Axonal Growth Abnormalities Underlying Ocular Cranial Nerve Disorders

Abstract

Abnormalities in cranial motor nerve development cause paralytic strabismus syndromes, collectively referred to as congenital cranial dysinnervation disorders (CCDDs), in which patients cannot fully move their eyes. These disorders can arise by two mechanisms: 1. defective motor neuron specification, usually by loss of a transcription factor necessary for brainstem patterning, or 2. axon growth and guidance abnormalities of the oculomotor, trochlear and abducens nerves. This review focuses on our current understanding of axon guidance mechanisms in the cranial motor nerves and how disease-causing mutations disrupt axon targeting. Abnormalities of axon growth and guidance are often limited to a single nerve or subdivision, even when the causative gene is ubiquitously expressed. Additionally, when one nerve is absent, its normal target muscles attract other motor neurons. Study of these disorders highlights the complexities of axon guidance and how each population of neurons uses a unique but overlapping set of axon guidance pathways.

INTRODUCTION

Binocular vision depends on precise control of the movements of each eye. In a group of rare disorders, collectively termed congenital cranial dysinnervation disorders (CCDDs), patients cannot fully move one or both eyes, due to defects in innervation of the extraocular muscles (EOMs) (Engle 2006). Identification of the genetic causes and mechanisms of these disorders has provided insight into the molecular mechanisms regulating axon growth and guidance of the cranial motor nerves.

Eye movements

Eye movements are mediated by six extraocular muscles on each eye, which are innervated by three cranial nerves. Horizontal eye movements are relatively straightforward: the medial rectus (MR) adducts the eye (moves it nasally), and the lateral rectus (LR) abducts the eye (moves it temporally). Vertical and torsional eye movements are more complicated, because the muscle axes are not in line with the visual axis in primary position. The vertical eye muscles therefore have primary, secondary and tertiary effects, depending on the horizontal position of the eyes. The superior rectus (SR) is primarily an elevator, with secondary and tertiary actions of intorsion (rotation of the vertical axis nasally) and adduction. The inferior rectus (IR) is primarily a depressor, with secondary and tertiary actions of extorsion (rotation of the vertical axis temporally) and adduction. The superior oblique (SO) is primarily an intorter, with secondary and tertiary actions of depression and abduction. The inferior oblique (IO) is primarily an extorter, with secondary and tertiary actions of elevation and abduction.

Ocular motor neurons

The oculomotor nerve (cranial nerve 3 [CN3]) innervates the MR, SR, IR, and IO, as well as the levator palpabrae superioris (LPS), which elevates the eyelid. The motor nucleus originates in the midbrain (just above the midbrain-hindbrain junction) and axons travel ventrally into the orbit before separating into the superior and inferior divisions (see Figure 1). The superior division innervates the SR and LPS, and the inferior division innervates the MR, IR, and IO. After axon outgrowth, cell bodies of the superior division neurons cross the midline in the midbrain; innervation to the SR is therefore contralateral. The trochlear nerve (cranial nerve 4 [CN4]) innervates the SO. The motor nucleus originates in the hindbrain (just below the midbrain-hindbrain junction) and axons exit the nucleus dorsally, cross the midline at the dorsal aspect of the hindbrain, then turn ventrally and travel to the orbit to innervate the contralateral SO. The abducens nerve (cranial nerve 6 [CN6]) originates in rhombomere 5 of the hindbrain and axons project caudally and ventrally to innervate the LR.

Figure 1:

Schematic of nerve trajectories for oculomotor, trochlear, and abducens nerves. The oculomotor nerve (3) originates in the midbrain, travels ventrally to the orbit, and divides into a superior branch (pink), which innervates superior rectus (SR) and levator (not pictured), and an inferior branch (purple), which innervates the medial rectus (MR), inferior rectus (IR), and inferior oblique (IO). The trochlear nerve (4) originates at the midbrain-hindbrain junction, travels dorsally and crosses the midline (dashes lines) and then travels ventrally to innervate the contralateral superior oblique (SO). The abducens nerve originates in the hindbrain and travels ventrally to innervate the lateral rectus (LR). Figure created with BioRender.com.

Congenital Cranial Dysinnervation Disorders

Strabismus refers to any misalignment of the eyes and includes esotropia (eyes deviate in), exotropia (eyes deviate out), and vertical misalignments. The majority of strabismus patients can fully move both eyes, but a subset have paralytic strabismus, and cannot move one or both eyes in specific directions. Paralytic strabismus syndromes are part of a group of disorders termed CCDDs. CCDDs can be subdivided by the cranial nerve(s) involved and can present as isolated eye movement disorders or as part of a larger syndrome, with other neurological and non-neurological features, depending on the specific genetic mutation.

Congenital fibrosis of the extraocular muscles (CFEOM) results from abnormal development of CN3 and CN4. Patients with CFEOM present with ptosis (droopy eyelids) and inability to elevate their eyes. This leads to a compensatory chin-up head posture. Horizontal movements range from full to severely limited. There are three types of CFEOM, originally distinguished clinically and now distinguished genetically (Graeber et al. 2013). CFEOM1 presents as an isolated eye movement disorder and is bilateral and typically symmetric. Patients always have significant ptosis and cannot elevate their eyes above the midline. It is inherited in an autosomal dominant fashion (Engle et al. 1997, Yamada et al. 2003). CFEOM2 is bilateral and presents with ptosis, exotropia, and significant limitation of both vertical and horizontal eye movements. Patients can have associated retinal problems, but do not have other neurological symptoms (Bosley et al. 2006, Khan et al. 2014a). It is inherited in an autosomal recessive fashion. CFEOM3 has the most variable presentation. As in the other forms of CFEOM, patients have deficits in eye elevation, but it can be unilateral or bilateral, and in rare cases does not involve ptosis (Figure 2). Depending on the specific genetic mutation, CFEOM3 presents as either an isolated eye movement disorder or a syndrome involving severe restrictions in eye movements, exotropia, other cranial neuropathies, intellectual and social disabilities, brain malformations, and peripheral neuropathy (Tischfield et al. 2010). It is inherited in an autosomal dominant or de novo fashion.

Figure 2:

Clinical Findings in CFEOM and DRS. A: External motility photos in a patient with CFEOM3 and a TUBB3 mutation. Top row: attempted upgaze. Middle row, left to right: attemped right gaze, primary position, attempted left gaze. Bottom row: attempted downgaze. Note the right-sided ptosis and inability to elevate the right eye. The left eye has an abduction deficit. Panel A adapted from (Whitman et al. 2016). B: Bilateral DRS. Note the limited abduction of each eye and the globe retraction on attempted adduction. Panel B adapted from (Whitman & Engle 2017).

Duane retraction syndrome (DRS) results from abnormal development of CN6. Patients with DRS are unable to fully abduct one or both eyes and have globe retraction (inward movement of the eye and narrowing of the lid fissure) on adduction (Figure 2). There is also often some degree of restriction of adduction. Patients can also have up-shoots or down-shoots of the affected eye with adduction. In primary gaze, patients can be esotropic or exotropic, and often have a compensatory head turn. It is most often unilateral but can be bilateral; when unilateral, most often the left eye is affected. DRS is the most common CCDD, affecting approximately 1:1000 individuals and is usually isolated, although it is also associated with several specific syndromes, including Duane Radial Ray syndrome, Goldenhar syndrome, and HOXA1 syndrome (Barry et al. 1993).

Other CCDDs involving abnormalities of horizontal eye movements include horizontal gaze palsy with progressive scoliosis (HGPPS) and Moebius syndrome. In HGPPS, patients cannot abduct or adduct either eye, but there is no globe retraction. Patients also develop a severe, progressive scoliosis (Jen et al. 2004). Moebius syndrome involves cranial nerves 6 and 7, leading to abduction deficits and facial weakness. Moebius patients have preserved vertical eye movements (MacKinnon et al. 2014). CCDDs can also present with paralytic strabismus that does not fall neatly into one of the above categories or varies between members of the same family. Superior oblique palsy could also qualify as a CCDD (see sidebar).

Genetic Causes of CCDDs

CCDDs were originally believed to result from primary deficits in the EOMs themselves, but identification of the causative genes has shown that they result from improper development of the cranial motor neurons, with a secondary fibrosis of uninnervated muscle. There are two main pathophysiologic mechanisms for CCDDs: defects of either (1) neuronal specification or (2) axon growth/guidance of the cranial motor nerves.

Neuronal specification

Brainstem patterning requires expression of a specific sequence of transcription factors that define brainstem regions. Loss-of-function mutations in these transcription factors have been identified in several CCDDs, and lead to loss of specific motor nuclei in the brainstem. Homozygous loss of function mutations in PHOX2A, necessary for motor neuron specification at the midbrain-hindbrain junction, result in CFEOM2 and congenital absence of CN3 and CN4 (and their motor nuclei) (Nakano et al. 2001). Homozygous loss of function mutations of HOXA1, which is required for hindbrain patterning, leads to a syndrome including bilateral DRS, sensorineural deafness, and variably facial weakness, central hypoventilation, vascular malformations, and intellectual disabilities (Tischfield et al. 2005). Haploinsufficiency of SALL4 disrupts development of the abducens nucleus and distal upper limb and leads to autosomal dominant Duane radial ray syndrome (DRS with variable hand and arm malformations) (Al-Baradie et al. 2002, Kohlhase et al. 2002). SALL4 is required for embryonic stem cell maintenance and complete loss is embryonic lethal very early in development (Sakaki-Yumoto et al. 2006). Isolated DRS can be caused by heterozygous loss-of-function mutations in the hindbrain transcription factor MAFB; heterozygous mutations with a dominant negative effect lead to DRS plus hearing loss (Park et al. 2016). A specific MAFB mutation has been found in individuals with DRS, hearing loss, and Focal Segmental Glomerulosclerosis (Sato et al. 2018).

Axon guidance defects

Axon guidance defects of CN3, CN4 or CN6 are the other cause of CCDDs and will be the main focus of this review. CFEOM1 results from missense mutations in KIF21A, a kinesin motor protein that transports molecular cargos along microtubules (Yamada et al. 2003). CFEOM3 results from specific missense mutations in TUBB3, the neuron-specific beta tubulin monomer, a component of microtubules (Tischfield et al. 2010), or from specific missense mutations in another beta tubulin monomer, TUBB2B (Cederquist et al. 2012). In both CFEOM1 and CFEOM3, there is abnormal axon guidance of CN3 and sometimes CN4. DRS can be caused by heterozygous mutations in CHN1, which encodes the Rac-GAP protein a-2-chimaerin, which acts downstream of multiple signaling molecules (Miyake et al. 2008). Homozygous mutations in the axon guidance receptor ROBO3 lead to HGPPS, by disrupting axon guidance of commissural axons (Jen et al. 2004). Mutations in several other genes, including ECEL-1, ACKR3, COL25A1, and TUBB6 lead to improper axon guidance of cranial motor neurons and variable CCDD phenotypes (Dieterich et al. 2013, Fazeli et al. 2017, Shinwari et al. 2015, Whitman et al. 2019).

NORMAL DEVELOPMENT AND AXON GUIDANCE

To correctly innervate the EOMs, developing axons must exit the brainstem, fasciculate into nerve bundles, traverse specific paths through the mesenchyme, enter the developing orbit, and contact the appropriate target muscle. Additionally, trochlear axons cross the midline before traversing the mesenchyme. The neurons are born between embryonic day (E)8.0 and 11.5; their axons begin exiting the brainstem as early as E9.5 (Easter et al. 1993), reach the developing orbit by E11.5, and extend towards the EOMs by E12.5 (Michalak et al. 2017). During these developmental processes, each axon terminates in a growth cone, which responds to environmental cues to determine directional growth (Mason & Erskine 2000). When a developing nerve is destined to turn, branch, or innervate a target, its axons’ advancing growth cones enter a transient decision region where the growth cones pause, enlarge, and follow short convoluted trajectories (Tosney & Landmesser 1985). CN3 has two decision regions in the orbit, an inferior decision region where the inferior division fibers branch to their respective EOMs, and a superior decision region, where the superior division branches off to innervate the SR and LPS (Cheng et al. 2014, Michalak et al. 2017). The successful exit of a growth cone from a decision region is dependent on both the cell-autonomous state of the motor neuron and local environmental cues (Kalil et al. 2000). The local cues that guide ocular motor axons could be expressed on the surface of cells along the trajectory and/or be diffusible short- or long-range signals secreted by the mesenchyme or the target EOMs.

Role of EOMs on cranial motor neuron axon guidance

Mice lacking the transcription factors Myf5 and Mrf4, which are upstream of MyoD in muscle precursor differentiation, do not develop EOMs; skeletal and facial muscles develop normally (Sambasivan et al. 2009). In the absence of EOMs, ocular motor axons initially target to the correct position within the orbit, indicating that muscle-derived cues are not required for initial axon guidance steps (Michalak et al. 2017). Within the orbit, however, the axons fail to make terminal branches, the superior division of CN3 does not branch off the main trunk, and the nerves eventually die back (Michalak et al. 2017). The muscles, therefore, provide molecular cues that influence the axon’s behavior in their final decision region. The identity of these cues is unknown and a subject for future study. Humans with homozygous loss of function of MYF5 have hypoplastic to absent EOMs and resulting ophthalmoplegia; their ocular motor nerves have not been examined (Di Gioia et al. 2018).

Semaphorin signaling

Class 3 Semaphorins are a class of secreted molecules with important axon guidance roles in the peripheral nervous system (reviewed in (Masuda & Taniguchi 2016). Sema3A is expressed in the mesenchyme surrounding the EOMs and Sema3C is expressed in the EOMs, particularly the LR (Ferrario et al. 2012). Sema3F is highly expressed in the midbrain and hindbrain, except along the trajectory of CN4 (Giger et al. 2000). Class 3 semaphorins bind to Neuropilins (Npn), which complex with PlexinAs to signal. Oculomotor neurons express Npn-1 and -2 (Chilton & Guthrie 2003) and PlexinA1 and PlexinA2 (Ferrario et al. 2012). Sema3A or Sema3C application leads to growth cone collapse of CN3 axons in vitro, and knock-down of PlexinA1 in chick embryos led to CN3 axon guidance defects, including defasiculation, target over-shoots, and abnormal branches to the LR (Ferrario et al. 2012). CN3 targeting is normal, however, in Sema3A and Sema3C knockout mice (Feiner et al. 2001, Taniguchi et al. 1997). Trochlear neurons also express both Npn-1 and -2 (Chilton & Guthrie 2003). Npn-2 binds Sema3F and Sema3C but not Sema3A (Chen et al. 1997), and trochlear axons are repelled by Sema3F (Giger et al. 2000). In Npn2 knockout mice, CN4 is absent and CN3 is defasciculated but projects to the normal target field. The trochlear motor neurons are present, but their projections are disorganized (Chen et al. 2000, Giger et al. 2000). Sema3F knock out mice have a strikingly similar phenotype: CN3 is defasciculated and CN4 is mostly absent (Sahay et al. 2003).

Abducens neurons express Npn-1, not Npn-2 (Chilton & Guthrie 2003). Loss of Npn-1 in motor neurons leads to a thin, defasciculated abducens nerve that targets correctly (Huettl & Huber 2011). CN6 axons are repelled by Sema3A but attracted to Sema3C. Given the high expression of Sema3C in LR, this may be a mechanism to guide abducens axons to their target (Ferrario et al. 2012). No abnormalities in abducens targeting were found in Sema3C knockout mice, however (Feiner et al. 2001). It should be noted, however, that for all the mice described here, the nerve trajectories were examined grossly in whole mounts, without the detailed examination of terminal branching now possible.

Netrin/Slit/Robo signaling

Netrin and Slit are axon guidance molecules originally identified for their role in midline crossing of commissural axons, but now known to have multiple roles in axon guidance (reviewed in (Huber et al. 2003). Netrin signaling plays a role in proper dorsal guidance of the trochlear axons, but that role can be overcome by genetic modifiers. In vitro, netrin-1 repels trochlear axons (Colamarino & Tessier-Lavigne 1995), and mice lacking the repulsive netrin receptor Unc5c have variable misprojections of the trochlear nerve ventrally (Burgess et al. 2006). In contrast, netrin-1 knockout mice have normal trochlear projections (Serafini et al. 1996). The phenotype of the Unc5 mutants is modified by background strain; severe phenotypes are noted on a B6 background, but not on a SJL background. This is linked to a modifier locus on chromosome 17 (Burgess et al. 2006). Netrin-1 mice were not reported on different background strains. The modification of ocular motor phenotypes by background strain has also been seen in CCDD mouse models (Cheng et al. 2014).

The transcription factor Nkx6-1 is a transcriptional repressor induced by Shh signaling from the floor plate and is expressed in a large proportion of developing oculomotor and trochlear neurons. Loss of Nkx6-1 leads to absence of CN4 and stalling of CN3, with dorsal growth (away from the eye) of CN3 before it stalls, and is associated with upregulation and earlier expression of Unc5c in the motor nuclei (Prakash et al. 2009). Loss of Nkx6-1 is also associated with upregulation and expanded area of expression of Robo1, and decreased expression of Slit2 in the mantal zone of the hindbrain (Prakash et al. 2009). The importance of these changes to the CN3 and CN4 phenotypes is unclear, because Slit/Robo signaling is involved in axon guidance and midline avoidance of the branchiomotor and visceral motor neurons in the hindbrain, but not oculomotor, trochlear, and abducens (somatic motor) neurons (Hammond et al. 2005).

Cell body migration

The cell bodies of the CN3 superior division axons cross the midline in the hindbrain, starting at E13.5 in mice, after the axons have already reached the orbit. The superior division is thus innervated contralaterally. Although Robo/Slit signaling does not guide CN3 axons, it does play a role in cell body migration. Mice deficient in both Slit1/2 or Robo1/2 have early migration of the cell bodies, as early as E10.5 (Bjorke et al. 2016). Similarly, the actin-binding protein drebrin has a role in midline crossing but not axon targeting of CN3. Blocking expression of drebrin with short hairpin RNA (shRNA) in chick embryos blocked the formation of the leading process across the midline and resulted in no migrating cells. Blocking expression of drebrin had no effect on axon targeting of CN3; overexpression led to changes in growth cone morphology, but not axon targeting (Dun et al. 2012).

CFEOM – MICROTUBULE/MOTOR PROTEIN INTERACTIONS

KIF21A

When congenital fibrosis of the extraocular muscles was named, it was believed to be a primary muscle disorder, but it is now known to be a primary neurological disorder, with secondary fibrosis of the muscles. Autopsy of an individual with CFEOM1 (and later shown to have a KIF21A mutation) showed absence of the superior division of CN3 and its motor neurons (Engle et al. 1997). MRI of fourteen individuals with CFEOM1 and KIF21A mutations from six families demonstrated profound hypoplasia of the SR and LPS muscles, and hypoplasia and misdirection of all the motor nerves in the orbit. Interestingly, in some subjects the lateral rectus appears to be innervated by a branch of the oculomotor nerve (Demer et al. 2005).

CFEOM1 results from heterozygous missense mutations in the kinesin-4 family member KIF21A (Yamada et al. 2003). KIF21A is a motor protein that transports molecular cargos anterogradely along microtubules in an ATP-dependent manner (Marszalek et al. 1999). KIF21A interacts with Kank1, a regulator of actin polymerization (Kakinuma & Kiyama 2009) and stabilizes microtubules by reducing the polymerization rate and inhibiting catastrophes in growth cones (van der Vaart et al. 2013). KIF21A is expressed widely, in cell bodies and processes in both the central and peripheral nervous systems as well as skeletal muscle and EOMs, starting in early development and continuing into adulthood (Desai et al. 2012). There were no changes in KIF21A expression in individuals with CFEOM1 (Desai et al. 2012).

Mice with knock in of the most common human KIF21A mutation, R954W (R943W in mouse), display ptosis and globe retraction (Cheng et al. 2014). The distal oculomotor nerve is thin, but proximally there is thickening that ends in a distinct “bulb.” This bulb is composed of stalled axons of the superior division, with enlarged growth cones, increased number of filopodia, and degenerating axons. Inferior division axons continue on a straight path through the bulb. The bulb resembles a decision region where axons pause and change direction. Distally, the superior division is severely hypoplastic and the inferior division displays some aberrant branching (Cheng et al. 2014).

KIF21A is composed of an amino terminal motor domain, a central stalk domain, and a carboxy terminal domain containing WD40 repeats (Figure 3). CFEOM1-causing mutations are in specific amino acids in the third coiled-coil region of the stalk or in the motor domain (Chan et al. 2007). One deletion of a single amino acid in the coiled-coil region has also been reported (Wang et al. 2011). The stalk region forms an intramolecular antiparallel coiled-coil domain; binding of the coiled-coil regulatory region to the motor domain auto-inhibits the protein by preventing binding of the motor domain to alpha tubulin (Bianchi et al. 2016). The R954W amino acid substitution, which is in the stalk, abolishes auto-inhibition of the protein and leads to increased association between KIF21A and microtubules (Cheng et al. 2014, van der Vaart et al. 2013). All CFEOM1-causing mutations in the stalk disrupt binding to the motor domain, some by destabilizing the antiparallel coiled-coil structure, some through interaction of amino acid side chains (Bianchi et al. 2016). The three reported CFEOM1-causing mutations in the motor domain are at the interface where the coiled-coil and motor domain bind, and do not interfere with microtubule binding but abolish binding to the coiled-coil domain (Bianchi et al. 2016, van der Vaart et al. 2013). Further evidence that KIF21A mutations lead to loss of auto-inhibition, CFEOM1-causing KIF21A mutations enhance the interaction with Kank1, leading to increased translocation of Kank1 to the membrane (Kakinuma & Kiyama 2009).

Figure 3:

Predicted KIF21A protein structure. The amino acid residues altered by heterozygous mutations are depicted by arrows. Adapted from (Chan et al. 2007).

The specific mechanism by which decreased auto-inhibition of KIF21A leads to CFEOM has not been established, and it remains unclear why the superior division of CN3 is specifically vulnerable to these mutations. Kif21a interacts with microtubule associated protein 1b (Map1b), and Map1b knock-out mice also have hypoplasia of the superior division of the oculomotor nerve (CFEOM pathology) (Cheng et al. 2014). In vitro, KIF21A suppresses growth cone dynamics and mutant KIF21A was more likely to accumulate in the growth cone (van der Vaart et al. 2013). CN3 axons from Kif21a knock-in mice are larger, with increased filopodia (Cheng et al. 2014). NCKX2, a neuron-specific K⁺-dependent Na⁺/Ca²⁺ exchanger, interacts specifically with the cargo-loading domain of KIF21A, and is transported down axons by KIF21A (Lee et al. 2012). Whether the interaction between KIF21A and NCKX2 is altered by CFEOM1-causing mutations has not been examined. NCKX2 knockout mice do not have any ocular phenotype or abnormal morphology of the oculomotor neurons (Lee et al. 2012, Li et al. 2006), but have not been evaluated specifically for axon guidance defects. Given that CFEOM results from decreased auto-inhibition of KIF21A, increased delivery of the cargo would be expected, so loss of NCKX2 would not be expected to cause CFEOM.

TUBB3/TUBB2B

As in CFEOM1, MRI in patients with CFEOM3 reveals hypoplasia of the oculomotor nerve and the muscles it innervates (Demer et al. 2010). Consistent with the clinical heterogeneity, imaging findings were more variable and more likely to be asymmetric.

CFEOM3 results from heterozygous missense mutations in TUBB3, the neuron-specific beta tubulin monomer (Tischfield et al. 2010) or TUBB2B, another beta tubulin (Cederquist et al. 2012). Alpha and beta tubulin monomers, encoded by separate genes, form heterodimers, which are then assembled into microtubules, which form the cytoskeleton (Lopata & Cleveland 1987). Missense mutations in TUBB3 cause CFEOM3 (Tischfield et al. 2010), malformations of cortical development (MCD) (Poirier et al. 2010), or, in rare cases, both (Whitman et al. 2016). Missense mutations in TUBB2B cause polymicrogyria (Guerrini et al. 2012, Jaglin et al. 2009) and, in rare cases, CFEOM with polymicrogyria (Cederquist et al. 2012, Romaniello et al. 2012). With exquisite genotype-phenotype correlations, CFEOM3 presents as either an isolated eye movement disorder or as a syndrome with other neurological features, including, in specific patterns depending on the specific mutation, severe eye movement restriction, congenital facial weakness, intellectual and social disabilities, progressive peripheral neuropathy, Kallmann syndrome (anosmia with hypogonadotropic hypogonadism), hypoplasia of the corpus callosum and anterior commissure, basal ganglia malformations, and congenital joint contractures (Tischfield et al. 2010). The best characterized mutation presents with a recognizable constellation of features, the TUBB3 E410K syndrome, which also includes cyclic vomiting (Chew et al. 2013).

Eleven separate missense mutations in TUBB3 and two missense mutations in TUBB2B have been reported to cause CFEOM3 (Cederquist et al. 2012, MacKinnon et al. 2014, Romaniello et al. 2012, Tischfield et al. 2010, Whitman et al. 2016). Tubulin has three structural domains: N-terminal, intermediate, and C-terminal (Figure 4). The N-terminal domain contains the GTP-binding pocket. The intermediate domain interacts with the N-terminal domain for structural rearrangements resulting from hydrolysis of GTP and mediates longitudinal and lateral interactions important for heterodimer and microtubule stability. The C-terminal domain interacts with motor proteins (kinesin and dynein) and other microtubule associated proteins (MAPs) (Tischfield et al. 2011). Most residues associated with CFEOM, including E410, D417, R380 and R262 are in the C-terminal domain, in the alpha helices associated with binding to MAPs. R262 and D417 are predicted to form a hydrogen bond. Mutations associated with MCDs are in residues near the GTP-binding pocket or associated with intra- and inter-heterodimer interfaces. The G71 and G98 residues associated with both phenotypes are near the GTP-binding pocket (Whitman et al. 2016). The two TUBB2B mutations, which both cause polymicrogyria with CFEOM3, are E421K (C-terminal) and G140A (GTP binding site) (Cederquist et al. 2012, Romaniello et al. 2012). E421, E410, D417 and R262 are critical for kinesin-microtubule interaction in vitro (Minoura et al. 2016, Uchimura et al. 2010, Uchimura et al. 2006). CFEOM-causing mutations disrupt kinesin binding. In yeast, both the E421K and the E410K mutations significantly decrease binding of the motor protein Kip3p (Cederquist et al. 2012). In neurons, E410K and D417H mutations disrupt kinesin binding and localization of kinesin to the axon tips (Niwa et al. 2013). In vitro, R262H mutations abolish kinesin binding (Minoura et al. 2016).

Figure 4:

A. Two-dimensional structure of TUBB3, showing the protein domains. The sites of CFEOM-causing mutations are indicated. B. Three-dimensional structure of TUBB3, showing position of disease-causing substitutions. Substitutions associated with CFEOM3 (blue) are most often in the C-terminal domain. Substitutions associated with MCD (green) are often in residues that contact the GTP nucleotide or are at contact surfaces between the intra-and inter-heterodimers. Substitions G71R and G98S (red) cause both CFEOM3 and MCD. Substitions at A302 (purple) can cause CFEOM (A302T) or MCD (A302V). Figure adapted from (Whitman et al. 2016).

In mice homozygous for the R262C substitution, there is misrouting of CN3 and CN4 (Tischfield et al. 2010). Total expression levels of TUBB3 are decreased and microtubules display greater stability. Additionally, there is a significant decrease in the amount of Kif21a that copurifies with microtubules from these mice (Tischfield et al. 2010). In vitro, R262C and R62Q TUBB3 had the lowest incorporation into microtubules, whereas A302T, R380C, R262H, D417H, D417N, and E410K incorporate into microtubules at rates similar to wild-type TUBB3 (Tischfield et al. 2010). Interestingly, patients with R262C and R62Q have the mildest phenotypes, perhaps because less mutant tubulin is incorporated into the microtubules. Loss of TUBB3 does not cause any developmental phenotypes (Latremoliere et al. 2018).

Introducing CFEOM-causing mutations into yeast beta tubulin disrupts microtubule function and confers resistance to depolymerization. Only R62Q and R380C haploid spores are viable, although all mutations lead to viable heterozygous diploids (Cederquist et al. 2012, Tischfield et al. 2010). All the mutations alter microtubule dynamics, but in different ways. Substitutions A302T, R62Q, and R380C resulted in microtubules that were more stable with prolonged paused states instead of growing and shortening. Substitutions R262C, R262H, E410K and E421K resulted in microtubules with prolonged, slow growth and more rapid complete disassembly than wild-type microtubules (Cederquist et al. 2012, Tischfield et al. 2010).

When mouse cortical projection neurons are electroporated with Tubb2b E421K or TUBB3 E410K or D417H, there is a decrease in homotopic connectivity across the corpus callosum, and axons fail to innervate their targets. This is not a result of abnormal neuronal migration (Cederquist et al. 2012, Niwa et al. 2013). Similarly, cortical projection neurons transfected with TUBB3 R262H or TUBB3 R262A (not a patient mutation) show decreased axon elongation. This can be rescued for R262A by co-transfecting a kinesin with the microtubule binding site mutated (Minoura et al. 2016). Transfection of TUBB3 E410K or D417H in hippocampal neurons disrupted axonal transport of vesicles, and in DRG neurons disrupted transport of mitochondria (Niwa et al. 2013).

Netrin-1 can act as either an attractive or a repulsive signal, by acting through different receptors. TUBB3 has been shown to act downstream of both netrin-1 attractive and repulsive signaling through direct interaction between TUBB3 and the netrin receptors DCC, DSCAM, and UNC5C, respectively (Huang et al. 2015, Qu et al. 2013, Shao et al. 2017). Knockdown of TUBB3 abolished netrin mediated outgrowth of cortical neurons (Qu et al. 2013) and netrin mediated repulsion of cerebellar neurons (Shao et al. 2017). TUBB3 mutants G82R, T178M, A302V, A302T, R262C, M323V, D417H and D417N reduced the interaction between TUBB3 and DCC, but E205K, M388V, R62Q, and E410K did not (Huang et al. 2018). In primary cortical neurons, R262C and A302V TUBB3 mutations block netrin-1 induced neurite outgrowth and axon branching, in a gain-of-function manner. WT TUBB3 localized to the peripheral grown cone including lamellipodia and filopodia, but R262C and A302V mutant TUBB3 mainly localized to the core of the growth cone, where they would be unavailable to directly interact with DCC on the membrane (Huang et al. 2018). For repulsive signaling, TUBB3 mutants A302T, M323V, R262C, R62Q and D417H reduced the interaction between TUBB3 and UNC5C, but E410K, R380C, M388V, R262H, E205K, G82R, and T178M did not (Shao et al. 2019). In cerebellar neurons, R262C or R62Q TUBB3 abolished netrin-1 mediated growth cone collapse and axon repulsion (Shao et al. 2019). How these results relate to the CFEOM3 phenotype is unclear, as only some of the CFEOM-causing mutations disrupt UNC5C or DCC interaction. The differences between different mutations may help explain some of the genotype-phenotype correlation among non-ocular phenotypes seen with CFEOM3, although the mutations with the most severe non-ocular phenotypes (particularly E410K) do not alter DCC or UNC5C interactions.

DUANE SYNDROME/HORIZONTAL GAZE PALSY

CHN1

DRS results from abnormal development of CN6. In most cases, DRS is sporadic, but a subset of patients with autosomal dominant DRS have missense mutations in the RacGAP protein α2-chimaerin, encoded by the gene CHN1 (Miyake et al. 2008). CHN1 encodes both α1- and α2-chimaerin, via alternate splicing, but several of the DRS-causing mutations are in the portion of the gene exclusive to α2-chimaerin. These missense mutations lead to over-activation of the protein, by disrupting residues associated with maintaining the molecule in a closed conformation. Enhancement of RacGAP activity is only seen after activation of the protein, however (Miyake et al. 2008, Nugent et al. 2017). Patients with CHN1 mutations are more likely to have bilateral DRS, and can have associated vertical anomalies, even in the absence of DRS, indicating involvement of CN6, CN3 and CN4 (Miyake et al. 2008, Miyake et al. 2011). MRI of DRS patients with CHN1 mutations shows hypoplasia or absence of CN6, innervation of the LR by CN3, and abnormalities of the EOMs (Demer et al. 2007).

The role of α2-chimaerin in CN3 axon guidance was first examined in chick oculomotor neurons. Knockdown of α2-chimaerin leads to defasciculation, overshoots, and ectopic branches of CN3 to the LR (Ferrario et al. 2012), whereas over expression of mutant α2-chimaerin causes axon stalling (Miyake et al. 2008). Overexpression of mutant α2-chimaerin also rescues the phenotype of PlexinA knockdown, indicating α2-chimaerin is downstream of Plexin (Sema) signaling (Ferrario et al. 2012). In zebrafish, expression of mutant α2-chimaerin leads to CN3 stalling, but loss of α2-chimaerin leads to ectopic branching (Clark et al. 2013).

Mice with knock-in of a DRS-causing CHN1 mutation display globe retraction and axon stalling of the abducens nerve, which then results in abducens motor neuron loss (Nugent et al. 2017). In contrast, CHN1 knock-out mice display a distinct abducens phenotype, axonal wandering, showing that DRS results from gain-of-function, rather than loss-of-function (Nugent et al. 2017). α2-chimaerin acts downstream of EphA4 in developing motor neurons to elicit growth-cone collapse (Iwasato et al 2007, Shi et al 2007, Wegmeyer et al 2007, Beg et al, 2007, Kao et al 2015), and EphA4 knock-out embryos have abducens phenotypes that are similar to CHN1 knockouts, although with some differences (Nugent et al. 2017). Eph/ephrin signaling can be both forward (downstream of the Eph receptor) and reverse (downstream of the ephrin, with a co-receptor). Abducens neurons and the mesenchyme they migrate through express both EphA4 and ephrin-A5, and abducens neurons participate in both forward and reverse Eph/ephrin signaling. In abducens neurons, α2-chimaerin acts downstream of EphA4 to mediate ephrin-A5 repulsion. Wild-type α2-chimaerin is recruited downstream of only forward Eph signaling, but mutant α2-chimaerin is also recruited downstream of reverse ephrin signaling (Nugent et al. 2017). CHN1 knock-in mice also display misrouting of trochlear axons and C1 axons. The trochlear misrouting is unaffected by removing EphA4, indicating that in trochlear neurons α2-chimaerin is downstream of another receptor. In contrast, C1 misrouting was completely rescued by elimination of EphA4, indicating C1 neurons use only Eph forward signaling, not ephrin reverse signaling (Nugent et al. 2017).

Primary versus secondary axon defects

Globe retraction in DRS results from co-contracture of the MR and LR on adduction, because of aberrant innervation of the LR by branches of CN3 (Hotchkiss et al. 1980, Huber 1974, Miller et al. 1982). In CHN1 knock-in mice, the lateral rectus is innervated by branches from the oculomotor nerve (Nugent et al. 2017). Two hypotheses exist to explain this aberrant branching: 1. A primary defect in CN3 axon development, or 2. A secondary defect, in which otherwise normal CN3 axons are attracted to uninnervated LR muscle in the absence of CN6. α2-chimaerin is expressed in CN3, and, as described above, disruption of α2-chimaerin signaling in chick or zebrafish OMNs leads to CN3 phenotypes, including stalling, defasciculation, overshoots, and ectopic branching (Clark et al. 2013, Ferrario et al. 2012, Miyake et al. 2008). MAFB knockout mice, however, also display aberrant innervation of the LR by branches of CN3 (Park et al. 2016) and allow us to resolve this definitively in favor of the second hypothesis. MAFB is a basic leucine zipper transcription factor expressed in rhombomeres 5 and 6 and required for hindbrain patterning (Cordes & Barsh 1994, Kim et al. 2005, Sadl et al. 2003). It is not expressed in developing OMNs (Eichmann et al. 1997, Kim et al. 2005), and therefore the aberrant branches of CN3 to the LR do not result from cell-autonomous changes in CN3 axons, but rather are secondary, due to lack of CN6 innervation. The mechanism by which the uninnervated LR muscle attracts CN3 motor axons remains to be elucidated.

ROBO3

Homozygous mutations in ROBO3 lead to HGPPS, in which patients are unable to move their eyes horizontally and develop a debilitating progressive scoliosis (Jen et al. 2004). On MRI, the abducens nuclei appear hypoplastic, but both CN3 and CN6 are present and innervate the orbit normally. Patients with HGPPS have uncrossed corticospinal and dorsal column-medial lemniscal pathways, as well as an abnormal midline cleft in the medulla, without apparent motor or sensory deficits (Bosley et al. 2005, Jen et al. 2004). ROBO3 is one of the Roundabout family of Slit receptors, and is necessary for commissural axons to cross the midline. Commissural axons are initially attracted to the midline, but after crossing must lose that attraction and switch to repulsion, so they can continue on their trajectory without recrossing. Unlike Robo1 and Robo2, which are expressed on commissural axons after midline crossing, Robo3 is expressed on pre-crossing commissural axons and downregulated after crossing (Sabatier et al. 2004). Robo3 knockout mice lack commissures in the spinal cord, hindbrain and midbrain (Marillat et al. 2004, Sabatier et al. 2004).

It was first proposed that Robo3 inhibited the repulsive signal of Robo1/2 receptors to allow commissural axons to approach the midline, and Robo1/2/3 triple knockouts display partial rescue (Sabatier et al. 2004), but recent studies support a model where Robo3 potentiates midline attraction (reviewed in (Friocourt & Chedotal 2017). Robo3 does not bind Slit with high affinity, and although it does not bind netrin-1, it forms a complex with DCC and is phosphorylated in the presence of netrin-1. Robo3 may therefore mediate netrin-1 attraction to the midline (Zelina et al 2014).

In mice with specific knockout of Robo3 at the level of the abducens nucleus in rhombomere 5 (via Krox20cre), abducens motor neurons project normally to the LR, but the internuclear commissure between the abducens and contralateral oculomotor nucleus is severely reduced. Those mice also have impaired horizontal eye movement reflexes, but normal vertical movements (Renier et al. 2010). Thus, HGPPS results from abnormal axon guidance of the internuclear commissure, rather than the cranial motor nerves.

Cadherins/Protocadherins

The cadherin superfamily of cell adhesion molecules, which include classical cadherins and protocadherins (Pcdh), are homophilic adhesion molecules important in recognizing self/non-self (reviewed in (Sanes & Zipursky 2020). Cadherin superfamily cell adhesion molecules are expressed in specific patterns in cranial motor nuclei and function in formation of distinct nuclei in the hindbrain (Astick et al. 2014). Mutations in N-cadherin (CDH2) result in a neurodevelopmental syndrome that involves intellectual disability, agenesis/hypoplasia of the corpus callosum, ocular, cardiac, and genital abnormalities, and, variably, DRS (Accogli et al. 2019). Pcdh17 is expressed in the abducens nucleus of adult and developing zebrafish (Liu et al. 2015, Liu et al. 2009). Pcdh17 mutant zebrafish display a variety of developmental abnormalities, including reduced axon growth and misdirected abducens axons in ~20% of nerves (Asakawa & Kawakami 2018). Expression of a dominant negative Pcdh17 in the abducens nucleus led to clumping of the axons ventrally, with no rostral extension of the axons. Consistent with a role of Pcdh17 in self-avoidance, when expression was limited to a subset of cells (via injection into embryos), the axons formed clumps in ~40% of resulting nerves (Asakawa & Kawakami 2018). The dominant negative may have had more significant phenotypes than the knockout because of functional redundancy of other Pcdh family members. Although protocadherins have not been linked to CCDDs in humans, it has been suggested that Pcdh17 may be downstream of Mafb (Asakawa & Kawakami 2018).

OCULOMOTOR SYNKINESIS

Oculomotor synkinesis is an unintended eye or eyelid movement when another eye or face movement is attempted. The best-known form is Marcus Gunn Jaw Winking, in which a ptotic eyelid elevates with sucking or chewing. This is believed to result from aberrant innervation of the eyelid by motor trigeminal fibers, which normally innervate the muscles of mastication, and are thus activated by sucking or chewing. Oculomotor synkinesis can be isolated, or accompany other eye movement disorders, including CFEOM1 (Yamada et al. 2005). Additionally, the globe retraction seen in DRS can be considered a form of oculomotor synkinesis.

Three mouse models of oculomotor synkinesis have been recently described, and one gene for isolated oculomotor synkinesis in humans has been identified (Whitman et al. 2019, Whitman et al. 2018). Humans with a homozygous missense mutation in the chemokine receptor ACKR3 (CXCR7) have ptosis, and display ipsilateral eyelid elevation upon abduction (Whitman et al. 2019). ACKR3 binds the chemokine CXCL12, which is also the ligand for the chemokine receptor CXCR4 (Quinn et al. 2018). CXCR4 has several functions in nervous system development, including in neuronal migration and axon guidance (Li & Ransohoff 2008). In the spinal cord, CXCR4 signaling is required for initial pathfinding of ventral motor neurons (Lieberam et al. 2005). CXCL12 promotes axon outgrowth of oculomotor neurons and has a mild chemoattractant effect on CN3 axons (Ferrario et al. 2012, Lerner et al. 2010). Inhibition of CXCR4 in an oculomotor slice culture causes dorsal projection of axons that had not yet exited the midbrain, while axons that have already exited into the neuroepithelium grow normally toward the eye (Whitman et al. 2018). Cxcr4 and Cxcl12 knockout mice recapitulate that phenotype, with oculomotor axons projecting dorsally, rather than ventrally (Figure 5) (Lerner et al. 2010, Whitman et al. 2018). In both lines of mice, CN3 does not reach the orbit, and there is aberrant innervation of the EOMs from axons of the motor trigeminal nerve, which are aberrantly following the sensory trigeminal pathway (Whitman et al. 2018). There is, however, variability between embryos in the degree of aberrant innervation. Unfortunately, Cxcr4 and Cxcl12 knockouts die at birth, so their eye movements cannot be assessed.

Figure 5:

Axon guidance defects in Cxcr4 and Ackr3 knockout mice. Maximum intensity projections of whole mount embryos in which motor axons are labeled GFP (IslMNGFP, green) and counterstained with anti-smooth muscle actin to label muscles and arteries (red). (A) shows the normal trajectory of CN3, CN4, and CN6 at E12.5 from the brainstem to the orbit, and (C and E) show each nerve projecting to its target muscles within the orbit E12.5 (C) and at E13.5 (E). (A) and E13.5 (E). CN3 enters the orbit, the superior division branches to the SR (yellow arrowhead in E), an inferior decision region forms between the MR and IR (yellow arrow in E) and a branch extends to the IO. CN4 projects to the SO. CN6 projects to the LR. (B) E12.5 Cxcr4cko/cko:Isl-cre embryo has loss of Cxcr4 specifically in motor neurons, and shows the dorsal projection and stalling of CN3 in the midbrain, whereas CN4 projects normally. (D) CN3 is absent from the orbit and fibers of the motor trigeminal nerve aberrantly project to the EOMs (stars). Midbrain is enlarged in A’ and B’. F. Example of an Ackr3KO/KO with Duane syndrome pathology. CN6 is absent. CN3 is thin, has scant projections to the SR and aberrant branches to the LR (yellow arrow). G. Central orbital view of an E13.5 Hb9:GFP embryo (CN6 is labeled in green but not CN3 or CN4) shows the normal trajectory of CN6 to the LR. H. Example of an E13.5 Ackr3KO/KO embryo in which CN6 not only innervates the LR but also has an aberrant branch that divides to innervate other EOMs that are normally innervated by CN3 (yellow arrow). SR, superior rectus; SO, superior oblique, MR, medial rectus; IR, inferior rectus; IO, inferior oblique; LR, lateral rectus. A-D were adapted from (Whitman et al. 2018) published by ARVO and E-H were adapted from (Whitman et al. 2019), published by Oxford University Press.

ACKR3 functions as a scavenger receptor, regulating the amount of CXCL12 available for CXCR4 binding (Abe et al. 2015, Boldajipour et al. 2008, Naumann et al. 2010). The ACKR3 mutation associated with oculomotor synkinesis results in a protein with a lower binding affinity for CXCL12. Ackr3 knockout mice display misrouting of CN3, CN6, and motor trigeminal nerves, with significant variability between embryos and even between the two sides of the same embryo (Whitman et al. 2019). CN3 phenotypes range from complete dorsal misrouting, as seen with loss of Cxcr4 or Cxcl12 (Whitman et al. 2018), to formation of a thin nerve projecting ventrally to the orbit, often with peripheral misprojections (Whitman et al. 2019). This thin nerve always arises from the rostral portion of the nucleus; axons from the caudal aspect of the nucleus always project dorsally. Motor neurons in the rostral portion of the nucleus are born first and their axons form the inferior division of CN3; caudal motor neurons are born later and their axons form the superior division of CN3 (Cheng et al. 2014, Greaney et al. 2017). The caudal motor neurons are the subpopulation whose cell bodies later cross the midline and are more sensitive to multiple different CCDD-causing mutations.

In embryos with complete misrouting of CN3, CN6 sometimes projects to the EOMs normally innervated by CN3. In other embryos, CN6 was completely absent, and CN3 axons aberrantly project to the LR (DRS pathology) (Figure 5). In Ackr3 knockout embryos with DRS pathology, the misinnervating CN3 fibers came exclusively from the inferior decision region (Whitman et al. 2019). In MafB and Chn1 DRS models, by contrast, aberrant projections from CN3 to the LR come from both the superior and inferior decision regions (Nugent et al. 2017, Park et al. 2016). This may be because the residual CN3 in Ackr3 knockouts is thin and derived from the rostral neurons, which form the inferior division. One individual heterozygous for the ACKR3 mutation has DRS, while other heterozygous individuals have normal ocular motility. It remains unresolved whether DRS represents variable penetrance of the heterozygous variant or if the etiology of this individual’s DRS is unrelated (Whitman et al. 2019).

Because ACKR3 sequesters CXCL12, loss of ACKR3 function results in more CXCL12 available for CXCR4 binding. Excess levels of CXCL12, however, lead to CXCR4 receptor downregulation and decreased CXCR4 signaling (Abe et al. 2015, Abe et al. 2014, Lewellis et al. 2013). Consistent with this, expression of CXCR4 protein (but not Cxcr4 mRNA) in and around the oculomotor nucleus is decreased with loss of Ackr3 (Whitman et al. 2019). The profound changes in CN3 axon trajectory seen with changes in CXCR4 expression are similar to those discussed above that result from loss of Nkx6-1 (Prakash et al. 2009). There may, therefore, be attractive factors dorsally or repulsive factors ventrally whose actions are normally overcome. The identity of such factors remains to be determined.

OTHER CCDDS – VARIABLE PHENOTYPES

COL25A1

Recessive COL25A1 mutations have been identified in patients with congenital ptosis or DRS (phenotypes vary even within the same family) (Khan & Al-Mesfer 2015, Shinwari et al. 2015). In Col25a1 knockout mouse embryos, motor axon bundles reach their target muscles but fail to extend into the muscle fascicles, including on EOMs, resulting in massive cell death of motor neurons (Munezane et al. 2019, Tanaka et al. 2014). Col25a1 is expressed in both developing motor neurons and muscle (Tanaka et al. 2014), but muscle-specific loss of Col25a1 leads to similar phenotypes as the full knockout, indicating that Col25a1 expression on muscle drives motor neuron invasion (Munezane et al. 2019). Col25a1 is transiently induced in uninnervated myotubes during development, and then is downregulated by nerve-induced muscle excitation. Motor axons are attracted to cells expressing Col25a1, which interacts with receptor protein tyrosine phosphatases PTPσ and PTP δ on the motor axons. This interaction is reduced by the CCDD-causing mutations, and motor axons are not attracted to cells expressing mutant Col25a1 (Munezane et al. 2019). As with other CCDD mutations, it is not known why the human phenotype is limited to the oculomotor system.

ECEL1

Autosomal recessive mutations in ECEL1 lead to distal arthrogryposis with variable ocular phenotypes (Dieterich et al. 2013, McMillin et al. 2013), at least some of which are consistent with CCDDs (Khan et al. 2014b, Shaaban et al. 2014). The majority of patients have ptosis, and some have ophthalmoplegia similar to CFEOM3 or DRS (Khan et al. 2014b, Shaaban et al. 2014). ECEL1 is a member of the neprilysin family of zinc metalloendopeptidases, primarily expressed in the central and peripheral nervous system, including the cranial motor neurons, from early development to adulthood (Nagata et al. 2006, Nagata et al. 2010, Valdenaire et al. 1999). It has been suggested that zinc metalloendopeptidases control axon growth by cleaving axon guidance receptors and ligands (Bai & Pfaff 2011). Ecel1 knockout mice have defective formation of the neuromuscular junction in skeletal muscles and the diaphragm (Nagata et al. 2010), with defective axonal arborization of the motor nerves, a phenotype also seen with knock in of the patient mutation C760R (Nagata et al. 2016). Patients with the G607S substitution have significant ophthalmoplegia and less severe contractures (Shaaban et al. 2014), suggesting that that substitution might have a different mechanism or effect than other mutations. The G607S substitution leads to splice variants which induce a premature stop codon and nonsense mediated decay of the mRNA, leading to extremely low levels of the protein. The C760R substitution leads to altered localization of the protein: it was present in the cell bodies, but not axons of motor neurons (Nagata et al. 2017). Despite these differences, mice with either the C760R or G607S substitution show stalling and/or wandering of the abducens nerve, with variability in the phenotype (Nagata et al. 2017).

MOEBIUS/FACIAL WEAKNESS

Moebius syndrome involves defects in CN6 and CN7, with resulting abduction deficits and facial palsy. Most cases are sporadic, and those cases that appear to be familial often do not meet full diagnostic criteria, often having facial palsy without abduction deficit (MacKinnon et al. 2014). One such family is interesting, however, because it potentially expands the range of tubulinopathies. A large family with autosomal dominant bilateral facial palsy, ptosis, velopharyngeal dysfunction and reduced gag reflex segregate a missense mutation in TUBB6, a class V tubulin. Introduction of this mutation to yeast tubulin led to increased sensitivity to the microtubule destabilizer, benomyl (Fazeli et al. 2017). Occasionally, patients with TUBB3 mutations and CFEOM3 with facial palsy are first misdiagnosed with Moebius syndrome (MacKinnon et al. 2014).

CONCLUSION

The oculomotor system is a relatively simple system in which to study axon guidance mechanisms – seven muscles are innervated by three cranial nerves, and eye movements provide an easy read-out of proper or improper innervation. Even within this “simple” system, however, the mechanisms of axon guidance are complex. Multiple signaling mechanisms are active, with different roles in different subsets of neurons. For reasons that remain unclear, the oculomotor, trochlear, and abducens neurons are selectively vulnerable to mutations in widely expressed molecules, and specifically to mutations that alter the auto-regulation of those molecules. For many mutations, there are also additional genetic modifiers, with significant phenotypic variability between individuals (in humans and mouse models) and strain differences in phenotypic expression in mouse models. EOMs lacking their normal innervation attract motor axons that would not normally target those muscles, through unknown mechanisms. Thus, there remain many unanswered questions before we have a full understanding of all the mechanisms of axon guidance in the oculomotor system.

Sidebar – Is congenital superior oblique palsy a CCDD?

Superior oblique palsy is a relatively common form of strabismus caused by underaction of the superior oblique muscle that results in vertical strabismus that varies with gaze position. Reliable identification of CN4 requires high resolution 3T MRI, but it has now been shown that CN4 is absent in a significant proportion of SO palsy patients (Kim & Hwang 2010, Lee et al. 2014, Yang et al. 2012). SO palsy can, therefore, be classified as a CCDD. No causative genes have been identified, and it is not known if it results from axon guidance defects.