The KLP-7 Residue S546 Is a Putative Aurora Kinase Site Required for Microtubule Regulation at the Centrosome in C. elegans

Regulation of microtubule dynamics is essential for many cellular processes, including proper assembly and function of the mitotic spindle. The kinesin-13 microtubule-depolymerizing enzymes provide one mechanism to regulate microtubule behaviour temporally and spatially. Vertebrate MCAK locates to chromatin, kinetochores, spindle poles, microtubule tips, and the cytoplasm, implying that the regulation of kinesin-13 activity and subcellular targeting is complex. Phosphorylation of kinesin-13 by Aurora kinase inhibits microtubule depolymerization activity and some Aurora phosphorylation sites on kinesin-13 are required for subcellular localization. Herein, we determine that a C. elegans deletion mutant klp-7(tm2143) causes meiotic and mitotic defects that are consistent with an increase in the amount of microtubules in the cytoplasmic and spindle regions of meiotic embryos, and an increase in microtubules emanating from centrosomes. We show that KLP-7 is phosphorylated by Aurora A and Aurora B kinases in vitro, and that the phosphorylation by Aurora A is stimulated by TPXL-1. Using a structure-function approach, we establish that one putative Aurora kinase site, S546, within the C-terminal part of the core domain is required for the function, but not subcellular localization, of KLP-7 in vivo. Furthermore, FRAP analysis reveals microtubule-dependent differences in the turnover of KLP-7(S546A) and KLP-7(S546E) mutant proteins at the centrosome, suggesting a possible mechanism for the regulation of KLP-7 by Aurora kinase.


Introduction
The microtubule (MT) cytoskeleton is required for multiple essential cellular processes including cell division, polarization, and transportation of intracellular cargo. Remodeling of MT structures occurs through changes in MT growth and shrinkage, via the addition or removal of α/β tubulin subunits. The regulation of MT dynamics occurs spatially and temporally as demands on the MT cytoskeleton change throughout the cell cycle. Many proteins interact with MTs to regulate polymer behaviour, for example, the microtubule-associated protein XMAP215 promotes MT polymerization [1][2][3] whereas MCAK (mitotic centromere-associated kinesin) deletion spanning two exons that is expected to remove the first 48 amino acids of the kinesin motor domain [31] and introduce a frame-shift [32]. Therefore, even if the klp-7(tm2143) transcript were translated, it would not likely yield a functional KLP-7 protein. Anti-KLP-7 antibodies have been generated against an N-terminal region of the protein (a.a. 14-190 of K11D9.1a [27]); tm2143 is expected to remove only the last 94 amino acids of this region. With these antibodies, KLP-7 was not detected in klp-7(tm2143) worms by western blot (Fig 1A) or immunofluorescence (S1 Fig), consistent with the idea that tm2143 is a null mutation. Surprisingly, many progeny from tm2143 homozygous worms survive to adulthood and are fertile. Furthermore, tm2143 lethality was temperature-sensitive, with 35% embryonic viability at permissive temperatures and 8% at the restrictive temperature (Table 1). klp-7(tm2143) worms also exhibited a higher number of spontaneous male progeny (1% vs. 0.2% at 25°C), likely indicating meiotic defects due to nondisjunction of the X chromosome [33].
To observe the MT cytoskeleton and chromatin during female meiosis, we examined wildtype and klp-7(tm2143) embryos expressing GFP::tubulin and mCherry::histone transgenes by filming one-cell embryos in utero. Compared to wild-type, klp-7(tm2143) embryos exhibited ectopic MTs in regions surrounding the meiotic spindle ( Fig 1B; S1 and S2 Movies), and near the cortex of the cell. To quantify the effect, we measured the amount of GFP-tubulin fluorescence near the cortex and compared it to a region in the center of the cell (Fig 1C). The ectopic MTs in klp-7(tm2143) embryos were observed throughout meiosis I and II. During wild-type anaphase I and II, a MT bundle normally forms between the segregating chromosomes and is restricted to the midzone region [34,35]. In both anaphase I and II in klp-7(tm2143) embryos, many MTs were observed beyond the spindle midzone region. Although the chromosomes separated during both meiosis I and II (n = 18), the spatial distribution of chromosomes was more variable in klp-7(tm2143). This was especially evident between telophase I and metaphase II, and during telophase II (Fig 1B). Polar body extrusion failed in many klp-7(tm2143) embryos (7/18 MI; 14/18 MII) compared to wild-type controls (0/11 MI; 1/11 MII).
We examined the subcellular location of KLP-7 in female meiosis by immunostaining. During meiosis I and II, KLP-7 localized to chromosomes, and appeared concentrated between paired chromosomes, similar to what has been observed for members of the chromosomal passenger complex [35]. We also observed KLP-7 concentrated at the spindle poles in metaphase I and II. KLP-7 remained associated with chromosomes in anaphase and telophase. During late anaphase II, KLP-7 also localized to the spindle midzone between separating chromosomes ( Fig 1D). KLP-7 immunofluorescence was also visible throughout the cytoplasm of meiotic embryos, consistent with a role of KLP-7 in regulating cytoplasmic MTs during meiosis.
Previous examination of klp-7(RNAi) embryos in mitosis indicated three prevalent phenotypes: an increase in the number of astral MTs, an abrupt anaphase with higher than normal centrosome separation rates, and a reduction in the amount of spindle midzone MTs during anaphase [24][25][26]. We observed similar defects with the klp-7(tm2143) mutant (Figs 2 and 3; S3 and S4 Movies). klp-7(tm2143) mutant embryos that expressed a GFP::KLP-7(WT) transgene exhibited normal anaphase chromatid separation rates, indicating the utility of a structure-function approach to study KLP-7 in vivo (Fig 2; S5 Movie).

KLP-7 is required for proper midzone microtubule formation
An increase in astral MT levels in klp-7(tm2143) embryos could cause increased pulling forces during anaphase, and thus explain the spindle breakage [26] (Fig 2A and 2B). However, these embryos also displayed fewer midzone MTs during anaphase (Fig 3A). Considering that KLP-7 locates to kinetochores, it is possible that KLP-7 contributes directly to the proper formation of the spindle midzone MTs [24,28]. In this case, the spindle-break phenotype in klp-7 demonstrates that KLP-7 is not detected in klp-7(tm2143) worm lysate. (B) Wild-type (upper panels, also S1 Movie) and klp-7(tm2143) (lower panels, also S2 Movie) meiotic spindles during meiosis I and II are displayed. GFP::tubulin is green and mCherry::histone is magenta. In klp-7(tm2143) embryos, many long MTs were visible near the meiotic spindle (arrowhead), and chromosome scattering was often observed after anaphase. Scale bar is 5 μm. (C) MTs were more prevalent near the cortex in meiotic klp-7(tm2143) embryos. The average ratio of cortical GFP::tubulin to a central region (a:b) in metaphase embryos is shown for WT (n = 16) and klp-7(tm2143) (n = 24). Error bars are SEM. (D) Wild-type embryos were fixed and stained to visualize MTs (green), KLP-7 (red) and DNA (blue). KLP-7 associates with chromosomes and spindle poles. Higher magnification insets show a higher concentration of KLP-7 at the region between meiotic chromosomes in metaphase I and II (arrows), and to the spindle midzone between the separating chromosomes in anaphase II (arrowhead). Scale bars are 5 μm. (tm2143) embryos could be due to a weakened resistance to anaphase force. To distinguish between these possibilities, we reduced anaphase forces by depleting the G-protein regulator, GPR-1/2, in both wild-type and klp-7(tm2143) embryos. Consistent with previous studies [36][37][38], klp-7(tm2143) embryos showed a significant reduction in the amount of midzone MTs compared to wild type (Fig 3B, p = 0.0001; two-tailed Student's t-test). RNAi against gpr-1/2 in the klp-7(tm2143) background significantly increased midzone MTs compared with klp-7 (tm2143) (p = 0.019; two-tailed Student's t-test), and gpr-1/2(RNAi) embryos showed the highest level of midzone MTs (Fig 3B). It is well-established that gpr-1/2(RNAi) causes a reduction in anaphase pole separation rates [36][37][38], and our results further indicate that the amount of midzone MTs is inversely correlated with spindle-pulling force.

KLP-7 is phosphorylated in vivo
In vertebrate systems, MCAK localization to chromatin and spindle poles is regulated via phosphorylation by Aurora B kinase and Aurora A kinase, respectively [8,10,16,17,23]. Aurora B kinase phosphorylates MCAK to inhibit MT depolymerization activity at the chromatin/kinetochore in cultured cells and Xenopus egg extracts [8,10,16] and Aurora A kinase phosphorylates MCAK at the spindle poles to promote Ran-dependent spindle bipolarity [23]. KLP-7 has twelve putative Aurora kinase sites, as determined by the group-based phosphorylation predicting and scoring method [39]. Therefore, we reasoned that the Aurora kinases could be important regulators of KLP-7 in C. elegans. In worms, Aurora B kinase (AIR-2) is a chromosomal passenger complex protein implicated in meiotic and mitotic spindle functions and Aurora A kinase (AIR-1) is involved in centrosome maturation and mitotic spindle assembly [29,30,[40][41][42][43][44]. Because of the overlapping locations of KLP-7 with AIR-1 and AIR-2 at centrosomes and chromosomes respectively, we sought to determine whether Aurora kinases play a role in either localizing or regulating KLP-7. Severe phenotypes and pleiotropy associated with air-1(RNAi) and air-2(RNAi) prevented the use of a simple genetic experiment to determine if the Aurora kinases are in the same pathway as KLP-7 for MT regulation. Therefore, we used a structure-function analysis of KLP-7 to test the hypothesis that worm kinesin-13 is regulated by the Aurora kinases in vivo. We first looked for evidence of in vivo phosphorylation of KLP-7 using Phos-tag SDS-PAGE followed by immunoblotting with anti-KLP-7 antibodies. Immunoblots of wild-type lysates revealed two bands that migrated more slowly than expected for KLP-7 (predicted MW of 80 KDa). Treatment of the lysate with protein phosphatase resulted in a single band at 80 KDa, which likely represented non-phosphorylated KLP-7 ( Fig 4A). We concluded that the KLP-7 protein was phosphorylated in vivo, possibly at multiple sites.
To determine whether the mobility of KLP-7 was altered by in vivo Aurora kinase activity, we collected lysates from wild-type, air-1(RNAi), air-2(RNAi) and air-1(RNAi);air-2(RNAi) worms. For each RNAi experiment, expected phenotypes were observed by fluorescence microscopy, confirming effective knockdown. Using this approach, we did not observe an obvious change in KLP-7 mobility on Phos-tag SDS PAGE (Fig 4B), suggesting that the RNAi knockdown did not significantly alter the phosphorylation state of KLP-7, or that subtle changes in phosphorylation occurred that were not detectable via the Phos-tag assay.
Aurora kinase activity has a subtle effect on the mobility of endogenous KLP-7 via 2D gel electrophoresis In order to resolve any differentially phosphorylated forms of KLP-7, we next looked for Aurora-dependent mobility shifts in KLP-7 via 2-D gel electrophoresis, followed by immunoblotting with anti-KLP-7 antibodies. Using this approach, we observed many distinct protein species (Fig 4C), which were sensitive to phosphatase treatment (S2 Fig). The migration pattern of KLP-7 suggested that endogenous KLP-7 protein might undergo multiple modifications. air-1(RNAi) resulted in a relative increase in at least one species of KLP-7, consistent with reduced phosphorylation (Fig 4C; right side of isoelectric axis). air-2(or207ts) caused many changes in the overall pattern of KLP-7 migration, some of which could represent increased or decreased phosphorylation. Although the complex migration pattern of KLP-7 suggested that other regulatory enzymes probably also modify the protein, we concluded that both Aurora kinases likely contribute to KLP-7 modification in vivo.

AIR-1 and AIR-2 directly phosphorylate KLP-7 in vitro
In order to verify that KLP-7 is an Aurora substrate, we performed in vitro kinase assays on bacterially-expressed GST-KLP-7 proteins. Full-length KLP-7 was insoluble, therefore, we expressed KLP-7 fragments that covered clusters of putative Aurora sites at the N-terminus KLP-7-N (a.a. 2-189) and the C-terminus with part of the core domain, KLP-7-CC (a.a. 487-689) (Fig 5A). We first tested the N-terminal fragment for phosphorylation. Previous work showed that the C. elegans INCENP protein, ICP-1, promoted AIR-2 phosphorylation of maltose binding protein [45]. Using an AIR-2/ICP-1 enzyme cocktail in an in vitro kinase assay, we observed that N-terminal KLP-7 was efficiently phosphorylated (Fig 5B and S3 Fig). In contrast, AIR-1 only weakly phosphorylated this fragment ( Fig 5C and S3 Fig). This suggested that either KLP-7 was not an effective substrate for AIR-1 or that, similar to AIR-2 kinase, AIR-1 required an accessory protein for efficient in vitro activity. TPX2 (Targeting Protein for Xenopus Klp2) has been shown to target Aurora A to spindles in both human cells and C. elegans, and to promote Aurora A kinase activity on histone H3 in vitro [46][47][48]. AIR-1 and TPXL-1 co-localize along astral MTs, with enrichment near the centrosome [48]. We tested TPXL-1 for its ability to facilitate phosphorylation of KLP-7 by AIR-1 in vitro. Indeed, AIR-1 together with TPXL-1 efficiently phosphorylated the N-terminal KLP-7 fragment (Fig 5C and S3 Fig).
We repeated the kinase assays using a C-terminal fragment that contained part of the core domain KLP-7-CC (a.a. 487-689) and we observed similar TPXL-1-dependent activity for AIR-1 kinase ( Because the C-terminal fragment appeared less soluble in the bacterial lysate, and was less phosphorylated by AIR-1 compared to the N-terminal fragment, we concluded that this fragment might not be an effective substrate due to non-native conformation. Therefore, we tested a smaller fragment (CC2) and observed efficient phosphorylation by AIR-2 kinase (Fig 5E and  S3 Fig). We concluded that both AIR-1 and AIR-2 directly phosphorylate KLP-7 in vitro, and that TPXL-1 promotes AIR-1-dependent phosphorylation of KLP-7.
Mutational analysis of putative Aurora kinase sites reveals a role for S546 in KLP-7 function in vivo In order to test the putative Aurora phosphorylation sites for their role in KLP-7 function in vivo, we mutated transgenes and tested their ability to rescue the klp-7(tm2143) deletion mutant. GFP-tagged KLP-7 was used to monitor changes in the intracellular location of the mutated proteins. We were unable to obtain every possible mutant combination, however, we tested multiple residues within clusters of kinase sites for both non-phosphorylation (serine or threonine to alanine, S/T to A) and constitutive phosphorylation (serine or threonine to glutamic acid, S/T to E) (S1 Table). We scored each of the stable transgenic lines for embryonic viability. At 25°C, klp-7(tm2143) worms exhibited low embryonic viability (15.9±4.1% survival). Compared to klp-7(tm2143), transgenic worms, including KLP-7(WT), T119A, T119E, T182E, N4A, N4E, S539A, S539E, S538A539A, S538ES539E and 8A showed increased embryonic viability, ranging from 40% to 90% (Table 2). This suggested that these transgenes retained some function. In contrast, four transgenic worm strains exhibited very low embryonic viability (16-30%; Table 2). In all of these strains, the S546 residue was mutated (S546A, S546E, T182ES539ES546E, and 10E). Western blots of lysates from transgenic animals probed with KLP-7 revealed that the fusion proteins were expressed, suggesting that the loss of embryonic viability was not due to a lack of protein expression (S5 Fig).
We next examined GFP::KLP-7 location in the various mutants using fluorescence microscopy ( Fig 6A and S6 Fig). Our results indicated that S546A, S546E, and T182ES539ES546E fusion proteins exhibited intracellular localization patterns that were similar to wild-type GFP:: KLP-7 ( Fig 6A). In contrast, the GFP::KLP-7(10E) mutant, predicted to mimic constitutive phosphorylation at 10 putative Aurora kinase sites (S1 Table), exhibited only cytoplasmic fluorescence in the one-cell embryos. Regulation of KLP-7 at the Centrosome The rescue experiments indicated that residue S546 was important for the function of KLP-7. Because KLP-7 activity inversely correlated with centrosomal MT levels and spindle pole separation rates in anaphase, we tested whether these parameters were specifically influenced by changes to S546. To quantify the effect, we measured the velocity of the posterior centrosome during the first 40 seconds of anaphase. We found that, while other mutant forms of KLP-7 rescued the fast centrosome movements that gave rise to the spindle-break phenotypes (S7 Fig), the GFP::KLP-7(S546A) and GFP::KLP-7(S546E) mutations still exhibited fast centrosome movements in anaphase (Fig 6B). Furthermore, neither mutation rescued the overproduction of centrosomal MTs characteristic of klp-7(tm2143) (Figs 2B and 6C). Therefore, the S546 residue is likely important for KLP-7 function at the centrosome.

KLP-7 is required for the proper formation and organization of microtubules
Although klp-7(tm2143) is an apparent null mutation, it causes temperature-sensitive lethality. Because MT growth in C. elegans is inherently temperature-sensitive [26], it is conceivable that the loss of a MT regulator could also cause variable phenotypic severity at different temperatures. Our characterization of klp-7(tm2143) mutants indicated that C. elegans kinesin-13 normally limits MT formation within the cytoplasm and spindles during meiosis as well as at the centrosomes in mitosis. One exception to this generalization is that loss of KLP-7 activity also caused a decrease in spindle midzone MTs during anaphase of mitosis. KLP-7 could be required to generate the spindle midzone MTs directly, however, this property seems inconsistent with the expected molecular role of kinesin-13 proteins to depolymerize MTs. Alternatively, the reduction of spindle midzone MTs associated with KLP-7 loss could be an indirect consequence of increased pulling forces during anaphase [24,25].
It has been previously proposed that anaphase spindle movements in C. elegans one-cell embryos utilize a limited number of cortically-located force generators [49]. In this context, excess microtubules due to klp-7(tm2143) might not be expected to increase the anaphase pulling force. However, klp-7(tm2143) could simply increase the rate at which new MTs contact a limiting number of cortical force generators. Indeed, it has been suggested that altering MTcontact frequency with the cortex could modulate anaphase spindle movements if individual MT-cortex interactions occur via a brief touch-and-pull mechanism [50].
Inhibition of MCAK causes defects in chromosome alignment in meiotic Xenopus egg extracts and mouse oocytes, as well as chromosome segregation defects resulting from increased MT-kinetochore attachment errors [14,16,51]. Loss of KLP-7 resulted in excessive cortical MTs in meiotic embryos and a disorganized MT network associated with female meiotic spindles. It is not clear what relationship, if any, the excess cortical MTs have to the Phosphorylated KLP-7 (green) exchanges rapidly between centrosomes and cytoplasm, by a MT-independent mechanism. Dephosphorylated KLP-7 (magenta) locates to the centrosome but must undergo phosphorylation (1) and subsequent dephosphorylation (2) before depolymerizing MTs and exiting the centrosome. The phosphorylation of KLP-7 at the centrosome could occur via TPXL-1-activated Aurora kinase A (AIR-1), based on in vitro data and data from Ozlü et al., [48]. embryonic lethality associated with klp-7(tm2143). However, excess MTs near the spindle, regardless of their point of origin, could physically displace chromosomes and contribute to chromosome scattering observed in meiosis I and II.

Phosphorylation of KLP-7 by Aurora kinases
Our data showed that endogenous KLP-7 is phosphorylated both in vitro and in vivo. Although the alteration to KLP-7 mobility upon AIR-1 loss was subtle, it is possible that only a subset of KLP-7 at the centrosomes is directly phosphorylated by AIR-1 in vivo. Alternatively, perhaps only one or a few potential KLP-7 sites are utilized by AIR-1 to regulate the protein, in which case only subtle mobility shifts would be expected via 2D gel analysis. This is consistent with our data that implicated only S546 in having a strong influence on KLP-7 function in vivo. In addition, other kinases have been shown to phosphorylate MCAK and regulate its localization and/or activity in mammalian cells, including Cdk1 (Cyclin-dependent kinase 1), PLK1 (Pololike kinase 1), and PAK1 (p21 activated kinase) [52][53][54]. Worm homologues exist for all three kinases and they might also contribute to the regulation of KLP-7 in vivo.

Regulation of KLP-7 activity by Aurora kinases
The kinase assays demonstrated that KLP-7 is a direct substrate of AIR-1 and AIR-2 kinases in vitro. AIR-1 on its own exhibited weak kinase activity on KLP-7, however, the addition of TPXL-1 resulted in efficient phosphorylation of KLP-7 by AIR-1. Our results showing that TPXL-1 stimulated the phosphorylation of KLP-7 by Aurora kinase in vitro, together with the fact that KLP-7, AIR-1 and TPXL-1 all locate at the centrosomes [48], suggest that KLP-7 is a physiological substrate of the active AIR-1/TPXL-1 complex.
Previous work established that Aurora kinases phosphorylate MCAK to regulate its activity with spatial and temporal precision. In the chromatin region, Aurora B kinase phosphorylates MCAK at S196 to inhibit its MT depolymerization activity, and at T95 and S110 to regulate its location to chromatin [8,10,16,17]. In Xenopus egg extracts without chromatin and centrosomes, Aurora A kinase phosphorylates MCAK at S196 to promote pole focusing and at S719 to promote its localization to spindle poles for bipolar spindle formation [23]. In this study, we found that individual Aurora kinase sites and groups of up to 8 sites (e.g., T119E, T119A, S538ES539E, S538AS539A, N4E, N4A, 8A, and T182ES539ES546E) could be altered without affecting the localization of KLP-7 in C. elegans embryos. Although GFP::KLP-7 fluorescence is specifically reduced at the centrosomes in air-1(RNAi) mitotic embryos (S6 Fig), air-1(RNAi) embryos are defective in overall centrosome maturation [44], making it difficult to make conclusions about the role of phosphorylation on KLP-7 localization when the kinase is depleted. The GFP::KLP-7 protein that contained 10 mutated Aurora sites (10E) failed to localize to kinetochores and centrosomes. Although we could not rule out the possibility of incorrect folding of the 10E protein in vivo, it is possible that dephosphorylation of KLP-7 at multiple Aurora sites is required for proper localization in mitosis.

A role for S546 in KLP-7 function
Despite localizing normally, GFP::KLP-7(S546A) and GFP::KLP-7(S546E) transgenes failed to rescue klp-7(tm2143) embryonic lethality or spindle breakage. The predicted Aurora kinase site lies within the catalytic core and is well conserved among C. elegans, Xenopus, mouse and human (Fig 6D). S546 is located in the α-helix 5 of the catalytic core domain, which is likely important for kinesin-MT binding [55,56]. Phosphorylation at S546 may introduce a negative charge to this region and thus interfere with KLP-7's interaction with MTs. Direct evidence for phosphorylation of this particular residue by Aurora is still needed; however, the phosphomutations reveal an important function for this site.
It is well established that phosphorylation regulates the precise location and function of kinesin-13 family proteins, however, few reports have identified functions for phosphorylation sites in the core domain [8,10,16,17,23,57]. The Drosophila kinesin-13 protein, KLP10A is phosphorylated by casein kinase 1α in the core domain (Drosophila S573, C. elegans S555). Phosphorylation at this site alters KLP10A association with the MT lattice [57]. One Cdk1 phosphorylation site (T537) has also been identified in the vicinity of S542 (the human equivalent of the worm S546 site) in human cells [52]. Phosphorylation of this site by Cdk1 is required for proper mitotic spindle formation, likely by attenuating MCAK MT-depolymerizing activity and promoting the release of MCAK from the spindle poles [52]. Although C. elegans KLP-7 does not have an equivalent Cdk1 site, S546 could perform a similar function.

A model for KLP-7 regulation at the centrosome
The FRAP results revealed a distinct difference between GFP::KLP-7(S546A) and GFP::KLP-7 (S546E) mutants. When MTs were depolymerized with nocodazole, GFP::KLP-7(S546A) and GFP::KLP-7(WT) recovery was 2-fold slower compared to untreated embryos, whereas the GFP::KLP-7(S546E) mutant embryos showed no change upon treatment. This suggests that turnover of KLP-7(S546A) at the centrosome is at least somewhat dependent on MTs. Furthermore, in the nocodazole-treated cells, a significant portion of centrosomal KLP-7 is likely not phosphorylated on S546, because the WT protein exhibited recovery rates that were similar to the S546A mutant. However, it is difficult to explain why the S546A mutant did not rescue the MT defects associated with klp-7(tm2143). In order to reconcile this result with the FRAP data, we suggest that KLP-7 undergoes a 2-step process for normal activity (Fig 7D). In this model, S546E and S546A can both locate to the centrosome, but neither form is able to depolymerize centrosomal MTs directly. We suggest that an initial phosphorylation step is required at the centrosome, likely by AIR-1 kinase/TPXL-1 (step 1 in Fig 7D). Then, subsequent dephosphorylation at S546 (step 2 in Fig 7D) would promote MT depolymerization. In this manner, cycles of phosphorylation and dephosphorylation would be required for the normal activity of the protein in vivo, explaining why both S546A and S546E exhibit loss-of-function phenotypes. An initial phosphorylation step could allow precise temporal or spatial regulation of a centrosomespecific pool of KLP-7, in order to prevent inappropriate activation of the protein by Aurora kinases at, for instance, other regions of the cell. Further work is required to elucidate the precise mechanism by which S546 alters KLP-7 function at the centrosome.

Phos-tag SDS-PAGE
Phos-tag SDS-PAGE was performed as previously described [63]. Proteins were separated by an 8% polyacrylamide gel containing 25 μM Phos-tag acrylamide (NARD Institute) and 25 μM MnCl 2 , and were transferred to nitrocellulose membrane (Pierce, Thermo Scientific). The resulting membrane was blocked with 10% non-fat milk in TBST, and probed with a rabbit anti-KLP-7 polyclonal antibody (1 μg/mL) [27]. The membrane was washed and incubated with a horseradish peroxidise-conjugated goat anti-rabbit antibody (Bio-Rad Laboratories). ECL western blotting substrate (Pierce, Thermo Scientific) was prepared according to the manufacturer's instructions and applied to immunoblots. The membrane was exposed to Kodak BioMax XAR film for detection.

2-D gel electrophoresis
Immobiline DryStrip (7 cm; pH 7-11 NL for Fig 4 and pH 6-11 for S2 Fig) gels, IPG buffer (pH 7-11 NL, and pH 6-11) and DeStreak Rehydration Solution was used for 2D gel experiments (GE Healthcare). Lysates from N2, air-1(RNAi) and air-2(or207ts) worms were extracted by sonication in 2D sample buffer (8 M Urea, 4% w/v CHAPS, 2% v/v IPG buffer, 40 mM DTT) supplemented with protease inhibitor cocktail (Mini-Complete, Roche Applied Science). The protein concentration was determined by performing the Bradford assay with protein assay dye reagent (Bio-Rad Laboratories). 120 μg of total protein was loaded to each gel strip by rehydration. Isoelectric focusing (IEF) was performed according to the manufacturer's instructions. Protein samples were further separated on 10% polyacrylamide gels and transferred to nitrocellulose membranes. The resulting immunoblots were probed with affinitypurified anti-KLP-7 antibody. To dephosphorylate proteins for 2D gel analysis, 100 μg of whole-worm lysate was treated with 2000 U of lambda phosphatase and then prepared in 2D sample buffer.

Immunofluorescence of stained embryos and live-cell imaging
Confocal microscopy was performed with an X81 motorized inverted microscope (Olympus) equipped with a Yokogawa CSU-10 spinning disk system. Images were collected with a 60× objective (NA 1.42) and a Hamamatsu Orca R2 camera controlled by MetaMorph software (Molecular Devices).
To quantify centrosomal MTs in metaphase embryos, GFP::KLP-7(S546E) or GFP::KLP-7 (S546A) worms were fixed and stained under the same conditions with the wild-type control worms. DM1A (Sigma-Aldrich) and rabbit polyclonal anti-γ-tubulin antibodies [44] were used at 1 μg/mL and 8 μg/mL respectively. In a single frame where the centrosome is in focus, the centrosomal MT level is defined as the fluorescence intensity of a circular region at the centrosome subtracting the intensity at a background cytoplasmic area. The relative intensity for each centrosome was calculated as: (integrated intensity of one centrosome/the average integrated intensity of control centrosomes) x 100.
GFP::tubulin fluorescence at the midzone of anaphase embryos was quantified from integrated fluorescence intensities within a rectangular region (4.65 x 3.72 μm) from a single confocal plane.
The spindle-snap phenotype was scored by first dissecting worms in 5 μL egg buffer (188 mM NaCl, 48 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , and 25 mM HEPES, pH 7.3). Embryos were then mounted on a thin 2% agarose pad and imaged for GFP::KLP-7(WT or mutants) from the start of mitotic metaphase with 5 second intervals. The environment temperature during imaging was controlled at 25 ± 0.5°C according to previous description [65]. The maximum velocity of the posterior centrosome during the first 40 seconds of mitotic anaphase was determined.
To perform in utero imaging of meiotic embryos, worms expressing GFP::tubulin and mCherry-histone were immobilized with 4 mM tetramisole hydrochloride (Sigma-Aldrich) and mounted on a thin 2% agarose pad. 20 second time-lapse images were collected with a 60× silicone oil objective (NA 1.3) from fertilization to the end of meiosis II. At each time point, 3 images were taken at an interval of 1 μm to visualize the structure of the meiotic spindles.

Nocodazole treatment and FRAP
To create permeabilized embryos, L4-adult worms were grown for 12-16 h at 20°C on ptr-1 RNAi-feeding plates [66,67]. Worms were dissected in embryo buffer and mounted on agarose pads with a coverslip for imaging. To depolymerize MTs, nocodazole was added (10 μM final concentration) just before mounting and allowed to incubate for at least 30 seconds [68]. Onecell embryos in metaphase were used for all photobleaching experiments. Only embryos that displayed a lack of anaphase pole separation after the FRAP experiment were deemed affected by nocodazole and used for analysis.
Photobleaching experiments were performed on a spinning disk confocal as described above. The bleaching laser (400 mW; Power Technology) was controlled by Metamorph Software using the Mosaic targeted illumination interface. Sufficient photobleaching of a circle (30 pixel diameter) was achieved with a targeted widefield beam at maximum laser power for 3 seconds. Confocal images were acquired at 1 second intervals both before photobleaching (5 frames), and during the recovery phase (90 frames). Fluorescence intensity of GFP::KLP-7 was measured with Metamorph software at the centrosome and at two background locations to correct for photobleaching resulting from image acquisition. Curves (S8 Fig) were generated with easyFRAP software, using full-scale normalization and double term curve-fitting [69].  Table 1.

Supporting Information
(XLSX) S9 File. Raw data of the results presented in Table 2.