During asymmetric cell division, the mitotic spindle must be properly oriented to ensure the asymmetric segregation of cell fate determinants into only one of the two daughter cells. In Drosophila neuroblasts, spindle orientation requires heterotrimeric G proteins and the Gα binding partner Pins, but how the Pins-Gαi complex interacts with the mitotic spindle is unclear. This study shows that Pins binds directly to the microtubule binding protein Mushroom body defect (Mud), the Drosophila homolog of Nuclear Mitotic Apparatus (NuMA) protein. Like NuMA, Mud can bind to microtubules and enhance microtubule polymerization. mud mutants form functional spindles and the neuroblasts are correctly polarized. Consistent with this, Brat and Numb form crescents in mud mutant neuroblasts, but the spindle is not aligned with them. Mitotic spindles in neuroblasts fail to align with the polarity axis. Therefore, the spindle orientation defect is a direct consequence of Mud loss of function. mud mutation can lead to symmetric segregation of the cell fate determinants Brat and Prospero, resulting in the misspecification of daughter cell fates and tumor-like overproliferation in the Drosophila nervous system. The data suggest a model in which asymmetrically localized Pins-Gαi complexes regulate spindle orientation by directly binding to Mud (Bowman, 2006; Izumi, 2006; Siller, 2006).

Asymmetric cell division makes an important contribution to cell fate diversity in multicellular organisms. In asymmetric division, cells create an internal polarity axis and localize cell fate determinants to one pole. Alignment of the mitotic spindle along the axis of polarity causes the determinants to segregate into one of the two daughter cells, making each daughter cell different from its sibling. How the mitotic spindle coordinates with the polarity axis is unclear (Bowman, 2006).

Drosophila neuroblasts are a well-studied model system for asymmetric cell division. Neuroblasts undergo repeated rounds of asymmetric division, generating a larger apical cell that retains neuroblast characteristics and a smaller, basal ganglion mother cell (GMC) that divides only once more to generate two neurons. During each division, the cell fate determinants Numb, Prospero, and Brat segregate into the GMC with the help of the adaptor proteins Miranda and Pon (Partner of Numb). Asymmetric segregation of all of these proteins requires the PDZ domain proteins Bazooka (Baz, the fly homolog of Par-3) and Par-6. Together with the kinase aPKC, they form the Par complex, which is localized at the apical neuroblast cortex. It is thought that aPKC phosphorylates and inactivates the cytoskeletal protein Lethal (2) Giant Larvae (Lgl) on the apical cell cortex. Since Lgl is required for the cortical localization of cell fate determinants, this could explain why the determinants concentrate at the basal side in a Par complex-dependent manner (Bowman, 2006).

In addition to Baz, Par-6, and aPKC, several other proteins are part of the apical complex. The Armadillo repeat protein Inscuteable (Insc) binds to Baz and aPKC. Its main function seems to be in spindle orientation since the ectopic expression of Insc in epithelial cells can trigger a reorientation of the spindle along the apical-basal axis. Insc, in turn, interacts with the N terminus of Pins (Partner of Inscuteable), an adaptor protein that contains three GoLoco motifs in its C terminus. GoLoco motifs are unique in their ability to bind heterotrimeric G protein α subunits in their GDP bound form and catalyze the dissociation of βγ subunits in a receptor-independent manner. Insc, Pins, and its binding partner Gαi, as well as the Par proteins, are all required for apical-basal orientation of the mitotic spindle, suggesting that the apical complex generates an attachment site for astral microtubules to anchor the spindle in an apical-basal orientation. Since the individual members of the complex depend on one another for apical localization, it is not clear which complex member makes the molecular link with spindle microtubules (Bowman, 2006).

The role of heterotrimeric G proteins in spindle positioning is conserved in other organisms. During the first division of the C. elegans zygote, the G proteins GOA-1 and GPA-16 as well as two nearly identical GoLoco motif binding partners called GPR-1 and -2 are essential for posterior displacement of the mitotic spindle. In this case, GPR-1 and -2 bind to LIN-5, a coiled-coil protein that is also required for correct spindle positioning. Since LIN-5 localizes to the mitotic spindle, it is a good candidate for the molecule that connects G proteins to the mitotic spindle in C. elegans. G proteins are also required for spindle orientation in vertebrates. They bind to the vertebrate homolog of Pins, which, in turn, interacts with NuMA (Du, 2004; Du, 2001), a microtubule binding protein that is essential for proper organization of the mitotic spindle (Fant, 2004). NuMA can enhance microtubule polymerization (Du, 2002) and interacts with the minus end-directed motor dynein (Merdes, 1996), and both effects could provide a mechanistic explanation for a potential role in spindle orientation. However, the pleiotropic effects of NuMA in vertebrates and the lack of invertebrate homologs have so far made it impossible to directly address a requirement of NuMA for spindle orientation. It was proposed that Insc might be a functional NuMA homolog in flies, but the recent identification of Insc homologs in C. elegans and mice make this unlikely. This study identifies the Drosophila protein Mushroom Body Defect (Mud) and the C. elegans protein LIN-5 as sequence homologs of NuMA in invertebrates. Mud is shown to bind to Pins and Gαi, and, like NuMA, Mud can enhance microtubule polymerization. mud mutants have defects in spindle orientation leading to missegregation of cell fate determinants and failed asymmetric divisions that produce excess neuroblasts. These results indicate that Mud closes the gap between heterotrimeric G proteins and the mitotic spindle and regulates spindle orientation in asymmetric cell division (Bowman, 2006).

The role of heterotrimeric G proteins in asymmetric cell division is well studied in Drosophila. In embryonic neuroblasts, G proteins make three major contributions: (1) maintenance of the apical localization of Inscuteable and the Par complex, (2) regulation of spindle orientation at metaphase, and (3) generation of spindle asymmetry at anaphase. It is thought that both free Gβγ and Pins-Gαi, as well as Par complex members Baz and aPKC, have a role to play in the control of spindle asymmetry. Whether G proteins can directly regulate spindle orientation is less clear because of the complexity of G protein phenotypes. Misregulation of G proteins can cause Insc and Par complex delocalization as well as spindle orientation defects. As a result, it is difficult to determine whether it is actually G proteins that are responsible for spindle misorientation, or whether the orientation defect is a secondary consequence of a general loss of polarity. mud mutants, however, show spindle misorientation without Insc or Par delocalization. Since Mud binds to Pins and localizes asymmetrically in neuroblasts, this suggests that Pins-Gαi regulates spindle orientation through its interaction with Mud (Bowman, 2006).

In vertebrates, the Pins-Gαi complex is proposed to control the attachment of astral microtubules to the cortex through its interaction with NuMA. This model of spindle positioning is supported by an experiment in which overexpressed Pins causes spindle rocking movements that can be inhibited by coexpressing a short fragment of NuMA or disrupting astral microtubules with low concentrations of nocodazole (Du, 2004). In Drosophila, astral microtubules are also important for spindle positioning. Mutations in centrosomin and asterless prevent the formation of centrosomes and astral microtubules, and neuroblasts in these mutant backgrounds often fail to coordinate the mitotic spindle with the crescent of cell fate determinants at metaphase. Abolishing astral microtubules pharmacologically produces similar results. It is proposed that Mud forms a complex with Pins and Gαi that regulates the attachment of astral microtubules to the cortex, and that this regulation is necessary for the mitotic spindle to assume the correct orientation in asymmetric cell division. In mud mutants, faulty microtubule-cortical attachment results in a failure to coordinate the mitotic spindle with the axis of polarity. Accordingly, the spindle assumes orientations that do not align with the crescents of Insc and Miranda, and regulators of cell size as well as cell fate determinants can be inherited symmetrically (Bowman, 2006).

The identification of Mud and LIN-5 as NuMA homologs indicates that three different model organisms use NuMA-like proteins to regulate spindle movements. During the first division of the C. elegans zygote, the mitotic spindle is set up along the A/P axis in the center of the cell. In anaphase, the spindle rocks vigorously as the posterior centrosome is displaced toward the posterior cortex. Following this division, mitosis begins in the daughter cells, which initially align their centrosomes transverse to the A/P axis. However, the spindle in the posterior cell eventually rotates 90° and orients along the A/P axis. These spindle rocking and displacement movements require the NuMA-like protein LIN-5. Because LIN-5 is found in a complex with the Pins-like GoLoco motif proteins GPR-1 and GPR-2, and because the phenotype of GPR-1/-2 loss of function is nearly identical to that of LIN-5 (Srinivasan, 2003), it is thought that LIN-5 and GPR-1/-2 act together to generate the forces required for spindle rocking and spindle orientation in mitosis (Bowman, 2006).

In rodents, NuMA, mammalian Inscuteable (mInsc), and G proteins regulate spindle orientation in the asymmetric division of self-renewing stem cells. Epidermal stem cells localize mInsc, NuMA, and Pins to an apical crescent and align the spindle parallel to the apical-basal axis. If apical localization of Pins and NuMA is disrupted, spindle orientation becomes randomized (Lechler, 2005). In the developing neocortex, neural progenitors divide with their spindles orthogonal to the apical-basal axis for symmetric divisions and parallel to this axis for asymmetric divisions. Reliable coordination of the spindle with the apical-basal axis during asymmetric division requires mInsc, free Gβγ, and the Pins-like protein AGS3. If the function of any of these proteins is compromised, asymmetric divisions fail because of misoriented spindles (Sanada, 2005; Zigman, 2005). Furthermore, NuMA and Pins can create spindle-rocking movements during mitotis (Du, 2004). This work shows that the NuMA-like protein Mud forms a complex with Pins and Gαi and is required for spindle orientation in asymmetrically dividing Drosophila neuroblasts. Taken together, these studies strongly suggest that asymmetric cell divisions in C. elegans, Drosophila, and vertebrates all use NuMA-Pins-Gαi complexes to regulate spindle orientation (Bowman, 2006).

In mud mutants, failure of asymmetric division leads to an expansion of the neuroblast pool. This places mud with lgl and brat in a class of genes in which zygotic loss of function produces ectopic neuroblasts. Because of the interaction of Pins with Mud, pins mutants could also be expected to have defective spindle orientation and symmetric divisions that produce two neuroblasts. Surprisingly, pins mutant neuroblasts do not overproliferate (Lee, 2006a). In fact, they exhibit a mild underproliferation phenotype (Bowman, 2006).

How can the difference in the proliferative behavior of mud and pins mutant neuroblasts be explained? First, the possibility that in addition to regulating spindle orientation, Mud directly inhibits proliferation by an unknown mechanism cannot be excluded. Since the overproliferation in mud mutants is mild compared to that in lgl or brat mutants (Betschinger, 2006; Lee, 2006a; Lee, 2006b), this seems unlikely. Second, Pins could be acting redundantly with Loco to regulate spindle orientation, so a potential pins mutant overproliferation is masked by the presence of Loco. Since Mud-C does not bind to Loco under the same conditions with which it binds to Pins, the notion that Loco substitutes for Pins by interacting with Mud is questionable. Alternatively, the proliferative differences could be explained by the localization of aPKC. A recent study in larval neuroblasts suggests that inheritance of cortical aPKC can confer the ability to self-renew (Lee, 2006a). Since work in embryos has shown that Pins is required to maintain the apical localization of the Par complex, it follows that in pins mutant brains, aPKC localizes weakly to the cortex and cytoplasm of metaphase neuroblasts (Lee, 2006a). By contrast, aPKC forms a cortical crescent in mud mutants. In this model, pins mutant daughter cells inheriting cytoplasmic aPKC are more likely to exit the cell cycle, while, in mud mutants, the daughter cells inheriting cortical aPKC continue to proliferate as neuroblasts. The data neither prove nor disprove this hypothesis (Bowman, 2006).

As in Drosophila, genetic lesions affecting NuMA cause overproliferation and cancer in humans. Acute promyelocytic leukemia (APL) can be caused by a chromosomal translocation that creates a NuMA fusion protein (Wells, 1997). Expression of this fusion protein in mice causes a leukemia indistinguishable from human APL (Sukhai, 2004). In addition, variations in the NuMA gene region on chromosome 11 have been associated with breast cancer susceptibility (Kammerer, 2005). NuMA is part of a conserved heterotrimeric complex that regulates spindle orientation, and, consequently, the NuMA-like protein Mud can influence proliferation in the asymmetrically dividing and self-renewing neuroblasts of Drosophila. In light of the cancer stem cell hypothesis, which proposes that a small fraction of cells in a tumor have the ability to proliferate and self-renew, the evolutionary conservation of protein complexes that regulate spindle orientation and proliferation suggests that Drosophila neuroblasts are useful as a cancer stem cell model (Bowman, 2006).