REVIEWS

Patterning and guidance of cranial motor neurons

Sarah Guthrie Abstract | The cranial motor control muscles involved in eye, head and movements, feeding, speech and facial expression. The generic and specific properties of cranial motor neurons depend on a matrix of rostrocaudal and dorsoventral patterning information. Repertoires of transcription factors, including Hox , confer generic and specific properties on motor neurons, and endow subpopulations at various axial levels with the ability to navigate to their targets. Cranial motor axon projections are guided by diffusible cues and aided by guideposts, such as exit points, glial cells and muscle primordia. The recent identification of genes that are mutated in human cranial dysinnervation disorders is now shedding light on the functional consequences of perturbations of cranial development.

Neural tube In humans, the cell bodies of cranial motor neurons lie experiments. The specificity of cranial motor neuron The primordium of the nervous in the , and their extend through the projections is governed by rostrocaudal and dorsoven- system. to control muscles in the head and neck. tral patterning mechanisms that produce a diversity of Other (including fish, chicks and mice) show motor neuron subpopulations with distinct differentia- Floor plate The ventral midline structure of a high degree of conservation in both the arrangement tion programmes. Some of the guidance molecules that the CNS. It has a role in of brainstem motor neurons and the muscles they inner- are involved in elaborating axon projections have also patterning and axon guidance. vate. Developing motor axons perform a spectacular feat, been characterized. However, many important ques- navigating over long distances from the CNS to their tions remain. The unique features of the differentiation Branchial arches targets in the periphery. programmes of each of the cranial nerves are only partly Repeated bars of mesenchymal that Early in development, the neural tube acquires a series of characterized. In particular, we know little about how contribute to the lower jaw and swellings at its rostral end, presaging the development patterning genes dictate the repertoires of receptors on neck; each contains a of the , the and the . axons, or how these receptors determine axon pathfind- cartilaginous component, a Caudally, the neural tube remains narrow and elongates ing behaviour to particular muscle targets. Deciphering muscular component, a nerve and an . to form the . Motor neurons differentiate these molecular mechanisms is a major challenge in ventrally, on either side of a midline structure, the floor developmental neurobiology. plate. Cranial motor neurons reside in the midbrain and In this context, it is fascinating that cranial dysinner- the hindbrain (which together constitute the brainstem), vation disorders, which reflect abnormalities of one or where they are partitioned into a series of nuclei. By more cranial nerves, are starting to be genetically charac- contrast, spinal motor neurons form a number of dis- terized in humans6. Clinical studies, together with studies continuous columns along the length of the cord (for in animal models, are now providing fresh impetus to reviews, see refs 1,2). Cranial motor axons follow dorsal understand how normal and abnormal cranial nerve wir- or ventral pathways from the brainstem; the axial posi- ing develops. In this Review, I describe the latest findings tion of this site of exit in turn dictates their peripheral in cranial motor neuron patterning and axon guidance, paths to muscles of the eye, tongue, branchial arches or to focusing mainly on mouse and chick studies (as zebrafish parasympathetic ganglia3 (FIG. 1; TABLE 1). studies have been reviewed elsewhere: see REF. 5). MRC Centre for Surprisingly, despite the functional significance of Developmental Neurobiology, cranial motor nerves, an understanding of the molecular Motor nuclei form at distinct axial levels King’s College, Guy’s Campus, mechanisms that underlie their development is only just Cranial motor neurons comprise three subsets: bran- London, SE1 1UL, UK. 4,5 e‑mail: starting to emerge . Exciting progress has been made in chiomotor (BM), visceral motor (VM) and somatic [email protected] understanding cranial motor neuron development, par- motor (SM) neurons (FIG. 2; TABLE 1). Early in develop- doi:10.1038/nrn2254 ticularly from gain-of-function and loss-of-function ment, these neurons arise in longitudinal progenitor

nature reviews | volume 8 | november 2007 | 859 © 2007 Nature Publishing Group

REVIEWS

r1, r2 and r3 (in mice) or r2 and r3 (in chicks), the facial (nucleus VII; BM and VM neurons) lies in r4 and r5, the glossopharyngeal nucleus (nucleus IX; BM MB and VM neurons) lies in r6 (in mice) or r6 and r7 (in chicks), and the vagus nucleus (nucleus X; BM and VM neurons) and cranial accessory nucleus (XI; BM FB III neurons) occupy r7 and r8 (REF.10) (FIG. 2). In the cau- dal hindbrain, the (nucleus VI, SM IV neurons) occupies r5 in mice and r5 and r6 in chicks, with the extended (nucleus XII, HB SM neurons) found in r8. In both mice and chicks, the facial motor neurons of the BM subtype are segregated V in r4, and those of the VM subtype are segregated in r5 (REFS 11,12). In all except avian species, the facial bran- VII/VII chiomotor (FBM) neurons are born in r4 and then undertake a striking caudal migration to r6 (REFS 10,13), BAs VI OV unlike most BM and VM neuron somata, which migrate dorsally14. Rhombomere 4 also contains a population of IX vestibuloacoustic neurons, which are efferent to the hair XII cells of the ; a subset of these neurons (contral- ateral vestibuloacoustic neurons) translocate their cell X/XI bodies across the midline15. Following their exit into the periphery, cranial motor axons converge to form components of the cranial nerves (FIG. 1). BM axons travel, through the trigeminal, facial, glossopharyngeal, vagus and cranial accessory nerves, towards branchial arches 1, 2, 3, 4 and 6, respectively, where they innervate muscles of Figure 1 | Cranial nerves in the chick embryo. A lateral the jaw and muscles that control facial expression, as view of cranial nerves in the chickNatur embryoe Revie wsat embryonic| Neuroscienc e well as the and the . VM axons project day four, showing the pathways from the hindbrain (HB), on the right, into the branchial arches (BAs) and other head towards , the neurons of structures (the midbrain (MB) and the forebrain (FB)), on the which supply salivary and lacrimal glands, smooth left. Roman numerals denote the nerves: III, oculomotor; IV, muscle and visceral organs. Oculomotor, trochlear trochlear; V, trigeminal; VI, abducens; VII/VIII, facial/ and abducens SM neurons innervate the six eye mus- vestibuloacoustic; IX, glossopharyngeal; X, vagus; XI, cles, with an additional oculomotor VM component cranial accessory; XII, hypoglossal. OV, otic vesicle. Figure synapsing at the ciliary . Hypoglossal neurons modified, with permission, from ref. 3  (1990) Wiley-Liss. project rostrally through the floor of the pharynx to the tongue muscles. Cranial motor nuclei conform to a theme, sharing common features, such as morphology domains in the hindbrain basal plate: BM and VM and initial axon trajectory, but nevertheless possess- neuronal somata migrate dorsally into the alar plate, ing distinct positional identity, synaptic targets and whereas SM somata remain ventral (in the basal functions. plate). BM and VM axons extend dorsally through the neuroepithelium to large common exit points, whereas Rostrocaudal patterning of the brainstem Basal plate The ventral half of the SM axons leave the neuroepithelium ventrally in small The midbrain is divided into a series of ‘arcs’, which neuroepithelium. groups (FIG. 2c), with the exception of trochlear SM have been proposed to underlie the differentiation of axons, which grow dorsally and cross the dorsal mid- nuclei and are distinguished by the expression of vari- Alar plate line at the midbrain–hindbrain boundary to project ous genes and other molecular markers16–18. The dorsal half of the contralaterally. The most medial arc contains oculomotor neurons, neuroepithelium. Individual motor nuclei can contain one or more and fibroblast growth factor 8 (FGF8), produced by Neuroepithelium of BM, VM and SM neuron subsets. The oculomotor the midbrain–hindbrain boundary, has been proposed A part of the early nervous nucleus (nucleus III), which contains SM and VM to dictate the rostrocaudal position of the oculomo- system that consists of dividing neurons, lies most rostrally in the midbrain. Along the tor nucleus, because misexpression of FGF8 shifts the progenitors arranged in a 17 columnar epithelium. rostrocaudal axis, the hindbrain is divided into rhom- nucleus rostrally . Differentiating oculomotor neurons bomeres, segmental entities that contain repeating sets express the homeobox gene paired-like homeobox 2a Homeobox of neurons with distinct differentiation programmes (Phox2a)16, which is an important determinant of ocu- A conserved 180 at different axial levels7–9. Motor nuclei differentiate lomotor identity, as oculomotor neurons (as well as sequence that encodes in individual rhombomeres or pairs of rhombomeres. trochlear motor neurons) are absent in Phox2a-mutant homeodomain regions of 19 that are involved in Rostral rhombomere one (r1) contains the trochlear mice . Further details of the transcriptional hierarchy binding to DNA and regulating nucleus (nucleus IV), which contains SM neurons. The that underlies oculomotor neuron determination remain transcription. trigeminal nucleus (nucleus V; BM neurons) occupies to be discovered.

860 | november 2007 | volume 8 www.nature.com/reviews/neuro © 2007 Nature Publishing Group

REVIEWS

Table 1 | Motor components of the cranial nerves and their targets in humans Nerve Subtype Nucleus Target muscles or ganglia III Somatic motor Oculomotor Superior, inferior and medial recti muscles; inferior oblique, levator palpebrae superioris Visceral motor Edinger-Westphal IV Somatic motor Trochlear Superior oblique V Branchiomotor Trigeminal motor Muscles of mastication, tensor tympani, anterior belly of digastric, others VI Somatic motor Abducens VII Branchiomotor Facial motor Muscles of facial expression, stapedius, posterior belly of digastric Visceral motor Superior salivatory Pterygopalatine/sphenopalatine ganglion, IX Branchiomotor Visceral motor Inferior salivatory X Branchiomotor Nucleus ambiguus Laryngeal and pharyngeal muscles Visceral motor Dorsal motor Non-striated muscle of thoracic and abdominal viscera Cranial XI Branchiomotor Nucleus ambiguus Laryngeal and pharyngeal muscles Spinal XI Branchiomotor Accessory nucleus, Sternocleidomastoid and muscles cervical spinal cord XII Somatic motor Hypoglossal Tongue muscles

In the hindbrain, a large number of transcription in a particular rhombomere, as well as the timing of the factors and other genes pattern rhombomere territories onset of the expression and the expression level, dictates through their segmental expression9,13,20. Many of these segmentation and segment identity at that axial level. genes regulate motor neuron development, either directly The patterns of expression in the hindbrain or indirectly. For example, the transcription are established, at least in part, by the diffusible action factor early growth response 2 (EGR2; also known as of FGF8 and retinoic acid (RA) at the rostral and caudal KROX20) is expressed early in r3 and r5 in the mouse21 ends of the hindbrain, respectively28,29. Rostrally, FGF8 and determines many of the features of odd-numbered that is produced by the midbrain–hindbrain boundary rhombomeres. In Egr2-mutant mice, r3 and r5 are sets the rostral boundary of Hoxa2 expression30. RA that largely missing, depleting the motor neurons at these is derived from the mesodermal somites, which flank the levels22. The MAFB (also known as caudal hindbrain, is thought to specify the caudal hind- Kreisler) is similarly expressed early and regulates r5 and (r5 to r8) through the induction of Hox genes, in r6 development23,24; in Mafb-mutant mice, r5 and r6 are a dose-dependent manner28. Experimental depletion of lost, causing the deletion of the r5 VM facial and SM RA in mice and chicks supports this idea: it causes caudal abducens neurons25. rhombomeres to assume more rostral identities (either Both Egr2 and Mafb lie upstream of, and activate r3 or r4)31,32. In some cases, the expression of caudal the transcription of, Hox genes, which contain an Hox genes is induced through upstream Retinoic Acid Antennapedia-class homeobox sequence and have a pre- Response Elements (RAREs)33,34. RA is implicated in the eminent role in hindbrain patterning. In vertebrates there production of SM neurons in the spinal cord35,36, raising are four Hox-gene clusters (named a–d) on four separate the possibility that it might also generate SM neurons in chromosomes26. There are 13 paralogue groups, (although r5 to r8 (the abducens and hypoglossal nuclei). Somatic no individual cluster contains all 13 genes). Paralogous motor neurons differentiate throughout the rostrocaudal genes (for example, Hoxa3 and Hoxb3) often exhibit over- extent of chick hindbrain explants following application of lapping functions; indeed, dosage-dependent effects have RA37, whereas a reduction in RA signalling in the zebrafish been shown in the case of Hoxa3 and Hoxd3 (Ref. 27). Hox produces a loss of hindbrain cranial motor neurons38. The genes show nested domains of expression in the hindbrain absence of SM neurons in the rostral hindbrain might be (FIG. 3), with each paralogue group expressed from the spi- maintained by the action of the RA‑degrading enzyme nal cord rostrally to particular rhombomere boundaries. CYP26, as inhibition of CYP26 caused the differentiation Thus, group 2, 3 and 4 domains end at the of SM neurons throughout hindbrain explants37. Thus, r2–r3, r4–r5 and r6–r7 boundaries, respectively (although FGF8 and RA, through their role in patterning Hox genes, Hoxa2 expression terminates at the r1–r2 boundary). and possibly through independent inductive mechanisms, Group 1 genes show an anomalous rostral restriction at pattern motor neurons such that BM and VM neurons the r3–r4 boundary, and Hoxb1 is expressed at high levels differentiate throughout the hindbrain, whereas SM in r4. The combination of Hox genes that are expressed neurons are restricted to caudal rhombomeres.

nature reviews | neuroscience volume 8 | november 2007 | 861 © 2007 Nature Publishing Group

REVIEWS

a Chick at embryonic day 4 b Mouse at embryonic day 11.5 MB MB III III

HB IV HB IV

r1 r1

gV r2 FP V gV r2 FP V c Transverse section r3 r3 of chick branchial region CVA gVII r4 VII/VIII gVII r4 CVA VII/VIII aVI aVI r5 r5 VE gVIII VIVI gVIII VIVI r6 IX r6 IX OV OV FP G r7 r7 X, XI X, XI gIX gIX r8 r8 PX gX gX XII XII BA

Figure 2 | Motor neuron organization in the brainstem. a | The organization of motor neurons in a flat- mounted chick brainstem at embryonic day four (E4). b | The organization of motor neurons inNa a turflat-mountede Reviews | Neur mouseoscienc e brainstem at E11.5. In parts a and b, rhombomere (r) levels are indicated; branchiomotor and visceral motor neurons are shown in red; somatic motor neurons are shown in blue. Roman numbers alone denote the nerves: III, oculomotor; IV, trochlear; V, trigeminal; VI, accessory abducens; VII/VIII, facial/vestibuloacoustic; IX, glossopharyngeal; X, vagus; XI, cranial accessory; XII, hypoglossal. Roman numerals with the letter g denote the cranial ganglia: gV, ; gVII, ; gVIII, vestibuloacoustic ganglion; gIX, petrosal ganglion; gX, nodose ganglion. c | Cranial motor axon pathways in a transverse section of the branchial region in the chick embryo at E4 (for details, see REF. 10). aVI, accessory abducens nucleus; BA, branchial arch; CVA, contralateral vestibuloacoustic neurons; FP, floor plate; G, cranial sensory ganglion; OV, otic vesicle; PX, pharynx; VE, ventricle.

Hox genes control motor neuron identity Hoxa2 regulates trigeminal motor neuron differentiation. As mentioned above, Hox genes are key controllers of Hoxa2 is expressed up to the r1–r2 boundary and is rostrocaudal patterning in the head, including hind- the only Hox gene to be expressed in r2, whereas in brain segmentation and rhombomere identity4,39. As r3 it is co-expressed with Hoxb2 (Refs 39,43). In both well as being expressed in neuroepithelial domains mice and chicks, only r4 second branchial arch, not r2, (FIG. 3), Hox genes are expressed in neural crest cells, (first branchial arch) neural crest cells express Hoxa2, which emigrate predominantly from even-numbered and loss of Hoxa2 in mice produces a striking transfor- rhombomeres into the branchial arches (FIG. 4), generat- mation of the second branchial arch into a first-arch ing skeletal tissues and cranial ganglia40–42. As a result, phenotype44–46. In Hoxa2-mutant mice, presumptive several Hox-mutant mice show defects in the pattern- trigeminal motor neurons in r3, and a subset in r2, ing of branchial arch-derived structures39. There is project through the r4 exit point46. This suggests that positional registration, that is, motor neurons project trigeminal motor neurons misroute to the transformed into peripheral territory that expresses the same Hox second arch, either because of a change in motor neuron repertoire as their rhombomere of origin. One expecta- identity and/or because of a change in axon guidance tion of this pattern is that Hox genes dictate the expres- cues. The r4-derived is also reduced in sion of matching axon guidance receptors and ligands, size later in development, possibly owing to a loss of in neurons and their targets, respectively. But, hitherto, second branchial arch character and arch-derived fac- there is little data on this issue. Nevertheless, a number tors46. When Hoxa2 is ectopically expressed in chick r1, of studies with Hox-mutant mice have shown specific which normally lacks motor neurons, trigeminal defects in the generation and patterning of cranial neurons are generated47, supporting the idea that motor neurons. Loss of rostrally expressed Hox genes Hoxa2 specifies trigeminal motor neurons. In Hoxb2- in paralogue groups 1 and 2 leads to patterning defects mutant mice there is also a mis-specification of some of the trigeminal and facial BM and VM nuclei, which trigeminal motor neurons in r3, which project through occupy rostral rhombomeres, whereas SM neurons in the r4 exit point, but the effect on trigeminal axon caudal rhombomeres are affected by the loss of group 3 projections is less striking than with Hoxa2 muta- paralogues. tion (Ref. 48). It is likely, therefore, that r2 trigeminal

862 | november 2007 | volume 8 www.nature.com/reviews/neuro © 2007 Nature Publishing Group

REVIEWS

54 MB into the first arch . Global Hoxb1 overexpression appears to cause the coordinated transformation of motor neurons and the first arch, such that they develop an FBM-like second arch identity. The molecular significance of this

HB positional registration in Hox gene expression between 2 BM neurons and their innervation territory was unclear for some time. However, recent studies show that Hoxb1-

r1 Hoxa positive, r4-derived neural crest cells preferentially give rise to Schwann cells, which might provide the guid-

r2 Hoxb2 1 ance cues and/or survival factors that are needed for the 55

Egr2 3

Hoxa r3 projection and maintenance of the facial . a3

x Conditional deletion of Hoxb1 only in the neural crest

Ho Hoxd Hoxb1 r4 Hoxb3 results in a significant proportion of animals showing Egr2 Mafb r5 facial paralysis, indicative of a loss of facial motor neurons 4 4 4 (as in Hoxb1-null mice)56. The facial nerve fails to branch 56

r6 Hoxb Mafb Hoxd Hoxa correctly and eventually the facial motor neurons die . FBM axons might require an interaction with Schwann r7 cells for their guidance and survival, perhaps through the production by the Schwann cells of neurotrophic factors. r8 An intriguing piece of evidence in favour of this idea is the prolonged survival of mis-specified FBM neurons in Hoxb1-mutant mice that are also mutant for Bax (a gene that is required in cell death pathways that are associated with neurotrophic factor deprivation)57. Figure 3 | Expression patterns of Hox genes in the vertebrate hindbrain. The As with the pairing of Hoxa2 and Hoxb2 in trigeminal domains of mRNA expression of the Hox genes Egr2 and MafbNa turin thee Re hindbrainviews | Neur ofoscienc the e mouse and chick embryo, at embryonic day 11.5 (E11.5) and E4, respectively. The bars neuron specification, Hoxa1 and Hoxb1 appear to labelled with different Hox genes show the genes’ expression domains, which extend act synergistically to pattern facial motor neurons. from the caudal hindbrain up to particular rhombomere (r) boundaries4,168. Darker Hoxa1;Hoxb1 double-mutants reflect this fact, and shading indicates higher levels of expression. HB, hindbrain; MB, midbrain. show more extensive FBM neuron defects than either of the single mutants58,59. Hoxb2 also functions in FBM specification, possibly because it is a downstream target neurons are patterned by Hoxa2, whereas those in r3 of Hoxb1 (Refs 48,60). The network of regulatory inter- are patterned by a combination of Hoxa2 and Hoxb2. actions that has been revealed so far for r4 provides a glimpse of how daunting it will be to achieve a thorough Hoxb1 is a key regulator of FBM neurons in r4. Extensive understanding of hindbrain motor neuron patterning. insight has been gained into the genetic hierarchy that Following the activation of Hoxa1 and Hoxb1 by RA, underlies FBM neuron development in r4. Hoxa1 and Hoxa1 transactivates Hoxb1, Hoxb1 auto-regulates its Hoxb1 are expressed up to the r3–r4 boundary in divid- own expression and activates Hoxb2 and Hoxa2 (Ref. 61). ing progenitors, but the onset of Hoxa1 expression comes Mutation studies have also shown that the genes Gata2, earlier (at embryonic day eight), and after neuronal dif- Gata3 and T-box 20 (Tbx20) are among the downstream ferentiation, only Hoxb1 is maintained at high levels in r4 targets of Hoxb1 in FBM specification62,63. and in the postmitotic FBM neurons40. In Hoxa1-mutant One outstanding question is how the specificity of Hox mice, rhombomeric segmentation of r3 to r8 is disrupted, proteins in binding to particular DNA target regions is the r3–r4 boundary does not form correctly, r4 is reduced conferred. An answer might lie in the fact that Hox gene- in size and r5 is absent49–51. By embryonic day 18.5, this encoded homeoproteins form complexes with specific early role of Hoxa1 in rhombomere segmentation and cofactors, possibly restricting the proteins’ binding spe- identity leads to loss of the and reduction cificity. These cofactors are the Pbx and Meis homeopro- or loss of the facial nerve. By contrast, Hoxb1 has a later teins, which form a tripartite complex with Hox proteins role in FBM neuron specification, reflected in the obser- to regulate downstream transcription64. In pbx4-mutant vation that in Hoxb1-mutant mice, segmentation occurs zebrafish, FBM neuron development is defective, and normally but there is a striking absence of FBM neurons the resultant phenotype is identical to that which results (which fail to migrate caudally to r6 and eventually die); from a deficiency in hoxb1a, the zebrafish Hoxb1 homo- CVA neurons are also missing52,53. Motor neurons that do logue65,66. Complete elimination of maternal and zygotic differentiate in r4 project axons to the first branchial arch, Pbx function produces a hindbrain r1 ‘ground state’ in suggesting that they default to a trigeminal-like identity. r2 to r7, reflected by a lack of characteristic molecular The instructive role of Hoxb1 was demonstrated in chicks markers and neuronal types, especially BM and VM neu- by overexpressing Hoxb1 in r2. This converts trigeminal rons67. The specificity of the interactions between Pbx motor neurons to an FBM fate, resulting in anomalous proteins and Hox proteins remains to be characterized, facial-like axon projections to the second branchial arch54. but HOXB1–Pbx- complexes have been shown to However, when Hoxb1 is misexpressed in both r2 and the bind to specific regulatory regions of the Hoxb1 and Hoxb2 first branchial arch, ectopic FBM axons navigate from r2 genes to regulate their expression in r4 (Refs 60,68).

nature reviews | neuroscience volume 8 | november 2007 | 863 © 2007 Nature Publishing Group

REVIEWS

specific dorsoventral progenitor domains76. The expres- sion domains of these transcription factors show cross- r1 MB repressive interactions at their boundaries, consolidating r2 their identity and generating groups of postmitotic neu- HB FB rons that express repertoires of transcription factors that r3 are involved in their further specification77. This model r4 has given insight into the mechanism of differentiation of cranial motor neurons, the progenitors of which r5 occupy distinct dorsoventral domains. r6 OV In the hindbrain, the progenitor domain that flanks the floor plate (known as the p3 domain) gives rise to r7 Md BM and VM neurons78,79, whereas the dorsally adjoin- BA2 Mx ing progenitor domain (known as the pMN domain) (FIG. 5a) BA3 generates SM neurons . P3 and pMN progenitors BA1 express distinct ‘codes’ of transcription factors (FIG. 5), BA4 most of which act as transcriptional repressors to control neuronal identity80–83. The key regulators of BM and VM neuron fate are suspected to be Nkx2.2 and Nkx2.9, but BA4 BA3 BA2 BA1 mutation of either gene on its own leaves these neurons Hox 1 78,84 Hox 2 intact, probably because of redundancy . Nkx6.1 and Hox 3 Nkx6.2 are not required to specify BM and VM neurons Hox 4 per se, but they act to repress alternative interneuron Figure 4 | Patterns of neural crest migration and fates, and in double mutants, BM and VM neurons branchial arch Hox gene expressionNature Revie inws chick | Neur andoscienc e express some interneuron markers79. Loss-of-function mouse embryos. A schematic diagram of a chick head at of Nkx6.1 alone or with Nkx6.2 also results in aberrant embryonic day two, showing pathways of neural crest 79,85 migration in the chick and mouse embryo and patterns of BM and VM neuron migration and axon pathfinding . Hox gene expression in the branchial arches (BAs)42,102,169,170. In the pMN domain, loss-of-function of Nkx6.1 and/or FB, forebrain; HB, hindbrain; MB, midbrain; Md, mandibular Nkx6.2, or of paired box gene 6 (Pax6), deletes abducens part of BA1; Mx, maxillary part of BA1; OV, otic vesicle; and hypoglossal SM neurons80,82,83,86. Combined Nkx r, rhombomere. gene activity in turn induces the expression of the basic helix-loop-helix transcription-factor-encoding gene oligodendrocyte transcription factor 2 (Olig2), which Hox3 genes regulate SM neuron differentiation. Hox3 coordinates generic neuronal differentiation with motor paralogues are good candidates for genes that regulate neuron subtype87,88 and induces the homeobox-containing the production of SM neurons in r5 to r8. Hoxb3, for gene MNR2. example, is expressed at high levels in r5 and r6 in As a result of these early specification events, postmi- chicks, but only in r5 in mice69,70, corresponding with totic SM neurons express a combination of the homeo­ the location of abducens neurons in these two species. box-containing genes MNR2 (which is chick-specific), Knockout studies have shown that abducens neurons are motor neuron and pancreas homeobox 1 (MNX1; also lost in Hoxa3;Hoxb3 mouse double-mutants, but can be known as HB9) and the LIM-homeobox genes Islet 1 ectopically induced by rostral expression of Hoxa3 in (Isl1), Isl2, Lhx3 and Lhx4 refs 73,87,89,90 (FIG. 5b). chicks71,72. Future studies might reveal whether Hox4 Interestingly, hindbrain SM neurons show variations paralogues, which are expressed in a caudal domain that on this pattern, for example, subsets of abducens SM extends up to the r6–r7 boundary, specify hypoglossal neurons express different combinations of these genes91. motor neurons in this region. MNR2 and MNX1 are involved in specifying SM neuron fate and repressing interneuron fates, respectively90,92,93. Dorsoventral patterning of Hox Lhx3 and Lhx4 are determinants of ventral pathway Combinations of other homeobox-containing tran- choice; SM neurons, including abducens and hypoglos- scription factors specify progenitor domains along the sal subpopulations, are absent in Lhx3;Lhx4 double dorsoventral axis of the brainstem1,73. Sonic hedgehog knockouts, and BM and VM neurons that misexpress protein (SHH) has been proposed to form a ventral-to- these transcription factors extend axons ventrally rather dorsal gradient that induces dose-dependent neuronal than dorsally94. Like SM neurons, postmitotic BM and differentiation. Although this model is based largely on VM neurons express Isl1 but, unlike SMs, they also experiments on the spinal cord, it is thought to apply to express Tbx20 (Ref. 95), Phox2a and Phox2b, with Phox2b the hindbrain as well, and both cranial and spinal motor expressed first19 (FIG. 5b). Phox2b is a key gene in BM and neurons are missing in Shh–/– mouse mutants74. SHH also VM neuron generation96, and in Phox2b-mutant mice, controls the differentiation of the midbrain arc, which all BM and VM neurons are absent97. Conversely, in mice contains oculomotor neurons16. The model proposes that lack Phox2a, which is expressed before Phox2b in that graded SHH signalling produces the graded activ- the oculomotor and trochlear nuclei, both of these nuclei ity of Gli transcription factors75, which in turn activate are missing19. The generation of knock-in mouse lines, or repress the expression of homeodomain proteins in in which Phox2b was replaced by the Phox2a locus, or

864 | november 2007 | volume 8 www.nature.com/reviews/neuro © 2007 Nature Publishing Group

REVIEWS

a b pMN SM neurons Pax6, Olig2, MNX1, MNR2, Nkx6.1, NKx6.2. Islet-1, Islet-2, Lhx3, Lhx4. p3 pMN Nkx2.2, Nkx2.9, BM/VM neurons Nkx6.1, Nkx6.2. p3 Islet-1, Phox2b, Phox2a, Tbx20.

FP FP SM neurons BM and VM neurons SHH N N Figure 5 | Dorsoventral patterning of cranial motor neurons. a | A schematic diagram of a transverse section through the hindbrain, showing pMN and p3 progenitor domains, which give rise to branchiomotor (BM) and visceral motor (VM) neurons, and somatic motor (SM) neurons, respectively. The arrows extending from the floor plate (FP) and Nature Reviews | Neuroscience the notochord (N) show the presumed diffusion of Sonic hedghog protein (SHH) during motor neuron differentiation. For the pMN and p3 domains, the repertoires of transcription factors that are expressed by the progenitors are listed. b | A schematic diagram of a transverse section through the hindbrain, showing the location of postmitotic cranial motor neurons following dorsal migration by BM and VM neurons. The repertoires of transcription factors that are expressed by the neurons are listed. Lhx, LIM homeobox protein; MNX1, motor neuron and pancreas homeobox 1; Olig2, oligodendrocyte transcription factor 2; Pax6, paired box gene 6; Phox2b, paired-like homeobox 2b; Tbx20, T-box 20.

vice versa, has revealed that the two genes are not func- into the branchial arch muscles, along paths that were tionally equivalent98. Oculomotor and trochlear motor previously followed by CPM cells; these CPM cells neurons are only partially rescued by the substitution migrate dorsoventrally to form the branchial muscle of Phox2b for Phox2a, and in the Phox2a into Phox2b plates103, which later split up into a characteristic set of knock-in line, FBM neurons differentiate but fail to muscles104. migrate correctly 98. Phox2b is regulated by converging streams of rostro- Motor axons are repelled from the midline. The initial caudal and dorsoventral patterning information, which phase of cranial motor axon extension involves repul- are dependent on Hox gene function and Nkx2.2 and sion of all subtypes of cranial motor axons by the floor Nkx6 gene function, respectively. Loss-of-function stud- plate105,106. Candidates for the mediation of this effect ies reveal that Nkx genes cooperate with Hoxb1 to main- are the axon guidance molecules netrin 1, the Slit pro- tain Phox2b expression in r4 and favour an FBM over teins and SEMA3A, all of which can repel BM and VM a serotonergic neuronal fate99. Hoxb1, Hoxb2 or Hoxa2 axons in vitro107–109. Cranial motor neurons express the misexpression can all activate Phox2b ectopically in UNC5A (which mediates the repellent effect ventral regions of the hindbrain, but co-electroporation of netrin 1), the Slit receptors ROBO1 and ROBO2, of Hoxb1 or Hoxa2 with Nkx2.2 is required to generate and the SEMA3A receptor neuropilin 1 (Refs 109–112). ectopic motor neurons in dorsal regions100. An upstream However, only netrin 1 and the Slit proteins are highly enhancer region that can be transactivated by Hoxb1, expressed in the brainstem floor plate at the time of Hoxb2 or Hoxa2 has recently been identified in the axon extension109,110,113–115, suggesting that these are the Phox2b gene; this region contains conserved Pbx and relevant repellents in vivo. In particular, attenuation of Meis protein binding sites, and its transcriptional activity Slit–ROBO signalling in chicks or mice in vivo leads to can be enhanced by Pbx and Meis cofactors99. BM and VM axon pathfinding defects, suggesting an in vivo role for the Slit proteins109. A third Robo receptor, Diffusible cues guide cranial motor axons ROBO3 (also known as Rig1), has been shown to have These early genetic programmes specify cranial motor a role in midline crossing of some neuronal types116, but neurons and dictate their responses to local guidance might not be expressed in motor neurons. cues in the developing head (TABLE 2). The first pathway There is as yet no clear evidence for a role for netrin 1 choice made by cranial motor axons is whether to project motor neuron repulsion in vivo, although trochlear motor ventrally in small groups into the perinotochordal mes- neuron cell bodies enter the floor plate in netrin 1 Cranial paraxial mesoderm The population of mesoderm enchyme (in the case of SM neurons) or dorsally (in the mutants, suggesting a loss of repulsion from the mid- 117 cells that originates adjacent to case of BM and VM neurons) through large common exit line . Motor axon pathways have not been investigated the brainstem and gives rise points. SM axons then innervate eye muscles derived from in Unc5a mutants, but there are peripheral motor axon to many head muscles. the prechordal mesoderm and cranial paraxial mesoderm guidance defects in mutants for a related Unc5 family (CPM), or tongue muscles derived from somites 1–4 member, Unc5c118,119. Unc5c is also prematurely expressed Sphenopalatine ganglion (Ref. 10) 85 A parasympathetic ganglion . VM axons supply parasympathetic ganglia, such in FBM neurons that lack Nkx6.1 and fail to migrate , that is innervated by VM as the ciliary, sphenopalatine and otic ganglia, which are suggesting that regulation of Unc5 family members neurons of the facial nerve. neural crest-derived structures101,102. BM axons project might be important in aspects of FBM guidance and

nature reviews | neuroscience volume 8 | november 2007 | 865 © 2007 Nature Publishing Group

REVIEWS

Table 2 | Molecules involved in cranial motor axon guidance Molecule Effect on cranial motor axon guidance References HOXB1 In Hoxb1 mutants, facial branchiomotor neurons project aberrantly to the first 52–54,56 branchial arch. When Hoxb1 is ablated in the neural crest only, the facial nerve fails to branch correctly. Ectopic expression of Hoxb1 in r2 converts trigeminal motor neurons into facial motor neurons HOXA2 In Hoxa2 mutants, trigeminal motor axons misroute into the second branchial arch 46 HOXB2 In Hoxb2 mutants, a subset of trigeminal motor axons misroute into the second 48 branchial arch NKX6.1, NKX6.2 In Nkx6.1 mutants, branchiomotor axons mistarget exit points and pathfind 79,85 aberrantly LHX3, LHX4 Direct cranial motor axons along ventral trajectories 94 PHOX2A In Phox2a mutants, visceral motor axons fail to project due to an absence of 138 ganglionic targets Netrin 1 Repels branchiomotor axons in vitro 107,108 UNC5C In Unc5c mutants, trochlear motor axons project aberrantly 118 Slits Repel branchiomotor axons in vitro. In Slit1;Slit2 double mutants, branchiomotor 109 axons aberrantly enter the midline Robos In Robo1 and Robo2 mutants, or in chick embryos that express dominant-negative 109 Robo receptors, branchiomotor axons enter the midline and fail to reach exit points SEMA3A Repels branchiomotor, visceral motor and somatic motor axons in vitro. Sema3A- 108,134 mutant mice show defasciculation of cranial motor nerves Neuropilin 1 Neuropilin 1 mutants show defasciculation of cranial motor nerves 135 SEMA3F SEMA3F repels trochlear motor axons in vitro or in the chick embryo in vivo. In 149,151,152 Sema3F mutants, the fails to exit and the is defasciculated Neuropilin 2 In Neuropilin 2 mutants, the trochlear nerve fails to exit and the oculomotor nerve 149,150 is defasciculated FGF8 FGF8 attracts trochlear axons, causing them to exit the brain 148 Ephrin As Overexpression of ephrin As in branchial muscle causes trigeminal motor axon 143 branching defects EPHAs Trigeminal motor neurons that express dominant-negative EPHAs show axon 143 branching defects SDF‑1/CXCL12 In SDF‑1 mutants, abducens and hypoglossal axons project dorsally rather than 124 ventrally CXCR4 The phenotype of CXCR4 mutants is identical to that of SDF‑1 mutants 124

migration. In some contexts, Unc5a can also bind to the CXCL12), which is expressed in the mesenchyme that DCC (deleted in colorectal carcinoma) receptor, which underlies the neuroepithelium, at least in the trunk124. normally mediates netrin 1 attraction, to mediate repul- The receptor for SDF‑1, CXCR4, is expressed by spinal sion120. In the fly embryo, UNC5A and DCC–UNC5A and cranial SM neurons, but not BM or VM neurons, and mediate short-range and long-range repulsion of dorsally in CXCR4 mutants, spinal, abducens and hypoglossal SM projecting motor neurons, respectively121,122. A similar axons fail to project ventrally and instead extend across possible role of netrin 1 in the rat embryo is implied by the floor plate or project dorsally124. In other neuronal the finding that BM and VM neurons express Unc5a types, SDF‑1 has been shown to attenuate responses during early axon projection but co-express Dcc during to repellents, such as Semaphorins or Slits, and so in later axon extension110; however, this requires functional motor axons, this molecule might attenuate the effects of confirmation. Semaphorins or of unidentified SM floor plate-derived It is currently unknown what molecules mediate chemorepellents124,125. the floor plate repulsion of SM axons110, because only BM and VM axons are repelled by netrin 1 and Slit pro- The exit point is a key guidepost. The next step in path- teins108,109. SM neurons do respond to SEMA3A, which is finding is projection to the exit point. For BM and VM expressed by the notochord and might have a function in axons, exit from the hindbrain appears to require the spacing apart the exiting abducens and hypoglossal SM presence of cranial sensory ganglia, which are apposed to axons108,123 (FIG. 2c). Cranial SM axon exit from the hind- the large dorsal exit points. Ablation of these structures brain depends on the chemokine SDF‑1 (also known as leads to a reduction in peripheral axon projections126.

866 | november 2007 | volume 8 www.nature.com/reviews/neuro © 2007 Nature Publishing Group

REVIEWS

The most proximal cells of the ganglion are specialized chemoattract trigeminal sensory axons137. Therefore, boundary cap cells which, in chicks, arise from a late- other neurotrophic factors are likely candidates in BM emigrating population of cadherin‑7-expressing neural and VM axon guidance. crest cells127,128. Transplantation experiments in chick The divergence of VM axons from their BM neigh- embryos demonstrated that, when an odd-numbered bours, for example, in the facial nerve, points to the rhombomere is rotated 180o, most BM axons reorien- existence of distinct VM guidance cues. Indeed, r5- tate to navigate rostrally to their correct exit points129. derived VM axons navigate to their parasympathetic However, evidence for an exit point-derived chemoat- ganglion targets by following chemoattractant signals; tractant has not been forthcoming, and the alar plate genetic removal of the sphenopalatine ganglion in itself is chemorepellent130, possibly due to a narrow Phox2a mutants results in the specific loss of the relevant stripe of Slit2 expression at the rhombic lip109. Thus, BM branch of the facial nerve138. The relevant chemoattract- and VM axon tracts might be hemmed in by floor plate ant molecules have not been identified, but the glial cell repellents medially and alar plate repellents dorsally. line-derived neurotrophic factor (GDNF) family of The sensory ganglia exert only a weak chemoattract- neurotrophic factors, which is implicated in the guid- ant influence on BM axons in vitro130, and the role of ance and arborization of autonomic neurons, might be boundary cap cells has yet to be explored, but this role involved139. is likely to be pivotal, judging by the intriguing finding that these cells limit spinal motor neuron emigration Motor pools and their target muscles from the neural tube131. Little is known about how cranial motor axons recog- nize their target structures at appropriate axial levels. A balance of attraction and repulsion in the periphery. For the spinal motor axon–limb system it has been pro- Cranial motor axon behaviour in the periphery depends posed that major guidance information comes from the on a balance of positive and negative influences. The lateral plate mesenchyme140. In the head, embryological perinotochordal mesenchyme, which is derived from the evidence strongly suggests that the neural crest from neural crest and differentiates into cartilage, repels spinal a particular axial level patterns muscles, the myogenic motor axons132. Although this role has not been directly progenitors of which migrate from the same level102. demonstrated for cranial SM axons, motor axons avoid Fate mapping has shown that neural crest cells form the forming cartilage of the branchial arches, and the connective tissue sheaths and skeletal attachment points parachordal cartilages, by channelling into the develop- around muscles derived from the same axial level42. The ing muscle plate103. Branchial arch neural crest cells also neural crest might thus provide specific guidance infor- express Semaphorins, including SEMA3A, and in mouse mation for BM axons and/or induce branchial muscles mutants that lack Sema3A or neuropilin 1, cranial motor to produce such guidance signals. However, the nature axons become defasciculated and invade inappropriate of these guidance signals is currently unknown. regions of the periphery133–135. The oculomotor nerve is Following the rostrocaudal inversion of an odd- normal in Sema3A mutants, consistent with the observa- numbered rhombomere, trigeminal BM axons that tion that oculomotor axons are not repelled by SEMA3A project to incorrect target muscles are eliminated, sug- in vitro108,134. Skeletogenic cells that express Semaphorins gesting a specific recognition between BM neurons and therefore influence peripheral axon pathways, and might their targets141. Studies in fish and chicks suggest that interact with growing nerves to influence the position- r2- and r3-derived regions of the trigeminal nucleus ing of foramina, which are the mature pathways contribute to separate subnuclei with different synaptic of the cranial nerves. Secreted Semaphorins, especially targets142,143. Rhombomere 2- and r3-derived trigeminal SEMA3A and SEMA3C, are expressed in particular motor neurons express high and low levels of ephrin subsets of cranial motor neurons, where their function A receptors, respectively, whereas r3 target muscle remains to be explored111,112. showed patterned ephrin A expression143,144. Expression Chemoattraction by the branchial arches can also of dominant-negative ephrin A receptors in r3 trigemi- strongly orient the trajectories of BM and VM axons130. nal motor neurons, or overexpression of ephrin As in Hepatocyte growth factor (HGF) is expressed by the their target, led to aberrant axon branching patterns, branchial muscle plate at the time of axon extension and suggesting a role for ephrins in this topographic axon accounts for most of this activity, as HGF presented on targeting143. beads is chemoattractant, and HGF-blocking antibodies can eliminate a large portion of the arch-mediated che- Extraocular muscle innervation moattraction. However, in HGF-mutant mice, BM and The six eye muscles that rotate the eyeball and pro- VM axon trajectories are normal, and only SM hypoglos- vide fine control of visual tracking movements are sal pathways are affected, suggesting that there are other innervated by three nerves. This is a promising sys- branchial arch-derived chemoattractants130. Indeed, at tem in which to study the mechanism that controls least three other neurotrophic factors are capable of nerve–muscle targeting, as well as a model that is of promoting cranial motor neuron outgrowth before they clinical relevance to humans. The abducens nerve become crucial factors for survival136. Brain-derived innervates the lateral rectus muscle, the trochlear Rhombic neurotrophic factor (BDNF) has a strong effect on motor nerve innervates the superior oblique (the dorsal 136 The structure at the dorsal axon outgrowth , is expressed in the branchial arches oblique in chicks) and the oculomotor nerve inner- extreme of the hindbrain. during early axon pathfinding, and has been shown to vates the remaining four muscles, the medial, dorsal

nature reviews | neuroscience volume 8 | november 2007 | 867 © 2007 Nature Publishing Group

REVIEWS

and ventral recti, and the inferior oblique. In chicks, and one is an axon guidance molecule (ROBO3), the lateral rectus primordium differentiates preco- supporting the notion of a developmental basis for ciously relative to the other , in a congenital CDDs. position ventral to r2 and r3 (Ref. 145). From embry- Mutations in HOXA1 have been shown to result in onic day three onwards, the abducens nerve extends two overlapping syndromes, Bosley-Salih-Alorainy syn- rostrally from its origin in r5 and r6, contacting the drome (BSAS) and Athabascan brainstem dysgenesis lateral rectus primordium by embryonic day four146. syndrome (ABDS)154. These disorders are character- A double immunohistochemical study contrasted the ized by impaired horizontal eye movements, ear and early abducens outgrowth with the delayed outgrowth vascular defects and, in some cases, autism or mental of the oculomotor nerve, which contacts its first and retardation. In Hoxa1-knockout mice, the abducens most distant muscle target, the ventral oblique (VO), nerve is lacking50, and so it is likely that abducens devel- by embryonic day five147. After the oculomotor nerve opment and consequent innervation of the lateral rectus reaches the VO, branches appear along the oculomotor muscle is aberrant in BSAS and ABDS patients. Clinical nerve to its other targets (the medial, dorsal and characterization of a loss of ocular motility, especially ventral recti). It is not known whether this occurs abduction (which is mediated by the abducens nerve), by interstitial branching of pre-existing axons or by broadly agrees with this interpretation159. Another de novo growth of neurons. However, FGF8 and the congenital CDD, type 2 (DURS2), Semaphorins have been shown to function in trochlear is also characterized by an absence or reduction of the and oculomotor initial projections. The dorsal exit of abducens nerve, whereas in some cases the oculomo- trochlear axons from the neuroepithelium depends tor nerve branches aberrantly into the lateral rectus on attraction by the midbrain–hindbrain boundary muscle160,161. This innervation pattern bears a striking and FGF8 (Ref. 148). SEMA3F is expressed rostral and resemblance to the pattern that is seen in Hoxa3- and caudal to the midbrain–hindbrain boundary and acts Hoxb3-mutant mice, which lack the abducens nerve through its receptor neuropilin 2 to guide the exit and and manifest aberrant innervation of the lateral rectus initial trajectory of trochlear and oculomotor nerves in through another nerve of uncertain origin71. However, both mice and chicks149–152. SEMA3F repels trochlear there is no published genetic evidence to link Hox3 or Congenital fibromatosis of the extraocular muscles axons in vitro, and in neuropilin‑2- or Sema3F-mutant other genes with DURS2. (CFEOM). A group of mice, trochlear axons fail to exit the brain correctly CFEOM2, a CFEOM variant which involves fibro- congenital syndromes that and the oculomotor nerve is defasciculated. The role sis of the extraocular muscles and severe loss of ocular involve cranial nerve miswiring of SEMA3F in extraocular innervation thus provides motility, has been linked to mutations in PHOX2A155. and paralysis or paresis of the an interesting counterpart to the role of SEMA3A in Neuroimaging data show that patients lack the ocu- extraocular muscles, often 162 associated with drooping of the guiding the branchiomotor nerves. Guidance cues for lomotor and trochlear nerves , a phenotype that 19 upper eyelid. later growth and branching and for targeting of the is identical to that of Phox2a-mouse mutants . This nerves III, IV and VI to specific extraocular muscles suggests that the primary defect is neural, and that the Duane syndrome remain to be characterized. extraocular muscle atrophy that occurs in CFEOM A congenital disorder characterized by might occur secondarily to the lack of innervation. impeded horizontal eye Dysinnervation disorders in humans Similarly, patients with CFEOM1 were found to have movements that resut from It will be vital to unravel in more detail the principles hypoplasia of the oculomotor nerve, and in some cases miswiring of the eye muscles. that underlie extraocular muscle wiring, in order of the abducens nerve163. CFEOM1 has been shown to to understand a group of human syndromes termed result from mutations in the kinesin motor protein Horizontal gaze palsy with 6,153 156 progressive scoliosis cranial dysinnervation disorders (CDDs ). These KIF21A , which has a role in anterograde axonal 164 (HGPPS). A rare congenital disorders are characterized most notably by deficits transport . syndrome that is characterized of horizontal eye movements (complex ), The rare syndrome HGPPS involves the absence of by the absence of conjugate and include defects such as congenital fibromatosis the conjugate horizontal eye movements that are medi- horizontal eye movements and of the extraocular muscles (CFEOM) Duane syndrome by deformities in the spine. , , ated by the abducens and oculomotor nerves, as well as horizontal gaze palsy with progressive scoliosis (HGPPS), scoliosis, and is caused by mutations in ROBO3 (Ref. 158). Möbius syndrome Möbius syndrome and Marcus Gunn syndrome. In some ROBO3 is required for axons to cross the midline and A rare congenital disorder cases these disorders are congenital, and it has been form commissures116, which are absent in the caused by abnormal proposed that they are caused primarily by defects in of HGPPS patients. It is likely that horizontal eye move- development of the cranial nerves, which results in paresis cranial motor neuron development and axon naviga- ment disturbances result from the failure of supranuclear or paralysis of the facial tion. Strabismus (squinting) often results from an tracts, such as the paramedian pontine reticular forma- muscles and, in some cases, imbalance in the function of the lateral (innervated by tion, to cross the midline and innervate the abducens other abnormalities. the abducens) and medial (innervated by oculomotor) and oculomotor nuclei158.

Marcus Gunn syndrome recti muscles, which respectively abduct and adduct A combination of clinical genetics and developmen- Also known as jaw-winking the eyeball. Familial studies, including linkage screens, tal neurobiology might in future further illuminate syndrome. It consists of an and mapping to candidate genetic loci have implicated the causes of CDDs in humans. In some CDDs, such elevation or depression of the five different genes in five separate syndromes. These as Möbius syndrome, an association with autism has eyelid on chewing and/or are HOXA1, PHOX2A, SALL4, KIF21A and ROBO3 been proposed165, suggesting that disorders of cranial suckling, and is thought to be 154–158 caused by aberrant innervation (also known as RIG1) . Three of these genes motor neuron development might have far-reaching of branches of the trigeminal – HOXA1, PHOX2A and SALL4 – are transcription significance for understanding human disorders of and oculomotor nerves. factors, one is involved in axonal transport (KIF21A) brain wiring.

868 | november 2007 | volume 8 www.nature.com/reviews/neuro © 2007 Nature Publishing Group

REVIEWS

Future prospects and challenges arch muscles must be elucidated; recent studies point Great progress been made in understanding how rostro- to differences in the molecular programme of differen- caudal and dorsoventral patterning processes regulate tiation in subpopulations of head mesoderm cells166,167. cranial motor neuron specification. However, there is a It is also currently unknown which combinations of sizeable gap in our knowledge that concerns how combi- neurotrophic factors maintain the survival of different nations of transcription factors govern the expression of cranial subpopulations; such factors might also have a axon guidance receptors and motor axon pathway deci- role in axon guidance and branching. Understanding sions. There are few characterized head-specific guid- the molecular hierarchy of development in motor neu- ance cues, which stems from the lack of basic knowledge rons is an exciting and compelling problem, with many about tissue patterning in the head. Glial cells are likely ramifications for human health. We need to elucidate to have a key role in cranial motor axon guidance, but the specific genetic hierarchies, signalling cascades and the idea that they might be molecularly heterogeneous pathfinding strategies of the cranial nerves in order to has not been addressed. Similarly, the specific charac- address clinical problems, such as cranial nerve palsies, ter of the extraocular versus the tongue and branchial and pan-motor neuron diseases.

1. Price, S. R. & Briscoe, J. The generation and 20. Lumsden, A. & Krumlauf, R. Patterning the vertebrate neuron development in the zebrafish hindbrain. diversification of spinal motor neurons: signals and neuraxis. Science 274, 1109–1115 (1996). Dev. Biol. 271, 119–129 (2004). responses. Mech. Dev. 121, 1103–1115 (2004). 21. Wilkinson, D. G., Bhatt, S., Chavrier, P., Bravo, R. & 39. Krumlauf, R. et al. Hox homeobox genes and 2. Briscoe, J. & Novitch, B. G. Regulatory pathways Charnay, P. Segment-specific expression of a regionalisation of the . J. Neurobiol. linking progenitor patterning, cell fates and zinc-finger gene in the developing nervous system of 24, 1328–1340 (1993). neurogenesis in the ventral neural tube. Philos. Trans. the mouse. Nature 337, 461–464 (1989). 40. Hunt, P., Wilkinson, D. & Krumlauf, R. Patterning the R. Soc. Lond. B Biol. Sci. 5 Feb 2007 (doi:10.1098/ 22. Schneider-Maunoury, S., Seitanidou, T., Charnay, P. & vertebrate head: murine Hox 2 genes mark distinct rstb.2006.2012). Lumsden, A. Segmental and neuronal architecture of subpopulations of premigratory and migrating cranial 3. Kuratani, S. & Tanaka, S. Peripheral development of the hindbrain of Krox‑20 mouse mutants. neural crest. Development 112, 43–50 (1991). avian trigeminal nerves. Am. J. Anat. 187, 65–80 Development 124, 1215–1226 (1997). 41. Graham, A., Maden, M. & Krumlauf, R. The murine (1990). 23. Cordes, S. P. & Barsh, G. S. The mouse segmentation Hox‑2 genes display dynamic dorsoventral patterns of 4. Cordes, S. P. Molecular genetics of cranial nerve gene kr encodes a novel basic domain- expression during development in mouse. Nature Rev. Neurosci. 2, transcription factor. Cell 79, 1025–1034 (1994). development. Development 112, 255–264 (1991). 611–623 (2001). 24. Moens, C. B., Cordes, S. P., Giorgianni, M. W., 42. Kontges, G. & Lumsden, A. Rhombencephalic neural 5. Chandrasekhar, A. Turning heads: development of Barsh, G. S. & Kimmel, C. B. Equivalence in the crest segmentation is preserved throughout craniofacial vertebrate branchiomotor neurons. Dev. Dyn. 229, genetic control of hindbrain segmentation in fish and ontogeny. Development 122, 3229–3242 (1996). 143–161 (2004). mouse. Development 125, 381–391 (1998). 43. Prince, V. & Lumsden, A. Hoxa‑2 expression in normal 6. Engle, E. C. The genetic basis of complex strabismus. 25. McKay, I. J., Lewis, J. & Lumsden, A. Organization and and transposed rhombomeres: independent Pediatr. Res. 59, 343–348 (2006). development of facial motor neurons in the kreisler regulation in the neural tube and neural crest. 7. Lumsden, A. & Keynes, R. Segmental patterns of mutant mouse. Eur. J. Neurosci. 9, 1499–1506 Development 120, 911–923 (1994). neuronal development in the chick hindbrain. (1997). 44. Gendron-Maguire, M., Mallo, M., Zhang, M. & Gridley, T. Nature 337, 424–428 (1989). 26. McGinnis, W. & Krumlauf, R. Homeobox genes and Hoxa‑2 mutant mice exhibit homeotic transformation This landmark paper describes for the first time axial patterning. Cell 68, 283–302 (1992). of skeletal elements derived from . the segmental organization of the hindbrain into 27. Greer, J. M., Puetz, J., Thomas, K. R. & Capecchi, Cell 75, 1317–1331 (1993). rhombomeres, including the segmental M. R. Maintenance of functional equivalence during 45. Rijli, F. M. et al. A homeotic transformation is arrangement of cranial motor neurons. paralogous Hox gene evolution. Nature 403, generated in the rostral branchial region of the head 8. Clarke, J. D. & Lumsden, A. Segmental repetition of 661–665 (2000). by disruption of Hoxa‑2, which acts as a selector gene. neuronal phenotype sets in the chick embryo 28. Glover, J. C., Renaud, J. S. & Rijli, F. M. Retinoic acid Cell 75, 1333–1349 (1993). hindbrain. Development 118, 151–162 (1993). and hindbrain patterning. J. Neurobiol. 66, 705–725 46. Gavalas, A., Davenne, M., Lumsden, A., Chambon, P. & 9. Lumsden, A. Segmentation and compartition in the (2006). Rijli, F. M. Role of Hoxa‑2 in axon pathfinding and early avian hindbrain. Mech. Dev. 121, 1081–1088 29. Rhinn, M., Picker, A. & Brand, M. Global and local rostral hindbrain patterning. Development 124, (2004). mechanisms of forebrain and midbrain patterning. 3693–3702 (1997). 10. Jacob, J., Hacker, A. & Guthrie, S. Mechanisms and Curr. Opin. Neurobiol. 16, 5–12 (2006). 47. Jungbluth, S., Bell, E. & Lumsden, A. Specification of molecules in motor neuron specification and axon 30. Irving, C. & Mason, I. Signalling by FGF8 from the distinct motor neuron identities by the singular pathfinding. Bioessays 23, 582–595 (2001). isthmus patterns anterior hindbrain and establishes activities of individual Hox genes. Development 126, 11. Gilland, E. & Baker, R. Conservation of neuroepithelial the anterior limit of Hox gene expression. 2751–2758 (1999). and mesodermal segments in the embryonic vertebrate Development 127, 177–186 (2000). 48. Davenne, M. et al. Hoxa2 and Hoxb2 control head. Acta Anat. (Basel) 148, 110–123 (1993). 31. Niederreither, K., Vermot, J., Schuhbaur, B., dorsoventral patterns of neuronal development in the 12. Jacob, J. & Guthrie, S. Facial visceral motor neurons Chambon, P. & Dolle, P. Retinoic acid synthesis and rostral hindbrain. Neuron 22, 677–691 (1999). display specific rhombomere origin and axon hindbrain patterning in the mouse embryo. 49. Carpenter, E. M., Goddard, J. M., Chisaka, O., Manley, pathfinding behavior in the chick. J. Neurosci. 20, Development 127, 75–85 (2000). N. R. & Capecchi, M. R. Loss of Hox‑A1 (Hox‑1.6) 7664–7671 (2000). 32. Dupe, V. & Lumsden, A. Hindbrain patterning involves function results in the reorganization of the murine 13. Moens, C. B. & Prince, V. E. Constructing the graded responses to retinoic acid signalling. hindbrain. Development 118, 1063–1075 (1993). hindbrain: insights from the zebrafish. Dev. Dyn. 224, Development 128, 2199–2208 (2001). 50. Mark, M. et al. Two rhombomeres are altered in Hoxa‑1 1–17 (2002). 33. Morrison, A., Ariza-McNaughton, L., Gould, A., mutant mice. Development 119, 319–338 (1993). 14. Simon, H., Guthrie, S. & Lumsden, A. Regulation of Featherstone, M. & Krumlauf, R. HOXD4 and 51. Barrow, J. R., Stadler, H. S. & Capecchi, M. R. Roles of SC1/DM-GRASP during the migration of motor regulation of the group 4 paralog genes. Development Hoxa1 and Hoxa2 in patterning the early hindbrain of neurons in the chick embryo brain stem. J. Neurobiol. 124, 3135–3146 (1997). the mouse. Development 127, 933–944 (2000). 25, 1129–1143 (1994). 34. Gould, A., Itasaki, N. & Krumlauf, R. Initiation of 52. Goddard, J. M., Rossel, M., Manley, N. R. & 15. Simon, H. & Lumsden, A. Rhombomere-specific origin rhombomeric Hoxb4 expression requires induction by Capecchi, M. R. Mice with targeted disruption of of the contralateral vestibulo-acoustic efferent neurons somites and a retinoid pathway. Neuron 21, 39–51 Hoxb‑1 fail to form the motor nucleus of the VIIth and their migration across the embryonic midline. (1998). nerve. Development 122, 3217–3228 (1996). Neuron 11, 209–220 (1993). 35. Sockanathan, S. & Jessell, T. M. Motor neuron-derived 53. Studer, M., Lumsden, A., Ariza-McNaughton, L., 16. Agarwala, S., Sanders, T. A. & Ragsdale, C. W. Sonic retinoid signaling specifies the subtype identity of Bradley, A. & Krumlauf, R. Altered segmental identity hedgehog control of size and shape in midbrain spinal motor neurons. Cell 94, 503–514 (1998). and abnormal migration of motor neurons in mice pattern formation. Science 291, 2147–2150 (2001). 36. Wilson, L., Gale, E., Chambers, D. & Maden, M. lacking Hoxb‑1. Nature 384, 630–634 (1996). 17. Agarwala, S. & Ragsdale, C. W. A role for midbrain Retinoic acid and the control of dorsoventral This paper, together with reference 52, shows that arcs in nucleogenesis. Development 129, 5779–5788 patterning in the avian spinal cord. Dev. Biol. 269, facial branchiomotor (FBM) neurons in Hoxb1- (2002). 433–446 (2004). mutant mice fail to differente and migrate and 18. Sanders, T. A., Lumsden, A. & Ragsdale, C. W. Arcuate 37. Guidato, S., Barrett, C. & Guthrie, S. Patterning of eventually die. plan of chick midbrain development. J. Neurosci. 22, motor neurons by retinoic acid in the chick embryo 54. Bell, E., Wingate, R. J. & Lumsden, A. Homeotic 10742–10750 (2002). hindbrain in vitro. Mol. Cell. Neurosci. 23, 81–95 transformation of rhombomere identity after localized 19. Pattyn, A., Morin, X., Cremer, H., Goridis, C. & Brunet, (2003). Hoxb1 misexpression. Science 284, 2168–2171 (1999). J. F. Expression and interactions of the two closely 38. Begemann, G., Marx, M., Mebus, K., Meyer, A. & This paper elegantly demonstrates that related homeobox genes Phox2a and Phox2b during Bastmeyer, M. Beyond the neckless phenotype: misexpression of Hoxb1 in presumptive trigeminal neurogenesis. Development 124, 4065–4075 (1997). influence of reduced retinoic acid signaling on motor motor neurons converts them to an FBM fate.

nature reviews | neuroscience volume 8 | november 2007 | 869 © 2007 Nature Publishing Group

REVIEWS

55. Arenkiel, B. R., Gaufo, G. O. & Capecchi, M. R. Hoxb1 generating different neuronal types in response to This important paper shows the sequential neural crest preferentially form glia of the PNS. Dev. SHH signalling. This model provides a conceptual generation of BM, VM and serotonergic neurons Dyn. 227, 379–386 (2003). framework for understanding motor neuron from the V3 domain in the hindbrain, and the 56. Arenkiel, B. R., Tvrdik, P., Gaufo, G. O. & Capecchi, differentiation in the hindbrain. genetic control mechanisms that control this M. R. Hoxb1 functions in both motoneurons and in 77. Shirasaki, R. & Pfaff, S. L. Transcriptional codes and process at different axial levels. tissues of the periphery to establish and maintain the the control of neuronal identity. Annu. Rev. Neurosci. 100. Samad, O. A. et al. Integration of anteroposterior and proper neuronal circuitry. Genes Dev. 18, 1539–1552 25, 251–281 (2002). dorsoventral regulation of Phox2b transcription in (2004). 78. Briscoe, J. & Ericson, J. The specification of neuronal cranial motoneuron progenitors by homeodomain This intriguing paper uses conditional deletion of identity by graded Sonic hedgehog signalling. Semin. proteins. Development 131, 4071–4083 (2004). Hoxb1 in the neural crest to show that r4 neural- Cell Dev. Biol. 10, 353–362 (1999). This is an important paper in which the genes that crest-derived Schwann cells have a role in the 79. Pattyn, A., Vallstedt, A., Dias, J. M., Sander, M. & regulate the patterning of the dorsoventral and maintenance and branching of FBM neurons. Ericson, J. Complementary roles for Nkx6 and Nkx2 rostrocaudal axes are found to converge in 57. Gavalas, A., Ruhrberg, C., Livet, J., Henderson, C. E. & class proteins in the establishment of motoneuron regulating PHOX2B expression and BM and VM Krumlauf, R. Neuronal defects in the hindbrain of identity in the hindbrain. Development 130, neuron differentiation in r4. Hoxa1, Hoxb1 and Hoxb2 mutants reflect regulatory 4149–4159 (2003). 101. Noden, D. M. Interactions and fates of avian interactions among these Hox genes. Development 80. Ericson, J. et al. Pax6 controls progenitor cell identity craniofacial mesenchyme. Development 103 (Suppl.) 130, 5663–5679 (2003). and neuronal fate in response to graded Shh signaling. 121–140 (1988). 58. Gavalas, A. et al. Hoxa1 and Hoxb1 synergize in Cell 90, 169–180 (1997). 102. Noden, D. M. & Trainor, P. A. Relations and patterning the hindbrain, cranial nerves and second 81. Briscoe, J. et al. Homeobox gene Nkx2.2 and interactions between cranial mesoderm and neural . Development 125, 1123–1136 specification of neuronal identity by graded Sonic crest populations. J. Anat. 207, 575–601 (2005). (1998). hedgehog signalling. Nature 398, 622–627 (1999). 103. Hacker, A. & Guthrie, S. A distinct developmental 59. Rossel, M. & Capecchi, M. R. Mice mutant for both 82. Sander, M. et al. Ventral neural patterning by Nkx programme for the cranial paraxial mesoderm in the Hoxa1 and Hoxb1 show extensive remodeling of the homeobox genes: Nkx6.1 controls somatic motor chick embryo. Development 125, 3461–3472 hindbrain and defects in craniofacial development. neuron and ventral interneuron fates. Genes Dev. 14, (1998). Development 126, 5027–5040 (1999). 2134–2139 (2000). 104. McClearn, D. & Noden, D. M. Ontogeny of 60. Maconochie, M. K. et al. Cross-regulation in the 83. Vallstedt, A. et al. Different levels of repressor activity architectural complexity in embryonic quail visceral mouse HoxB complex: the expression of Hoxb2 in assign redundant and specific roles to Nkx6 genes in arch muscles. Am. J. Anat. 183, 277–293 (1988). rhombomere 4 is regulated by Hoxb1. Genes Dev. 11, motor neuron and interneuron specification. Neuron 105. Guthrie, S. & Pini, A. Chemorepulsion of developing 1885–1895 (1997). 31, 743–755 (2001). motor axons by the floor plate. Neuron 14, 61. Tumpel, S. et al. Expression of Hoxa2 in rhombomere 4 84. Pabst, O., Rummelies, J., Winter, B. & Arnold, H. H. 1117–1130 (1995). is regulated by a conserved cross-regulatory Targeted disruption of the homeobox gene Nkx2.9 106. Tucker, A., Lumsden, A. & Guthrie, S. Cranial motor mechanism dependent upon Hoxb1. Dev. Biol. 302, reveals a role in development of the spinal accessory axons respond differently to the floor plate and 646–660 (2007). nerve. Development 130, 1193–1202 (2003). sensory ganglia in collagen gel co-cultures. 62. Pata, I. et al. The transcription factor GATA3 is a 85. Muller, M., Jabs, N., Lorke, D. E., Fritzsch, B. & Eur. J. Neurosci. 8, 906–916 (1996). downstream effector of Hoxb1 specification in Sander, M. Nkx6.1 controls migration and axon 107. Colamarino, S. A. & Tessier-Lavigne, M. The axonal rhombomere 4. Development 126, 5523–5531 pathfinding of cranial branchio-motoneurons. chemoattractant netrin‑1 is also a chemorepellent for (1999). Development 130, 5815–5826 (2003). trochlear motor axons. Cell 81, 621–629 (1995). 63. Song, M. R. et al. T‑box transcription factor Tbx20 86. Osumi, N. et al. Pax‑6 is involved in the specification 108. Varela-Echavarria, A., Tucker, A., Puschel, A. W. & regulates a genetic program for cranial motor neuron of hindbrain motor neuron subtype. Development Guthrie, S. Motor axon subpopulations respond cell body migration. Development 133, 4945–4955 124, 2961–2972 (1997). differentially to the chemorepellents netrin‑1 and (2006). 87. Novitch, B. G., Chen, A. I. & Jessell, T. M. Coordinate semaphorin D. Neuron 18, 193–207 (1997). 64. Moens, C. B. & Selleri, L. Hox cofactors in vertebrate regulation of motor neuron subtype identity and This paper presents in vitro evidence for a development. Dev. Biol. 291, 193–206 (2006). pan-neuronal properties by the bHLH repressor Olig2. differential effect of netrin 1 and SEMA3A, with 65. Cooper, K. L., Leisenring, W. M. & Moens, C. B. Neuron 31, 773–789 (2001). netrin 1 repelling only BM and VM axons and Autonomous and nonautonomous functions for 88. Lu, Q. R. et al. Common developmental requirement SEMA3A repelling BM and VM and somatic cranial Hox/Pbx in branchiomotor neuron development. for Olig function indicates a motor neuron/ motor neuron axons. Dev. Biol. 253, 200–213 (2003). oligodendrocyte connection. Cell 109, 75–86 (2002). 109. Hammond, R. et al. Slit-mediated repulsion is a key 66. McClintock, J. M., Kheirbek, M. A. & Prince, V. E. 89. Tsuchida, T. et al. Topographic organization of regulator of motor axon pathfinding in the hindbrain. Knockdown of duplicated zebrafish genes embryonic motor neurons defined by expression of Development 132, 4483–4495 (2005). reveals distinct roles in hindbrain patterning and a LIM homeobox genes. Cell 79, 957–970 (1994). This study uses a combination of electroporation in novel mechanism of duplicate gene retention. 90. Tanabe, Y., William, C. & Jessell, T. M. Specification of chicks, in vitro assays and, importantly, in vivo Development 129, 2339–2354 (2002). motor neuron identity by the MNR2 homeodomain evidence, to demonstrate a role for Slit–ROBO 67. Waskiewicz, A. J., Rikhof, H. A. & Moens, C. B. protein. Cell 95, 67–80 (1998). signalling in the repulsion of BM and VM axons Eliminating zebrafish pbx proteins reveals a hindbrain 91. Varela-Echavarria, A., Pfaff, S. L. & Guthrie, S. away from the midline and towards their exit ground state. Dev. Cell 3, 723–733 (2002). Differential expression of LIM homeobox genes among points. 68. Popperl, H. et al. Segmental expression of Hoxb‑1 is motor neuron subpopulations in the developing chick 110. Barrett, C. & Guthrie, S. Expression patterns of the controlled by a highly conserved autoregulatory loop brain stem. Mol. Cell. Neurosci. 8, 242–257 (1996). netrin receptor UNC5H1 among developing motor dependent upon exd/pbx. Cell 81, 1031–1042 92. William, C. M., Tanabe, Y. & Jessell, T. M. Regulation neurons in the embryonic rat hindbrain. Mech. Dev. (1995). of motor neuron subtype identity by repressor activity 106, 163–166 (2001). 69. Manzanares, M. et al. Segmental regulation of Hoxb‑3 of Mnx class homeodomain proteins. Development 111. Chilton, J. K. & Guthrie, S. Cranial expression of class by kreisler. Nature 387, 191–195 (1997). 130, 1523–1536 (2003). 3 secreted semaphorins and their neuropilin 70. Manzanares, M. et al. Conserved and distinct roles of 93. Thaler, J. et al. Active suppression of interneuron receptors. Dev. Dyn. 228, 726–733 (2003). kreisler in regulation of the paralogous Hoxa3 and programs within developing motor neurons revealed 112. Melendez-Herrera, E. & Varela-Echavarria, A. Hoxb3 genes. Development 126, 759–769 (1999). by analysis of homeodomain factor HB9. Neuron 23, Expression of secreted semaphorins and their 71. Gaufo, G. O., Thomas, K. R. & Capecchi, M. R. Hox3 675–687 (1999). receptors in specific neuromeres, boundaries, and genes coordinate mechanisms of genetic suppression 94. Sharma, K. et al. LIM homeodomain factors Lhx3 and neuronal groups in the developing mouse and chick and activation in the generation of branchial and Lhx4 assign subtype identities for motor neurons. brain. Brain Res. 1067, 126–137 (2006). somatic motoneurons. Development 130, 5191–5201 Cell 95, 817–828 (1998). 113. Kennedy, T. E., Serafini, T., de la Torre, J. R. & (2003). 95. Kraus, F., Haenig, B. & Kispert, A. Cloning and Tessier-Lavigne, M. Netrins are diffusible chemotropic 72. Guidato, S., Prin, F. & Guthrie, S. Somatic expression analysis of the mouse T‑box gene Tbx20. factors for commissural axons in the embryonic spinal motoneurone specification in the hindbrain: the Mech. Dev. 100, 87–91 (2001). cord. Cell 78, 425–435 (1994). influence of somite-derived signals, retinoic acid and 96. Brunet, J. F. & Pattyn, A. Phox2 genes - from 114. Kennedy, T. E., Wang, H., Marshall, W. & Hoxa3. Development 130, 2981–2996 (2003). patterning to connectivity. Curr. Opin. Genet. Dev. 12, Tessier-Lavigne, M. Axon guidance by diffusible 73. Jessell, T. M. Neuronal specification in the spinal cord: 435–440 (2002). chemoattractants: a gradient of netrin protein in the inductive signals and transcriptional codes. Nature 97. Pattyn, A., Hirsch, M., Goridis, C. & Brunet, J. F. developing spinal cord. J. Neurosci. 26, 8866–8874 Rev. Genet. 1, 20–29 (2000). Control of hindbrain motor neuron differentiation by (2006). 74. Litingtung, Y. & Chiang, C. Control of Shh activity and the homeobox gene Phox2b. Development 127, 115. Brose, K. et al. Slit proteins bind Robo receptors and signaling in the neural tube. Dev. Dyn. 219, 143–154 1349–1358 (2000). have an evolutionarily conserved role in repulsive axon (2000). This paper uses knockout mice to show that guidance. Cell 96, 795–806 (1999). 75. Stamataki, D., Ulloa, F., Tsoni, S. V., Mynett, A. & PHOX2B has an important role in controlling the 116. Sabatier, C. et al. The divergent Robo family protein Briscoe, J. A gradient of Gli activity mediates graded differentiation of BM and VM neuronal types in rig‑1/Robo3 is a negative regulator of slit Sonic hedgehog signaling in the neural tube. Genes the hindbrain. In these mice, the neurons fail to responsiveness required for midline crossing by Dev. 19, 626–641 (2005). differentiate and subsequently die. commissural axons. Cell 117, 157–169 (2004). 76. Briscoe, J., Pierani, A., Jessell, T. M. & Ericson, J. A 98. Coppola, E., Pattyn, A., Guthrie, S. C., Goridis, C. & 117. Serafini, T. et al. Netrin‑1 is required for commissural homeodomain protein code specifies progenitor cell Studer, M. Reciprocal gene replacements reveal axon guidance in the developing vertebrate nervous identity and neuronal fate in the ventral neural tube. unique functions for Phox2 genes during neural system. Cell 87, 1001–1014 (1996). Cell 101, 435–445 (2000). differentiation. EMBO J. 24, 4392–4403 (2005). 118. Burgess, R. W., Jucius, T. J. & Ackerman, S. L. Motor This important study shows that progenitor 99. Pattyn, A. et al. Coordinated temporal and spatial axon guidance of the mammalian trochlear and domains along the dorsoventral axis of the spinal control of motor neuron and serotonergic neuron phrenic nerves: dependence on the netrin receptor cord that express different combinations of generation from a common pool of CNS progenitors. Unc5c and modifier loci. J. Neurosci. 26, 5756–5766 homeodomain proteins are of key importance in Genes Dev. 17, 729–737 (2003). (2006).

870 | november 2007 | volume 8 www.nature.com/reviews/neuro © 2007 Nature Publishing Group

REVIEWS

119. Engelkamp, D. Cloning of three mouse Unc5 genes 138. Jacob, J., Tiveron, M. C., Brunet, J. F. & Guthrie, S. 157. Yamada, K. et al. Heterozygous mutations of the and their expression patterns at mid-gestation. Role of the target in the pathfinding of facial visceral kinesin KIF21A in congenital fibrosis of the extraocular Mech. Dev. 118, 191–197 (2002). motor axons. Mol. Cell. Neurosci. 16, 14–26 (2000). muscles type 1 (CFEOM1). Nature Genet. 35, 120. Hong, K. et al. A ligand-gated association between 139. Young, H. M., Anderson, R. B. & Anderson, C. R. 318–321 (2003). cytoplasmic domains of UNC5 and DCC family Guidance cues involved in the development of the 158. Jen, J. C. et al. Mutations in a human ROBO gene receptors converts netrin-induced growth cone peripheral . Auton. disrupt hindbrain axon pathway crossing and attraction to repulsion. Cell 97, 927–941 (1999). Neurosci. 112, 1–14 (2004). morphogenesis. Science 304, 1509–1513 (2004). 121. Keleman, K. & Dickson, B. J. Short- and long-range 140. Tosney, K. W. & Landmesser, L. T. Pattern and 159. Bosley, T. M. et al. Clinical characterization of the repulsion by the Drosophila Unc5 netrin receptor. specificity of axonal outgrowth following varying HOXA1 syndrome BSAS variant. 69, Neuron 32, 605–617 (2001). degrees of chick limb bud ablation. J. Neurosci. 4, 1245–1253 (2007). 122. Labrador, J. P. et al. The homeobox transcription 2518–2527 (1984). 160. Demer, J. L., Ortube, M. C., Engle, E. C. & Thacker, N. factor even-skipped regulates netrin-receptor 141. Warrilow, J. & Guthrie, S. Rhombomere origin plays a High-resolution magnetic resonance imaging expression to control dorsal motor-axon projections in role in the specificity of cranial motor axon demonstrates abnormalities of motor nerves and Drosophila. Curr. Biol. 15, 1413–1419 (2005). projections in the chick. Eur. J. Neurosci. 11, extraocular muscles in patients with neuropathic 123. Anderson, C. N. et al. Molecular analysis of axon 1403–1413 (1999). strabismus. J. Aapos 10, 135–142 (2006). repulsion by the notochord. Development 130, 142. Higashijima, S., Hotta, Y. & Okamoto, H. Visualization 161. Demer, J. L., Clark, R. A., Lim, K. H. & Engle, E. C. 1123–1133 (2003). of cranial motor neurons in live transgenic zebrafish Magnetic resonance imaging evidence for 124. Lieberam, I., Agalliu, D., Nagasawa, T., Ericson, J. & expressing green fluorescent protein under the control widespread orbital dysinnervation in dominant Jessell, T. M. A Cxcl12-Cxcr4 chemokine signaling of the Islet‑1 promoter/enhancer. J. Neurosci. 20, Duane’s retraction syndrome linked to the DURS2 pathway defines the initial trajectory of mammalian 206–218 (2000). locus. Invest. Ophthalmol. Vis. Sci. 48, 194–202 motor axons. Neuron 47, 667–679 (2005). 143. Prin, F., Ng, K. E., Thaker, U., Drescher, U. & (2007). 125. Chalasani, S. H., Sabelko, K. A., Sunshine, M. J., Guthrie, S. Ephrin-As play a rhombomere-specific role 162. Bosley, T. M. et al. Neurological features of congenital Littman, D. R. & Raper, J. A. A chemokine, SDF‑1, in trigeminal motor axon projections in the chick fibrosis of the extraocular muscles type 2 with reduces the effectiveness of multiple axonal repellents embryo. Dev. Biol. 279, 402–419 (2005). mutations in PHOX2A. Brain 129, 2363–2374 and is required for normal axon pathfinding. 144. Kury, P., Gale, N., Connor, R., Pasquale, E. & (2006). J. Neurosci. 23, 1360–1371 (2003). Guthrie, S. Eph receptors and ephrin expression in cranial 163. Demer, J. L., Clark, R. A. & Engle, E. C. Magnetic 126. Moody, S. A. & Heaton, M. B. Developmental motor neurons and the branchial arches of the chick resonance imaging evidence for widespread orbital relationships between trigeminal ganglia and embryo. Mol. Cell. Neurosci. 15, 123–140 (2000). dysinnervation in congenital fibrosis of extraocular trigeminal motoneurons in chick embryos. II. 145. Noden, D. M., Marcucio, R., Borycki, A. G. & muscles due to mutations in KIF21A. Invest. Ganglion axon ingrowth guides motoneuron migration. Emerson, C. P., Jr. Differentiation of avian craniofacial Ophthalmol. Vis. Sci. 46, 530–539 (2005). J. Comp. Neurol. 213, 344–349 (1983). muscles. I. Patterns of early regulatory gene 164. Marszalek, J. R., Weiner, J. A., Farlow, S. J., Chun, J. & 127. Niederlander, C. & Lumsden, A. Late emigrating expression and myosin heavy chain synthesis. Goldstein, L. S. Novel dendritic kinesin sorting neural crest cells migrate specifically to the exit points Dev. Dyn. 216, 96–112 (1999). identified by different process targeting of two related of cranial branchiomotor nerves. Development 122, 146. Wahl, C. M., Noden, D. M. & Baker, R. Developmental kinesins: KIF21A and KIF21B. J. Cell Biol. 145, 2367–2374 (1996). relations between sixth nerve motor neurons and their 469–479 (1999). 128. Ju, M. J., Aroca, P., Luo, J., Puelles, L. & Redies, C. targets in the chick embryo. Dev. Dyn. 201, 191–202 165. Miller, M. T. et al. Autism associated with conditions Molecular profiling indicates avian branchiomotor (1994). characterized by developmental errors in early nuclei invade the hindbrain alar plate. Neuroscience 147. Chilton, J. K. & Guthrie, S. Development of oculomotor embryogenesis: a mini review. Int. J. Dev. Neurosci. 128, 785–796 (2004). axon projections in the chick embryo. J. Comp. Neurol. 23, 201–219 (2005). 129. Guthrie, S. & Lumsden, A. Motor neuron pathfinding 472, 308–317 (2004). 166. Bothe, I. & Dietrich, S. The molecular setup of the following rhombomere reversals in the chick embryo 148. Irving, C., Malhas, A., Guthrie, S. & Mason, I. avian head mesoderm and its implication for hindbrain. Development 114, 663–673 (1992). Establishing the trochlear motor axon trajectory: role craniofacial myogenesis. Dev. Dyn. 235, 2845–2860 130. Caton, A. et al. The branchial arches and HGF are of the isthmic organiser and Fgf8. Development 129, (2006). growth-promoting and chemoattractant for cranial 5389–5398 (2002). 167. Porter, J. D. et al. Distinctive morphological and motor axons. Development 127, 1751–1766 149. Giger, R. J. et al. Neuropilin‑2 is required in vivo for gene/protein expression signatures during (2000). selective axon guidance responses to secreted myogenesis in novel cell lines from extraocular and 131. Vermeren, M. et al. Integrity of developing spinal semaphorins. Neuron 25, 29–41 (2000). hindlimb muscle. Physiol. Genomics 24, 264–275 motor columns is regulated by neural crest Together with references 150–152, this paper (2006). derivatives at motor exit points. Neuron 37, 403–415 establishes a role for SEMA3F–neuropilin 2 168. Trainor, P. A. & Krumlauf, R. Patterning the cranial (2003). interactions in the initial guidance of the neural crest: hindbrain segmentation and Hox gene 132. Tosney, K. W. & Oakley, R. A. The perinotochordal oculomotor and trochlear nerves, in this case using plasticity. Nature Rev. Neurosci. 1, 116–124 mesenchyme acts as a barrier to axon advance in the neuropilin 2 mouse mutants. (2000). chick embryo: implications for a general mechanism 150. Chen, H. et al. Neuropilin‑2 regulates the development 169. Lumsden, A., Sprawson, N. & Graham, A. Segmental of axonal guidance. Exp. Neurol. 109, 75–89 of selective cranial and sensory nerves and origin and migration of neural crest cells in the (1990). hippocampal mossy fiber projections. Neuron 25, hindbrain region of the chick embryo. Development 133. Puschel, A. W., Adams, R. H. & Betz, H. Murine 43–56 (2000). 113, 1281–1291 (1991). semaphorin D/collapsin is a member of a diverse gene 151. Sahay, A., Molliver, M. E., Ginty, D. D. & Kolodkin, 170. Santagati, F. & Rijli, F. M. Cranial neural crest and the family and creates domains inhibitory for axonal A. L. Semaphorin 3F is critical for development of building of the vertebrate head. Nature Rev. Neurosci. extension. Neuron 14, 941–948 (1995). circuitry and is required in neurons for 4, 806–818 (2003). 134. Taniguchi, M. et al. Disruption of Semaphorin III/D selective CNS axon guidance events. J. Neurosci. 23, gene causes severe abnormality in peripheral nerve 6671–6680 (2003). Acknowledgements projection. Neuron 19, 519–530 (1997). 152. Watanabe, Y., Toyoda, R. & Nakamura, H. Navigation Thanks to A. Lumsden and R. Knight for critical comments on Together with reference 133, this paper describes of trochlear motor axons along the midbrain-hindbrain the manuscript. the defects in cranial nerve fasciculation and boundary by neuropilin 2. Development 131, pathfinding in SEMA3A and neuropilin 1 mutants, 681–692 (2004). showing an important role for this ligand–receptor 153. Gutowski, N. J. Duane’s syndrome. Eur. J. Neurol. 7, DATABASES pair in peripheral nerve projections. 145–149 (2000). Gene: http://www.ncbi.nlm.nih.gov/entrez/query. 135. Kitsukawa, T. et al. Neuropilin-semaphorin III/D- 154. Tischfield, M. A. et al. Homozygous HOXA1 mutations fcgi?db=gene mediated chemorepulsive signals play a crucial role in disrupt human brainstem, inner ear, cardiovascular BDNF | cadherin‑7 | CXCR4 | CYP26 | DCC | EGR2 | FGF8 | peripheral nerve projection in mice. Neuron 19, and cognitive development. Nature Genet. 37, Gata2 | Gata3 | HGF | Hoxa1 | HOXA1 | Hoxa2 | Hoxa3 | Hoxb1 | 995–1005 (1997). 1035–1037 (2005). hoxb1a | Hoxb2 | Hoxb3 | Hoxd3 | Isl1 | Isl2 | KIF21A | Lhx3 | Lhx4 136. Naeem, A., Abbas, L. & Guthrie, S. Comparison of the 155. Nakano, M. et al. Homozygous mutations in ARIX | MAFB | MNR2 | MNX1 | netrin 1 | neuropilin 1 | neuropilin 2 | effects of HGF, BDNF, CT‑1, CNTF, and the branchial (PHOX2A) result in congenital fibrosis of the Nkx2.2 | Nkx2.9 | Nkx6.1 | Nkx6.2 | Olig2 | Pax6 | pbx4 | Phox2a | arches on the growth of embryonic cranial motor extraocular muscles type 2. Nature Genet. 29, PHOX2A | ROBO1 | ROBO2 | ROBO3 | ROBO3 | SALL4 | SDF‑1 neurons. J. Neurobiol. 51, 101–114 (2002). 315–320 (2001). 137. O’Connor, R. & Tessier-Lavigne, M. Identification of 156. Al-Baradie, R. et al. Duane radial ray syndrome | SEMA3A | SEMA3C | SEMA3F | Slit2 | SHH | Tbx20 | UNC5A | maxillary factor, a maxillary process-derived (Okihiro syndrome) maps to 20q13 and results from Unc5c chemoattractant for developing trigeminal sensory mutations in SALL4, a new member of the SAL family. ALL LINKS ARE ACTIVE IN THE ONLINE PDF axons. Neuron 24, 165–178 (1999). Am. J. Hum. Genet. 71, 1195–1199 (2002).

nature reviews | neuroscience volume 8 | november 2007 | 871 © 2007 Nature Publishing Group