The mitotic kinesin-14 KlpA contains a context-dependent directionality switch

Kinesin-14s are commonly known as nonprocessive minus end-directed microtubule motors that function mainly for mitotic spindle assembly. Here we show using total internal reflection fluorescence microscopy that KlpA—a kinesin-14 from Aspergillus nidulans—is a context-dependent bidirectional motor. KlpA exhibits plus end-directed processive motility on single microtubules, but reverts to canonical minus end-directed motility when anchored on the surface in microtubule-gliding experiments or interacting with a pair of microtubules in microtubule-sliding experiments. Plus end-directed processive motility of KlpA on single microtubules depends on its N-terminal nonmotor microtubule-binding tail, as KlpA without the tail is nonprocessive and minus end-directed. We suggest that the tail is a de facto directionality switch for KlpA motility: when the tail binds to the same microtubule as the motor domain, KlpA is a plus end-directed processive motor; in contrast, when the tail detaches from the microtubule to which the motor domain binds, KlpA becomes minus end-directed.


Introduction
The mitotic spindle is a microtubule-based bipolar machine in eukaryotes that separates duplicated chromosomes to ensure that daughter cells each receive proper genetic material during cell division. Several different kinesin motor proteins are orchestrated inside the mitotic spindle for its assembly and maintenance 1,2 . Of all mitotic kinesins, kinesin-14s are commonly considered to be nonprocessive minus end-directed microtubule motors [3][4][5][6][7][8][9][10][11][12] . While mitotic kinesin-14s are nonessential for normal cells, loss of the kinesin-14 Pkl1 in fission yeast Schizosaccharomyces pombe has been shown to cause erroneous chromosome segregation 13 . In cancer cells, the human kinesin-14 HSET/KIFC1 is needed for clustering multiple centrosomes, a process crucial for cancer cell proliferation and survival 14 .
KlpA is a mitotic kinesin-14 from the filamentous fungus Aspergillus nidulans 15 . It is worth noting that A. nidulans is also the model organism for the discovery of BimC, the founding member of mitotic kinesin-5s 16 . Like mitotic kinesin-14s in other eukaryotic cells 11,17,18 , KlpA counteracts the function of BimC 15 . Similar to the fission yeast kinesin-14 Pkl1 19 , KlpA is nonessential in wildtype cells but its loss becomes synthetically lethal with gamma tubulin mutations 20 . KlpA is an attractive model protein for dissecting the mechanism and function of kinesin-14s, as its loss-of-function mutations can be conveniently isolated as suppressors of the bimC4 mutation 21 . However, compared with other mitotic kinesin-14s such as Ncd from Drosophila melanogaster and Kar3 from Saccharomyces cerevisiae, KlpA is much less well studied.
In this study, we report our in vitro characterization of KlpA motility using total internal reflection fluorescence (TIRF) microscopy. KlpA unexpectedly moves processively toward the plus ends on individual microtubules as a single homodimer and switches to the canonical minus end-directed motility inside microtubule bundles. Thus, KlpA is a context-dependent bidirectional kinesin-14, making it distinct from all other kinesin-14s that have been examined to date. Furthermore, our results suggest that KlpA contains an N-terminal nonmotor microtubule-binding domain that not only enables the motor for plus end-directed processive motility but also acts a switch for controlling its directionality in different cellular contexts.
These findings shed new light on KlpA motor mechanism and provide a molecular view of how KlpA may be regulated for mitotic spindle assembly and maintenance.

KlpA glides microtubules with canonical minus end-directed motility
We set out to determine the directionality of KlpA in vitro using TIRF microscopy. To that end, we purified the recombinant full-length KlpA tagged with an N-terminal green fluorescent protein (GFP-KlpA, Fig. 1a, b). Since KlpA substitutes for Kar3 in S. cerevisiae 15 and Kar3 forms a heterodimer with the nonmotor proteins Cik1 or Vik1 22 , we performed two different assays -hydrodynamic analysis and single-molecule photobleaching -to determine the oligomerization status of KlpA. The hydrodynamic analysis yielded a molecular weight that is close to the theoretical value of a GFP-KlpA homodimer ( Supplementary Fig. 1a, b). The photobleaching assay showed that the GFP fluorescence of GFP-KlpA was predominantly photobleached in a single step or two steps ( Supplementary Fig. 1c,  We next performed a microtubule-gliding assay to determine the directionality of KlpA (Fig. 1c). Briefly, GFP-KlpA molecules were immobilized on the coverslip via an N-terminal polyhistidine-tag, and KlpA directionality was deduced from the motion of polarity-marked microtubules. The assay showed that GFP-KlpA caused polarity-marked microtubules to move with the bright plus ends leading ( Fig. 1d and Supplementary Video 1). In a control experiment using the plus end-directed human conventional kinesin hKHC 26 , microtubules were driven to move with the bright plus ends trailing (Supplementary Fig. 2 and Supplementary Video 2).
Taken together, these results demonstrate that KlpA, anchored on the surface via its N-terminus, is a minus end-directed motor protein, in agreement with a previous study using KlpA from clarified bacterial lysates 20 .

Single KlpA molecules move processively toward the plus ends on individual microtubules
We wanted to determine whether KlpA is a typical kinesin-14 that lacks the ability to move processively on individual microtubules as a single homodimer. To address this, we performed an in vitro motility assay to visualize the movement of KlpA molecules on surfaceimmobilized polarity-marked microtubules (Fig. 2a). The assay was first performed at relatively high input levels of GFP-KlpA (≥ 4.5 nM). Contrary to the notion of kinesin-14s as minus enddirected motors, GFP-KlpA molecules unexpectedly formed a steady flux to accumulate at the microtubule plus ends (yellow arrow, Fig. 2b and Supplementary Video 3). Occasionally, there were GFP-KlpA particles moving toward the microtubule minus ends (white arrow, Fig. 2b), but these minus end-directed particles were significantly brighter than the ones moving toward the plus ends, implying that they were aggregates rather than simple homodimers. Since GFP-KlpA appeared to move processively toward the microtubule plus ends (Fig. 2b), we repeated the in vitro motility assay at lower protein input levels (≤ 0.2 nM) so that the motile behavior of individual GFP-KlpA molecules could be distinguished. The assay showed that individual GFP-KlpA molecules moved preferentially toward the microtubule plus ends in a processive manner ( Fig. 2c and Supplementary Video 4) with a mean velocity of ~320 ± 90 nm/s (mean ± s.d., n = 249, Fig. 2d) and a characteristic run-length of 8.8 ± 0.2 µm (mean ± s.e., n = 249, Fig. 2e). This run-length likely was an underestimate, as most KlpA molecules reached the microtubule plus ends. Together, these results demonstrate that KlpA, in direct contrast to all other kinesin-14s examined to date, is a processive plus end-directed kinesin.

KlpA exhibits opposite directional preference inside and outside microtubule overlaps
From the opposite directional preference exhibited by GFP-KlpA in the ensemble microtubule assay (Fig. 1d and Supplementary Fig. 3c) and the single-molecule motility experiments (Fig. 2b, c), we inferred that KlpA contains a context-dependent mechanism to switch directions on the microtubule 32 . We thus directly compared the motility of GFP-KlpA inside and outside the microtubule overlap on the same track microtubule using a microtubuletransport assay (Fig. 4a), as has been done previously for S. cerevisiae kinesin-5 Cin8 32 . Briefly, in this assay the track (blue) and cargo (red) microtubules were both polarity-marked but labeled with different dyes; track microtubules were first immobilized on a coverslip inside the motility chamber and bound with purified GFP-KlpA molecules; and cargo microtubules were added into the chamber before three-color time-lapse imaging was acquired to simultaneously visualize the motility of GFP-KlpA molecules and cargo microtubules on the same track microtubules. Like KlpA, GFP-KlpA was also able to slide antiparallel microtubules relative to each other (Fig. 4b) and to statically crosslink parallel microtubules (Fig. 4c). In both scenarios, when outside the microtubule overlap regions, GFP-KlpA molecules showed a plus enddirected flux and accumulated at the plus end on the track microtubule (yellow arrow, Fig. 4b, c and Supplementary Video 9 and 10). This matches the behavior of GFP-KlpA on individual microtubules (Fig. 2b). In contrast, inside the antiparallel microtubule overlap regions, GFP-KlpA molecules carried the cargo microtubule toward the minus end of the track microtubule (white arrow, Fig. 4b and Supplementary Video 9). In the parallel orientation, the cargo microtubule remained stationary on the track microtubule, but GFP-KlpA molecules moved preferentially toward and gradually accumulated at the minus end inside the parallel microtubule overlap (white arrow, Fig. 4c and Supplementary Video 10). This is similar to the observation that Ncd preferentially accumulates at the minus ends between statically crosslinked parallel microtubules 24 . Collectively, these results demonstrate that KlpA can, depending on context, display opposite directional preferences on the same microtubule: it is plus end-directed outside the microtubule overlap regions and minus end-directed inside the microtubule overlap regions regardless the relative microtubule polarity.

Discussion
Kinesin-14 has been an intriguing kinesin subfamily since the discovery of its founding member Ncd 33,34 , because all kinesin-14s studied to date are exclusively minus end-directed in the microtubule-gliding experiments 23,25,[33][34][35][36][37][38][39] . With the lone exception of Kar3, no other kinesin-14 has been shown to be able to generate processive motility directly on the surface of individual microtubules as a single homodimer. In vitro, it has been shown that Kar3 generates processive minus end-directed motility on individual microtubules by forming a heterodimer with its associated light chains Vik1 or Cik1 22,40 . By revealing KlpA as a kinesin-14 that demonstrates both processive plus end-directed motility on individual microtubules and contextdependent directional switching, our study further expands the diversity of kinesin-14s.
How does KlpA achieve the observed context-dependent directional switching? Our results show that while the full-length KlpA clearly moves processively toward the plus ends on individual microtubules (Fig. 2b, c), a truncated KlpA lacking the N-terminal nonmotor MTBD is unable to produce processive motility (Fig. 3c) but does retain the ability to glide microtubules with minus end-directed motility (Fig. 3b). There are several important implications from these observations. First, the motor core of KlpA without the nonmotor MTBD is inherently minus end-directed, which is in agreement with the notion that all kinesin-14s share a highly conserved neck linker that serves as the minus end directionality determinant 26,[29][30][31] . Second, the nonmotor MTBD is required for plus end-directed KlpA motility on individual microtubules. We suggest that the nonmotor MTBD is a de facto switch for controlling the direction of KlpA motility: KlpA is plus end-directed kinesin-14 motor when the switch-like nonmotor MTBD and the motor domain both bind to the same microtubule, and it reverses to become a nonprocessive minus end-directed motor when the switch-like nonmotor MTBD is detached from the microtubule to which its motor domain binds. This could explain the minus end-directed motility of KlpA anchored on the coverslip via the N-terminus (Fig. 1d) or inside microtubule bundles (Fig. 4b, c), because in both cases the switch-like nonmotor MTBD is in effect detached from the microtubule to which its motor domain binds. Future studies will need to determine the structural basis of how the nonmotor MTBD enables KlpA to move with plus end-directed processive motility. We speculate that positioning of the nonmotor MTBD relative to the motor domain on the microtubule may favor KlpA to search the next binding site between steps toward the microtubule plus ends.
Our findings provide a molecular view for how KlpA motility may be regulated inside the mitotic spindle (Fig. 5). While other mitotic kinesin-14s appear to depend on partner proteins to localize to the spindle midzone for antagonizing the action of kinesin-5s 10,37,41,42 , KlpA can in principle autonomously localize to the spindle midzone via its inherent plus end-directed motility by having both the nonmotor MTBD and the motor domain on the same microtubule (Fig. 5a). Inside the antiparallel microtubule overlaps at the spindle midzone (Fig. 5b) or the parallel microtubule overlaps near the spindle poles (Fig. 5c), KlpA switches to become minus end-directed as the switch-like nonmotor MTBD and the motor domain bind to two different microtubules. This apparent directional plasticity suggests that other proteins could exist to regulate KlpA motility via intermolecular interactions that interfere with the binding of the switch-like nonmotor MTBD to microtubules. A recent study shows that Pkl1 -a mitotic kinesin-14 from the fission yeast -forms a complex with Msd1 and Wdr8 for translocating to and anchoring at the spindle poles 43 . The homologs of both Msd1 and Wdr8 are also present in A. nidulans 44 . Thus, it is plausible that binding of Msd1 and Wdr8-like proteins to KlpA could dislodge its N-terminal nonmotor MTBD from the surface of microtubules to activate the kinesin for minus end-directed motility both on individual microtubules (Fig. 5d) and at the spindle poles (Fig. 5e).
Several mitotic kinesin-5s were recently shown to be context-dependent bidirectional motor proteins 32,[45][46][47] , suggesting that context-dependent directional switching likely is evolutionarily conserved among kinesin-5s. Our current work on KlpA provides the first evidence to suggest that context-dependent directional switching could also exist among some, if not all, mitotic kinesin-14s. The mechanism and regulation of bidirectional mitotic kinesins will be an important subject for future studies.

Methods
Detailed methods are described in Supplementary Information.

Author Contributions
W.Q. conceived, designed and supervised the study; P.A.K. and X.X. provided conceptual suggestions; A.R.P. and K.-F.T. performed the experiments; K.-F.T. and P.W. contributed all KlpA constructs. All authors participated in discussing the results. W.Q. wrote the manuscript with input from all authors.

Author Information
The authors declare no competing financial interests. Correspondence and requests for materials should be addressed to Weihong.Qiu@physics.oregonstate.edu.