The protein kinase Aurora-A is required for centrosome maturation, spindle assembly, and asymmetric protein localization during mitosis. Borealis (Bora, so named for aurora borealis to indicate its similarity with aurora-A) is a conserved protein that is required for the activation of Aurora-A at the onset of mitosis. In the Drosophila peripheral nervous system, bora mutants show defects during asymmetric cell division identical to those observed in aurora-A. Furthermore, overexpression of bora can rescue defects caused by mutations in aurora-A. Bora is conserved in vertebrates, and both Drosophila and human Bora can bind to Aurora-A and activate the kinase in vitro. In interphase cells, Bora is a nuclear protein, but upon entry into mitosis, Bora is excluded from the nucleus and translocates into the cytoplasm in a Cdc2-dependent manner. A model is presented here in which activation of Cdc2 initiates the release of Bora into the cytoplasm where it can bind and activate Aurora-A (Hutterer, 2006).

Cell division involves the coordinated execution of several distinct steps. First, chromosomes condense and the nuclear envelope breaks down. Then, the mitotic spindle forms, sister chromatids separate, and chromosomes segregate into the two daughter cells. Finally, mitosis finishes with cytokinesis, the actual division of the cell into two separate daughter cells. Mitosis involves the sequential activation of several protein kinases that are required for all or a subset of these mitotic events: while Cdc2 is a master regulator of mitosis and is required for the initiation of mitosis, kinases of the Aurora and Polo families are responsible for distinct subsets of mitotic events. How these kinases are activated and how they regulate individual mitotic events is not very well understood (Hutterer, 2006).

Aurora kinases were originally identified in Drosophila, but homologs were later found in all eukaryotic organisms. While yeast contains only a single Aurora kinase called Ipl1p, at least two families with distinct functions and subcellular localizations can be distinguished in multicellular organisms: Aurora-A is concentrated on the spindle and on centrosomes and is required for centrosome maturation and spindle assembly, while Aurora-B is localized on chromosomes and on the central spindle and is involved in chromosome condensation, kinetochore-microtubule attachment and cytokinesis. Aurora-B is part of a multimeric complex containing the so-called chromosome passenger proteins INCENP, survivin, and borealin. The individual members of that complex are codependent for their subcellular localization, and their role is to direct Aurora-B to its correct localization within the cell. Consistent with the conserved function and localization of Aurora-B, all members of the complex are conserved in evolution (in C. elegans, they are called ICP-1, BIR-1, and CSC-1. Binding partners have also been identified for Aurora-A, but in this case, their evolutionary conservation is less clear (Kufer, 2003). TPX2 is a microtubule binding protein required for spindle assembly. It can bind Aurora-A and activate the kinase via an N-terminal domain. Upon TPX2 RNAi, Aurora-A fails to localize to the spindle whereas its centrosome localization is unaffected (Kufer, 2002). Since the interaction of TPX2 with Aurora-A is stimulated by the small GTPase Ran, a model was proposed in which activated Ran is generated by condensed chromatin and locally activates Aurora-A, thereby stabilizing microtubules. Although a putative C. elegans TPX2 homolog was identified, the homology does not extend over the whole protein and no homologs are present in other invertebrates, including Drosophila. Another Aurora-A binding partner is the LIM domain protein Ajuba. Like TPX2, Ajuba can activate Aurora-A, but again, no homologs have been identified in invertebrates (Hutterer, 2006 and references therein).

Besides its role in centrosome maturation and spindle assembly, Aurora-A has a special function during asymmetric cell division (Berdnik, 2002). To divide asymmetrically, some cells are capable of segregating cell fate determinants into one of their two daughter cells (Betschinger, 2004). Asymmetric cell divisions are particularly well understood in Drosophila external sensory (ES) organs where they contribute to the formation of four different cell types from a single sensory organ precursor (SOP) cell. The SOP cell divides into a pIIa and a pIIb cell. Later, pIIa gives rise to the two outer cells, while pIIb generates the two inner cells of the organ. During each division, the cell fate determinant Numb localizes asymmetrically and segregates into one of the two daughter cells where it regulates cell fate by repressing Notch signaling. In numb mutants, Notch is not repressed and abnormal ES organs with too many outer and no inner cells are formed. A similar phenotype is observed in aurora-A mutants. In these mutants, Numb does not localize asymmetrically and is not segregated into one of the two daughter cells (Berdnik, 2002). Since asymmetric Numb localization requires actin, but not microtubules, this phenotype is not an indirect consequence of the centrosome maturation and spindle assembly defects that are also observed in aurora-A. Thus, besides its role in regulating microtubules, Aurora-A also regulates actin-dependent mitotic processes (Hutterer, 2006 and references therein).

Despite its functional conservation, a conserved pathway for the activation of Aurora-A is not known. This study describes the identification of Bora, an interaction partner of Aurora-A that is conserved from C. elegans to humans. Bora was identified due to its phenotypic similarity to aurora-A and it has been shown that bora overexpression can partially rescue aurora-A mutants. Bora binds to Aurora-A and can activate the kinase in vitro. Bora is a nuclear protein that translocates into the cytoplasm upon activation of Cdc2, suggesting that its subcellular localization might contribute to the regulation of Aurora-A. These results describe a regulator of Aurora-A that is conserved from Drosophila to humans and suggest a potential mechanism for the sequential activation of Cdc2 and Aurora-A (Hutterer, 2006).

Thus the conserved Aurora-A binding partner Bora is essential for Aurora-A to perform centrosome maturation, spindle assembly, and asymmetric protein localization during mitosis in Drosophila. Bora can activate Aurora-A in vitro. Bora is a nuclear protein that is excluded from the nucleus during prophase in a Cdc2-dependent manner. Nuclear retention of Bora might help to keep Aurora-A inactive during interphase. When Cdc2 becomes activated, Bora is released into the cytoplasm where it can bind and activate Aurora-A. This hypothesis could provide a molecular explanation for previous results which have demonstrated that Cdc2 is crucial for the activation of Aurora-A (Marumoto, 2003; Maton, 2003). Since Bora is a substrate for Cdc2 in vitro and—at least in vertebrates—a fraction of Cdc2 has been reported to be nuclear, it is conceivable that direct phosphorylation of Bora might facilitate its exclusion from the nucleus. However, nuclear release of Bora is not the only mechanism by which its activation of Aurora-A is regulated since the bora mutant phenotype can also be rescued by Bora fused to a myristylation signal, which keeps the protein in the cytoplasm, or fused to a nuclear localization signal, which retains the protein in the nucleus until nuclear envelope breakdown (Hutterer, 2006).

Although in Drosophila, Bora so far is the only known activator of Aurora-A, several in vitro activators of Aurora-A have been identified in other organisms. In vertebrates, TPX2 prevents PP1-dependent dephosphorylation and thereby locks the kinase in its active conformation (Bayliss, 2003). The activation of Aurora-A by Cdc2 is PP1 independent (Maton, 2005), and, therefore, TPX2 is unlikely to participate in this particular event. Furthermore, TPX2 is only required for a subset of Aurora-A-dependent processes: TPX2 inactivation by RNAi causes spindle defects and loss of Aurora-A from the mitotic spindle, but centrosome maturation is normal, and the centrosome pool of the kinase is unaffected. TPX2/Aurora-A binding is stimulated by the small GTPase Ran, which in turn is activated by RCC1, an exchange factor that is located on condensed chromatin and is involved in microtubule nucleation and spindle formation. Thus, unlike Bora, TPX2 seems to be specifically responsible for the spindle assembly function of Aurora-A. So far, no TPX2 homolog has been identified in Drosophila. Whether this is due to a low level of sequence similarity that escapes standard homology searches or whether it reflects a fundamental difference in Aurora-A function between organisms is currently unclear (Hutterer, 2006).

One protein that might be generally required for Aurora-A activation is Ajuba (Hirota, 2003). Upon Ajuba RNAi, Aurora-A fails to be activated. In HeLa cells, this leads to a cell-cycle block in G2 and prevents entry into mitosis. However, since ajuba null mutant mice are completely viable (Pratt, 2005) and keratinocytes from these mice have no cell-cycle block, the significance of these RNAi experiments is unclear. Furthermore, no Ajuba homologs are found in C. elegans or Drosophila, suggesting that a functional connection between Ajuba and Bora is unlikely (Hutterer, 2006).

More recently, two other activation pathways for Aurora-A have been described. The focal adhesion protein HEF1 binds to Aurora-A and is required and sufficient for Aurora-A activation (Pugacheva, 2005). The protein kinase PAK relocalizes to centrosomes during mitosis where it is activated and in turn phosphorylates and activates Aurora-A (Zhao, 2005). Since PAK is a part of focal adhesion complexes, both pathways might be part of a mechanism establishing crosstalk between cell adhesion and the mitotic apparatus (Cotteret, 2005). However, PAK inhibition only delays centrosome maturation, suggesting that this pathway is not a crucial regulator of the G2/M functions of Aurora-A. In Drosophila, both PAK and HEF1 are conserved, but the PAK mutant phenotype does not suggest any requirement of the kinase for mitosis. Taken together, these observations suggest that Bora does not participate in any of the known pathways but is more globally involved in the activation of Aurora-A (Hutterer, 2006).

Like Aurora-A, Bora is required for actin-dependent asymmetric protein localization during mitosis (Berdnik, 2002). It is thought that the polarized localization of the kinase aPKC leads to asymmetric phosphorylation of the cytoskeletal protein Lgl. Since phosphorylation inactivates Lgl and Lgl is essential for establishing a cortical binding site for cell fate determinants, those determinants accumulate exclusively on the side of the cortex that is free of aPKC. Aurora-A could act at several points in this pathway: either the cortical binding site could already be polarized in interphase and activation of Aurora-A could establish its affinity for cell-fate determinants, or alternatively, Aurora-A could regulate the activity of aPKC. In this case, aPKC would be asymmetric but inactive in interphase and its activation in prophase would initiate asymmetric localization of cell-fate determinants. At the moment, it is not possible to distinguish between these possibilities, but identification of the Aurora-A substrates relevant for asymmetric protein localization should clarify its mode of action. In any case, the observation that Cdc2 is essential for asymmetric determinant localization as well (Tio, 2001) is consistent with a model where Cdc2 is required for the Bora-dependent activation of Aurora-A (Hutterer, 2006).

Aurora-A has been implicated in carcinogenesis. It is overexpressed in a number of cancers and its overexpression results in polyploidy or cells containing multiple centrosomes. Aurora-A has therefore been used as a drug target for cancer therapy, and the identification of Bora offers an alternative route for the discovery of Aurora-A selective inhibitors. The human Bora homolog (annotated as FLJ22624) is located on chromosome 13 in a region that contains a breast cancer susceptibility gene and has been implicated in a variety of malignant tumors. Future studies will reveal whether it is involved in carcinogenesis as well (Hutterer, 2006).

GENE STRUCTURE

cDNA clone length - 2004 bp

Bases in 5' UTR - 93

Exons - 2

Bases in 3' UTR - 281

PROTEIN STRUCTURE

Amino Acids - 539

Structural Domains

Bora has no obvious protein domains of known function or structure. Blast searches reveal homologs in other insect species and a bioinformatics analysis identifies sequence homologs in all vertebrate species, including humans. Conservation of Bora is highest in an N-terminal domain extending approximately from aa 65 to aa 247 of the Drosophila protein, while the rest of the protein is less conserved. Mouse Bora has been annotated as BAE24669, and human Bora is annotated as LOC79866 and located at 13q22.1. Bora is also conserved in C. elegans, where it is encoded by gene F57C2.6, but no homologs were detected in unicellular organisms (Hutterer, 2006).

EVOLUTIONARY HOMOLOGS

To determine whether the requirement for activation of Aurora-A by Bora is conserved between flies and vertebrates, whether loss of human Bora leads to mitotic defects was tested. The gene was silenced in mammalian U2OS cells by siRNA, and a significant reduction of HsBora mRNA was detected 48 hr after siRNA transfection. In contrast to cells treated with a control siRNA, cells treated with siRNAs against Bora frequently displayed multipolar spindles in mitosis, a phenotype that is also observed upon TPX2 RNAi and after injection of antibodies blocking Aurora-A function. Taken together, these experiments suggest that Bora is a key activator of Aurora-A that is functionally conserved between Drosophila and vertebrates (Hutterer, 2006).

date revised: 22 February 2007

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.