F-box protein FBXB-65 regulates anterograde transport of the kinesin-3 motor UNC-104 through a PTM near its cargo-binding PH domain

Axonal transport is essential for establishing and maintaining neuron structure and function (Maday et al., 2014). The molecular motors kinesin and dynein drive long-distance axonal transport of cargo synthesized in the neuronal cell body. Caenorhabditis elegans UNC-104 and its homolog KIF1A, kinesin-3 family members, are essential for pre-SV exit from the cell body in multiple model systems (Hall and Hedgecock, 1991; Kumar et al., 2010; Okada and Hirokawa, 1999; Pack-Chung et al., 2007). Mutations in C. elegans UNC-104 corresponding to those found in KIF1A that are associated with hereditary spastic paraplegia hyperactivate UNC-104 by increasing the rate at which the motor associates with microtubules, and hence depleting motors from the cell body and causing them to accumulate at neuronal distal ends (Chiba et al., 2019; Cong et al., 2021; Niwa et al., 2016). Kinesin-1 achieves its steady-state distribution in neurons based solely on attaching and detaching cargo and intermittent diffusion (the loose-bucket brigade model) (Blasius et al., 2013). Dysregulation of kinesin distribution is often associated with perturbed distribution of its cargo (Chiba et al., 2019; Cong et al., 2021). Altered cargo distribution in turn can lead to neuronal dysfunction. Thus, regulating UNC-104 and KIF1A distribution is essential for maintaining neuronal homeostasis and preventing the progression of neurological diseases.

Levels of UNC-104 at synapses are regulated by ubiquitylation (Kumar et al., 2010). Many studies have demonstrated that ubiquitin and ubiquitin-like modifiers regulate protein function independently from degradation (Dupré and Haguenauer-Tsapis, 2001; Goo et al., 2015; Govers et al., 1999; Lin et al., 2011; van Delft et al., 1997). Post-translational modifications (PTMs), such as phosphorylation, alter the ability of kinesin-1 and kinesin-2 to bind cargo (Guillaud et al., 2008; Liang et al., 2014; Matthies et al., 1993; Yoshimura et al., 2010), and alter the ability of kinesin-3 to move or bind its adapters (Gan et al., 2020; Hummel and Hoogenraad, 2021; Kevenaar et al., 2016). PTMs on motors or motor adapters might also regulate the ability of a kinesin to switch between inactive and active conformations (Cai et al., 2007; Hammond et al., 2009). The role of kinesin ubiquitylation and other PTMs remains poorly understood (Hong et al., 2018).

Both ubiquitin and ubiquitin-like PTMs are added to lysine residues on protein substrates (Mattiroli and Sixma, 2014). Ubiquitylation and its related ubiquitin-like modifications sequentially use E1, E2 and E3 enzymes to achieve substrate specificity (Nagy and Dikic, 2010). In C. elegans, uba-1 is the sole E1 ubiquitin-activating enzyme and thus regulates all ubiquitylation (Kulkarni and Smith, 2008). Attachment of ubiquitin or ubiquitin-like modifications can mutually depend on each other via: (1) using the same family of E3 ligases to attach different modifications (Chu and Yang, 2011; Oved et al., 2006; Schmidt and Dikic, 2005), (2) attaching an ubiquitin-like modification to an ubiquitin E3 ligase to control its activity (Liu and Xirodimas, 2010; Ohh et al., 2002; Ohki et al., 2009; Oved et al., 2006; Tatham et al., 2008; Uzunova et al., 2007), and (3) attaching an ubiquitin-like modification to the substrate, altering its ability to be ubiquitylated (Sun et al., 2007; Xie et al., 2007). Additionally, multiple E3s might target different sites on the same substrate (Haglund et al., 2003). In summary, the interplay between ubiquitin and ubiquitin-like modifiers can result in complex regulation of substrate modification.

We have previously shown that UNC-104 is degraded in a ubiquitin-dependent manner in C. elegans neurons (Kumar et al., 2010). In our current study, using a combination of in vivo experiments and analytical theory, we show that a putative E3 ligase, encoded by fbxb-65, regulates UNC-104 cargo binding and motor distribution in C. elegans neurons.

UNC-104 and its mammalian ortholog KIF1A are ubiquitylated (Kumar et al., 2010; Ordureau et al., 2020; Oshikawa et al., 2012; Sarraf et al., 2013; Vogl et al., 2020) and rapidly turned over (Cohen et al., 2013; Fornasiero et al., 2018; Huang et al., 2020; Mathieson et al., 2018). A temperature-sensitive strain of uba-1(it129ts) grown at a restrictive temperature causes UNC-104::GFP accumulation at synapses (Kumar et al., 2010). We screened for E3 ligases that might act downstream of UBA-1 to regulate UNC-104 synaptic levels, localization or distribution.

We carried out a touch receptor neuron (TRN)-specific RNAi screen targeting predicted neuronally enriched genes encoding known E3 ligase domains based on data from GExplore 1.8 (Hutter and Suh, 2016; Calixto et al., 2010) and posterior lateral microtubule neuron (PLM)-specific gene expression NCBI SRR databases (SRR3481679, SRR2054422 and SRR2054423). 90 genes in these different families of E3s were available in the Ahringer RNAi library (Kamath et al., 2003). Based on their orthologous domains, these proteins might act as either ubiquitin or other ubiquitin-like modifier E3 ligases. TRN-specific RNAi of uba-1 caused an increase of UNC-104::GFP at both the PLM distal end and at synapses (Fig. 1A; Fig. S1A,D,E). The phenotypes upon knockdown of the 90 E3s observed fell into the following classes: (1) 45 E3s did not cause any significant change in UNC-104::GFP distribution, (2) six E3s caused UNC-104::GFP accumulation at the distal end, (3) 28 E3s caused an increase in UNC-104::GFP at synapses, and (4) 11 E3s caused an increase in UNC-104::GFP levels at both synapses and the distal end (Table S4).

Only knockdown of four of the 90 E3s tested led to no significant change in UNC-104 RNA levels and thus these four E3s might regulate UNC-104 post-translationally (Table S5). Two of these, fbxb-65 and fbxa-81, increased UNC-104 distal tip accumulation. Bifluorescence complementation (BiFC) between UNC-104 and FBXB-65 revealed fluorescence in the cell bodies and neuronal processes of the nerve ring (Fig. 1B,C), in contrast to UNC-104 BiFC with itself, which shows the majority of signal at the distal ends of the neuronal processes (Fig. 1B,C) (Hsu et al., 2011). Owing to the above finding that FBXB-65 interacts with UNC-104, and its orthology to the F-box domain associated with multiple E3 ligases (Fig. S1F; Kipreos and Pagano, 2000), we focused on understanding the role of fbxb-65 in regulation of UNC-104.

We characterized the UNC-104::GFP distribution phenotype in both uba-1 and fbxb-65 RNAi knockdown animals. The TRN-specific knockdown of uba-1 increases TRN-expressed UNC-104::GFP levels in 90% of animals at the PLM distal end compared to that for empty vector RNAi (hereafter referred to as control) animals (Fig. 1A). fbxb-65 RNAi causes distal end accumulation of UNC-104::GFP in 62% of animals (Fig. 1A). In control animals, UNC-104::GFP shows a steep increase in intensity only in the last 5 μm of the neuronal process (Fig. S1A,D). In contrast, in uba-1 RNAi and fbxb-65 RNAi animals, UNC-104::GFP intensity increases in the last 50 μm of the PLM process, with the greatest accumulation at neuronal distal tips (Fig. S1A). Additionally, TRN-specific expression of FBXB-65::GFP shows diffuse expression throughout the neuron (Fig. S1B). Thus, FBXB-65 could potentially influence UNC-104 in the entire neuron controlling steady-state motor distribution.

UNC-104::GFP levels increased at the PLM synapse in 85% of uba-1 RNAi animals compared to in 42% of fbxb-65 RNAi animals (Fig. 1A; Fig. S1E, Table S14). Pan-neuronal knockdown of uba-1 caused a 3-fold increase of total UNC-104 levels, whereas a pan-neuronal knockdown of fbxb-65 did not alter UNC-104 levels (Fig. 1D,E, Table S6).

UBA-1 specifically catalyzes ubiquitin activation in C. elegans (Kulkarni and Smith, 2008). However, E3 ligases like FBXB-65 might enable modification by other ubiquitin-like modifiers as well (Kanei-Ishii et al., 2012). Using a pan anti-ubiquitin antibody on immunoprecipitated UNC-104, we show that uba-1 RNAi caused a significant reduction (58%), whereas fbxb-65 RNAi caused a non-significant reduction (34%), in ubiquitylated UNC-104 levels (Fig. 1F,G; Table S6). The larger reduction seen in uba-1 RNAi compared to fbxb-65 RNAi suggest multiple ubiquitylation pathways might modify UNC-104.

The above findings indicate that fbxb-65 knockdown causes only a subset of the uba-1 RNAi phenotypes, namely UNC-104 accumulation at the distal ends; hence, fbxb-65 might predominantly regulate UNC-104 along the neuronal process rather than at synapses.

Single motor movement properties can change the neuronal process distribution of the motor (Chiba et al., 2019; Cong et al., 2021). Therefore, we examined movement of UNC-104::GFP in the TRNs (Klopfenstein et al., 2002; Kumar et al., 2010; Wu et al., 2016; Zhou et al., 2001).

Using kymographs of our movies of RNAi-treated C. elegans, we measured the flux, velocity and net displacement of UNC-104::GFP puncta (Fig. 2A; Movie 1). The average net displacement (7 μm), the fraction of UNC-104::GFP trajectories with run length >7 μm and velocity (1 μm s⁻¹) were not significantly changed in fbxb-65 RNAi animals (Fig. 2B,C; Fig. S2B, Tables S7, S15). There was a 43% increase in UNC-104::GFP anterograde flux, and a 140% increase in UNC-104::GFP retrograde flux in fbxb-65 RNAi animals (Fig. 2D,E, Table S7). Additionally, the intensity of both anterogradely and retrogradely moving UNC-104::GFP puncta was 148% greater in fbxb-65 RNAi animals (Fig. 2F; Fig. S2C,D, Table S7, Movies 1, 2).

In summary, knocking down fbxb-65 increased the intensity of moving UNC-104::GFP trajectories in addition to causing more UNC-104::GFP particles to move. These moving UNC-104::GFP particles likely reflect motors bound to their cargo.

fbxb-65 RNAi led to an increase in flux of UNC-104. Net movement of motors along the axon can also be assessed by an injury/axotomy-based assay (Abay et al., 2017; Kumar et al., 2010; Li et al., 1999; Okada et al., 1995; Rao et al., 2008). uba-1 mutants show increased UNC-104::GFP intensity, representing increased number of motors, at the distal cut site compared to that in wild type animals (Kumar et al., 2010), suggesting an increase in retrograde movement of UNC-104 at longer time scales in uba-1 mutants. To assess whether this assay also revealed increased UNC-104 movement in fbxb-65 RNAi animals, we examined accumulation of UNC-104 in the first 120 s after injury as well as 1 h post-injury.

We cut the neuronal process ∼100 μm away from the cell body and tracked UNC-104::GFP accumulation at the cut sites proximal and distal to the cell body for the first 120 s post-axotomy (Fig. 3A,B). In fbxb-65 RNAi animals, the UNC-104::GFP proximal cut site accumulation was 1.5-fold faster and 40% higher compared to in the control (Fig. 3C–E; Table S8). The UNC-104::GFP intensity at the distal cut site in fbxb-65 RNAi animals remained largely unchanged, which contrasts with the increase at the proximal cut site (Fig. S2A). At 1 h post-injury, UNC-104::GFP intensity at the proximal cut site continued to remain high, with fbxb-65 RNAi animals showing a 75% higher intensity than controls (Fig. 3F,G); the distal cut site showed a 70% increase in intensity (Fig. 3H; Table S8). Thus, fbxb-65 knockdown increases both anterograde and retrograde movement of UNC-104.

At 1 h post-injury, uba-1 RNAi led to a greater accumulation of UNC-104::GFP, with an increase of 95% and 35% at the distal and proximal cut sites, respectively, compared to that in the control (Fig. 3F–H; Table S8) (Kumar et al., 2010). At 120 s post-injury, there was a 2-fold faster and 50% higher accumulation of UNC-104::GFP in the proximal cut site in uba-1 RNAi animals compared to that in the control (Fig. 3C–E; Table S8). Immediately post-injury, UNC-104::GFP intensity at the distal cut site reduced in uba-1 RNAi animals (Fig. S2A). Thus, uba-1 knockdown increases the anterograde movement of UNC-104 at shorter time scales but has a larger effect on the net retrograde movement at longer time scales.

Therefore, fbxb-65 regulates UNC-104 bidirectional flux within the neuronal process. uba-1 increases UNC-104 retrograde flux (Fig. 3F), possibly due to a decrease in UNC-104 degradation at synapses (Kumar et al., 2010). The increased accumulation of UNC-104 at the neuronal ends of the TRNs might arise from altered net movement of the motors in fbxb-65 knockdown animals.

The above expression of P_s(n) (Eqn 3) fits well with the probabilities obtained from the intensity distributions of moving UNC-104::GFP puncta for the control (Fig. 2G) and fbxb-65 RNAi (Fig. 2H) with γ=1 and μ_control>μ_fbxb-65 (Table S19). Given that μ_control>μ_fbxb-65, b₊ is smaller in the control and larger in fbxb-65 RNAi, suggesting that fbxb-65 knockdown can increase the ability of UNC-104 to bind cargo to facilitate movement. Furthermore, given that γ=1 fits most of our data well, it is expected that FBXB-65 regulated UNC-104 influences the cooperative recruitment rate b₊ of UNC-104 on the cargo surface and does not alter the detachment rate for UNC-104 (β).

UNC-104 contains 114 lysine residues that can bond ubiquitin or other ubiquitin-like modifiers. Given that fbxb-65 does not appear to regulate overall ubiquitylation levels of UNC-104, fbxb-65 could potentially alter a single monoubiquitin addition or indirectly affect attachment of other ubiquitin-like modifications. The addition of ubiquitin or ubiquitin-like modifications will change UNC-104 mobility only by ∼8 kDa, making it difficult to identify the change on a western blot given that UNC-104 is ∼180 kDa. Thus, we expressed multiple different UNC-104 fragments spanning the UNC-104 coding region to identify the region that demonstrates an fbxb-65-dependent increase in molecular mass.

We expressed four different fragments, encompassing the motor domain, the coiled-coil domain, which is important for UNC-104 dimerization, the stalk domain and the cargo-binding PH domain (Fig. 4A). The construct containing the PH domain fragment exhibits two distinct bands separated by ∼8 kDa (Fig. 4B), which is the size expected for monoubiquitin or monoubiquitin-like modifications (e.g. SUMO, NEDD8, URM1 and UFM1). Furthermore, the relative ratio of the higher molecular weight (HMW) to lower molecular weight (LMW) PH fragments greatly reduces upon pan-neural fbxb-65 RNAi (Fig. 4C,D), suggesting that fbxb-65 might regulate a PTM of UNC-104 close to its cargo-binding PH domain. We also assessed whether the HMW PH fragment is ubiquitylated using anti-ubiquitin antibodies. However, no signal was obtained at the HMW PH size, suggesting that this region is unlikely to be a ubiquitin-conjugated UNC-104 fragment (Fig. S3C). Given that the FBXB-65-dependent HMW band is not recognized by anti-ubiquitin antibodies and because fbxb-65 knockdown does not significantly change the UNC-104 ubiquitylated fraction (Fig. 1F,G), the modification might be ubiquitin-like and depend on other ubiquitin-like E1 enzymes. C. elegans has five E1 activating enzymes, one each for ubiquitin, SUMO, NEDD8, URM1 and UFM1, encoded by genes uba-1, uba-2, rfl-1, moc-3 and uba-5, respectively. Pan-neuronal knockdown of these E1 genes does not show a significant reduction in the modification of the PH-containing fragment, except for uba-1 RNAi, which shows a 20% reduction (Fig. S3A,B; Table S16). We could not confirm the nature of the modification on PH-containing fragment due to the lack of antibodies that recognize the C. elegans ubiquitin-like modifiers.

To find the site of this PTM near the PH domain, we created four sub-deletions of the PH domain-containing fragment (Δ1, Δ2, Δ3 and Δ4), all of which contain several lysine residues (Fig. 4E; Table S9). Upon deletion of the first 36 amino acids (aa) (Δ1) of the PH-containing fragment (Fig. S3D), the HMW band is not seen (Fig. 4E; Table S9). Thus, the PH-containing fragment is modified in the Δ1 region; however, with the limitation that the nature of the ∼8 kDa modification is as yet unidentified.

In summary, fbxb-65 regulates an unidentified PTM in the 1386–1421 aa region N-terminal to the cargo-binding PH domain of the UNC-104 motor, leading to the presence of a HMW band. FBXB-65 might either directly modify UNC-104 to attach a ubiquitin-like modification or regulate the modification of UNC-104 through other E3 ligases that attach other 8 kDa ubiquitin-like modifications.

UNC-104 is essential for transporting heterogeneous pre-SVs out of the neuronal cell body (Hall and Hedgecock, 1991). Given that we observed an increase in UNC-104 anterograde flux and particle intensity upon fbxb-65 knockdown, we examined whether knockdown of fbxb-65 led to increased anterograde transport of the pre-SV UNC-104 cargo RAB-3.

Compared to control animals, fbxb-65 RNAi animals showed an increased number of GFP::RAB-3 puncta at the PLM distal end (Fig. 5A). GFP::RAB-3-labeled cargo redistributed towards the end of the neuronal process in fbxb-65 RNAi animals (8 puncta per 40 μm) with fewer puncta in the proximal process (5 puncta per 40 μm) compared to in control animals (6 puncta per 40 μm and 7.5 puncta per 40 μm, respectively) (Fig. 5B,D; Table S10). In contrast to what was seen in the neuronal process, there was no significant change in GFP::RAB-3 intensity at the synapses between fbxb-65 and control RNAi animals (Fig. 5C; Table S10). Therefore, fbxb-65 RNAi leads to the accumulation of the UNC-104 cargo GFP::RAB-3 towards the distal ends without affecting its distribution at the synapses of the PLM neuron.

Previous studies in vitro suggest that higher numbers and increased activity of motors on cargo can be associated with longer single runs of cargo (Furuta et al., 2013; Vershinin et al., 2007). Thus, we examined movement properties of UNC-104-dependent GFP::RAB-3-tagged pre-SVs in the PLM (Fig. 5E). There was an increase in the average anterograde run length and the fraction of GFP::RAB-3 with run lengths >2 μm in fbxb-65 RNAi compared to the control RNAi (Fig. 5G; Fig. S4B, Tables S10, S17). GFP::RAB-3-tagged vesicles had a similar velocity (0.6 μm s⁻¹) and reduced flux (6 vesicles per 20 μm per 10 s) compared to that in control animals (Fig. 5F; Fig. S4A, Tables S10, S17). In contrast, the retrograde GFP::RAB-3 flux, displacement and velocity were largely unchanged by fbxb-65 RNAi (Fig. S4C–E).

In conclusion, fbxb-65 RNAi decreases the number of GFP::RAB-3 puncta near the cell body and increases it at the distal tip. GFP::RAB-3 pre-SVs show increased run-length in fbxb-65 RNAi animals, which is likely from increased numbers of motors. The increased number of moving UNC-104::GFP particles is probably associated with pre-SVs, influencing movement and distribution of both motor and cargo.

To verify the fbxb-65 RNAi phenotypes, we generated a strain containing a CRISPR-mediated deletion of fbxb-65, named fbxb-65(syb7320). The deletion removes the first 272 aa out of the total of 307 aa and covers the entire F-box domain present from aa 6–57. Thus, this allele likely acts as a null and is hereafter referred to as fbxb-65(0) (Fig. S5A). We also generated a CRISPR-mediated deletion of the 36 aa (Δ1) near the UNC-104 PH domain, which is required for FBXB-65-dependent modification, named unc-104(syb7293), hereafter referred to as unc-104(Δ1) (Fig. S5A). We assessed the effects of these mutants on UNC-104::GFP CRISPR knock-in animals to examine whether the endogenous motor was affected in a similar manner to the TRN-specific transgenically expressed UNC-104::GFP.

The fbxb-65(0) animals showed PLM distal end accumulation of CRISPR knock-in UNC-104::GFP, similar to what was seen with fbxb-65 RNAi (Fig. S5B). In addition, UNC-104::GFP knock-in particles accumulated at the distal ends of the anterior lateral microtubule (ALM) neuron and other neurons that terminated in the head in fbxb-65(0), which was rescued upon pan-neuronal overexpression of FBXB-65::VN173 (Fig. 6A). Similar to what was seen with fbxb-65 RNAi, there was an increase in the average number (1.2 per 20 μm per 10 s) and particle intensity [11 arbitrary units (AU)] of moving UNC-104::GFP particles in fbxb-65(0) compared to what was seen in control animals (0.9 per 20 μm per 10 s and 7 AU, respectively) (Fig. 6C,D; Table S11, Movies 3, 4). The run lengths of UNC-104::GFP trajectories were largely unaltered (12.5 μm) as seen in fbxb-65 RNAi (Fig. S5C, Table S18). Similar to what was seen with fbxb-65 RNAi, the intensity histogram of UNC-104::GFP trajectories in fbxb-65(0) animals are right shifted compared to their controls with μ_control>μ_fbxb-65 (Fig. S5D,E). We also observe a similar increase (30%) in the membrane-bound fraction of UNC-104 in fbxb-65(0) compared to wild type (Fig. S5G,H). Thus, CRISPR knock-in UNC-104::GFP particles with fbxb-65(0) phenocopies the intensity distribution and movement properties of TRN-specific UNC-104::GFP particles with fbxb-65 RNAi.

To study the genetic interaction between fbxb-65 and uba-1, we built a neuronal RNAi-sensitive strain with fbxb-65(0) to assess UNC-104::GFP distribution upon loss of both fbxb-65 and uba-1. As reported previously (Fig. 1A), we observed an increased UNC-104::GFP distal tip accumulation upon loss of either fbxb-65 or uba-1 (Fig. 6E). Upon dual loss of fbxb-65 and uba-1, the accumulation of UNC-104::GFP did not increase further at the distal end (Fig. 6E). Distal tip accumulation of UNC-104::GFP was no worse upon dual loss of uba-1 and fbxb-65 compared to the single mutant fbxb-65(0) suggesting that UNC-104::GFP accumulation is FBXB-65 dependent. In contrast, synaptic UNC-104::GFP accumulation significantly increased only upon knockdown of uba-1, and not upon loss of fbxb-65 (Fig. S5F). Given that UBA-1 is a known E1 ubiquitin-activating enzyme that acts upstream of E3 ligases, we infer that fbxb-65 acts downstream of uba-1 to regulate the accumulation of UNC-104::GFP at the distal end.

We further assessed whether fbxb-65(0) altered GFP::RAB-3 transport. The average run length of GFP::RAB-3 in fbxb-65(0) (8.6 μm) was increased compared to that in wild-type animals (6.4 μm) (Fig. 6G,I; Table S11). GFP::RAB-3 run lengths in both fbxb-65(0) and fbxb-65 RNAi were increased compared to their respective controls (Figs 5G, 6I; Tables S10, S11). The flux (2.2 per 20 μm per 10 s) and velocity (0.9 μm s⁻¹) of GFP::RAB-3 in fbxb-65(0) were similar to those in wild-type animals (Fig. S5J,K; Table S18).

We assessed whether deletion of the 36 aa FBXB-65-regulated region of UNC-104 also altered GFP::RAB-3 distribution and transport properties akin to fbxb-65(0). The allele unc-104(Δ1) did not alter the levels of UNC-104 compared to wild type, in contrast to what was observed with unc-104(e1265) (Fig. S5L). The number of GFP::RAB-3 puncta at the distal ends of the neuron (8 per 40 μm) and GFP::RAB-3 run length (7.5 μm) was increased in unc-104(Δ1) compared to those in wild-type animals (6 per 40 μm and 6.4 μm, respectively) (Fig. 6F,H,I; Table S11). The velocity and flux of GFP::RAB-3 were largely unchanged in unc-104(Δ1) compared to in wild-type animals, similar to what was seen with fbxb-65(0) animals (Fig. S5J,K; Table S18). Thus, both fbxb-65(0) and unc-104(Δ1) have similar effects on GFP::RAB-3-associated pre-SV movement, similar to fbxb-65 RNAi.

In summary, fbxb-65(0) phenocopies fbxb-65 RNAi and unc-104(Δ1) (Table S20). Together, these data suggest that loss of the UNC-104 PTM, owing to loss of fbxb-65, increases UNC-104 association with cargo, thereby increasing cargo run length and redistribution of both motors and cargo to the end of the neuron.

The distance cargo travels in the axon depends on net motor activity (Hayashi et al., 2019; Niwa et al., 2017). In partial loss of function unc-104(e1265tb120) mutants with lower levels of motors, pre-SV-associated proteins travel into the axon but do not reach the synapse (Kumar et al., 2010). In sam-4 mutants with both lower UNC-104 levels and accumulation of the motor in the cell body, pre-SV-associated proteins accumulate near the cell body and very little reaches the synapse (Zheng et al., 2014). UNC-104 overexpression can suppress the pre-SV transport defects of mutants of the UNC-104 regulator sam-4 (Zheng et al., 2014). We tested whether fbxb-65, which controls the number of UNC-104 on pre-SVs, phenocopies what is seen with UNC-104 overexpression and rescues both unc-104(e1265tb120) and sam-4(js415) phenotypes.

In unc-104(e1265tb120) animals, GFP::RAB-3 was present up to ∼60 μm into the PLM neuronal process unlike in wild-type animals, whereas GFP::RAB-3 was present along the entire ∼350 μm PLM neuronal process (Figs 5A, 7A). In the double mutant unc-104(e1265tb120); fbxb-65(0) animals, GFP::RAB-3 traveled further into the neuronal process and was present until ∼80 μm (Fig. 7A). Likewise, in sam-4(js415) animals treated with control RNAi, GFP::RAB-3 accumulated close to the cell body and only 40% animals showed GFP::RAB-3 at the branch point (200–300 μm from the cell body) or at synapses (Fig. 7C) (Zheng et al., 2014). Overexpression of UNC-104 in sam-4(js415) reduced GFP::RAB-3 accumulation near the cell body (Fig. 7D; Table S12) and led to an increased number of animals with GFP::RAB-3 at the branch point (80%) compared to sam-4(js415) (Fig. 7E; Table S12) (Zheng et al., 2014). In sam-4(js415) animals treated with fbxb-65 RNAi, we found a similar decrease in GFP::RAB-3 accumulation near the cell body and an increase in the number of animals with GFP::RAB-3 present at the branch point (87%) (Fig. 7D,E; Table S12). Likewise, in sam-4(js415); fbxb-65(0) double mutant animals, the accumulation of GFP::RAB-3 puncta in the proximal neuronal process was reduced compared to that seen in sam-4(js415) animals (Fig. 7F,G; Table S12). Thus, consistent with increased UNC-104 on the cargo surface, loss or reduction of fbxb-65 leads to presence of GFP::RAB-3 further into the neuronal process in the sam-4(js415) background.

We used unc-104(Δ1) as an allele of unc-104 that is independent of fbxb-65-mediated regulation and potentially encodes UNC-104 motors that associate in greater numbers with cargo to assess cargo movement. In the double mutant unc-104(Δ1) sam-4(js415), there were fewer GFP::RAB-3 puncta in the proximal process (Fig. 7F,G; Table S12) with an increase in synaptic GFP::RAB-3 (Fig. 7H; Fig. S5I, Tables S12, S18) compared to what was seen in sam-4(js415) animals. Thus, the motor encoded by unc-104(Δ1) might associate more with pre-SVs since they are incapable of being modified by FBXB-65. The form of UNC-104 that is independent of FBXB-65 regulation suppresses sam-4(js415) and leads to increased GFP::RAB-3 at synapses.

To summarize, increasing the number of motors capable of associating with pre-SVs both by removing/reducing FBXB-65 and making UNC-104 independent of FBXB-65-mediated regulation leads to improved pre-SV transport in sam-4(js415) and unc-104(e1265tb120) animals. The PTM of UNC-104 by FBXB-65 might functionally titrate the levels of motors on the pre-SV cargo surface and regulate pre-SV cargo distribution.

Levels and types of motors are critical in navigating complex microtubule arrangements (Bergman et al., 2018; Tymanskyj et al., 2022). The PLM TRN has a synaptic branch that drops from the main process (Fig. 8A). pre-SVs turning into the synaptic branch depends on numbers of UNC-104 motors. UNC-104 overexpression leads to an increased incidence of pre-SVs going straight to the non-synaptic neuronal terminus in the PLM neuron (Vasudevan et al., 2024). Given that the numbers of UNC-104 molecules on the pre-SV cargo surface are likely increased in fbxb-65(0) and unc-104(Δ1) animals, we tested whether they controlled the behavior of vesicles at PLM TRN branch points.

In the single mutants unc-104(Δ1) and fbxb-65(0) there was an increased number of SNG-1::mNeonGreen-marked pre-SVs going straight toward the asynaptic distal end, along with a concomitant decrease in vesicles turning toward the synapse compared to those in wild-type animals (Fig. 8B,D; Table S13). The double mutant unc-104(Δ1); fbxb-65(0) has a similar phenotype to both the single mutants (Fig. 8B,D; Table S13). Given that the double mutant phenotype is similar to that of both single mutants, we suggest that FBXB-65-regulated PTM of UNC-104 controls synaptic vesicle turning to the branch. The percentage of vesicles pausing at the branch point are similar across wild-type, fbxb-65(0), unc-104(Δ1) and the double mutant unc-104(Δ1); fbxb-65(0) animals (Fig. 8C). These phenotypes resemble those that have been shown for UNC-104 overexpression in the PLM (Vasudevan et al., 2024). Therefore, disruption of fbxb-65-mediated regulation of UNC-104 modification might lead to mistargeting of SVs across branch points away from their intended destination, the synapse.

Increased levels of UNC-104 via transgenic overexpression increases median lifespan of C. elegans from 18 days in wild type to 21 days (Li et al., 2016). Given that we observe increased UNC-104 transport away from the cell body upon loss of fbxb-65, and loss of fbxb-65 acts in a similar manner to UNC-104 overexpression to control localization of pre-SVs (Fig. 8B,D), we tested whether mutants with reduced PTM of UNC-104 also had altered lifespan.

We observed a similar increase to that shown previously (Li et al., 2016) in C. elegans median lifespan upon UNC-104 overexpression, from 19.5 days in wild-type to 22 days (Fig. 8E). Interestingly, the single mutants fbxb-65(0), unc-104(Δ1) and the double mutant fbxb-65(0); unc-104(Δ1) all showed a significantly increased median lifespan (20.5 days) compared to wild-type worms (Fig. 8E). Given that both mutants behave similarly, and the double mutant is similar to both single mutants, we suggest that the FBXB-65-dependent pathway regulates worm lifespan through UNC-104 (Fig. 8F,G). fbxb-65(0) acts in a similar manner to UNC-104::GFP overexpression in increasing lifespan. One potential mechanism is through improving cargo transport.

PTM of motors regulates motor activity (Liang et al., 2014), motor–cargo interactions (Guillaud et al., 2008; Matthies et al., 1993) and motor conformation (Espeut et al., 2008). Here, we show that an unidentified PTM near the UNC-104 cargo-binding PH domain is regulated by a putative E3 ligase FBXB-65. This modification reduces the number of motors on the cargo surface, thereby altering anterograde motor and cargo movement. The significance of FBXB-65-mediated regulation is revealed when cargo transport is compromised. Our findings suggest that a PTM regulates motor–cargo binding to change the number of motors on the cargo surface. Our theoretical model suggests that cooperative binding of UNC-104 might account for some of our observations. Our inability to identify the precise nature of the PTM is a limitation of our study.

An increased number of kinesins on cargo increases cargo run lengths in vitro, although not in vivo (Beeg et al., 2008; Shubeita et al., 2008; Wilson et al., 2021). In vivo increases in cargo run lengths and flux arise from PTMs, such as phosphorylation, of either the motor or its adapter (Padzik et al., 2016; Prowse et al., 2023). The UNC-104 motor binds to its cargo through phosphoinositides (Klopfenstein and Vale, 2004; Kumar et al., 2010; Rizalar et al., 2023), with ∼132 phosphoinositides present on each synaptic vesicle (Takamori et al., 2006). Dictyostelium UNC-104 motors cluster and display higher processivity with increasing levels of phosphatidylinositol 4,5-bisphosphate (PIP₂) (Klopfenstein et al., 2002). Developing a theoretical model to explain the distribution of UNC-104 cluster intensity, we show that UNC-104 can bind cooperatively, resulting in higher motor intensities than predicted by a model with non-cooperative motor binding. The FBXB-65-dependent PTM of UNC-104 regulates UNC-104 levels on the cargo surface by reducing the propensity of UNC-104 to bind to the cargo surface. KIF1A phosphorylation at the cargo binding domain regulates the ability of motor to bind cargo (Hummel and Hoogenraad, 2021). In C. elegans, modification of UNC-104 by a FBXB-65-dependent pathway might play an analogous role or provide an additional mode of regulation.

uba-1 acts upstream of fbxb-65 and in turn FBXB-65 acts to (1) directly add an 8 kDa modification (potentially SUMO, NEDD8, URM1 or UFM1) near the UNC-104 PH domain or (2) acts on a SUMO, NEDD8, UFM1 and/or URM1 E3 ligases, which in turn modifies the region near the UNC-104 PH domain (Fig. 8F). The unidentified FBXB-65-dependent 8 kDa PTM occurs close to the cargo-binding PH domain that might sterically hinder motor–cargo association. However, it is formally possible that this modification stabilizes the closed state of the motor, indirectly preventing cargo binding. Loss of this PTM might also lead to non-specific interaction of inactive UNC-104 with multiple cargo including synaptic vesicles, thereby accounting for increased bidirectional flux (Fig. 2D,E).

Movement of synaptic vesicles can be impeded due to the narrow and complex intracellular environments enroute to the synapse (Sabharwal and Koushika, 2019; Sood et al., 2018). Increased KIF1A promotes synaptogenesis (Kondo et al., 2012) and increased UNC-104 improves behavior in aged animals (Li et al., 2016). The number of motors on the cargo surface is likely to be crucial for how far the cargo travels along the axon, how cargo navigates obstacles and for the steady-state distribution of cargo (Chen et al., 2020). Some individuals with KIF1A-associated neurological disorders carry gain-of-function KIF1A mutations leading to increased motor velocity or landing rates on microtubules, which lead to aberrant accumulation of synaptic vesicle proteins at neuronal distal ends (Chiba et al., 2019; Gabrych et al., 2019). Similarly, we observed increased pre-SVs at distal ends in fbxb-65 animals (Fig. 5B,D). Increased attachment of motor to cargo might lead to phenotypes similar to those seen in motors defective in auto-inhibition. Such regulation might, in part, be mediated by FBXB-65, which interacts with UNC-104 throughout the neuronal process (Figs 1B,C; Fig. S1B). In situations where cargo transport is compromised, for instance upon reduced cargo binding in UNC-104(D1497N M1540I) (Kumar et al., 2010; Maeder et al., 2014), reduced cargo processivity in sam-4(js415) (Zheng et al., 2014), or in aging animals with reduced UNC-104 levels (Li et al., 2016), not having the PTM can improve cargo transport and bypass these phenotypes. However, unregulated UNC-104 pre-SV association upon reduced modification leads to mistargeting of pre-SV cargo away from synaptic regions (Fig. 8B,D). Intriguingly, while the retrograde UNC-104::GFP flux increased (Fig. 2E), the retrograde movement of pre-SVs was unaffected in fbxb-65 (Fig. S4C–E). This increased flux might arise from inactive UNC-104 associated with a variety of retrogradely moving cargo, a subset of which might associate with pre-SVs. Therefore, although numerous factors influence cargo run lengths and distribution in vivo, increased motors on the cargo surface might serve as a mechanism to overcome local obstacles, and regulation of motors on cargo might facilitate efficient transport of cargo to their intended destination.

Ubiquitin, initially discovered for its involvement in ATP-dependent protein degradation (Ciechanover et al., 1980; Hershko et al., 1980), plays a crucial role in regulating protein turnover in neurons (Speese et al., 2003). Previous studies have consistently observed an accelerated turnover of KIF1A, surpassing the average lifetime of other neuronal proteins and even other members of the kinesin family (Cohen et al., 2013; Fornasiero et al., 2018; Huang et al., 2020; Mathieson et al., 2018). A prior study demonstrates ubiquitin-dependent degradation of UNC-104 (Kumar et al., 2010). UBA-1 hence controls two independent pathways, synaptic UNC-104 degradation and UNC-104 PTM. The latter is FBXB-65 dependent, where FBXB-65 acts, either directly through UNC-104 association or indirectly through other ubiquitin-like E3s, to attach an unidentified 8 kDa PTM near the PH domain of UNC-104 (Fig. 8F,G). Notably, these modifications, mediated by E3 ligases (Chu and Yang, 2011; Oved et al., 2006; Schmidt and Dikic, 2005), might elicit distinct outcomes in different neuronal compartments, such as the synapse versus in the neuronal process, indicating potential cross-talk between various modification types. Although ubiquitylation can regulate synaptic UNC-104 levels, another as-yet-unidentified PTM influences the movement of UNC-104, and together, these modifications allow the neuron to finetune motor activity and levels for cargo transport.

In conclusion, we propose that the levels of UNC-104 on the pre-SV surface are controlled by a FBXB-65-dependent PTM near the UNC-104 PH domain. Although increased UNC-104 levels on pre-SVs might bypass transport defects, regulating UNC-104 levels is essential for correct targeting of pre-SVs to the synapse and preventing mis-accumulation of pre-SVs at neuronal ends. The lack of appropriate targeting with too much transport may be detrimental to neuronal health (Gabrych et al., 2019).

Worms were maintained on nematode growth medium (NGM) agar at 20°C, and spotted with the Escherichia coli strain OP50 (Brenner, 1974). RNAi experiments were conducted by transferring adults to an NGM plate, supplemented with 100 μg μl⁻¹ ampicillin and 1 mM IPTG (Biobasic Catalog IB0168), 1 day after seeding with dsRNA-expressing E. coli HT115 isolated from the Ahringer C. elegans RNAi feeding library or the control Empty Vector L4440 RNAi. We use cell-specific SID-1 (a dsRNA transporter) overexpression along with a sid-1 mutant as previously standardized for both TRN-specific and pan-neuronal RNAi to ensure most efficient RNAi in neurons (Calixto et al., 2010). For all imaging experiments on animals treated with RNAi, we used a TRN-specific RNAi-sensitive strain expressing SID-1 under the control of the mec-18 promoter, which only expresses within the six TRNs, also engineered to express the required markers. For all biochemistry experiments on animals treated with RNAi, we used a we used a pan neuronal RNAi-sensitive strain expressing SID-1 under the control of the unc-119 promoter, also engineered to express the required markers. Larval stage 4 (L4) hermaphrodite worms were used for all experiments unless stated otherwise. For worm lifespan analysis, gravid adults of required genotypes were placed on six different NGM plates until 30 eggs were spotted on each. These worms were transferred every 2 days after reaching the adult stage until the worms stopped moving over an entire day and stopped responding to touch. Transgenic worms were generated by microinjecting into wild-type N2 using an Eppendorf^® FemtoJet microinjection system. Injection cocktails were composed of 50 ng μl⁻¹ co-injection marker, the required concentration of reporter DNA, and made up to 200 ng μl⁻¹ with pBluescript vector. The strains PHX7320 fbxb-65(syb7320) and PHX7293 unc-104b(syb7293) were generated by SunyBiotech's CRISPR service. Other strains were created using crosses by standard approaches. All strains used are listed in Table S1.

L4 animals were taken and anesthetized with 5 mM tetramisole (Brenner, 1974) and laid on a 5% agar pad. Live imaging of UNC-104::GFP (TT2440) and GFP::RAB-3 (TT2775) RNAi strains was performed using an Olympus IX83 fitted with a Yokogawa CSU-X1 excited with a 488 nm solid-state laser at 15% of 1 mW laser power at objective using a 100×/1.4 NA DIC oil objective with a pixel size of ∼129 nm per pixel and imaged with a Hamamatsu EM-CCD camera.

Live imaging of the strain with endogenously tagged UNC-104::GFP was performed on an Olympus IX83 fitted with a Yokogawa CSU-W1 excited with a 473 nm solid-state laser at 15% of 3 mW laser power at objective using a 100×/1.4 NA DIC oil objective and imaged with a Prime BSI sCMOS camera configured with 2×2 binning leading to a pixel size of ∼130 nm per pixel.

To generate kymographs, the movies were taken with an exposure time of 300 ms leading to a frame rate of 3 frames per second for a total time of ∼3 min.

Static images of UNC-104::GFP and GFP::RAB-3 were taken on an epifluorescence IX73 scope illuminated with 100% output from a 120 W X-cite mercury-arc lamp and imaged using a Photometrics Evolve 512 EMCCD camera and a 100×/1.4 NA DIC oil objective with a pixel size of 158 nm per pixel and exposure of 150 ms. UNC-104 and FBXB-65 BiFC images were taken as part of a Z-stack with a step size of 400 nm on an Olympus IX83 fitted with a Yokogawa CSU-W1 excited with a 473 nm solid-state laser at 15% of 3 mW laser power at objective using a 100×/1.4 NA DIC oil objective and imaged with a Prime BSI sCMOS camera configured with 2×2 binning leading to a pixel size of ∼130 nm per pixel.

Transgenic lines were generated by following standard techniques (Stinchcomb et al., 1985) using an Olympus IX53 equipped with 20× and 40× air objectives, Narishige M-152 micromanipulator and an Eppendorf Femtojet II microinjector. The F2 progenies that inherited and stably expressed the extrachromosomal transgene with >70% transmission was used for biochemistry experiments. All strains generated are listed in Table S1.

All image panels used for representation and analysis of time lapse movies were generated using Fiji-ImageJ v1.52p (Schindelin et al., 2012). Experimental kymographs were built using KymoResliceWide v.0.5 plugin for ImageJ (https://github.com/ekatrukha/KymoResliceWide) using the average intensity measured across a polyline of width 3 in a region ∼100 μm distal to the PLM cell body. The kymographs were then manually annotated by researchers who were aware of the experimental conditions to include all events with a meaningful velocity (<5 μm s⁻¹ and >0.1 μm s⁻¹) manually to derive estimates of flux and run lengths. A cargo was counted as moving if it had been displaced by at least 3 pixels in successive time frames.

For estimating the intensity of a single UNC-104::GFP trajectory, the average intensity of a trajectory was subtracted from the average intensity of the same trajectory three pixels above (corresponding to 1 s prior) on a kymograph with time progression vertically downwards. Regions of the trajectory that had a pause were not used in the analysis.

The imaging was performed using an LSM 710 system with the TRN-specific RNAi strain TT2440 expressing UNC-104::GFP anesthetized in 5 mM tetramisole on a 5% agar with a 63×/1.4 NA DIC M27 oil objective at 3× zoom with a pixel size of 90 nm and a 488 nm Argon laser. Images were acquired every 500 ms in the ALM, or in the PLM at ∼70 μm or ∼280 μm away from the cell body. Ablation was undertaken using the Zen software and a femtosecond MaiTai Ti: Sapphire laser (Spectra Physics) mode-locked at 800 nm at 80% power (400 mW maximum at objective) with 20 iterations post acquiring five pre-bleach images. The imaging was undertaken for at least 200 frames post-bleaching.

Post-acquisition, movies were analyzed using FIJI software, by drawing a rectangular region of interest (ROI) of ∼1×2 μm to cover the proximal and distal cut sites. The intensity at each time point was exported along with the ROI of bleaching and the ablation time point. The data were then normalized to pre-ablation intensity values and plotted with an in-house script generated in Python (available upon request).

RNA was isolated from TT3185 worms using TRIzol reagent (Thermo Fisher Scientific, catalog no. 15596018). Briefly, a worm pellet was freeze-cracked and RNA was extracted using a standard protocol (Rio et al., 2010). The isolated RNA (1 μg) was converted into cDNA using Superscript IV RT polymerase (Thermo Fisher Scientific, catalog no. 18090010) according to the manufacturer's protocol using random hexamers. The resultant cDNA was diluted 1:20 and used for quantitative PCR (qPCR).

For qPCR, primer efficiency was first calculated by amplifying cDNA template with KAPA 2× SYBR Master mix (Roche catalog no. 07959362001) and quantifying the SYBR intensity using a Roche LightCycler 480 at three different concentrations of cDNA 10-fold apart (1:1, 1:19, 1:199 dilutions in nuclease free water). Only primers with an efficiency greater than 95% were subsequently used. For each qPCR, controls for wild-type N2 worms and actin were set as negative control and standard, respectively. Each reaction was performed in triplicates. The final fold change was determined using the ΔΔCt method. All primers are listed in Table S2.

UNC-104 was divided into four different fragments by performing PCR of the pSN8 construct (Niwa et al., 2016). Primers (Table S2) were generated to clone a 5× Myc tag in the C-terminal of the fragments using in-fusion PCR. The combined fragment with Myc tag was then ligated with the vector backbone containing a pan-neuronal rab-3 promoter pHW393 [Prab-3::GAL4-SK(DBD)::VP64::let-858 3'UTR; Addgene plasmid #85583 deposited by Paul Sternberg; Wang et al., 2017]. Further deletions in the PH domain-containing fragments were made using PCR followed by in vivo recombination in the E. coli strain DH5α. All reagents are available upon request. All plasmids along with their construction details are listed in Table S3.

Worms were harvested from three almost-starved 60 mm NGM dishes by washing three times in M9 buffer (Brenner, 1974). Protein preparation from a mixed population of worms was adapted from a previous study (Shaham, 2006). Immunoprecipitation (IP) was undertaken using Myc trap agarose beads sourced from Chromotek (yta) in homogenization buffer [50 mM HEPES-KOH, pH 7.6, 1 mM EDTA,140 mM KCl, 0.5% NP-40, 10% glycerol and 1× Protease Inhibitor Cocktail (Sigma 11836170001)]. Post IP, the beads were rinsed three times with homogenization buffer and boiled with Laemmli buffer. An equal amount of total protein (5 μg) in this lysate, assessed using a Pierce™ BCA Protein Assay Kit, was subjected to denaturing PAGE on either an 8% (for whole UNC-104) or 12% (for fragments of UNC-104) gel and then transferred onto a supported nitrocellulose membrane. UNC-104 ubiquitin levels were assessed on blots with a normalized loading of UNC-104 protein amount assessed using the same lysate run on a previous blot probed for total UNC-104 intensity. The blots were probed with mouse anti-myc (AE010) antibody (ABclonal, lot no. 3500014031; dilution used 1:2500) to visualize the entire UNC-104 or UNC-104 fragments, mouse anti-ubiquitin (FK2) antibody (Sigma-Aldrich, lot no. 3845622; dilution used 1:1000) to visualize ubiquitin, and rabbit anti-actin (AC026) antibody (ABclonal, lot no. 9100026001; dilution used 1:5000) to visualize β-actin (Fig. S6).

Statistical analyses were performed using GraphPad Prism 9 or the SciPy stats module of Python 3.8 (Virtanen et al., 2020). The Shapiro–Wilk test was used to examine the normality of various distributions. Sample size estimates were calculated using G*Power 3.1.9.7 (Faul et al., 2009). Single comparisons between normally distributed data were performed using unpaired two-tailed Student's t-test and those between non-normal distributions were performed using the Mann–Whitney U-test. For multiple comparisons, one-way ANOVA with Dunnett's multiple comparisons test was used for normal data, and Mann–Whitney–Wilcoxon test with Bonferroni multiple comparisons correction was used for non-normal data. All data are represented as violin plots and have individual data points marked with filled circles along with the median (dashed line), 25th and 75th percentile marked (dotted lines). Table S21 contains all values used to generate all the graphs in the manuscript.

We thank Mei Ding for the strain xdKi3, Kavya S. Pillai for generating and characterizing the RNAi sensitive strain sid-1(pk3321) him-5(e1490); uIs71; jsIs1111; lin-15B(n744), Amal Mathew for generating and characterizing the strain sid-1(pk3321) him-5(e1490); uIs71; jsIs821; lin-15B(n744) and creating the rab-3p::UNC-104 construct. Debasish Chaudhuri thanks ICTS-TIFR, Bangalore, for an Associateship. This research was supported in part by the International Centre for Theoretical Sciences (ICTS) for the program ‘Statistical Biological Physics: from Single Molecule to Cell’ (code: ICTS/SBP2022/10). We thank Amruta Vasudevan for thoughtful discussions. Some strains were provided by the Caenorhabditis Genetics Center (CGC), which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440). The theoretical model work in this paper was performed by A.S., A.N. and D.C.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261553.reviewer-comments.pdf