CENP-meta, an Essential Kinetochore Kinesin Required for the Maintenance of Metaphase Chromosome Alignment in Drosophila

Precise chromosome partitioning during mitosis requires a specialized microtubule-based structure, the mitotic spindle. The spindle interacts with chromosomes most strongly via a protein complex called the kinetochore, which forms at the centromere of each chromosome during cell division (see McIntosh and Hering 1991; Nicklas 1997; Rieder 1982; Rieder and Salmon 1998). A number of important kinetochore components have been identified. Among these are several motor proteins from the kinesin superfamily including CENP-E (Yen et al. 1992; Wood et al. 1997), the MCAK/XKCM1 family (Wordeman and Mitchison 1995; Walczak et al. 1996; Maney et al. 1998), and cytoplasmic dynein and its associated components (Steuer et al. 1990; Pfarr et al. 1990). Besides motor enzymes, a number of other proteins are at least transiently associated with kinetochores during mitosis. These include rod (Karess and Glover 1989; Starr et al., 1998), the MAP kinases ppERK (Shapiro et al. 1998; Zecevic et al. 1998) and ppMEK (Shapiro et al. 1998), and an APC activator protein fizzy (called cdc20/cdh1 in other organisms; Kallio et al. 1998). Likewise, the protein kinase polo (Logarinho and Sunkel 1998) and a number of components of the highly conserved spindle assembly checkpoint, including Bub1 (Taylor and McKeon 1997; Taylor et al. 1998; Basu et al. 1999), Bub3 (Taylor et al. 1998; Basu et al. 1998; Martinez-Exposito et al. 1999), Mad1 (Jin et al. 1998; Chen et al. 1998), and Mad2 (Waters et al. 1998; Chen et al. 1996, Chen et al. 1998), are transiently kinetochore-bound.

CENP-E was first identified as a protein that is present on the kinetochores of mitotic cells during chromosome movement (Compton et al. 1991; Yen et al. 1991). The sequence of mammalian CENP-E revealed significant sequence similarity between the NH₂-terminal domain of the 312-kD CENP-E protein and the motor domain of members of the kinesin superfamily (Yen et al. 1992). Mammalian CENP-E accumulates in the cytoplasm of cells in late G2, and upon nuclear envelope breakdown, it rapidly associates with kinetochores where it remains until the end of anaphase-A (Brown et al. 1996). CENP-E then relocates to the spindle midzone where it concentrates in the developing midbody before nearly quantitative degradation at the end of mitosis (Brown et al. 1994).

The presence of this kinesin-like motor at the kinetochore during mitosis makes CENP-E an obvious candidate for providing motive force for any of a number of different chromosome movements. Consistent with this view, microinjection of cells with antibodies specific for CENP-E caused a delay in anaphase onset (Yen et al. 1991) and a misalignment of chromosomes (Schaar et al. 1997). Similarly, both immunodepletion and inhibitory antibody addition experiments with Xenopus cell-free extracts caused failure in chromosome alignment at metaphase (Wood et al. 1997), while antisense oligonucleotide meditated suppression of CENP-E accumulation in mammalian cultured cells yields chronically mono-oriented chromosomes with spindles flattened along the plane of the substrate (Yao et al. 2000).

Additional observations suggest the possibility that CENP-E may have other roles during mitosis. For example, the relocalization of CENP-E to the midzone at anaphase-B may indicate an additional role for this plus-end motor (Wood et al. 1997) in spindle elongation and/or cytokinesis. CENP-E has also been implicated as part of, or a target of, the spindle assembly checkpoint based on a physical interaction with the spindle assembly protein hBubR1 (Chan et al. 1998, Chan et al. 1999; Yao et al. 2000). Indeed, in Xenopus extracts CENP-E is required for establishing and maintaining this checkpoint (Abrieu, A., J.A. Kahana, K.W. Wood, and D.W. Cleveland, manuscript submitted for publication). CENP-E is also a target of phosphorylation by the map kinase ppERK, an active kinase found at the kinetochore (Shapiro et al. 1998; Zecevic et al. 1998). CENP-E has also been implicated in the dynamic attachment of kinetochores to disassembling microtubules in vitro (Lombillo et al. 1995). Together, the available data suggest a general model in which plus-end directed kinesins, such as CENP-E (Wood et al. 1997), are required during chromosome congression to power the movement of the trailing kinetochores towards the unstable microtubule plus ends. In addition, such kinesins may tether the kinetochore to the spindle microtubules (Lombillo et al. 1995; Wood et al. 1997) and perhaps play a crucial role in signaling the mitotic checkpoint machinery about the status of chromosome capture and/or alignment.

To examine the role of kinetochore-associated kinesin motors in mitosis in vivo, especially in distinguishing whether kinetochore kinesins like CENP-E are required for the establishment of chromosome congression or its maintenance once chromosomes are properly aligned, we have now used genetic methods to determine the in vivo consequences of removal of one of two kinesin relatives of CENP-E in Drosophila.

A small region corresponding to amino acids (aa) 81–130 of the amino terminus of both CENP-meta and CENP-ana was identified in a PCR-based screen for new members of the kinesin superfamily in Drosophila. A 198-bp oligonucleotide probe (58 nucleotides [nts] of intron 2 followed by codons for aa 87–130 of both proteins) was synthesized corresponding to this sequence and used to probe a Drosophila genomic P1 library filter (Genome Systems, Inc.). All DNA hybridizations were performed at 60°C according to Church and Gilbert 1984. This probe identified two P1s, DSO5055 and DSO3675, both of which map to chromosome 2L(32D-E). The P1 clones, DSO5055 and DSO3675, were generously provided by the Karpen Lab (Salk Institute, La Jolla, CA) and all subsequent subcloning was done using P1 DSO5055. A genomic contig was assembled, using overlapping hybridization and sequence alignments, for the region of DSO5055 that contains the entirety of CENP-meta and nucleotides (nts) 1–4,162 of CENP-ana. Additional P1 clone (DSO5055) and BAC clone R43K24 covering this region were sequenced by the Berkeley Drosophila Genome project (BDGP). These allowed us to assemble a >300-kb genomic contig covering this region of chromosome 2 (see Fig. 1, see also BDGP contig B28C1-33E8).

cDNA clones for CENP-meta and CENP-ana were isolated from a Drosophila embryonic λZAP library (Stratagene) and a Drosophila 0–4 h embryonic library (a kind gift from Nick Brown; Brown and Kafatos 1988). Some portions of cDNA were identified using reverse transcription of total RNA isolated from S2 cells (Schneider 1972) or embryos with SuperScript II (GIBCO BRL), followed by PCR amplification using either the proof-reading ELONGase enzyme (GIBCO BRL) or PFU polymerase (Stratagene) and subcloning. All clones were sequenced using automated sequencing (Applied Biosystems, PerkinElmer). Sequences were compiled using DNA Strider software.

Complete predicted coding sequences for CENP-meta and CENP-ana are available from GenBank/EMBL/DDBJ under accession numbers AF220353 and AF220354, respectively.

P-element lines from the Bloomington stock center and Tod Laverty at BDGP were screened for insertions at 32E in the CENP-meta gene. Genomic sequences around the P-element were isolated by plasmid rescue. Plasmid rescue was used to identify P-element l(2)04431 within the CENP-meta region. Subsequent sequencing of the rescued sequences and of inverse PCR (invPCR) products (BDGP protocol) containing the same flanking sequences allowed precise assignment of the insertion point of l(2)04431 within exon 10 of CENP-meta gene. Excision, transposition, and deletion lines were obtained by crossing the l(2)04431 strain into a background containing the Δ2-3 transposase source. Transposition of l(2)04431 in line CENP-meta^Δ deleted ∼5 kb of genomic sequence in the CENP-meta gene just upstream of the 5′ end of the initial P-element insertion site.

To generate germ-line clones of CENP-meta, cmet-FRT chromosomes were constructed by recombination of P[ry+;neo^R-FRT] 40A; ry with cmet/SM1; ry lines. Germ-line clones were induced using FRT-ovoD1 and hsFLPase following Chou and Perrimon 1996. Clones containing chromosomes illuminated by histone-2A-var-GFP (a kind gift of Rob Saint, University of Adelaide; Clarkson and Saint 1999) were generated using FRT-cmet/SM1;hisvar-GFP/TM6B and hsFLPase/Y, FRT-ovoD1/CyO. Real-time movies were made using manually dechorionated embryos in heptane glue (Santamaria, 1986). A Biorad MRC1024 laser-scanning confocal microscope was used to collect an image (average of three scans) every 10 s.

Total RNA was isolated from embryos and female adults using Trizol reagent (GIBCO BRL). 15 μg of total RNA from adult females and 30 μg of total RNA from embryos from a variety of strains were run on a 1.0% denaturing agarose/formaldehyde gel as described in Current Protocols in Molecular Biology. The RNA was transferred to Hybond (Amersham Pharmacia Biotech) and hybridized in Church and Gilbert buffer at 60°C, overnight. Initially the filter was hybridized with a 686-bp BglII-BamHI fragment (nt 6,248–6,934) specific for CENP-meta. After exposure to film the filter was stripped until there was no detectable signal. The filter was then hybridized with a 3-kb probe to the Drosophila mcm5 gene (Su et al. 1997) to verify equal loading. It was then stripped and probed with a 2.4-kb fragment (nt 3,022–5,425) from cDNA clone λZAP 4-2, which recognizes CENP-ana (and weakly cross-reacts with CENP-meta), then stripped and probed for PPT-2 with a genomic PCR fragment corresponding to the first four exons of PPT-2. All blots were exposed to XAR-XOMAT film (Kodak) at −80°C with intensifying screens.

The vector pQE 30 (QIAGEN) was used to express an NH₂-terminal hexahistidine fusion with aa 285–706 of CENP-meta. BamHI ends were added to a PCR fragment (nt 963–2,231) of CENP-meta λZAP 3-1 subclone by PCR with Pfu polymerase (Stratagene). The PCR product was digested with BamHI, ligated to BamHI digested pQE30, and the resulting plasmid, pQE3-1, was verified by sequencing. Expression constructs were transformed into E. coli strain M15 (QIAGEN). For protein expression, cells were grown to an OD₆₀₀ of 0.7, fusion protein was induced by the addition of isopropylthiogalactoside (IPTG) to 1 mM, and the cells were grown for an additional 5 h at 37°C. Cells were pelleted, lysed in 8 M urea and the fusion protein purified over a Ni-NTA agarose (QIAGEN) column according to the QIAGEN protocol.

Two rabbit antisera (nos. 6583 and 6584) were raised to an internal peptide sequence of CENP-meta. The peptide, SDKGQQKRRRTWC (aa 385–397), was synthesized by Genosys Biotechnologies, Inc., and coupled to KLH via the COOH-terminal cysteine. The KLH-coupled peptide was used to immunize two rabbits at Lampire Biological Laboratories. Antiserum from rabbit 6584 had a significantly high titer against the CENP-meta fusion protein (pQE3-1) and was used for all subsequent work. Antibodies were affinity-purified over a column containing the appropriate fusion protein coupled to cyanogen bromide activated Sepharose (Amersham Pharmacia Biotech), eluting with 200 mM glycine-HCl (pH 2.3) + 0.5 M NaCl. The antibody was neutralized and concentrated into PBS (1.8 mM NaH₂PO₄, 8.4 mM Na₂HPO₄, 10 mM KCL, and 150 mM NaCl) using an Ultrafree-15 centrifugal filter (Millipore), stabilized with 40% glycerol, and stored at 4°C.

Schneider (S2) cells (Schneider 1972) were obtained from Invitrogen and grown at room temperature in DES Complete medium with l-glutamine (Invitrogen), supplemented with 10% heat-inactivated fetal bovine serum (Hyclone).

CENP-meta was immunolocalized in S2 cells attached to coverslips precoated for 5 min at room temperature with 0.1% poly-d-lysine. Nonadherent cells were washed off coverslips in PBS and attached cells were fixed in methanol for 10 min at −20°C. Cells were washed in PBS and incubated for 1 h in blocking buffer (0.2 M glycine, 2.5% fetal bovine serum, and 0.1% Triton X-100, in PBS). Primary antibody incubations were done using 1:100 dilution (∼14 μg/ml) affinity-purified CENP-meta antisera no. 6584 either alone or in combination with 1:100 dilution of mouse anti-alpha tubulin monoclonal antibody DM1A (Sigma-Aldrich) for 1 h at room temperature. Cells were washed in PBS and incubated in ALEXA-Red–conjugated goat anti–rabbit and/or ALEXA-Green–conjugated goat anti–mouse secondary antibodies (Molecular Probes) for 1 h at room temperature. Coverslips were washed, incubated with Hoechst 33528 for 2 min, washed again, and mounted in Citrofluor (Ted Pella). Standard fluorescent images were collected using a Princeton Instrument cooled CCD mounted on a Zeiss Axioplan microscope controlled by Metamorph software (Universal Imaging Corp.). Deconvolved images were acquired on a Leica DMRXA/RFA/V automated microscope with a Cooke Sensicam digital camera and processed using Slidebook software (Intelligent Imaging Innovations). Additional image processing was performed using Metamorph (Universal Imaging Corp.) and Adobe Photoshop software.

Whole brains from 3rd instar larvae were fixed and immunostained with monoclonal antibody YL1/2 (1:10 dilution of a culture supernatant; Harlan Bioproducts) to visualize tubulin as described in Glover and Gonzalez 1993. To localize CENP-meta, embryos were fixed in −20°C methanol and stained with the no. 6584 antibody to CENP-meta. Embryos from germ-line clones were fixed with 37% formaldehyde and treated with the tubulin and Rb188 antibodies (to stain centrosomes) as well as DAPI. These samples were observed using a Bio-Rad MRC1024 laser-scanning microscope.

Cells visible in “squash” preparations of larval brains were analyzed following protocol 14 of Glover and Gonzalez 1993. Squashed preparations were then stained with 0.5 mg/ml DAPI and viewed on a Zeiss Axiophot. A field is defined by the area visible using such optics. At least 300 fields from a minimum of eight brains were examined for each genotype and the results were plotted as figures per field in order to minimize effects of different numbers of cells per field. To immunostain chromosomes cooled to dry ice temperature in brain squashes, the samples were dehydrated in methanol for 30 min, then rehydrated in PBS + 0.1% Triton X-100 and processed for immunofluorescence as described for whole brains.

To prepare extracts for immunoprecipitation (IP), 1.8 × 10⁷ S2 cells per IP were pelleted and washed once in PBS. Cell pellets were then resuspended in 1 ml RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, and 50 mM Tris, pH 8.0) plus protease inhibitors (0.5 mM AEBSF and 10 μg/ml each aprotinin, pepstatin A, leupeptin, and soybean trypsin inhibitor) and passed 10 times through a 26-gauge needle to shear DNA. The lysate was then incubated on ice for 3 h. To the lysate, 10 μl of affinity-purified (1.4 mg/ml) antibody or 1 μl of pre-immune serum (∼16 mg/ml) was added, and the lysates were rotated for 2 h at 4°C. 20 μl of Pansorbin beads (Calbiochem-Novabiochem) were added and lysates rotated for 2 h at room temperature. Pansorbin beads were spun down and washed 3× in RIPA buffer and then resuspended in 4× sample buffer (50 mM Tris, pH 6.8, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, and 0.05% bromophenol blue), boiled 10 min, spun 5 min, and loaded onto 10% SDS-PAGE gel. Protein was transferred to nitrocellulose BA83, as previously described (Harlow and Lane 1988), and then processed for ECL as per manufacturer's instructions (Amersham Pharmacia Biotech). For immunoblots, affinity-purified no. 6584 was used at 14 μg/ml, HRP-conjugated goat anti–rabbit antibody was used at 1:20,000–50,000, and ECL development reagents were used at 0.5× the recommended concentration.

Real-time movies were made using manually dechorionated cmet clone embryos in heptane glue. A Bio-Rad MRC1024 laser-scanning confocal microscope was used to collect an image (average of three scans) every 10 s. Files containing real-time series of images were uploaded to NIH image and saved as quicktime movies. Videos 1–3, further depicting Fig. 6 and Fig. 7 are. To ensure a good resolution of the movies, please check that the monitor of your computer is set on millions of colors or true colors (32 bits). The videos show embryos from which images were taken to form Fig. 6 and Fig. 7. Refer to the respective figure legends for further explanation.

Degenerate PCR, followed by cDNA and genomic library screening, was used to identify two new Drosophila kinesins (Fig. 1 A), both of which map to chromosome 2 at the cytological position 32D-E. We designated these genes CENP-meta (cmet) and CENP-ana (cana). The predicted CENP-meta protein is 2,244 amino acids (257 kD), encoded by a gene comprised of 11 exons and that shares 55 nt of 3′ untranslated region (utr) with the 3′ utr of the abrupt gene, which is transcribed in the opposite direction (Fig. 1 A). The CENP-meta sequence contains a putative nuclear localization sequence (aa 2,185–2,190) and a predicted tyrosine phosphorylation site (aa 962-970). The CENP-ana gene encodes a predicted protein of 1,931 aa (222 kD).

The predicted protein structures of both CENP-meta and CENP-ana are similar to those of a variety of kinesins, including CENP-E proteins from human (HCENP-E) and Xenopus (XCENP-E). CENP-meta and CENP-ana both contain an NH₂-terminal kinesin-like motor domain linked to a small globular tail domain by a rod domain predicted to form a long discontinuous coiled-coil (Fig. 1 B). The proteins encoded by CENP-meta and CENP-ana show considerable sequence similarity throughout their motor and stalk domains (42% identity overall), but they lack any significant sequence similarity to HCENP-E and XCENP-E in their the stalk and tail domains. Nonetheless, a variety of sequence comparison algorithms (ALIGN, NCBI BLAST, GCG PILEUP, and BCM multiple sequence alignment tools, see Materials and Methods) revealed that the motor domains of CENP-meta and CENP-ana are more similar to the motor domains from HCENP-E and XCENP-E than they are to other kinesins (our observations and Case, R., University of California, San Francisco, personal communication). Thus, CENP-meta and CENP-ana appear to be members of the same family of kinesins as HCENP-E and XCENP-E.

To determine the localization of CENP-meta throughout the cell cycle, a polyclonal antiserum was raised to a 13-mer peptide (SDKGQQKRRRTWC) from the motor domain of CENP-meta (meta peptide aa 385–397, Fig. 1 A). Affinity-purified serum specifically recognizes and immunoprecipitates two protein species in extracts made from either S2 cells or from whole animals. The major species is a protein of ∼180 kD (Fig. 2 C, lane 2); a second species of ∼250 kD, which corresponds to the predicted full-length protein constitutes only a minor component (Fig. 2 C, lane 2, asterisk). The 250-kD band, in both unfractionated S2 cell lysates and immunoprecipitations, is extremely sensitive to proteolysis. It is only recovered from lysates prepared in the presence of very high concentrations of protease inhibitors and can only be easily detected after concentration of the protein by immunoprecipitation from large cell numbers. Preimmune serum does not recognize these bands in lysates from S2 cells (data not shown) nor does it immunoprecipitate either protein species (Fig. 2 C, lane 1). Hence, the 250-kD band is likely to be the full-length CENP-meta protein, which is reduced by proteolysis to the lower molecular mass form. These polypeptides represent CENP-meta, and not CENP-ana, since immunoblotting against a fusion protein that contains the corresponding domain of CENP-ana (aa 230–665) reacts with the antibody ∼140 times more weakly than the analogous fusion protein from CENP-meta (not shown).

In Drosophila S2 tissue culture cells (Fig. 2 A), syncytial embryos (Fig. 2 B), and squashed larval brain chromosome preparations (Fig. 2 D), immunolocalization with affinity-purified antibody revealed that CENP-meta is associated with the centromeric region of chromosomes, presumably kinetochores, during metaphase. This was most apparent in isolated chromosomes (Fig. 2 D), where CENP-meta was restricted to the primary constriction of each chromosome pair, a finding consistent with the localization of CENP-E to the corona fibers of the kinetochore (Cooke et al. 1997; Yao et al. 1997). Preimmune serum gave only faint (non-specific) cytoplasmic staining regardless of the stage of the cell cycle.

Examination of cycling cells revealed that CENP-meta is associated with kinetochores during all stages of the cell cycle (Fig. 2 A), a behavior that differs from that of HCENP-E and XCENP-E, which have been reported to associate with the kinetochore only upon nuclear envelope breakdown, and then is degraded at the end of mitosis (Brown et al. 1996; Wood et al., 1996). During interphase, CENP-meta is found as a tight cluster of approximately four nuclear dots, perhaps revealing a kinetochore structure that remains intact through the cell cycle on each of the four Drosophila chromosomes. At mitosis, as chromosomes congress to the metaphase plate, CENP-meta lies at the junction between the chromosomes and the ends of the microtubules along the metaphase plate. During anaphase and telophase, as sister chromatids and then daughter cells separate, CENP-meta is again present as a cluster of four or so dots, just as in interphase cells. At no time during the cell cycle was CENP-meta observed anywhere but the centromeric/kinetochore region of the chromosomes.

Late larval lethal P-element insertions were previously mapped to the area of chromosome 2 around 32D-E (Spradling et al. 1998). Using plasmid rescue and inverse PCR the precise insertion site of P-element l(2)04431 was mapped to the 5′ end of exon 10 (nt 5,986), which lies at the beginning of the coding region for the globular tail domain of CENP-meta. Additional alleles were generated by transposase-mediated mobilization of the P-element during male meiosis. This yielded several new alleles, including cmet^Δ and cmet^lx1. Since CENP-meta and abrupt overlap, we confirmed that cmet^Δ and cmet^lx1 fully complement abrupt null mutants. We also found that the lethality of the cmet⁰⁴⁴³¹ chromosome can be reverted by precise excision of the l(2)04431 P-element; hence, no other lethal mutations reside on this chromosome.

Blots of total RNA from adult females and embryos of various genotypes were probed for CENP-meta, CENP-ana, and the nearby PPT-2 gene (as well as the mcm5 gene as a loading control). The results demonstrate that cmet^Δ is a null allele for CENP-meta at the RNA level (Fig. 3, upper panels), but that this allele does not affect the size or abundance of RNAs for CENP-ana (middle panels) or PPT-2 (data not shown).

To determine the phenotype of mutations in CENP-meta third instar larval brains from animals homozygous or heterozygous for CENP-meta mutations were examined. Analysis of at least 300 fields of fixed, squashed brains from cmet^lx1 or cmet^Δ animals yielded a striking increase in the fraction of mitotic cells compared with phenotypically wild-type controls (Fig. 4 E). While the squashing procedure makes strict interpretation of chromosome alignment difficult, the mitotic cells from mutant animals very frequently contained misaligned chromosomes (Fig. 4, A–C). These data suggest a role for CENP-meta in chromosome congression, or the maintenance of that congression, in agreement with the model proposed earlier from experiments showing misalignment of chromosomes in spindles assembled in vitro using Xenopus egg extracts depleted of CENP-E (Wood et al. 1997). Additionally, the chromosomes sometimes appear to be undercondensed, perhaps as a secondary consequence of mitotic arrest. Other mitotic abnormalities, such as premature sister chromatid separation, circular figures and polyploidy, were no more numerous in homozygous mutant brains than in heterozygous controls. Heterozygotes were indistinguishable from homozygous wild-type strains for all parameters tested.

To ensure that the chromosome positioning defects seen in CENP-meta mutants did not arise from mechanical disruption during squash preparation, and to examine spindle morphology, whole larval brains were examined using indirect immunofluorescence and laser-scanning confocal microscopy. Chromosomes deficient for CENP-meta still apparently bound to bipolar spindles, presumably through kinetochore attachments, but as before, the chromosomes were not as tightly arrayed on a metaphase plate (Fig. 4 F) as in wild-type (Fig. 4 G).

Using techniques as described above, a second late larval lethal P-element insertion, l(2)00716, was sited in the CENP-ana gene (Fig. 1 A). As with embryos carrying mutations in CENP-meta, embryos carrying the P-element displayed an increased mitotic index when fixed, squashed brains were analyzed. However, rather than an increased frequency of prometaphase/metaphase seen with CENP-meta mutants (Fig. 4 E), a dramatic increase in anaphase figures was detected (data not shown), suggesting premature anaphase entry or delay during anaphase. Initial efforts to generate CENP-ana–specific antibodies and additional CENP-ana alleles were not successful; more detailed analysis of the functional characteristics of CENP-ana, thus, awaits identification of such reagents.

That CENP-meta mutants can survive until pupal stage may reflect a maternal pool of CENP-meta sufficient for early development but that is exhausted during larval cell division. To ascertain whether CENP-meta has an essential role at earlier developmental stages, a strain was constructed in which the CENP-meta product was removed from early embryos as a consequence of cmet^lx1 gene disruption in the female germline (referred to hereafter as cmet^lx1 mutant embryos). Examination of fixed and stained cmet^lx1 embryos revealed that chromosomes failed to align or maintain alignment (Fig. 5B and Fig. E) and that there were numerous polyploid nuclei as judged by increased intensity of staining with DNA binding dyes. Such defects were observed as early as the eighth embryonic mitosis and became more prominent during later cell cycles in the early Drosophila syncytia. Anaphase defects, such as unequal anaphases lacking centrosomes, were also observed at low frequency (<10% of mitoses). These latter cases were consistent with secondary defects arising from cell cycle progression in the absence of proper chromosome positioning.

Further, as expected for an early mitotic defect of this type (e.g., the dal mutant; Sullivan et al. 1990), the absence of CENP-meta correlated with aberrantly aligned chromosome sets, and these fell into the interior of the embryo, leaving their centrosomes behind (examples of orphan centrosomes indicated by yellow arrowheads in Fig. 5B and Fig. E). Examination of embryos derived from germlines homozygous for cmet^Δ and constructed by similar means (see Materials and Methods) was more difficult. Such embryos were produced at much lower frequencies and 20% of these embryos showed defects, such as unusual overall shape, which are strongly suggestive of abnormal oogenesis. However, in the few embryos obtained, prominent chromosome alignment defects could be seen (Fig. 5C and Fig. F), as well as many nuclei with an abnormally high chromosome content. CP190, usually a marker for centrosomes (Fig. 5A and Fig. B; arrowed nucleus in Fig. 5 C), became redistributed around the chromatin (Fig. 5 C, arrowheads).

To examine more directly how the absence of CENP-meta affects chromosome movement, chromosomes in CENP-meta deficient embryos were marked by introducing a stably inherited histone-2A-variant-GFP (Clarkson and Saint 1999). Real-time images of these embryos (videos available) revealed that congression of the fluorescently marked chromosomes appears to proceed relatively normally in cmet^lx1 mutant embryos, but the chromosomes do not stably persist at the metaphase plate. Sporadically they move off the plate before anaphase onset (e.g., pseudocolored chromosomes in Fig. 6, A–C; see also supplemental movies corresponding to the images for Fig. 6, A–C). These are errors in the maintenance of congression because this misalignment usually occurred after initial congression and always before the disjoining of sister chromosomes.

Examination of movies from 16 embryos left no doubt that anaphase onset takes places on schedule, and moreover, it takes place simultaneously for both aligned and misaligned chromosomes. This can be clearly seen for several nuclei where a chromatid pair closer to one pole than the other disjoins contemporaneously with its aligned brethren, and subsequently one chromatid passes across the metaphase plate en route to the more distant pole (e.g., blue or pink pseudo-colored chromosomes; Fig. 6A and Fig. B). For other nuclei (Fig. 6 C), a misaligned chromosome never disjoins, presumably reflecting the failure of one of the kinetochores of the chromosome pair to reassociate successfully with microtubules from the opposite pole. It is probably this mis-segregation event that results in the previously noted aneuploidy and lethality. In no case did the absence of CENP-meta result in premature sister chromatid separation of aligned or misaligned chromosomes and the duration of anaphase was not significantly extended. In cells of wild-type embryos carrying the histone-2A-var-GFP, chromosomes stay aligned at the metaphase plate and sister chromatids segregate cleanly (Fig. 6 D).

Consistent with the more severe defects observed with fixed examples of cmet^Δ, movies made with this allele revealed defects at an earlier nuclear cycle (in many cases before nuclear migration to the cortex was complete) and a higher proportion of polyploid cells. Again, maintenance of chromosome congression was clearly aberrant (Fig. 7; videos further depicting these data available). For example, in the series of images taken over 300 s in Fig. 7 B, tight metaphase alignment of most chromosomes was not maintained, with one chromosome (pseudo-colored green) markedly losing alignment and staying misaligned through initiation of anaphase (at between 150 and 200 s). In this example, upon anaphase onset, both sister chromatids segregated to the same pole. This is also seen in many other examples as illustrated in Fig. 7 A (see chromosome pseudo-colored red) and Fig. 7 C (see chromosome pseudo-colored orange). Indeed, in the cmet^Δ embryos, examples where misaligned chromatids successfully disjoined and segregated separately at anaphase were much more rare (1 of 12) than for cmet^lx1 (7 of 36). Anaphase duration was not detectably extended.

The localization of CENP-meta to the kinetochore of chromosomes and the phenotypes of mutants in it suggest that this motor may be a functional homologue of vertebrate CENP-E. That the product of a second gene, CENP-ana, shares a similarly high degree of sequence identity in the motor domain, but no sequence similarity in the tail domains is reminiscent of the bimC family of kinesins, in which members are unambiguously identified by functional analysis, even though some lack similarity in their tail domains. That there are two CENP-E–like proteins in flies may anticipate the situation in mammals. Whereas a single chromosomal locus at 4q24 has been reported for human CENP-E (Testa et al. 1994), a non-syntenic location for CENP-E on mouse chromosome 6 has been reported (Fowler et al. 1998), and interspecies back-crossing in mice has identified CENP-E–related loci on both chromosomes 3 and 6 (the former of which is syntenic with human 4q24; Putkey, F., D.W. Cleveland, N. Jenkins and N. Copeland, unpublished observations).

To earlier perturbations of CENP-E function, all of which lead to a disruption in chromosome alignment, our real-time observation of mitosis in mutants lacking CENP-meta demonstrates a specific role for CENP-E in maintenance, as opposed to establishment of alignment. Whether CENP-ana has a similar function has not yet been determined; nevertheless, it is readily apparent that the functions of CENP-meta and CENP-ana cannot be identical, since removal of CENP-meta alone causes mitotic disturbance and lethality. What emerges clearly from our data is that multiple motors must be required to generate a stable metaphase alignment since removal of only one, CENP-meta, allows initial congression of many chromosomes, but disturbs its maintenance. In this view, the role of CENP-meta in maintenance of a metaphase plate is to generate a balancing force at the kinetochore, pushing the chromosome away from the pole.

Two additional arguments support the conclusion that CENP-meta is not required to maintain the attachment of microtubules to kinetochores. First, most chromosomes obviously retain bi-oriented attachment to microtubules in the absence of CENP-meta, as seen by the proper disjunction of the majority of chromosomes at anaphase onset. Second, among those chromosomes that do not maintain normal alignment, leaving the metaphase plate abnormally in the absence of CENP-meta, there are several examples where the misaligned chromatid pair disjoins at the normal timing of anaphase with one sister successfully travelling to each pole. Thus, even the misaligned chromosomes in some cases appear to retain (or recover) bipolar attachment, suggesting that failure to remain at the metaphase plate is not solely a result of losing attachment to one of the two poles.

Despite the demonstration of a plus-end–directed, ATP-dependent microtubule motor activity for both an NH₂-terminal fragment of CENP-E (Wood et al. 1997) and the full-length polypeptide expressed using baculovirus (Kahana, J.A., and D.W. Cleveland, unpublished), it has been suggested that kinetochore kinesins of the CENP-E family might be associated with a minus-end–directed microtubule motor activity (Thrower et al. 1995) or might couple chromosomes to depolymerizing microtubules under some circumstances (Lombillo et al. 1995). Thus, CENP-E related motors may be important kinetochore components for generating anaphase chromosome movement. Two observations suggest that CENP-meta either does not have such an activity, or that it is not essential for anaphase: (a) anaphase chromosome movement appears normal in mutants lacking CENP-meta; and (b) some misaligned chromosomes that departed the metaphase plate prematurely in cmet^Δ cells can still traverse the entire spindle at anaphase. These chromosomes act as though they not only retained an attachment to the far pole, but that the attachment could properly support, or generate, minus-end–directed movement. Thus, CENP-meta, and by extension CENP-E, may not be essential components of the minus-end force generating mechanism, although they could still play a redundant role with dyneins, MCAK, or other CENP-E relatives such as CENP-ana.

In yeast and vertebrate systems, drugs that cause spindle damage have revealed the presence of a checkpoint that prevents sister chromatid separation when the microtubule spindle is aberrant or missing. The checkpoint is mediated through a diffusible signal of Mad2/Cdc20 released from unattached kinetochores and which inhibits activation of the anaphase promoting complex required for the degradation of regulators of chromatid linkage (e.g., Fang, 1999; Morgan 1999). In mammals, the checkpoint kinase BubR1 (Cahill et al. 1998; Taylor et al. 1998) has been shown to interact with CENP-E (Chan et al. 1998, Chan et al. 1999), forming a nearly stoichiometric, soluble complex with CENP-E in mitotically arrested cells (Yao et al. 2000). Moreover, immunodepletion of CENP-E from frog extracts that can cycle between interphase and mitosis eliminates the ability to activate or maintain the mitotic checkpoint (Abrieu, A., J.A. Kahana, K.W. Wood, and D.W. Cleveland, manuscript submitted for publication) further implicating CENP-E, or one of its relatives, as an essential component for directly activating and/or silencing kinetochore-dependent, BubR1-dependent, signaling that arrests the cycle before anaphase.

For Drosophila, treatment of dividing Drosophila cells and mitotic syncytial embryos with microtubule depolymerizing agents, such as colchicine and nocodazole, has been reported to yield mitotic arrest with high levels of cyclin. In syncytial embryos the arrest is coordinate, and chromatin is trapped in a prometaphase-like configuration (Zalokar and Erk 1976; Foe and Alberts 1983; Foe et al. 1994). Our data suggest that this system is defective in mutants lacking CENP-meta. This tentative conclusion is based on the observation that mislocalized chromosomes, some of which seem not to be stably bi-oriented, do not appear to cause a delay in anaphase onset. However, the robustness of the checkpoint in these early cycles (i.e., relative to that in typical somatic divisions) remains to be firmly established. Thus, further work is needed to resolve how CENP-E and its relatives participate in this checkpoint and whether it is required to detect all types of mitotic abnormalities.

The authors wish to thank the members (both past and present) of the Cleveland, Goldstein, McIntosh, Winey, and Karpen labs for excellent technical and generous scientific advice. In addition, we wish to thank Harold A. Fisk and Hilary Snaith for their unwavering emotional support and many years of critical scientific discussions.

This work was supported by grant GM29513 to D.W. Cleveland and funds from the Howard Hughes Medical Institute to L.S.B. Goldstein. D.W. Cleveland receives salary support from the Ludwig Institute for Cancer Research. L.S.B. Goldstein is an Investigator of the Howard Hughes Medical Institute. J.K. Yucel was supported by a postdoctoral fellowship from the National Institutes of Health and from the Ludwig Institute for Cancer Research. A.J.V. Philp was supported by a Fulbright Cancer Research Fellowship and by the Howard Hughes Medical Institute.