news and views
Cdc42 & GSK-3: signals at the crossroads
f
focus on
cell polarity
Adrian Harwood and Vania M. M. Braga
Cell polarity is regulated by independent pathways that are controlled by Wnt- or Cdc42-mediated signalling. Now, glycogen synthase kinase 3 (GSK-3), an established component of the Wnt pathway, is shown to interact with Par6–protein kinase C ζ (PKCζ), a complex that transduces Cdc42 signals. This exciting result suggests a potential point of conversion between these previously distinct pathways.
olarity is a fundamental cellular feature and underlies a host of biological functions. Reorganization of the microtubule network and regulation of its dynamics are essential for the establishment and maintenance of polarity. A recent study by Etienne-Manneville and Hall1 reveals that during astrocyte migration, the small GTPase Cdc42 triggers microtubule reorganization through GSK-3 and adenomatous polyposis coli (APC), two proteins previously associated with Wnt signalling. Previously, both small GTPases and Wnt signalling have been independently implicated in the regulation of polarity. This work now shows how GSK-3 may sit at the interchange between these ubiquitous signalling pathways. Polarity reflects an asymmetry within the cell that is usually achieved through a differential distribution of organelles, cytoplasmic and membrane proteins, and projections from the cell surface (such as protrusions, microvillae or even axons). Despite the universal nature of cell polarity, basic questions remain unanswered. For instance, it is not known how the polarizing signal is initiated at a particular subcellular location and how asymmetry is then propagated and maintained across the cell. It was work in the nematode worm Caenorhabditis elegans that initially characterized the interaction of Cdc42 with the Par3–Par6–atypical PKC complex during cell polarizing events. In worms, determination of somatic and germ cell fate requires an asymmetric cell division during the first cell cleavage. Essential to this process is the repositioning of the mitotic spindle by PAR3, PAR6 and the atypical PKC, PKC-3 (ref. 2). In addition, symmetric positioning of the spindle is determined by restricting the localization of PAR3–PAR6–PKC-3 to the anterior pole of the zygote, where they interact with Cdc42. Mutation of any of these genes abolishes the asymmetric division and disturbs partitioning of cell-fate determinants. Cdc42 is a member of the Rho GTPase subfamily and participates in many signalling
P
?
Cdc42
Rac
Protrusion formation Par6
PKCζ
Protrusion orientation GSK-3
EB1
?
Microtubule ends stabilization?
APC
Dynein
MTOC polarization
Figure 1 Polarization of migrating astrocytes. After wounding of an astrocyte monolayer, cells protrude their membrane towards the wound, re-orientate the MTOC and migrate to close the empty space. Formation of protrusions requires Rac and Cdc42 function. Cdc42 is transiently activated at the leading edge of astrocytes (facing the wound). Whether Rac is activated downstream of Cdc42 or through independent pathways is not known. Next, Cdc42-dependent translocation of Par6–PKCζζ to the leading edge is accompanied by activation of PKCζζ. GSK-3 (which is found constitutively associated with Par6–PKCζζ) is inactivated by phosphorylation and dissociates from the complex; however, inactive GSK-3 is also found at the leading edge. This restricted spatial inactivation of GSK-3 may allow APC to stabilize the growing end of microtubules and provides a mechanism for MTOC polarization and protrusion formation (which strictly requires the microtubule network). Additional microtubule-binding proteins, such as dynein (microtubule-based motor) and EB1 (interacts with APC), may also participate in these processes. Interestingly, PKCζζ and GSK-3 seem to regulate the orientation of the protrusions towards the wound. The orientation of protrusions has not been evaluated during inhibition of APC, so it is unclear if APC is also important in this process. Nevertheless, these results suggest that protrusion formation, orientation and MTOC polarization are distinct processes regulated by different mechanisms, but tightly coordinated1,5,6. Dashed lines indicate that Cdc42 has been shown to interact directly with Par6 in epithelial cells, but not in astrocytes1. Although GSK-3 interacts with the Par6–PKCζζ complex, it is not known whether the latter kinase is responsible for phosphorylation and inactivation of GSK-3. EB1 interacts with APC13; in astrocytes, however, EB1 localization at the leading edge is independent of Cdc42 (ref. 1).
pathways, but is particularly important in cytoskeletal remodelling3. The Cdc42 interaction with a Par3–Par6–atypical PKC complex is conserved throughout animal evolution and is required for more than asymmetric division. For example, it localizes to cell–cell junctions and is crucial for the generation of apicobasal polarity in
NATURE CELL BIOLOGY VOL 5 APRIL 2003 www.nature.com/naturecellbiology © 2003 Nature Publishing Group
epithelial cells3. In developing neurons, Cdc42–Par3–Par6–atypical PKC is required to specify and maintain the polarity of the nascent axon4. In addition, components of this complex have been implicated in cell polarization during migration5. Migrating cells protrude membrane and re-orientate their Golgi apparatus and 275
news and views a
b Canonical Wnt signalling Wnt
c Microtubule dynamics Wnt7a
Fz
Axin
MOM-2 Fz
Dsh
Dsh
GSK-3
GSK-3
APC β-catenin
Gene expression
Early nematode development
MOM-5
GSK-3
MAP1B
Microtubule dynamics
Wrm-1
Gene expression
Mitotic spindle rotation
Figure 2 Wnt signalling — a simplified scheme. a, In the canonical Wnt signalling pathway, the Wnt ligand binds to the receptor Frizzled (Fz) and inhibits GSK-3 action through Dishevelled (Dsh). GSK-3 phosphorylates β-catenin through a complex containing Axin and APC. Phosphorylated β-catenin is degraded, blocking its effect on gene expression. Two unconventional (non-canonical) Wnt signalling pathways interact with the microtubule cytoskeleton. b, In cerebellar axons, Wnt7a inhibits GSK-3 activity to increase microtubule dynamics. Inhibition of GSK-3 decreases phosphorylation of the microtubule-associated protein MAP1B to reduce microtubule stability. c, During early development of the nematode embryo, the EMS cell divides to produce daughter cells that differentiate into intestinal or pharyngeal and body wall muscle cells. This process is regulated by binding of a Wnt signal (MOM-2) to an Fz receptor (MOM-5), which functions positively on GSK-3. In this pathway, GSK-3 controls gene expression through a β-catenin homologue (Wrm-1) and regulates the plane of cell division through a 90°° rotation of the mitotic spindle.
microtubule organizing centre (MTOC) towards their direction of movement. Although these processes are coordinated, they may be regulated by different mechanisms1,5,6. Etienne-Manneville and Hall investigated migration of primary astrocyte cells in monolayer culture by using a wounding assay. When a micropipette is scraped through the monolayer, cells adjacent to the wound re-orientate and then migrate to fill vacated space while remaining attached to their neighbours. In contrast to fibroblasts and epithelial cells, astrocytes migrate by extending highly elongated protrusions at the wound border; this process is driven by the microtubule cytoskeleton, rather than by microfilaments3. At the molecular level, we know that the astrocyte wound response results in the relocalization of Cdc42 to the leading edge of the cell and the formation of a complex between Par6 and the atypical PKC, PKCζ (Fig. 1)5. Interestingly, and in contrast to other systems, this relocalization during migration does not involve Par3. Dominant-negative PKCζ and Par6 all inhibit microtubule re-orientation, but not membrane protrusion (Fig. 1)5. Drawing on previous indications of a link between GSK-3 and microtubule organization and dynamics, EtienneManneville and Hall have now investigated the role of GSK-3 in astrocyte migration. Two lines of evidence connect GSK-3 and 276
microtubules: regulation of microtubule stability and the Wnt-dependent polarization processes. GSK-3 regulates microtubule stability by modulating the function of other microtubule-binding proteins, such as MAP1B and the APC protein7,8. APC binds to the plus ends of microtubules and stabilizes the growing ends, an activity that is decreased by phosphorylation of GSK-3 (ref. 8). In the canonical Wnt pathway, GSK-3 forms a complex with Axin and APC, and regulates gene expression through degradation of the signalling protein β-catenin (Fig. 2). However, there are also non-canonical Wnt pathways through which GSK-3 regulates the microtubule cytoskeleton, such as the Wnt pathway that functions during outgrowth of cerebellar axons and rotation of the mitotic spindle during early development in C. elegans (Fig. 2)7,9. Consistent with a regulatory role in mitosis, GSK-3 is present along spindle microtubules, but inactivated at the centrosome (the MTOC present during mitosis)10. Inhibition of GSK-3 increases the length of astral microtubules and causes defects in chromosomal alignment. GSK-3 is therefore an ideal target to regulate the position, orientation and dynamics of the mitotic spindle. APC is also required during mitosis, where it attaches mitotic chromosomes to the microtubules within the spindle. Loss of the carboxyl terminus of APC removes its ability to bind microtubules and results
in chromosomal instability. Such truncations of APC are frequently observed in colorectal cancer, which is associated with a high degree of genetic instability11. Etienne-Manneville and Hall now find that in migrating astrocytes, GSK-3 is present in a complex with Par6 and PKCζ1. Wounding also results in a Cdc42-dependent inhibition of GSK-3 through phosphorylation on Ser 9, which correlates with its dissociation from Par6. Inhibition of either Cdc42 or PKCζ, or expression of a constitutively active version of GSK-3, blocks reorientation of the MTOC. Cdc42, Par6–PKCζ and phosphorylated GSK-3 all concentrate at the leading edge, precisely where a polarizing signal should be active. Blocking GSK-3 or PKCζ, however, does not abolish formation of membrane protrusions. Instead, they become randomly orientated, suggesting an overlap with a second polarization mechanism. This new role of GSK-3 in cell polarization is all the more remarkable when we consider its downstream targets. Both β-catenin and APC localize to the leading edge in response to wounding, and the authors find that this depends on activation of Cdc42 and PKCζ or inactivation of GSK-3 (ref. 1). The localization of β-catenin can be perturbed by disruption of actin filaments without affecting cell polarity, suggesting that it does not regulate MTOC orientation. However, expression of a truncated APC protein lacking the microtubule interaction domain blocks MTOC re-orientation. This establishes a signalling pathway from Cdc42–Par6–PKCζ-mediated inhibition of GSK-3 to the potential regulation of microtubule dynamics by APC. Paradoxically, both activation and inhibition of GSK-3 block MTOC re-orientation1. This phenomenon has previously been highlighted through the use of dominant-negative and constitutively active Cdc42 (ref. 5) and further ties Cdc42 function to the regulation of GSK-3. The data also supports the notion that a tightly regulated and localized activation of signalling is essential for generating polarity. These results could be explained by a requirement for Cdc42 to cycle between an ‘on’ and ‘off’ state to relay its signal. However, an alternative scenario, in which Cdc42 and GSK-3 have both positive and negative effects on MTOC polarization, cannot currently be discounted. Naturally, these results raise a number of questions. How does the Cdc42–Par6–PKCζ complex specifically localize to the leading edge? In other systems, Par3 is involved in the restricted localization of Par6–PKCζ, but this does not seem to be the case during astrocyte migration. One proposal entails the integrin-mediated stimulation of Src-like kinases5; whatever the mechanism, it needs to take into account the activation kinetics of the various components. Which kinase is responsible for GSK-3 phosphorylation and
NATURE CELL BIOLOGY VOL 5 APRIL 2003 www.nature.com/naturecellbiology © 2003 Nature Publishing Group
news and views inactivation? It is tempting to suggest PKCζ, but there is currently no direct biochemical evidence for this. Which other Cdc42 targets could be involved in membrane protrusion initiation? Cdc42 is necessary to form protrusions, whereas the other components of the pathway are not (Fig. 1). Instead, PKCζ and GSK-3 are required to regulate the orientation of protrusions towards the wound (Fig. 1). So, which other additional signalling pathways control the positioning of protrusions? Is APC also involved? At a more general level, Cdc42 is now a hot candidate for regulating microtubule dynamics1,12. Thus, Cdc42 and the other family members Rho and Rac can regulate the dynamics of both microtubule and intermediate filaments13,14, functions beyond their established role in actin cytoskeleton. To reorganize the MTOC in migrating astrocytes, polarization signals initiated by Cdc42–Par6–PKCζ use GSK-3 and APC, two components of the Wnt pathway. Clearly, astrocyte migration need not require a Wnt
signal to generate the polarizing signal. Could two independent pathways that regulate polarization use the same proteins to modulate microtubules (Figs 1, 2)? Or could it be that Cdc42 and Wnt pathways interact with each other? Interestingly, a similarity between Cdc42–Par6–PKCζ and unconventional Wnt signal pathways is that β-catenin is not required for pathways that result in reorganization of microtubules. Could the formation of a complex between Par6–PKCζ and GSK-3 represent an alternative mode of β-catenin and APC regulation? Obviously, it will be interesting to re-examine Wnt signalling for an interaction with Cdc42 and PKCζ (and vice versa). Whatever the answers to these questions, the new results derived from astrocyte migration certainly point us in new directions for unravelling the regulation of cell polarity by Cdc42 and Wnt. Adrain Harwood is in the MRC Laboratory of Molecular Cell Biology & Department of Biology, University College London, Gower Street, London, WC1E 6BT.
Leading the pack
Vania M. M. Braga is in the Cell and Molecular Biology Section, Division of Biomedical Sciences, Faculty of Medicine, Imperial College London, Sir Alexander Fleming Building, London e-mail:
[email protected] 1. 2. 3. 4. 5. 6.
Etienne-Manneville, S. & Hall, A. Nature 421, 753–756 (2003). Ahringer, J. Curr. Opin. Cell Biol. 15, 73–81 (2003). Etienne-Manneville, S. & Hall, A. Nature 420, 629–635 (2002). Shi, S. H., Jan, L. Y. & Jan, Y. N. Cell 112, 63–75 2002 Etienne-Manneville, S. & Hall, A. Cell 106, 489–496 (2001). Magdalena, J., Millard, T. H. & Machesky, L. M. J. Cell Sci. 116, 743–756 (2003). 7. Lucas, F., Goold, R., Gordon-Weeks, P. & Salinas, P. J. Cell Sci. 111, 1351–1361 (1998). 8. Zumbrunn, J., Kinoshita, K., Hyman, A. A. & Nathke, I. S. Curr. Biol. 11, 44–49 (2001). 9. Schlesinger, A. et al. Genes Dev. 13, 2028–2038 (1999). 10. Wakefield, J. G., Stephens, D. J. & Tavare, J. M. J. Cell Sci. 116, 637–646 (2003). 11. Fodde, R. et al. Nature Cell Biol. 2, 433–438 (2001). 12. Palazzo, A. F. et al. Curr. Biol. 11, 1536–1541 (2001). 13. Gundersen, G. G. & Cook, T. A. Cur. Opin. Cell Biol. 11, 81–94 (1999). 14. Braga, V. M. M. Curr. Opin. Cell Biol. 14, 546–556 (2002).
ACKNOWLEDGEMENTS A.J.H. is a Wellcome Senior Fellow; V.M.M.B. is a Medical Research Council Senior Research Fellow.
f
The ability of cells to polarize and migrate up a chemical gradient is key to a variety of processes, from axon guidance to leukocyte homing. This ability to chemotax is also key to the survival of the social amoebae Dictyostelium discoideum during times of starvation, and studies in this system have provided important insights into the molecular mechanisms directing chemotaxis. One key question in the field is how migrating cells amplify chemo-attractant gradients, and new work published by Kriebel et al. (Cell 112, 549–560 (2003)) now provides a model for this. They show that D. discoideum cells amplify the gradient of the chemo-attractant cyclic AMP by asymmetrically localizing the enzyme responsible for its generation. On exposure to a cAMP chemo-attractant gradient, D. discoideum cells polarize, begin to migrate towards the gradient source, and then line up head-to-tail in streams to form aggregates that are key to their survival. Now, Kriebel et al. propose that by localizing adenylyl cyclase (ACA) — the enzyme responsible for generating cAMP — to the uropod during polarization, migrating cells can then themselves produce cAMP from their tail. In this way, they provide an amplified cAMP signal for the cells behind to follow. Cells lacking ACA can polarize when exposed to a chemoattractant gradient, but cannot then line up to form streams. To see where ACA localizes in the cell, the authors fused ACA to yellow fluorescent protein (YFP) and found that after polarization, ACA is highly enriched at the uropod. Intriguingly, ACA is also present in intracellular vesicles, suggesting that vesicular transport may be involved in its localization or in the signal transduction pathways that result in its activation. So what is the pathway that mediates ACA localization? Neither CRAC (cytosolic regulator of adenylyl cyclase), which activates ACA, nor PKA (protein kinase A), a target of ACA, seem to be involved. By perturbing the actin cytoskeleton with drug treatments, the authors find that the actin cytoskeleton is important for ACA localization, but the exact mechanism by which this takes place remains to be established. The enzymatic activity of ACA also affects its localization, as cells expressing a
focus on
cell polarity
D. discoideum amobae chemotax towards a gradient source. As they migrate, D. discoideum line up in a head-to-tail fashion to form aggregates. Imaging ACA–YFP reveals plasma membrane labelling that is highly enriched at the uropod of polarized cells. Image provided by C. Parent.
constitutively active mutant of ACA are also unable to stream. The importance of this work is that it suggests an amplifying mechanism by which migrating cells responding to a chemoattractant can then themselves become a source of chemoattractant for other cells to follow. This signal relay may also be common to other mammalian cell types. In leukocytes, for example, it is known that migrating cells secrete chemo-attractant in response to chemo-attractant exposure. This work, therefore, may provide a basis for future studies into the relay mechanisms used by diverse chemotactic cell types. ALISON SCHULDT
NATURE CELL BIOLOGY VOL 5 APRIL 2003 www.nature.com/naturecellbiology © 2003 Nature Publishing Group
277
