Microtubule-associated proteins (MAPs) ensure the fidelity of chromosome segregation by controlling microtubule (MT) dynamics and mitotic spindle stability. However, many aspects of MAP function and regulation are poorly understood in a developmental context. We show that mars, which encodes a Drosophila melanogaster member of the hepatoma up-regulated protein family of MAPs, is essential for MT stabilization during early embryogenesis. As well as associating with spindle MTs in vivo, Mars binds directly to protein phosphatase 1 (PP1) and coimmunoprecipitates from embryo extracts with minispindles and Drosophila transforming acidic coiled-coil (dTACC), two MAPs that function as spindle assembly factors. Disruption of binding to PP1 or loss of mars function results in elevated levels of phosphorylated dTACC on spindles. A nonphosphorylatable form of dTACC is capable of rescuing the lethality of mars mutants. We propose that Mars mediates spatially controlled dephosphorylation of dTACC, which is critical for spindle stabilization.
Microtubule-associated proteins (MAPs) ensure the fidelity of chromosome segregation during cell division by controlling the formation and stability of spindle microtubules (MTs). Because disruption of spindle formation can promote genomic instability, an understanding of MAP function and regulation is central to dissecting basic mechanisms of tumorigenesis and would be invaluable in designing new therapies for the treatment of cancer. Although much progress has been made in understanding the functions of spindle-associated MAPs in the last few years, many aspects of their role or regulation remain to be fully elucidated. Human hepatoma up-regulated protein (HURP) has been described as a highly charged MAP that can bind directly to MTs in vitro and enhance their polymerization (Santarella et al., 2007). In vivo, HURP is part of a Ran-dependent complex that stabilizes mitotic MTs and is required for the formation and function of bipolar mitotic spindles (Koffa et al., 2006; Sillje et al., 2006; Wong and Fang, 2006). However, it is not known how HURP-associated proteins functionally interact with one another in a developmental context to support normal cellular function.
Mars, a D. melanogaster sequence homologue of HURP, was previously identified as a protein phosphatase 1 (PP1) binding protein, implicating reversible phosphorylation in the control of Mars or Mars-associated proteins (Bennett and Alphey, 2004; Yang et al., 2005). In this paper, we report the essential role of mars during early embryogenesis, its interactions with other MAPs, and its key role in promoting protein dephosphorylation on mitotic spindles to ensure spindle stability.
Results And Discussion
HURP is a component of the mitotic spindle apparatus. To determine the cell cycle distribution of Mars, we generated a Mars-specific antibody and used it to stain syncytial embryos undergoing nuclear division. In prophase, Mars antibody staining was predominantly around the centrosome. In metaphase and anaphase, Mars was localized on spindle MTs in a gradient along the pole-to-pole axis with more intense staining at the centrosome-proximal regions (Fig. 1 A).
This is distinct from the distribution of HURP, which has been shown to be predominantly at chromatin-proximal regions until telophase, when levels sharply decline (Koffa et al., 2006; Sillje et al., 2006; Wong and Fang, 2006). In telophase, we observed a low level of discrete Mars staining at the midbody (not depicted), but the majority of Mars protein appeared to be spread over the nuclear envelope where it persisted during interphase (Fig. 1 A and Fig. S1). Double staining for Mars and either Klp10A, which is primarily localized at focused minus ends where it promotes depolymerization and poleward flux (Rogers et al., 2004), or γ-tubulin, which marks the face of the centrosome and nucleates MT polymerization (Jeng and Stearns, 1999), confirmed that Mars is localized at MT minus ends but not at the centrosome (Fig. 1, B and C). Mars' localization during mitosis was completely disrupted upon treatment with colchicine to depolymerize MTs, indicating that Mars associates with spindle MTs (Fig. 1 D).
To determine the in vivo role of mars, we generated a null allele of mars, mars1, by imprecise excision of a P element transposon (referred to as marsP hereafter), which we found inserted in the mars 5′ untranslated region (Fig. 2, A and C).
marsP flies express full-length Mars protein at a much lower level than wild type (Fig. 2 B). mars1 flies fail to produce Mars protein, which is consistent with molecular analysis revealing that ∼0.84 kb of the coding region, including the translation start site, is deleted in this mutant (Fig. 2, B and C). mars mutant flies are viable but female sterile. Notably, eggs laid by mars1, marsP, and mars1/P flies show a greatly reduced ability to hatch (Fig. 3 A). As we were able to visualize sperm tails in early-arrested embryos laid by mars mutant females (unpublished data), we conclude that the sterility observed in mars mutants is caused by arrest in embryogenesis after fertilization. The viability of embryos laid by mars mutant mothers was restored by moderate overexpression of marsWT (wild-type mars) in the female germline (Fig. 3 A), indicating that the failure of mars mutant embryos to develop is caused by disruption of the mars transcription unit.
To determine the cause of the lethality of mars mutant embryos, we fixed embryos from wild-type and mars mutant females and examined the distribution of nuclei and MTs. Embryos lacking maternal mars arrested during early embryogenesis after no more than five nuclear divisions. 81.9% (n = 144) of 15–45-min embryos laid by mars1 mothers exhibited at least two discrete DNA-containing regions (Fig. 3, B and C). The first of these was localized to the embryonic cortex and resembled a polar body, most likely containing the unused products of meiosis II (Wilson and Borisy, 1998). One or more additional DNA-containing regions, each surrounded by a bipolar spindle, were also observed more centrally, indicating that the vast majority of mutant embryos pass through meiosis to form one or more mitotic figure. Notably, the spindle structures in mars1 mutant embryos were very small and weak, albeit still bipolar (Fig. 3 C). Most spindles had at least one detached centrosome, possibly because of weakened spindle–centrosome interactions (Fig. 3, C–E). We also observed unaligned chromosomes in the mars1 mutant, which is indicative of insufficient MT attachment or tension (Fig. S2 A). Embryos laid by mars1/P mothers resembled those from mars1 mothers, except the phenotypes were somewhat less severe (Fig. 3, C–E). Quantitation of α-tubulin staining revealed no significant difference between astral MTs in embryos laid by wild-type and mars1/P mothers (P = 0.416), whereas cold-resistant kinetochore MTs were destabilized in embryos laid by mars1/P mothers (Fig. S2 B). Approximately 51% of embryos laid by marsP homozygous mothers (n = 264) failed to develop beyond embryogenesis. Many showed terminal phenotypes at or shortly after gastrulation, presumably as a consequence of primary defects during the early cleavage divisions. Collectively, analysis of loss-of-function mutants suggest that mars has a role in spindle MT stabilization. HURP has bundling activity in vitro and in vivo (Koffa et al., 2006; Sillje et al., 2006; Wong and Fang, 2006).
Correspondingly, strong overexpression of marsWT in syncytial embryos resulted in enlarged spindles with ectopic MT fibers (Fig. 3 F), suggesting that mars is limiting for MT bundling.
mars was originally isolated from a two-hybrid screen for putative PP1 binding proteins and contains a canonical PP1 binding motif (K/R,x,V,x,F; Bennett and Alphey, 2004; Bennett et al., 2006). To verify interactions with PP1, we performed pulldown experiments between Mars and PP1α87B, the major D. melanogaster PP1 isoform (Dombrádi et al., 1990; Baksa et al., 1993; Kirchner et al., 2007). We found that GST-tagged full-length MarsWT interacted efficiently with HA-tagged PP1α87B from crude D. melanogaster embryo extracts (Fig. 4 A), whereas cells expressing GST alone did not bind (not depicted). We also observed efficient binding between endogenous PP1 and FLAG-Myc (FM)–tagged MarsWT by coimmunoprecipitation from D. melanogaster embryo extracts (Fig. 4 B). To assess the importance of the putative PP1 binding motif for interaction with PP1, we tested a mutant form of Mars in which phenylalanine 839 was replaced with alanine (MarsF839A) in our pulldown assay. The ability of MarsF839A to bind PP1 was greatly reduced compared with wild-type Mars (Fig. 4 A), indicating that Phe839 is crucial for interaction with PP1. Immunoprecipitation with an HURP antibody followed by immunoblotting with PP1 antibody showed that endogenous human PP1 and HURP also interacted efficiently with each other in HeLa cell extracts (Fig. 4 C), suggesting that binding to PP1 is an evolutionarily conserved property of HURP proteins.
Binding to PP1 prompted us to test functional interactions between mars and PP1 in vivo. The ability of embryos laid by mars or PP1a87B heterozygotes to hatch resembled that of the wild type (Fig. 4 D). Embryos laid by flies transheterozygous for mars and PP1a87B showed a significantly reduced hatch ratio (Fig. 4 D). PP1 is a pleiotropic enzyme. To test the specific role of Mars-bound PP1, we introduced FM-tagged MarsF839A into flies to create Mars complexes lacking PP1. Ectopic expression of marsF839A at comparable levels to those of marsWT failed to rescue the embryonic lethality of embryos laid by mars1 or mars1/P mothers (Fig. 4 D). Collectively, these data suggest that binding to PP1 is critical for mars function. Identical distributions of MarsF839A and MarsWT, as determined by FLAG antibody staining, indicate that PP1 binding is not necessary for normal Mars localization on the mitotic spindle (Fig. 4 E).
HURP has been shown to interact with other MAPs, including TPX2 and XMAP215 in Xenopus laevis (Koffa et al., 2006), suggesting that its role in the stabilization of spindle MTs may be partly mediated via interactions with other spindle assembly factors.
In immunoprecipitation assays, we found that Mars associates with two proteins that are known to cooperate with each other to stabilize MTs during cell division (Lappin et al., 2002): the D. melanogaster XMAP215 homologue encoded by minispindles (msps) and Drosophila transforming acidic coiled-coil (dTACC) protein (Fig. 5 A). In syncytial embryos, dTACC is concentrated at centrosomes during mitosis but is also found on spindle MTs (Gergely et al., 2000) where it colocalizes with Mars (unpublished data). Mars staining was not affected in embryos that produce no detectable dTACC protein (Fig. S3 A; Lee et al., 2001). Conversely, the ability of dTACC to associate with MTs was largely unaffected in mars1/P mutant embryos (Fig. 5 B). Therefore, although Mars and dTACC associate with one another, they do not appear to be dependent on each other for their localization.
Phosphorylation of dTACC on Ser863 is critical for stabilization of the minus ends of centrosome-associated MTs during mitosis (Barros et al., 2005). Although dTACC is found on both the centrosome and mitotic spindle, phosphorylated dTACC (p-dTACC) is tightly localized to the centrosomes (Barros et al., 2005), suggesting that once phosphorylated, p-dTACC is either unable to exchange with the soluble pool of dTACC or is rapidly dephosphorylated when it leaves the centrosome. The role of dephosphorylated TACC is not known, but it may function to stabilize MTs through lateral interactions with MTs or interactions with MT plus ends. The localization of Mars toward the minus ends of spindle MTs and association with both dTACC and PP1 prompted us to examine the involvement of Mars in maintaining low levels of p-dTACC on the spindle.
To examine the effect of Mars on dTACC phosphorylation, we stained mars mutant embryos with an antibody that specifically recognizes dTACC phosphorylated at Ser 863 (p-dTACC). mars mutant embryos showed increased levels of p-dTACC on the mitotic spindles compared with the wild type (Fig. 5, B–D). On careful examination of these mutants, we noticed some spindles that looked normal but possessed elevated levels of p-dTACC (Fig. S3 B), indicating that increased p-dTACC was not simply a secondary consequence of aberrant spindle structure. Levels and distribution of total dTACC appeared normal in mars mutants (Fig. 5 B), although we cannot rule out that global ratios of DTACC/p-DTACC are affected. We used marsF839A to examine whether Mars promotes the dephosphorylation of dTACC by binding to PP1. Embryos with moderate levels of ectopic marsWT in embryos laid by either mars1 or mars1/P mothers were essentially wild type in appearance and had little or no p-dTACC staining on mitotic spindles. In contrast, mars mutant embryos ectopically expressing marsF839A at comparable levels to those of ectopic marsWT retained elevated p-dTACC staining on spindles (Fig. 5 B and not depicted).
To test whether promoting the dephosphorylation of dTACC is a critical function of mars, we examined whether the sterility of mars mutants could be rescued by a nonphosphorylatable form of dTACC (dTACCS863L) expressed under control of the dTACC promoter (Barros et al., 2005). dTACCS863L, but not dTACCWT(wild-type dTACC), restored a normal distribution of p-dTACC staining on mitotic spindles (Fig. 5, B–C). Quantification of p-dTACC staining confirmed that dTACCS863L restored a normal ratio of spindle/centrosomal p-dTACC staining in mars mutant embryos (Fig. 5 D). When we examined hatching of these embryos, we found that the lethality of embryos laid by mars1/P mothers was rescued by dTACCS863L but not dTACCWT (Fig. 5 D). Collectively, these data indicate that dephosphorylation of dTACC on the spindle is an essential function of mars. Homozygous mars1 mutants were not rescued by dTACCS863L (unpublished data), suggesting that residual Mars protein in mars1/P embryos may play a dTACC-independent role, such as MT bundling, or that the level of ectopic dTACCS863L was insufficient to compensate for elevated p-dTACC in a mars1 background.
In summary, we have shown that mars, which encodes a D. melanogaster sequence homologue of HURP, is critical for mitotic spindle structure and chromosome segregation during early embryogenesis. The primary defect in mars mutants appears to be loss of spindle MT stability, whereas overexpression of mars leads to the production of enlarged spindles with ectopic MTs. These data are consistent with a role for mars in MT bundling/stabilization similar to that described for its human homologue HURP. However, our identification of Mars as an interacting subunit of PP1 suggests a novel mechanism by which this family of proteins can maintain normal spindle structure in vivo. Binding of PP1 to Mars implicates PP1 in dephosphorylation of Mars or a Mars-associated protein.
Although dTACC may be a substrate of PP1, it is also possible that Mars-bound PP1 may indirectly stimulate dephosphorylation of dTACC on the spindle by activating another protein phosphatase or inactivating a dTACC kinase such as Aurora-A, a known target of PP1 during mitosis (Katayama et al., 2001).
Our genetic experiments indicate that promoting dephosphorylation of dTACC on mitotic spindles is an essential role of Mars. Why is it important to maintain dephosphorylated TACC on the spindle? One possibility is that mars functions to ensure that MT stabilization mediated by p-dTACC only occurs at the centrosome, allowing a more dynamic spindle. This is hard to test however, because the mechanism by which p-dTACC stabilizes MTs is unclear. The effect of dTACC on MT assembly appears to be mediated by effector proteins, such as Msps, as TACC has not been described to possess MT stabilizing activity on its own (Kinoshita et al., 2005). Phosphorylation of dTACC might help activate effectors because dTACC mutated at Ser863 is able to recruit Msps to the centrosome but not promote MT assembly (Barros et al., 2005). However, it is not known which aspect of Msps activity is affected by dTACC phosphorylation or to what extent the effect of dTACC phosphorylation is context dependent. It is conceivable that dTACC stabilizes spindle MTs by establishing lateral interactions with MTs or interactions with plus ends and that these functions of dTACC are impaired when phosphorylated at Ser863.
Is mars a functional homologue of HURP? We have confirmed that various aspects of mars and HURP function are conserved, including spindle stabilization and binding of Mars to Msps and PP1. However, Mars and HURP display apparently distinct spindle localizations, suggesting that there may be differences in how these proteins are used during cell division. This may reflect a wider difference in the organization of MAPs that control MT stability and the formation of bipolar spindles in flies and humans.
Spindle defects caused by lack of TACC phosphorylation or by alterations in TACC or HURP protein levels may lead to genetic instability and are implicated in cancer progression (Raff, 2002; Barros et al., 2005; Brittle and Ohkura, 2005). Our data indicate that spatially controlled dephosphorylation also plays a positive role in TACC function, suggesting that deregulation of either phosphorylation or dephosphorylation of TACC may also be involved in the molecular pathology of cancer by compromising the fidelity of chromosome segregation.
Materials And Methods
EP(2)2477, referred to here as marsP, is a homozygous viable P element insertion in the 5′ untranslated region of mars. GFP-dTACCWT and GFP-dTACCS863L (gift from J. Raff, The Gurdon Institute, Cambridge, England, UK) have been previously described (Barros et al., 2005). Other fly stains are described in FlyBase (http://www.flybase.org).
Isolation and characterization of a
Isolation of a null allele of mars by P element excision from marsP was performed as follows. Jumpstarter y w/Y; isogenic marsP/CyO, P(Delta2-3) males were crossed with y w; Tft/CyO females. From each cross, only one w revertant male, y w; mars*/CyO, in which the P element was excised, was individually crossed back to y w; Tft/CyO females. To determine the molecular lesion in the mars1 mutant, genomic DNA surrounding the original marsP insertion site was amplified from mars1 homozygotes by PCR using flanking primers and sequenced.
We used unpaired two-tailed t tests to compare mean hatch ratios of eggs from different strains and unpaired one-tailed t tests to compare mean number of mitotic spindles in wild-type and mars1 mutant embryos.
Site-directed mutagenesis and ectopic expression
marsF839A was constructed by PCR-based site-directed mutagenesis. For ectopic expression in flies, full-length marsWT and marsF839A were subcloned into pPFMW (Drosophila Genomics Resource Center, Indiana University, USA), a modified version of pUASP (Rorth, 1998) that contains an N-terminal 3xFLAG 6xMyc (FM) tag. UASP-FM-mars flies were made by P element–mediated germline transformation into a w1118 stain by Genetic Services, Inc. Embryos were provided with moderate levels of tagged MarsWT and MarsF839A by ectopic expression of UASP-FM-mars transgenes in the germline using arm-GAL4 (Sanson et al., 1996) or nanos-GAL4VP16 (Van Doren et al., 1998).
Immunoprecipitation and GST-pulldown experiments
We subjected lysates from arm-GAL4 or arm-GAL4 UASP-FM-mars flies to immunoprecipitation with dTACC (gift from J. Raff; Gergely et al., 2000) or Myc antibody (A14 rabbit polyclonal; Santa Cruz Biotechnology, Inc.). After adsorption on protein G bound to GammaBind Plus Sepharose (GE Healthcare), we analyzed immunoprecipitates and total cell extracts by immunoblotting with Myc (9E10 mouse monoclonal; Santa Cruz Biotechnology, Inc.), α-tubulin (DM1a; Sigma-Aldrich), Msps (gift from H. Ohkura, University of Edinburgh, Edinburgh, Scotland, UK; Cullen et al., 1999), PP1α87B (gift from P.T. Cohen, University of Dundee, Dundee, Scotland, UK; Helps et al., 2001), or dTACC (Gergely et al., 2000) antibodies. For GST-pulldown experiments, full-length mars ORF was subcloned into pDEST-15 (Invitrogen) for expression in E. coli in frame with an N-terminal GST tag. marsF839A was made by PCR-based site-directed mutagenesis and subcloned into pDEST-15 in the same way. Constructs were sequenced to confirm that they contained no sequence errors. To test binding to PP1, bacterial cell lysates expressing GST-tagged MarsWT or MarsF839A were incubated with arm-GAL4 UAS-HA-PP1α87B (Vereshchagina et al., 2004) D. melanogaster embryo extracts, and GST-labeled protein was precipitated with GST Bind Resin (EMD). Precipitates were examined by immunoblotting with HA antibody (12CA5; Roche). To test binding between HURP and PP1, HeLa cell nuclear extracts were subjected to immunoprecipitation with HURP antibody (gift from I.W. Mattaj, European Molecular Biology Laboratory, Heidelberg, Germany; Koffa et al., 2006). Precipitates were analyzed by immunoblotting with PP1 antibody (Helps et al., 2001).
Mars antibodies and immunofluorescence
Anti-peptide antibodies against Mars were raised by Eurogentec in rabbits by simultaneous immunization with two peptides: LVPEGTKTPPRRESN (residues 512–526) and TLRNRRVNLRPSSEFM (residues 906–921). Embryos were fixed with either methanol or formaldehyde and were processed for immunofluorescence as described previously (Huang and Raff, 1999). Colchicine treatment of embryos before fixation was as previously described (Gergely et al., 2000). Antibodies used for indirect immunofluorescence were as follows: FLAG (rabbit polyclonal; Sigma-Aldrich), KLP10A (gift from D. Sharp, Albert Einstein College of Medicine, Bronx, NY; Rogers et al., 2004), dTACC (Gergely et al., 2000), p-dTACC (gift from J. Raff; Barros et al., 2005), α-tubulin (DM1a; Sigma-Aldrich) and γ-tubulin (rabbit polyclonal or GTU88 monoclonal; Sigma-Aldrich). Secondary antibodies conjugated with Alexa Fluor 488 (Invitrogen), Cy3, or Cy5 (Jackson ImmunoResearch Laboratories) were used at 1:500–1,000 dilutions. DNA was counterstained with 1 μg/ml propidium iodide (Sigma-Aldrich).
Image acquisition and processing
Fixed embryos, mounted in 85% glycerol and 2.5% n-propylgallate in PBS, were examined using a microscope (Eclipse E800; Nikon) with either a 40× 1.3 NA Plan Fluor or a 60× 1.4 NA Plan Apo objective and a scanning confocal system (Radiance Plus; Bio-Rad Laboratories) equipped with LaserSharp 2000 software (Bio-Rad Laboratories). Images were imported to Photoshop (Adobe) and adjusted for brightness and contrast uniformly across entire fields. Quantitation of astral MTs was performed by making maximum intensity projections of 8–12 image stacks that were taken at 0.5-μm intervals from 15–90-min methanol-fixed embryos stained with α-tubulin, γ-tubulin, and DNA. The projections were imported into AQM Advance 6 software (Kinetic Imaging Ltd.), and mean α-tubulin intensities were measured within a circle around each centrosome from spindles in metaphase. The mean pixel intensity of astral MTs from at least 25 embryos of each genotype was calculated. All measurements were corrected for variation in staining conditions by subtracting the background intensity. A similar approach was taken to quantify p-dTACC staining, except that measurements were also taken at four points on the spindle, both proximal and distal to the centrosome. Linescans were generated using LSM510 software (Carl Zeiss, Inc.). The mean intensity through the center of the unprocessed spindle images, parallel to the long axis of the structure, was calculated for each channel of fluorescence by the software.
Online supplemental material
Fig. S1 shows individual channel images of Mars distribution during mitosis. Fig. S2 shows both misaligned chromosomes on mitotic spindles in embryos laid by mars mutant mothers and reduction of cold-resistant kinetochore MTs in embryos laid by mars1/P mothers. Fig. S3 shows both the spindle localization of Mars in dTACC mutant embryos and elevated p-dTACC spindle staining on a normal looking spindle from an embryo laid by mars1/P.
© 2008 Tan et al. This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.jcb.org/misc/terms.shtml). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/).
Abbreviations used in this paper: dTACC, Drosophila transforming acidic coiled coil; FM, FLAG-Myc; HURP, hepatoma up-regulated protein; MAP, microtubule-associated protein; Msps, minispindles; MT, microtubule; p-dTACC; phosphorylated dTACC; PP1, protein phosphatase 1.
We thank P.T. Cohen for PP1α87B antibody, I.W. Mattaj for HURP antibody, T. Murphy and the Drosophila Genomics Resource Center for plasmid vectors, H. Ohkura for Msps antibody, J. Raff for dTACC fly strains and antibodies, D. Sharp for KLP10A antibody, and K. Clifton and E. Taylor for technical assistance. We thank the Centre for Cell Imaging at Liverpool University for assistance with image quantitation and members of the Bennett laboratory, M. Bollen, J-Y. Huang, and J. Wakefield, for advice and critical comments on the manuscript.
This work was funded by grant BBS/B/11869 from the Biotechnology and Biological Sciences Research Council to D. Bennett, with additional support from Cancer Research UK and the Royal Society. E. Lyulcheva was supported by an Oxford University European Scatcherd Scholarship and a Phizackerley Senior Scholarship at Balliol College, Oxford. D. Bennett was supported by a Todd-Bird Research Fellowship at New College, Oxford.