Actin-Binding Proteins Part 3 (Molecular Biology)

4.1. Arp

The first reports describing a family of actin-related proteins, the Arps, appeared a few years ago (2, 209). These are highly conserved proteins that share a 30-60% homology with actin, but are functionally distinct from actin. Two of these actin-related proteins, Arp2 (44 kD) and Arp 3 (47 kD) have attracted a great deal of attenion, since they were discovered to exist in a large complex together with five other proteins (40 kD, 35 kD, 19 kD, 18 kD, and 15 kD) that could be isolated by affinity chromatography on immobilized profilin (210). The Arp2/Arp3 complex has been reported to nucleate actin filament formation, to bind along the sides of actin filaments and to express filament crosslinking activity. A similar complex was identified in searching for factors that initiate actin assembly at the surface of Listeria (211). The Arp2/Arp3 complex is highly conserved among eukaryotes. Null mutations are lethal. It is localized in the cortex of amoebae and yeast and in the lamellipodia of higher eukaryotes (212, 213). In yeast, the Arp2/Arp3 complex is required for the integrity and motility of actin patches and for endocytosis (214). Thus the complex appears to play a major role in actin-based motility.

It has now been demonstrated that the Arp2/Arp3 complex isolated from Acanthamoeba binds profilin (215) and to the (-)-ends of actin filaments (168). It is not an efficient nucleator of actin polymerization. Capped filaments grow by the addition of monomers to the (+)-end. The complex can be seen by electron microscopy to attach the (-)-end of one filament to the side of another filament. This suggests that the Arp2/Arp3 complex might control the assembly of a branching network of actin filaments in the advancing lamellae of motile cells. Similar branching of actin filaments have been recognized earlier in electron microscopic pictures of the weave of microfilaments in advancing lamellae in moving keratinocytes (36). The cellular concentrations of the Arp2/Arp3 complex has been estimated to be high enough to cap the (-)-ends of all filaments in a cell. These observations imply that models assuming that actin filaments in vivo undergo an ADF/cofilin-aided depolymerization from the (-)-end during motile activity may be too simplistic, since there must also be mechanisms for the controlled uncapping of the (-)-ends of the filaments.

Further studies of the Listeria system have shown that the bacterial protein ActA, which is essential for the actin polymerization-dependent propulsion of the bacterium through the cytoplasm, acts synergistically with the Arp2/Arp3 complex in vitro. Working together, they eliminate the lag phase in the polymerization of actin and increase the initial rate of assembly 50-fold (216). This activity has been shown to reside in the N-terminal domain of the ActA protein. A cellular protein that corresponds to ActA has not yet been identified, but zyxin has been pointed out as a plausible candidate.

Capping protein Z (CapZ) is a heterodimeric protein, which was first detected in skeletal muscle sarcomers, where it is associated with actin filaments at the Z-line end of sarcomers, i.e. at the (+)-end of the filament (217). In vitro, it binds with high affinityto the (+)-end of actin filaments. A homologous protein that exists in non-muscle cells, capping protein b2 (CPb2), is enriched at the edge of the advancing lamellum of spreading platelets. By binding to the (+)-end of actin filaments, it blocks the addition of actin monomers to that end, and since actin filaments in vivo appear to grow by assembly primarily at that end, these capping proteins may be important regulators of actin polymerization. The fact that CPb2 is necessary for the proper organization of actin bundles in the morphogenesis of bristles in Drosophila, the view is strengthened that CPb2, gelsolin in some cases, and Cap G may play active roles in actin-based motility in response to signaling. There is evidence also in this case that polyphosphoinositides are involved in the regulation of the protein.

4.2. Adducin

In erythrocytes there is, in addition to CapZ, a second actin filament (+)-end capping protein, adducin (218, 225). This protein is associated with the short erythrocyte actin filaments (one a,b heterodimer per filament). Adducin has been shown to bind to spectrin:actin complexes in the cell cortex, and promote the binding between these two proteins. It also appears to be able to bind directly to actin filaments and to bundle them. Adducin has a relatively low affinity for the filament (+)-end, but even so it appears to be the capping protein bound to the erythrocyte actin filaments, and not CapZ. Its binding to actin filaments is downregulated by calcium and calmodulin, an effect possibly modulated by reversible phosphorylation of the protein. The complex relationship between adducin and the spectrin:actin network indicates that adducin represents a new type of regulated actin filament-binding and barbed end capping protein. The entire molecule is required for capping activity, and the association of adducin with actin filaments seems to be regulated by Rho-associated kinase and myosin phosphatase.

Gelsolin is an abundant and ubiquitously expressed actin-binding protein, which confers calcium sensitivity to the regulation of the dynamics of the microfilament system (219). (See separate article Gelsolin.) For references on specific aspects of gelsolin, see (219-223).

Tropomodulin is a 43 kD protein that caps the (-)-ends of actin filaments in erythrocyte cortical network and in striated muscle sarcomeres (225). The interaction is particularly tight when tropomyosin is bound simultaneously. It is thought that tropomodulin is important in controlling the length of the actin filaments in these situations.

5. Architectural Elements of the Actin Cortex

The dynamic changes in cell surface morphology and motile behaviour, seen in response to growth hormones and other cell stimulating agents, reflect changes in the formation, organization, and activity of the microfilament system. Actin-binding proteins that cross-link actin filaments into tightly packed bundles and more loosely organized networks are necessary for the formation of the weave of microfilaments seen in membrane lamellae and filopodia and in the rest of the submembraneous zone around the cell. Actin-based surface projections on cells take many forms, such as the elaborate stereocilia of cochlear hair cells, microvilli of intestinal epithelium, and the membrane ruffles and filopodia on moving fibroblasts. The formation of these structures depends on the activity of actin-crosslinking proteins. Important members of this class of proteins belong to the spectrin superfamily which are reviewed in ref. 237.

Villin is closely related to gelsolin, an actin severing protein (described above), but distinguished from gelsolin by the presence of an additional domain comprised of 76-amino acid residues, called the "headpiece" (224-229). The headpiece contains an additional actin-binding site, which confers actin filament bundling activity to the protein. Villin is found in the microvilli of certain types of epithelia. Crosslinking of actin filaments occurs at low Ca concentrations. Like gelsolin, villin 2+ severs actin filaments in the presence of micromolar concentrations of Ca , and after breakage of the actin filament, villin remains bound to the (+)-end of one of the fragments. The actin-binding properties, tissue distribution and expression during cell differentiation suggest that villin is important in organizing the actin filament core of microvilli in epithelial cells forming brush borders. In support of this, villin expressed transiently in transfected fibroblasts, results in loss of stress fibers and appearance of large numbers of microvillar protrusions on the dorsal surface of the cells. Microinjection of high concentrations of villin into cells that normally lack this protein, lose their stress fibres, and incorporate the villin into cell surface microspikes and large microvilli, an activity which was shown to depend on the presence of the C-terminal head piece.

The N-terminal domain of villin (14T) consists of 126 amino acid residues. It binds calcium and actin. The structure of this domain has been determined by NMR (230). The 76-amino acid residue long, C-terminal domain of villin is an actin-binding module that is also used in another actin-bundling protein dematin (band 4.9), but whose core domain is unrelated to villin. The 3-dimensional structure of the stably-folded 35 residue long subdomain of the villin headpiece has been determined by NMR (231). It is the smallest autonomously folding protein unit found.

Fimbrin belongs to the plastin family of actin-binding proteins (31, 232, 237). Members of this family are present in all types of eukaryotic cells. Fimbrin is the smallest of the actin crosslinking proteins. It consists merely of an N-terminal EF-domain and two actin-binding domains, each composed of two CH domains. Fimbrin is present together with villin in the microvilli of the intestinal brush border, and in the membrane lamellae and microspikes of most other eukaryotic cell types. A fimbrin-binding site on actin has been suggested from analysis of actin mutations suppressing fimbrin mutations (233). The structure of the N-terminal actin-crosslinking domain of fimbrin has been solved by crystallography (234), and electron cryomicroscopic images of fimbrin-decorated actin filaments has been used to localize the fimbrin-binding site on the surface of the actin filament (235).

In a suggested scenario for the formation of a microvillus (236), the first phase consists of the "streaming" of actin filaments from electron-dense plaques on the apical plasma membrane. These filaments then gathered into loose bundles that extend into rudimentary microvilli. Villin may contribute to an initial bundling of the filaments. Ezrin (see section 3), a protein controlled by tyrosine phosphorylation, links the microvillar actin filaments to the membrane components. During a second phase, the microvilli elongate and become highly organized into a full core of hexagonally-packed actin filaments. This phase correlates in time with the localization of fimbrin to the core of the microvillus. In this way a stable surface protrusion is thought to take form. Similar schemes may apply to the development of surface protrusions on the surface of other types of cells.

ABP-280, filamin, and ABP-120, are all rod-like, dimeric, actin filament crosslinking proteins, where each monomer has an N-terminal actin-binding CH-domain and a long repeat region consisting of tandemly repeated, 100 amino acid residue long, homologous segments (237). The length of the repeat region is different in the different ABPs. The ABP120 was found in Dictyostelium discoideum. The vertebrate variants, ABP-280 and filamin, are major cellular proteins, and products of different genes. Filamin is located specifically in the Z-lines structure of the sarcomeres in muscles. ABP-280 is a major cellular protein in non-muscle cells, where it is present in the cell cortex. The function of ABP-280 appears to be to cross-link actin filaments into orthogonal networks. The 3D-structure of the repeat region of ABP-120 (gelation factor, 6 repeats) has been determined by NMR spectroscopy, which shows it to have an immunoglobulin-like fold (226). Both filamin and ABP-280 (24 repeats) share conserved residues that form the core of the gelation factor repetitive segment structure as determined by NMR. This distinguishes these actin crosslinking proteins from the spectrins and a-actinins, which have tandem repeats of an a-helical domain.

On the basis of their actin filament crosslinking activity, depending on the presence of a calponin homology CH domain, and their elongated shape, as seen by electron microscopy, these proteins have been grouped together with the spectrin and fimbrin families into a superfamily. It is true that the actin binding domains of all these proteins are homologous. However, the presence or absence of repeat domains, and the widely different structures of the repeat domains of the spectrin and filamin families, rather suggests that the three families should be considered as separate protein families employing a common actin-binding domain, since it is possible that the different families will turn out to have quite different functions in relation to the actin filament system.

Spectrin (fodrin) was first isolated as part of the matrix proteins of the erythrocyte plasma membrane (237-240). Underneath the lipid bilayer of these cells, and linked to transmembrane proteins glycophorin and the ion channel band 3, there is a regular organization of units consisting of five or six spectrin molecules attached to a short actin filament. The spectrin molecules are attached to spectrins of adjacent units to form a sheet of five- and six-sided polygons. The membrane skeletons, and isolated junctional complexes, reproducibly contain four proteins in the molar ratios of 1 spectrin dimer:2-3 actin monomers:1 band 4.1 protein:0.1-0.5 protein 4.9 (protein designations refer to mobility on gel electrophoresis). In addition tropomyosin, tropomyosin-binding protein and adducin are thought to be part of the junctional complex.

Spectrin consists of two subunits, an a and a b subunit, which form heterodimers, which in turn associate to form tetramers in the membrane lattice. The b-chain has an N-terminal actin binding CH domain and 12 a-helical repeat domains, whose structure has been determined by NMR spectroscopy (241). The a-chain associates in an antiparallel fashion with the b-chain. It has a C-terminal EF-hand domain, which is close to the actin-binding site of the b-chain followed towards the N-terminus by a binding site for protein 4.1, and a Src-like sequence in the middle of the somewhat longer repeat region. Between repeat 11 and 12, there is a calmodulin-binding sequence inserted. The structures of the spectrin repeat, the phosphoinositide-binding pleckstrin (PH) domain, the SH3 domain, and the CH domain have all been determined (241-246).

The spectrin-actin based network of proteins endows the erythrocyte with an exquisite elasticity which makes it possible for it to go through quite drastic shape changes. However, it is clear that the structural role played by the spectrin:actin system is not its sole function. The close relationship between the spectrin-actin system and transmembrane ion channels implies that the range of controlling functions is wider. The spectrin-actin system is not unique to the erythrocytes, but exists also in other tissues and in cells of as widely separated organisms as Drosophila, Acanthamoeba, Dictyostelium, echinoderms, and possibly also in higher plants.

Alfa-actinin is a classical skeletal muscle protein identified a long time ago (237, 247). It is highly enriched at the Z-disc-end of the actin filaments in the sarcomeres of the myofibrils. It is now known also to be an integral part of the microfilament system in nonmuscle cells, where an interaction between transmembrane b1 integrins and a-actinin has been demonstrated (see Fig. 2). There are several isoforms of a-actinin, which are products of three different genes. In vitro, it crosslinks actin filaments into regular ladder-like structures. It is therefore thought to contribute to the organization of actin in cells. Alfa-actinin is an elongated homodimer (200-215 kD) with antiparallel arrangement of the two subunits. Each subunit consists of a CH-domain, 4 a-helical domains ("spectrin repeats"), and an EF-hand domain. The binding of nonmuscle a-actinin to actin filaments is modulated by binding to the EF-domain, whereas the muscle isoform of a-actinin is insensitive toThe 3D-structure of the CH-domain is known by analogy to the structure of the corresponding domain from calponin solved by x-ray crystallography, and the structure of the repeated domain is known by analogy to the known a-helical repeat domain in spectrin, and the EF-hand can be compared with calmodulin.

Alfa-actinin can be extracted from myofibrils without major destruction of the sarcomeric organization of the actin filaments in the myofibril, and when added back to the myofibril preparation it rebinds at the Z-disc. This implies that a-actinin may serve some function, other than a structural, in relation to the actin filaments. In nonmuscle cells, a-actinin is found enriched in spots along stress fibres and in cell adhesion plaques.

Dystrophin is the product of the gene responsible for the X-linked myopathies Duchenne and Becker muscular dystrophy. It is an elongated protein of high molecular weight (430 kD) present in low amounts in muscle and nerve cells. A closely related protein, utrophin, is more widely distributed, and there are many dystrophin/utrophin isoforms produced, due to the operation of alternative promotors and alternative splicing. The structure of these proteins place them in the spectrin/a-actinin family of proteins. They are multidomain proteins; the largest isoform is comprised of 3 N-terminal and 2 internal actin-binding sites, a peptide-binding WW domain, a calcium-binding EF domain, a zinc finger domain. The N-terminal actin binding region has sequence homology with the calponin homology CH domains of the spectrin/a-actinin family of proteins.

Dystrophin/utrophin connect cortical actin filaments with transmembrane proteins, dystroglycans and sarcoglycans, which in turn associate with extracellular matrix proteins, laminin and merosin, respectively. Skeletal muscle dystrophin can be purified from muscle cells as a large multiprotein dystrophin-glycoprotein complex, which stabilizes actin filaments in vitro through lateral associations. Both dystrophin and utrophin bind with higher affinity to b- than to a-actin. Comprehensive reviews on the structure and function of the dystrophin/utrophins are found in ref 237, 248.

6. Ion Channel-associated Actin-binding Proteins

The spectrin-actin network in erythrocytes is involved in anchoring ion exchange proteins in the membrane via the linker protein ankyrin, a critical link in maintaining the characteristic biconcave shape of the cell (249). In axons, an interaction between actin, fodrin, and ankyrin appears to be responsible for concentrating sodium channels at the nodes of Ranvier (249). At the neuromuscular junction, the clustering of acetycholine receptors appears to be mediated by spectrin, actin, and the rapsyn/43 kD actin-binding protein (250). Various lines of evidence suggest that intact actin filaments, and perhaps actin polymerization itself, are required for calcium regulation of the postsynaptic response of NMDA receptors in the central nervous system (251). Postsynaptic excitatory synapses are most often found on the small actin-rich budlike structures, called spines, that protrude from the dendrites of highly arborized neurons such as cerebellar Purkinje cells (252). Dendritic spines have pronounced electron-dense (as seen by electron microscopy) post-synaptic densities, called PSD’s, that are the likely sites of proteins linking receptors to actin. Candidate proteins include PSD-95/SAP90, chapsyn/PSD-93, SAP102, and alpha-actinin-2. Only alpha-actinin has a demonstrated actin binding affinity, and it has been suggested that clustering for some classes of receptors might be mediated by a looser, more "corral-like", entrapment mechanism (253). The PSD contains brain dystrophin, an isoform of the muscle protein originally identified as the mutated gene product involved in muscular dystrophy that is known to link transmembrane ion channels to actin filaments (254).

The precise role for actin in the functioning of ion channels is unclear. The provision of clustered anchorage sites may confer cooperativity in the opening of channels, or support the formation of multiprotein complexes capable of integrating different signalling or environmental factors. It is not unlikely that the actin network might passively or actively transmit forces to the channels providing the impetus behind stretch-activated gating phenomena.

7. Myosin as an Actin-binding Protein

The myosins belong to a large superfamily of proteins, which together with actin transduce chemical energy to force generating tension and movements in the eukaryotic cells. Myosin II is the prime partner of actin in the generation of force in muscle, as well as in nonmuscle cells. Generally myosins are thought of as the force generators, and therefore referred to as motor molecules. However, the actual mechanism of force generation is still unknown. Although, the ATP-binding head domain is largely conserved, there are many different molecular forms of myosin serving in as diverse functions as muscle contraction, cell motility, membrane traffic, and sensory perception. Myosin hydrolysis ATP to ADP.Pi still bound to the head domain. Interaction with actin filaments brings about product release, and in the presence of actin filaments, there is a rapid ATP-dependent cycling of heads on and off the actin filaments as ATP is hydrolysed. A current review of the myosin superfamily and their functional involvements is found in ref. 255.

8. Regulators of the Actomyosin Interaction

The N-terminus of actin is involved in the binding of a large number of actin binding proteins, including myosin (S1), tropomyosin, troponin I, a-actinin, caldesmon, gelsolin, cofilin, actobindin. This actin interface is important for muscle contraction, since it is implicated not only in the activation of the myosin ATPase, but also in the regulation of actomyosin interaction by interacting with troponin I.

Tropomyosin is an elongated, dimeric, coiled-coil a-helical protein that binds along the actin filament (see special entry). The tropomyosin molecule has a sevenfold repeat of nonpolar and polar amino acid residues that bind seven actin monomers in the filament. Multiple genes for tropomyosin and alternative splicing can generate a number of tropomyosins that are differentially expressed during development and in different cell types. Muscle cells express 1-2 and most vertebrate nonmuscle cells 3-8 tropomyosin isoforms. Tropomyosin in complex with the heterotrimeric protein troponin makes muscle contraction Ca dependent, but the function of tropomyosin in nonmuscle cells has not been clarified. However, the involvement of tropomyosin in the control of chemomechanical transduction also in nonmuscle cells is indicated by the occurrence of tropomyosin in tightly bundled and contractile organizations of actin filaments (stress fibres). For further information see special entry on tropomyosin.

Troponin is a heterotrimeric protein complex that interacts directly with actin through one of its subunits. Together with tropomyosin it confers calcium sensitivity to muscle contraction in striated muscles (see special entry).

Caldesmon is involved in the regulation of the actomyosin interaction in smooth muscle (256, 257). It inhibits actomyosin ATPase and filament sliding in vitro has been shown to bind to subdomain 2 of actin. Three-dimensional image reconstruction of reconstituted thin filaments consisting of actin and smooth muscle tropomyosin, both with and without a caldesmon derivative added has been reported. In filaments containing the caldesmon derivative, tropomyosin was found in a different position as compared to the situation in the absence of the caldesmon. The observations suggest that caldesmon causes changes in the relationship between tropomyosin and the actin filament, which are different from those seen with troponin.

The giant modular protein nebulin (mw 600- 900 kD) spans the whole length of the thin filament of the striated muscle sarcomeres in vertebrates, and has been proposed to function as a "ruler" for control of the length of actin filaments in sarcomeres (258-260). It is thought to bind and stabilize F-actin. It comprises 2-3 % of the myofibrillar protein mass of skeletal muscles. There are tissue and development-specific isoforms. The C-terminal part of human nebulin is anchored in the sarcomere Z-disk and contains an SH3 domain. The nebulin SH3 sequence from several species has been determined and found strikingly conserved. Its 3D-structure has been determined in solution by NMR spectroscopy, and its interaction with poly(L-proline) has been modelled. Nebulin consists of nearly 200 tandem repeats of homologous modules about 35 residues long. These are organized into about 20 tandem superrepeats consisting of 7 different modules, and this superrepeat segment is flanked near the N- and C-termini by single-repeat regions containing 8 modules of the same type. It has been proposed that the 35 residue module is the basic actin-binding domain and that the superrepeats reflect tropomyosin/troponin binding sites along the nebulin molecule. Exactly how the nebulin molecules are linked to actin filaments in the sarcomere of muscle is not yet known. Nebulin promotes actin nucleation and stabilizes actin filaments. Crosslinking experiments have identified the first two residues in actin to be involved in binding nebulin.

9. Future Directions

One of the most remarkable aspects of the contractile apparatus in muscle cells is the capacity of the system to ramp up its power output in response to increased demands for mechanical work. This phenomenon is known as the Fenn Effect, and it suggests that the acto-myosin system can increase its ATPase activity in response to higher imposed loads. It is apparently the lattice organization of filaments into sarcomeres that somehow enables the effect of external loads to be transmitted directly to the proteins producing the biochemical reactions that produce work from ATP hydrolysis. It remains to be seen if the bundles and meshworks found in non-muscle cells display similar, or even more unusual, chemomechanical feedback mechanisms. Muscle cells also perform work at nearly 100% thermodynamic efficiency, and it is interesting to ask whether the free energy transduction pathways in the cytoplasm are equally optimized for their tasks.

The newly discovered connections between the microfilament system and signal transduction raise a host of questions that will be the subject of future research. What roles do mechanical forces play in transmitting signals from the surfaces of cells to their nuclei? What regulates the assembly of these focal sites of signalling potential and what leads to their breakup after a signalling cascade has outlived its usefulness?

The large number of actin-binding proteins obscures the possible branching points that must have occurred during evolution to establish this incredibly diverse dynamical system. The analysis of evolutionary relationships in actin-binding proteins from amoeba, plants, and mammals will be of extraordinary interest for understanding how the immune system and the mammalian nervous system developed as specializations of more primitive motile mechanisms.