Characterizing human KIF1Bβ motor activity by single-molecule motility assays and Caenorhabditis elegans genetics

Neurons have long neuronal processes called axons (Luo, 2020). The morphology and function of an axon depend on the function of motor proteins that transport cargo organelles by moving in a processive and directional manner along microtubules (Hirokawa et al., 2009). The kinesin superfamily proteins (KIFs) and cytoplasmic dynein serve as molecular motors for anterograde and retrograde axonal transport, respectively (Holzbaur and Scherer, 2011). Each KIF binds to a specific organelle or protein complex and transports them toward the axon terminal (Hirokawa et al., 2009).

Synaptic vesicles containing neurotransmitters are essential for synaptic functions (Luo, 2020). Mature synaptic vesicles are generated at the synaptic terminal. However, the components of synaptic vesicles are primarily synthesized in the cell body and then transported to synapses along the axon (Chiba et al., 2023; Hirokawa et al., 2009). The vesicular organelle responsible for this axonal transport, known as a synaptic vesicle precursor, is carried by KIF1A and UNC-104 orthologs in all animals analyzed so far (Chiba et al., 2023). These motor proteins belong to the kinesin-3 family (Hirokawa et al., 2009). UNC-104, a Caenorhabditis elegans kinesin, is the founding member of the kinesin-3 family (Hall and Hedgecock, 1991; Otsuka et al., 1991). Mutations in the unc-104 gene induce mislocalization of synapses in C. elegans (Hall and Hedgecock, 1991; Otsuka et al., 1991). Mammals have two orthologs of the UNC-104 motor in the genome, known as KIF1A and KIF1Bβ (Niwa et al., 2008; Okada et al., 1995; Zhao et al., 2001). KIF1A has been identified as a molecular motor responsible for transporting synaptic vesicle precursors in mice (Okada et al., 1995). The genomic locus of KIF1B generates two isoforms of motor proteins in mammals: KIF1Bα and KIF1Bβ (Zhao et al., 2001). Previous studies have demonstrated that KIF1Bα is involved in the transport of mitochondria, whereas KIF1Bβ, structurally similar to KIF1A, also plays a role in transporting synaptic vesicle precursors in mammals (Nangaku et al., 1994; Niwa et al., 2008; Zhao et al., 2001). In addition to the axonal transport of synaptic vesicle precursors, studies in flies and worms suggest that KIF1A and UNC-104 orthologs transport active zone precursors and mature synaptic vesicles (Chiba et al., 2023; Pack-Chung et al., 2007). Furthermore, KIF1A and KIF1Bβ transport other neuronal cargo, such as dense-core vesicles and IGF1 receptor-containing vesicles, in the mammalian axon (Stucchi et al., 2018; Xu et al., 2018). KIF1Bβ, together with KIF5 proteins, is also implicated in lysosome transport in HeLa cells (Guardia et al., 2016). Consistent with this function in non-neuronal cells, KIF1Bβ exhibits ubiquitous expression, whereas KIF1A is dominantly expressed in neurons in mice (Okada et al., 1995; Zhao et al., 2001).

Owing to the importance of KIFs in axonal morphology and function, mutations in KIFs are often associated with human neuropathies (Baron et al., 2022; Boyle et al., 2021; Budaitis et al., 2021; Ebbing et al., 2008; Esmaeeli Nieh et al., 2015; Hirokawa et al., 2010; Holzbaur and Scherer, 2011; Klebe et al., 2012; Nakano et al., 2022; Pant et al., 2022; Zhao et al., 2001). Mutations in human KIF1A and human KIF1Bβ lead to neuronal diseases (Budaitis et al., 2021; Morikawa et al., 2022; Xu et al., 2018; Zhao et al., 2001). The KIF1A genomic locus tends to mutate at a relatively higher frequency than other loci, and genetic variants of KIF1A are associated with a spectrum of congenital genetic disorders called KIF1A-associated neurological disorder (KAND) (Boyle et al., 2021; Chiba et al., 2023). Mutations in KIF1Bβ, such as the Q98L mutation, are associated with a peripheral neuropathy known as Charcot–Marie–Tooth type 2A type 1(CMT2A1), probably caused by reduced axonal transport in affected neurons (Zhao et al., 2001). KIF1Bβ has also been identified as a candidate 1p36 tumor suppressor that regulates apoptosis in sympathetic neurons (Munirajan et al., 2008). One possible mechanism of tumor suppression is that KIF1Bβ regulates calcineurin activity and mitochondrial dynamics (Li et al., 2016).

Numerous studies have revealed the biochemical properties of both wild-type KIF1A and disease-associated KIF1A (Anazawa et al., 2022; Boyle et al., 2021; Chiba et al., 2019; Esmaeeli Nieh et al., 2015; Guardia et al., 2016; Kita et al., 2023; Lam et al., 2021). The phenotypes of loss of unc-104 function mutant worms can be rescued by the expression of human KIF1A (Chiba et al., 2019). Taking advantage of these systems, we have previously developed simple methods to determine the axonal transport capability of disease-associated KIF1A (Lam et al., 2021). However, to our knowledge only one study has analyzed the ATPase activity of KIF1Bβ (Zhao et al., 2001). Biophysical motile parameters of KIF1Bβ have not yet been determined. There is a need for the development of straightforward and rapid genetic methods to assess the transport activity of disease-associated KIF1Bβ. In this study, we have established genetic methods to determine whether a mutation in KIF1Bβ results in a loss of transport activity. Furthermore, we have measured the motile parameters of KIF1Bβ in vitro.

KIF1Bβ, along with KIF1A, is a mammalian ortholog of the UNC-104 motor and has a very similar domain architecture (Fig. 1A) (Hirokawa et al., 2009; Niwa et al., 2008). We have previously demonstrated that the expression of human KIF1A can rescue the body movement defects in unc-104(e1265), a widely used loss-of-function allele of unc-104 (Chiba et al., 2019; Hall and Hedgecock, 1991). In this article, we refer to this allele as unc-104(lf). Here, we conducted a similar experiment using human KIF1Bβ (Niwa et al., 2008). Human KIF1Bβ cDNA was fused with the unc-104 promoter (a neuron-specific promoter; Chiba et al., 2019), and expressed in unc-104(lf). We found that the expression of human KIF1Bβ in worm neurons could rescue the body movement defects observed in the unc-104(lf) allele (Fig. 1B-D). We have previously shown that KIF1A with disease-associated loss-of-function mutations cannot rescue unc-104(lf) (Chiba et al., 2019). To determine whether we can detect disease-associate defects in KIF1Bβ, we introduced a CMT2A1-associated Q98L mutation into KIF1Bβ cDNA and tested whether this mutant version of KIF1Bβ could rescue the unc-104(lf) mutant or not. The KIF1Bβ(Q98L) mutation is located in the P loop that constitutes the ATP-binding pocket. Although Q98 does not directly bind to ATP, the residue forms a hydrogen bond with N337 and stabilizes the structure of the motor domain (Fig. S1). We found that KIF1Bβ(Q98L) was unable to rescue the body movement defects (Fig. 1E,F), but the same amount of wild-type KIF1Bβ could rescue it (Fig. 1D,F). When we injected a tenfold higher amount of plasmid DNA, it could weakly rescue the body movement defects in the unc-104(lf) mutant (Fig. 1F). These results suggest that KIF1Bβ(Q98L) is a loss-of-function mutation.

We proceeded to analyze whether the expression of human KIF1Bβ could rescue the synaptic defects in unc-104(lf) worms. The DA9 neuron is a polarized neuron with distinct regions, including a dendrite, cell body and axon (Fig. 2A,B) (Klassen and Shen, 2007). En passant synapses are formed along the axon of the DA9 neuron, making it a valuable model for studying axonal transport and synaptogenesis (Balseiro-Gómez et al., 2022; Glomb et al., 2023; Higashida and Niwa, 2023; Klassen and Shen, 2007; Klassen et al., 2010; Niwa et al., 2016; Wu et al., 2013). Previous studies have shown that synaptic vesicles, visualized by GFP::RAB-3, are mislocalized in the dendrite and cell body of the DA9 neuron in unc-104(lf) worms, whereas synaptic vesicles are localized to the dorsal axon in the wild-type DA9 neuron (Niwa et al., 2016; Wu et al., 2013) (Fig. 2B,C). When human KIF1Bβ was expressed in unc-104(lf) mutant worms using the unc-104 promoter, the localization of synaptic vesicles along the axon in transgenic strains was restored and we could not distinguish them from the wild-type neuron (Fig. 2D). However, when human KIF1Bβ(Q98L) was expressed in unc-104(lf) mutants, synaptic vesicles remained mislocalized along the dendrite (Fig. 2E). Even in a highly expressing strain, KIF1Bβ(Q98L) only partially rescued the synaptic defects (Fig. 2F). We conducted statistical analyses by counting the number of GFP::RAB-3 puncta in the axon and the commissure and dendritic region (Fig. 2G,H). Expression of wild-type KIF1Bβ, but not KIF1Bβ(Q98L), almost completely restored the number of GFP::RAB-3 puncta to the wild-type level in the axon (Fig. 2G). When KIF1Bβ(Q98L) was strongly expressed, some GFP::RAB-3 puncta were observed along the dorsal axon (Fig. 2G). Similarly, in the commissure and dendrite expression of wild-type KIF1Bβ, but not KIF1Bβ(Q98L), reduced the number of mislocalized GFP::RAB-3 puncta (Fig. 2H). Even in the strongly expressed strain, KIF1Bβ(Q98L) did not significantly reduce dendritic mislocalization (Fig. 2H). The results from these assays suggest that human KIF1Bβ is capable of performing axonal transport in the worm neuron, and this assay is useful to detect reduced transport activity caused by disease-associated KIF1Bβ mutations.

Detailed physical parameters of KIF1Bβ in vitro have remained largely elusive. This is in contrast to KIF1A, for which a number of studies have revealed physical parameters (Boyle et al., 2021; Budaitis et al., 2021; Chiba et al., 2019; Kita et al., 2023; Lam et al., 2021; Okada and Hirokawa, 1999; Soppina et al., 2014; Tomishige et al., 2002). Therefore, we measured the motile parameters of KIF1Bβ using a total internal reflection fluorescence (TIRF) assay. We expressed and purified recombinant human KIF1Bβ fused with the superfolder GFP (sfGFP) fluorescent protein (KIF1Bβ-FL) using Sf9 insect cells and baculovirus (Fig. S2). Additionally, we analyzed a deletion mutant of KIF1Bβ fused with sfGFP that consisted of the motor domain, neck coiled-coil domain, and the forkhead-associated (FHA), coiled-coil 1 (CC1) and coiled-coil 2 (CC2) domains [KIF1Bβ(1-721)] (Fig. 3A; Figs S2, S3), as similar deletion mutants of human KIF1A and worm UNC-104 have been analyzed in previous studies (Hammond et al., 2009; Kita et al., 2024; Tomishige et al., 2002). Both KIF1Bβ-FL and KIF1Bβ(1-721) showed processive movement on microtubules (Fig. 3B,C). The velocities of KIF1Bβ-FL and KIF1Bβ(1-721) were 0.82±0.44 µm/s and 0.79±0.34 µm/s, respectively (mean±s.d.) (Fig. 3D; n=342 and 425 molecules; no statistically significant difference by unpaired two-tailed t-test, P=0.2546). The run lengths of KIF1Bβ-FL and KIF1Bβ(1-721) were 7.7±6.7 µm and 8.0±6.7 µm (mean±s.d.) (Fig. 3E; n=342 and 425 molecules; no significant difference by Mann–Whitney test, P=0.9097). These results indicate that the properties of the motor domain were not affected by deletion of the tail domain. In contrast, the landing rate showed a significant difference. The landing rates of KIF1Bβ-FL and KIF1Bβ(1-721) were 0.0016±0.0006 µm⁻¹ s⁻¹ at 2 nM and 0.0062±0.0023 µm⁻¹ s⁻¹ at 0.5 nM (mean±s.d.), respectively (Fig. 3F). The concentrations could not be matched because when one was adjusted to an ideal concentration for measurement the other either became saturated or fell below the limit concentration. Although the concentration of KIF1Bβ(1-721) was lower than KIF1Bβ-FL in the measurement, the landing rate of KIF1Bβ(1-721) was higher (n=31 and 34 microtubules; P<0.0001, Mann–Whitney test). This result suggests that KIF1Bβ(1-721) is capable of binding to microtubules much more frequently than KIF1Bβ-FL, indicating that the domain encoding KIF1Bβ(722-1770) inhibits the association of KIF1Bβ-FL with microtubules. This is consistent with the predicted KIF1Bβ structure obtained by AlphaFold2 (Fig. S3), as described in the Discussion.

It has been demonstrated that physical parameters affected by a disease-associated mutation correlate with human symptoms in the case of KIF1A (Boyle et al., 2021). In contrast, in the case of KIF1Bβ, the Q98L variant, which has been found in a Japanese family with CMT2A1, has only been studied by an ATPase assay (Zhao et al., 2001). Therefore, we conducted a comparison of the activity of wild-type KIF1Bβ(1-721) [KIF1Bβ(1-721)wt] and KIF1Bβ(1-721)(Q98L) using single-molecule assays. In contrast to KIF1Bβ(1-721)wt, which exhibits processive movement on microtubules, KIF1Bβ(1-721)(Q98L) displayed slower movement on microtubules (Fig. 4A-C). The velocities of KIF1Bβ(1-721)wt and KIF1Bβ(1-721)(Q98L) were 0.79±0.34 µm/s and 0.33±0.18 µm/s, respectively (Fig. 4C; mean±s.d.; n=425 and 208 molecules; P<0.0001 by unpaired two-tailed t-test). The run lengths of KIF1Bβ(1-721)wt and KIF1Bβ(1-721)(Q98L) were 8.0±6.9 µm and 6.2±4.4 µm (Fig. 4D; mean±s.d.; n=425 and 208 molecules; P=0.0465 by Mann–Whitney test). Moreover, the landing rates of KIF1Bβ(1-721)wt and KIF1Bβ(1-721)(Q98L) were 0.0062±0.0023 µm⁻¹ s⁻¹ at 2 nM and 0.0025±0.0008 µm⁻¹ sec⁻¹ at 2 nM (mean±s.d.), respectively (Fig. 4E; n=34 and 37 microtubules; P<0.0001 by Mann–Whitney test). Collectively, these data suggest that the KIF1Bβ(Q98L) mutation reduces a broad range of motile parameters in the human KIF1Bβ motor protein.

The residue KIF1Bβ(Q98L) is well conserved in worm UNC-104 and is equivalent to UNC-104(Q94L) (Fig. 5A). We next introduced this mutation into C. elegans using CRISPR/Cas9 (Arribere et al., 2014; Ghanta et al., 2021). Introduction of the mutation was confirmed by genomic PCR followed by restriction enzyme digestion and Sanger sequencing (Fig. S4). We then observed the macroscopic phenotypes of homozygous worms with the Q94L mutation (Fig. 5B,C). Unlike strong loss-of-function alleles of unc-104 and KIF1Bβ(Q98L)-rescued strains (Fig. 1C,E), unc-104(Q94L) mutant worms exhibited movement on feeder plates (Fig. 5B,C). Unlike KAND model unc-104 mutants (Anazawa et al., 2022), the body size of the mutant worms was comparable to that of wild-type worms (Fig. 5B,C). However, the movement of mutant worms was slower than that of wild-type worms. To analyze worm movement quantitively, we counted the number of body bends in a water droplet (Fig. 5D). The assay confirmed that body movements were 30% less frequent in unc-104(Q94L) mutant worms than in wild type (Fig. 5D). Next, we observed the synaptic vesicle marker RAB-3 in the DA9 neuron of unc-104(Q94L) worms (Fig. 5E-G). We counted the number of GFP::RAB-3 puncta in the commissure and dendrite (Fig. 5G). Because no accumulation was observed in the wild-type background, we consider the number to reflect the severity of the phenotype. Whereas accumulation of GFP::RAB-3 was not observed in the commissure and dendrite in wild-type worms, 23% of unc-104(Q94L) mutant worms exhibited mis-accumulation of GFP::RAB-3 in the commissure and dendrite (Fig. 5E-G). These data suggests that unc-104(Q94L) is a hypomorphic allele of unc-104.

We have previously also observed weak mis-accumulation of GFP::RAB-3 in the dendrite in gain-of-function alleles of unc-104 (Niwa et al., 2016). To exclude the possibility that unc-104(Q94L) might be a gain-of-function allele and to confirm that unc-104(Q94L) is a hypomorphic allele, we crossed unc-104(Q94L) worms and unc-104(lf) to generate transheterozygotes, unc-104(Q94L)/unc-104(lf) (Fig. 6). As shown in Figs 2B,C and 5F, the synaptic marker RAB-3 localized in the dorsal axon of the DA9 neuron of wild-type worms (Fig. 6A), mislocalized to the dendrite in unc-104(lf) worms (Fig. 6B), and partially mislocalized to the dendrite and commissure in unc-104(Q94L) worms (Fig. 6C). When wild-type or unc-104(Q94L) worms were crossed with unc-104(lf) worms, whereas wild type/unc-104(lf) heterozygotes [+/unc-104(lf) in the figure] did not exhibit GFP::RAB-3 mislocalization in the DA9 neuron, unc-104(Q94L)/unc-104(lf) transheterozygotes exhibited misaccumulation of GFP::RAB-3 in the dendrite and commissure (Fig. 6D-E′). To confirm the phenotype, we counted the number of GFP::RAB-3 puncta in the commissure and dendrite (Fig. 6F). We found that the mislocalization of GFP::RAB-3 was increased in unc-104(Q94L)/unc-104(lf) transheterozygotes compared with wild-type/unc-104(lf) heterozygotes and unc-104(Q94L) homozygotes. These data again suggest that unc-104(Q94L) is a hypomorphic allele of unc-104.

To investigate the mechanism underlying the phenotype of unc-104(Q94L) mutants, we analyzed axonal transport of synaptic vesicle precursors using GFP::RAB-3. We found that the velocity of anterograde axonal transport was slower in unc-104(Q94L) mutant worms compared with wild-type worms (Fig. 7A-C). The velocities of anterograde axonal transport were 1.64±0.40 and 0.93±0.29 µm/s (mean±s.d.; P<0.0001 by unpaired two-tailed t-test), respectively, in wild-type and unc-104(Q94L). The velocity of retrograde axonal transport was not significantly affected (Fig. 7D). We next calculated the frequency of anterograde and retrograde axonal transport per second by counting the number of vesicles passing a single point on the axon (Fig. 7E,F). We found that the frequency of anterograde transport (P=0.0102 in Mann–Whitney test), but not retrograde transport (P=0.4992 in Mann–Whitney test), was reduced by 5% in unc-104(Q94L), consistent with the weak mislocalization of synaptic vesicles in unc-104(Q94L).

The role of both KIF1A and KIF1Bβ as axonal motors responsible for the transport of synaptic vesicle precursors in mammals has been shown (Niwa et al., 2008; Okada et al., 1995; Zhao et al., 2001). Although KIF1A has been subjected to extensive analysis of its biochemical and biophysical properties, the examination of KIF1Bβ has been relatively limited. Although the ATPase activity of KIF1Bβ has been measured (Zhao et al., 2001), the biophysical properties of human KIF1Bβ have not been determined. In this study, we analyzed biophysical parameters of KIF1Bβ using single-molecule motility assays. The velocity, run length, and landing rate of KIF1Bβ are almost identical to those of KIF1A (Table S1) (Chiba et al., 2019), which is consistent with the redundant functions of these two motors. Similar to KIF1A, KIF1Bβ can transport synaptic materials in worm neurons and can rescue defects in unc-104(lf) mutant worms (Fig. 1A,B).

In our single-molecule assays conducted in the absence of cargo, we observed that KIF1Bβ(1-721) exhibited more frequent microtubule binding compared with KIF1Bβ-FL (Fig. 3). This property is similar to the behavior of KIF1A reported previously (Budaitis et al., 2021). Although the precise mechanism underlying this phenomenon needs further analysis, this is likely to be related to the regulatory mechanism known as autoinhibition, which is a common feature among motor proteins (Verhey and Hammond, 2009). Kinesin motors are generally folded in the absence of cargo molecules and kept in an inactive state. For instance, kinesin-1 is known to adopt a folded conformation in the absence of cargo (Tan et al., 2023; Weijman et al., 2022). A common property of kinesin-3 motors is that the CC1 and CC2 domains fold toward the motor domain and inhibit the activity of the motor domain (Ren et al., 2018; Wang et al., 2022). Although the structural details of full-length KIF1Bβ orthologs have not been revealed, a recent study has clarified the crystal structure of the full-length KLP-6 (Wang et al., 2022), a kinesin-3 family motor unique to worms. Their study revealed that, not only the CC1 and CC2 domains, but also more C-terminal tail domains of KLP-6 are folded toward the motor domain and inhibit the activity of KLP-6 (Wang et al., 2022). In light of this finding, it is likely that KIF1Bβ(722-1770), the C-terminal region of KIF1Bβ containing cargo-binding domains, may inhibit the activity of KIF1Bβ. Consistent with this idea, the AlphaFold2 prediction suggests that the region encoded by KIF1Bβ(722-1770) adopts a folded conformation toward KIF1Bβ(1-721) (Fig. S3), which is similar to observations in kinesin-1 (Tan et al., 2023; Weijman et al., 2022). According to the AlphaFold Protein Structure Database (Varadi et al., 2022), the architecture is conserved in other kinesin-3 family members, including KIF1A (UNC-104), KIF13A, KIF13B and KIF16B. To understand fully the mechanism of the autoinhibition of KIF1Bβ and its orthologs, further studies focusing on elucidating the high-resolution structure of the full-length proteins will be essential.

The ongoing revolution in whole-genome sequence technology is revealing a lot of disease-associated mutations in KIF1A and KIF1Bβ (Boyle et al., 2021; Xu et al., 2018). When previously uncharacterized single nucleotide polymorphisms are identified in human, it is essential to analyze these mutations. In the case of KIF1A mutations, we and others have established genetic and biochemical assays to test whether a mutation results in a loss of function (Anazawa et al., 2022; Boyle et al., 2021; Budaitis et al., 2021; Chiba et al., 2019; Lam et al., 2021). Using these assays, numerous disease-associated KIF1A mutations have been characterized. Comprehensive biochemical analysis showed that such disease-associated mutations in KIF1A affect various motile parameters of the KIF1A motor (Boyle et al., 2021). In contrast, methods to analyze KIF1Bβ mutations were yet to be developed. Although the expression of KIF1Bβ(Q98L) cannot strongly rescue unc-104(lf) worms, introduction of the analogous Q94L mutation into the endogenous unc-104 locus results in weak effects (Figs 1, 2, 5 and 7). The discrepancy between transgenic rescue and genome editing is unusual compared with previous results obtained from KIF1A analysis. One possibility is that KIF1Bβ(Q98L) mRNA or protein is not properly expressed in worms. Another possibility is that KIF1Bβ(Q98L) is unstable in the worm cytoplasm. Structural analysis predicts that the KIF1B(Q98L) [UNC-104(Q94L)] mutation disrupts a critical hydrogen bond, stabilizing the motor domain core, which implies reduced stability for these mutant proteins compared with their wild-type counterparts. Furthermore, because of differences such as optimal temperatures and the presence/absence of chaperones, mammalian proteins cannot fold as easily in the worm cytoplasm, compared with native worm proteins. Because of these factors, the KIF1Bβ(Q98L) mutation might have an accelerated effect in worm cytoplasm. Discriminating these possibilities requires further analysis, including single-copy insertion and accurate comparison of the expression level. Nevertheless, we propose that transgenic rescue in unc-104 mutants, as a convenient screening method to test a new pathological allele in KIF1Bβ, identifies a loss of function. To investigate the abnormalities caused by KIF1Bβ mutations in detail, it will be necessary to perform multidimensional analysis, including rescue experiments using DNA injection at multiple concentrations, and to employ genome editing.

Although it has been reported that the KIF1Bβ(Q98L) variant is associated with CMT2A1 in a Japanese family, other studies have shown that CMT2A is caused by mutations in MFN2, a protein essential for mitochondrial fusion (Züchner et al., 2004). The KIF1B locus is in close proximity to the MFN2 locus on the human chromosome 1 (Züchner et al., 2004). This makes it unclear whether or not the KIF1Bβ(Q98L) mutation really affects neuronal function or not (Züchner et al., 2004). Our in vivo and in vitro assays gave compelling evidence demonstrating that the KIF1Bβ(Q98L) mutation reduces motor activity and disrupts the axonal transport of synaptic vesicle precursors. Thus, it is reasonable to conclude that the KIF1Bβ(Q98L) mutation has an adverse impact on neuronal function in human neurons (Zhao et al., 2001). The velocity of KIF1Bβ(Q98L) in vitro as well as the velocity of axonal transport in unc-104(Q94L) worms are significantly slower. Although the velocity of axonal transport is significantly affected in unc-104(Q94L) mutant worms, these mutants do not exhibit the unc phenotype. In contrast, other disease-model unc-104 mutants, such as unc-104(R9Q), unc-104(R251Q) and unc-104(P298L), exhibit a strong unc phenotype (Anazawa et al., 2022). These worms with a strong unc phenotype commonly showed significantly reduced frequency of axonal transport, whereas this parameter was weakly affected in the unc-104(Q94L) mutants (Fig. 7E). This suggests that the frequency of axonal transport may be more fundamental to synaptogenesis, whereas the contribution of the velocity may be relatively limited. These worm phenotypes appear to correlate with symptoms in humans. KIF1B(Q94L) variants are associated with late-onset motor neuron defects in a Japanese family, and KIF1A(R11Q), KIF1A(R254Q) and KIF1A(P305L) variants lead to early-onset neuronal defects (Zhao et al., 2001; Boyle et al., 2021). The velocity of axonal transport may be important for the maintenance of synapses at the late stage. To determine comprehensively the fundamental axonal transport parameters for synaptogenesis and synaptic maintenance, further analysis of additional mutations with varying axonal transport parameters is required. Moreover, mathematical models that explain the relationship between axonal transport parameters and synaptic morphology and functions are needed. We and others have shown that the velocity of heterodimers composed of a wild-type motor and a slow motor is slower than that of wild-type dimers (Kaseda et al., 2002; Kita et al., 2024). This effect would cause CMT2A1 through autosomal dominant inheritance (Zhao et al., 2001). Interestingly, mutations in KIF1Bβ have been linked to neuroblastoma (Munirajan et al., 2008). However, whether these mutations affect the transport activity of KIF1Bβ has not been determined. The system we have developed here would be useful to elucidate the molecular mechanism underlying these phenomena and to advance our understanding of KIF1Bβ-related pathologies.

C. elegans strains were maintained as described previously (Brenner, 1974). N2 wild-type worms and OP50 feeder bacteria were obtained from the C. elegans genetic center (CGC) (Minneapolis, MN, USA). wyIs85 and unc-104(e1265); wyIs85 were described previously (Anazawa et al., 2022). Plasmids to transform worms and their Addgene IDs are described in Table S2. Human KIF1Bβ cDNA encoding isoform c of KIF1Bβ was obtained from the Kazusa DNA Institute (Chiba, Japan). We found mutations and insertions in Kazusa's cDNA clone. First, mutations were corrected by PCR-based mutagenesis (Stratagene). Second, insertions in the Kazusa's clone were deleted by PCR to obtain cDNA encoding KIF1Bβ isoform b. cDNA encoding KIF1Bβ isoform b was inserted to a Punc-104 encoding vector that has been previously described (Chiba et al., 2019). Transformation of C. elegans was performed by plasmid injection as described (Mello et al., 1991). For the rescue experiments, 1 ng/µl of Punc-104::KIF1Bß, 1 ng/µl of Punc-104::KIF1Bß(Q98L) or 10 ng/µl of Punc-104::KIF1Bß(Q98L) were injected (Fig. 1). Strains used in this study are described in Table S3. The swim test was performed as described previously (Pierce-Shimomura et al., 2008).

The unc-104(Q94L) mutation was introduced using the Alt-R^® CRISPR/Cas9 system (Integrated DNA Technologies). Alt-R^® CRISPR-Cas9 tracrRNA (1072533), crRNA for unc-104 and Alt-R^®S.p. Cas9 (1081058) were purchased from Integrated DNA Technologies. The target sequence was 5′-GCATACGGTCAAACAGGATC-3′ (Fig. S4). The repair template was 5′-AATAATTGACTTCTAGGTATAATGTCTGCATTTTTGCATACGGTCTAACCGGATCCGGAAAATCATATACAATGATGGGAAAAGCCAATG-3′.

Injection mix was prepared as described with a slight modification (Ghanta et al., 2021). We excluded the rol-6 marker because successful injections should produce mutants with unc phenotypes due to deletion mutations of unc-104 caused by repair failures. Three days after the injection, we selected a few plates that contained strong unc worms. Then, unc mutant worms as well as superficially wild-type worms were singled from these plates. Seven days later, plates that contained the unc-104(Q94L) allele were selected by genomic PCR and BamHI (NEB) digestion. Of these plates, 30% contained either unc-104(Q94L) heterozygotes or homozygotes. It was revealed that worms with a strong unc phenotype were deletion mutants of unc-104.

Statistical analyses were performed using GraphPad Prism version 9. Statistical methods are described in the figure legends. Graphs were prepared using GraphPad Prism version 9, exported in TIFF format, and aligned using Adobe Illustrator 2021.

Reagents were purchased from Nacalai Tesque (Kyoto, Japan), unless described. Plasmids to express recombinant KIF1Bβ and their Addgene IDs are described in Table S2. Sf9 cells (Thermo Fisher Scientific) were maintained in Sf900™ II SFM (Thermo Fisher Scientific) at 27°C. DH10Bac (Thermo Fisher Scientific) were used to generate bacmids. To prepare baculovirus, 2×10⁶ cells of Sf9 cells were transferred to each well of a 6-well plate (Falcon). After the cells attached to the bottom of the dishes, about ∼5 μg of bacmid were transfected using 5 μl of TransIT^®-Insect transfection reagent (Takara Bio Inc.). Five days after initial transfection, the culture media were collected and spun at 3000 g for 10 min to obtain the supernatant containing recombinant baculovirus (P1). To express recombinant proteins, 200 ml of Sf9 cells (2×10⁶ cells/ml) were infected with 100 µl of P1 virus and cultured for 65 h at 27°C. Cells were harvested and stocked at −80°C. Sf9 cells were resuspended in 30 ml of lysis buffer [50 mM HEPES-KOH, pH 7.5, 150 mM KCH₃COO, 2 mM MgSO₄, 1 mM EGTA, 10% glycerol (volume/volume)] along with 1 mM DTT, 1 mM PMSF, 0.1 mM ATP and 0.5% Triton X-100 (volume/volume). After incubating on ice for 10 min, lysates were cleared by ultracentrifugation (80,000 g, 20 min, 4°C). Lysate was loaded on Streptactin-XT resin (IBA Lifesciences) (bead volume: 2 ml). The resin was washed with 40 ml wash buffer (50 mM HEPES-KOH, pH 8.0, 450 mM KCH₃COO, 2 mM MgSO₄, 1 mM EGTA, 10% glycerol). Protein was eluted with 40 ml elution buffer (50 mM HEPES-KOH, pH 8.0, 150 mM KCH₃COO, 2 mM MgSO₄, 1 mM EGTA, 10% glycerol, 100 mM biotin). Eluted fractions were further separated using an NGC chromatography system (Bio-Rad) equipped with a Superdex 200 Increase 10/300 GL column (Cytiva). Peak fractions were collected and frozen.

TIRF assays were performed as described (Chiba et al., 2019; Kita et al., 2024). Glass chambers were prepared by acid washing as previously described (Chiba et al., 2022). Glass chambers were coated with PLL-PEG-biotin (SuSoS). Polymerized microtubules equivalent to 70 µg/ml of tubulin were flowed into streptavidin-adsorbed flow chambers and allowed to adhere for 5-10 min. Unbound microtubules were washed away using assay buffer [90 mM HEPES-KOH pH 7.4, 50 mM KCH₃COO, 2 mM Mg(CH₃COO)₂, 1 mM EGTA, 10% glycerol, 0.1 mg/ml biotin–BSA, 0.2 mg/ml kappa-casein, 0.5% Pluronic F127 (volume/volume), 2 mM ATP, and an oxygen scavenging system composed of PCA/PCD/Trolox]. Purified motor protein was diluted to the indicated concentrations in the assay buffer. Then, the solution was flowed into the glass chamber. An ECLIPSE Ti2-E microscope equipped with a CFI Apochromat TIRF 100XC Oil objective lens, an Andor iXion life 897 camera and a Ti2-LAPP illumination system (Nikon) was used to observe single-molecule motility. NIS-Elements AR software ver. 5.2 (Nikon) was used to control the system.

For quantification, kymographs were made using ImageJ software. To determine velocity and run length, unidirectional lines >3 pixels along the x-axis (equivalent to approximately 480 nm) were analyzed as processive runs and the other signals were not analyzed. For the landing rate determination, we counted the motor that bound to microtubules within the observation time window.

Local Colabfold was obtained from GitHub (https://github.com/YoshitakaMo/localcolabfold) and used as described (Mirdita et al., 2022). We employed the ‘—amber’ option for structural refinement and used templates from the Protein Data Bank (PDB) for the prediction process. We set the number of prediction cycles to ten, while all other parameters remained at their default settings. PDB files obtained from predictions were analyzed using UCSF Chimera (https://www.cgl.ucsf.edu/chimera/), and the results were exported as TIFF files (Fig. S3).

During the preparation of this manuscript, we used ChatGPT in order to check English grammar and improve English writing. After using this tool, we reviewed and edited the content as needed and take full responsibility for the content of the publication.

We thank the members of the Niwa lab (Tohoku University). Some worm strains and OP50 were obtained from the CGC. We would like to thank FRIScore (Tohoku University) for supporting worm transformation and imaging experiments.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261783.reviewer-comments.pdf