Identification of key residues that regulate the interaction of kinesins with microtubule ends

Abstract Kinesins are molecular motors that use energy derived from ATP turnover to walk along microtubules, or when at the microtubule end, regulate growth or shrinkage. All kinesins that regulate microtubule dynamics have long residence times at microtubule ends, whereas those that only walk have short end‐residence times. Here, we identify key amino acids involved in end binding by showing that when critical residues from Kinesin‐13, which depolymerises microtubules, are introduced into Kinesin‐1, a walking kinesin with no effect on microtubule dynamics, the end‐residence time is increased up to several‐fold. This indicates that the interface between the kinesin motor domain and the microtubule is malleable and can be tuned to favour either lattice or end binding.

classic cargo carriers. The translocating activity and cargo carrying function of the Kinesin-1 family does not require the ability to recognise the microtubule end and this ability has not been detected for a Kinesin-1. Here we show that the α4-helix of the kinesin motor domain is a critical region in regulating the balance between microtubule lattice and microtubule end binding. Substitution of  family specific residues from the α4-helix into equivalent positions in a Kinesin-1 confers the ability to distinguish the microtubule end from the lattice, such that the microtubule end-residence times of Kinesin-1 mutants are increased several-fold.
2 | RESULTS 2.1 | Substitution of Kinesin-13 residues into a Kinesin-1 motor domain increases microtubule endresidence time A key characteristic of kinesins that regulate microtubule dynamics is to recognise the microtubule end as distinct from the microtubule lattice. This ability enables regulating kinesins to reside for extended times at microtubule ends and places them at the correct location to influence microtubule growth and/or shrinkage dynamics. The α4-helix of the motor domain plays a major role in the interface between the kinesin motor domain and the microtubule. Three residues from the α4-helix of the microtubule depolymerising Kinesin-13, MCAK, (K524, E525 and R528) are critical to its ability to reside at the microtubule end and essential for depolymerase activity (Patel et al., 2016). To determine if these residues can increase microtubule end binding for a purely translocating kinesin, we substituted the Kinesin-13 residues into a Kinesin-1, rkin430. The Kinesin-1 motor domain has not been shown to recognise the microtubule end and accordingly, we measured an end-residence time for rkin430 of ≤0.46 ± 0.01 s, which is over four-fold shorter than MCAK (Figure 1a).
The residues critical to MCAK's microtubule end recognition ability, K524, E525 and R526, correspond by sequence alignment to the rkin430 residues G262, N263 and S266 (Figure 1b) and the structures of the Kinesin-1 and Kinesin-13 motor domains confirmed the spatial correspondence of these residues (Figure 1c). We therefore created the rkin430 variants G262K, N263E and S266R to substitute these Kinesin-13 family specific residues into a Kinesin-1. We also made a triple mutant containing all three individual mutations.  Figure S1a). Increased residence times resulting from these substitutions are specific to the microtubule end: lattice residence was either decreased or unaffected for each of these kinesin-1 variants (Supporting Information Table S1). The substitution S266R had the largest effect on end-residence time, causing a ≥ three-fold increase relative to wild-type Kinesin-1. Kymographs show that S266R stays on the microtubule end for an average of 5 frames

| Increased end residence is not due to misfolding or disruption of translocation activity and is specifically favoured by Kinesin-13 residues
To determine whether the amino acid substitutions had a deleterious effect on the functionality of the Kinesin-1 motor domain, we measured the ability of each mutant to turnover ATP. The basal ATPase rate was not significantly different to wild type for any of the mutants (Table 1) and there was no correlation between ATPase rate and microtubule end-residence time (Supporting Information Figure S1b).
The ability of the Kinesin-1 mutants to turnover ATP at an equivalent rate to wild type indicates that the motor domain is correctly folded and functional.
Each of the amino acid substitutions caused a moderate reduction in translocation velocity (Table 1). However, there was no relationship between translocation velocity and end-residence time ( Figure 2c). This indicates that the increased end-residence times do not simply result from slower translocation velocities causing the mutants to take longer to move over the microtubule end. Taken together, these data indicate that substitution of Kinesin-13 specific residues into the α4 helix of Kinesin-1 caused a specific increase in microtubule endresidence time.
To establish whether the observed increased end-residence times were specifically due to introduction of Kinesin-13 residues rather than loss of Kinesin-1 residues, we created the rkin430 variant S266A. The microtubule end-residence time for this variant was increased relative to wild-type Kinesin-1 (Table 1 and Figure 2b) but not to the same degree as the increase caused by introduction of the Kinesin-13 family specific residue at this position. The end residence of S266A was 0.9 s compared to 1.4 s for the Kinesin-13 specific substitution, S266R. These data suggest not only that the Kinesin-13 residue at this position specifically favours microtubule end binding over lattice binding, but also that the Kinesin-1 residue disfavours binding at the microtubule end.

| Increased microtubule end residence alone does not confer depolymerase activity
The ability to recognise the microtubule end is a property that dis-

| DISCUSSION
Despite high sequence and structural conservation of the superfamily defining motor domain, kinesins from different families display a T A B L E 1 Compiled data for microtubule end residence, velocity, run length, ATPase rate and depolymerisation activity for WT-rkin430 and variants  The Kinesin-13 family specific residues K524, E525 and R528 are not only charged but also larger than the equivalent Kinesin-1 residues. The impact of this is to create a bulkier α4-helix (Figure 3a,b) in the Kinesin-13 motor domain relative to Kinesin-1. The Kinesin-1 α4-helix forms a cone-like shape, with the diameter at the C-terminal end being smaller than at the N-terminal end (Figure 3a), whereas the Kinesin-13 α4-helix is more cylindrical due to the presence of these bulkier family specific residues at the C-terminus (Figure 3b).
Structures of the kinesin motor domain in complex with tubulin  (Gigant et al., 2013;Wang et al., 2017) show that the α4-helix contacts the α/β-tubulin heterodimer at the interface between the α and β subunits, the so-called "intradimer groove" (Figure 3c,d). The bulky nature of the Kinesin-13 residues may enhance sensing of tubulin curvature by responding to the reduction of space at the interface between the α and β subunits in a curved tubulin conformation such as that proposed to form at the microtubule end compared to a straight tubulin conformation found within the microtubule lattice.
Whilst introduction of the Kinesin-13 residues into the Kinesin-1 motor domain increased microtubule end-residence time relative to wild type, none of the variants had significant depolymerisation activity. The end-residence times for the Kinesin-1 variants are shorter than for the specialist microtubule depolymerising kinesin, MCAK, and it is possible that a longer microtubule end residence is required for depolymerase activity. However, it is more likely that structural elements in addition to the α4 helix, such as the Kinesin-13 specific extended Loop 2, are also required for depolymerisation activity (Ogawa, Nitta, Okada, & Hirokawa, 2004;Patel et al., 2016;Shipley et al., 2004). It is possible that the α4-helix is principally a curvature sensing region, which is required to bind at the microtubule end, whilst the region of the Kinesin-13 motor domain that actively removes tubulin causing depolymerisation is located elsewhere.

| Single molecule TIRF assays
Single molecules of rkin430-GFP and variants were observed on immobilised, GMPCPP-stabilised, 25% rhodamine-labelled microtubules in BRB12 (12 mM PIPES/KOH pH 6.9, 1 mM EGTA, 2 mM MgCl 2, ) plus 1 mM ATP, 0.1% Tween 20, 0.1 mg/ml BSA and antifade (1% 2-mercaptoethanol, 40 mM glucose, 40 mg/ml glucose oxidase, 16 mg/ml catalase) using TIRF microscopy (Patel et al., 2016). Images were collected at a frame rate of 2.7 Hz. Time spent by single kinesin molecules at the microtubule end and on the microtubule lattice was measured in FIJI (Schindelin et al., 2012) using kymographs of individual microtubules. The microtubule end was defined as the final pixel at either end of a microtubule in the rhodamine channel with a fluorescence signal above the background subtracted threshold. An event was defined as a pixel containing fluorescence intensity above background in the GFP channel. Events were considered discrete when separated by ≥1 non-event pixel in either the vertical (time) or horizontal (distance) axis. Translocating events were defined as events which moved in a unidirectional manner for at least 3 pixels in the vertical (>750 ms). When events crossed and it was not possible to identify individual events, they were discounted from the analysis. All other events were captured and classified.

| ATPase assays
Reactions were initiated by addition of 1 μM Kinesin to BRB12 buffer containing 2 mM ATP and samples taken every 5 min, quenched with an equal volume of ice-cold 0.6 M perchloric acid, neutralised with Tris/KOH and clarified by centrifugation. The progression of the reaction was monitored by separating ADP from ATP by HPLC (Friel, Bagshaw, & Howard, 2011).

| Depolymerisation assays
GMPCPP-stabilised, rhodamine-labelled microtubules were incubated with 40 nM kinesin and 1 mM ATP in BRB12 for 20 min. The reactions were then flowed into channels made from poly-lysine coated cover glasses and imaged. Microtubule depolymerisation was also monitored by light scattering at 350 nm. Kinesin protein was added to microtubules at final concentrations of 50 nM and 1 μM, respectively, in the buffer BRB80 pH 6.9, 75 mM KCl, 1 mM MgATP, 1 mM DTT, 200 μg/ml BSA. The light scatter signal was recorded at 5 s intervals using a Hitachi F2500 fluorimeter.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.