN was reduced in Hoxa2−/− and Hoxa2−/−;Hoxb2−/− mutant mice, and Hoxa2 directly binds to the Robo2 locus in P19 cells. These results suggest that Robo2 receptors are direct transcriptional targets of Hoxa2 in PN-migrating cells (Fig. 1B) (Geisen et al. 2008).
Local cellular interactions mediated by cell-adhesion molecules and trophic factors regulate the migration of cerebellar granule neurons (CGN) in a process that also seems to be transcriptionally regulated. Granule cells in the early postnatal cerebellum migrate radially from the external germinal layer to the internal granule cell layer (Fig. 1C). At these stages, the nuclear factor I (NFI) proteins NFIA, NFIB, and NFIX are highly expressed in radially migrating CGN (Wang et al. 2007). In vivo analysis of Nfia−/− mutants and in vitro experiments using a dominant repressor form of NFI (NF1/ENR) revealed prominent defects in CGN migration (Wang et al. 2007). NFI transcription factors seem to cell autonomously control CGN migration because NFI/EnR infection also inhibited the migration of dissociated CGN. Several lines of evidence suggest that NFI function involves the direct regulation of molecules regulating cell–cell contacts, such as EphrinB1 and N-cadherin. Thus, inhibition of EphrinB1 and N-cadherin partially mimics the loss of NFI function in the cerebellum and expression of EphrinB1 and N-cadherin is reduced in the absence of NFI. Moreover, NFI factors bind to the N-cadherin and EphrinB1 promoters in CGNs (Wang et al. 2007) (Fig. 1C).
Another factor involved in regulating the migration of CGNs is Barhl1, a HD transcription factor highly expressed in both cerebellar granule progenitors and migratory cells. Analysis of the Barhl1 mutants revealed a disruption in the radial translocation of CGNs, which otherwise appear to be properly specified, and a prominent increase in CGNs’ cell death (Li et al. 2004). These defects might be caused by a reduction in the expression of Neurotrphin-3 (NT-3) in the cerebellum of Barhl1−/− mice, as NT3 has been shown to be required for the migration and survival of cerebellar granule cells (Minichiello and Klein 1996; Bates et al. 1999; Li et al. 2004). However, a direct link between Barhl1 and NT-3 in cerebellar granule cells remains to be established.
The Migration of Telencephalic Interneurons Is Transcriptionally Regulated
Neurons in the developing ventral telencephalon (i.e., the subpallium) undergo extensive migrations to reach their target territories (Marín and Rubenstein 2001), and recent studies have started to reveal a prominent contribution of transcriptional regulation in this process. The medial ganglionic eminence (MGE) is a proliferative structure located in the subpallium that give rise, among other neuronal populations, to interneurons that migrate tangentially to the striatum and cerebral cortex (Lavdas et al. 1999; Sussel et al. 1999; Wichterle et al. 1999, 2001; Marín et al. 2000). Neuropilin-1 (Nrp1) and Neuropilin-2 (Nrp2), the binding receptors for the striatal repulsive molecules Semaphorin-3A and Semaphorin-3F, respectively, are expressed by MGE-derived cortical and absent from MGE-derived striatal interneurons, mediating thereby the segregation of these populations (Marín et al. 2001). Recently, the HD transcription factor Nkx2-1 has been shown to modulate the selective responsiveness of MGE-derived interneurons to class 3 semaphorins (Fig. 2A). All MGE progenitor cells express Nkx2-1, which plays a crucial role in their specification (Sussel et al. 1999). Interestingly, Nkx2-1 expression is maintained in MGE-derived postmitotic striatal interneurons but is downregulated by cortical migrating cells (Nóbrega-Pereira et al. 2008) (Fig. 2A). Forced expression of Nkx2-1 in all MGE-derived cells prevented the migration of interneurons to the cortex. Conversely, conditional deletion of Nkx2-1 in MGE-derived postmitotic cells resulted in a reduction in the number of striatal interneurons (Nóbrega-Pereira et al. 2008). Nkx2-1 mediates the sorting of these interneuronal populations by negatively regulating their sensitivity to class 3 semaphorins; MGE-migrating cells expressing Nkx2-1 are no longer repelled by a source of semaphorins and exhibit a significant reduction in the expression of Nrp2. Moreover, Nkx2-1 directly binds to the Nrp2 locus in MGE cells, and interaction with this sequence is sufficient to repress transcription in vitro, suggesting a direct and cell-autonomous role for Nkx2-1 in controlling the expression of Nrp2 in MGE-derived cells (Nóbrega-Pereira et al. 2008) (Fig. 2A).
Dlx1 and Dlx2 HD transcription factors have also been described as potential negative regulators of Nrp2 expression in MGE cells. Dlx1 and Dlx2 bind to a specific region of the Nrp2 locus in vivo, and interaction with this genomic region promotes repression in vitro (Le et al. 2007). However, because these factors are expressed both by MGE progenitor cells and in migrating neurons (Eisenstat et al. 1999), the functional relevance of this interaction for the segregation of MGE-derived cells is unclear.
In addition to guidance molecules, there is increasing evidence that proteins related to the actin cytoskeleton and microtubule-associated proteins, such as Lis1, doublecortin (Dcx), and Dclk, also control the tangential migration of MGE-derived interneurons (Kappeler et al. 2006; Friocourt et al. 2007). So, as for the guidance receptors, it is conceivable that specific transcription factors may regulate the precise expression of these proteins in migrating neurons. Analysis of Dlx1/2 double mutants revealed a role for these transcription factors in controlling the differentiation and tangential migration of subpallial-derived γ-aminobutyric acidergic interneurons (Anderson et al. 1997; Marín et al. 2000; Pleasure et al. 2000), but the molecular mechanisms operating downstream of Dlx1/2 remain unclear. A recent study has shown that Dlx1/2-deficient MGE cells have increased expression of several cytoskeleton regulators, such as microtubule-associated protein 2 (MAP2), Tau1, growth associated protein 43 (GAP43), and p-21-activated serine/threonine kinase 3 (PAK3), which may in turn cause their premature differentiation and block their migration (Cobos et al. 2007). For example, the p-21-activated serine/threonine kinase PAK3, a downstream effector of the Rho family of GTPases (Bokoch 2003), is primarily expressed in postmigratory MGE-derived cells, in which it controls the growth of axons and dendrites. Thus, premature expression of PAK3 in migratory MGE-derived cells causes excessive neurite length, which in turn blocks their normal movement (Cobos et al. 2007). This work strongly suggests that Dlx1/2 transcription factors promote tangential migration, in part, through negatively regulating the neurite differentiation program in migratory neurons (Fig. 2A).
Role of Transcription Factors in the Migration of Cortical Projection Neurons
Cortical projection neurons are born from progenitor cells located in the dorsal telencephalon (i.e., the pallium) and migrate radially toward the pial surface to form the layers of the developing cortex (Kriegstein and Noctor 2004) (Fig. 2B). Regulation of the actomyosin and microtubule cytoskeletons, adhesion molecules, and nonreceptor kinases is believed to regulate the migration of cortical projection neurons (Marín and Rubenstein 2003; Ayala et al. 2007). The bHLH transcription factors neurogenin1 (Ngn1) and neurogenin2 (Ngn2) are well known for their proneural activity in several regions of the central nervous system, including the pallium (Guillemot 2007). In addition, they have been proposed to potentiate cortical neuronal migration (Fig. 2B). For instance, Ngn1/2 may enhance cortical cell migration by increasing the expression of p35 and Dcx and diminishing the expression of the small GTPase RhoA, key modulators of the actin and microtubule cytoskeleton (Ge et al. 2006). Consistent with this notion, the migration defects observed in the cortex of Ngn2 mutants can be rescued by reducing the function of RhoA (Hand et al. 2005). These genes are likely direct targets of Ngn1 and Ngn2 because both factors bind directly to E-box elements (i.e., consensus binding sites for bHLH factors) located in the Dcx promoter in cortical neurons and are able to induce and repress, respectively, the expression of the Dcx and RhoA promoters in vitro (Ge et al. 2006) (Fig. 2B). This mechanism provides a conceptual framework to understand how transcription factors simultaneously control the expression of several genes in migrating neurons.
The expression of RhoA, Dcx, and p35 is only partially affected in the cortex of Ngn2 single and Ngn1/2 double mutants, suggesting that other genes contribute to regulate their expression. Rnd2, another member of the Rho family of small GTPases, is transiently expressed by cortical migrating neurons and has been identified as a potential Ngn2 downstream target in a recent genomic screen (Heng et al. 2008) (Fig. 2B). Knockdown and overexpression of Rnd cause striking defects in the radial migration of projection neurons. In particular, deregulation of Rnd activity leads to the persistence of an immature multipolar morphology, which may prevent the migration of projection neurons (Nakamura et al. 2006; Heng et al. 2008). As previously shown for a dominant-negative form of RhoA, weak and transient expression of Rnd2 in newborn Ngn2-deficient cortical neurons rescued the morphological and migratory abnormalities found in Ngn2 mutants, reinforcing the view that Rnd2 acts as a downstream mediator of Ngn2 during cortical migration (Heng et al. 2008). Furthermore, ChIP assays suggest that Ngn2 directly regulates Rnd2 expression in newborn cortical projection neurons (Heng et al. 2008) (Fig. 2B).
Migrating cortical projection neurons populate the developing cortex in an “inside–out” pattern, through which deep cortical layers contain cells that become postmitotic earlier than cells populating relatively more superficial layers (Rakic 1974). This process is regulated through at least 2 closely related signaling pathways, one involving the glycoprotein Reelin and the intracellular adaptor protein Dab1 and another implicating the cyclin-dependent kinase (Cdk5) (Marín and Rubenstein 2003; Ayala et al. 2007). These pathways appear to be transcriptionally regulated in migrating cortical neurons by the class III POU-domain transcription factors Brn1 and Brn2 (Fig. 2B). Mice lacking Brn1 and Brn2 exhibit an inversion in cortical lamination that resembles the defects observed in mice deficient in Cdk5 or Reelin signaling (McEvilly et al. 2002; Sugitani et al. 2002). In addition to a role in the proliferation of late cortical progenitors (Sugitani et al. 2002), Brn1/2 cell autonomously control the correct laminar allocation of projection neurons by inducing the expression of the Cdk5-activating subunits p35 and p39 and the Reelin cytoplasmatic effector Dab1 in migrating neurons. Expression of these genes is reduced in the cortex of Brn1/2 mutants, and Brn1/2 are able to bind and activate the transcription of the p35 and p39 promoters in vitro (McEvilly et al. 2002; Sugitani et al. 2002) (Fig. 2B).
Transcription Factors as Multitasking Regulators during Brain Development
It has been typically assumed that different families of transcription factors control distinct events during development, such as the specification of progenitor cells or the differentiation of neuronal populations. This prediction, however, turned out to be largely incorrect (Guillemot 2007). Thus, whereas some transcription factors seem to function primarily in the differentiation of neurons (e.g., LIM-HD transcription factors), many others regulate distinct events at different developmental stages (Shirasaki and Pfaff 2002; Müller et al. 2003; Garcia and Jessell 2008). For instance, the HD transcription factors Nkx2-1 and Hoxa2 are first required for the early specification of specific neuronal populations in the developing telencephalon and hindbrain, respectively, and later regulate the migration of the same neurons (Geisen et al. 2008; Nóbrega-Pereira et al. 2008). An obvious question that emerges from this complexity is how time and context-dependent transcriptional selectivity is achieved for a given transcription factor. Transcription factors are able to recognize and bind to specific DNA target sequences through the interaction of its DNA-binding motif with unique nucleotide sequences present in the promoter of selected genes (Damante et al. 1996). In addition, several other mechanisms have been proposed to further modulate transcriptional selectivity, including posttranslational modifications (e.g., phosphorylation or acetylation) of specific amino acid residues in protein- or DNA-binding motifs (Yang et al. 2004; Hand et al. 2005). Interactions with specific coregulators have also been described, such as the cooperative binding of the Mash1 and Brn transcription factors to adjacent cis sequences in the Delta1 promoter in cortical neurons or the interaction of Hox transcription factors with Pbx cofactors (Samad et al. 2004; Castro et al. 2006). However, how these mechanisms control the switch in transcriptional selectivity is still unclear. Recently, the Ngn1/2 transcription factors have been proposed to control neurogenesis and migration of cortical projection neurons through partially distinct mechanisms. A tyrosine (T) to alanine (A) replacement in the position 241 of the Ngn2 protein (Y241A-Ngn2) blocks cortical cell migration without changing Ngn2 ability to regulate the neurogenic NeuroD promoter, whereas the mutated AQ-Ngn1/2, containing 2 amino acid substitutions in the C-terminus DNA-binding domain, fails to activate the neurogenic transcriptional program but still promotes neuronal migration (Hand et al. 2005; Ge et al. 2006). Phosphorylation of the Y241 residue was proposed to mediate the Ngn1/2 migratory promoting activity by displacing the CREB (cAMP [cyclic adenosine monophosphate] response element binding) binding protein (CBP) coactivator, which in turn interacts with the RhoA and Dcx promoters during migration. In contrast, Ngn1/2 proneural function does not seem to heavily rely on CBP function and instead requires binding to E-box elements. In any case, Ngn1/2 function in neuronal migration also depends on their binding to E-box elements (Hand et al. 2005; Ge et al. 2006; Heng et al. 2008), suggesting that the segregation of the migratory and proneural functions of Ngn1/2 is not exclusively regulated by these mechanisms. In sum, even for the same developmental process, transcription factors use a combination of strategies to recognize and modulate the expression of different effector genes.
As for axon guidance, increasing evidence suggests that the precise migration, guidance, and allocation of different neuronal populations is controlled by the precise transcriptional regulation of effector genes regulating a variety of events, from cytoskeleton dynamics to the response to guidance cues. It is also now evident that the same transcription factor can regulate several steps during the development of a given neuronal population. Understanding the mechanisms that modulate transcriptional selectivity will further uncover the multitasking potential of transcription factors during brain development.
Spanish Ministry of Education and Science (BFU2005-04773/BMC, CONSOLIDER CSD2007-00023); EURYI program; Predoctoral fellowship from the Foundation for Science and Technology (POCI 2010/FSE), Portugal, to SNP.
We apologize to colleagues whose work was not discussed due to space limitations. Conflict of Interest: None declared.