SEARCH

SEARCH BY CITATION

Keywords:

  • actin;
  • annexin;
  • cytoskeleton

Abstract

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References

The actin cytoskeleton is a malleable framework of polymerised actin monomers that may be rapidly restructured to enable diverse cellular activities such as motility, endocytosis and cytokinesis. The regulation of actin dynamics involves the coordinated activity of numerous proteins, among which members of the annexin family of Ca2+- and phospholipid-binding proteins play an important role. Although the roles of annexins in actin dynamics are not understood at a mechanistic level, annexins have the requisite properties to integrate Ca2+-signaling with actin dynamics at membrane contact sites. In this review we discuss the current state of knowledge on this topic, and consider how and where annexins may fit into the complex molecular machinery that regulates the actin cytoskeleton.

Annexins are a family of cytosolic Ca2+-binding proteins characterized by non-EF hand or type II Ca2+ binding sites. Ca2+-binding enables annexins to interact with membranes containing acidic phospholipids, and Ca2+-dependent phospholipid binding is considered the unifying biochemical property in the annexin family. Ca2+-, and thereby Ca2+-dependent, phospholipid binding is mediated through a protein domain common to all annexins, the annexin core. Annexin cores are composed of four (eight in annexin 6) annexin repeats, these being tandem homologous segments of 70–80 amino acids. Crystal structures of annexin cores show that they are compact and highly α-helical entities forming slightly curved discs with a convex, Ca2+/membrane-binding surface and a concave side facing towards the cytoplasm in membrane-bound annexins. The high resolution structures also reveal that it is the architecture of the type II Ca2+ sites which enables annexin proteins to associate with the cytosolic face of cellular membranes. Ca2+ ions bound in these sites provide a bridge between the protein and phosphoryl moieties of the glycerophospholipid backbone that enables annexins to peripherally interact with membrane surfaces. Annexins differ in their N-terminal domains, which precede the core and often mediate binding to specific protein ligands. In some cases the N-terminal domains can probably fold as distinct structural moieties, whereas in the majority of annexins, particularly those with N-terminal sequences of only 10 or 20 residues, these regions are either unfolded or fold back on the protein core (1–3). Structural and biochemical data support the view that the annexin core is a module that mediates a unique type of Ca2+-regulated and in most cases reversible membrane binding. The N-terminal domain can modulate this membrane binding activity and participates in specifying the cellular target membrane for a given annexin and the type of membrane interaction (4).

Annexins were first identified in higher vertebrates but have now been described in a wide range of species including invertebrates, plants, insects and protists (5). Usually more than one annexin is expressed in a given cell type, indicating that different members of the family serve distinct and different functions. However, annexins are rather similar in structure, biochemical properties and subcellular distribution, suggesting that some annexins could functionally replace one another and possibly explaining the subtle phenotypes observed in some of the recent knockout animals.

The most detailed functional characterization has focussed on the vertebrate annexins, which can be grouped into 12 subfamilies (4). Some of these subfamilies exhibit distinctive features in their N-terminal domains, in particular annexins 1 and 2, which harbour phosphorylation sites for different signal transducing kinases and also binding sites for two EF-hand Ca2+-binding proteins (S100A11 and S100A10); annexins 7 and 11, which have an extended N-terminal domain rich in GYP repeats capable of interacting with the EF-hand proteins sorcin and S100A6, respectively, and annexin 13, which is N-terminally myristoylated. The different vertebrate annexins have been implicated in a variety of cellular responses to stimulus-induced elevations in intracellular Ca2+ levels. Often these responses involve membrane-related events ranging from the control of membrane skeleton architecture to membrane trafficking and the regulation of ion flux across membranes (4). Such functions are consistent with both the generic ability of annexins to bind to membranes in a Ca2+-regulated manner, and observations that annexins interact with components of the actin cytoskeleton.

Interactions with Actin

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References

Although not a common property within the annexin family, several annexins have been shown to interact directly with polymerized actin in vitro, which correlates with the functions that have been proposed for these annexins in mediating, stabilizing and/or regulating membrane–actin interactions. Annexin 2, both as a monomer and in the heterotetrameric complex with S100A10, was the first annexin shown to be capable of binding to and also bundling actin filaments in a Ca2+-dependent manner (6,7). The Kd for F-actin binding is approximately 0.2 μm, with half maximal binding occurring at submicromolar Ca2+ concentrations in the case of the annexin 2-S100A10 complex and a stoichiometry of one annexin 2 molecule per two actin monomers (8,9). Studies employing synthetic peptides and C-terminally truncated annexin 2 mutants revealed the importance of the very C-terminal sequence LLYLCGGDD for F-actin binding and also showed that a sequence located further N-terminal in the fourth annexin repeat (VLIRIMVSR) appears to be involved in actin filament bundling (10,11). Interestingly the C-terminal sequence required for F-actin binding is conserved in a number of other annexins, e.g. annexin 4, which do not bind F-actin (12). Thus additional parameters probably determine the F-actin binding in annexin 2. F-actin bundling by the annexin 2-S100A10 heterotetramer can probably occur when the two annexin subunits of the complex bind to different actin filaments. Bundling mediated by the monomer, on the other hand, can only be explained by the presence of two F-actin binding sites per annexin 2 molecule or by dimerization or oligomerization of F-actin bound annexin 2.

Despite its ability to bundle actin filaments, annexin 2 is not found associated with prominent actin bundles or cables within cells, such as microvilli or stress fibers. However, more dynamic actin structures, in particular those associated with cellular membranes during, e.g. phagocytosis, pinocytosis and cell migration, contain annexin 2 and probably require the protein (see below). This suggests that annexin 2 is not a general F-actin bundling factor but that the bundling activity observed in vitro reflects a somewhat related function in the assembly of actin structures at cellular membranes. In fact, accumulating evidence indicates that a cellular membrane is required for efficient recruitment of annexin 2 to actin assembly points. This membrane typically contains lipids and proteins characteristic of raft domains (13–16). Moreover, the activity of annexin 2 at these actin assembly points is probably tightly regulated, e.g. by phosphorylation at Tyr-23 in the N-terminal domain, which is catalysed by src-like kinases and inhibits F-actin binding and bundling (17). In view of these observations, it is possible that src-family kinases targeted to raft domains can turn off an actin-assembly activity mediated by annexin 2, which may in turn contribute to the altered actin dynamics characteristic of many tumour cells.

Annexin 1 also binds and bundles actin filaments in a Ca2+-dependent manner, although higher Ca2+-concentrations seem to be required to establish this interaction (18,19). The actin binding observed for annexin 1 in vitro could relate to the finding that the protein co-localizes with F-actin to EGF-induced membrane ruffles and to phagosomal membranes (20,21). Annexin 1 may also influence actin polymerization via a direct interaction with profilin. Although this has only been demonstrated in vitro, the interaction was reported to modulate both the ability of annexin 1 to bind phospholipids and the effects of profilin on actin polymerization (22,23). A direct interaction with a specific actin isoform, γ-actin, has been reported for annexin 5. This might be of particular importance in platelets where annexin 5 is abundant and relocates to the cytoskeleton following stimulus-induced Ca2+ elevation (24). Whether and how this relates to activation-triggered shape changes and granule secretion of platelets remains to be established.

Annexin 6 is another annexin that has been reported to bind actin filaments in a Ca2+-dependent manner (25,26). Again, this binding could be responsible for the observed colocalization of annexin 6 and F-actin at membrane ruffles, microspikes and certain secretory processes (26,27). Moreover, actin binding which occurs at the pointed ends of actin filaments and/or a specific Ca2+-dependent association of annexin 6 with lipid raft domains have been implicated in mediating a relocation of the protein from the cytoplasm to plasma membrane domains in smooth muscle cells (28,29). Such annexin 6 recruitment has been proposed to be required for the dynamic reorganization of the sarcolemma during Ca2+-controlled smooth muscle contraction (30).

The common theme that emerges is that annexin–actin interactions are highly dynamic in nature and occur only in close proximity to cellular membranes. Through these interactions, annexins could participate in a number of cellular events discussed below.

Regulation of Actin Dynamics

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References

The directed polymerization of actin is involved in almost every aspect of eukaryotic cell biology. Actin association with cell–cell contact points (e.g. adherens junctions), focal adhesions and extracellular matrix-binding proteins such as integrins maintains the connectivity of cells with membranes and with neighbouring cells in tissues. Dynamic rearrangement of actin is associated with phagocytosis, pinocytosis, cytokinesis, migration, formation of filopods and lamellipods, endocytosis and exocytosis. Annexins are often identified in genetic or proteomic analyses of subcellular compartments involved in these processes and some annexins directly bind F-actin (see above), but their precise roles have been difficult to establish (Figure 1).

image

Figure 1. Schematic representation of putative interactions of annexins with actin. The figure identifies key points at which annexins may be involved in remodelling of the membrane cytoskeleton, in endocytosis, vesicle transport and motility. A: Nascent endosome. B: Phosphatidylinositol 4,5 bisphosphate. C: Monomeric annexin associated with actin filaments. D: Heterotetrameric and 2/p11 complexes cross-linking actin filaments. E: Actin filaments. F: Cross-linking of extracellular receptors recruits G-proteins (G) and phosphatidylinositol kinases that generate inositol phospholipids. H: Rocketing macropinosome. I: Annexin association with dynamic actin-rich filopods. J: Annexins associated with membrane-cytoskeletal structures in the phagosome. K: Shuttling of actin monomers between annexins and actin-sequestering proteins such as profilin and thymosin β4 may regulate the effective concentration of free G-actin. Original artwork by Matt Hayes.

In spite of these difficulties, drastic and apparently coincident redistribution of annexins and actin filaments in response to extracellular signals supports the idea of direct involvement of annexins in actin remodelling. In some cases the co-localization has been shown to be exquisitely precise. Annexin 2 localises to the vesicle cup and actin tail of rocketing macropinocytic vesicles induced by hyperosmotic shock (31). Disruption of annexin function with a dominant-negative chimeric annexin 2/S100 A10 fusion construct inhibited rocketing, implying that correct localization of annexin 2 is an absolute requirement for this process. These findings are consistent with observations that annexin 2 is required for the actin-dependent apical transport of a specific subset of exocytotic vesicles in polarized epithelial cells (32), and that annexin 2 is involved in the actin-dependent transport of GLUT4-containing secretory vesicles in adipocytes (33,34).

Annexin 2 and its binding partner S100A10 have been shown to associate with actin-rich pedestals formed in HeLa cells infected with non-invading enteropathogenic Escherichia coli (EPEC). Mutant EPECtir bacteria, which do not trigger actin pedestal formation but induce some actin accumulation, are also capable of recruiting annexin 2 (35), although the absolute requirement for annexin 2 in this process has not been demonstrated. Despite illustrating two quite distinct examples of annexin 2 involvement in the organization of actin at membrane contact sites, macropinosome formation and EPEC attachment both induce the synthesis of phosphatidylinositol-4,5-bisphosphate (PtdIns4,5P2), and annexin 2 has recently been shown to bind PtdIns4,5P2 with high specificity and affinity (15,16). In this respect, annexin 2 has similarities with other known actin modulating proteins such as profilin, which also bind PtdIns4,5P2.

The localization of annexins to actin-rich membrane subdomains is also a characteristic of phagocytosis. Annexins 1–5 have all been found associated with phagosomes isolated from J774 macrophages, and phosphorylation of an N-terminal serine has been implicated in recruitment of annexin 1 to the phagosomal protein complex (36,37). The differential recruitment of annexins to Mycobacterium avium-containing phagosomes as compared to those containing inert latex beads may form part of a mechanism by which the bacterium evades destruction by phago-lysosomal fusion. It appears that annexins 1, 4, 7 and 11 are less effectively recruited to mycobacterium-containing phagosomes, whilst annexin 2 levels are conserved. This suggests that annexin 2 has a critical and possibly direct role in phagocytosis, and that other annexins have a secondary (perhaps actin-independent) role, determining intracellular trafficking and maturation of the phagosome. Annexins 1, 3, 4, 7 and 11 have also been found on phagosomes initiated in a macrophage-like cell line as a result of Fc-receptor cross-linking. Stimulation of phagocytosis resulted in re-localization of some of the annexins to both phagosomal and non-phagosomal membranes, annexin 11 being predominantly associated with the phagosomes (38,39).

Annexin 2 has also been shown to be transiently recruited to phagosomes (in association with c-src) in osteoclast-like cells (40). Actin remodelling at the phagosome is complex, involving nucleation at the points at which the receptors are cross-linked, extensive elongation as the phagocytic cup is enlarged, deformation of the plasma membrane resulting in its invagination, cup closure, internalization of the phagosome and subsequent depolymerization of the actin ‘cage’. It is possible that several members of the family of annexins participate in this process, contributing to different aspects of it or performing redundant functions in different cell types or in association with different receptors.

Organization of the Membrane-Cytoskeleton

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References

The ability of annexins to bind both the negatively charged phospholipids of the inner leaflet of the plasma membrane and the cytoskeleton, is consistent with a role for annexins in the organization of the membrane-cytoskeleton. Annexin 2, for example, has been purified from complexes containing Rac1 at the site of cell–cell contacts in MDCK epithelial cells (41). Association of annexin 2 with nascent cell–cell contacts appears to require the products of phosphatidylinositol-3-kinase (PI3-kinase), but mature contacts probably do not have this requirement. Similarly, in mature monolayers of endothelial cells, annexin 2 physically associates with the tyrosine phosphatase SHP-2, the two proteins being concentrated in cholesterol-rich domains required for stabilization of adherens junctions (42,43).

Annexin 2 has also been shown to associate with the large protein AHNAK. This protein is found in both the nucleus and at desmosomes, where it associates with both G- and F-actin. It is also found at the cytosolic surface of the plasma membrane in MDCK cells, where it accumulates following cell–cell contact. Down-regulation of annexin 2 or its binding partner S100A10 inhibits re-localization of AHNAK (44). As AHNAK is a large protein it has the potential to interact simultaneously with numerous proteins and, although the characterization of such interactions will inevitably take time, it is possible that AHNAK-mediated organization of the actin cytoskeleton may be regulated by functional interactions with rho-family kinases. In this context it appears that annexin 2 could play the role of a scaffold protein linking plasma membrane domains with the actin polymerization machinery by recruiting AHNAK to the membrane.

Annexin 6 has been identified as a binding partner of calspectin (brain spectrin or fodrin) and in vitro experiments suggest that it can influence the F-actin bundling activity of this protein in a calcium- and phosphatidylserine-dependent manner (45). Thus it is possible that annexin 6 could regulate microfilament architecture, e.g. in neurites in response to calcium signals. In the endocytic pathway annexin 6 has been reported to be required for the protease-dependent dissociation of the clathrin lattice from the spectrin cytoskeleton during the final stages of coated pit budding (46). However, these observations are at odds with studies showing that endocytosis occurs normally in cells lacking annexin 6 (47), and indeed the absence of any harmful phenotype in annexin 6 null mutant mice argues against any important regulatory role for annexin 6 in endocytosis (48). Calcium-dependent shuttling of annexin 6, annexin 2 and actomyosin into lipid microdomains has been demonstrated in smooth muscle cells (49) and the authors suggest that annexin 2 may be able to translocate from actin stress fibres to the cell periphery in response to changes in cell plasticity.

Conclusions and Future Perspectives

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References

Despite the fact that F actin-binding was one of the first described characteristics of annexins 1 and 2, the significance and physiological relevance of this property have been difficult to establish, not least because these annexins do not generally co-localise with actin filaments in cells. What has recently become clear is that annexin 2 (annexin 1 has been less well studied in this context) is recruited to sites of actin polymerization that are intimately associated with membrane surfaces, and that interfering with annexin 2 function can lead to failure of certain types of actin polymerization. The challenge now is to understand what role annexin 2 performs at a mechanistic level. Given the large number of proteins already known to be involved in the regulation of actin dynamics, including capping proteins, severing proteins, nucleating proteins and branching proteins, it will be particularly interesting to discover the niche filled by annexin 2 in this complex molecular machine.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References

Work in the authors' laboratories is funded by The Wellcome Trust, The Medical Research Council and Fight for Sight as well as the Deutsche Forschungsgemeinschaft and the Interdisciplinary Center for Clinical Research of the University of Muenster Medical School.

References

  1. Top of page
  2. Abstract
  3. Interactions with Actin
  4. Regulation of Actin Dynamics
  5. Organization of the Membrane-Cytoskeleton
  6. Conclusions and Future Perspectives
  7. Acknowledgements
  8. References