The reorganization of the GC cytoskeleton during axon steering is ultimately enabled by a tight regulation of the MT-actin interplay (Gordon-Weeks, 2004; Lowery and Vactor, 2009; Dent et al., 2011). The targeting of the actin cytoskeleton, either by downregulation of actin isoforms or by depolymerization of F-actin using Cytochalasin B, reduces the size of the growth cone, hinders filopodial dynamics and abolishes the axonal turning response (Forscher and Smith, 1988; Zheng et al., 1996; Moradi et al., 2017). Despite the requirement of F-actin for axon steering, it has been shown that neurons can still elongate their axons after F-actin depolymerization or inhibition of myosin II (Bentley and Toroian-Raymond, 1986; Zheng et al., 1996; Bradke and Dotti, 1999; Hur et al., 2011b), indicating that F-actin filaments remodeling is essential for motility and steering but dispensable for axon extension. In contrast, MT dynamics is necessary for axon growth (Yamada et al., 1970; Bamburg et al., 1986; Mansfield and Gordon-Weeks, 1991; Tanaka et al., 1995; Yu and Baas, 1995; Rochlin et al., 1996; Tischfield et al., 2010) but the role of MTs in GC steering is yet a matter of discussion.

More than 30 years ago, pioneer studies already showed that MTs asymmetrically invade the P-domain of the GC in the direction of the turn and their reorganization is essential during GC maneuvers (Forscher and Smith, 1988; Sabry et al., 1991; Tanaka et al., 1995; Williamson et al., 1996). In addition, MTs entering the filopodia were found to be captured and stabilized by preventing shrinkage. These stabilized MTs allowed the flow of cytoplasmic organelles and increased the filopodia lifetimes, enabling directional axon outgrowth (Gordon-Weeks, 1991; Sabry et al., 1991; Geraldo et al., 2008). Ever since, different works have demonstrated that axon guidance signaling proteins influence MT dynamics at the GCs. For instance, bath incubation of Sema3A and netrin-1 in neuron cultures revealed that Sema3A treatments result in the collapse of MT networks, while netrin-1 incubation stimulates MT splaying and exploration in the GC peripheral region (Dent et al., 2004; Shao et al., 2017). Nerve growth factor (NGF) signaling in sensory neurons also redistributed the MTs in the distal part of axons (Zhou et al., 2004; Turney et al., 2016), and Wnt3a or Wnt5a treatments changed the organization and directionality of MT polymerization in the GC (Purro et al., 2008; Li et al., 2014). Micro-gradients of Brain-derived neurotrophic factor (BDNF) or Sema3A over GCs of sensory axons also biased the direction of MT growth toward the treated GC side or the opposite side, respectively (Pavez et al., 2019). In addition, genetic studies in Caenorhabditis elegans showed that UNC-6/Netrin signaling hampers MT accumulation in the GC and inhibits protrusion, through a molecular pathway that involves the repulsive Netrin receptor UNC-5, RHO-1/RhoA and UNC-33/CRPM (Gujar et al., 2019). Consistently with these findings, the inhibition of MT dynamics blunts GC turning in response to guidance cues. The exposure of GCs and axons to drugs that disrupt MT dynamics, such as taxol or nocodazole, reduced the GC activity and abolished the turning of GCs exposed to netrin-1 and glutamate gradients or substrate boundaries (Tanaka and Kirschner, 1995; Challacombe et al., 1997; Buck and Zheng, 2002; Suter et al., 2004). Yet, because the spatial distribution of MTs in the GC periphery is strongly influenced by actomyosin forces and F-actin dynamics, MTs have been classically relegated to simple supporters of actin-guided movements during axon navigation.

Now, mounting evidences support the idea that MTs and MAPs can also instruct axon pathfinding even when actin dynamics are not directly perturbed. Seminal studies performed 20 years-ago showed that the focal application of the MT stabilizing drug taxol to one side of the GC induces turning toward the drug source, whereas the application of the MT destabilizing drug nocodazole triggers GC turning away from the application side (Buck and Zheng, 2002). Experimental evidence also demonstrated that MT-initiated GC turning engages F-actin remodeling during the movement, since the inhibition of actin polymerization abolished the taxol-evoked attractive GC turning (Buck and Zheng, 2002). Subsequent works employing micro-scale chromophore-assisted laser inactivation (micro-CALI) have also revealed that several MAPs can trigger GC steering. Micro-CALI technique has been exploited to address the role of specific proteins in GC steering (Chang et al., 1995; Diefenbach et al., 2002), although it may have caveats, such as a rapid recovery dependent on protein diffusion or trafficking. The asymmetric inactivation of the MT-stabilizing protein MAP1B or the Adenomatous polyposis coli (APC) protein – a +TIP that stimulates MT polymerization – by micro-CALI in one side of the GC, led to the collapse of the irradiated side followed by GC turning toward the opposite direction (Mack et al., 2000; Koester et al., 2007). In contrast, the asymmetric inactivation of the MT extension-modulator CRMP2 or the MT-sliding motor Kinesin-5 in half of the GC by micro-CALI, resulted in GC turning toward the irradiated side (Higurashi et al., 2012; Nadar et al., 2012). Overall, these experiments support the idea that MT dynamics are not just required for guidance-evoked GC steering movements, but they play an instructive role in GC turning. Corroborating their instructive role in axon guidance signal transduction, evidences from Liu’s laboratory have proven that the neuron-specific tubulin isotype TUBB3 – polymerized in MTs – is a direct target of netrin-1 signaling (Qu et al., 2013; Huang et al., 2015; Shao et al., 2017). Netrin-1 is a dual guidance cue that can evoke attractive and repulsive responses by binding to its high-affinity GC receptors DCC and UNC5, respectively (Alcántara et al., 2000). It was reported that the exposure of cortical neuron cultures to netrin-1 induces GC chemoattraction and stimulates the MT dynamics in the GC. Both of these netrin-1 effects required the direct interaction of its receptor DCC with the neuron-specific tubulin isotype TUBB3, integrated in the MT polymer. Indeed, the interaction of TUBB3 with DCC was greatly increased by exogenous netrin-1 addition and this interaction was dependent on MT dynamics (Qu et al., 2013). Conversely, the GC repulsion induced by netrin-1 exposure through UNC5C receptor signaling also relied on TUBB3-UNC5C binding. UNC5C directly interacts with polymerized TUBB3 in vitro and both partially colocalize in the GC periphery of primary neurons. The focal application of Netrin-1 was found to disengage UNC5C-TUBB3 interaction in GCs and stimulate MT polymerization in the GC region distally to the netrin-1 source, promoting the repulsive response (Shao et al., 2017). Missense mutations of TUBB3 in humans are associated to an abnormal development of the corpus callosum, the anterior commissure, the corticospinal tracts or optic nerves in human patients (Table 1). Additionally, it was found that TUBB3 mutations impaired MT dynamics and abolished both netrin-1-evoked attractive and repulsive responses of cortical axons in vitro (Poirier et al., 2010; Tischfield et al., 2010; Whitman et al., 2016; Huang et al., 2018; Shao et al., 2019). Although the deficits caused by TUBB3 loss-of-function in neural circuits development may be compensated by the remaining β-tubulin isotypes (Latremoliere et al., 2018).

Other mutations in human α-and β-tubulin-encoding genes – such as TUBA1A, TUBB2B, TUBA8, TUBB4A, TUBB2A, TUBB – are linked to severe brain malformations and motor-cognitive disabilities, collectively refereed as tubulinopathies. These syndromes present gross brain malformations and an abnormal development of various nerve tracts, suggesting a putative role in axon guidance (Romaniello et al., 2018). Recent analysis performed on TUBA1A loss-of-function mice and cultured fibroblasts from TUBB-associated tubulinopathy patients have revealed an impaired MT dynamics and aberrant cytoskeleton configurations in the axonal GCs (Buscaglia et al., 2020; Sferra et al., 2020). However, it remains to be uncovered whether and which guidance molecules are involved in these MT-associated axon tract malformations.

In the shaft of cortical axons, MTs are formed de novo – nucleated – locally in an acentrosomal manner, by a mechanism involving the Augmin/HAUS complex and the γ-tubulin ring complex (γTuRC) that ensures the uniform polarity of the MT network (Sánchez-Huertas et al., 2016; Cunha-Ferreira et al., 2018). Although local events of MT nucleation have not been yet reported in the axonal GCs, acentrosomal γTuRC-dependent MT nucleation has been recently observed over endosomes in the dendritic GCs of invertebrate neurons (Liang et al., 2020; Yoong et al., 2020). These evidences, in combination with the following recent findings, allow to put forward the hypothesis of axon guidance signaling influencing local MT nucleation events in the distal axon.

The γTuRC-dependent MT nucleation in eukaryotic cells undergoes spatial and temporal regulation by means of additional MAPs, such as TPX2 and its activator RanGTP, and both proteins have been found to be enriched at neuritic tips (Chen et al., 2017; Huang et al., 2020; Liu et al., 2021). MT-bound TPX2 participates in MT nucleation at elongating neurite tips in cultured hippocampal neurons (Chen et al., 2017). RanGTP is transported anterogradely along the axons through actin waves, it colocalizes with actin-based structures in the axonal GC and enables local nucleation events at neurite tips (Chen et al., 2017; Huang et al., 2020). Actin waves (also known as growth cone-like waves) are dynamic cytoskeletal structures traveling anterogradely along the axon shaft. These waves are associated to transient MT generation activity along the axons, including an increase in MT polymerization and MT-based transport (Winans et al., 2016). Therefore, it is possible that RanGTP and TPX2 are transported to the GC, jointly with other MT nucleation machinery such as γ-TuRCs, to trigger local short-lived MT nucleation events.

New results also suggest that Wnt signaling could shape axonal MT configurations via regulation of local MT nucleation mechanisms. Weiner et al. (2020) showed that in Drosophila, some Wnt signaling proteins, such as Fz, LRP5/6 or Axin, recruit the MT core-nucleation protein γ-Tubulin to endosomes in the dendritic branch points, enabling local MT nucleation and indicating that extracellular Wnt signaling can regulate local MT nucleation in dendrites. In addition, two other recent studies have revealed that the Wnt pathway controls axon specification in developing neurons by organizing the polarity of MT networks both in the axon (Stanganello et al., 2019) and in non-axonal neurites (Puri et al., 2021). Plus, it is known that local MT nucleation contribute the MT arrangements in these compartments (Sánchez-Huertas et al., 2016; Cunha-Ferreira et al., 2018). Overall, these results convey a putative mechanism whereby extracellular Wnt signaling might control MT architecture in axons and dendrites via spatial-temporal control of MT nucleation in developing neurons. Hence, we believe that the contribution of local MT nucleation events in distal axons to guidance cue-instructed navigation should be further investigated.

Microtubules are heavily stabilized in the axonal shaft, whereas in the GC they are very dynamic. The stability status and polymerization rate of MTs in the axons rely on the activity of specific MAPs, such as MAP1B, tau or CRMP2, whose activities are directly regulated by axon guidance signaling pathways (Figure 2). MAP1B is a MT-stabilizing protein that associates with the lattice of dynamic MTs in the most distal region of the axon and in the GC. Studies of asymmetric laser inactivation in GCs together with genetic analyses revealed that the phosphorylated form of MAP1B is a direct effector of axon turning because selectively stabilizes MTs at the GC periphery (Black et al., 1994; Mack et al., 2000; Bouquet et al., 2004). MAP1B phosphorylation levels are increased in cortical neurons after Netrin1 treatment via GSK3β and CDK5 kinase activity. Consistently, growing axons from MAP1B-deficient CNS explants are irresponsive to netrin-1-induced chemoattraction. MAP1B mutant mice are viable but exhibit misguided cortical, thalamocortical and hippocampal axons (Table 1) (Meixner et al., 2000; Del Río et al., 2004). These dramatic axon wiring defects suggest that MAP1B is involved in additional axon guidance pathways, other than netrin-1. Indeed, the repulsive axonal guidance responses evoked by Draxin and Sema3A treatments also involve MAP1B in their downstream pathways. Draxin, which is an essential guidance cue for the development of forebrain commissural tracts, interacts with the netrin receptor DCC and activates the GSK3β-MAP1B pathway in order to induce a repulsive response in cortical axons (Meli et al., 2015). On the other hand, Sema3A treatment of hippocampal neurons increases MAP1B levels in distal axons in a local translation-dependent manner (Campbell and Holt, 2001; Li C. et al., 2009). Specifically, Sema3A induces the local degradation of the translational suppressor FMRP via the ubiquitin-proteasome pathway, which results in the increase of MAP1B mRNA-coding translation in the GC (Takabatake et al., 2020). Thus, it appears that MAP1B is a downstream mediator of both attractive and repulsive guidance cues. This high degree of MAP1B tunability could be entailed by its multiple phosphorylation sites (Kawasaki et al., 2018), sensitive to CDK5 and GSK3β activity, but further work is needed to understand the molecular mechanisms whereby MAP1B promotes GC steering.

Tau is another phospho-MAP that binds the MT lattice and stabilizes the MTs in the axon shaft, playing a critical role in axon specification, growth and branching. Tau hyperphosphorylation generally correlates with impaired MT binding and axonal MT cytoskeleton disruption (Cleveland et al., 1977; Binder et al., 1985; Drubin and Kirschner, 1986; Grundke-Iqbal et al., 1986). Similar to MAP1B, Tau is a downstream target of GSK3β and CDK5 - among other kinases (Guo et al., 2017). In particular, it was found that Sema3A treatment transiently increases phospho-tau levels in the GC of chick neuron cultures via CDK5-dependent phosphorylation previously to GC collapse (Sasaki et al., 2002). Also, Wnt5a promotes the reorganization of MTs in the GC of cortical neurons through CaMKII-dependent phosphorylation of tau within its MT-binding site (Ser262) in order to evoke a repulsive axonal response (Li et al., 2014). In opposition to this, the activation of EphB2 receptor by ephrin B1 reduces tau hyperphosphorylation through GSK3β inhibition in vivo in the CA3 hippocampal region of tau transgenic mice (Jiang et al., 2015). Moreover, Sema3C addition upregulates the total tau protein levels in cultured cerebellar granule neurons, preserving survival and stimulating neuritogenesis (Moreno-Flores et al., 2004). Interestingly, it was uncovered that tau promotes the co-alignment of MT and actin fibers in vitro, and stimulates the coordinated polymerization of both cytoskeleton networks (Elie et al., 2015). In line with this, it was recently reported that tau does not only decorates the lattice of stabilized MTs along the axon shaft but also associates to dynamic MTs aligned with actin filaments in the GC periphery of cortical neurons. Tau downregulation disrupted the MT bundling in the GC central domain, prevented MT invasion into the periphery and misoriented MT trajectories. Overall, tau loss-of-function inhibited the turning of cortical axons exposed to Wnt5a gradients (Biswas and Kalil, 2018).

The collapsin response mediator proteins (CRMPs) family are cytosolic phospho-MAPs that play important roles in the developing nervous system, including axon guidance (Nakamura et al., 2020). CRMP family name was given because its first member identified, CRMP2, was a molecular mediator of GC collapse upon stimulation with Sema3A (originally known as Collapsin) (Goshima et al., 1995). There are five human CRMPs (CRMP1-5), displaying different subcellular localization and cytoskeletal targets. Among them, CRMP2 localizes to the axon and the C-domain of the GC and controls MT polymerization/stability. Indeed, it has been observed that CRMP2 participates in axon specification, elongation, branching and guidance effect by several guidance cues (Inagaki et al., 2001; Lin et al., 2011; Higurashi et al., 2012; Yamashita and Goshima, 2012). When CRMP2 monomers are non-phosphorylated, they bind tubulin heterodimers and the complex is transported to the distal part of growing axons, by kinesin-1-dependent motor forces, to support MT polymerization and axon growth. Upon Sema3A stimulation, CRMP2 is sequentially phosphorylated at its C-terminal domain by CDK5 and GSK3β kinases, hampering its tubulin-binding properties and leading to GC collapse via MT destabilization (Figure 2). The Sema3A-induced CRMP2 inactivation is achieved by phosphorylation at Ser522 by CDK5, followed by GSK3β-dependent phosphorylation at Ser518, Thr514 and Thr509 (Fukata et al., 2002; Kimura et al., 2005; Cole et al., 2006). In addition to Sema3A, other repulsive guidance cues induced CRMP2 phosphorylation via GSK3 and/or Rho kinase to achieve GC collapse, these include Sema4D, RGMa and ephrinA5 (Arimura et al., 2005; Ito et al., 2006; Wang et al., 2013). Consistently, CRMP2 has been demonstrated to be essential for axon navigation in vivo because CRMP2KO mice exhibit axon guidance defects in peripheral nerves and in the corpus callosum (Ziak et al., 2020).

Mutations in the genes encoding the MAP doublecortin (DCX) account for the majority of the human cases of double cortex syndrome, which exhibits severe brain cortex malformations primarily attributable to neuronal migration and proliferation deficits (Gleeson et al., 1998; Bahi-Buisson et al., 2013). DCX is a MT-stabilizing phospho-protein abundant in the axonal GCs, which decorates the lattice of MTs invading the F-actin rich peripheral region of the GC (Moores et al., 2004; Tint et al., 2009). Interestingly, the double genetic deletion of DCX and its closest homolog protein doublecortin-like kinase1 (DCLK1) in mice led to widespread defects in axon tracts, affecting the corpus callosum, anterior commissure, subcortical fiber tracts and internal capsule. More specifically, the DCX mutant axons exhibit impaired transport, growth and are irresponsive to netrin-1-evoked chemoattraction, although the latter was suggested to stem from DCX regulatory effects on actin configurations (Deuel et al., 2006; Koizumi et al., 2006; Fu et al., 2013). This data suggests that DCX is required for guidance signaling-evoked axonal steering during nervous system development. Indeed, a recent study uncovered that DCX mediates the repulsive response of GCs upon Sema3A treatment (Bott et al., 2020). Bott et al. (2020) showed that DCX forms a complex with Nestin that enables DCX phosphorylation by CDK5/p35 downstream Sema3A signaling. They also demonstrated that DCX phosphorylation by CDK5 decreased its MT affinity and resulted in MT destabilization.

In addition to MT polymerization and stability, MT depolymerization and severing are also critical mechanisms for the arrangement of MT networks. Several of these MAPs have been involved in the transduction of axon guidance signaling. SCG10 (superior cervical ganglion-10)/Stathmin-2 is a neuron-specific member of the MT-destabilizing protein family of the stathmins. Stathmins bind tubulin dimers, sequestering them from growing plus-ends and thereby, promoting MT depolymerization (Charbaut et al., 2001; Grenningloh et al., 2004). SCG10 is considered an axon survival protein, highly enriched in the GCs C-domain of developing neurons, and its levels are dynamically regulated by local degradation and KIF1B-dependent axonal transport toward the GC (Shin et al., 2012; Drerup et al., 2016). Axon extension during neuron differentiation requires SGC10 activity, since its downregulation produces MT overstabilization and looping in the GC of hippocampal neurons (Morii et al., 2006). The repulsive protein EphB typically triggers GC collapse, but in the presence of laminin and L1 leads to paused GCs, which retain their normal filopodial dynamics and actin distribution. It was found that this guidance cue combination specifically reduced SCG10 levels in GC, which stimulated the invasion of long curved MTs into the GC periphery and led to GC pause (Figure 2) (Suh, 2004). Additionally, SCG10 interacts with the small RhoGTPase Rnd1, and this interaction enhances SCG10 MT destabilizing activity in neurons. Rnd1 is known to mediate the GC collapse induced by Sema4D-Plexin-B1 signaling in hippocampal neurons (Oinuma et al., 2004; Li Y.-H. et al., 2009), suggesting that SCG10 may also function downstream of the Sema4D signaling pathway.

On the other hand, the MT-severing enzymes cut MT fibers into shorter fragments, creating new local MT seeds and influencing axon branching (Sharp and Ross, 2012). Spastin is a MT-severing protein required for axon morphogenesis, associated to a degenerative disease of the corticospinal axon tracts, named Hereditary spastic paraplegia (Roll-Mecak and Vale, 2008). Recently, the alternative translation of spastin mRNA transcripts has been found to influence both motor neuron axon guidance and migration downstream of bone morphogenic protein (BMP) and neuropilin-1 signaling during zebrafish development (Jardin et al., 2018). Fidgetin-like-1 (Fignl1) is another MT-severing protein enriched in the growth cone of zebrafish growing axons, whose downregulation led to pathfinding defects in spinal motor axons and impaired larvae locomotion (Fassier et al., 2018), although no specific guidance proteins controlling Fignl1 activity have been identified.

Concerning MT polymerization inhibitors, the kinesin-4 family members KIF21A and KIF21B and the immotile kinesin-13 family member KIF2A, have been linked to neurodevelopmental malformations associated with axon growth and guidance defects in human patients (Table 1) (Yamada et al., 2003; Poirier et al., 2013; Asselin et al., 2020). More specifically, KIF21A is a gene risk factor for the CFEOM1 (congenital fibrosis of the extraocular muscles type-1), a developmental oculomotor nerve disorder. CFEOM1-associated Kif21a mutations in mice caused aberrant axon branching, stalling and misorientation defects in oculomotor nerves (Yamada et al., 2003; Cheng et al., 2014). It was reported that KIF21A decreases MT polymerization rate and suppresses MT plus-end catastrophes. KIF21A overexpression in hippocampal neurons slendered the GC morphology, stimulated axon growth and suppressed the repulsive axonal response to Sema3F (van der Vaart et al., 2013). In turn, KIF2A has been proposed to regulate axon pruning by preventing MT overstabilization in the GC. It was found that Kif2a^–/– mice exhibit an aberrant axonal overextension in hippocampal neurons, due to reduced MT depolymerization in the GCs (Homma et al., 2003; Maor-Nof et al., 2013).

As aforementioned, dynein-driven motor forces facilitate the entry of MTs into the GC periphery, influencing neurite initiation, axon outgrowth and steering (Dehmelt et al., 2006; Myers et al., 2006; Grabham et al., 2007). In support of dynein’s role in guidance-evoked GC movements, dynein loss-of-function experiments using RNAi or Cilibrevin D revealed an impairment in NGF-evoked filopodia formation and in GC turning over substrate boundaries. However, both dynein-driven MT-sliding into the GC periphery or MT-based retrograde transport of signaling endosomes could contribute to these instructed axon movements (Myers et al., 2006; Sainath and Gallo, 2015). Likewise, MT-based kinesin-dependent anterograde transport is necessary for axonal extension and steering. For instance, the MT-sliding activity of kinesin-5 – also called Eg5 or kif11 – inhibits the MT invasion into the GC periphery and it is required for GC turning in response to repulsive substrate boundaries. It was found that an asymmetric accumulation in the GC of the phosphorylated form of kinesin-5 precedes turning, and its acute inactivation in one side of the GC elicits the MT invasion into the hampered side and GC turning (Nadar et al., 2008, 2012).

A recent study has pinpointed the kinesin KIF13B as the molecular motor responsible of Myo X localization to axons upon netrin-1 stimulation. Myo X is an actin-based motor protein that transports lipids and transmembrane receptors, such as DCC, to the filopodia tip during axon pathfinding. It was found that netrin-1 signaling increases Myo X-KIF13B interaction and its anterograde MT-dependent transport along the axons, in order to stimulate axon initiation and axon branching in the cortical commissural projections (Yu et al., 2020). The kinesin family member 1 binding protein (KIF1BP) is also necessary for a proper development of the anterior commissures and the sympathetic innervation of the gut (Hirst et al., 2017). Mutations in the Kif1β gene, associated to the Charcot-Marie-Tooth peripheral neuropathy, have been found to prevent KIF1Bβ binding to the insulin-like growth factor 1 (IGF1) receptor IGF1R, involved in sensory axon guidance. These mutations blocked the MT-dependent axonal transport of IGF1R and inhibited IGF1-evoked axon outgrowth (Scolnick et al., 2008; Xu et al., 2018).

The kinesin-1 motor complex has also been suggested to participate in the netrin-1-evoked repulsive response in invertebrate motor axons and is a phosphorylation target of GSK3β, a major transduction hub of various guidance signaling pathways (Teulière et al., 2011; Banerjee et al., 2018). Furthermore, mutations in gene encoding the subunit KIF5C of the kinesin-1 complex (encoded by the Kif5 genes) have been linked to an abnormal development of the axon tracts of the corpus callosum and the internal capsule (Table 1) (Poirier et al., 2013; Michels et al., 2017). The recent analysis of a mutant mice lacking the kinesin-1 light chain KLC1 has revealed hypoplasia of the internal capsule tract, that includes corticofugal and thalamocortical axons. The innervation defects were found to be caused by an impaired kinesin-1-dependent axonal transport of the cannabinoid type-1 receptors (CB1R), and the subsequent axon unresponsiveness to endocannabinoids signaling (Saez et al., 2020).

Plus-end tracking proteins (+TIPs) regulate MT plus-end polymerization and stability, and mediate interactions between the MT ends and actin fibers, organelles and plasma membrane (van de Willige et al., 2016). Evidences obtained during the last 15 years have demonstrated that axon guidance signaling pathways directly target via regulation of +TIPs’ activity and localization (Figure 3) (Bearce et al., 2015; Cammarata et al., 2016; Voelzmann et al., 2016).

Microtubule end-binding (EB) proteins are the most abundant +TIPs in cells. EBs (EB1, EB2, and EB3) directly associate with MT plus-ends through their N-terminal calponin homology (CH) domain and are autonomous regulators of plus-end dynamics. MT tip-tracking of EBs mainly correlates with MT polymerization episodes, since it favors a continuous polymerization and reduces the number of catastrophes. Importantly, EBs and are also scaffold-providers for other +TIPs through their C-terminal domain, and for this reason EBs are considered master regulators of +TIP network. EB1 protein is ubiquitous, whereas EB3 is predominantly expressed by neurons, and both EB1 and EB3 are necessary for axon extension (Akhmanova and Steinmetz, 2015; van de Willige et al., 2016). Semaphorin4D can influence the EB3-labeled MT plus-ends polymerization dynamics in hippocampal neurons, and it was found that both EB1 and EB3 interact with the intracellular domains of the Plexin-A2, Plexin-B1, and Plexin-B3 Semaphorin family receptors (Laht et al., 2012, 2014). This data suggest that the semaphorin-plexin-EB pathway may regulate MT dynamics during axon pathfinding. When MT plus-ends enter the actin-rich P-domain of GCs, EB3 recruits the F-actin-binding protein Drebrin to couple growing MT tips to actin filaments. Hence, the EB3-Drebrin module facilitates the invasion of exploratory MTs into the GC periphery, and enables growth cone formation and neuritic elongation (Geraldo et al., 2008). It has been recently proposed that the EB1-Drebrin module at plus-ends interacts with the chemokine receptor type 4 (CXCR4) upon stromal cell-derived factor-1 (SDF-1) signaling. The chemokine SDF-1 can regulate axonal elongation and branching, and the SDF-1/CXCR4/Drebrin/EB1 pathway appears to be critical for SDF-1-induced MT cytoskeleton remodeling during neuronal motility (Pujol et al., 2005; Shan et al., 2021). Furthermore, EB3 colocalizes at the MT plus-ends within the GC filopodia with the stromal interacting molecule (STIM1), which is a calcium-sensing protein that mediates GC steering in response to various axon guidance cues (Mitchell et al., 2012; Pavez et al., 2019). New data revealed that EB3-STIM1 interaction at MT plus-ends is calcium-sensitive and STIM1 instructs the asymmetric invasion of EB3-labeled MT plus-end into the motile GC side downstream of BDNF or Sema3A signaling in sensory neuron cultures. Moreover, in vivo experiments in zebrafish showed that the EB3-STIM pathway regulates the axon guidance of spinal motor neurons (Pavez et al., 2019). In addition, EB1/3 proteins can bind MAP1B and Tau, and this interaction sequesters EBs from MT plus-ends and jeopardizes MT growth (Tortosa et al., 2013; Sayas et al., 2015). Because MAP1B and tau are downstream effectors of several axon guidance pathways, their interaction with EBs could indirectly influence the MT dynamics and the assembly of the +TIP complex on account of its dependence on EB scaffold.

One of the pioneer studies that assigned + TIPs a prominent role in axon guidance refers to the cytoplasmic linker associated protein (CLASP) and the Slit-evoked repellent response (Lee et al., 2004). CLASP decorates the MT plus-ends that polymerize over F-actin bundles in the GC periphery, and its overexpression causes MT overstabilization, looping and prevents their extension beyond the transition zone (Hur et al., 2011a). Orbit/MAST, the CLASP ortholog in invertebrates, is an Abelson tyrosine kinase (Abl) target downstream of Slit/Robo signaling that mediates repulsion. These evidences served the authors to propose that focal Slit stimulation in one side of the GC provokes asymmetric activation of the Abl-CLASP pathway and MT growth arrest, entailing a GC movement away of the source of Slit (Lee et al., 2004). CLASP is recruited to MT plus-ends through EB binding, although it contains tumor overexpressed gene (TOG) domains which can serve as tubulin-binding modules (Mimori-Kiyosue et al., 2005; Al-Bassam and Chang, 2011). Indeed, it has been observed that CLASP localization in the MTs can alternate between the plus-end and the MT lattice, based on its phosphorylation by GSK3β. These MT-binding activities determine the degree of MT protrusion and subsequent axon growth in an opposing manner. A high GSK3 kinase activity promotes CLASP dissociation from MT plus-ends, leading to MT destabilization and impaired axon growth, while a moderate GSK3 activity allows CLASP plus-end binding, promoting MT stabilization and axon extension. A low GSK3 activity leads to CLASP localization to the MT lattice, producing MT overstabilization and looping in the GCs, and axon growth attenuation (Akhmanova et al., 2001; Hur et al., 2011a). Given that GSK3 kinase activity is fine-tuned by many downstream axon guidance pathways, CLASP may also act as transducing factor of other extracellular guidance cues (Hur and Zhou, 2010).

APC (Adenomatous Polyposis Coli Protein) is a critical tumor suppressor, initially reported as Wnt-signaling regulator. In the Wnt pathway, APC forms a complex with GSK3 and other proteins to target and degrade the oncoprotein β-catenin (Stamos and Weis, 2013). In addition to this function, APC is an EB-binding +TIP that stabilizes the MT plus-ends and, similar to CLASP, this activity is abolished by GSK3β-mediated phosphorylation (Nakamura et al., 2001; Zumbrunn et al., 2001). In neurons, APC is transported toward the distal region of the growing axon by kinesin-1 motor forces, and distributes asymmetrically within the GC. Indeed, the local accumulation of APC in one side of the GC anticipates the steering movement of the axon in this axial direction (Koester et al., 2007; Ruane et al., 2016). It was demonstrated that the focal stimulation of GCs with Nerve Growth Factor (NGF) produces the localized inactivation of GSK3β via PI3K activity, which enables APC-dependent stabilization of MT plus-ends in GCs and rapid axon elongation (Zhou et al., 2004). Additionally, the treatment with the GC-pausing guidance cue Wnt3a led to altered MT growth directionality in the GC by misallocating APC from the MT plus-ends at the P-domain (Purro et al., 2008). In vivo, despite initial contradictory results obtained in Drosophila, APC has been shown to play an important role in neural circuits formation. APC mutant mice exhibit gross misrouting defects in the internal capsule, posterior commissure or thalamocortical axons, and APC-deficient neurons displayed an abnormal axonal arborization and curling at the tips (Rusan et al., 2008; Yokota et al., 2009; Jin et al., 2018). Besides its MT-stabilizing role at the plus-end, APC participates in the MT-based transport of mRNAs, such as those encoding β-actin, Tubb2b or the dynein complex subunit Lis1, toward the axon. Importantly, APC association with their mRNA targets to transport them along sensory axons is triggered by exogenous stimulation with NGF (Preitner et al., 2014; Villarin et al., 2016; Baumann et al., 2020).

APC2, APC’s brain specific homolog, is a MT-binding protein and contains a C-terminal region with MT tip-tracking properties. APC2 localizes to GCs of chick retinal axons and participates in retinotectal axon guidance through regulation of MT stability. Apc2-knockdown display an attenuated response to ephrin-A2 in retinal ganglion cells (Shintani et al., 2009; Kahn et al., 2018). Also in retinal neurons, APC2 has been identified as a direct target of the transcription factor Zic2, the main determinant of axon midline avoidance, which also regulates the guidance receptors EphB1 and Unc5c (Herrera et al., 2003, 2019b; Escalante et al., 2013; Kridsada et al., 2018; Murcia-Belmonte et al., 2019). In ipsilaterally projecting neurons, Apc2 expression is intrinsically downregulated by Zic2 likely to facilitate Wnt5a and ephrinB2-mediated axon repulsion at the optic chiasm (Morenilla-Palao et al., 2020).

Microtubule-actin crosslinking factor 1 (MACF1), also known as actin-crosslinking factor 7 (ACF7), is a large multidomain protein of the spectraplakin family, highly expressed in the nervous system. MACF1 interacts with MT plus-ends and enables MT capture by F-actin, facilitating MT polymerization over F-actin bundles at the cellular periphery (Kodama et al., 2003; Wu et al., 2008). MACF1 can directly interact with MTs through its C-terminal Gas2-related (GAR) domain or indirectly by EB binding, and simultaneously binds F-actin through its N-terminal calponin-homology (CH) domains. In addition, MACF1 has a C-terminal AAA-ATPase domain that can exert molecular forces over the MT cytoskeleton (Moffat et al., 2017). Genetic studies in Drosophila showed that MACF1 homolog protein Shot is required for axon extension and midline guidance, and that its MT plus-tip tracking enabled by EB1-binding is necessary to maintain an organized MT network in axons (Lee et al., 2007; Alves-Silva et al., 2012). Consistently, mammalian MACF1 also regulates neuronal MTs configurations and filopodia formation, a role dependent on both MACF1 F-actin- and MT-binding domains (Sanchez-Soriano et al., 2009). MACF1 mediates Wnt/GSK3β signaling, and its loss-of-function in mice phenocopied the early developmental defects observed in Wnt3^–/– embryos. Specifically, the conditional deletion of MACF1 in neural progenitors produced the agenesis of the anterior commissure and an abnormal development of the thalamocortical fibers and the hippocampal commissure in neonatal mice (Chen et al., 2006; Goryunov et al., 2010). Moreover, MACF1 downregulation in cortical early postmitotic neurons interfered with the normal arrangement of MTs and F-actin networks in neurites, inhibited neuron radial migration and disrupted callosal axon innervation (Ka et al., 2014; Ka and Kim, 2016). Interestingly, heterozygous missense mutations in the MT-binding GAR domain of MACF1 have been recently identified in human individuals exhibiting axonal midline crossing phenotypes, among other defects (Table 1) (Dobyns et al., 2018).

Neuron navigator-1 (NAV1) belongs to the +TIP family of Navigators (NAVs), which is represented by NAV1, NAV2, and NAV3 in mammals. NAVs are large proteins, carrying N-terminal calponin-homology (CH) domains and an intriguing C-terminal ATPase domain, which have been associated to axon outgrowth (Martínez-López et al., 2005; van Haren et al., 2009; McNeill et al., 2010; Abe et al., 2014). In particular, NAV1 expression was found to be largely restricted to the developing nervous system being enriched in the neuritic tips and GCs. Hindbrain neurons lacking NAV1 do not respond to Netrin-1, which suggested a function downstream of Netrin-1 signaling (Martínez-López et al., 2005; van Haren et al., 2014). It was recently described that, similar to CLASP or MACF1, NAV1 is an EB-dependent +TIP that can directly bind actin fibers in vitro, and data suggest that it crosslinks MT plus-ends to the F-actin network within the GCs from mammalian cortical neurons (Sánchez-Huertas et al., 2020). In the proposed model, EB proteins recruit NAV1 to the MT tip during polymerization inside F-actin-enriched regions. Following EB-complex disassembly and MT growth arrest, NAV1 switches to an EB-independent form of association with the MT plus-end and stabilizes it, reducing the frequency of MT shrinkage. Thereafter, paused plus-ends undergo retrograde translocation coupled to F-actin retrograde flow via MT-NAV1-F-actin crosslinking (Sánchez-Huertas et al., 2020). However, NAV1 sequence does not possess a CH domain for actin binding, neither GAR nor TOG domains for direct MT interaction. Hence, the specific NAV1 domains responsible for direct F-actin binding and whether NAV1-MT interaction requires an intermediary autonomous MT-binding protein, still remain to be elucidated. NAV1 was also found to mediate the chemoattractive response of cortical axons toward a source of netrin-1 and the radial migration of pyramidal neurons during in vivo corticogenesis (Sánchez-Huertas et al., 2020). NAV1 mRNA and protein levels are highly enriched in developing cortical layer V, mainly populated by projection neurons innervating subcortical targets, such as the brainstem or the spinal cord (Martínez-López et al., 2005; Sorensen et al., 2015). This observation suggests that NAV1 might be required for axonal navigation by layer V projection neurons in particular, and allows to hypothesize that ad hoc neuron cytoskeletal machinery may transduce guidance signaling differently in specific neuron subtypes.

Recent evidences suggest that the module formed by the +TIPs XMAP215 (chTOG or CKAP5 in mammalian cells) and transforming acidic coiled-coil 3 (TACC3) protein represent an unconventional EB-independent regulatory mechanism of MT plus-end dynamics downstream axon guidance signaling. XMAP215 is a conserved processive MT polymerase that catalyzes tubulin addition into the polymer while it tracks the MT plus-ends (Gard and Kirschner, 1987; Brouhard et al., 2008). Although XMAP215 and EB1 can act synergistically to promote MT growth, XMAP215 does not require EB proteins to track MT plus-ends because it binds MTs directly through its five N-terminal TOG domains. Indeed, XMAP215 locates to the extreme MT plus-end several tens of nanometers ahead of the region bound by EB1 and remains attached to the MT plus-end even during shrinkage events (Nakamura et al., 2012; Zanic et al., 2013; Maurer et al., 2014). XMAP215 downregulation greatly increases MT catastrophe frequency throughout the neuron cell body and compromises hippocampal axon growth (van der Vaart et al., 2012). While in most cellular contexts XMAP215 downregulation decreases MT plus-end growth, in GCs it accelerates MT plus-end velocities. This increase was proposed to arise from higher MT anterograde translocation rates in the GCs, likely due to the uncoupling between MT plus-ends and the F-actin retrograde flow in the absence of XMAP215 (Lowery et al., 2013). More recently, it was reported that XMAP215 directly binds actin fibers and it is necessary for MT-F-actin alignment in the GCs. Indeed, it has been demonstrated that XMAP215 regulates MT invasion into GC filopodia, influences GC morphology and protrusion, and mediates the repulsive response to ephrinA5 (Slater et al., 2019).

TACC3, first identified as a regulator of astral and spindle MT length, has been classified as + TIP on account of its binding to MT plus-ends through its TACC domain and assigned a role in plus-end dynamics and axon outgrowth (Gergely et al., 2000; Nwagbara et al., 2014). TACC3 interacts with XMAP215 in the distal region of MT plus-ends, and they are important for each other’s localization to the plus-end. Indeed, TACC3 and XMAP215 can rescue each other’s downregulation phenotypes in axon elongation, and it has been suggested that TACC3 strengthens the XMAP215-TACC3 complex binding to MTs in order to drive polymerization activity (Nwagbara et al., 2014; Erdogan et al., 2017). TACC3 is a phosphorylation target of the kinase Abl, whose activity is known to be regulated by axon guidance signaling (Kannan and Giniger, 2017). A TACC3 phospho-null mutant failed to localize at MT plus-ends in GCs, leading to an increase of MT invasion into the filopodia and impaired axon pathfinding. Interestingly, the overexpression of TACC3 interfered with the responsiveness of axons from Xenopus neurons explants upon Slit2 and Ephrin-A5 signaling (Erdogan et al., 2017, 2020).