A stop or go switch: glycogen synthase kinase 3β phosphorylation of the kinesin 1 motor domain at Ser314 halts motility without detaching from microtubules

In vitro, motor proteins are regulated by autoinhibition mediated by conformational changes of the head-tail interaction, which is released when motors bind cargo, activating motility. However, in vivo, motors are frequently bound to cargo and many regulatory factors are thought to play key roles during motility. Yet the molecular mechanisms of motor regulation are unknown and several key questions remain: How does the motor know when to bind microtubules (MTs) and/or specific cargo/organelles? How are motors turned on/off and how is continuous motility achieved and sustained?

Sequence analysis demonstrate that kinesin 1 subunits, kinesin heavy chain (KHC, motor domain) and kinesin light chain (KLC, cargo binding) have putative phosphorylation sites for several kinases: GSK3β, 5′ AMP-activated protein kinase (AMPK), casein kinase 2 (CK2) and c-Jun N-terminal kinase (JNK) (Hollenbeck, 1993; Schafer et al., 2009; Morfini et al., 2009a). However, the functional significance of the need for multiple kinases is unclear. Structural and biochemical studies indicate that purified kinesin 1 motors can adopt a folded/auto-inhibited conformation where KLC and/or the C-terminal tail of KHC interacts with the N-terminal motor domain causing motor inactivation (Muretta et al., 2018; Lee and Hollenbeck, 1995; Zhao et al., 2020). While JNK phosphorylation at S175 was shown to stabilize the auto-inhibited state of mammalian KHC (DeBerg et al., 2013), auto-inhibition disrupted MT binding and ATPase activity of kinesin 1 (Verhey et al., 1998; Verhey and Hammond, 2009). In contrast, phosphorylation of the inhibitory C-terminal tail by cyclin-dependent kinase 1 (Cdk1) relieved auto-inhibition of two mitotic kinesins Eg5 and CENPE, increasing MT binding and motility (Cahu et al., 2008; Espeut et al., 2008). In squid axoplasm, phosphorylation of KLC at an unknown site stimulated kinesin 1 release from cargoes/membranes (Morfini et al., 2002). However, in another system, phosphorylation of kinesin-associated proteins regulated kinesin 1 activity, but not membrane binding (McIlvain et al., 1994). In mammalian cells, phosphorylation of KHC by an unidentified kinase initiated MT binding and motility (Lee and Hollenbeck, 1995), but in cultured neurons, JNK3 phosphorylation of KHC decreased kinesin binding to MTs (DeBerg et al., 2013; Morfini et al., 2009b). Therefore, although it is clear that kinases and phosphorylation play key roles in motor regulation, discrepancies in published observations obscure our current understanding of the functional significance of how specific kinases control kinesin 1 movement.

Previously we proposed that GSK3β likely plays a key regulatory role during axonal transport in vivo (Banerjee et al., 2018; Dolma et al., 2014; Iacobucci and Gunawardena, 2018). Although excess active GSK3β perturbed axonal movement by increasing kinesin 1 binding to membranes, non-functional GSK3β had no effect and GSK3β loss/reduction decreased binding (Dolma et al., 2014). Furthermore, while active GSK3β associated with and phosphorylated KHC in vitro (Banerjee et al., 2018), the particular GSK3β phosphorylation site(s) are unknown.

KHC has a highly conserved globular head (N-terminus), a coiled-coil stalk and a C-terminal tail that associates with KLC and cargoes, including mitochondria (Fig. 1A) (Hirokawa, 1998; Ma and Taylor, 1995). The N-terminal region (aa1-410) is sufficient for MT-based motility (Stewart et al., 1993; Bloom et al., 1988), with aa1-339 functioning as a MT-activated ATPase (Stewart et al., 1993; Moyer et al., 1996, 1998). As biochemical analysis showed that endogenous fly KHC is phosphorylated and phosphorylation is eliminated by GSK3β loss/reduction (Fig. S1A), we used the Human Protein Reference Database-PhosphoMotif Finder (Amanchy et al., 2007) to isolate putative GSK3β-phosphorylation sites on Drosophila KHC. Three putative GSK3β-phosphorylation sites (serines 209, 242 and 314, Fig. 1A) which were 100% conserved with mammalian KHC (Fig. 1C) were identified. S209 is on α-helix 3a (α3a). S242 is near loop 11, which connects the MT- and ATP-binding domains (Fuger et al., 2012; Vale et al., 1996). S314 is on α6, which is near the neck linker that links motor heads (Kozielski et al., 1997; Hariharan and Hancock, 2009) (Fig. 1B,C).

To preclude phosphorylation, we mutated serines (S) 209, 242 or 314 to alanine (A) (Fig. 1C) and purified recombinant KHC proteins were assayed for GSK3β-mediated phosphorylation. KHCWT (KHC339/KHC410) and mutants KHCS209A, KHCS242A or KHCS314A (Fig. 1D) were incubated with active-GSK3β in the presence of γ-³²P-ATP. A γ-³²P positive band at ∼60 kDa corresponding to the size of KHC motor domain was seen in KHCWT (KHC339/KHC410), similar to endogenous fly KHC (Banerjee et al., 2018), indicating that the N-terminus of KHC is phosphorylated by GSK3β (Fig. 1E). KHCS209A and KHCS242A also showed γ-³²P-positive bands, which disappeared with GSK3β inhibitor CT990221 (Fig. 1E). Strikingly, γ-³²P incorporation was completely abolished in KHCS314A, demonstrating that S314 is likely phosphorylated by GSK3β (Fig. 1E). Furthermore, only the double mutant KHCS209A/S242A showed a γ-³²P positive band, whereas no bands were seen in KHCS209A/S314A, KHCS242A/S314A or the triple mutant KHCS209A/S242A/S314A (Fig. 1E). These results identify that GSK3β phosphorylates the KHC motor domain at S314.

In control MT-binding (Fig. 2A) experiments, MAPF (MT-associated protein fragment, positive control) was present in the pellet (P) fraction with tubulin, while BSA (bovine serum albumin, negative control) was found in the soluble (S) fraction (Fig. 2B). Endogenous KHC (khc^WT) isolated from wild-type flies bound to MT and was seen exclusively in the pellet (Fig. 2C). KHCS314A was also in the pellet, demonstrating that disrupting GSK3β phosphorylation does not affect MT binding (Fig. 2D). Although MT binding was unaffected in KHCS209A or KHCS242A (Fig. 2D), quantification of the ratio of band intensities in pellet to soluble did not significantly change (Table S1). These results are consistent with those from khc¹⁷ null mutant flies (S246F in the motor domain), in which MT binding is unaffected but the velocities of MTs are decreased (Brendza et al., 1999). Similarly, the N256S mutation or the JNK3 non-phosphorylatable mutation S175A on mammalian KHC did not influence MT binding, but decreased velocities (Ebbing et al., 2008). We found that although KHCS209A/S242A bound to MTs, the extent of MT binding in KHCS209A/S314A, KHCS242A/S314A and KHCS209A/S242A/S314A was reduced (Fig. S1B, Table S1). As site-directed mutagenesis could induce structural changes in the short 339/410 N-terminal fragment, we generated double (KINS209A/S314A and KINS242A/S314A) and triple (KINS209A/S242A/S314A) mutants in full-length KHC (975aa, Fig. S1B) (Yang et al., 1990), and found that all full-length mutant proteins bound MTs (Fig. S1CD, Table S1). Further MT binding was not disrupted in the GSK3β phospho-mimicking-KHCS314D mutant, where the serine (S) was replaced with aspartate (D) (Fig. 2D, Table S1). Therefore, disrupting S314 or GSK3β phosphorylation at S314 does not affect kinesin 1 attachment to MTs.

To test the prediction that GSK3β phosphorylation turns motors on/off, we evaluated motility dynamics of KHCS314A and KHCS314D using the in vitro MT gliding assay. Robust MT gliding at 0.68μm/s was observed for KHCWT, whereas KHCS314A showed a significant decrease in MT motility with an average velocity of 0.3 μm/s (P<0.05, Fig. 2EF, Movies 1, 2). The MT velocities observed for KHCWT were within previously reported ranges of 0.4-0.7 µm/s (Stewart et al., 1993; Woehlke et al., 1997). MT gliding velocities for KHCS209A and KHCS242A were indistinguishable from KHCWT (Fig. S2AB). The decreased velocities seen for KHCS314A correlated with decreased run lengths (P<0.005, Fig. 2G), indicating that kinesin 1 processivity was likely affected. The total distance traveled for KHCS314A was also significantly reduced compared with KHCWT (P<0.0005, Fig. 2H). Intriguingly, KHCS314D showed little motility, with a significant reduction in the average MT gliding velocities (P<0.005, Fig. 2E,F, Movie 3). A significant decrease in run length and in the total distance traveled was also observed (P<0.0005) (Fig. 2G,H). Our results are strikingly different from previous observations showing that JNK3 phosphorylation at S175 did not affect run length or processivity (Cahu et al., 2008). Therefore, although constitutive GSK3β phosphorylation at S314 likely acts as a STOP for kinesin 1 motility without dissociating from MTs, disrupting GSK3β phosphorylation at S314 affects kinesin 1 processivity.

The uncoordinated and decreased motility observed for KHCS314A could result from alterations in the rate at which the motor hydrolyzes ATP during motility. Indeed, KHCS314A showed a significant decrease in ATPase activity of 54% compared with KHCWT, while the GSK3β-phospho-mimetic-KHCS314D showed a much severe decrease of ∼37% (Fig. 2I). Although our results are consistent with those observed for kinesin null mutants, which showed decreased motility by lowering ATP turnover rate, without affecting MT binding (Brendza et al., 1999), our data are strikingly different from those observed for the phospho-defective and phospho-mimicking mutations of JNK3 at S175 (Cahu et al., 2008). Neither of these conditions altered the chemical kinetic cycle of ATP activity (Cahu et al., 2008). Therefore, disrupting GSK3β phosphorylation at S314 likely alters kinesin 1 processivity by reducing the ATP turnover rate, and decreasing run length and velocity (Fig. 2J,K).

As there is a considerable amount of discrepancy between what is observed for motor behaviors in vitro with those detailed in cells/neurons, to further isolate how GSK3β phosphorylation at S314 affects kinesin 1 processivity in vivo, we generated a fly containing S314A in endogenous fly KHC using CRISPR/Cas9, because S314D stopped motility. khc^S314A−/− homozygous larvae pupate but fail to eclose into adults, similar to other khc null mutants (Hurd and Saxton, 1996). These larvae show severe behavioral defects, including tail flipping and posterior paralytic phenotypes similar to khc null mutants (Fig. 3A, Movie 4) (Hurd and Saxton, 1996). Larval crawling velocities were significantly decreased compared with wild type (khc^WT) (P<0.005, Fig. 3D). Heterozygous khc^S314A−/+ larvae also showed significant decreases in crawling velocities (P<0.05, Fig. 3D). Segmental nerves from khc^S314A−/− larvae stained with the synaptic vesicle marker cysteine string protein (CSP) showed significant amounts of axonal blockages compared with khc^WT (P<0.0005, Fig. 3B,C). The extent of axonal blockages observed was similar to those seen in khc null mutants (Hurd and Saxton, 1996; Gunawardena and Goldstein, 2001), in the GSK3β null mutant (sgg^M11−/−, Weaver et al., 2013), in wild-type larvae fed with the GSK3β inhibitor (Fig. S4C) and in larvae expressing active GSK3β (Dolma et al., 2014). khc^S314A−/+ larvae also showed significant amounts of blockages (P<0.05, Fig. 3BC) demonstrating the severity of this mutation. The ratio of the average number of axonal blocks plotted against the ratio of the average larval crawling velocities indicates that the extent of blockages correlates with crawling defects (Fig. 3E). Additionally, synaptic morphological defects were seen in khc^S314A larval NMJs from segments A2 muscle 6/7, with significantly increased synaptic lengths compared with wild type (P<0.005, Fig. S3AB). khc null mutants also showed similar synaptic defects (Kang et al., 2014). A significant increase in the post-synaptic DLG maker was seen in khc^S314A NMJs (P<0.05, Fig. S3D), whereas the pre-synaptic HRP intensity was unaffected (Fig. S3E). Increased DLG staining likely indicates the presence of post-synaptic footprints due to nerve retraction (Fig. S3A) (Eaton et al., 2002), similar to what was previously observed in the N256S mutant (Amanchy et al., 2007). Therefore, the CRISPR/Cas9 khc^S314A fly mutant is phenotypically comparable with other loss-of-function khc mutants.

As mitochondria directly associate with KHC for their movement within axons, independent of KLC (Hurd and Saxton, 1996), we next evaluated mitochondrial motility in vivo in khc^S314A larvae. Larvae expressing mito-GFP with khc^S314A showed increased populations of stalled mitochondria compared with wild type (P<0.0005), which correlated with decreased populations of anterograde, retrograde and reversing mitochondria (Fig. 3G), similar to khc⁸. Similar to khc⁸, khc^S314A larvae also showed reductions in anterograde mitochondrial flux (P<0.005, Fig. 3H), indicating that few mitochondria enter the axon, perhaps due to reductions in the number of available active kinesin 1 motors. Mitochondrial switch frequencies were also significantly increased in khc^S314A compared with khc^WT (Fig. 3H), perhaps indicating the uncoupling of kinesin 1 and dynein motor activities. Although a significant number of mitochondria were stalled in khc^S314A, a small population showed anterograde motility. The overall anterograde mitochondrial velocities were significantly decreased (P<0.005, Fig. 3J), similar to other khc mutants (Pilling et al., 2006). However, in contrast to khc⁸ and other khc null mutants (Pilling et al., 2006), decreases in anterograde velocities did not correlate with decreases in run lengths (Fig. S4A). This observation differs from what was seen in in vitro MT gliding assays for recombinant KHC mutants (Fig. 2G) and for khc^S314A isolated from flies (Fig. S5). Although it is clear that motor behaviors in vitro do not recapitulate what is observed in vivo, the decreases seen in khc^S314A segmental velocities are likely due to increases in switch frequencies and decreases in anterograde pause frequencies (P<0.05, Fig. 3IK). Therefore, under physiological conditions, perhaps mitochondria bound to non-phosphorylatable khc^S314A motors are still associated and engaged on MTs, but have lost their ability to be controlled.

At least two mechanisms could explain how control of kinesin 1 motility is disrupted in vivo in khc^S314A larvae. Non-phosphorylatable khc^S314A could: (1) alter the competition between detachment/re-association of one motor head on MT during a step and/or (2) disrupt the rate of dissociation of the complex while only one head is bound. Although these possibilities may not be mutually exclusive, both can affect the overall processivity of kinesin 1. To test these predictions, we immunoprecipitated endogenous khc^S314A from CRISPR/Cas9 khc^S314A−/+ flies and tested MT binding and ATP hydrolysis. Isolated khc^S314A showed significantly decreased ATPase activity (P<0.005, Fig. 4D) and was present in the pellet with tubulin, but a substantial amount of khc^S314A was also observed in the soluble (Fig. 4A, Table S1). This observation is in contrast to recombinant KHCS314A in which MT binding was unaffected (Fig. 2D, Table S1). Several possibilities could explain these observations. Perhaps non-phosphorylatable khc^S314A (1) does not bind MTs strongly, (2) causes only one kinesin 1 head to detach from MTs, (3) produces unstable kinesin 1, which can fluctuate between the folded/inactive and the unfolded/active forms and/or (4) disrupts the coordinated associations with KLC and other accessory proteins that mediate membrane/organelle binding in vivo.

Intriguingly a high molecular weight (HMW) form of fly khc^S314A was seen exclusively bound to MTs (Fig. 4A), which likely do not represent the folded/inactive motors that do not bind MTs (McIlvain et al., 1994). The HMW, MT bound form of khc^S314A could represent open but ‘locked’ dimers stalled on MTs. Although our observations indicate that GSK3β-phosphorylation likely has no role in MT binding, as endogenous KHC isolated from constitutively active-GSK3β flies (sgg^ACTIVE) or loss of GSK3β function flies (sgg^[M11]−/+) bound MTs (Fig. S6B, Table S1), similar to KHCS314A and S314D (Fig. 2D, Table S1), we propose that disrupting GSK3β-phosphorylation on KHC affects the tightly controlled kinesin 1 motor behaviors by altering (1) KHC associations with KLC to build a functional kinesin 1 motor, and/or (2) kinesin 1 motor associations with MTs, membranes/organelles and/or regulatory factors in a physiological setting. Indeed, as previous work in cells suggested that binding of kinesin 1 to membranes activates motility and is likely dependent on the phosphorylation state of KLC (Schafer et al., 2009; DeBerg et al., 2013), disrupting motor binding to membranes/cargo could contribute to the motility defects in khc^S314A larvae. Previously GSKβ-mediated KLC phosphorylation was shown to release kinesin 1 from vesicles (Morfini et al., 2002), but other work suggested that KHC phosphorylation induces membrane association (Lee and Hollenbeck, 1995). We showed that binding of kinesin 1 to membranes is coupled to active GSK3β with excess active GSK3β increasing and mutant non-functional GSK3β (sgg^[M11]−/+) decreasing kinesin 1 binding to membranes (Dolma et al., 2014). Consistent with these observations, fractionated membranes isolated from khc^S314A−/+ showed significant decreases in both KHC and KLC binding to membranes (P<0.05 for both), compared with wild type (Fig. 4BC). The level of active GSK3β bound to membranes was also significantly decreased (P<0.05) (Fig. 4B,C). However, HMW KHC was still present with membranes further demonstrating that HMW KHC is not the folded/inactive soluble form. A significant decrease in dynein binding to membranes was also observed (P<0.005). These observations are consistent with our previous work, which showed that motor binding to membranes requires active GSK3β (Dolma et al., 2014). Increased motor binding to membranes was seen in sgg^ACTIVE, which was decreased in sgg^[M11]−/+ (Dolma et al., 2014). Therefore, KHC phosphorylation by GSK3β likely contributes to kinesin 1 attachment to vesicles/organelles.

In summary, we have identified a STOP/GO switch for kinesin 1 movement. Constitutive GSK3β phosphorylation at S314 in the KHC motor domain halts motility without perturbing MT binding (Fig. 2K). In contrast, disrupting GSK3β phosphorylation at S314A causes uncoordinated motility, by decreasing MT/membrane binding and reducing ATP hydrolysis, interfering with the tightly coordinated fine-tuning mechanism of detachment/attachment of motor heads from/to MTs and/or disassociation/association of the tail from/to membranes (Figs 2K and 4F). Although future work is needed to isolate how other kinases function together with GSK3β to coordinate and regulate kinesin 1 motility under physiological conditions, it is clear that a complex regulatory mechanism exists to precisely control motor protein activity within long narrow-caliber axons in a whole organism.

Transgenic line UAS-Mito-HA-GFP/Cyo was obtained from the Bloomington Stock Center (Table S2). pGAL4-62B SG26-1 (Table S2), which is expressed in eight motor neurons was used for neuronal expression of mitochondria (Gunawardena and Goldstein, 2001; Gunawardena et al., 2013). For generation of khc^S314A−/− homozygous larvae, khc^S314/CyO males were first crossed to SG9-GAL4;T(2:3),CyO,TM6B, Tb/Pin88K virgin females to obtain SG9-GAL4/y;khc^S314A/T(2:3),CyO,TM6B,Tb males. The chromosome carrying T(2:3),CyO,TM6B,Tb is referred to as B3 and carries the dominant markers, Hu, Tb and CyO. The larval Tb (tubby) marker is used to select larvae. The SG9-GAL4/y; khc^S314A/B3 males were crossed to SG9-GAL4/y; khc^S314A/B3 virgin females and the elongated (homozygous khc^S314A−/−), and tubby larvae (heterozygous khc^S314A/B3) were used for immunohistochemistry. For generation of SG9-GAL4/y;khc⁸/B3 males, khc⁸/CyO males were crossed to SG9-GAL4;B3/Pin88K virgin females to obtain SG9-GAL4/y;khc⁸/B3 males. The SG9-GAL4/y;khc^S314A/B3 and SG9-GAL4/y; khc⁸/B3 males were crossed to virgin females that were UAS-Mito-HA-GFP/CyO, and non-tubby male/female larvae (SG9-GAL4/+; khc^S314A/ UAS-Mito-HA-GFP and SG9-GAL4/y; khc⁸/ UAS-Mito-HA-GFP) were dissected for live imaging. Only 50% of the non-tubby larvae expressed the desired genotype. Sibling tubby larvae (SG9-GAL4/+; B3/UAS-Mito-HA-GFP) were evaluated as controls. Reciprocal crossings were also carried out to confirm observations. For biochemical analysis, the loss-of-function Drosophila GSK3β line sgg^[M−11] and the active GSK3β transgenic line UAS-SGG^S9A (sgg^ACTIVE) crossed to the pan neuronal driver APPL-GAL4 was used (Dolma et al., 2014). In all cases, flies were reared at 29°C and 60% humidity.

The S314A point mutation was generated in the nos-Cas9 flies (Bloomington Stock Center). The khc gene in these flies were first sequenced to check for polymorphisms in the khc allele. The S314A point mutation was generated via CRISPR/homology directed repair (HDR). Two gRNA target sites (cctttcagACATGAACGAGCATT and AGGAACTCATCGCTAACGCTCGG) were used. The sequence of the donor plasmid that served as template for the CRISPR-mediated HDR event is provided in Fig. S7.

The gRNAs and the donor plasmid were co-injected into the nos-Cas9 embryos. F1 crosses were set with B3/CyO. The presence of the mutation was confirmed by genomic PCR and stocks were established from positive F1 crosses.

Site-directed mutagenesis was used to introduce the amino acid exchanges S209A, S242A, S314A and S314D in pGEX-K410, pGEX-K339 and pET-KIN constructs containing 410aa or 339aa of the Drosophila KHC motor domain and full-length KHC, respectively. GeneArt Site-Directed Mutagenesis kit (ThermoFisher Scientific) was used for site- directed mutagenesis. The following primers were used: S209A fwd, ATGAACGAGCATTCTGCGCGATCCCACTCAG; S209A rev, CTGAGTGGGATCGCGCAGAATGCTCGTTCAT; S242A fwd, GTGGATTTGGCCGGTGCCGAGAAGGTTTCCA; S242A rev, TGGAAACCTTCTCGGCACCGGCCAAATCCAC; S314A fwd, TGCTGCTCTCCAGCCGCTTTCAACGAATCTGA; and S314A rev, TCAGATTCGTTGAAAGCGGCTGGAGAGCAGCA.

Purification of the GST-tagged proteins KHC 410, KHC 339, KHC S209A and KHC S252A was accomplished essentially as described previously for Stewart et al. (1993) (Table S2). Briefly, the GST fusion proteins were purified from BL21(DE3) E. coli lysates by affinity absorption to glutathione-agarose. The constructs were transformed into chemically competent BL21(DE3) E. coli. The bacterial cultures were induced with 0.5 mM isopropyl β-D-thiogalactopyranoside for 4 h at 22°C. Lysates were prepared by sonication of bacterial cells in 1 M NaCl in 1×PBS/0.5% Triton-X-100/1 mM DTT/1 mM phenylmethanesulfonyl fluoride (PMSF) at pH 7.2. Cellular debris and unlyzed cells were pelleted by centrifugation at 20,000 g for 20 min at 4^οC. The lysates were incubated with glutathione-sepharose beads (GE Healthcare) overnight at 4°C. The beads were collected by centrifugation and washed extensively in 1 M NaCl in 1× PBS/0.02% Triton-X-100/1 mM DTT. The GST-tagged proteins were eluted from the agarose beads with 15 mM glutathione, 200 mM NaCl, 0.1% Triton-X-100 and 1 mM DTT. Buffer exchange was performed using Amicon ultracentrifuge filter units. The proteins were transferred to PEM80 buffer (80 mM K₂PIPES, 1 mM EGTA, 1 mM MgC1₂ and 50 μM ATP, pH 6.9) with 1 mM DTT, for the MT gliding assay, frozen in liquid N2 in small aliquots, and stored at −70°C. The His-tagged KIN proteins (pET KIN S209A/S242A, pET KIN S242A/S314A, pET KIN S209A/S314A and pET KIN S209A/S242A/S314A) were induced using the same protocol. The lysis buffer contained 10 mM imidazole, 1 mM PMSF, 0.1% Triton-X-100, 1.3 mM benzamide and 1× PBS at pH 8. The lysates were incubated with Ni-NTA beads (GE Healthcare) and the beads were washed in wash buffer (20 mM imidazole and 1× PBS at pH 8). The his-tagged proteins were then eluted using elution buffer containing 250 mM imidazole and 1× PBS at pH 8.

Recombinant GST-GSK3β (SignalChem) was used for the in vitro kinase assay (Table S2). Recombinant KHC purified from E. coli were incubated with 50 ng GSK3β and 1mCi/100 γ32P-ATP for 30 min at 37°C. The reaction was terminated using 4×sample buffer. Control reactions containing 3 µM of GSK3β inhibitor CT99021 (Selleck) were carried out to evaluate the specificity of the GSK3β phosphorylation assay. Proteins were separated by SDS-PAGE, the gel was dried and sealed in saran wrap, and exposed to X-ray film overnight. After exposure, gels were stained with Coomassie Brilliant Blue to visualize proteins.

In vitro MT binding assay was performed following the instructions in the Microtubule Binding Protein Spin-Down Assay Biochem Kit (Cytoskeleton) (Table S2). Briefly, tubulin was polymerized to MTs by incubation in a 37°C water bath for 20 min in the presence of cushion buffer containing sucrose. 2 mM Taxol was used to stabilize the polymerized MTs. 5 μg of the recombinant KHC fusion proteins were incubated with the polymerized MTs at room temperature in the presence of General Tubulin Buffer (Cytoskeleton) supplemented with taxol. The reaction mixtures were placed on top of 100 μl Cushion Buffer and spun at 100,000 g at room temperature for 40 min. The uppermost layer of the supernatant was carefully removed. This is the soluble (S) fraction. The pellet fraction (P) containing tubulin was suspended in 1× Laemmli sample buffer. For all of the experiments, the soluble fraction was precipitated with 100% TCA. The pellet obtained after TCA precipitation was suspended in 1 M Tris-base (pH 10.4) and 1× Laemmli sample buffer. Both the S and P fraction were run on SDS-PAGE and visualized using Coommassie Brilliant Blue or western blot.

For gliding assays, the flow chambers were constructed as described previously (Tao et al., 2006). The coverslips were acid washed overnight and washed extensively with water for 30 min. Flow chambers were incubated for 5 min with anti-GST antibody [1:100 Invitrogen (Table S2)] GST tagged KHC in dilution buffer [10 mM ATP, 1 mg/ml BSA and 150 mM NaCl in BRB80 (80 mM PIPES.KOH at pH 6.8, 5 mM MgCl₂ and 1 mM EGTA)]. After washing with blocking buffer (1 mg/ml BSA in BRB80), the flow chamber was filled with rhodamine-labeled microtubules in motility buffer (2 mM ATP, 20 mM taxol, 1 mM DTT, 0.1 mg/ml glucose oxides, 0.37 mg/ml casein, 0.02 mg/ml catalase and 2.25 mg/ml glucose). All assays were performed at 25°C. Gliding microtubules were observed by a Nikon Eclipse TE 2000U microscope using the 40× objective. The trajectories of the MTs were traced and the gliding velocity, average run length and total distance traveled were calculated using the manual tracking plugin in NIH imageJ software.

A phosphate standard curve was plotted using the Cytophos reagent (Cytoskeleton) and the ATPase assay was performed following the instructions in the HTS kinesin ATPase Endpoint Assay Biochem Kit (BK053) (Cytoskeleton) (Table S2). Briefly, the purified proteins were diluted to ∼2 µg in kinesin reaction buffer [80 mM PIPES (pH 7), 1 mM EGTA and 1 mM MgCl₂] and incubated with polymerized MT in 96-well microtiter plates. ATP was added and the reaction was allowed to proceed for 5 min at room temperature. The reaction was terminated by adding Cytophos reagent to each well and the reactions were allowed to proceed at room temperature for exactly 10 min. The readings were obtained using a spectrophotometer at 650 nm. The ATPase activity was calculated using the formula: ATPase activity=(S_a×R_v)/(S_v×T), where S_a=concentration of phosphate (µM) generated in unknown sample well, R_v=reaction volume (µl), S_v=sample volume (µl) and T=reaction time

Third instar larvae were dissected and fixed, and segmental nerves were immunostained as previously described (Banerjee et al., 2018; Fye et al., 2010). Briefly, larvae were dissected in dissection buffer [2× stock containing 128 mM NaCl, 4 mM MgCl₂, 2 mM KCl, 5 mM HEPES and 36 mM sucrose (pH 7.2)]. Dissected larvae were fixed in 8% paraformaldehyde, washed with PBT [phosphate-buffered saline (PBS) supplemented with 0.1% Tween-20] and incubated overnight with antibodies against CSP (1:10, Developmental Studies Hybridoma Bank, Table S2). Larvae were incubated in secondary antibodies (Alexa anti-mouse 488, 1:100, Invitrogen) and mounted using Vectashield mounting medium (Vector Labs). For CT99021 feeding, larvae were fed food containing 5 µM and 10 µM CT99021 (Cell Signaling, Table S2) dissolved in 0.01% DMSO or food containing 0.01% DMSO alone, for 12-14 h prior to dissection and immunohistochemistry with CSP. Images of segmental nerves were collected using a Nikon Eclipse TE 2000U microscope using the 40× objective; NMJs and cell bodies were imaged using the 100× objective. Quantitative analysis on the extent of blockages was carried out by collecting six confocal optical images from larval neurons from the region directly below or posterior to the larval brain, where several segmental nerves are visible or come into focus through the optical series. For each genotype, five to seven animals were imaged, and nerves were analyzed over a length of 50 µm, using the threshold, density slice and particle analysis functions in NIH ImageJ software (Schneider et al., 2012) as previously described (Banerjee et al., 2018). For NMJ analysis, NMJs between muscles 6 and 7, and muscle 4 at segments A2 or A3 from six to eight animals were imaged. The threshold, density slice and particle analysis function in NIH image software were used to quantify the number of boutons and synaptic length. The HRP and DLG intensities were obtained using NIH ImageJ and graphed using an Excel worksheet.

Larvae were dissected and immediately imaged under physiological conditions as previously detailed (Kuznicki and Gunawardena, 2010). The motility of Mito-GFP was visualized within living larval segmental nerves using a Nikon Eclipse TE 2000U microscope using the ×90 objective. From each larva, four sets of movies at an imaging window frame size of 90 μm at 150 frames were taken from the mid-region of the larva at an exposure of 500 ms using a Cool Snap HQ cooled CCD camera (Photometrics) and the Metamorph imaging system (Molecular Devices). Kymographs were generated in Metamorph using the kymograph stack tool. The movies were imaged for each genotype at a spatial resolution of 0.126 μm/pixel. Movies were analyzed using a MATLAB-based particle tracker program (Yang et al., 2005) as previously detailed (Gunawardena et al., 2013). Vesicle trajectories were analyzed to obtain the overall distribution of cargo populations and individual vesicle movement behaviors (velocities, pause frequencies/durations and run lengths). Duration-weighted segmental velocity evaluates the average velocity behavior that vesicles exhibit per time spent moving. Pause frequency details the total number of times a cargo pauses divided by its total time in movement. Switch frequency is defined as the number of reversals per second (times/s, y-axis). Flux is a measure of the fraction of cargo moving per second. Pause duration evaluates the total time a cargo pauses within its directional track. Run length is a description of the total distance a cargo moves in a particular direction before pausing or changing direction.

Five milliliters of larval brains from khc^WT and khc^S314A−/+ flies were collected and homogenized in acetate buffer [10 mM HEPES (pH 7.4), 100 mM K acetate, 150 mM sucrose, 5 mM EGTA, 3 mM Mg acetate and 1 mM DTT] with proteinase inhibitor (Roche) and phosphatase inhibitor (Invitrogen). The homogenate was centrifuged at 1000 g for 10 min and the debris was discarded. The resulting PNS was brought to 40% sucrose. This mixture was overlaid with cushions of 35% and 8% sucrose, and the gradient was centrifuged at 50,000 g for 1.5 h in a TLS55 rotor (Beckman Coulter). Vesicles, membranous organelles and membrane-associated proteins were found at the 35/8 interface, while heavier membranes and mitochondria were found in the pellet. Equal amounts of protein from the PNS, 35/8 interface, soluble and pellet fractions were analyzed by western blotting. pY216 GSK3β monoclonal antibody (Abcam, ab75745, 1:1000, Table S2), anti-KHC polyclonal antibody (Cytoskeleton, AKIN01, 1:1000, Table S2), anti-KLC monoclonal antibody (a gift from L. S. Goldstein, University of California, San Diego, USA, 1:1000) (Gindhart et al., 1998, Table S2), anti-DIC monoclonal antibody (Abcam, ab255981, 1:1000, Table S2), anti-Rab5 polyclonal antibody (Abcam, ab31261, 1:1000, Table S2) and anti-tubulin monoclonal antibody (Invitrogen, 13-8000, 1:1000, Table S2) were used. Immunoreactions were detected using the SuperSignal West Femto Maximum sensitivity substrate (Invitrogen) and imaged using QuantityOne (Bio-Rad). Quantification analysis was performed using Imagelab software (Table S2). Relative intensities were calculated by dividing the intensity value for each sample by the intensity value of Rab5 and then normalized to wild type, so that wild type was 1. Statistical significance was calculated using a two-sample two-sided Student's t-test. Differences were considered significant at P=0.05, which means a 95% statistically significant correlation from three separate membranes from three independent experiments.

For preparation of Drosophila head extracts, 5 ml of fly heads from KHC WT, khc^S314A−/+, sgg^ACTIVE or sgg^[M11]−/+ flies were homogenized in acetate buffer as previously described (Banerjee et al., 2018; Dolma et al., 2014; Haghnia et al., 2007). The lysate was centrifuged at 1000 g for 10 min at 4°C. Concentrations of the extracts were determined using BCA (bicinchoninic acid) protein assay (Pierce). For immunoprecipitation, 2 mg of the fly head lysate was incubated overnight with 4 µg KHC antibody (Cytoskeleton) at 4°C. Protein A/G Magnetic Beads (Pierce) washed in wash buffer (Tris-buffered saline containing 0.05% Tween-20) was added to the mixture and incubated at room temperature for 1 h. Magnetic beads were then eluted in 100 µl low pH elution buffer (Pierce). The low pH was neutralized by adding 15 µL Tris (pH 8.8). The concentration of the KHC pull down was determined by BCA assay. Western blot analysis was used to evaluate the extent and purity of the KHC immunoprecipitation. For in vivo phosphorylation, KHC was first immunoprecipitated from wild-type and sgg^[M11]−/+ flies using KHC antibody. Using western blot analysis, the level of phosphorylated KHC was evaluated using the p-Ser/Thr antibody (Abcam).

The statistical analysis used for each experiment is indicated in each figure legend. First, power and sample size (n) calculations were performed on Minitab18 for each experimental paradigm: comparing two means from two samples, with two-sided equality to identify the sample size that corresponds to a power of 0.8 with α=0.05. For immunofluorescence analysis of axonal blockages, NMJ length and number of boutons statistical analysis was performed in Excel (Table S2), using the two-sample two-sided Student's t-test. Overlaid dot plots were constructed for all figures using OriginLab/OriginPro (Table S2). Differences were considered significant at a significance level of 0.05, which means a 95% statistically significant correlation for 5-10 individual larvae from several independent crosses. For western blots, quantification analysis was performed using Image Lab software. Data obtained from Image Lab was analyzed in Excel using both the two-sided Student's t-test and the Welch's t-test with no significant changes. Differences were considered significant at 0.05, which indicates a 95% statistically significant correlation for three separate membranes from three independent experiments. The statistical analysis used for in vivo motility analysis of mitochondria is detailed elsewhere (Krzystek et al., 2021). Briefly, to select the appropriate statistical test, data distributions for each transport dynamic analyzed were first checked for normality using the nortest package of R: the Lilliefors test and Anderson–Darling test. Statistical significance of normal distributions was calculated by one-way ANOVA/post-hoc analysis to reduce Type I error, followed by two-sample two-tailed Student's t-tests to compare individual groups in Excel and Minitab18. Statistical analysis reported in figures are from Student's t-tests, as results from ANOVA/post-hoc and Student's t-tests were consistent. Over 120 mitochondria were analyzed from five to six independent larvae.

We thank the members of the Gunawardena laboratory for constructive discussions, the Bloomington Drosophila Stock Center for fly lines, and Dr. Lawrence Goldstein for the KHC constructs and the KLC antibody. S.G. thanks Priyantha Karunaratne for constant support.