Insight into the regulation of axonal transport from the study of KIF1A-associated neurological disorder

Intracellular transport in neurons is fundamental for their function and homeostasis because most proteins required for the function of axons and nerve terminals are synthesized in the cell body and need to be transported to their sites of action (Hirokawa et al., 2009). Membrane organelles, protein complexes and mRNAs are transported bidirectionally along neuronal axons by molecular motor proteins (Fig. 1A). Kinesins, also known as kinesin superfamily proteins (KIFs), are molecular motors that use the chemical energy of ATP to transport cargoes along microtubules. In the axon, microtubule plus-ends point toward the axon terminal (Burton and Paige, 1981). KIFs generally move towards the plus-end of microtubules and therefore transport cargoes towards nerve endings. Each KIF has a conserved motor domain and a specific tail domain (Hirokawa et al., 2009). The conserved motor domain is essential for the movement along microtubules (Scholey et al., 1989; Yang et al., 1990), whereas the tail domain binds to cargoes. The cargo transported by a KIF is, therefore, determined by the tail domain. In addition to transporting cargoes, some KIFs can regulate microtubule dynamics in the cell by using motor domains that have microtubule stabilizing or destabilizing activities (Ogawa et al., 2004; Zhou et al., 2009; Cheng et al., 2014; van der Vaart et al., 2013; Taguchi et al., 2022; Niwa et al., 2012; Shima et al., 2018; Peet et al., 2018).

This Review focuses on one of the KIFs, KIF1A, because the functions of other kinesins have been reviewed in detail previously (Ou and Scholey, 2022; Niwa, 2015; Hirokawa et al., 2009). Here, we first outline the discovery of KIF1A and its cargoes, before discussing its regulation. Finally, we describe KIF1A-associated neurological disorder (KAND) and what has been learned from its analysis with regard to the motor activity and the regulation of KIF1A.

We briefly outline here how KIF1A and its orthologs were discovered. Kinesin (now also called the kinesin-1 family, comprising kinesins containing the heavy chains KIF5A, KIF5B and KIF5C) was originally discovered as an axonal transport motor in squid, cow and chick in the late 1980s (Vale et al., 1985; Brady, 1985), and since then many kinesin orthologs have been described. KIF1A belongs to the kinesin-3 subfamily, and a KIF1A ortholog was originally identified by molecular genetics in Caenorhabditis elegans. Among the worm mutants showing uncoordinated (unc) movement, unc-104 mutants had mutations in a gene encoding a kinesin ortholog (Otsuka et al., 1991; Hall and Hedgecock, 1991). UNC-104 has a conserved N-terminal domain, now known as the kinesin motor domain, but the tail domain is entirely different from that of kinesin-1 proteins (Fig. 1B), and has a phosphatidylinositol-binding pleckstrin homology (PH) domain. In an unc-104 allele, unc-104(e1265), the PH domain has a missense mutation (Klopfenstein and Vale, 2004; Kumar et al., 2010). The PH domain of KIF1A binds to artificial vesicles containing phosphatidylinositol 4,5-bisphosphate (PIP2) (Klopfenstein and Vale, 2004). Therefore, the PH domain is considered essential for association with cargo vesicles. Interestingly, unc-104 mutant worms have defective synaptic structures and a significantly reduced number of synaptic vesicles (SVs) at synapses; instead, SV-like clear vesicles accumulate in neuronal cell bodies in unc-104 mutant worms (Hall and Hedgecock, 1991). Some synapses are formed along the axon, but the number of active zones is reduced. Based on these observations, it has been suggested that UNC-104 is a molecular motor that transports either SVs or their precursors (Hall and Hedgecock, 1991).

Mammalian KIF1A was identified from mouse brain by degenerate PCR amplification of the motor domain sequence (Aizawa et al., 1992; Okada et al., 1995), and comparison of the sequence revealed that KIF1A is an ortholog of UNC-104. KIF1A binds to vesicles containing some SV membrane proteins, but some SV membrane proteins are in vesicles not transported by KIF1A (Okada et al., 1995). The function of KIF1A has been analyzed by gene targeting technology (Yonekawa et al., 1998). Similar to what is seen in unc-104 mutant worms, the number of SVs is reduced in KIF1A-knockout (KO) mice. Mammals have also a KIF1A homolog named KIF1Bβ (Zhao et al., 2001), and KIF1A and KIF1Bβ have redundant roles in mammalian neurons (Niwa et al., 2008). The ortholog of KIF1A in Drosophila melanogaster remained elusive until the late 2000s. In Drosophila, immaculate connections (imac) mutants have mutations in a gene encoding a KIF1A ortholog (Pack-Chung et al., 2007). In imac mutants, synaptic boutons do not correctly mature, and SVs and dense core vesicles (DCVs) are diminished at nerve endings, which contain very few active zones (Pack-Chung et al., 2007). Together, these findings indicate that imac is required for synaptic bouton formation and SV transport. Collectively, KIF1A is an evolutionarily conserved motor protein.

Given that the primary function of KIF1A is cargo transport, we will discuss in detail the cargoes transported by KIF1A (Fig. 2). Membrane proteins that are localized on SVs are originally synthesized in the cell body and transported to synapses (Nakata et al., 1998), whereas neurotransmitter-containing mature SVs are produced by endocytosis (Regnier-Vigouroux et al., 1991). Originally, it was suggested that KIF1A binds to and transports vesicles that carry some of the SV membrane proteins, but other SV membrane proteins are not part of these vesicles (Okada et al., 1995). We and others have therefore described the axonal transport by UNC-104/KIF1A as ‘axonal transport of SV precursors’ and not ‘axonal transport of (mature) SVs’ (Hirokawa et al., 2009; Okada et al., 1995; Niwa et al., 2016, 2008). However, mature SVs are also transported along the mammalian axon (Darcy et al., 2006), and work in C. elegans suggests that mature SVs are transported by UNC-104 (Edwards et al., 2015). In temperature-sensitive unc-104 mutants, mature SVs are generated and accumulate normally at synapses at the permissive temperature. When these temperature-sensitive mutant worms are transferred to a restrictive temperature, the localization of SVs changes from synapses to the cell body and dendrite, suggesting that the localization of mature SVs once generated at synapses is maintained by the UNC-104 activity (Edwards et al., 2015).

DCVs are also important organelles that localize to synapses. They contain peptidergic neurotransmitters and neuropeptides, but unlike SVs, they are produced in the cell body and transported to synapses (Tooze et al., 2001). KIF1A binds to DCVs (Stucchi et al., 2018), and accordingly, DCVs are not properly localized at synapses in worms and flies that have defects in UNC-104 or Imac (Hall and Hedgecock, 1991; Zahn et al., 2004; Pack-Chung et al., 2007). Furthermore, transport of BDNF-containing DCVs is reduced in KIF1A-KO mice and KIF1A-knockdown neurons (Kondo et al., 2012; Lo et al., 2011). Defects in this transport also lead to brain malformation in rats with KIF1A knockdown (Carabalona et al., 2016).

Active zone precursor vesicles, also called Piccolo–Bassoon transport vesicles, are transported along the axon (Shapira et al., 2003). It was originally believed that KIF1A does not transport active zone precursor vesicles in mammalian neurons (Okada et al., 1995). However, more recent studies from invertebrates have suggested that KIF1A might transport some of the active-zone proteins (Pack-Chung et al., 2007; Oliver et al., 2022). In Drosophila imac mutants, synapses do not properly mature at nerve endings, and observation of active zone markers during nerve development indicates that some of the active-zone proteins are transported by Imac (Pack-Chung et al., 2007). This is consistent with a recent study in C. elegans showing that neurexin, a presynaptic cell adhesion protein, is transported by UNC-104 (Oliver et al., 2022).

In the original unc-104 study (Hall and Hedgecock, 1991), a reduced number of synapses was described, but little attention was paid to this observation at the time. Subsequently, in KIF1A-KO mice, a reduced number of synapses was also observed (Yonekawa et al., 1998). Conversely, overexpression of KIF1A increases the number of synapses and facilitates experience-dependent learning in transgenic mice (Kondo et al., 2012). These data indicate that KIF1A is a molecular motor that is essential for correct synaptic development, rather than merely transporting SVs and SV precursors. Axonal transport of active-zone proteins appears to depend on KIF1A, although further analysis in mammalian systems is needed.

The axonal membrane receptor TrkA (also known as NTRK1) is also transported by KIF1A (Tanaka et al., 2016). TrkA is a nerve growth factor (NGF) receptor and is essential for sensory receptor function in dorsal root ganglion neurons, which is consistent with findings of human genetic diseases. Indeed, it has been shown that KIF1A mutation leads to hereditary sensory and autonomic neuropathy type 2 (HSAN2) (MIM: 614213) (Riviere et al., 2011), in which patients are insensitive to pain. A very similar sensory neuropathy is caused by mutations in defects in TrkA (MIM: 256800) (Greco et al., 1999).

A monomer-to-dimer transition model has been proposed for KIF1A activation, which is supported by many studies to date. KIFs walk on microtubules by the alternate use of two motor domains (Fig. 1A) (Hancock and Howard, 1998), and their coiled-coil domains thus form a dimer (Fig. 1B). Nevertheless, KIF1A was originally identified as a unique monomeric motor (Okada et al., 1995). Although KIF1A has several coiled-coil domains, purified full-length KIF1A exists as a monomer. Interestingly, an single-molecule assay has shown that the monomeric motor domain of KIF1A, generated by deleting the other domains, processively moves on microtubules by biased Brownian motion, suggesting that the coiled-coil domain is not essential for the motility (Okada and Hirokawa, 1999). This raised the question of whether the full-length KIF1A therefore transports cargo vesicles as a monomer by biased Brownian motion in the axon. However, unlike the monomeric motor domain of KIF1A, purified full-length KIF1A moves on microtubules in a unidirectional manner, rather than in the biased Brownian motion seen in single-molecule assays (Chiba et al., 2019; Budaitis et al., 2021). Interestingly, the motility of the KIF1A motor domain is greatly enhanced by artificial dimerization (Tomishige et al., 2002; Soppina et al., 2014). In fact, the unidirectional motility of dimerized KIF1A motor domains is similar to that of full-length KIF1A, suggesting that on microtubules, full-length KIF1A is a dimer that uses two motor domains to walk unidirectionally. Why then does recombinant full-length KIF1A purify as a monomer? Several molecular motors are inactivated by autoinhibition, prompting the idea that autoinhibited KIF1A is a monomer and activated KIF1A is a dimer (Fig. 3A) (Tomishige et al., 2002). This hypothesis is supported by several lines of evidence. KIF1A binds to PIP2 on cargo vesicles (Klopfenstein and Vale, 2004; Kumar et al., 2010). Increasing the local concentration of KIF1A by artificially creating PIP2 microdomains on liposomes causes KIF1A to dimerize and become activated (Tomishige et al., 2002). Cryo-electron microscopy has shown that an UNC-104 deletion mutant consisting of the motor domain (MD), neck coiled-coil (NC) and coiled-coil 1 (CC1) domains is monomeric in the AMPPNP-binding state (ATP-binding state) and dimeric in the nucleotide-free and ADP-binding states (Al-Bassam et al., 2003). These structures suggest that, in the monomeric state, CC1 binds to NC, which would inhibit dimerization, whereas in the dimeric state, CC1 and NC form homodimers (Al-Bassam et al., 2003). High-resolution intramolecular interactions have been revealed by X-ray crystallography of other kinesin-3 members, including KIF13B and KLP-6 (Fig. 3B,C); KIF13B and KLP-6 are paralogs of KIF1A/UNC-104 and have very similar MD-NC-CC1 sequences to that in KIF1A (Ren et al., 2018; Wang et al., 2022). The MD-NC-CC1 construct of KIF13B is a monomer in solution, and its crystal structure revealed that the CC1 forms intramolecular interactions with the NC and MD (Ren et al., 2018) (Fig. 3C). The neck linker is a short sequence between the MD and NC. Undocking of the neck linker and subsequent ADP release is an important step for KIFs to processively move on microtubules (Case et al., 2000). This critical step is inhibited in the monomeric state because the intramolecular interaction between CC1, NC and MD locks down the entire neck domain (Ren et al., 2018). Interestingly, this mechanism is conserved in kinesin-1 members, although kinesin-1 is a constitutive dimer and its autoinhibited structure is totally different (Kaan et al., 2011). Recent work has shown that full-length KLP-6 is monomeric in solution and undergoes similar intramolecular interactions to those found in KIF13B (Wang et al., 2022). This raised the question of whether these intramolecular interactions inhibit the motor activity in vivo? This is indeed the case, as the presence of the CC1 and CC2 domains negatively regulate the activity of KIF1A (Hammond et al., 2009). Moreover, we found gain-of-function unc-104 mutants, in which SVs are over-transported (Wu et al., 2013; Niwa et al., 2016). The mutated residues in these gain-of-function mutants, UNC-104(V6I), UNC-104(E412K) and UNC-104(E612K) are well conserved in KIF1A, which map to the MD, CC1 and CC2 domains, respectively (Fig. 3C; Niwa et al., 2016). Interestingly, the residue corresponding to UNC-104(E412)/KIF1A(E448) directly binds to the motor domain in the crystal structure of KIF13B (E399 in Fig. 3B,C; Ren et al., 2018). Taken together, CC1 and CC2 appear to have key roles in the regulation of KIF1A. In the case of kinesin-1 members, it is thought autoinhibition of the motor protein exists to prevent its activation without cargoes (Kaan et al., 2011). However, the over-transport phenotype in CC1 and CC2 mutant worms suggests that autoinhibition of KIF1A regulates the amount of cargo transport rather than merely preventing energy wastage. Unlike what is seen in kinesin-1 members, the KIF1A domains that are required for the autoinhibition, such as CC1 and CC2 domains, are located away from the cargo-binding stalk domain and PH domain in KIF1A (Fig. 1B). These structural differences might have allowed the regulation of the amount of cargo transport by autoinhibition in KIF1A.

Although the PH domain of KIF1A binds to PIP2, PIP2 is broadly distributed on the plasma membrane and is not specifically enriched at synapses, nor at SVs (van Rheenen et al., 2005). Therefore, other factors are thought to be required for the specific recognition of KIF1A cargo vesicles. Indeed, a mini-motor consisting of the motor domain and the PH domain cannot rescue unc-104-mutant worms, indicating that the stalk domain is the interface with adaptors and regulators (Klopfenstein and Vale, 2004).

The search for interacting proteins identified MADD (also known as DENN) as an adaptor that links KIF1A and RAB3-carrying SV precursors (Niwa et al., 2008; Hummel and Hoogenraad, 2021). Liprin-α, an active zone scaffolding protein, also binds to KIF1A (Stucchi et al., 2018; Shin et al., 2003; Wagner et al., 2009), and it has been suggested that binding to liprin-α promotes the dimerization of KIF1A (Wagner et al., 2009). In addition, calmodium (CaM) has been identified as a regulator for DCV transport, and Ca²⁺ activates KIF1A-dependent transport of DCVs through CaM (Stucchi et al., 2018).

A search for mutant worms with mislocalized SVs identified an arl-8 mutant (Klassen et al., 2010). ARL-8 in worms, and its homologs ARL8A and ARL8B in mammals, is a small GTPase that localizes on SVs and lysosomes (Klassen et al., 2010; Nakae et al., 2010; Hofmann and Munro, 2006). The GTP form of ARL-8/ARL8 binds to the CC3 domain of UNC-104/KIF1A (Wu et al., 2013; Niwa et al., 2016), and this binding might overcome the autoinhibition of UNC-104/KIF1A. In this pathway, BLOC-1 related complex (BORC) is an activator for ARL-8/ARL8 (Pu et al., 2015; Niwa et al., 2017). In mutant worms lacking BORC subunits, SVs are not properly transported to synapses in the axon. Expression of mammalian ARL8A or ARL8B can rescue defects in axonal transport of SV precursors in worm arl-8 mutants; however, it remains controversial whether this pathway is conserved in mammalian neurons (Klassen et al., 2010; De Pace et al., 2020).

In addition to these binding partners, polyglutamylation of α-tubulin has been identified as a positive regulator of KIF1A (Ikegami et al., 2007). Polyglutamylation of α-tubulin is reduced in ROSA22 mice, which have a mutation in the tubulin polyglutamylase complex. Entry of KIF1A into the axon, as well as the axonal transport of SV precursors, are reduced in ROSA22-derived neurons (Ikegami et al., 2007). In vitro assays show that the run length of KIF1A is longer on polyglutamylated microtubules than unmodified microtubules (Lessard et al., 2019). The negatively charged K-loop of KIF1A MD associates with the positively charged polygulamylated α-tubulin. This interaction inhibits the dissociation of KIF1A from the polyglutamylated microtubules. Moreover, microtubule-associated protein 9 (MAP9) increases the activity of KIF1A on microtubules (Monroy et al., 2020). Furthermore, in vitro, KIF1A binds to microtubules more frequently when MAP9 is localized on microtubules (Monroy et al., 2020).

In addition to the above positive factors, a negative regulator has also been described – KIF1-binding protein KIAA1279, now also called kinesin-binding protein (KBP or KIFBP) (Kevenaar et al., 2016). KBP was originally identified as a cargo adaptor for KIF1C (Wozniak et al., 2005). Human genetics data shows that homozygous mutations in KBP are linked to Goldberg–Shprintzen syndrome (GOSHS; MIM: 609460) (Brooks et al., 2005). GOSHS patients suffer from a broad spectrum of neuronal symptoms, including megacolon, impaired intellectual development, microcephaly and dysmorphic facial features (Brooks et al., 2005). Subsequent studies have shown that KBP directly binds to the motor domain of some KIFs, including KIF1A, and inhibits their motor activity, rather than serving as a cargo adaptor (Kevenaar et al., 2016; Malaby et al., 2019). Purified KBP inhibits the binding of KIF1A to microtubules. Overexpression of KBP in worm neurons inhibits axonal transport of SV precursors (Kevenaar et al., 2016). It remains elusive whether the axonal transport of SVs or their precursors is hyperactivated in neurons from GOSHS patients.

Mutations in human KIF1A were originally found in families suffering from HSAN2 (MIM: 614213) and hereditary spastic paraplegia (SPG) (Riviere et al., 2011; Klebe et al., 2012). For instance, a homozygous deletion mutation in the tail domain of KIF1A was found in a family suffering from HSAN2 (Riviere et al., 2011). SPGs are a heterogeneous group of lower motor neuron diseases that are characterized by spasticity in the lower limbs, and many causative genes for SPG have been identified. By analyzing SPG type 30 (SPG30) families, homozygous KIF1A(A255V) and KIF1A(R350G) mutations were found (MIM: 610357) (Klebe et al., 2012).

Whole-genome sequencing has identified numerous KIF1A mutations in patients suffering from a broad spectrum of neuronal diseases. KIF1A mutations have also been associated with neurodegeneration and spasticity with or without cerebellar atrophy or cortical visual impairment (NESCAV) syndrome (MIM: 614255) (Hamdan et al., 2011; Okamoto et al., 2014; Esmaeeli Nieh et al., 2015; Lee et al., 2015). Other studies have identified KIF1A mutations in progressive encephalopathy with edema, hypsarrhythmia and optic atrophy (PEHO) syndrome (Langlois et al., 2016). In most of the cases, patients do not have a family history. Because the common cause of these broad-spectrum of neurological diseases is KIF1A, these diseases are now collectively termed KIF1A-associated neurological disorder (KAND) (Boyle et al., 2021) (Fig. 4).

De novo mutations are a major cause of KAND. Interestingly, most disease-associated mutations are missense mutations in the motor domain of KIF1A, indicating that the motility of KIF1A is affected by disease mutations. This hypothesis has been experimentally analyzed by several in vitro assays, and the variants KIF1A(T99M), KIF1A(R216C) and KIF1A(E253K) have been shown to exhibit diminished activity in microtubule-gliding assays (Esmaeeli Nieh et al., 2015). KAND mutations have also been shown to strongly inhibit the motility of KIF1A on microtubules in single-molecule assays (Boyle et al., 2021). Depending on the mutated residue, disease-associated KIF1A mutants exhibit reduced microtubule association, reduced velocity and run length, or increased non-motile rigor microtubule association. For instance, the KIF1A(V8M) mutant has defects in force generation and is slower than wild-type KIF1A (Budaitis et al., 2021), whereas KIF1A(P305L) affects the association of KIF1A with microtubules (Lam et al., 2021). The P305 residue is conserved and part of an unusual 310 helix immediately close to the K-loop, which facilitates a high rate of association with microtubules (Okada and Hirokawa, 1999; Lam et al., 2021). The KIF1A(E239K) mutation over-stabilizes the intramolecular interaction between MD and neck domain, and as a result, KIF1A(E239K) disrupts the motor activity and induces neurological symptoms (Morikawa et al., 2022). Heterodimers composed of wild-type KIF1A and disease-associated KIF1A have reduced motility, suggesting that KAND mutations work in a dominant-negative manner (Anazawa et al., 2022). In addition to these in vitro studies, we have generated KAND model worms to analyze the effect of KAND mutations on axonal transport and synapses, with disease-associated mutations introduced into C. elegans by CRISPR/cas9 (Anazawa et al., 2022) (Fig. 5). Consistent with the in vitro assays, KIF1A(R11Q), KIF1A(R254Q) and KIF1A(P305L) mutant worms showed reduced axonal transport. There were fewer SVs at synapses, and instead, they accumulated in the cell body and in dendrites (Fig. 5). These worms show defects in body movement. Moreover, we found that these mutations work in a dominant-negative manner, rather than simply inducing a loss of function (Anazawa et al., 2022). Similar dominant-negative effects for a KAND mutation have been observed in rat brain experiments (Carabalona et al., 2016).

However, not all KAND-associated mutations result in loss of function. For instance, KIF1A(A255V), a familial mutation associated with SPG30, does not strongly affect the motility of KIF1A in vitro, despite the mutation being in the motor domain (Esmaeeli Nieh et al., 2015). C. elegans genetics identified a motor domain mutation, KIF1A(V8I), that results in the over-transport of SV precursors (Niwa et al., 2016). A subsequent study found that a similar mutation, KIF1A(V8M), caused familial SPG (Iqbal et al., 2017). These findings indicate that some KAND mutations might increase axonal transport, rather than reduce it. Indeed, in unc-104(V6M) and unc-104(A252V) mutant worms established by CRISPR/cas9, mimicking KAND-associated KIF1A(V8M) and KIF1A(A255V) mutations, respectively, SVs abnormally accumulate in the tips of axons, indicating hyperactivation of axonal transport (Fig. 5) (Chiba et al., 2019). In agreement with this, the number of SVs at each en passant synapse is lower than in wild type because of the increased transport (Chiba et al., 2019). This phenotype is clearly different from worms with loss of function of unc-104 and worm models of loss of KAND-related KIF1A function (Fig. 5) (Hall and Hedgecock, 1991; Anazawa et al., 2022). Moreover, single-molecule assays have shown that full-length KIF1A(V8M), KIF1A(R255V) and KIF1A(R350G) bind to microtubules and are able to move on them with high frequency, whereas wild-type KIF1A is autoinhibited and rarely binds to microtubules (Chiba et al., 2019). Some KAND-associated mutations might, therefore, unlock KIF1A autoinhibition and so increase axonal transport. Interestingly, some studies have shown that KIF1A(V8M), KIF1A(R255V) and KIF1A(R350G) show reduced physical parameters, including velocity, run length and force generation, in single-molecule assays (Budaitis et al., 2021; Guedes-Dias et al., 2019). Furthermore, experiments in cells show that intracellular transport mediated by KIF1A(V8M) is slower than that of wild-type KIF1A (Budaitis et al., 2021). How can this contradiction be explained? One possibility is that worm UNC-104 and/or worm neurons might be different from their human equivalents. Another possibility is that there is a threshold in the physical characteristics of KIF1A. A slightly slower transport or shorter run length might not substantially disturb axonal transport and neuronal function, whereas the effect of an elevated microtubule association might be more pronounced. The optimal physical parameters for axonal transport remain elusive, but analyzing KAND mutations might clarify these characteristics.

Great progress has been made in elucidating the function of KIF1A in axonal transport and synapse formation, and in determining the basis of KIF1A autoinhibition, as well as how mutations in KIF1A lead to neuronal diseases. However, many issues remain to be solved, as outlined below.

Firstly, how are the loading and unloading of cargo vesicles regulated? In gain-of-function unc-104 mutant worms, including KAND models, transport of SVs is increased, and SVs accumulate at the axonal tip (Niwa et al., 2016; Chiba et al., 2019). As a result, the number of SVs is reduced in en passant synapses (Fig. 5). This indicates that autoinhibition of KIF1A at the synapse is essential for cargo unloading and for determining the number of SVs at each synapse. However, it remains unclear how each synapse monitors the number of SVs and communicates this information to KIF1A in order to maintain a constant the number of SVs. Worm genetics is one of the powerful ways to address this problem. Other motors such as dynein, UNC-116/KIF5 (kinesin-1 proteins) and VAB-8/KIF26 (kinesin-11 proteins), are also involved in determining the localization of SVs (Ou et al., 2010; Balseiro-Gomez et al., 2022). In this context, the UNC-116(G274R) and UNC-116(E432K) mutations are interesting, as they partially suppress the phenotype of unc-104 loss-of-function (Kurup et al., 2015). UNC-116 (kinesin-1)-dependent axonal transport of SV precursors might be increased in these mutants. Analyzing the interplay between KIF1A and other motors, as well as microtubule-associated proteins, both in vivo and in vitro is another important area for future exploration (Derr et al., 2012; Gicking et al., 2022). Moreover, reconstitution of KIF1A activation and inactivation in vitro is required to fully understand how axonal transport by KIF1A is regulated. In the case of other motors, such as kinesin-1 and dynein, activation has been reconstituted in vitro (McKenney et al., 2014; Chiba et al., 2022). Whereas purified full-length dynein and kinesin-1 are inactive, mixing purified cargo adaptor proteins activates their motility on microtubules in single-molecule assays (McKenney et al., 2014; Chiba et al., 2022). Although activation of full-length KIF1A by KAND-associated point mutations has been observed (Chiba et al., 2019), such activation by KIF1A-binding proteins has not been accomplished in vitro. It will be interesting to test whether KIF1A binding proteins and activators are able to overcome the autoinhibition of full-length KIF1A.

A second outstanding issue is the full-length structure of KIF1A. Although the structures of KIF1A orthologs and paralogs have been solved (Al-Bassam et al., 2003; Ren et al., 2018; Wang et al., 2022), we still cannot understand from those why some KIF1A domains and KAND-associated residues are required for KIF1A autoinhibition. For instance, it is known that CC2 negatively regulates the activity of KIF1A and UNC-104 (Lee et al., 2004; Hammond et al., 2009; Wu et al., 2013; Niwa et al., 2016), but this domain is not included in the available KIF13B structure (Ren et al., 2018). Moreover, the CC2 domain of KLP-6 is much shorter than that of KIF1A and not well conserved (Wang et al., 2022). Furthermore, the KAND-associated KIF1A(V8M) mutation is thought to activate KIF1A by unlocking the autoinhibition; however, the current structures of kinesin-3 members do not show any intramolecular interactions that include this residue (Fig. 3C). It is possible that KIF1A(V8) binds to parts of the KIF1A structure that are not yet solved. The structures of KIF1A paralogs strongly support the current model that the inactive form of KIF1A is a monomer and the active form of KIF1A a dimer. Nevertheless, some data are not fully consistent with a monomer-to-dimer transition. Whereas several KAND mutations strongly activate KIF1A, we found these mutations do not significantly change the size of purified KIF1A in size exclusion chromatography (Chiba et al., 2019). In addition, fluorescent resonance energy transfer (FRET) assays indicate that KIF1A is always dimeric in the cytosol (Hammond et al., 2009). KIF1C, another paralogue of KIF1A, was thought to be monomeric as well, but a recent study has shown that KIF1C is in fact a constitutive dimer (Siddiqui et al., 2019). The MD-NC-CC1-FHA sequence of KIF1C is almost identical to that of KIF1A, KIF13B and KLP-6. Therefore, the full-length structure of KIF1A, both in the active and inactive forms, is needed to fully understand how KIF1A is activated and inactivated during axonal transport. An important step in this direction are advances in cryo-electron microscopy that now enable the structures of full-length motor proteins to be solved (Zhang et al., 2017; Yang et al., 2020; Scarff et al., 2020).

The final issue to be addressed is whether advances in our understanding of KIF1A can be exploited for treatment strategies for KAND. As discussed here, KIF1A variants are a frequent cause of congenital neurological disorders (Boyle et al., 2021; Pennings et al., 2020), and KIF1A.org (https://www.kif1a.org/), a patient association for KAND has been established with the aim to collaborate with researchers to investigate the molecular mechanisms underlying KAND and to develop treatment methods. Although much has been learned concerning the basic biology of KIF1A and axonal transport from the study of KAND, there are still no effective treatment strategies for KAND. As discussed above, the ectopic expression of KIF1A(R18W) in rat brain causes defects in brain morphogenesis, which is a good KAND model, and these defects could be partially rescued by ectopic administration of BDNF, a cargo of KIF1A (Carabalona et al., 2016). Thus, BDNF administration is a candidate treatment strategy. To expand the repertoire of possible candidates, better screening methods are required. Some disease model worms have been established, which might facilitate forward genetic screening and drug screening approaches. For instance, using a forward-genetic screening assay, we identified an intragenic mutation in KIF1A that acts as a suppressor (Anazawa et al., 2022). The suppressor mutation restores the axonal transport in KAND model worms. Although intragenic mutations are difficult to target for drug development, identification of extragenic suppressors could contribute to the identification of targets for KAND treatment. If the extragenic suppressor mutations are loss-of-function mutations for given factors, inhibitors for those factors could be used to treat KAND. Another possible approach is the use of induced pluripotent stem cells (iPSCs) obtained from KAND patients (Xiaojing et al., 2020). KIF1A.ORG is trying to establish further iPSCs with KAND mutations and distribute them to researchers. For instance, neurons generated from these iPSCs could be used to analyze axonal transport and might provide further knowledge of KIF1A-dependent axonal transport and synaptogenesis in humans, with the hope to develop treatment strategies for KAND.

We are very grateful to KIF1A.ORG and Dylan Verden for their assistance with our research and for providing a schematic figure showing symptoms of KAND patients (Fig. 4). We also thank the scientists and clinicians who are participating in the KIF1A.ORG monthly meeting for helpful discussion. We also thank Jeremy Allen, PhD, from Edanz for editing a draft of this manuscript.