The Drosophila Gene abnormal spindle Encodes a Novel Microtubule-associated Protein That Associates with the Polar Regions of the Mitotic Spindle

Chromosomal material corresponding to polytene map position 96A2196B10 was microdissected from Oregon R third instar larval salivary gland polytene chromosomes. Two fragments were microdissected from two chromosomes and were pooled. The DNA was extracted and cloned, as described (Scalenghe et al., 1985; Saunders et al., 1989), in the lambda insertion vector NM1149 (Murray, 1983). The recombinant clones were screened with labeled genomic DNA to identify clones containing repetitive DNA. Clones that appeared to be single-copy DNA were mapped by in situ hybridization to verify that they were derived from the asp region. A second chromosome microdissection was carried out using PCR amplification, as described previously (Saunders et al., 1989). In this case, two serial sections were performed through the 96A21-96B10 region of one chromosome. The amplifications yielded pools of DNA fragments of mean size 300 bp, as expected for Sau3A digestion. Microclone inserts and pools of DNA were labeled by random oligonucleotide priming and used to screen the cosmid libraries.

Genomic clones were isolated from lambda libraries in Lambda dash (Stratagene, La Jolla, CA), and cosmid libraries were those constructed using the vectors Smart2 (Speek et al., 1988) or Lorist6 (Siden-Kiamos et al., 1990). The initial screening was carried out with cloned DNA arising from the first microcloning experiment. Subsequent steps of the walk used terminal RNA probes synthesized from the promoters at the ends of the cloning vectors. Periodically, the progress of the walk was monitored by in situ hybridization as described (Saunders et al., 1989). In situ hybridization of fragments from the chromosome walk to In(3R+3L)ats chromosomes (with which the deficiency-bearing In(3R)Ubx^7LLats^R was synthesized; Gonzalez et al., 1989) enabled the location of the distal breakpoint of the deficiency In(3R)Ubx^7LLats^R to be determined. To map the distal breakpoint of Df(3R)Hd^γ1, in situ hybridization with fluorescent labels (Saunders, 1994) was used to visualize signals derived from one or both homologues of 3R.

cDNAs were subsequently isolated by using segments of the walk to screen a testis cDNA library, kindly provided by T. Hazelrigg that had been constructed by a combination of random and oligo-dT priming in the vector λZAP and also embryonic cDNA libraries (provided by Brown, N.; Brown and Kafatos, 1988).

Rescue experiments were carried out using two Drosophila melanogaster lines containing homozygous second chromosome insertions of the P element transformation vector plasmids pMBO1366 and pMBO1367, respectively (Shimell et al., 1991). Plasmid pMBO1367 contains an 18-kb SalI fragment that includes both the tld and asp transcripts; pMBO1366 contains a 14-kb SalI fragment that includes the tld transcript but only the 5′ half of the asp transcript. Females with the genotype +/+; asp¹ red/TM6B were crossed to 1367/1367; +/+ males. 1367/+; asp¹ red/+ males were selected from the F1 progeny and crossed to +/ +; asp¹ red/TM6B females to obtain asp¹ red/asp¹ red individuals. 1367; asp¹/asp¹ animals did not show any morphological defect and were fully fertile, indicating that the 1367 insertion rescues the asp¹ homozygous lethal phenotype. As a control, the same crosses were carried out using 1366 transformants. This insertion failed to rescue the asp¹ mutation.

Sequencing was carried out using the dideoxynucleotide chain termination procedure (Sambrook et al., 1989). Template clones for sequencing were produced by exonuclease III deletion (Sambrook et al., 1989) of inserts subcloned in Bluescript SK+ and KS+. Sequences were assembled using Microgenie (Beckmann Instruments, Inc., Fullerton, CA) and Lasergene (DNASTAR Inc., Madison, WI) software. Searches of GenBank/ EMBL/DDBJ database were conducted using BLAST with the MRC human genome mapping project computer. The protein sequence was analyzed using DOMAIN, SAPS, and COILS programs.

Plasmid pASP36 was constructed by inserting the NcoI–BamHI fragment from p6a (see Fig. 2) into the expression vector pET23d (Invitrogen Corp., San Diego, CA). Conditions for culture and induction were as described by the supplier. The polypeptide expressed by this clone represents the NH₂-terminal portion of the Asp protein and does not include the putative actin binding domain. This polypeptide was insoluble and was purified by preparing inclusion bodies, separating the proteins by polyacrylamide gel electrophoresis, excising the appropriate band from the gel, and electroeluting the protein as described by Leppard et al. (1983). Polyclonal sera were prepared by injecting rabbits as described (Harlow and Lane, 1988).

Plasmid pASP11 was constructed by inserting the BamHI–EcoRI fragment from p6a (see Fig. 2) into the expression vector pQE31 (Qiagen, Inc., Santa Clara, CA). Conditions for culture, induction, and denaturing purification on Ni–agarose columns (Qiagen, Inc.) were as described by the supplier.

Embryos were fixed and stained as described by Gonzalez and Glover (1993). Propidium iodide was used to visualize DNA; the rat monoclonal antibody YL1/2 (Kilmartin et al., 1982) was used to stain microtubules, and the rabbit polyclonal antibody Rb3133 was used to stain the Asp protein. Staining was revealed using Texas red–conjugated goat anti–rat antibody and fluorescein-conjugated goat anti–rabbit antibody. All commercial antibodies were obtained from Jackson ImmunoResearch Labs, Inc. (West Grove, PA). Images were collected using an Optiphot microscope (Nikon Inc., Garden City, NY) with a confocal scanning head (model 600; BioRad, Richmond, CA) and merged using Adobe Photoshop V2.5.

Microtubules were purified from 0–3-h-old Drosophila embryos essentially as described by Goldstein et al. (1986). Embryos were collected, dechorionated in 50% bleach, and washed with tap water and then with lysis buffer. About 3 ml of embryos were homogenized in 2 vol of ice cold lysis buffer (0.1 M Pipes/NaOH, pH 6.6, 5 mM EGTA, 1 mM MgSO₄, 0.9 M glycerol, 1 mM DTT, 1 mM PMSF, 1 mg/ml aprotinin, 1 mg/ml leupeptin, and 1 mg/ml pepstatin) with a Dounce homogenizer. The microtubules were depolymerized by incubation on ice for 15 min, and the extract was then centrifuged at 16,000 g for 30 min at 4°C. The supernatant was recentrifuged at 135,000 g for 90 min at 4°C. Microtubules in this latter supernatant were polymerized by addition of GTP, to a final concentration of 1 mM, addition of taxol, to a final concentration of 20 mM, and incubation at room temperature for 30 min. 3-ml aliquots of extract was layered on top of 3-ml 15% sucrose cushions prepared in lysis buffer supplemented with 20 mM taxol and 1 mM GTP. After centrifuging at 54,000 g for 30 min at 20°C using a swing out rotor, the pellet was resuspended in lysis buffer containing taxol and GTP. To extract the MAPs, the concentration of NaCl was adjusted to 0.4 M, and the samples were incubated for 30 min at 37°C. Microtubules were pelleted by centrifugation at top speed on a microcentrifuge for 30 min; the supernatant is considered to be the MAP fraction.

Proteins were blotted to PVDF membranes (Amersham Life Sciences, Pittsburgh, PA) after SDS-PAGE. Membranes were blocked with TBST (50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20) containing 5% nonfat dry milk for 1 h, washed for 15 min in TBST plus 1% nonfat dry milk, and washed for 15 min in PEMF (0.1 mM Pipes/NaOH, pH 6.6, 1 mM EGTA, 1 mM MgSO₄, 25 mM NaF) buffer plus 1 mM GTP. The blots were then incubated overnight at room temperature with 5 μg/ml Drosophila tubulin (purified as described) in PEMF supplemented with 1 mM GTP and 20 mM taxol. The bound microtubules were detected by standard Western blot procedures using the monoclonal antibody Bx69 diluted five times.

Previous genetic mapping studies (Gonzalez et al., 1989) mapped asp to the cytological interval 96A21-96B10 between the distal breakpoint of In(3R)Ubx^7LLats^R and the breakpoint of T(Y;3)B197 (Fig. 1). At the onset of this work there were no obvious molecular entry points into this region, and so we chose the technique of chromosome microdissection and microcloning to achieve this end. Two chromosomal segments corresponding to the asp interval were microdissected and DNA fragments extracted for cloning in the bacteriophage λ insertion vector λ1149. The chromosomal origins of these clones were confirmed by in situ hybridization, before using them to screen phage λ and cosmid genomic libraries (see Materials and Methods). A resulting chromosome walk of 130 kb was carried out and clones correlated with the polytene chromosome map by in situ hybridization. The distal breakpoint of In(3R)Ubx^7LLats^R, which does not uncover asp and defines the proximal boundary of the region known to contain asp, was mapped to a 5.8-kb BamHI fragment at the extreme proximal end of the walk. The distal end of the chromosome walk maps to the interband between 96A21-25 and 96B1-10. Subsequently, we found that a cytologically invisible deficiency Df(3R)Hd^γ1 uncovers not only the nearby zygotic embryonic lethal tolloid (tld), as reported by Shimell et al. (1991), but also asp.

To locate the asp gene, restriction fragments from the chromosome walk were used to screen a Drosophila testis cDNA library (a gift of Hazelrigg, T.). Genomic DNA corresponding to partial cDNA clones obtained from this screen was used in a secondary screen of an embryonic cDNA library (Brown and Kafatos, 1988). This yielded 6.5-kb cDNA clones corresponding to a transcription unit lying immediately distal to tld (Shimell et al., 1991). The tld^68–62 mutation results from a P element insertion associated with a deletion of ∼4 kb that removes the entire tld transcription unit and ∼1 kb of the 5′ end of the gene encoding the 6.5kb transcript. Shimell et al. (1991) showed that tld can be rescued by two constructs, pMBO1366 and pMBO1367. Although pMBO1366 rescues the early embryonic zygotic lethality of tld^68–62, the rescued flies exhibit cuticular defects. As such defects are associated with asp, it seemed that tld^68–62 could be a tld asp double mutant. Consequently we tested whether these same transformants would rescue the recessive lethality of asp¹. pMBO1366 contains tld but only the 5′ end of the candidate asp gene, whereas pMBO1367 contains both transcription units (see Materials and Methods; Fig. 1). We found that the pMBO1367 rescues the lethality of asp¹ mutants while pMBO1366 does not. Taken together, these data confirm the identity of the 6.5-kb transcription unit as asp.

Furthermore, this gene has a developmental pattern of expression typical of many genes essential for the cell cycle whose products have to be maternally provided to the embryo. We used EcoRI–EcoRI or NcoI–EcoRI fragments from the 6.5-kb cDNA clone p6a as hybridization probes on developmental Northern blots. One transcript of ∼6.5 kb was readily observed (Fig. 2) in mRNA prepared from embryos and females (Fig. 2, lanes 1, 2, and 6) but was present at greatly reduced levels in larval and adult male mRNA (Fig. 2, lanes 3–5 and 7). When a SalI– NotI fragment from the plasmid pMBO1367, encompassing around 14 kb upstream of the NotI site on asp, was used as a probe, we observed the same 6.5-kb transcript plus one 3.8-kb transcript in embryonic mRNA corresponding to tld (Shimell et al., 1991) and a 5.2-kb transcript presumably corresponding to the recently described tolkin gene (Finelli et al., 1995; and data not shown).

The 6.5-kb asp cDNA encodes a predicted polypeptide of 1,863–amino acid residues that has no homologues in the GenBank/EMBL/DDBJ database (Fig. 3,A). The asp protein is predominantly hydrophilic and strikingly basic, having a calculated pI of 10.8. Its secondary structure is predicted to be mostly α-helical. Analysis of the protein using the COILS program shows that short stretches of amino acids near the COOH terminus have the potential to form a coiled coil. There is a small sequence lying between residues 848 and 870 that has significant similarity to the core actin binding domain of a number of actin binding proteins, such as α-actinin (Noegel et al., 1987; Blanchard et al., 1989), fimbrin, spectrin, dystrophin (de Arruda et al., 1990; Matsudaira, 1991), and the Dictyostelium discoideum ABP120 (Bresnick et al. 1990; Fig. 3,B). These proteins either bundle actin filaments together (for example α-actinin) or attach actin filaments to other cellular structures. A second sequence lying between residues 938 and 968 corresponds to the conserved calmodulin binding (IQ) motif (Cheney and Mooseker, 1992) present in neuromodulin, a neuron-specific membrane-associated protein (Chapman et al., 1991); neurogranin, a neuron specific protein kinase C substrate (Baudier et al., 1991); the igloo gene product, a calmodulin-binding protein from the Drosophila central nervous system (Neel and Young, 1994); and in the “neck” regions of most forms of conventional and nonconventional myosin (for review see Cheney and Mooseker, 1992; Fig. 3 C). In addition Asp shows six consensus sites for phosphorylation by p34^cdc2 and four consensus sites for phosphorylation by MAP kinase. Interestingly, these are all clustered in the NH₂-terminal third of the molecule.

To further characterize the asp gene product, we raised rabbit polyclonal antibodies to a truncated Asp protein expressed in E. coli. The NcoI–BamHI fragment of cDNA 6a (Fig. 4) was subcloned into the expression vector pET23d. The resulting construct, pASP36, expresses a polypeptide corresponding to amino acid residues 1–512, with a predicted molecular weight of 55.9 kD. This polypeptide does not contain the putative actin binding domain of the Asp protein.

The resulting serum Rb3133 recognizes one polypeptide of ∼220 kD in immunoblots of third instar larval brains (Fig. 5). To confirm that Rb3133 reacts specifically with the Asp protein, brains from larvae of the transgenic strain 1366 (see Materials and Methods) were used in Western blot analysis (Fig. 5, lane 2). Flies transformed with pMBO1366 are expected to synthesize two forms of the Asp protein: full length protein of 220 kD, derived from the endogenous copy of asp at 96A and a truncated form derived from the transgene. As expected, the serum detects the 220-kD wild-type asp protein plus a 124-kD polypeptide corresponding to the COOH-terminal truncated form of the asp protein encoded by plasmid pMBO1366. When extracts from brains of asp¹ homozygous larvae were analyzed (Fig. 5, lane 3), the 220-kD protein is no longer observed, but the serum labels a polypeptide of ∼130 kD. As the polyclonal antibodies were raised against the NH₂ terminus of the Asp protein, it seems that the asp¹ mutation results in a COOH terminus truncation or in an internal deletion of the Asp protein.

As mutations in asp affect the behavior of spindle microtubules, we sought to determine whether the Asp protein was itself microtubule associated. We purified microtubules from Drosophila embryos and took aliquots at each stage of the purification for electrophoresis and blotting onto PVDF membranes. These blots were incubated with the polyclonal serum Rb3133 and with the monoclonal antibody Bx69, which detect the Asp protein and β-tubulin, respectively (Fig. 6, A and B). Enrichment in tubulin is paralleled by an enrichment in the Asp protein. Moreover, the Asp protein seems to bind microtubules with a very high affinity as it is not released from the microtubule pellet by salt conditions that are known to dissociate most MAPs (Fig. 6, lanes 6). Asp protein is also not liberated after incubation with ATP (data not shown).

We further assessed the ability of Asp protein to bind microtubules in overlay assays in which bacterially expressed Asp fragments were separated by SDS-PAGE and blotted to a membrane that was subsequently incubated with microtubules. Bound microtubules were then detected by probing with the monoclonal anti–β-tubulin antibody Bx69. These assays were performed using either total E. coli proteins from cells expressing Asp11 and Asp36 (Fig. 7, A and B) or the corresponding purified fusion proteins (Fig. 7,C). Microtubules were observed to bind to both Asp36 and Asp11, but binding to Asp36 appears to be of a higher affinity (Fig. 7, B and C). Binding seems to be specific, as no detectable binding of microtubules to total E. coli proteins was observed in strains that had not been induced to express the fusion proteins (Fig. 7, A and B). To ensure that differences in binding did not simply reflect differences in the degree of induction of the two fusion proteins in the bacteria, we repeated the microtubule binding assay using increasing amounts of the two purified fusion proteins (Fig. 7 C). Differences in the binding affinity can be seen at all protein concentrations; Asp11 binding to microtubules was barely detectable when 1.6 μg of protein are used, whereas microtubule binding to Asp36 is readily detectable when 0.4 μg of protein were blotted to the membrane. This shows that the NH₂-terminal third of the Asp protein binds more avidly to microtubules than does the central part of the molecule.

Mutations in asp affect the morphology of both the mitotic spindle at several developmental stages and the meiotic spindle. At all stages mutant spindle microtubules may be described as having a long and wavy appearance, and it is not uncommon to see the loss of bipolarity in the form of hemispindle structures. To determine whether Asp protein is a constituent of the wild-type spindle we used the Rb3133 antibody to localize the Asp protein with respect to microtubules in mitosis in syncytial embryos (Fig. 8). During interphase, Asp protein appears to be distributed throughout the cytoplasm (Fig. 8,a), but as the syncytium enters mitosis and the bipolar spindle is formed, Asp is seen in association with the polar regions (Fig. 8,b). This polar association becomes tighter throughout metaphase and anaphase (Fig. 8, c and d), but at telophase, as the chromatin is decondensing and the spindle begins to disassemble, the Asp protein appears to move away from the region occupied by the centrosome onto the central region of the spindle microtubules.

Several pieces of evidence confirm the molecular identity of the asp gene. The full length gene rescues asp in germline transformants, whereas a truncated gene does not. Furthermore, antibodies raised against a segment of the gene expressed in E. coli recognize a truncated form of Asp protein in asp¹ homozygotes. These same antibodies have allowed us to examine the subcellular distribution of the Asp protein in cytological preparations by immunostaining and biochemically after subcellular fractionation. These experiments show the Asp protein to copurify with microtubules after Taxol-induced tubulin polymerization and that it remains associated with the microtubules in salt concentrations that remove many other MAPs. That the Asp protein is a MAP is confirmed by its association with the polar regions of spindles in mitosis.

Asp is the first metazoan MAP to be identified through a genetic approach, and it is gratifying that the naming of the gene by Ripoll et al. (1985) was predictive of the localization and function of the protein. The protein is highly basic, and in its central region lies a putative actin-binding motif. Basic regions within other MAPs play a significant part in conferring the ability to bind microtubules. The Asp protein is no exception to this rule in that at least one segment of this basic molecule, a polypeptide comprised of the 512 NH₂-terminal amino acids, can be shown to bind microtubules in vitro. The central region of the protein binds microtubules poorly. Unfortunately, we have been unable to test the microtubule binding ability of the COOH-terminal region, as this segment of the protein appears unstable in both bacterial and insect cell expression systems.

It is instructive to compare the domain organization and microtubule binding properties of Asp with other MAPs that have been characterized. These are principally of mammalian origin, and most have been purified from brain as proteins associated with the stable microtubules of the axons and dendrites of neurons. In some cases, these proteins may be expressed specifically in brain. However, in many cases their presence in other tissues is being confirmed. MAP2, for example, is predominantly found in brain but has recently been shown to be present in rat testes (Loveland et al., 1996). In general, MAPs bind to microtubules to increase their stability. Studies of MAP2C in vivo, for example, show that it bundles and stabilizes microtubules in cultured fibroblasts (Umeyama et al., 1993). Moreover, in vitro studies show that MAP2 can decrease the frequency of transitions between the growth and shortening of microtubules leading to net microtubule growth (Kowalski and Williams, 1993; Itoh and Hotani, 1994). Like several other MAPs, MAP2 has a short “promoter” domain, which binds to microtubules to promote assembly, and a “projection” domain thought to interact with other cytoskeletal elements. Fellous et al. (1994) have digested MAP2 with chymotrypsin to remove the projection domain and find that the residual promoter domain differs from the full length protein in that it binds microtubules to organize them into clusters of spirals. Moreover, assembly of microtubules in the presence of the promoter domain alone is more resistant to certain depolymerizing drugs than assembly in the presence of the full length protein. Thus the projection domain would appear to modify the interactions of MAP2 with microtubules in their assembly. The Asp protein may be seen as being structurally similar and functionally analogous, having an NH₂-terminal microtubule binding domain and central and COOH-terminal regions with the potential to interact with other cytoskeletal components.

It may be not only that microtubule dynamics are affected by the asp mutation but possible that the function of other MAPs, in particular motor molecules, could be impaired if the mutant protein can still bind microtubules. The binding of MAP2, for example, is known to inhibit the motility of kinesin and cytoplasmic dynein, but this is reversed by phosphorylation of the protein (Lopez and Sheetz, 1995). This specifically requires a cAMP-dependent kinase that copurifies with MAP2, and it is not affected by phosphorylation with p34^cdc2 kinase, mitogen activated protein kinase, or the NIMA mitotic kinase. In contrast, phosphorylation of MAP4 by p34^cdc2 kinase abolishes its microtubule stabilizing activity, interestingly without altering its ability to bind microtubules (Ookata et al., 1995). MAP4 is the major MAP of many mammalian cell lines and is present along the length of both interphase and mitotic microtubules (Bulinski and Borisy, 1980; DeBrabander et al., 1981). There is evidence that MAP4 mediates the association of cyclin B with microtubules in both the starfish oocyte and in cultured mammalian cells (Ookata et al., 1992, 1993, 1995). It may therefore provide a general mechanism for destabilizing interphase microtubules upon entry into mitosis. Other MAPs could also have such a role, and as with the kinesin-like proteins, MAPs could be functionally redundant. A 230-kD protein, XMAP 230, has been purified from Xenopus oocytes and found to be present in all dividing cells (Andersen et al., 1995). It is localized to microtubules in interphase but dissociates from them upon entry into mitosis, to reassociate later in the mitotic cycle. It is phosphorylated by mitotic extracts whereupon it has a reduced affinity for microtubules. Upon binding it increases microtubule growth rate, decreases the rate of shrinking, and suppresses microtubule loss by “catastrophe.” It thus stabilizes interphase microtubules and can locally modify microtubule behavior during mitosis. As the Asp protein has numerous sites for phosphorylation by p34^cdc2 and mitogen-activated protein kinase, it is not inconceivable that cyclical phosphorylation could affect its properties throughout the mitotic cycle.

The presence of wild-type Asp protein in the polar regions of the spindle points to this being the site of its action and suggests that it may play a crucial role in regulating the dramatic changes of microtubule dynamics upon the entry into mitosis. Its presence in the astral microtubules may be important to mediate interactions between microtubules and the actin cytoskeleton. This function is suggested by the actin-binding motif in the central domain of the protein. This motif is homologous to ones seen in a number of actin bundling or gelation factors. These include α-actinin (Noegel et al., 1987; Blanchard et al., 1989), a filamentous actin cross-linking protein found in stress fibers and adhesion plaques in nonmuscle cells and in Z-discs in muscle cells; it has a single actin binding motif and can cross link actin filaments into a homodimer. Single actin binding motifs are also found on dystrophin and spectrin (de Arruda et al., 1990; Matsudaira, 1991). Fimbrin, an actin-associated protein found in microvilli and filopodia, on the other hand, has a duplicated actin binding motif that enables it to bundle actin filaments as a monomer (de Arruda et al., 1990). It has homology to plastins, cytoplasmic actin binding proteins, and the protein encoded by the yeast gene ABP7 or Sac6p (Adams et al., 1995). The association of microtubule and microfilament networks is nowhere more obvious than in the vicinity of centrosomally nucleated asters. In the syncytial blastoderm Drosophila embryo, for example, the actin caps that form above interphase nuclei are positioned by the centrosome and its associated microtubules. Moreover, the spatial coordination of mitosis and cytokinesis requires the contractile actin ring that will dictate the cleavage furrow to be correctly positioned with respect to the spindle. Classical experiments with echinoderm eggs showed that moving the spindle by micromanipulation resulted in the corresponding repositioning of the cleavage furrow. However, the experimental manipulation of such embryos to produce two mitotic spindles within a single cell led to the generation of a third cleavage furrow between the asters of the juxtaposed spindles (for review see Rappaport, 1986a,b). The nature of the molecular “signal” that positions the actin ring with respect to the asters is unknown, but it is possible that a molecule such as the Asp protein could fulfill this role.

The presence of a potential calmodulin binding site in the central region of the molecule implies that the Asp protein may be responsive to fluxes in calcium ions known to occur in the polar regions of the spindle. Preliminary experiments show that the motif present in Asp enables it to bind to calmodulin in a Ca²⁺-dependent manner (data not shown). Calmodulin is a member of the EF hand superfamily of proteins. There is strong evidence for a direct role for two members of this family in spindle pole body (SPB) function in budding yeast. Calmodulin itself has been shown by both genetic and molecular studies (Geiser et al., 1993; Stirling et al., 1994, 1996; Sundberg et al., 1996) to be essential for spindle function as a result of its interaction with the COOH terminus of Spc110p, an essential protein with a large coiled-coil domain that provides a spacer between the central and inner plaques of the SPB (Kilmartin, 1993). The second EF hand protein localized to the SPB is the product of the CDC31 gene, which associates with the KAR1 gene product to form a complex (Spang et al., 1993, 1995; Biggens and Rose, 1994; Vallen et al., 1994). Roles for these proteins in spindle function are likely to be conserved. Calmodulin is the major intracellular Ca²⁺ receptor and is involved in many cellular processes; it is known to be localized to the microtubules of metazoan spindles (e.g., Stemple et al., 1988), and it appears to be important in regulating the function of its dependent kinase and phosphatase in the metaphase– anaphase transition and in spindle behavior (Morin et al., 1994; Yoshida et al., 1994). CDC31 is the yeast homologue of centrin, a protein first identified in the Chlamydomonas basal body and subsequently found to be a universal component of centrosomes in animal cells (for review see Salisbury, 1995). There are therefore a large number of potential ways in which Asp protein function might be mediated through an association with one of these EF hand proteins.

It is evident that many questions are thrown open by our finding that the Asp protein is a MAP containing potential actin and calmodulin binding motifs. However, our work has now generated the molecular tools with which we can begin to address some of the above issues.

We would like to thank Fiona Cullen for sequencing; Mar Carmena and Daryl Henderson for their comments on the manuscript; Alvaro Tavares for his suggestions and help; and Dr. M. O'Connor for supplying lines transformed with pMBO1366 and pMBO1367.

This work was supported by research grants from Science and Engineering Research Council, Medical Research Council, Cancer Research Campaign, and the Human Capital and Mobility Programme of the European Union. Thomas Howard is grateful for a studentship from BBSRC and Maria do Carmo Avides for an European Molecular Biology Laboratory Long-Term Fellowship.