Actin-Binding Proteins Part 2 (Molecular Biology)

2. Signal Transduction and Restructuring of the Actin Microfilament System

Cell surface receptors continuously monitor the extracellular milieu and pass signals to the cytoplasm by generating second messengers of various kinds which either stimulate or inhibit the activity in the microfilament system. Addition of growth factors to serum-starved cultured cells causes the immediate outgrowth of membrane lamellae and filopodia, events that depend on the polymerization and organization of actin. About 60 seconds later, there is a generalized increase in motile activity of the cells, including restructuring of the actin-containing stress fiber system and translocation of the cell. Observations of this kind have been made both with epidermal growth factor (EGF) (43) and platelet-derived growth factor (26). In the case of EGF, a significant fraction of the receptors was found in direct association with microfilaments, and a binding site for actin on the receptor has been identified (44, 45).

The phosphatidylinositol-cycle (PI cycle) is coupled to activation of the microfilament system (46). Ligand-induced, receptor-mediated activation of kinases generate phosphatidyl inositol 4,5-bisphosphate (PtdIns 4,5-P2) and phosphatidyl inositol 3,4,5-trisphosphate (PtdIns 3,4,5-P3) (47, 48), both of which affect the activity of actin-binding proteins. Profilin, gelsolin, a-actinin, vinculin, ERM family of proteins, and myosin all bind PtdIns 4,5-P2, and/or PtdIns 4,5-P3. Stimulation of platelets with low concentrations of thrombin causes a transient increase in polyphosphoinositides during the first 10 sec (49). This is accompanied by an equally transient transformation of unpolymerized actin into filaments. This initial polymerization is over in about 10-20 sec (49). The immediate precursor for this polymerization appears to be profilin:actin, since increased amounts of free profilin appears in cell extracts after stimulation. Within 2 min, these filaments become organized into supramolecular structures by the action of crosslinking proteins stabilizing the protrusions seen on the platelet surface (50-52). Simultaneous activation of phospholipase Cg-1 results in the hydrolysis of PtdIns 4,5-P2, releasing the second messenger inositol trisphosphate, which causes release oftmpFF-328_thumb[2]ions from intracellular Ca stores (53). Calcium ions are involved in activation of the actomyosin interaction and thereby force generation.


In addition to connections to the phosphatidyl inositol signal transduction cycle, the microfilament system is linked to the cAMP-dependent signalling system. Agonists causing increasing levels of cAMP often inhibit actin polymerization (54). A connection between adenylate kinase pathways and the actin monomer-binding protein profilin was suggested by genetic studies showing that increased levels of profilin could compensate for the deletion of the C-terminal domain CAP of the cyclase (55, 56). In platelets, the actin-binding protein, vasodilator-stimulated protein VASP is phosphorylated by cAMP- and cGMP-dependent kinases (57), which then correlates with the inhibition of microfilament reorganization.

The proteins involved in controlling the organization and function of the microfilament system are part of multiprotein signal transduction complexes. Several of these proteins have been identified as products of protooncogenes, ie proteins which in mutated form may cause cancer (58, 59). In these signal transduction complexes the proteins are linked together by adaptor modules in linker proteins such as SH2, SH3, and WW domains which recognize specific features of short polypeptide stretches, especially phosphotyrosine-containing and polyproline sequences (60). An example is the oncogene product Vav that has several such binding motifs suggesting how it links signal transduction at the cell surface to restructure the cortical cytoplasm (61). Vav possesses many of the homology domains commonly found in signal transduction complexes, comprising a phosphotyrosine-binding SH2 domain, two proline-rich peptide recognizing SH3 domains, a phosphatidyl inositol 4,5-bisphosphate recognizing PH domain, a guanine nucleotide exchange factor DH-domain, two cysteine-rich nuclear transcriptionally important LIM domains, and an actin-binding calponin homology domain. The clearest demonstration that these functional linkages actually lead to nuclear transcription is that the production of Interleukin-2 (IL2) requires Vav-dependent actin cap formation in activated T-cells (62, 63).

Outgrowth of filopodia and membrane lamellae, and the formation of stress fibers, which are manifestations of the activation of the microfilament system by different kinds of ligand:receptor interactions, all depend on the activation of members of the Rho subfamily of small GTPases for their control. The small GTPase Cdc42 appears to control formation of filopodia, Rac controls membrane lamellae and Rho stress fibers (64). There is only little information so far as to how the receptor-mediated transmembrane signal is relayed to the GTPases and what their downstream targets are. It has been shown with permeabilized platelets that ligation of the thrombin receptor, and activated Rac, uncap actin filament (+)-ends through phosphoinositide synthesis (65). Some of the microfilament-controlling, signal transduction elements mentioned below interact with small GTPases. Interestingly, moesin (see below), an actin-binding protein linking actin filaments to transmembrane receptors, is required for Rho and Rac effects on the actin filament system (66).

VASP belongs to a growing collection of related proteins that are involved in stimulation of assembly of actin filaments in vivo (67-69). In moving cells, it is found in highly dynamic membrane regions and in integrin-rich adhesion plaques. Molecular cloning of VASP has established that it has both a central proline-rich core, which interacts directly with profilin (68, 70), and an actin-binding site. In cells, the primary site of growth of actin filaments is thought to be at advancing cell edges, and profilin:actin heterodimers are believed to be guided into this site by VASP (71). High concentrations of VASP appear to be targetted to focal adhesion sites by zyxin (72). Zyxin also binds a-actinin and Vav suggesting that orderly assembly of crosslinked actin networks can proceed at these locales. In addition to binding profilin and zyxin, VASP binds to the SH3 domains of the src-family of protein kinases (69), thus integrating signal transduction via phosphorylation with the ordered assembly of cortical structures in locally specialized sites (63). VASP and vinculin complexes can be immunoprecipitated out of cell extracts with either antivinculin or antiVASP antibodies, and recent experiments have demonstrated that the interaction between the two proteins is enhanced by PtdIns 4,5-bisphosphate binding to vinculin (73).

VASP has come to prominence owing to its relevance to the understanding of the movement of the bacterium Listeria through the host cell cytoplasm (74). Listeria expresses a proline-rich protein on its surface called ActA (zyxin-like), which binds to the modular proline-binding SH3-domain of VASP. Apparently, VASP can attract profilin:actin complexes to sites of polymerization on the surface of the bacterium, generating a propulsive force. There are many examples of pathogenic microorganisms that have acquired host cell genes that code for proteins that modulate the behavior of the microfilament system for the purpose of invasion, locomotion in the cytoplasm, and cell-to-cell spreading.

WASP is a multidomain, actin-binding protein, which is defective in patients afflicted with Wiskott-Aldrich syndrome (75, 76), a disease characterized by severe thromifocytopenia and immunodeficiency. WASP is specific to hematopoietic cells, which in case of WAS have a paucity of surface filopodia and other shape abnormalities (77). The small GTPases Cdc42 and Rac bind directly to WASP (78-80). WASP is also a binding partner for Src family of protein tyrosine kinases (81).

A protein homologous to WASP, N-WASP, is more widely distributed. N-WASP is concentrated at nerve terminal regions in the post synaptic density, and coexpression of Cdc42 and N-WASP induces extremely long actin-based filopodia in cells (82, 83). Its involvement in the formation of membrane lamellae and filopodia may be controlled by phosphoinositides in the plasma membrane. It contains a pleckstrin homology PH domain and a cofilin homology domain. Furthermore, N-WASP has been shown to be essential for the actin polymerization-dependent movement of the bacterium Shigella flexneri (84).

WIP is a recently discovered WASP interacting protein that is important for cortical actin assembly (85). WIP contains both profilin- and actin-binding regions, which like in VASP, are near each other in the protein sequence suggesting that they constitute the binding site for a profilin:actin heterodimer to be added to a growing actin filament. The small signal transduction GTPase Cdc42 binds to WASP (78, 79), but not to WIP. Thus, WIP function may be regulated by Cdc42 via WASP. Stimuli that activate Cdc42 may target the WASP-WIP complex to the actin filament system via interactions between the WH1 domain of WASP and the proline-rich ABM1 motifs of structural proteins such as zyxin and vinculin. The presence of SH3 binding motifs in both WIP and WASP, and the capacity to bind to the adaptor protein Nck (86), suggest that the WASP-WIP complex couples additional signaling pathways to the actin filament system.

Formins are profilin and actin-binding proteins from S. cerevisiae. They direct the determination of cell polarization and cytokinesis (87 for refs.). In response to a-mating factor, the formin Bni1p forms complexes with the active form of Cdc42, actin, profilin, and actin associated protein Aip3p (88-90). These proteins localize to the tips of mating projections suggesting that the formin is a target for Cdc42, linking the pheromone response pathway to activation of the actin system.

The formin family of proteins is also important in limb and kidney development in vertebrates (91). Mutations in formins lead to defects in the contractile ring formation during cytokinesis, causing the appearance of multinucleated cells. Formins are localized both in the cytoplasm and the nucleus of the cells. FH-proteins, defined by the presence of ‘formin homology’ regions, are important for a number of actin-dependent processes, including polarized cell growth and cytokinesis. They are large, probably multi-domain, proteins and their function may in part be mediated by an interaction with profilin.

IQGAP is yet another actin-binding, multidomain, regulatory protein (92). It was originally described as a 190-kD protein with extensive sequence similarity to the catalytic domain of RasGaps (91). These proteins control the activity of small regulatory GTPases by stimulating their GTPase activity. Subsequently an IQGAP2 was described as a liver-specific protein with a 62% homology to IQGAP1 (93). However, so far no GAP activity has been found in relation to the GTPases examined. Both IQGAPs have several copies of a ca 50 amino acid long internal repeat (IR), a single WW (SH3-like) domain presumably binding to proline-rich sequences, a number of calmodulin-binding IQ motifs, and a putative actin-interacting calponin homology CH domain. In transient transfection of COS-7 cells with epitope-tagged Cdc42, it was demonstrated that IQGAP is present in the advancing lamellae of motile cells, and a major in vivo target of activated Cdc42 (94). The IQGAP1 binds to filamentous actin in vitro via its N-terminal domain and cross-links actin filaments in a Cdc42-dependent manner (95). Recently IQGAP1 was implicated in the regulation of E-cadherin-mediated cell-cell adhesion (96).

In fission yeast, Schizosaccharomycespombe, the rng2 gene codes for an IQGAP-related protein, rng2p. The rng2p is located in the actomyosin ring and the spindle pole body and is required for the assembly of the actomyosin ring at cytokinesis (97).

Paxillin is not known as an actin-binding protein, but it is involved in the control of the microfilament system at the adhesion contacts. It was detected as a 68-kD phosphotyrosine-containing protein in RSV-transformed cells, and was later purified from smooth muscle cells (98, 99). Many features of paxillin suggest that it functions in signalling events at the adhesive membrane. It is concentrated at focal adhesions, and it is tyrosine phosphorylated in response to extracellular matrix binding and cell transformation. It binds in vitro to the elongated tail domain of vinculin, and thereby links vinculin to the SH3 domain of the pp60csrc, a component of focal contacts. Integrins, which are the principal transmembrane proteins at these sites, do not seem to bind directly to paxillin, and paxillin does not seem to bind either a-actinin or actin, but there is evidence for direct association of paxillin to growth factor receptors. Adhesion of cells to an extracellular matrix like fibronectin or laminin via transmembrane integrins results in increased tyrosine phosphorylation of paxillin, and a similar increase in phosphorylation takes place in connection with clustering of integrins in the plane of the membrane with antibodies. This phosphorylation appears to be caused by activation of the focal adhesion-associated tyrosine kinase, pp125FAK, and tyrosine phosphorylation of paxillin and pp125FAK initiated by cell adhesion is accompanied by the appearance of organized actin-containing stress fibres. Mutagenesis of paxillin has identified one of the paxillin LIM domains as being responsible for targetting this protein to focal contacts (100)

Zyxin is a low-abundance phosphoprotein localized at sites of cell-substratum adhesion in fibroblasts (72). It has an architecture of an intracellular signal transducer with a proline-rich domain, a nuclear export signal, and three copies of the LIM motif. LIM is a double zinc finger domain found in many proteins that play central roles in regulation of cell differentiation. Zyxin interacts with a-actinin, members of the cysteine rich protein CRP family, proteins that display Src homology 3 (SH3 domains) and Vasp/Ena family members, which in turn bind to both actin and profilin. Zyxin and its partners have been implicated in the spatial control of actin filament assembly as well as in pathways important for cell differentiation. Based on its repertoire of binding partners and its behaviour, zyxin is thought to serve an organizing centre for the assembly of multimeric protein machines that function both at sites of cell adhesion and in the nucleus.

In summary, the major signal transduction pathways and mechanisms of activation have been shown to be linked to the changes in the dynamic restructuring of the actin cortex: the phosphatidyl inositol cycle and the cAMP second messenger systems and the small GTPase coupled signal transduction mechanisms. All of this points to the growing realization that cytoplasmic signal transduction and transcriptional events in the nucleus are so interwoven that discussions of sequential control by one system or the other are oversimplifications. The idea that actin cortical movements are "downstream" events of "upstream" effectors is similarly unjustified, since actively motile cells are required for the generation of the signal itself.

There are perplexing aspects of polymerization-based propulsion of bacterial pathogens, and by implication filopodial extension. WASP and N-WASP have actin-binding sequences. It is not clear what the role of these proteins is in the regulation of actin polymerization at focal growth sites. It is known that the polar actin filaments are oriented such that their fast growing ends are "pushing" against the moving bacterial wall, or against the membranes of the advancing edges of lamellipodia and filopodia. Actin-binding proteins such as WASP are anchored in these nascent growth complexes (to ActA in the case of Listeria). The problem is to explain what happens when an incoming actin monomer is presented as a profilin:actin precursor to the end of the growing filament, since it would appear to be blocked by WASP. Another problem is to account for the free energy of ATP hydrolysis that is made available by the actin polymerization reaction.

3. Further Links of Actin Filaments to the Extracellular Matrix

Cortactin is the name of two related proteins p80 and p85, which cross-link actin filaments close to the plasma membrane. Cortactin was first recognized as a substrate for tyrosine-phoshorylation in pp60src-transformed cells, but its presence has also been demonstrated in a variety of nontransformed cells (101). Cortactins have internal repeats of 37 amino acid residues in the N-terminal end of the molecule, which are necessary for their binding to actin filaments. In the C-terminus there is an SH3-domain, and preceding that in the sequence there are proline and serine/threonine rich regions. These latter parts of the molecule are not needed for actin-binding, and their functions remain to be elucidated.

In non-transformed cells, cortactin is normally phosphorylated on serine and threonine, but becomes transiently phosphorylated on tyrosine in response to growth factor stimulation, and on transformation by activated cSrc. In platelets, cortactin becomes transiently tyrosine-phosphorylated in reponse to stimulation with thrombin. However, tyrosine phosphorylation of cortactin does not appear to affect its ability to bind to actin filaments, although it does influence its localization to the cortical actin filament system.

Stimulation of cultured cells with various growth factors and platelets with thrombin results in translocation of cortactin into the actin filament system of advancing lamellae and filopodia (102, 103). Activation of fibroblast growth factor receptor-1 results in an association of the activated receptor with cSrc, and a correlated association of cortactin with SH2 domain of cSrc. The redistribution from the cytoplasm into advancing lamellae (ruffles and lamellipodia) appears to depend on the Rac-1-induced activation of the serine/threonine kinase PAK1, a downstream effector of Rac1 and Cdc42 (104). These GTPases are involved in the control of kinases in the phosphatidyl inositol signalling pathway, and there is evidence that PtdIns 4,5-bisphosphate could be involved in the association of cortactin with the activated microfilament system (105).

Eps8 is a putative actin-binding protein which resembles cortactin in its distribution in the cell (106). It has an unusual Src homology-3 domain in the carboxy terminus, formed by an intertwined dimer of Esp8 molecules which may affect its peptide specificity. Eps8 is mostly localized in the perinuclear region. However, on stimulation of cells with growth factors, it accumulates in the peripheral cell extensions, where it is found colocalized with cortactin and filamentous actin. The Eps8 pool associated with newly formed lamellipodia and membrane lamellae appears to be mostly detergent-insoluble, as is the tyrosine phosphorylated population of Eps8 molecules generated by activation of vSrc kinase, suggesting Eps8 recruitment to specific sites during cell remodeling.

Ezrin, radixin, moesin (ERM) belong to the band 4.1 superfamily on the basis of their homology to the erythrocyte band 4.1, a protein that connects the actin/spectrin network to the erythrocyte membrane protein glycophorin C (31, 107, 108). Like cortactin, the ERM proteins appear to play structural and regulatory roles in stabilizing specialized organizations of actin filaments in connection with the plasma membrane both during development and in adult tissues. They are generally present in microvilli, filopodia, and membrane ruffles, sometimes together, but more often differentially distributed between different types of cells. They link transmembrane proteins with the cortical actin filaments, a process governed by signal transduction involving the Rho-GTPase (109). Ezrin has been shown to bind specifically to nonmuscle b-actin with high affinity (110).

The N-terminal domain of ezrin associates with plasma membrane components, and the C-terminal half connects these structures with the submembranous actin filaments. Purified ezrin exists in an inactive conformation that requires activation to expose sites that allow it to associate with the membrane components and with actin filament (107). It has been shown that binding of ezrin to the transmembrane hyaluronan receptor CD44 (111) requires activation of ezrin by PtdIns 4,5-bisphosphate binding to the N-terminal domain of ezrin (112-114). These are not the only transmembrane proteins binding ERM, since ERM proteins have a membrane location also in cells that do not express CD44. It is now known that ERM proteins, in addition to CD44, also bind to intercellular adhesion molecules, ICAM-1 and ICAM-2. A positively charged amino acid cluster of CD44, CD43, and ICAM2 positioned close to the lipid bilayer has been implicated in the binding (115).

A group of phosphorylated, 50-55 kDa, ezrin-binding protein has been purified from human placenta and bovine brain by affinity chromatography on immobilized N-terminal domains of ezrin or moesin (31). Isolated EBP50 has two peptide-binding PDZ domains in the N-teminal half of the molecule with capacity to bind to cytoplasmic tails of transmembrane proteins. The EBP50 can be coprecipitated with ERM and may serve to bind ERM members to one or more integral membrane proteins.

Cells opened up by the use of the detergent digitonin are still amenable to activation with non-hydrolyzable analogues of GTP. Addition of guanosine 5 ‘ -O-(3-thiotriphosphate), GTPgS, to such permeabilized cell models induces assembly of both actin filaments and focal adhesion complexes through activation of endogenous Rho and Rac (66). The sensitivity to GTPgS is lost after a few minutes of incubation, but can be restored by the addition of the actin-binding protein moesin indicating that moesin is required for the effect of Rho and Rac on actin filament system. This is an important step towards the resolution of the signal relay which initiates the reorganization of the microfilament system.

Talin is an elongated, flexible actin-binding protein first recognized in smooth muscle (116, 117) and platelets (118, 119). It is present in focal contacts in tissue cultured cells, in adhesion plaques formed by activated platelets, in myotendinous junctions, and in dense plaques of smooth muscle, but is absent from cell:cell interaction sites (for refs, see 38). In motile cells talin is present in membrane ruffles. It is a major protein in platelets constituting more than 3% of the total protein. In response to activation, it is redistributed to the actin-rich platelet cortex (120). Thus talin is clearly of central importance for the association of actin filaments to membrane components to sites where cells make close contact with extracellular matrix proteins, and it has been shown to interact with the cytoplasmic domains of transmembrane b1-integrin (116, 121).

Talin has a molecular mass of of 270 kD as estimated from the gene sequence (122). It is about 60 nm in length, and appears to be composed of a series of globular domains as judged by electron microscopy (123). In solution, it forms homodimers by antiparallel association of the monomers. It binds to integrin, vinculin, and actin in vitro. There are reports that talin can nucleate, cap and crosslink actin filaments, and that its activities may be controlled by phosphorylation by both serine/threonine and tyrosine kinases (124, 125). An N-terminal domain of talin is homologous to ezrin (122), and may be the part of the molecule which associates with membrane lipids. The larger C-terminal domain contains the binding site for vinculin (see (38) and (125) for further refs.). Disruption of the talin gene in embryonic stem cells leads to a disturbance in the formation of vinculin and paxillin-containing focal contacts and cell spreading (126). The talin (-/-) cells do form embryoid bodies, and notably two morphologically distinct cell types have been observed that were able to spread and form focal adhesion-like structures with vinculin and paxillin on fibronectin. Thus, although talin was essential for the expression of b1-integrin, a subset of differentiated cells managed without it. What substitutes for talin in these cells is unclear.

Vinculin is a 117-kD, actin-binding protein found concentrated at all sites of attachment of filamentous actin to transmembrane components of adherens junctions (38, 127-131). At these sites, it enters into multiple interactions with other proteins at the cytoplasmic face of the contacts. It is critically required as a structural component in the formation of the junctions. On ligand-induced clustering of integrins, vinculin is recruited from the cytoplasm to sites of integrin-ligand interaction, where it is a substrate for tyrosine phosphorylation by pp60src.

There is electron microscopic evidence that the vinculin molecule can attain different conformational states. It is found either as a compact, globular molecule, or it is unfolded into a globular head with an extended tail (132-134). In the compact globular state, its actin-binding capacity is masked by an intramolecular association of the 95-kD head and 30-kD tail domains. The actin-binding site can be exposed by binding of acidic phospholipids to the protein suggesting that the affinity of vinculin for target molecules is regulated by modulation of the head-to-tail interaction, and that this may be used in the control of the assembly of adherens junctions (73, 135). The phospholipid PtdIns 4,5-bisphosphate binds to two descrete regions of the vinculin tail, disrupts the intramolecular head-tail interaction and induces vinculin oligomerization. The binding of PtdIns 4,5-bisphosphate to vinculin also enhances the association of vinculin to VASP the actin- and profilin-binding protein present in focal adhesion sites. The C-terminal tail region contains a binding site for the focal adhesion protein paxillin (136). Furthermore, vinculin has been identified as part of the cadherin-catenin junctional complex (137-139).

Tensin is a dimer of two 200 kD polypeptide chains. It is found in different types of cell contacts, including focal adhesions, zonula adherens, intercalated discs, and myotendinous and neuromuscluar junctions (140-143). It contains three distinct, actin-binding sites per monomer and appears to cap as well as cross-link actin filaments. Tensin also binds vinculin and possibly tyrosine kinase receptors through a Src homology 2 (SH2) domain (144). It has been suggested that a tensin dimer with its 6 actin-binding sites might wrap around the end of an actin filament, and bind it to a tyrosine kinase receptor through its SH2 domain. Induction of the formation of a focal contact, by the binding of extracellular ligands to integrins, causes tensin to be phosphorylated on tyrosine. Insertin, an actin-binding protein suggested to be involved in the actual incorporation of monomers into filaments, has been recognized as a proteolytic breakdown product of tensin (145).

4. Proteins Controlling Polymerization of Actin

DNase I (mw 31 000) was the first actin monomer-binding protein found (12). However it is doubtful whether DNase I plays a role as an actin monomer-binding protein in the cell. Since the enzyme is produced by acinar cells in the pancreas and secreted into the duodenum, it has been thought of as a digestive enzyme. However, intracellular forms of the enzyme have been observed, and there is evidence that it might be involved in programmed cell death, apoptosis. (146).

The crystallization of DNase I was reported in 1948 by Kunitz (147). The same year Laskowsky and collaborators described the presence of a proteinaceous inhibitor in tissues from vertebrates (148, 149). Two DNase I inhibitors (I and II) were isolated and crystallized in 1966 (13, 150). They were shown to form gelifying high molecular weight aggregates. Searching for the identity of the inhibitors eventually led to the finding that actin was the ubiquitous inhibitor of DNase I (12). Analysis of the crystals of the DNase I inhibitor II revealed that they contained actin together with an equimolar amount of a small protein, now known as profilin (14, 15). Later, the inhibitor I and II were identified as profilin in complex with g-actin and b-actin, respectively (151).

DNase I is a glycoprotein. It forms a stable 1:1 complex with monomeric actin, and when mixed with equimolar amounts filamentous actin, it causes depolymerization of the filaments and formation of the 1:1 complex with actin (152-154). Kabsch and coworkers provided the first insights into the structure of actin by solving the structure of the DNase I:actin complex (155; see special entry on actin). The high resolution structure of DNase I alone and complexed to an oligonucleotide has been reported as well (156, 157). The current model of the actin filament was arrived at using the structure of the actin monomer, as it occurs in the DNase I:actin crystal, together with diffraction data obtained with oriented gels of F-actin (158). This model is claimed to fit well into reconstructions of actin filaments from electron microscopic images (159, 160), and reconstructions of actin filaments with bound actin-binding proteins are interpreted in terms of this model (see below). However, doubts about this model being a relevant representation of the actin filament have been expressed, and an alternative way of deriving a model of F-actin has been suggested from the analysis of the organization of actin monomers in crystals of profilin and actin (161, 162). To ascertain the validity of the current model, independent information about the orientation of the actin monomer in the actin filament needs to be acquired.

Profilin (mw 12 000-15 000) is an abundant actin monomer-binding protein in eukaryotic cells (163, review). (See separate article Profilin.) Particularly high concentrations of profilin are found in lymphoid cells and brain cells. In lymphoid tissues the concentration of unpolymerized actin is about 100 |iM and in platelets it may be as a high as 200 |iM. In both cases about 50% of the unpolymerized actin can be accounted for by profilin:b-actin and profilin:g-actin isoforms. The remaining unpolymerized actin appears to be sequestered by b-thymosins (see below). For references on specific aspects of profilin, see (164-189).

Observations point at profilin being an important factor in the control of the release of inositol trisphosphate and thus of the generation of Ca pulses in the cell. The primary targets for Ca regulation in cells is the actomyosin system with global changes in structure and activity as the result. It is now known that many of the microfilament-associated proteins interact with components formed in the phosphatidylinositol-cycle as a result of receptor-mediated activation of the cell, and that these interactions modulate the activity of these proteinstmpFF-329_thumb[2]actin. Although the exact physiological roles of these interactions remain to be elucidated, it suggests that the phosphatidylinositol-cycle is directly involved in controlling the microfilament-based motility cycle (see Fig. 3).

Beta thymosins constitute a conserved family of polypeptides (mw 5 000), the members of which were originally isolated from mammalian thymus and thought to be important for the immunoregulatory activities of this organ (190). Homologues of Tb4 have been found also in distantly related organisms (191). The thymosin b4 isoform is abundant in mammalian tissues, and has attracted much attention after the discovery of its association with unpolymerized actin in platelet extracts. It has a poorly defined structure as determined by NMR in solution (192). Binding studies suggested that it binds to the (+)-end of the actin monomer in competition with profilin. However, crosslinking studies imply a more complex interaction involving sites at both ends of the actin molecule (193).

Binding of thymosin b4 to actin monomers inhibits incorporation of actin monomers at both ends of the actin filament. Thymosin b4 has a rather low affinity for actintmpFF-330_thumb[2], but since it is present in high concentrations in cells (in platelets ca 500 |iM) it is considered to be the most important actin sequestering protein. In the presence of polymerization-inhibiting concentrations of thymosine b4, addition of profilin overcomes the thymosin effect, due to the higher affinity of profilin for actin tmpFF-331_thumb[2], and the fact that profilin:actin can add actin monomers onto the fast growing ends of actin filaments (see Profilin).

In vitro, thymosin can be shown to bind also to actin filaments and decrease the critical concentration for actin polymerization (194), and experiments on cultured cells indicate that b-thymosins may not be just actin sequestering factors (195). In the NIH3T3 cell line thymosin b10 is the predominant beta thymosin isoform, being present in about equimolar concentration to proflin and ADF/cofilin. Overexpression of thymosin b10 in these cells led to an increase in the cellular content of polymerized actin without any change in the total actin content of the cells. An intimate relationship with the regulation of the dynamic functioning of the microfilament system is also indicated by the increase in motility caused by overexpressing thymosin b10. Consonant with this an isoform of beta thymosine, thymosin b15, is upregulated in some forms of malignant cells (196).

The ADF/cofilin family of proteins include actin depolymerizing factor, cofilin, destrin, depactin, actophorin (197-199). These proteins are found in all kinds of eukaryotes, and when tested genetically they have proved to be essential. In vitro, they can form 1:1 complexes with actin monomers as well as with actin in filamentous form. The structure of the prototype ADF/cofilin, actophorin from Acanthamoeba have been solved by crystallography (200) and there are NMR structures of yeast cofilin and destrin (201, 202). It has recently been recognized that an ADF homology domain is present in each member of a newly identified protein family consisting of the ADF/cofilins, the twinfilins, and the drebrin/Abp1s (203).

In the cell, most of the ADF/cofilin is found in the perinuclear area, but there are also significant amounts in highly motile advancing lamella and filopodia, where the reorganization of actin through polymerization/depolymerization is most intensive. The interaction between ADF/cofilin and actin is sensitive to the pH of the medium, and as in the case of profilin, polyphosphoinositides interfer with the ADF:actin interaction. Physiologically, however, the most important mechanism controlling the activity of ADF/cofilin depends on phosphorylation/dephosphorylation of the protein. Phosphorylation of ADF/cofilin at a site near the N-terminus (S3) blocks its binding to actin. In cell extracts about 60% of the protein is phosphorylated, and in connection to receptor-mediated stimulation of cells dephosphorylation of ADF/cofilin takes place. The phosphatase involved in this reaction has not been identified, but a specific protein kinase, LIM kinase, has been found responsible for the in vivo phosphorylation of ADF/cofilin (204, 205).

The turnover of actin filaments inside living cells appears to be up to 100-fold faster than in in vitro experiments with actin alone. It has now been demonstrated in in vitro as well as in in vivo experiments that ADF/cofilin can accelerate the turnover of actin filaments (17, 199, 206-208). ADF/cofilin binds with higher affinity to ADP-containing actin, monomeric as well as filamentous, and facilitation of filament growth by active ADF/cofilin is likely to be brought about by increased depolymerisation of the actin filaments at their proximal ADP-containing ends increasing treadmilling activity. Nucleotide exchange, possibly enhanced by profilin, would in this model precede the reallocation of the ATP-containing precursor to the fast growing, distal (+)-end of the population of growing filaments. The presence of ADF in lamellipodia, where filaments are long, argues against a severing function in vivo, and the end-specific effects of ADF on filament assembly seen in vitro also indicate that the severing activity of ADF/cofilin is less important in vivo.

Twinfilin is a newly discovered protein composed of two cofilin-like regions (209). It was identified and characterized as an actin monomer-binding protein in budding yeast. Genes coding for homologous proteins have been recognized also in Caenorhabditis elegans, humans and mice. The two halves of the molecule twinfilin I and II resemble the corresponding domain from other species more than they resemble each other. Thus, twinfilins have evolved from a common ancestor and the twinfilins represent a single protein family (199). Twinfilin does not appear to act as an actin filament depolymerizing factor, but there are in vivo observations that suggest its involvement in the control of the dynamics of the microfilament system.

Next post:

Previous post: