news and views - Nature

news and views

Cytokinesis: mind the GAP Pier Paolo D’Avino and David M. Glover Cytokinesis ensures proper partitioning of genomic and cytoplasmic material between dividing cells. A key regulator of this process is the centralspindlin complex. Two recent papers report that GAP activity of one of the members of this complex regulates the function of Rho family GTPases during cytokinesis. Cytokinesis is the final act of cell division, when a single cell is split into two daughter cells. This separation is accomplished through the formation and ingression of a cleavage furrow (CF) between the separating anaphase chromosomes. CF ingression is driven by the assembly and contraction of actomyosin filaments in the form of a highly organized and dynamic structure known as the actomyosin contractile ring (CR). In metazoans, CR assembly and constriction require the activity of the small GTPase RhoA, which, like all small GTPases, can exist in two states: an inactive or GDP-bound form, and an active or GTP-bound state. Cycling between these alternative states is controlled by activators, guanine nucleotide exchange factors (GEFs), and inhibitors, GTPase activating proteins (GAPs). A key factor responsible for RhoA activation at the cleavage site is an evolutionarily conserved two-protein complex termed centralspindlin1. This complex is composed of a plusend directed microtubule motor protein, known as ZEN‑4 in Caenorhabditis elegans, Pavarotti (Pav) in Drosophila melanogaster and MKLP1 in mammals, and a Rho family GAP called CYK‑4 in C. elegans, RacGAP50C in Drosophila and MgcRacGAP in vertebrates (Fig. 1). This complex is essential for the assembly of a microtubule structure, the central spindle, which forms between the separating anaphase chromosomes (hence its name). Through the activity of its motor component, centralspindlin localizes to Pier Paolo D’Avino is in the Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QP, UK. David M. Glover is in the Cancer Research UK Cell Cycle Genetics Research Group, Department of Genetics, University of Cambridge, Downing Street, Cambridge CB2 3EH, UK. e‑mail: [email protected]; [email protected]

112

the plus-ends of both equatorial and central spindle microtubules (Fig. 1). Evidence from Drosophila and mammals indicates that the RacGAP component of centralspindlin binds and activates a RhoGEF known as Pebble (Pbl) in Drosophila and ECT2 in mammals, which is ultimately responsible for RhoA activation at the cleavage site2–6 (Fig.1). However, a longstanding question in the field has been why a GAP molecule, which is an inhibitor of small GTPases, is required as part of the process to activate RhoA and promote CR formation and constriction. Two recent papers7,8 add to the ongoing debate about the specificity of the CYK‑4 family of RacGAP proteins: one group provides evidence to further support the notion that this protein inhibits the Rac sub-family of Rho-related GTPases7, the other that it restricts the activity of RhoA itself at the CF8. Some years ago we reported a series of genetic experiments indicating that RacGAP50C, the Drosophila CYK‑4 molecule, was able to inhibit the function of the Rac GTPases9. From these data we proposed a model in which the RacGAP of centralspindlin is required to simultaneously inhibit Rac GTPases and activate RhoA at the cleavage site. Subsequently, we further speculated that inhibition of Rac GTPases could be important for reorganization of the actomyosin meshwork at the cortex (possibly by inactivating crosslinking factors), thereby allowing RhoA to rearrange the actomyosin filaments in dense bundles to drive constriction of the CR10. Subsequent finding by Zavortink and colleagues11 that a RacGAP50C variant mutated in its GAP domain could not rescue the cytokinesis failure of RacGAP50C/tum mutants further indicated that GAP activity of this molecule was essential for CF ingression.

In a recent report, Canman et al.7 provide further support for this ‘Rac inactivation’ model, showing that Rac inhibition by the GAP activity of CYK‑4 is essential for cytokinesis in C. elegans. The authors studied the consequences of point mutations that substituted two residues predicted to have an important role in the structure of the CYK‑4 GAP domain. They showed by live imaging that CF ingression was impaired in these potentially GAP-inactive mutants, similarly to other CYK‑4 loss-of-function alleles. However, this phenotype was independent of the role of CYK‑4 in central spindle assembly because this structure was intact and the other centralspindlin component ZEN‑4 localized properly. Canman et al. then found that depletion of RhoA by RNA interference (RNAi) could enhance the CF defects of these presumptive GAP-inactive CYK‑4 mutants, whereas depletion of the two Rac proteins rescued CF ingression. These results indicate that GAP activity of CYK‑4 is necessary to downregulate Rac GTPases at the cleavage site to promote CF ingression, consistent with previous models based on data from both Drosophila and mammals9,12. The authors, however, went a step further and showed that either co-depletion of the Rac effectors WASp and WAVE, or RNAi knockdown of their downstream target, the Arp2/3 complex, could also rescue the CF ingression defects of CYK‑4 mutants. They concluded that CYK‑4-directed downregulation of Rac at the cleavage site is important to reduce the activity of the Arp2/3 complex (Fig.1). As the Arp2/3 complex promotes branching of actomyosin filaments, these authors suggested that inactivation of Rac is necessary to ultimately reduce the formation of a branched actomyosin meshwork at the CF, which may interfere with CR constriction.

nature cell biology volume 11 | number 2 | FEBRUARY 2009 © 2009 Macmillan Publishers Limited. All rights reserved.

news and views Rho flux

Rac inactivation

Rho–GTP Rac

WAVE WASp

Arp2/3

Rho–GDP

Microtubule Pbl/ECT2 CYK-4, RacGAP50C, MgcRacGAP ZEN-4, Pav, MKLP1

Equatorial microtubules

Central spindle microtubules

Figure 1 The role of the RacGAP component of the centralspindlin complex in regulating the activity of Rho family GTPases during cytokinesis. Centralspindlin moves to the plus ends of equatorial and central spindle microtubules through the activity of its motor component (ZEN‑4, Pav, MKLP1). The RacGAP component (CYK‑4, RacGAP50C, MgcRacGAP) interacts with the RhoGEF Pbl/ECT2 to promote RhoA activation at the cleavage site. GAP activity of this protein has also been proposed to inhibit 1) RhoA activity at the cleavage site, thereby promoting its flux through the GTPase cycle (Rho flux model), and 2) Rac GTPases and consequently their effectors WAVE and WASp and their target the Arp2/3 complex (Rac inactivation model).

Despite the strength of the genetic data, neither our study in Drosophila9 nor that of Canman et al. in C. elegans7 directly examined the biochemical activity of RhoA and Rac GTPases. However, this was undertaken in a recent study by Miller and Bement8, who showed that the GAP activity of MgcRacGAP is necessary to inhibit RhoA in the vicinity of the CF in Xenopus laevis embryos. These authors inhibited expression of the MgcRacGAP gene in Xenopus embryos using antisense morpholino oligonucleotides (MO) targeting its 5ʹ-UTR. These MO‑treated embryos showed cytokinesis failure that could be rescued by injecting wild-type MgcRacGAP mRNA lacking the 5ʹ-UTR, but not by injecting two mutated mRNAs: one containing a substitution in an arginine residue essential for GAP activity (R384A)13, and the other depleted of the entire GAP domain (ΔGAP). As both mutant proteins localized to the central spindle, interacted with MKLP1 and did not cause defects in central spindle assembly, these results indicate that the GAP activity of MgcRacGAP was necessary to rescue the cytokinesis defects of MO‑treated embryos. To monitor the activity of members of the three sub-families of Rho GTPases (RhoA,

Rac and Cdc42) in cells expressing the GAPmutated version of MgcRacGAP, the authors used specific GFP probes. They found no detectable effect on the activity of Rac and Cdc42, whereas the zones of RhoA activity at the CF were broader and more intense than in control cells. Cells expressing MgcRacGAPΔGAP mRNA also showed lateral oscillations of the RhoA activity zone. The broader RhoA activity zones detected in these GAP mutants resembled those of cells expressing a constitutively active RhoA and were observed either in the presence or absence of the endogenous MgcRacGAP, indicating a strong dominant effect. Surprisingly, however, these mutants also caused an increase in RhoA activity away from the CF, suggesting a potential pleiotropic effect. From these data, Miller and Bement8 proposed that the GAP domain of MgcRacGAP has a dual function: it can inhibit RhoA activity at the CF to promote its flux through the GTPase cycle (Fig. 1), and it may be also necessary to anchor this GTPase at the CR during furrow ingression. A common conclusion of these two new studies7,8 is that GAP activity of the centralspindlin complex is essential for cytokinesis in both worms and vertebrates, confirming

earlier findings in Drosophila and mammals11,14. However, it is difficult to reconcile the Rac inactivation and Rho flux models. Perhaps they need not be mutually exclusive and the RacGAP component of centralspindlin can inactivate both RhoA and Rac GTPases. We cannot exclude the possibility that these apparently contrasting functions may be relevant in different cell types or in differing regions of the cell cortex. For example, some cell types may posses an intricate actomyosin network at the future cleavage site, which could interfere with the assembly and constriction of the CR. In such cells, Rac inactivation would be essential to disassemble the actin network and allow CF ingression. On the other hand, other cell types may lack this dense cortical actomyosin meshwork and in this case it may be more important to tightly regulate the flux of RhoA through its GTPase cycle. We also cannot eliminate the possibility of evolutionary differences between species. In support of these hypotheses, increasing evidence indicates that cytokinesis is an extremely robust process that can use different signalling pathways to regulate a single mechanism. One implication of the Rho flux model8 is that the antagonistic activities of the RacGAP component of centralspindlin and the RhoGEF Pbl/ECT2 must ultimately result in RhoA activation, otherwise CR assembly and constriction could not occur. This suggests that GAP activity must be weaker than its counteracting GEF activity, although active RhoA could be stabilized by interaction with its effectors. Consistently, biochemical studies in both vertebrates and invertebrates indicate that the RacGAP component of centralspindlin is less active against RhoA than it is against Rac and Cdc42 (refs 13, 15, 16). However, one of these studies 13 also showed that in vitro, the R384A mutation in MgcRacGAP abolished its GAP activity against Rac but had no effect on RhoA, a finding that would support the Rac inactivation rather than the Rho flux model. The difficulty in reconciling these findings into a single model suggests that the GAP domain of the RacGAP component of centralspindlin may well have two quite distinct functions. Further studies will be required to harmonize the genetic and biochemical observations that led to these two proposed functions for the GAP activity of what is undoubtedly a master regulator of cytokinesis.


113

news and views 1. Mishima, M., Kaitna, S. & Glotzer, M. Dev. Cell 2, 41–54 (2002). 2. Somers, W. G. & Saint, R. Dev Cell 4, 29–39 (2003). 3. Zhao, W. M. & Fang, G. Proc. Natl Acad. Sci USA 102, 13158–13163 (2005). 4. Yuce, O., Piekny, A. & Glotzer, M. J. Cell Biol. 170, 571–582 (2005). 5. Nishimura, Y. & Yonemura, S. J. Cell Sci. 119, 104– 114 (2006). 6. Kamijo, K. et al. Mol. Biol. Cell 17, 43–55 (2006).

7. Canman, J. C. et al. Science 322, 1543–1546 (2008). 8. Miller, A. L. & Bement, W. M. Nature Cell Biol. 11, 71–77 (2008) 9. D’Avino, P. P., Savoian, M. S. & Glover, D. M. J. Cell Biol. 166, 61–71 (2004). 10. D’Avino, P. P., Savoian, M. S. & Glover, D. M. J. Cell Sci. 118, 1549–1558 (2005). 11. Zavortink, M., Contreras, N., Addy, T., Bejsovec, A. & Saint, R. J. Cell Sci. 118, 5381–5392 (2005).

12. Yoshizaki, H. et al. J. Biol. Chem. 279, 44756–44762 (2004). 13. Kawashima, T. et al. Blood 96, 2116–2124 (2000). 14. Hirose, K., Kawashima, T., Iwamoto, I., Nosaka, T. & Kitamura, T. J. Biol. Chem. 276, 5821–5828 (2001). 15. Toure, A. et al. J. Biol. Chem. 273, 6019–6023 (1998). 16. Jantsch-Plunger, V. et al. J. Cell Biol. 149, 1391–1404 (2000).

Alzheimer’s dementia by circulation disorders: when trees hide the forest Carlos G. Dotti and Bart De Strooper Deposition of amyloid β-peptide in cerebral vessel walls, termed cerebral amyloid angiopathy (CAA), enhances the cognitive deficits associated with Alzheimer’s disease. The molecular details by which circulatory defects with hypoxia alter peptide clearance, contributing to brain deposition and AD, are beginning to be elucidated. Alzheimer’s disease (AD), the most common form of dementia, is characterized by the presence of hyperphosphorylated tau filaments in neurons (neurofibrillary tangles), and aggregates of amyloid β‑peptide (Aβ) in the brain parenchyma (amyloid plaques) and in the walls of small brain arteries, leading to cerebral amyloid angiopathy (CAA). The degree of amyloid deposition ranges from a thin ring of amyloid in the vessel wall to large plaque-like extrusions into the brain parenchyma. CAA is also associated with local loss of neurons, synaptic abnormalities, microglial activation and microhaemorrhage. Clearly, such defects will alter neuronal and synaptic function and even at its earliest stage, amyloid deposits around brain vessels could certainly interfere with the dynamic adaptation of cerebral blood flow (CBF) to changing brain function. Blood flow in the brain is known to vary according to local metabolic demands, and reductions in CBF are associated with decreased cerebral protein synthesis, changes in intracellular pH, accumulation of toxic by-products in brain interstitial fluid and eventually, inability of neurons to fire action potentials1. A question that arises, therefore, is to what extent vascular diseases could Carlos G. Dotti and Bart De Strooper are in the VIB Department of Molecular and Developmental Genetics (VIB11) and Center for Human Genetics, Katholieke Universiteit Herestraat 49, PO Box 603 3000, Leuven, Leuven, Belgium. e-mail: [email protected]

114

directly underlie CAA and therefore trigger AD, which is different from the canonical view that CAA is a consequence of a primary neuronal defect. Although there is no doubt that a number of AD cases can be explained by a primary defect in neuronal function, there is also substantial evidence to support the notion that circulatory defects may have a primordial, initiating, role2‑4. However, the molecular pathways by which a peripheral vascular defect could lead to brain amyloid deposition have not been clearly defined. On page 143 of this issue, Bell et al. provide a molecular mechanism that could explain how vascular defects may lead to reduced amyloid clearance, and thus Alzheimer’s pathology, by showing that hypoxia in vascular smooth muscle cells (VSMCs) of meningeal arterioles induces transcription factors that regulate the expression of the low-density lipoprotein receptor related protein 1 (LRP1, ref. 5), a major efflux transporter for Aβ across the blood brain barrier6. Blood-brain barrier endothelial cells and pericytes have an active role in transferring soluble Aβ from the brain to the blood, by LRP1 binding to Aβ at the ablumenal (brain) side of the blood-brain barrier (BBB) followed by transcytosis4. Additionally, Aβ can be removed from the brain by VSMCs in brain arterioles, which rapidly and efficiently internalize exogenous Aβ by a mechanism relying on the LRP1 pathway7,8. Given that VSMCs are important in the control of CBF, through

the contraction and relaxation of arterioles and small pial arteries, it is conceivable that any perturbation in the physiology of VSMCs could have a major effect in CAA and AD. There are several ways one can imagine VSMC dysfunction leading to AD symptoms. The classical, neuronal-initiated hypothesis of AD would predict that VSMC defects arises as a result of excess Aβ release from axonal terminals into the vicinity of the neurovascular (arteriolar) unit (Fig. 1). This would lead to local toxicity, reduced clearance and increased amyloid deposition, perturbation of CBF control and eventually, haemorrhage and infarction (that is, CAA). Consistent with this possibility, it has been observed that Aβ constricts arteries and counteracts activity-mediated vasodilatation9,10, and that topical application of soluble Aβ produces cerebrovascular alterations in normal mice11. Interestingly, in mice expressing the Swedish mutation of APP, alterations in the microcirculation precede the appearance of amyloid plaque deposits and is followed by cognitive deficits12, indicating that an excess of Aβ could lead to CAA through direct perturbation of amyloid clearance by VSMCs. The presence of excess Aβ in the vicinity of brain vessels could also lead to CAA in capillaries of the blood-brain barrier, not only in small arterioles with VSMCs. At the blood-brain barrier, an excess of neuronal amyloid peptide would saturate the endothelial LRP1 uptake pathway, resulting in accumulation of toxic
