To better understand the role of the tail domain on kinesin-5 function, we first evaluated the effect of the tail domain on MT-stimulated ATP hydrolysis by the kinesin-5 motor domain. To circumvent the added complexity of MT crosslinking and sliding of the full length protein (Figure 1A), we generated constructs consisting of the motor-neck-linker domains (termed ‘motor’; residues 1–356), the isolated tail (termed ‘tail’; residues 913–1056), and a fusion construct in which the motor-neck-linker is linked at its C-terminus to the tail via an 8-residue linker (termed ‘motor-tail fusion’) (Figure 1B–D). We generated and studied both human-Eg5 and Dm-KLP61F constructs in parallel to characterize the conservation of features across these two well-studied orthologs (Table 1; Figure 1B–D; Figure 1—figure supplement 1K–M). For these assays, increasing levels of MTs were added to the motor, to a mixture of motor and tail, or to the motor-tail fusion and resulting MT stimulated ATP hydrolysis rates were measured (Materials and methods; Figure 1B–D). The KLP61F motors robustly hydrolyzed ATP in response to increasing amounts of MTs (k_cat7.1 s⁻¹ and K_m680 nM; Figure 1B; Table 1). The addition of equimolar KLP61F tail to a mixture of the motor and MTs led to a two-fold decrease in k_cat (3.5 s⁻¹ vs 7.1 s⁻¹) but produced little change in K_m (756 nM vs 680 nM) suggesting that the tail does not competitively interfere with motor-MT binding but rather modulates MT-stimulated ATP hydrolysis (Figure 1C). The motor-tail fusion exhibited a 2.2-fold decrease in k_cat compared to the motor alone (3.4 s⁻¹ vs 7.1 s⁻¹) and a 15-fold decrease in K_m compared to that for the motor alone (39 nM vs 680 nM) (Figure 1D). We observed similar behavior for human Eg5 motor and tail constructs. Although, the Eg5 motor displays an eight-fold increase in K_m compared to KLP61F motor at 50 mM K-acetate condition (3849 nM vs 680) suggesting the Eg5 motor is slightly sensitive to ionic strength. Therefore, we studied Eg5 constructs at both 20 mM KCl and 50 mM K-acetate ionic strength conditions. At 20 mM KCl condition, the Eg5 tail decreases the MT stimulated Eg5 motor ATP hydrolysis rate by 32% compared to Eg5 motor alone (K_cat7.6 s⁻¹ vs 5.3 s^-’;Table 1; Figure 1—figure supplement 1K). The decrease in homogeneous solubility of the Eg5 tail construct compared to the KLP61F tail is likely responsible for this mild decrease in MT-stimulated ATPase. An Eg5 motor-tail fusion construct exhibited a two-fold decrease in MT-stimulated ATPase K_cat (3.47 s⁻¹ vs 7.55 s⁻¹) nearly identical to a parallel construct for Klp61F (Figure 1—figure supplement 1L). The tail-induced decrease in MT-stimulated ATPase activity, however, was somewhat reduced at a 50 mM K-acetate condition (5.71 s⁻¹ vs 6.81 s⁻¹; Table 1; Figure 1—figure supplement 1M), suggesting that in this condition the tail-to-motor regulation is weakened (see next section). Our studies reveal that the kinesin-5 tail domain decreases the MT-stimulated ATP hydrolysis rate of the motor domain in either an isolated or a fused configuration and that this feature is conserved across human Eg5 and Dm-KLP61F.

To understand the biochemical basis for the kinesin-5 tail-mediated regulation of motor domain MT-stimulated ATP hydrolysis, we studied the binding of kinesin-5 motor domain with and without the MT lattice. First, we measured the capacity of the motor domain to co-purify and co-elute with a C-terminally strep-II tagged KLP61F tail domain construct onto streptactin XT resin (See Materials and methods; Figure 1E; Figure 1—figure supplement 1A). We compared the binding activities at two ionic strength conditions (25 and 75 mM KCl) for which motility assays indicated distinct modes of tail-dependent motility regulation (see below; Figures 3–4). We also tested how the binding is influenced by distinct nucleotides that trap the motor domain in different steps of the ATP hydrolysis cycle (Figure 1E). At 25 mM KCl, the motor co-eluted with the tail in the 2 mM ADP and in the nucleotide free state (in presence of Apyrase enzyme) but very poorly co-eluted with the tail in the presence of 2 mM AMPPNP (Figure 1E). Furthermore, the proportion of Klp61F motor co-eluting with the tail decreased dramatically at 75 mM KCl conditions (Figure 1E). Higher concentration of motor and tail did not change the molar ratio of the eluting proteins (Figure 1—figure supplement 1A).

Next, we reconstituted binding of the Klp61F and Eg5 motor and tail constructs onto Paclitaxel-stabilized MTs in different nucleotide state conditions and analyzed the binding using MT co-sedimentation assays (Figure 1F–I; Figure 1—figure supplement 1B–H). We again used conditions that mimic each step of the kinesin ATP hydrolysis cycle (Figure 1F–I; Figure 1—figure supplement 1B–H). We also compared the binding activities at two ionic strength conditions (25 and 75 mM KCl) as described above. The KLP61F tail and motor individually bound MTs in the presence of 2 mM ADP, as has been shown previously (Figure 1—figure supplement 1G–H; Weinger et al., 2011). The molar ratio of the bound tail and motor to MTs (polymerized tubulin) are roughly ~0.5 and~1 per αβ-tubulin at 25 mM KCl and decreased at 75 mM KCl (Figure 1—figure supplement 1F,H). MT co-sedimentation of motor and tail together in the presence of 2 mM AMPPNP or ADP.AlF4 revealed that the tail is displaced from MTs, while the motor bound with high affinity to saturation (Figure 1F; Figure 1—figure supplement 1B–C). Even at the lowest molar ratio of motor to MTs, where there is an abundance of unoccupied MT lattice sites, the tail does not co-sediment with MTs and does not compete with the motor for MT binding sites.

In the ADP and nucleotide free states, the KLP61F motor domain recruits the tail domain more effectively to the MT lattice despite the different motor affinities for MTs. In the presence of 2 mM ADP, the motor binds MTs with low affinity in a concentration-dependent manner, and its binding molar ratio is higher at 25 mM KCl compared to 75 mM KCl (Figure 1G; Figure 1—figure supplement 1D). At 25 mM KCl, a higher amount of motor binds MTs, resulting in a higher amounts of tail being recruited into the pellet, approaching a molar ratio of ~1 motor per αβ-tubulin MT lattice site (Figure 1G; Figure 1—figure supplement 1D). At 75 mM KCl, both motor and tail amounts bound to MTs decreased and their molar ratio decreased (Figure 1I). MT co-sedimentation in the presence of apyrase (1 U/ml) to promote thenucleotide-free state of the motor revealed a high affinity of the motor for MTs at both 25 and 75 mM KCl (Figure 1H). Under the nucleotide-free condition, higher levels of tail are recruited in a manner that correlates with the amount of motor bound to MT in the pellet fraction (Figure 1I). Quantitative densitometry analyses indicates the molar ratios of tail recruited to the MT-bound fraction are highest in the ADP and nucleotide-free state in contrast to the AMPPNP and ADP.AlF4 states (Figure 1I). We studied analogous human Eg5 motor and tail constructs in the AMPPNP, ADP, and nucleotide free states which revealed essentially similar patterns of behavior to those described above for KLP61F (Figure 1—figure supplement 1I–J). The human Eg5 motor recruits the Eg5 tail to the MT-bound fraction in the presence of 2 mM ADP. Despite the high affinity of the Eg5 motor for MTs in the presence of 2 mM AMPPNP, the tail remains unbound and in the soluble fraction (Figure 1—figure supplement 1H–J). This suggest that the Eg5 tail binds to Eg5 motor in the presence of 2 mM ADP, but its binding to the Eg5 motor in presence of 2 mM AMPPNP is similar but weaker than the KLP61F tail (Figure 1—figure supplement 1I–J). Together, our studies revealed that the ATP and ADP.Pi states of the kinesin-5 motor domains exhibit a high binding affinity for MTs, but a low affinity for the tail domain, while also displacing the isolated tail from MTs. In contrast, in the ADP and nucleotide-free states the kinesin-5 motor exhibits increased affinity for the tail, leading the tail to be recruited to the MT bound fraction despite the difference in the motor-MT affinity in these two states (Figure 1J).

We used cryo-electron microscopy (cryo-EM) to investigate the kinesin-5 tail-motor interface and its role in motor mediated ATP hydrolysis. We collected cryo-EM images of KLP61F motor decorated onto MTs in the AMPPNP state, and the KLP61F motor and tail decorated onto MTs in the nucleotide free state. Class averages for the MT-decorated segments in the two states clearly revealed repeating motordensities decorating the MT lattice sites (Figure 2E). New and previously described image analysis strategies (see Materials and methods) were used to calculate and refine structures of MTs decorated with the KLP61F motor in the AMPPNP state and the KLP61F motor and tail complex in the nucleotide-free state to ~4.4 Å. We adopted a new strategy to apply local classification to the motor and tail densities in the nucleotide free state, which involved classifying small patch of the MT lattice (Materials and methods). Only 70% of the data were utilized for this refinement step (Materials and methods). However, this new strategy generally improved the resolution of the motor domain density but did not enhance the occupancy or resolution for the tail (Figure 2A–F; Figure 2—figure supplement 1A–C; Table 2; see Materials and methods). Additional density, which we attribute to the tail domain, is observed on top of the motor domain in the nucleotide-free motor and tail-decorated MTs. Despite the size of the tail domain (80–100 residues), this density is small and accounts for only about one third of this mass (Figure 2—figure supplement 1B–C,H). This suggests that two thirds of the tail is either unstructured, or forms a flexibly attached and separate domain; a major part of the tail region is rendered invisible after averaging. Furthermore, the density attributed to the tail region cannot be interpreted in terms of secondary structure as there is a substantial drop in the resolution to ~8 Å at the motor-tail interface (Figure 2—figure supplement 1B–C; Video 1). In keeping with the biochemical data, we do not observe any ordered additional density on the MT lattice, consistent with the idea that the tail does not bind the MT lattice in the presence of the motor domain. Overall, the data strongly suggest that the tail interacts loosely with the underlying motor domain, binding directly to its α0 helix hairpin element, located on its MT minus end facing side (Figure 2D–F). Comparison of the KLP61F motor AMPPNP structure with the motor-tail nucleotide-free structure revealed conformational changes within the motor domain leading a major rotation in the α0 helix hairpin (Figure 2D–F; Video 1), and suggests how elements of the nucleotide-free state form an effective binding site for the tail (Figure 2C). One end of the α0 helix is connected to the phosphate binding loop (p-loop) (Figure 2G) and allows possible feedback between the nucleotide pocket and the tail binding site.

Model building for both the kinesin-5 AMPPNP and nucleotide-free motor domain structures revealed conformational changes in α0-helix hairpin that may regulate the affinity of the tail within the motor domain (Table 3; Figure 2C,F; Video 1). Compared to kinesin-1, the kinesin-5 motor structure revealed the conserved kinesin fold with two longer and reorganized class-specific L6 and L8 loops which form part of the motor MT-binding interface (Figure 2—figure supplement 1E–F-G), similar to the map of the recently-determined S. pombe and U. maydis kinesin-5 motor domain structures (von Loeffelholz and Moores, 2019; von Loeffelholz et al., 2019). Structurally, the kinesin-5 motor domain can be divided to three subdomains, comparable to those seen in kinesin-1: the N-terminal subdomain, the upper subdomain, and the MT-binding subdomain (Figure 2FH-I; Cao et al., 2014; Shang et al., 2014). The MT-bound subdomain consists of the α4-helix L11 and L12, as described for kinesin-1 (Figure 2I–J; Figure 2—figure supplement 1E–F-G; Video 1). As in kinesin-1, the N-terminal subdomain rotates around the MT-bound subdomain in the kinesin-5 nucleotide-free state, leading to a reorganization of the MT minus-end facing end of the motor (Figure 2H–I; Figure 2—figure supplement 1F–G). In the nucleotide-free state, the N-terminal subdomain (blue) rotates upward with respect to the upper subdomain (pink) and the MT bound α4 helix (switch-II), and L11 (red) to open the switch I loop in the active site (Figure 2H–J). Compared to the AMPPNP state, the N-terminal subdomain in the nucleotide-free state rotates upward by 20° causing the α0-helix hairpin, which lies at its extreme tip, to move upward by 10 Å from the MT surface (Figure 2H–J; Video 1). This N-terminal subdomain rotation repositions the switch I, switch II, and P-loops, resulting in an open active site (Figure 2H–J), whereas in the AMPPNP state, the α0-helix is positioned downward and 10 Å closer to the MT lattice, (Figure 2H–J) due to ATP binding and active site closure. Our structures suggest that binding of the tail domain to the N-terminal subdomain likely stabilizes the upward α0-helix conformation which prevents the P-loop and switch I//II from engaging ATP (Figure 2J; Video 1). Our models suggest that the tail interface stabilizes this upward N-terminal subdomain motor conformation leading to a decrease in the ability of the active site to bind incoming ATP. This suggests that the motor-α0-helix may cycle between ‘on’ and ‘off’ states while binding the tail to result in a slower ATP hydrolysis cycle.

We next set out to examine the role of the tail to motor interface in kinesin-5 motility, we reconstituted the motility of homotetrameric human Eg5 motors along individual MTs. We studied two human Eg5 constructs that either contain or exclude the tail domain (Figure 3A). We purified full-length Eg5 (termed FL-Eg5, residues 1–1056), full-length Eg5 with a C-terminal GFP (termed FL-Eg5-GFP), and mutant Eg5 with the tail domain deleted with a C-terminal GFP (termed Eg5-Δtail-GFP, residues 1–912) (Figure 3A–B). These motors were expressed in insect cells and purified using StrepII-tag affinity, followed by size exclusion chromatography (Figure 3—figure supplement 1A–B; see Materials and methods). Deleting the tail domain does not alter the shape or the homotetrameric oligomerization of Eg5, as was described previously (Acar et al., 2013; Weinger et al., 2011; Figure 3B; Figure 3—figure supplement 1A–B). Using TIRF microscopy, we reconstituted FL-Eg5-GFP and Eg5-Δtail-GFP motility along GMPCPP or Paclitaxel-stabilized AlexaF-633 labeled MTs (Figure 3A, lower panel). Both FL-Eg5-GFP and Eg5-Δtail-GFP exhibited robust and processive motility along MTs in 25 mMHEPES pH 7.5 containing 25 to 100 mM KCl (Figure 3D–E; Video 2). The processive motility we observe in pH 7.5 condition is highly homogeneous in contrast to the non-processive diffusive motility that has been seen at 80 mM PIPES pH 6.8 (BRB-80) with a range of 0–100 mM KCl conditions (Kapitein et al., 2008; Kapitein et al., 2005; data not shown). These observations, coupled with previous work on other kinesin motors, suggest that buffer composition, which influences pH and ionic strength, is critical for observing robust Eg5 motor motility along single MTs and affects landing rates along MTs (Kapitein et al., 2008).

We analyzed the motility properties of FL-Eg5-GFP and Eg5-Δtail-GFP motors at 25 mM HEPES pH 7.5 buffer conditions with added ionic strengths of 25, 50 and 100 mM KCl. At 25 mM KCl, FL-Eg5-GFP exhibited active yet extremely slow velocities (7 nm/s; Table 4; Figure 3D, left panel; Video 3), which was roughly four times lower than the velocity of Eg5-Δtail-GFP (31 nm/s; Figure 3E, left panel; Video 2). Quantitative analysis of motility events revealed mono-disperse distributions of velocities for the two motors that fit single Gaussians with single average values (Figure 3F–G, top panels). At 50 mM KCl, FL-Eg5-GFP exhibited a more than three-fold increase in average velocity (24 nm/s) compared to 25 mM KCl conditions (Figure 3D, middle panel). In contrast, Eg5-Δtail-GFP exhibited a nearly identical average velocity (35 nm/s) at 50 mM KCl, which is indistinguishable from what was observed at 25 mM KCl (Table 4; Figure 3G, middle panel). At 100 mM KCl, FL-Eg5-GFP motors exhibited two velocities (24 and 41 nm/s), with a bimodal distribution representing 60% and 40% of total, respectively (Figure 3D right panel; Video 3; Figure 3F, lower panel). Eg5-Δtail-GFP motors also exhibited two velocities at 100 mM KCl (36 nm/s and 55 nm/s) that were 20% higher than FL-Eg5-GFP and showed a similar bimodal distribution representing 85% and 15% of the total, respectively (Table 4; Figure 3D right panel; Video 3; Figure 3G, lower panel). At the 50 and 100 mM KCl conditions, fewer motor landing events were observed compared to 25 mM KCl (Figure 3D–E, right panels) for both FL-Eg5-GFP and Eg5-∆Tail-GFP. Thus, FL-Eg5 tetramers strongly respond to ionic strength, with 25 mM KCl significantly reducing its velocity, while Eg5-∆tail-GFP remains largely impervious to these effects. At 50 to 100 mM KCl, Eg5-Δtail-GFP consistently exhibited 20% higher velocities compared to FL-Eg5-GFP with bimodal distributions of slow and fast motors (Table 4).

We next explored the potential for FL-Eg5-GFP and Eg5-Δtail-GFP motile particles to form higher-order oligomers and sought to determine whether they form clusters based on the relative fluorescence intensities of motile GFP-fused motors (Figure 3F–G, inset graphs). Kymographs show that FL-Eg5-GFP motors in 25 mM KCl and Eg5-Δtail-GFP motors in 25 and 100 mM KCl conditions exhibited a homogeneous distribution of low-intensity values that was well fit by a single Gaussian (Figure 3D–E, left panels; Video 3; Table 4; Figure 3F,G insert graphs in top panels). In contrast, FL-Eg5-GFP particles exhibited both bright and dim intensity value distributions and a bimodal intensity value distribution at 100 mM KCl (Figure 3D, right panel; Video 3) which can be fitted by two overlapping Gaussians. FL-Eg5-GFP particles displayed high intensity values that were on average four-fold higher than the dim motile particles observed at 25 mM KCl, and low intensity values that were roughly two-folds higher than the dim particles at 25 mM KCl (Figure 3F, insert graph in bottom panel; Table 4). Conversely, at 100 mM KCl, Eg5-Δtail-GFP motile particles exhibited a narrow intensity distribution that closely matched Eg5-Δtail-GFP motile particle distributions at 25 mM KCl (Figure 3G, insert graph in bottom panel; Table 4). Thus, FL-Eg5-GFP and Eg5-Δtail-GFP both show motility as homogeneous particles, likely individual homotetramers, with little clustering at 25 mM KCl. At 100 mM KCl, FL-Eg5-GFP motors form clusters consisting of up to two to four homotetramers at 100 mM KCl, unlike the Eg5-Δtail-GFP, which remained as individual homotetramers under all conditions tested. The tail domain may thus mediate inter-homotetramer interactions that induce clustering of Eg5 motors on the MT.

Analysis of the FL-Eg5-GFP and Eg5-Δtail-GFP motility run lengths revealed that the tail influences the processive motility of kinesin-5. At 25 mM KCl, both FL-Eg5-GFP and Eg5-Δtail-GFP motors exhibit run lengths that exceed the experimental imaging time with both motors eventually concentrating at MT plus-ends and were thus were not quantifiable (Figure 3—figure supplement 1C; Table 4). In contrast, at 50 mM KCl we observe measurable run lengths for FL-Eg5-GFP and Eg5-Δtail-GFP motility events that are identical and in the range of 13–14 μm. At 100 mM KCl, FL-Eg5-GFP retained long run lengths with an average of 13 μm, while Eg5-Δtail-GFP showed a 45% decrease in run lengths to an average of 8 μm (Figure 3C; Table 4). These data suggest that the tail plays a role in maintaining homotetramer-MT interactions, with deletion of the tail resulting in a less processive motility cycle and a decrease in its affinity to the MT lattice.

To understand the impact of the tail-to-motor interaction on kinesin-5 MT motility under relative MT sliding conditions, we reconstituted Eg5 MT sliding in vitro using three-color TIRF assays in 25 mM HEPES, pH 7.5 with 25–50 mM KCl conditions (Figure 4A). AlexaF-633 and biotin-labeled MTs were attached to PEG-treated glass surfaces via a neutravidin to biotin linkage (anchored MT, red). AlexaF-560 labeled MTs (free MT yellow) and FL-Eg5-GFP or Eg5-Δtail-GFP motors were then added to reconstitute the crosslinking and sliding of free-MTs (yellow) along the anchored MTs (red) (Figure 4A). Under these conditions we observed that 10–20 nM FL-Eg5-GFP motors promote robust MT sliding. FL-Eg5-GFP motors were able to bind along the anchored MTs, crosslink free-MTs (yellow), and zipper MTs to form anti-parallel MT sliding zones (Figure 4A-left; Figure 4—figure supplement 1; Video 4). Within a sliding event between two antiparallel MTs, FL-Eg5-GFP motors undergo slow motility similar to those observed along single MTs (Figure 3) and concentrate within newly formed two-MT bundles to produce sliding zones (green) (Figure 4B-left). To determine the effects of ionic strength on sliding, we measured motor and MT velocities at increasing solution ionic strength (25–50 mM KCl) (Figure 4C; Table 5). At 50 mM KCl, FL-Eg5-GFP motors displayed a two-fold increase in velocity similar to the effect along single MTs (Figure 3). This increase in motor velocity on anchored MTs matched a two-fold increase in the sliding rate of the free-MTs along the anchored MTs (Figure 4C; Table 5). Our experiments therefore show a correlation between the tail-mediated decrease in kinesin-5’s velocity at lower ionic strength, and the velocity of the MTs in anti-parallel MT sliding (Table 5; Figure 4C).

We next studied the Eg5-Δtail-GFP in the MT sliding assays at similar motor concentrations (10–20 nM). Strikingly, the Eg5-Δtail-GFP motors crosslinked the free-MT only at focal points and were unable to completely zipper the free MT along the anchored MT (Figure 4A, right; Figure 4B right; Video 4). This MT zippering defect by the Eg5-Δtail-GFP motors lead the free MTs to ‘scissor’ along the anchored MTs due to Brownian motion at the crosslinking point (Figure 4B-right; Video 4). The scissoring behavior occurred most frequently near or at MT plus-ends, where the Eg5-Δtail-GFP motors concentrate (Figure 4B, lower right, second and third panels). We next studied the effect of increasing Eg5-Δtail-GFP motor concentrations on the formation of MT sliding zones. A ten-fold higher concentration of Eg5-Δtail-GFP (200 nM) led to a higher number of MT sliding events relative to scissoring events (Figure 4D). At 20 nM Eg5-Δtail-GFP, only 6% of MT crosslinking events transitioned toward MT sliding events and the remaining 94% of events retained a scissoring defect. In contrast, at 200 nM Eg5-Δtail-GFP about 45% of crosslinking events transitioned into sliding events and 55% remained only scissoring events (Figure 4D). A possible reason for this is tail-mediated regulation leading to slower motility, which concentrates Eg5 motors in the overlap zone and facilitates MT zippering, as suggested by the fact that a high concentration of Eg5-Δtail-GFP did not fully restore the zippering of two MTs into sliding zones.

To dissect the role of the tail-motor interface in kinesin-5 engagement within the MT sliding zones, we next studied how FL-Eg5-GFP or Eg5-Δtail-GFP motors behave outside and within active sliding zones formed by non-fluorescent FL-Eg5 (Figure 4E). FL-Eg5-GFP or Eg5-Δtail-GFP were spiked in at 1 nM into MT sliding assays containing 20 nM unlabeled FL-Eg5 to visualize single motors along the anchored MT and within active MT sliding zones (Figure 4E; see Materials and methods). FL-Eg5-GFP motors exhibit slow plus-end directed motility along anchored MTs with infrequent direction reversals, with their average velocity decreasing within MT sliding zones (5–7 nm/sec) (Figure 4H). In contrast, Eg5-Δtail-GFP motors exhibit an increased plus-end directed motility (10 nm/sec) along the anchored MT (Figure 4I). Upon entering MT sliding zones and interacting with both MTs, Eg5-Δtail-GFP motility remained rapid but the motors transitioned to bi-directional motility with frequent reversals (Figure 4I). We interpret these direction reversals to be uncoordinated MT plus-end directed motility along either of the two anti-parallel MTs within sliding zones that potentially indicate a higher rate of unbinding from the MT (Figure 4G, third panel from the left; Video 5). The Eg5-Δtail-GFP velocity is roughly two-fold higher than that of FL-Eg5 GFP along single MTs and within MT sliding zones. Eg5-∆tail-GFP also exhibited a three-fold shorter average run length compared to FL-Eg5-GFP within MT sliding zones (Table 5; Figure 4F right panels; Figure 4I; Video 5). The kinesin-5 tail thus is critical for Eg5 motors to engage within the MT sliding zones by modulating a unique motor-MT association which decreases motor unbinding rates, and potentially mediates coupling between the two MT-bound ends of kinesin-5. The bi-directional motility of Eg5-∆tail-GFP and its frequent direction reversals suggest that the two ends of the motor are undergoing rapid uncoordinated ATP hydrolysis cycles as they bind both MTs within sliding zones.

In order to understand how the tail regulates the ability of kinesin-5 to slide apart antiparallel MTs, we next measured the forces generated by FL-Eg5-GFP or Eg5-Δtail-GFP on the free MT during sliding events using optical trapping combined with TIRF assays as previously described (Shimamoto et al., 2015; Figure 5A). Here, we reconstituted anti-parallel pairs of MTs similar to those described above (Figure 4), but with an additional step of attaching polystyrene beads coated with nucleotide-free kinesin-1 motor domain to the ends of Rhodamine-labeled MTs (red). These bead-associated free-MTs were reconstituted to form anti-parallel overlaps within flow chambers containing Hilyte-647-labeled MTs (purple) anchored to the slide surface and either FL-Eg5-GFP or Eg5-Δtail-GFP (green). These sliding MT pairs were then simultaneously imaged using TIRF microscopy while measuring the force exerted on the bead using an optical trap (Figure 5A). Under these conditions, 3–10 nM FL-Eg5-GFP produced robust MT sliding events (Figure 5B left panel; Video 6). In contrast, for 3–10 nM Eg5-Δtail-GFP, MT sliding was very rarely observed, and the majority of crosslinking events resulted in free MTs undergoing ‘scissoring’ movements about a single point on the anchored MT, similar to our previous observations in TIRF sliding assays (Figure 4D). In order to recruit comparable amounts of Eg5-Δtail-GFP motors within the overlap regions and to promote relative MT sliding, we increased the amount of Eg5-Δtail-GFP in the chamber by 50–100-fold (Figure 4—figure supplement 1D; Figure 5B; Video 6). Under these conditions, regions of MT overlap exhibited increased recruitment of kinesin-5 molecules (Figure 5B; Video 6).

To measure the pushing forces generated by MT sliding under these conditions, the Rhodamine-labeled MT-bound polystyrene bead was optically trapped upon observation of a sliding event (Figure 5C). FL-Eg5-GFP-mediated MT sliding exhibited a steady increase in the force produced until a maximum ‘plateau’ force was reached, frequently resulting in many 10 s of pN of force across the MT sliding pair. This behavior was quite similar in both timescale and force magnitude to reports of Xenopus Eg5 pushing forces measured in a similar assay (Shimamoto et al., 2015). In contrast, Eg5-Δtail-GFP-mediated MT pairs exhibited short excursions of force increase and reached significantly lower ‘plateau’ values with lower total forces generated overall (Figure 5C).

For each individual MT sliding event examined, we also calculated the integrated intensity of GFP signal within the overlap region as defined by the distance between MT plus-ends (dashed gray lines, Figure 5B). This served as a proportional readout of the amount of motors localized within the overlap region. By comparing a number of MT sliding zones, we determined there is an approximately linear relationship between the relative MT overlap zone length and the number of motors within this region (Figure 5D). Furthermore, a comparable concentration of FL-Eg5-GFP and Eg5-Δtail-GFP were measured within overlap zones, suggesting that similar numbers of motors were engaged within overlaps of similar size, despite the different concentration used in the assays (Figure 5D).

We also examined the relationship between the magnitude of the force plateau reached in each individual MT sliding event and the length of the MT overlap zone during force generation. For both Eg5-GFP and Eg5-Δtail-GFP we observed a nearly linear increase in plateau force relative to MT overlap zone length (Figure 5E). However, the slopes of these relationships differed significantly between the two constructs. Here the slopes represent the plateau pushing force generated per μm length of sliding zone: the Eg5-Δtail-GFP generates about 0.6 ± 0.3 pN per μm while the FL-Eg5-GFP generates about 4.0 ± 0.6 pN per μm, indicating about a seven-fold difference in MT sliding forces (Figure 5F). The pushing forces produced within sliding anti-parallel MT bundles by ensembles of Eg5-Δtail-GFP are roughly seven-fold lower than that those generated by comparable numbers of FL-Eg5-GFP motors. Together, these data indicate that the tail domain is critical for homotetrameric kinesin-5 motor sliding behavior and mediates regulation of force production during MT sliding events.

We next determined the role of the kinesin-5 tail domain in motor localization of Eg5 in metaphase and anaphase in HeLa cells. Cells were transfected with human GFP-α-tubulin to visualize MTs, and either FL-Eg5-mCherry (FL-Eg5-mCh), Eg5-Δtail-mCherry (Eg5-Δtail-mCh), or mCherry alone (mCh). Expression of each construct in HeLa cells was assessed by western blot (Figure 6A). In fixed metaphase cells, FL-Eg5-mCh localized to spindle MTs. In contrast, localization of Eg5-Δtail-mCh was more diffuse, with increased cytoplasmic signal (Figure 6B). This difference in localization was quantified as the ratio of mCh signal localized to the mitotic spindle and mCh signal in the cytoplasm. In cells with comparable mCh-construct expression levels (Figure 6C, left), this spindle-to-cytoplasm ratio was high for FL-Eg5-mCh (2.39 ± 0.08, mean ± SEM) and significantly reduced for Eg5-Δtail-mCh (1.37 ± 0.02, p<0.0001, ANOVA with Tukey’s multiple comparisons test), indicating that deletion of the tail causes a defect in localization to metaphase spindle MTs (Figure 6C, right). Treatment with the compound BRD-9876, which locks Eg5 in a nucleotide-free-like state (Chen et al., 2017), only partially rescued the localization of Eg5-Δtail-mCh to metaphase spindle MTs, potentially due to defects in both the on and off-rates (Figure 6D and E).

Live cell imaging was used to assess whether this tail deletion localization defect was present in anaphase as well as metaphase mitotic spindles. Measurement of spindle to cytoplasm mCh intensity ratios in anaphase showed a similar reduction, indicating decreased localization to spindle MTs, for Eg5-Δtail-mCh compared to FL-Eg5-mCh as seen in metaphase cells (ratios 1.36 ± 0.04 and 2.21 ± 0.17, mean ± SEM, respectively, p<0.0001, ANOVA with Tukey’s multiple comparisons test) (Figure 6F and G). As HeLa cells progressed from metaphase to anaphase, the deletion of the tail consistently reduced the localization of Eg5 to spindle MTs (Figure 6F and H). These data support a critical role for the tail domain in kinesin-5 mitotic spindle MT localization.

Our studies suggest that the tail domain regulates processive hand-over-hand stepping during kinesin-5 motility (Figure 7B–D; Vale and Milligan, 2000). Well-studied MT plus-end-directed kinesin motors (i.e. kinesin-1) undergo hand-over-hand walking motility along MT protofilaments utilizing dimeric motor domains coupled by their neck linkers and neck coil-coils. During hand-over-hand motility, one motor domain of a dimeric kinesin binds the MT lattice in a trailing position, while the other motor domain occupies a leading position, 8 nm apart along two consecutive αβ-tubulins within MT protofilaments (Figure 7D). The trailing motor is in an ATP-like state with its neck-linker docked, while the lead motor initially binds weakly, leading to ADP dissociation and a nucleotide-free state (Figure 7D). The hand over hand cycle of dimeric kinesin-1 motors result in 8 nm steps toward the MT plus-ends (Vale and Milligan, 2000). Our studies suggest a new form of regulation for kinesin-5 hand-over-hand motility, where the tail domain decreases ATP binding and stabilizes the ADP or nucleotide-free states of the lead motor domain (Figure 7C–D). The kinesin-5 tail binds the α0-helix within the motor domain, stabilizing its N-terminal subdomain in an upward conformation and resulting in an open active site that slows its ATP binding capacity (Figures 1–3; Figure 7C–D). Our model is the simplest explanation for all the data presented here. The tail-motor regulatory mechanism, described here, is likely to be the source of kinesin-5’s conserved sliding force-generating capabilities and is essential for its anti-parallel MT sliding activity. In yeast kinesin-5 motors, such as Cin8, these interactions may also regulate the reversal of direction from minus-end-directed motility along single MTs to plus-end-directed motility within sliding zones as tail deletion in Cin8 interferes with directionality reversal (Düselder et al., 2015).

The tail-motor interface may stiffen both kinesin-5-MT interactions at both ends of the homotetrameric filament within MT sliding zones and thus may improve force transmission between the two bipolar ends (Figures 5 and 7B–D). Our data suggests that the tail-motor interaction increases the time each end of the kinesin-5 homotetramer spends in the dual motor-bound state. The tail may promote a high affinity state for the motor domains at each of the homotetramer (Figure 7C–D). In the absence of the tail domain, Eg5-Δtail-GFP motors are unable to effectively engage both sliding MTs and exhibit reversal in motility direction by moving to the MT plus-end on either sliding MTs asynchronously. This leads to a poor capacity for generating MT sliding and pushing forces (Figures 4–5). Our data also indicate that this interaction may be responsible for clustering multiple homotetramers into larger complexes under certain conditions (Figure 7A). Clusters of up to four kinesin-5 homotetramers were observed in the yeast ortholog, Cin8, where cluster formation induced motility direction reversal and generated sites for the capture of free MTs to promote MT sliding (Shapira et al., 2017). We suggest that the clustering of kinesin-5 motors may serve to coordinate the motor motility cycles of groups of motors within MT sliding zones (Figure 7A). Cooperativity may also be a result of multiple motors stalled in a traffic jam due to slow stepping, a model previously suggested for Cut7, the S. pombe ortholog (Britto et al., 2016).

The kinesin-5 tail domain is essential for mitotic spindle assembly and elongation functions. Our studies confirm that the mitotic localization defect in Eg5-Δtail-GFP may relate to its rapid binding and unbinding from spindle MTs; trapping Eg5-Δtail-GFP motor in the nucleotide-free state using chemical strategies dramatically enhances their MT spindle localization (Figure 6). A recent study identified Eg5 mutations in cultured cells that result in resistance to Eg5 inhibitors (Sturgill et al., 2016). Among these is a mutant cell strain that possesses an Eg5 mutant with motor domain mutation that traps it in the nucleotide-free state that forms a near-normal bipolar metaphase spindle and for which correct mitotic organization is restored in the presence of specific compounds (Sturgill et al., 2016). This data highlights the importance of the stabilization of the nucleotide-free state during the force generation cycle of kinesin-5.

The tail-motor interface is essential for the stable localization of kinesin-5 motors to mitotic spindle MTs by also acting as a regulatory site for phosphorylation. The tail domain contains the conserved BimC box, a mitotic CyclinB dependent kinase phosphorylation consensus site (Blangy et al., 1995; Sharp et al., 1999). This motif is conserved across kinesin-5 motors and its mitotic phosphorylation was shown to promote the accumulation of kinesin-5 motors at the mitotic spindle midzone and to mediate spindle elongation in anaphase (Sharp et al., 1999). It remains unclear how phosphorylation of the BimC box influences the kinesin-5 tail-motor interaction. It is possible that phosphorylation enhances motor regulation by the tail and further slows MT-based motility to increase MT sliding efficiency during anaphase. Future studies of kinesin-5 motors phosphorylated at the BimC box will provide further clues for kinesin-5 regulation throughout the cell cycle.

We present a new mechanism for the regulation of the kinesin-5 motor via its tail domain during plus-end-directed motility along MTs. The tail binds the motor domain to stabilize it in the high affinity MT-bound nucleotide-free state. This regulatory role is critical for slowing kinesin-5 motility along MTs allowing motors to crosslink and generate force to slide apart antiparallel MTs, an aspect of kinesin-5 that is essential for cell division to occur.

For in vitro motility studies, we generated constructs for expression insect cells full length coding regions for Human Eg5 (KIF11)(FL-Eg5: residues 1–1056) and C-terminally truncated Eg5 (Eg5-Δtail; residues 1–920) were inserted in pFastbac1 vector in frame of a C-terminal StrepII tag or fused to monomeric superfolder-GFP(msf-GFP) and a C-terminal Strep-II tag.

For in vivo imaging studies, A pcDNA-3.1 plasmid containing Eg5 full-length fused to mCherry (Eg5 FL-mCh) inserted between the EcoR1 and Not1 sites was used as a template. Plasmid pAT4206, encoding Eg5-Δtail, which encoded residues 1–912, fused to mCherry (Eg5 ΔTail-mCh), was generated using the Phusion Site-Directed Mutagenesis Kit (Thermo Scientific). These clones contained silent mutations for siRNA-resistance: T2124C, C2130T, G2133T, and G2136A. The forward and reverse primer sequences used to generate Eg5 ΔTail-mCh were 5’-GGAGCGCCAATGGTGAGCAA-3’ and 5’-AAAGCAATTAAGCTTAGTCAAACCAATTTT-3’, respectively.

For Kinetic and structural studies, coding regions for isolated Dm KLP61F and Human Eg5 motor constructs (KLP61F motor residues 1–369; Eg5 motor residues 1–360) and the tail domains for these proteins (KLP61F tail residues 906–1016 and Eg5 residues 920–1056) and KLP61F motor-tail fusion (KLP61F residues 1–369 fused to 906–1016 and Eg5 residues 1–360 fused to residues 920–1056) were inserted in frame with histidine-tag in pET v2 vector (macrolab UC-berkeley). For affinity co-purification studies, a KLP61F tail region (residues 906–1016) was fused to a C-terminal StrepII tag.

For expression FL-Eg5-GFP, FL-Eg5 and Eg5-Δtail-GFP was carried out using baculoviral expression system. Briefly 200 to 500 mL of Spodoptera frugiperda (Sf9) cells, were infected with third passaged virus for each construct and expressed for 60–72 hr. Virus-Infected Sf9 Cells were centrifuged at 1500 rpm and then washed and incubated with lysis buffer (50 mM HEPES 300 mM KCl, 10 mM beta-meractoptoethanol, 1 mM MgCl2, 0.2 mM ATP) in the presence of 0.5% Triton X100. Cells were lysed by extrusion using a dounce homogenizer and then were centrifuged at 40,000 rpm using Ti45 rotor in ultracentrifuge (Beckman). The lysate was passaged on a Streptactin XT resin (IBA lifesciences) equilibrated with lysis buffer, washed extensively and then eluted with lysis buffer supplemented with 100 mM D-Biotin. Purified Eg5 proteins were concentrated and loaded onto Superose-6 (10/300) column (GE Health Care) equilibrated onto AKTA system and eluted in 0.5 mL fractions (Figure 3—figure supplement 1A–B). Fractions were evaluated using SDS-PAGE for protein quality and Eg5 protein containing fractions were either used immediately for motility experiments or were snap frozen with 15% glycerol in lysis buffer using liquid nitrogen. We did not observe any difference in the FL-Eg5-GFP, FL-Eg5 and Eg5-Δtail-GFP activities in the frozen or freshly prepared settings (Figure 3—figure supplement 1A–B).

For kinetic and structural cryo-EM studies, Dm-KLP61F and human Eg5 motor, tail and motor-tail constructs were expressed using T7 expression in SoluBL21 cells. Briefly, cells were grown at 37°C and induced for expression using 0.5 mM Isopropyl thio-beta-glucoside overnight at 19°C. Cells were centrifuged and lysed in Lysis buffer using a microfluidizer. Proteins were purified using IDA-nickel affinity chromatography (Macherey Nagel), washed extensively and then eluted with lysis buffer supplemented with 200 mM Imidazole. The purified fractions were loaded onto a Superdex 200 (16/6) size exclusion column equilibrated with lysis buffer at 1 mL fractions. SDS-PAGE was used to evaluate fractions and concentrated pure proteins were incubated with 15% glycerol before freezing in liquid nitrogen.

MT-stimulated ATPase activity for KLP61F and Eg5 motor, tail, motor-tail fusion and Eg5-motor and tail constructs were carried out by measuring the free phosphate production rate in 50 mM Potassium acetate (K-acetate) or 20 mM KCl, 25 mM HEPES, 5 mM magnesium acetate, 1 mM EGTA, pH 7.5 buffer in the presence of a minimum of a 5-fold molar excess range of MT concentrations, using a commercially-available kit (EnzChek, Molecular Probes) at 20^o C. The Eg5 motor, tail and motor-tail fusion were additionally studied at 20 mM KCl due to the 10-fold higher Eg5 motor MT stimulated ATP hydrolysis K_m in 50 mM K-Acetate. These comparisons suggest that the Eg5 tail regulation of motor ATPase is slightly weaker at 50 mM K-Acetate conditions (Figure 1—figure supplement 1L–M).

For affinity co-purification for the motor and tail, 5–10 μmoles of KLP61F tail-StrepII tagged protein was incubated with 5–10 μmoles of the KLP61F motor domain in incubation buffer (25 mM HEPES, 25–75 mM KCl 5 mM MgCl₂ and 1 mM EGTA, 2% glycerol). The mixture (Load) was then incubated with 200 μl of StrepXT resin (IBA-biotech). The flow-through fraction from the resin was collected (FL). The resin was washed with two column volumes of incubation buffer. The bound fraction was then eluted (Elute) with incubation buffer + 50 mM biotin. The mixture was analyzed by SDS PAGE. Quantitative densitometry was carried out to obtain the motor/tail molar ratio using Biorad Image Lab software (Biorad). Motor and tail band intensities were measured, and the intensity data was converted to μmol values and then molar ratios were calculated (Figure 1E, Figure 1—figure supplement 1A).

To measure MT co-sedimentation activity for KLP61F and Eg5 motor and tail constructs, we prepared 5 mg/ml MTs in 5% DMSO by polymerization at 37 °C and then stabilized with Paclitaxel (sigma), MT co-sedimentation assays were carried out by mixing 0.01, 0.025, 0.05, 0.1, 0.25 up to 0.05 μmol of KLP61F or Eg5 motor with or without their respective tail domains, in 25 mM HEPES pH 7.5, 5 mM MgCl2, 1% glycerol, 25 mM or 75 mM KCl in the presence of 2 mM AMPPNP, 2 mM ADP, 2 mM ADP.AlF4 or 10U Apyrase to mimic the nucleotide-free state. The five to six different concentration mixtures of motors, tail indicated above were mixed with MTs in these conditions were incubated at 25°C for 20 min and then centrifuged at 18 k for 25 min at 25 °C. A control condition is usually included in which co-sedimentation is carried out without MTs (-MTs). The supernatant fractions were then removed and mixed with SDS sample buffer. The pellets were resuspended with SDS sample buffer. Equal amounts of each supernatant and pellet fractions for 0.01, 0.025, 0.05, 0.1, 0.25 μmol and -MT conditions were analyzed by SDS PAGE (Figure 1F–H; Figure 1—figure supplement 1),. Quantitative densitometry was carried out using Biorad Image Lab software (Biorad) to determine the saturated molar ratios of the motor and tail to polymerized tubulin in MTs. The intensities for each band was measured in comparison to a local background control. The intensity data were converted to μmoles and then molar ratios were calculated for each lane average ratios were calculated for all saturated lanes.

MTs were prepared by polymerizing 5 mg/ml tubulin (Cytoskeleton, Denver, CO) in BRB80 buffer (80 mM PIPES, pH 6.8, 1 mM EGTA, 4 mM MgCl2, 2 mM GTP, 9% dimethyl sulfoxide) for 30 min at 36° C. Paclitaxel was added at 250 μM before further incubation of 30 min at 36°C. The polymerized MTs were then incubated at room temperature for several hours or overnight before use. For grid preparations, KLP61F motor and KLP61F tail proteins were mixed in 0.5% binding buffer (25 mM HEPES, 35 mM potassium acetate plus 2 mM ADP to a final concentration of 0.1 mg/mL and 0.04 mg/mL respectively). For grid preparation of motor alone KLP61F motor (10 mg/mL) was diluted in binding buffer (50 mM HEPES, 70 mM potassium acetate plus 2 mM AMPNP). All samples were prepared on 1.2/1.3 400-mesh grids (Electron Microscopy Services). Grids were glow-discharged before sample application. The cryo-samples were prepared using a manual plunger, which was placed in a homemade humidity chamber that varied between 80% and 90% relative humidity. A 4 μl amount of the MTs at ∼0.5 μM in 80 mM PIPES, pH 6.8, 4 mM MgCl2, and 1 mM EGTA supplemented with 20 μM Paclitaxel was allowed to absorb for 2 min, and then 4 μl of the KLP61F motor and tail domains were added to the grid. After a short incubation of 2 min, 0.5 μL of Apyrase (25 units) was added and after a short incubation of 3 min the sample was blotted (from the back side of the grid) and plunged into liquid ethane. This procedure was repeated for the motor alone preparations without the addition of the apyrase step.

Images of frozen-hydrated KLP61F motor decorated MTs in the AMPPNP state or KLP61F motor and tail decorated MTs in the nucleotide-free state (see Table 2) were collected on a Titan Krios (FEI, Hillsboro, OR) operating at 300 keV equipped with a K2 Summit direct electron detector (Gatan, Pleasanton, CA). The data were acquired using the Leginon automated data acquisition (Suloway et al., 2005). Image processing was performed within the Appion processing environment (Lander et al., 2009). Movies were collected at a nominal magnification of 22500 × with a physical pixel size of 1.31 Å/pixel. Movies were acquired using a dose rate of ~7.96 and 8.3 electrons/pixel/second over 8.25 s yielding a cumulative dose of ~38 and 40 electrons/Å2 (respectively). The MotionCor frame alignment program (Hirschi et al., 2017; Li et al., 2013) was used to motion-correct. Aligned images were used for CTF determination using CTFFIND4 (Rohou and Grigorieff, 2015) and only micrographs yielding CC estimates better than 0.5 at 4 Å resolution were kept. MT segments were manually selected, and overlapping segments were extracted with a spacing of 80 Å along the filament. Binned boxed segments (2.62 Å/pixel, 192 pixel box size) were then subjected to reference-free 2D classification using multivariate statistical analysis (MSA) and multi-reference alignment (MRA) (Hirschi et al., 2017). Particles in classes that did not clearly show an 80 Å layer line were excluded from further processing.

For cryo-EM reconstruction, undecorated 13,14- and 15-protofilament MT densities (Sui and Downing, 2010) were used as initial models for all preliminary reconstructions. We used the IHRSR procedure (Egelman, 2007) for multi-model projection matching of MT specimens with various numbers of protofilaments (Alushin et al., 2014), using libraries from the EMAN2 image processing package (Tang et al., 2007). After each round of projection matching, an asymmetric back-projection is generated of aligned segments, and the helical parameters (rise and twist) describing the monomeric tubulin lattice are calculated. These helical parameters are used to generate and average 13, 14 and 15 symmetry-related copies of the asymmetric reconstruction, and the resulting models were used for projection matching during the next round of refinement. The number of particles falling into the different helical families varied. Helical families that had enough segments were further refined. Final refinement of MT segment alignment parameters was performed in FREALIGN (Grigorieff, 2007 without further refinement of helical parameters. FSC curves were used to estimate the resolution of each reconstruction, using a cutoff of 0.143. To better display the high-resolution features, we applied a B-factor of 200 Å, using the program bfactor (http://grigoriefflab.janelia.org). The final statistics for all data sets were described in Table 2.

In order to enhance the total mass of the mobile tail density and improve the resolution of the MT decorated with KLP61F motor and tail in the nucleotide-free state, an additional round of ‘MT-patch refinement’ processing was performed enhance conformational homogeneity. The same motion-corrected micrographs and boxes were used, but defocus parameters were re-estimated using GCTF (Zhang, 2016). MTs were then sorted into 13, 14, and 15 protofilament MTs using reference alignment as previously described (Shang et al., 2014). Of the 29,274 starting particles, roughly two thirds (19,128) MT particles corresponding to the 14 protofilament symmetry were selected for further processing. MTs were then refined using RELION helical processing (He and Scheres, 2017). Initially, the asymmetric unit was defined as one full 82 Å repeat (consisting of 13 tubulin dimers), using an initial estimate of zero for the helical twist. Local symmetry searches were performed during the refinement to optimize these parameters. Following refinement, the particle coordinates were smoothed as previously described (Huehn et al., 2018). After MT refinement, an additional protofilament refinement step was performed in an attempt to increase resolution of the final volume by correcting for distortions in the MT lattice. To do this, a wedge mask is applied to the final MT volume, resulting in a MT missing a single protofilament. This volume was then rotated and subtracted thirteen times from each experimental image to generate a stack of protofilament particles, with one particle for every tubulin dimer in the imaged filament. In this case, 267,792 protofilament particles were obtained from the original stack of 19,128 MT particles. Protofilament particles alignment parameters were initialized using Euler angles derived from the MT refinement step and subjected to further, local refinement using RELION. The final resolution was computed using the RELION post-processing module with a soft-edged mask. A more detailed description of the protofilament refinement method is currently being prepared for publication (Debs et al. manuscript in preparation). In order to more accurately estimate the resolution of each region of the reconstructed density, a local resolution calculation was performed using the ‘blocres’ function in the Bsoft processing package (Heymann and Belnap, 2007). This analysis revealed that the majority of the tubulin density is in the range of 3.5–4.5 Å, while the kinesin portion ranges from 5 to 6 Å resolution and tail density is around 8 Å resolution (Figure 2—figure supplement 1A-C). Model building was performed using the programs Coot and UCSF chimera using the kinesin-5 structural model (Turner et al., 2001). All the ligands were included in the model except the taxol molecule. The model was adjusted using the secondary structure elements in the density maps and analyzed for clash score using Coot and Molprobity (Table 3). The maps were compared using the programs UCSF chimera and Coot to determine the transitions of various elements.

Kinesin-5 MT stimulated motility was reconstituted as follows: Flow chambers were assembled from N 1.5 glass coverslips (0.16 mm thick; Ted Pella) that were cleaned with the Piranha protocol and functionalized with 2 mg/mL PEG-2000-silane containing 2 μg/mL biotin-PEG-3400-silane (Laysan Bio) suspended in 80% at pH 1 (Henty-Ridilla et al., 2016). After the flow chamber was assembled, 0.1 mg/mL NeutrAvidin (Thermofisher) was used to functionalize surfaces. Biotin and Alexa-Fluor-633-labeled porcine tubulin were generated in the laboratory as described (Al-Bassam, 2014) and were polymerized using the non-hydrolysable GTP analog guanosine-5’-[(α,β)-methyleno] triphosphate (termed GMPCPP; Jena Biosciences) or using the MT stabilizing drug, Paclitaxel (sigma). These MTs (100–200 μg/mL in BRB-80: 80 mM PIPES, 1 mM MgCl₂ and 1 mM ETGA; pH 6.8, 1% glycerol, 0.5% pluronic-F127, 0.3 mg/ml casein, 3 mM BME, 4 mM ATP-MgCl2) were flowed into chambers and attached to glass via biotin-neutravidin linkage. Flow chambers were then extensively washed with imaging buffer (25 mM HEPES, 25–100 mM KCl, pH 7.5, 10 mM beta-mercatopethanol; 1% glycerol, 0.5% Pluronic-F127, 0.3 mg/ml casein, 3 mM BME, 4 mM ATP-MgCl2). Kinesin-5 MT-stimulated motility was reconstituted at 25°C by injecting 1–20 nM FL-Eg5-GFP combined with a photobleach-correction mixture into flow chambers (Telley et al., 2011). Movies were captured in TIRF mode using a Nikon Eclipse Ti microscope using 1.5 Na objective and an Andor IXon3 EM-CCD operating with three (488 nm, 560 nm and 640 nm) emission filters using alternating filter wheel in 2 s increments operated using elements software (Nikon).

To study MT sliding activities in vitro, flow chambers were prepared as described above and either Paclitaxel or GMPCPP stabilized AlexaF-633 and biotin labeled MTs were anchored along their surface via Biotin-Neutravidin linkage. A mixture of 1–20 nM FL-Eg5-GFP or 20–200 nM Eg5-Δtail-GFP were mixed with 100–200 μg/ml AlexaF-560 labeled MTs and injected into these flow chambers, after being equilibrated with imaging buffer. Imaging was initiated as described above almost immediately and areas of MT sliding events were identified through search. At 3–20 nM FL-Eg5-GFP robust free MT crosslinking (yellow) was observed, followed by zippering along anchored MT (red) and then MT sliding was consistently observed (Figure 4B, Videos 2–3). In contrast, 3–20 nM Eg5-Δtail-GFP MT crosslinking was often observed but free-MTs (yellow) did not zipper along the anchored MT (red) and remain in scissoring motion for extensive periods (Figure 4B).

For MT sliding and kinesin-5 spiking studies, MT sliding experiments were performed as described above with the exception of using 20 nM FL-Eg5. Eg5-GFP spiking was carried out by the addition of either 1 nM FL-Eg5-GFP or 1 nM Eg5-Δtail-GFP with 100–200 μg/ml AlexaF-560-MTs in imaging buffer conditions. TIRF Imaging was carried out as described above.

Image movie stacks were preprocessed with photobleach correction and image stabilization plugins using the program FIJI (Schindelin et al., 2012). For motility along individual MTs, individual FL-Eg5-GFP or Eg5-Δtail-GFP motor motility events were identified along anchored MTs based on kymographs in generated for multiple channels. The FIJI plugin, trackmate, (Schindelin et al., 2012) was used to measure particle motility rates and identify their run lengths. Large collections of motile events for FL-Eg5-GFP or Eg5-Δtail-GFP conditions were collected for 25, 50, and 100 mM KCl conditions (Table 4). Average MT parameters were determined by frequency binning the motility events in a range conditions and then fitting these events using Gaussian distributions using the program Prism (Table 4). In general, all parameters fit single Gaussian distributions. Run lengths were fitted using exponential decay to identify the half-length for each motor condition. T-tests were performed to determine significance of the differences observed.

For motor fluorescence intensity quantifications, kymographs were manually analyzed using the line tool in FIJI, a line was placed over the initial signal of an individual Eg5 molecule and an intensity profile was generated and recorded in Microsoft Excel. The line was extended to include an area of the kymograph where a fluorescent signal was absent in order to measure the local surrounding background signal. This background measurement was subtracted from the initial fluorescence intensity of the molecules signal in Microsoft Excel. Only molecules that were observed to have landed on the MT during the observation period, and that were motile were used for quantification. The intensity data were frequency binned and Gaussian fit using Prism.

For MT sliding and MT sliding spiking assays, image analysis was carried out as described above with the exception of visualizing the free-MT sliding motility with respect to the anchored MT using 560 nm emission channel. The motility patterns FL-Eg5-GFP or Eg5-Δtail-GFP motor particles were studied with respect to sliding zone (along both the free and anchored MT) using the FIJI plugin, Trackmate, to determine motor velocities and their changes in motility direction inside or outside the sliding zone.

To study Eg5 MT sliding forces in optical trapping, flow chambers were prepared as described above and previously (Shimamoto et al., 2015). Paclitaxel stabilized Hilyte-649 and biotin labeled MTs were anchored along their surface via Biotin-Neutravidin attachment. Polystyrene beads were coated with kinesin-1 nucleotide-free mutant and linked to Rhodamine labeled MTs (bead attached free-MTs). These bead attached free-MTs were then mixed with 1–20 nM FL-Eg5-GFP or 1–500 nM Eg5-Δtail-GFP and injected into these flow chambers, after being equilibrated with imaging buffer. The beads attached free MTs were observed to interact with the anchored MTs and locked into the optical trap to measure the forces. Generally, 3–10 nM FL-Eg5-GFP was sufficient to observe MT sliding events, while 3–10 nM Eg5-Δtail-GFP rarely produced sliding events, and mostly crosslinked without zippering into sliding zones forming scissoring events. At 200–500 nM Eg5-Δtail-GFP, we observed sufficient MT sliding events. For each event, the length of the sliding zone, the total Eg5-GFP intensity, and plateau pushing forces developed were measured and used for scaled comparisons (Figure 5B). The Eg5-GFP intensity scaled linearly with the MT overlap MT sliding zone length without significant difference between FL-Eg5-GFP or Eg5-Δtail-GFP conditions, despite the difference in the concentration using in the assay (Figure 5D). The plateau forces generated by FL-Eg5-GFP scaled linearly with the size of the overlap MT sliding zone length, with seven-fold difference in the slope of the same comparison for Eg5-Δtail-GFP data.

HeLa cells were cultured in Minimal Essential Media-α (Gibco) supplemented with 10% fetal bovine serum (Gibco). HeLa cells were authenticated by STR genotyping by the Vermont Integrative Genomics Resource. Cells were maintained at 37°C with 5% CO₂. Transient transfections of plasmid DNA were performed via electroporation using a Nucleofector 4D system, pulse code CN114, and Cell Line SE reagents (Lonza). Cells were plated onto 12 mm glass coverslips (Electron Microscopy Sciences) for fixed cell immunofluorescence, 4-chamber 35 mm glass-bottom dishes (Greiner Bio-One) for live cell imaging, or 60 mm polystyrene tissue culture dishes for lysate collection.

To assess the protein levels in vivo, cells were arrested in 100 μM monastrol (Selleckchem) overnight and lysed in PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 4 mM MgSO₄) with Halt Protease and Phosphatase Inhibitor cocktail (Thermo-Fisher) on ice. Lysates were extracted on ice for 10 min and centrifuged at 21,130 x g for 10 min. An equal volume of 4X Laemmli buffer (Bio-Rad) was added to the supernatant and samples were heated to 95°C for 10 min. Lysates were separated by electrophoresis on 4–15% Tris-glycine polyacrylamide gels (Bio-Rad) and transferred to polyvinylidene fluoride membranes (Bio-Rad). Membranes were blocked in Odyssey blocking reagent (LI-COR) diluted 1:1 in tris-buffered saline for 1 hr, incubated with rabbit anti-mCh (diluted 1:1000, AbCam) and mouse anti-GAPDH (diluted 1:10,000, Thermo-Fisher) primary antibodies overnight, and incubated with IRDye 800- and IRDye 680-tagged fluorescent secondary antibodies (LI-COR) for 1 hr. Blot fluorescence was imaged using an Odyssey CLx system (LI-COR) and analyzed using Image Studio Lite (LI-COR).

Fixed and live cell imaging was performed using a Nikon Ti-E inverted microscope controlled by NIS Elements software (Nikon Instruments) with a Plan APO 60X/1.42 NA oil immersion objective or APO 100X/1.49 NA oil immersion objective (Nikon Instruments), Spectra-X light engine (Lumencore), and Clara CCD camera (Andor). Image processing was performed using NIS Elements (Nikon Instruments) and ImageJ (NIH). Data analysis and statistical comparisons were performed using Excel (Microsoft) and Prism (GraphPad Software).

For assessment of mCh and Eg5-mCh expression levels and localization at metaphase in fixed HeLa cells, cells were treated with 20 μM MG132 (Selleckchem) 2 hr prior to fixation. Cells were fixed for 10 min in 1% paraformaldehyde (Electron Microscopy Sciences) in ice-cold methanol (Thermo-Fisher). Cells were blocked using 20% goat serum in antibody-diluting buffer (AbDil, 1X tris-buffered saline with 2% bovine serum albumin, 0.1% Triton-X 100, and 0.1% sodium azide) for 1 hr, incubated with mouse anti-α-tubulin primary antibodies (DM1a, Sigma-Aldrich, diluted 1:750 in AbDil) for 1 hr, and incubated in fluorescent secondary antibodies conjugated to Alexa Fluor 488 or 647 (Life Technologies, diluted 1:500 in AbDil) for 1 hr. Cells were mounted in ProLong Gold with DAPI (Thermo-Fisher). Expression levels of mCh-tagged proteins were compared by drawing elliptical regions of interest (ROIs) around mitotic cells using α-tubulin staining, measuring mCh fluorescence intensity within the cellular ROIs, and subtracting averaged intensity from two background ROIs containing no visible cells. Localization was assessed by defining an ROI as the spindle based on α-tubulin staining and an ROI as cytoplasm by subtracting this spindle ROI from an ellipse that encompassed the cell. Intensity of mCh signal was measured in both the spindle and cytoplasm ROIs, and a ratio of spindle to cytoplasm intensity calculated.

For live cell imaging, growth media was exchanged for CO₂-Independent Media (Gibco) supplemented with 10% fetal bovine serum (Gibco) and penicillin/streptomycin (Gibco). For assessment of localization after treatment with BRD-9876 (Tocris Bioscience), HeLa cells were treated with 20 μM MG132 (Selleckchem) for 2 hr prior to imaging. Cells were imaged prior to drug addition, 1 min after addition of 5 μM BRD-9876, and subsequently once every 5 min. For assessment of localization in anaphase, cells in metaphase were identified and imaged at 2 min intervals through anaphase. For both live cell assays, localization of proteins to the spindle was quantified as described for fixed cell imaging, with the spindle ROI defined by GFP-tubulin signal.