BIOLOGICAL OVERVIEW

Aurora-A kinase is found at centrosomes and on microtubules of the mitotic spindle and is essential for setting up a functional mitotic spindle. Drosophila Aurora-A, the founding member of this kinase family, is also required for centrosome separation, and its complete absence leads to the formation of monopolar spindles with a characteristic circular chromosome arrangement (Glover, 1995). Disruption of the function of the A-type Aurora kinase of Drosophila by mutation or RNAi leads to a reduction in the length of astral microtubules in syncytial embryos, larval neuroblasts, and cultured S2 cells. In neuroblasts, it can also lead to loss of an organized centrosome and its associated aster from one of the spindle poles, whereas the centrosome at the other pole has multiple centrioles. When centrosomes are present at the poles of aurA mutants or aurA RNAi spindles, they retain many antigens but are missing the Drosophila counterpart of mammalian transforming acidic coiled coil (TACC) proteins, D-TACC (Transforming acidic coiled-coil protein). A subpopulation of the total Aurora A is present in a complex with D-TACC, which is a substrate for the kinase. It is proposed that one of the functions of Aurora A kinase is to direct centrosomal organization such that D-TACC complexed to the MSPS/XMAP215 microtubule-associated protein (mini spindles) may be recruited, and thus modulate the behavior of astral microtubules (Giet, 2002).

In addition to the Aurora-A function in centrosome maturation, Aurora A also functions in asymmetric protein localization during mitosis (see Effects of Mutation). Using photobleaching of a GFP-Aurora fusion protein, it has been show that two rapidly exchanging pools of Aurora-A are present, one in the cytoplasm and a second at the centrosome. These pools might carry out the two functions. Activation of the Aurora-A kinase at the onset of mitosis is required for the actin-dependent asymmetric localization of Numb. Aurora-A is, as described in this section, involved in centrosome maturation and spindle assembly, indicating that Aurora A regulates both actin- and microtubule-dependent processes in mitotic cells (Berdnik, 2002).

Correct regulation of the organization and dynamics of microtubules is an essential aspect of entry into M-phase. Microtubules are nucleated by the gamma-tubulin ring complex, the amount of which increases markedly at the centrosome upon entry into mitosis. Microtubule nucleation at the centrosome requires the cooperation of other microtubule-associated proteins (MAPs), notably the Abnormal spindle (Asp) in Drosophila. MAPs also play a central role in regulating microtubule dynamics. For example, Xenopus XMAP215 promotes the elongation rate of microtubules at their plus ends (less so at the minus ends), and appears to counteract the catastrophe-promoting activity (the transition from polymerization to a depolymerization) promoted by XKCM1. The counterpart of XMAP215 in Drosophila is encoded by the gene minispindles, mutations that appear to destabilize spindle microtubules which become small and associated with single chromosomes. The MSPS protein forms a complex with the Drosophila counterpart of mammalian transforming acidic coiled coil (TACC) proteins, the centrosomally associated protein D-TACC. Injections of antibodies against D-TACC or mutations in the d-tacc gene result in centrosomal microtubules that are abnormally short, as well as in the accumulation of mitotic defects. The D-TACC protein is found at the spindle poles and its recruitment of MSPS protein has been postulated to stabilize centrosomal microtubules. In the acentriolar spindles of female meiosis, both the motor protein Ncd and D-TACC are required for the proper localization of MSPS (Giet, 2002 and references therein).

The Aurora-related enzymes constitute a major family of mitotic kinases (Giet, 1999a). Most is known about the B-type subfamily, which first localizes to condensing chromosomes and centromeres, and subsequently to the central spindle and the midbody in anaphase and telophase, respectively. The Aurora B protein kinase is the functional subunit of a complex containing INCENP, and the BIR-1/survivin protein is required for its localization; it is required in chromosome segregation and it phosphorylates histone H3, which correlates with recruitment of the condensin complex. In contrast, less is known of the exact function of the Aurora A–type kinases, although ectopic expression of the human enzyme leads to aneuploidy, centrosome amplification, and transformation (Bischoff, 1998; Zhou, 1998). Mutations in the aurora A gene of Drosophila melanogaster lead to formation of spindles with abnormally organized poles, including characteristic monopolar structures (Glover, 1995). Bipolar spindles having abnormally organized centrosomes, and microtubules have been observed after double-stranded RNA mediated interference directed against the air1 gene of Caenorhabditis elegans (Schumacher, 1998). A catalytically inactive or truncated version of the pEg2 Aurora A-like kinase promotes the collapse of spindles assembled in Xenopus egg extracts (Roghi, 1998). This would be consistent with the known ability of Aurora A kinase to phosphorylate the kinesin-like protein XlEg5 that is also required for spindle assembly and stability (Giet, 1999b). To gain a better understanding of the role of Aurora A kinase, the requirement for the enzyme in organizing spindle poles in different mutant alleles was examined. In mutant cells or after RNA interference to eliminate the enzyme, the spindle poles have abnormal organization and abnormally short arrays of astral microtubules. The latter defect correlates with the loss of D-TACC from centrosomes. D-TACC interacts with a subpopulation of Aurora A in vivo, and D-TACC is a substrate of the kinase. A model in which one of the centrosomal functions of Aurora A kinase is to control microtubule dynamics at the spindle poles by regulating the recruitment of D-TACC and its associated MAP, the minispindles/XMAP215 protein (Giet, 2002).

A role for the Drosophila Aurora A in regulating centrosome behavior has suggested undertaking an analysis of the mutant phenotype, a characteristic of which is the generation of circular arrays of mitotic chromosomes around a single body of apparently duplicated centrosomes (Glover, 1995). It is possible that such a phenotype could arise through two different routes. One possibility, that Aurora A function is required to maintain spindle pole separation, receives support from the observation that bipolar mitotic spindles formed in frog egg extracts collapse following the addition of dominant negative mutant forms of Eg2, the Xenopus A-type Aurora kinase (Giet, 2000). However, the present observations of multiple centrioles at the spindle poles in aurora A mutant cells suggests that at least in these cells there has been a failure of centriole segregation at the onset of mitosis. An allelic series of mutations offers the possibility of studying multiple functions of a protein and its role at different developmental stages where the cell cycle may be under differing modes of regulation. This study concentrated on two mutants, one of which displays a mitotic phenotype in syncytial embryos that undertake rapidly alternating S and M phases, and the other in larval neuroblasts, cells that undergo conventional cell cycles with active checkpoints. In both situations focus was placed on events at the spindle poles that provide a common aspect of mutant phenotype in these differing cell cycles. In the first case, the centrosomes and astral microtubules were examined in syncytial embryos derived from mothers homozygous for aurA²⁸⁷, a weak hypomorphic aurA mutant that produces poorly functional protein that allows repeated, but increasingly abnormal mitoses in the syncytium. In the second case, the spindle poles were examined in aurA^e209, a strongly hypomorphic mutant that shows an equally elevated mitotic index, whether the mutation is homozygous or hemizygous. In each mutant a diminution of the length and number of astral microtubules was observed even though the ultimate consequences for mitotic progression differ markedly in these two circumstances; the embryonic mitotic cycles can continue whereas the larval cycles are blocked at metaphase. The origins of the monopolar spindles seen in larval neuroblasts are still unclear. The increase in proportion of such structures seen in the presence of two copies rather than a single copy of the aurA^e209 allele (in both cases in the absence of wild-type protein) points toward a neomorphic function for the mutant protein. Consistent with this is the finding that no monopolar spindles were seen following air-1 RNAi in C. elegans (Schumacher, 1998). The interphase-like arrays of microtubules seen in the air-1-depleted embryos resemble those seen in the cytoplasm of aurA²⁸⁷-derived embryos. However, depletion of Aurora A by RNAi does not block mitotic progression in the embryonic cell line S2, but nevertheless it does lead to defects at the spindle poles similar to those in the aurA mutants (Giet, 2002).

Examination of the ultrastructure of the spindle poles in aurA^e209 mutant larval neuroblasts has revealed that when centrosomes are present they are comprised of multiple centrioles surrounded by electron-dense material having a distribution similar to centrosomal antigens revealed by light microscopy. These abnormal centrosomes contain core centrosomal antigens such as centrosomin, proteins required for the nucleation of microtubules such as gamma-tubulin and Asp and which can be removed by salt washes from the core centrosome, and proteins such as CP-190 whose association with the centrosome may be mediated through an interaction with gamma-tubulin. However, a characteristic of the aurA^e209 mutant centrosomes, also seen in the aurA²⁸⁷-derived embryos and after aurA RNAi in S2 cells, is the reduced amounts of the D-TACC protein and the MSPS proteins, known to be required to maintain the length and/or number of astral microtubules (Cullen, 1999; Gergely, 2000a; Lee, 2001). Thus, the failure of D-TACC-MSPS complex to be recruited to the centrosome when Aurora A kinase function is compromised provides an explanation for the correlation of the apparent diminution of length and/or number of astral microtubules in such circumstances (Giet, 2002).

D-TACC is not only associated with centrosomes, but also with spindle microtubules. This latter association, which appears not to require Aurora A kinase, has been shown to depend on binding to the MSPS MAP, a member of the ch-TOG/XMAP 215 family of proteins (Cullen, 2001; Lee, 2001). Aurora A kinase can bind to the D-TACC protein in Drosophila, as can the orthologs of these proteins in human cells. However, in extracts of Drosophila embryos, only a minor proportion of each protein appears to be associated in the same complex, probably reflecting the fact that both Aurora A and D-TACC proteins are supplied as an abundant maternal dowry that is only used for mitosis as development proceeds. Aurora A is capable of phosphorylating D-TACC, that this phosphorylation is required to recruit D-TACC to the centrosome early in mitosis. However, it is equally possible that Aurora A phosphorylates other centrosomal proteins in such a way as to facilitate the recruitment of D-TACC. The association of D-TACC with the centrosome could occur either independently or when it is already complexed with MSPS. In either case, the docking of the D-TACC-MSPS complex to the centrosome could then allow the complex access to microtubules nucleated by the gamma-tubulin ring complex in concert with the Asp protein. Association of MSPS may then promote the growth of the microtubules at both minus and plus ends as is normally seen in mitotic asters. This is suggested by the known properties of the Xenopus counterpart of MSPS, XMAP215, which promotes microtubule elongation rates strongly at the plus end but also at the minus end. Thus, the complex may also be carried on the extending microtubule giving it the appearance of accumulating near the plus ends as well as on the centrosome. Such localization of D-TACC has been observed near to the putative plus ends of microtubules on the spindles of S2 cells. It is also possible that the D-TACC-MSPS complex acts to prevent ejection of microtubules from the centrosome (Giet, 2002).

Although both hypomorphic and null alleles of d-tacc lead to female sterility, the gene is not essential for the larval division cycles. Thus, homozygote mutants can transit the earliest stages of development using wild-type protein from their heterozygous mothers and develop to adulthood. One possible explanation is that a need for D-TACC to target MSPS to centrosomes and thereby provide an efficient means of stabilizing astral microtubules is of particular importance in the rapid division cycles of the syncytial embryo. The longer cell cycle of larval cells coupled with their strong metaphase checkpoint could permit time to correctly assemble spindle poles in the absence of D-TACC protein. However, strong hypomorphic aurora A mutants do arrest at metaphase, pointing toward additional functions of the Aurora A enzyme beyond D-TACC recruitment, possibly in aspects of the metaphase-anaphase transition itself (Giet, 2002).

Ultrastructural studies have shown that bipolar spindles in aurA^e209 neuroblasts are missing centrioles from one of their poles and have multiple centrioles and pericentriolar material at the other pole. These observations correlate well with observations made by immunostaining with the light microscope. Such studies reveal many mitotic figures in which multiple centrosomal antigens were missing from one of the poles and multiple bodies containing centrosomal antigens were present at the other. The ability to make a stabilized and focused spindle pole in the absence of centrosomes is well known. Such focused spindle poles have been shown to form in the absence of centrosomes both in Xenopus extracts and in Drosophila through the concerted action of microtubule motors and MAPS to organize and stabilize focused microtubule minus ends. Such acentriolar spindle poles are also seen in female meiosis in Drosophila where the minus end directed motor Ncd is essential to organize the poles. Cullen (2001) has proposed that complexes of D-TACC and MSPS at the acentriolar poles of the spindles of female meiosis could stabilize the bipolar structure, thus accounting for its loss of bipolarity in MSPS or d-tacc mutants. Such a function is unlikely to be essential to maintain the mitotic spindle in larval neuroblasts, which can adopt a stable bipolar structure in the absence of Aurora A function and hence D-TACC accumulation, and indeed in the absence of centrosomes (Giet, 2002).

Could the failure to recruit the D-TACC protein to the centrosome also explain the accumulation of replicated centrioles at the spindle poles and their failure to segregate? The dispersed distribution of centrosomal antigens at the spindle pole is explained at the ultrastructural level by the finding of multiple centrioles surrounded by electron-dense pericentriolar material. It is possible that the presence of D-TACC is required to maintain aspects of the structural integrity of the centrosome, since the molecule is endowed with a coiled coil region. It has also been demonstrated to form polymers that could be of structural importance (Gergely, 2000a). Overexpression of the TACC domain of D-TACC alone results in the formation of TACC aggregates that bind MSPS and nucleate asters of microtubules (Lee, 2001). Overexpression of the human counterpart of D-TACC, HsTACC3, in mammalian cells also leads to the formation of aggregates to bind ch-TOG and appears to increase the numbers of microtubules (Gergely, 2000a). This apparent tendency of the TACC proteins to form large polymers could perhaps account for the cytoplasmic accumulation of D-TACC aggregates in aurA mutant neuroblasts. These show some tendency to cluster over the spindle microtubules. However, their failure to affect spindle structure might reflect the lower levels of MSPS that are found in this tissue compared with the syncytial embryo. It may be particularly important during mitosis that this tendency of D-TACC to aggregate is controlled. Aurora A kinase could fulfil such a role by permitting recruitment of D-TACC once mitosis is underway and when the centriole pairs have separated before prophase. Regulation of the behavior of astral microtubules may be important at the time that centrosomes are migrating around the nuclear envelope before the nuclear lamins depolymerize. The length, density, and dynamics of these microtubules may be essential for the migration of the centrosome not only around the nuclear envelope, but in other developmental processes. Indeed, one developmental failure seen in d-tacc mutant embryos is the failure of centrosomes to migrate to the cortex of the syncytial embryo in cycles 9 and 10 (Gergely, 2000b), a process known to be microtubule dependent. Failure of centrosomes to migrate to opposite sides of the nucleus could explain the origins of monoastral biploar spindles. It is proposed that accumulation of microtubule nucleating centers at one pole could in its extreme disrupt the inherent tendency to form a bipolar structure leading to formation of monopolar spindles. The failure of duplicated centrioles to segregate to both ends of the mitotic spindle as it forms raises the possibility that D-TACC/MSPS recruitment may also be required to stably associate the replicated centriole pair (Giet, 2002).

D-TACC is only the second substrate of the A-type Aurora kinases to be identified, the first being the Eg5 kinesin-like protein (Giet, 1999a). Undoubtedly there exist many others, and indeed it is possible that D-TACC is only one of several centrosomal substrates of the Aurora A kinase that may play a role in facilitating the equitable segregation of centrioles to the spindle poles. Indeed as d-tacc does not appear to be essential for viability it may be functionally redundant and so the larval lethality shown by strongly hypomorphic alleles of aurA may reflect a role for the kinase in modifying other mitotic targets. Identifying other substrates of the Aurora A kinases and evaluating their roles in mitotic progression remains a future challenge (Giet, 2002).

GENE STRUCTURE

Female sterile mutations of aurora are allelic to mutations in the lethal complementation group ck10. This complementation group lies in a cytogenetic interval, 87A7-A9, that contains eight transcription units. A 250 bp region upstream of both aur and a divergent transcription unit corresponds to the site of a specific chromatin structure (scs') previously proposed to be a barrier to insulate enhancers of the major hsp70 gene at 87A7 (Glover, 1995).

cDNA clone length - 1597

Bases in 5' UTR - 136

Exons - 3

Bases in 3' UTR - 372

PROTEIN STRUCTURE

Amino Acids - 421

Structural Domains

Aurora is a protein kinase and contains a tyrosine kinase catalytic domain and a serine/threonine protein kinase family active site (Glover, 1995).

date revised: 20 May 2002

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

Please e-mail comments/corrections to brodyt@codon.nih.gov