Tyrosines in the Kinesin-5 head domain are necessary for phosphorylation by Wee1 and for mitotic spindle integrity

SUMMARY

Mitotic spindle assembly and maintenance relies on Kinesin-5 motors that act as bipolar homotetramers to cross-link microtubules [1–5]. Kinesin-5 motors have been subject to extensive structure-function analysis [5], but the regulation of their activity in the context of mitotic progression remains less well understood [2]. We report that Drosophila Kinesin-5 (KLP61F) is regulated by Drosophila Wee1 (dWee1). Wee1 tyrosine kinases are known to regulate mitotic entry via inhibitory phosphorylation on Cdk1 [6–10]. Recently, we showed that dWee1 also plays a role in mitotic spindle positioning through γ-tubulin and spindle fidelity through an unknown mechanism [11]. Here we investigated whether a KLP61F-dWee1 interaction could explain the latter role for dWee1. We found that dWee1 phosphorylates KLP61F in vitro on three tyrosines within the head domain, the catalytic region that mediates movement along microtubules. In vivo, KLP61F with tyrosine-to-phenylalanine mutations fails to complement a klp61f mutant and dominantly induces spindle defects similar to ones seen in dwee1 mutants. We propose that phosphorylation of the KLP61F catalytic domain by dWee1 is important for the motor’s function. This study identifies a second substrate for a Wee1 kinase and provides evidence for phospho-regulation of a kinesin in the head domain.

RESULTS AND DISCUSSION

dWee1 and KLP61F interact at endogenous protein levels

Previously, we identified KLP61F in mass spectrometric analysis of HA-dWee1-containing protein complexes [11] (data not shown). To confirm that dWee1 and KLP61F interact at physiological levels, we assayed for co-immunoprecipitation of endogenous proteins (Figure 1). The specificity of a previously described anti-KLP61F serum [12] and a newly generated anti-dWee1 antibody were first confirmed with western blots of extracts from respective mutants (Figure S1). The anti-dWee1 antibody recognized a doublet (Figure S1C), similarly seen for human Wee1 [13]. This doublet was present in immunoprecipitates obtained with the anti-KLP61F serum (Figure 1). Also, the anti-dWee1 antibody immunoprecipitated dWee1 and co-precipitated KLP61F from syncytial embryos (data not shown).

Figure 1. Endogenous KLP61F and dWee1 interact.

Extracts from syncytial wild type embryos were immunoprecipitated with anti-KLP61F serum or mock-precipitated (no Ab IP) and western blotted for KLP61F (A) or dWee1 (B). Samples equivalent to 0.6% and 1.2% of starting extracts were loaded for comparison.

GST-dWee1 phosphorylates His-KLP61F within the head domain

Because dWee1 is a tyrosine kinase, we next investigated whether dWee1 could phosphorylate KLP61F. Incubation of recombinant, purified His-KLP61F with GST-dWee1 resulted in phosphorylation (Figure 2B) and recognition of the former by an anti-phosphotyrosine antibody (Figure 2C). Autophosphorylation by dWee1 occurred as expected [14, 15]. Mass spectrometry was used to identify phosphorylated peptides of His-KLP61F after GST-dWee1 kinase assays. We achieved 57% coverage of KLP61F and identified 4 phosphopeptides containing a single tyrosine (see Experimental Procedures in Supplemental Data). Three of these peptides are in the head domain and contain Y23, Y152, or Y207 as their single tyrosine (Figures S1A–S1C). The fourth is in the BimC box, a conserved region of ~20 amino acids in the tail domain, and contains Y927. To confirm that these regions of KLP61F are important for phosphorylation by dWee1, we generated purified polypeptides containing either the head domain (His-Head^WT) or the BimC box (His-BimC; Figure 2A). Both were phosphorylated by GST-dWee1, but only His-Head^WT was recognized by an anti-phosphotyrosine antibody after phosphorylation by GST-dWee1 (Figures 2D–2E). Therefore, we focused our analysis on the head domain.

Figure 2. GST-dWee1 phosphorylates KLP61F on tyrosines in the head domain in vitro.

(A) Schematic representation of KLP61F with head domain (red), coiled-coil (blue) and tail domain that contains the BimC box (yellow). His-tagged polypeptides used in kinase assays are depicted. (B) His-KLP61F was incubated with varying amounts of GST-dWee1 in kinase reactions in vitro and analyzed for ³²P incorporation. (C) Reaction with equimolar GST-dWee1 and His-KLP61F was western blotted with anti-phosphotyrosine antibody after in vitro kinase assays. (D–E) GST-dWee1 and His-KLP61F polypeptides were incubated in in vitro kinase reactions and analyzed for ³²P incorporation (left panels) and subsequently by western blots for phosphotyrosines (right panels). (F) KLP61F head domain with Y23, Y152 and Y207 mutated to phenylalanines, His-Head^3YF, was incubated with GST-dWee1 in in vitro kinase reactions and analyzed for ³²P incorporation. none = no kinesin; GST-KHC = conventional Kinesin Heavy Chain head domain; His-Head^WT = wild type head domain. (G–H) Western blots for GST (G) or His tags (H) show equivalent amounts of substrates in kinase reactions in (F).

Tyrosine(s) in head domain are required for phosphorylation by dWee1

Mutation of the putative phospho-acceptor residues Y23, Y152, and Y207 in the head domain to phenylalanines resulted in greatly diminished phosphorylation by dWee1 (“3YF” mutant; Figures 2F and 2H). It is unlikely that these mutations resulted in protein mis-folding; mutant head domain binds ATP in preliminary studies (not shown) and, more importantly, full-length KLP61F with 3YF mutations shows activity in vivo (described in detail below). The head domain of Drosophila Kinesin-1 motor, Kinesin Heavy Chain (GST-KHC), lacks tyrosines at the corresponding residues and was also a poor substrate for dWee1 (Figure S1A–S1C, Figure 2F–2G). Thus, phosphorylation in the head domain by dWee1 appears specific to KLP61F. Of the three putative phospho-acceptor tyrosines in KLP61F, Y207 is conserved in metazoan Kinesin-5 motors (Figure S1C). Alignment of residues flanking Y207 of Kinesin-5 motors and Y15 of Cdk1 homologs from Drosophila, human and Xenopus reveal several conserved residues (Figure S1D). These may comprise a Wee1 consensus if Y207 serves as a phospho-acceptor for Wee1 in other Kinesin-5 motors.

Potential phospho-acceptor tyrosine(s) are important in vivo

klp61f³ is a well characterized loss-of-function allele with a transposon insertion in the 5’UTR that greatly reduces protein expression [16, 17] (Figure S1A). Neuroblasts from klp61f³ homozygous larvae have mitotic spindle defects, chromosome segregation failure, and polyploid nuclei that result in death at the pupal stage [16, 17]. A full-length Myc-KLP61F^WT transgene expressed constitutively from the ubiquitin promoter has been shown to rescue the lethality and cytological defects of klp61f³ homozygous larvae [16]. To determine the importance of Y23, Y152, and Y207 in vivo, we used an identical expression system to express a full-length Myc-KLP61F^3YF. Of the 5 independent transgenic lines generated, we analyzed one that expresses Myc-KLP61F^3YF at similar levels to Myc-KLP61F^WT (Figure S1A). Unlike Myc-KLP61F^WT, Myc-KLP61F^3YF did not rescue the lethality of klp61f³ homozygotes or the incidence of polyploidy in larval neuroblasts (Figures 3A–3B), suggesting that Myc-KLP61F^3YF has reduced function. These data indicate that Y23, Y152, and/or Y207 are important for KLP61F function in larval cells.

Figure 3. Phospho-accepting tyrosines in the head domain are important in vivo.

(A–B) Pupal lethality of klp61f³ mutants (A) and level of aneuploidy in klp61f³ mutant neuroblasts (B) are rescued by a Myc-KLP61F^WT transgene ([WT]; klp61f³) but not a Myc-KLP61F^3YF transgene ([3YF]; klp61f³). Data is represented as percentage of dead pupae (mean +/− SD from 3 embryo collections) in (A) and as percentage of aneuploid nuclei (mean +/− SD of nuclei from 3 larval brains) in (B). * indicates p < 0.01.

Embryos with reduced klp61f have spindle defects

Next, we asked whether Y23, Y152, and Y207 are also important for KLP61F function in embryonic syncytial cycles where dwee1 has been implicated in maintaining spindle fidelity [11]. Drosophila embryogenesis begins with 13 nuclear divisions driven by maternally contributed products. These divisions occur synchronously in a common cytoplasm called a syncytium and alternate between S and M phases without gap phases. The first nine syncytial divisions occur in the interior of the embryo (interior divisions), after which nuclei migrate outward such that cycles 11–13 occur in a monolayer directly beneath the cortex (cortical divisions). Drosophila embryos that lack maternally provided dwee1 die as syncytial embryos, consistent with other Wee1 homologs being essential for metazoan embryogenesis [18–20].

Embryos from Drosophila females that are hemizygous for the strongest extant allele of dwee1, dwee1^ES1, (to be called “dwee1 mutant embryos”) show multiple spindle defects [10, 11]. As expected, these defects included ones known to result from elevated Cdk1 activity and premature mitotic entry. Surprisingly, these defects also included two “dwee1-specific defects” that could not be explained by premature mitotic entry or bulk elevation in Cdk1 activity: multiple microtubule organizing centers (MTOCs) within a single spindle and microtubule spurs that extend to neighboring spindles leading to collisions [11].

To investigate a possible link between dWee1 and KLP61F during syncytial cycles, we wanted to compare mitotic spindles in dwee1 mutant embryos to mitotic spindles in embryos with reduced KLP61F. Homozygotes of strong klp61f mutants do not survive to adulthood [17], but a recent study showed that syncytial embryos from klp61f³ heterozygous mothers show a phenotype, namely, abnormal spindle length [1]. Therefore, we examined these embryos (to be called “klp61f embryos”) for possible additional spindle defects. We found that klp61f embryos have reduced KLP61F (Figure S3) and ~20% of spindles were abnormal (Table 1). While the incidence of spindle defects was unexpectedly high for embryos that typically survive to adulthood, the majority of these problems occurred during the interior divisions. Defective syncytial nuclei are typically culled into the interior yolk mass and do not contribute to subsequent embryogenesis [21], which may explain the viability of klp61f embryos. The most frequent defects were anastral and monopolar spindles (Figures 4B, 4C; quantified in 4D; Table 1 includes a complete listing of all spindle defects). Such defects have been reported in larval neuroblasts from klp61f mutants or embryos injected with anti-KLP61F antibodies [17, 22].

Table 1.

Percent of Spindle Defects in Syncytial Embryos

	Wild-Type	klp61f	[WT]g	[3YF]h
Interior Divisions (Cycles 1–9)	n=201	n=532	n=287	n=410
Monastrala	0.5	1.88	0	0
Anastralb	0.5	4.89 *	0.7 *	0
Promiscious MTsc	0	0.75	0.35	0.49
Multi MTOCd	1.99	3.76	5.57	11.75 *
Monopolare	0	6.77 **	0.35 **	8.29 **
TOTAL	2.99	18.05	6.97	20.53
Cortical Divisions (Cycles 11–13)	n=118	n=104	n=191	n=175
Monopolar	0.85	0	0.52	1.14
Promiscuous MTs (Meta)	0	1.83	4.19	1.71
Promiscuous MTs (Ana)	0	0	2.09	6.86 ^
TOTAL	0.85	1.83	7.78	9.71
TOTAL DEFECTS	3.84	19.88	14.75	30.24

n represents the number of spindles assayed from 30 (interior) or 3 (cortical) embryos

^abipolar spindles with only one centrosome

^bbipolar spindles with no centrosomes

^cmicrotubule spurs or interactions between spindles

^dbipolar spindles with multiple MTOCs

^emonopolar spindles

^fstatistically compared to wild type embryos

^gstatistically compared to klp61f embryos

^hstatistically compared to [WT] embryos

MT = microtubules

^**0.001>p

^*0.01>p

^{^}0.05>p

Figure 4. Mitotic spindle defects in embryos containing KLP61F^WT or KLP61F^3YF.

Embryos from mothers that are wild type, klp61f³ heterozygotes (klp61f), klp61f³ heterozygotes with Myc-KLP61F^WT ([WT]) and klp61f³ heterozygotes with Myc-KLP61F^3YF ([3YF]) were fixed and stained for DNA (blue), α-tubulin (green), and centrosomin (red). Representative images show a normal spindle (A) and defects significantly different among the genotypes: a monopolar spindle (B), an anastral spindle (C), and spindles with multiple MTOCs (F–G) from interior divisions, and a promiscuous microtubule interaction during anaphase in cortical divisions (H). Scale bar = 5 µm. Defects in (A–C) and (F–H) are quantified in (D) and (E) respectively. See Table 1 for numbers. * p < 0.01, ** p < 0.001, and ^ p < 0.05. (I) A model for regulation of KLP61F to maintain mitotic spindle integrity and prevent promiscuous MT interactions. dWee1 may regulate KLP61F activity to bundle parallel (not shown) and/or anti-parallel microtubules to create a more robust spindle (left). Without dWee1 regulation, KLP61F activity is reduced on the spindle, leading to an unstable spindle with microtubule spurs. These microtubule spurs can then interact with neighboring spindles in a syncytium (spur-spindle interaction is not depicted). Reduced KLP61F activity on the spindle is depicted as reduced protein levels on the spindle, but it remains possible that similar level of protein associate with the spindle but display reduced activity.

Potential phospho-acceptor tyrosines are important for the prevention of monopolar spindles

Restoring KLP61F with a maternal transgene encoding Myc-KLP61F^WT rescued both anastral and monopolar spindles in klp61f³ embryos (“[WT]” in Figure 4D). To date, it remains unclear what aspect of KLP61F function normally prevents anastral spindles [17]. Nonetheless, anastral spindles in klp61f³ embryos were similarly reduced when mothers carried a transgene encoding Myc-KLP61F^3YF (“[3YF]” in Figure 4D). This suggests that Myc-KLP61F^3YF is at least partially active. Consistent with this idea, Myc-KLP61F^3YF also localized to mitotic spindles similar to endogenous KLP61F and Myc-KLP61F^WT(Figure S3B) [12]. These data, together with the preliminary finding that His-Head^3YF can bind ATP in vitro (data not shown), suggest strongly that the 3YF mutations do not create a general protein-folding problem for KLP61F.

The Myc-KLP61F^3YF transgene, however, did not rescue the monopolar spindle defect in klp61f³ embryos (“[3YF]” in Figure 4D). The role of KLP61F in the prevention of monopolar spindles has been explained by its ability to antagonize minus-end directed motors [4, 22]. In the absence of KLP61F, the latter brings separated centrosomes together; this collapses the spindle from a bipolar to a monopolar structure [22]. Consistent with the idea that monopolar spindles in klp61f³ embryos result from reduced KLP61F, such defects were rescued in [WT]; klp61f³ embryos (Figure 4D). Monopolar spindles, however, were not rescued in [3YF]; klp61f³ embryos (Figure 4D). Thus, potential phospho-acceptor tyrosines in the KLP61F head domain appear to be important for this aspect of KLP61F function.

KLP61F^3YF induces spindle defects in a dominant manner

The presence of maternal Myc-KLP61F^3YF produced two additional spindle defects in klp61f embryos. These defects were also seen in dwee1 mutant embryos and are ‘dwee1-specific’, that is, independent of Cdk1 mis-regulation [11]. One defect was multiple MTOCs per spindle, as defined by staining for centrosomin (Figures 4F and 4G). The incidence of spindles with multiple MTOCs was not significantly different among embryos from wild type, klp61f³/+ and klp61f³/+ mothers carrying the Myc-KLP61F^WT transgene (“wild type”, “klp61f” and “[WT]” respectively in Figure 4E). In contrast, embryos from klp61f³/+ mothers carrying the Myc-KLP61F^3YF transgene displayed a significantly higher incidence of spindles with multiple MTOC (p <0.01) (“[3YF]” in Figures 4E; Table 1). Spindles with multiple MTOCs were also seen in dwee1 mutant embryos both during interior and cortical divisions ([11] and data not shown).

The second dwee1-specific defect induced by Myc-KLP61F^3YF in klp61f³ embryos was promiscuous microtubules, which included microtubule spurs and microtubules interacting between neighboring spindles, during cortical divisions (Figure 4H). Since embryo size remains constant, increases in nuclear density leads to crowding of spindles. During cortical divisions, cortical actin descends to form ‘furrows’ separating neighboring spindles. These actin furrows retract during metaphase and are absent in subsequent stages of mitosis, and yet spindles do not collide [23]. We had proposed that dWee1 plays a role, via phospho-regulation of γ-tubulin, in positioning nuclei/spindles within the protection of these furrows to prevent spindle collisions [11]. However, in dwee1 mutant embryos, spindles collide throughout mitosis [11], suggesting that dWee1 acts in another, furrow-independent mechanism that normally prevents spindle interactions after metaphase.

During cortical divisions, we saw promiscuous microtubules in metaphase in klp61f³ embryos whether or not the mother carried a transgene; the incidences were not significantly different among klp61f³, [WT]; klp61f³ and [3YF]; klp61f³ embryos (Table 1). However, in anaphase, [3YF]; klp61f³ embryos, but not [WT]; klp61f³ embryos, showed a significant increase in spindles with promiscuous microtubules (p < 0.05) (Figure 4E, Table 1). This could be because in [WT]; klp61f³ embryos, regulation by dWee1 ensures optimal activation of transgenic KLP61F and proper spindle integrity (Model in Figure 4I). In [3YF]; klp61f³ embryos, the level of transgenic KLP61F is similar to that in [WT]; klp61f³ embryos, but the lacks regulation by dWee1. Further more, it appears that having elevated KLP61F that cannot be regulated (in [3YF]; klp61f³) is worse than having reduced KLP61F that can be regulated because klp61f³ embryos do not show anaphase spindle interactions.

The incidence of promiscuous microtubules in [3YF]; klp61f³ embryos at ~7% was lower than the ~28% in dwee1 mutant embryos (Table 1) [11]. This could be because dWee1 acts through both KLP61F and nuclear/spindle positioning [11] to prevent collisions. Also, the presence of endogenous KLP61F may prevent a stronger effect by Myc-KLP61F^3YF. Since KLP61F forms tetramers [2, 4], Myc-KLP61F^3YF may complex with endogenous KLP61F. The resulting complex could be compromised in function and could also explain why Myc-KLP61F^3YF acts in a dominant negative manner.

These results lead us to conclude that mutation of tyrosines in KLP61F, which are important for in vitro phosphorylation by dWee1, phenocopies spindle defects seen in dwee1 mutant embryos in vivo. We suggest that in the absence of phosphorylation by dWee1, Myc-KLP61F^3YF is less functional. This would explain the failure of Myc-KLP61F^3YF to rescue monopolar spindles in klp61f³ embryos or to rescue polyploidy in klp61f³ mutant larvae. Reduced ability of KLP61F to cross-link and translocate along microtubules would result in loss of spindle bipolarity and chromosome segregation failure. In addition, a decreased ability of Myc-KLP61F^3YF to cross-link microtubules may lead to reduced organization within the spindle. Resulting microtubule spurs could then interact with neighboring spindles in a syncytium, especially during anaphase when spindles are no longer protected by actin furrows.

Multiple roles for dWee1

Based on these and previous data, dWee1 appears to regulate mitosis in three ways during syncytial divisions. It (negatively) regulates Cdk1 to time the entry into mitosis [10]; it regulates γ-TuRC, perhaps indirectly, to position the mitotic spindle directly beneath the cortex [11]; finally, it (positively) regulates KLP61F to preserve spindle integrity and prevent microtubule interaction between neighbors (Figure 4I). Regulation of KLP61F by dWee1, we propose, is important for bundling parallel and/or anti-parallel microtubules and, consequently, generating a robust bipolar spindle. Without dWee1 regulation, KLP61F activity would be reduced, leading to unstable spindles, microtubule spurs, and interaction between neighboring spindles in a syncytium.

Outside the syncytium, mitotic regulation by dWee1 also appears to be important, particularly, during somatic cell cycles when the phospho-acceptor tyrosines of KLP61F are required for larval viability and chromosome segregation in neuroblasts. Consistent with this idea, dWee1 is also necessary to prevent mitotic defects in neuroblasts [10]. dwee1, however, is dispensable for larval growth as homozygous or hemizygous dwee1 mutants survive to adulthood, whereas Myc-KLP61F^3YF cannot support larval growth. We speculate that, another kinase may partially substitute for dWee1 in the larva. Precedence for overlapping function between Wee1 and another kinase (Myt1) is seen in Cdk1 regulation [24].

Implications of phosphorylation within the KLP61F head domain

Our studies show that the head domain of KLP61F is phosphorylated by dWee1. Regulation of a specific kinesin via phosphorylation in this domain may appear unlikely since it is highly conserved among all kinesins. However, we found that modification by dWee1 may be specific to the KLP61F head domain since the KHC head domain was a poor substrate. Previous studies show that domains outside of the Kinesin-5 head domain are also phosphorylated [25–28]. Cdk1 phosphorylates Kinsin-5 on a threonine in the “BimC box”, and is important for spindle localization of the motor [3, 26, 29]. Nek6, a NIMA kinase, also phosphorylates Kinesin-5 in the tail domain, possibly to regulate its function at the spindle poles [28]. The coiled-coil domain of Xenopus Kinesin-5 is also phosphorylated on a non-conserved serine by Aurora kinase, but does not appear to be important for spindle assembly [27, 29]. If and how these phosphorylation events affect activity of the head domain remains unknown. Direct regulation via phosphorylation within catalytic regions is not unheard of. Indeed, Wee1 kinases inhibit Cdk1 activity by phosphorylation in the ATP-binding domain [30]. The structure of KLP61F remains to be solved, but in the structure for human Kinesin 5, the amino acid that correspond to Y207 of KLP61F, is located near regions involved in nucleotide-sensing [31] (Supplemental Figure 4). The amino acid that corresponds to another potential phospho-acceptor, Y152, is near regions of KLP61F important for microtubule interaction [32] (Supplemental Figure 4). Exactly how phosphorylation in the head domain by dWee1 alters KLP61F activity remains to be investigated, but it is intriguing that two of the potential phospho-acceptor tyrosines may reside near regions important for kinesin activity.