INTRODUCTION

Top Previous Next References

A diverse group of intracellular microorganisms, including Listeria monocytogenes, Shigella spp., spotted fever group Rickettsia spp., and vaccinia virus, utilize actin-based motility to move within and spread between mammalian host cells. L. monocytogenes is a gram-positive bacillus that enters the human host via the intestine and can cause meningitis, fetal death, and diarrhea. Shigella spp. infect cells of the intestine and cause diarrhea and dysentery. Spotted fever group Rickettsia spp. are fastidious obligate intracellular coccobacillary organisms that cause Rocky Mountain spotted fever and related diseases. Vaccinia virus is a poxvirus that is the vaccine against smallpox.

These organisms have in common a pathogenic life cycle that involves a stage within the cytoplasm of mammalian host cells (Fig. 1). The bacterial organisms induce uptake into an endocytic vacuole, while vaccinia virus enters by fusion. Bacteria gain access to the cell cytoplasm by lysing the vacuole, whereas vaccinia virus enters directly into the cytoplasm. Once in the cytoplasm, each of these microbes recruits to its surface host actin and other cytoskeletal proteins and activates the assembly of an actin tail.

View larger version (39K): [in this window] [in a new window]   FIG. 1.   Pathogenesis of Shigella (representative of the pathogenesis of Listeria and Rickettsia as well). 1, Shigella organisms (solid ellipses) enter mammalian host cells by inducing phagocytosis. 2 to 4, After entry, the bacterium is within a phagocytic vacuole (step 2), which it lyses (step 3), thereby releasing it into the cytoplasm of the host cell (step 4). 5, the bacterium assembles an actin tail on one pole. Assembly of the actin tail propels it through the cell cytoplasm. 6, Actin tail assembly also enables it to form a protrusion from the cell surface. The protrusion contacts the membrane of the adjacent cell and is taken up, along with the bacterium within it. 7 to 9, The bacterium is then within a double-membrane vacuole, which it lyses, thereby releasing it into the cytoplasm of the adjacent cell. 10, The bacterium again assembles an actin tail that propels it through the cell.

The continuous assembly of an actin tail provides sufficient force to propel the organisms through the cytoplasm of the infected cell and into adjacent cells. Passage of Listeria, Shigella, and Rickettsia into adjacent cells occurs via membrane protrusions that form when the bacterium pushes out against the cell membrane (Fig. 1). These protrusions are engulfed by the adjacent cell, placing the bacterium into a double-membrane-bound vacuole (Fig. 1). The bacterium lyses the double membranes and is thereby released into the cytoplasm of the adjacent cell. Vaccinia virus also forms protrusions from the cell; however, in contrast to Listeria, Shigella, and Rickettsia, the viral particle fuses with the membrane at the tip of the protrusion and is thereby released into the extracellular space.

Actin-based motility is essential to the virulence of Listeria, Shigella, and vaccinia virus, and although its role in virulence has not been directly tested for Rickettsia, it is presumably important there as well. For Listeria, Shigella, and vaccinia virus, genetic deletion or modification of the genes that encode proteins known to be required for actin-based motility markedly attenuates the organism. For Rickettsia, genetic systems for manipulating genes do not yet exist.

As will become evident in this review, each of these organisms utilizes preexisting mammalian pathways of actin rearrangement to induce its own actin-based motility. In particular, each organism has evolved a protein(s) that binds and activates one or more of the components of a mammalian actin assembly pathway, thereby inducing the cascade of activation of all downstream effector molecules, with the end result being de novo actin nucleation, polymerization, and cross-linking. Particularly remarkable is that while all of these microbes use the same or overlapping pathways, each intercepts the pathway at a different step. In addition, the microbial molecules involved are all distinctly different. Taken together, these observations suggest that each of these microbes separately and convergently evolved a mechanism to utilize the cellular actin assembly machinery. The current understanding of the molecular mechanisms of microbial actin-based motility is the subject of this review.

To better understand microbial actin-based motility, the current understanding of the principles of actin cytoskeletal dynamics will be reviewed here. Several excellent reviews on this subject have recently been published (8, 19, 29, 31, 62, 125, 129, 184, 199).

Actin forms the scaffold of the cell's supportive structures. It is assisted in this by a large number of proteins known collectively as actin-binding proteins or actin-associated proteins. In many cell types, actin is the most abundant protein, constituting more than 5% of total cellular protein. Actin exists in two forms, monomeric (G-actin), which is soluble in the cytosol, and filamentous (F-actin). "Filamentous actin" refers to thin, flexible, helically arranged homopolymers of actin (Fig. 2A) that may be up to hundreds of micrometers in length.

View larger version (28K): [in this window] [in a new window]   FIG. 2.   Actin filament structure and assembly dynamics. (A) Asymmetric homopolymer of helically arranged actin. The pointed or minus end (left) and barbed or plus end (right) are indicated. ATP bound to the actin monomer is hydrolyzed to ADP shortly after addition to the filament. (B) Actin polymerization curve. An idealized curve from a pyrene actin assay is shown. Actin is polymerized in vitro under experimental test conditions. At time zero, potassium and magnesium are added. The lag phase represents the time required for actin nucleation. The rapid polymerization phase represents the time during which short filaments elongate. Steady state represents an equilibrium between growth of the filaments due to monomer addition and shortening of the filaments due to loss of monomer.

Assembly is such that actin filaments are asymmetric, with two structurally distinct ends. ATP-bound monomers polymerize onto the barbed or plus end of the filament at up to 10 times the rate at which they polymerize onto the pointed or minus end. A large pool of filamentous actin is in dynamic equilibrium with the pool of monomeric actin, while certain actin filaments (such as those in microvilli) are quite stable. Trimers of actin (actin nuclei) are the nucleating structures for the spontaneous polymerization of monomeric actin into filaments. Actin filaments are generally found associated with other actin filaments, cross-linked into bundles or aggregates by one of several actin cross-linking proteins.

Each actin molecule is a single polypeptide of 395 amino acid residues. Actin filaments are approximately 8 nm in diameter. Each actin monomer is tightly bound by a molecule of ATP. Following addition of the monomer to the barbed end of the filament, ATP is hydrolyzed to ADP (Fig. 2A). Monomers that dissociate from the pointed end of the filament are therefore bound by ADP, which can then be exchanged for ATP in a process that is slow. The regenerated ATP-bound monomer is then competent for addition to a filament barbed end. The process of continually adding monomers to the barbed end while continually dissociating monomers from the pointed end is known as treadmilling.

The free actin monomer concentration at which the proportion of cellular actin that is polymerized is constant (i.e., the critical concentration) is approximately 0.2 µM (approximately 8 µg/ml). The concentration of monomeric actin in the cell is actually much greater than this, typically about 50 to 200 µM (2 to 8 mg/ml). To maintain this high concentration of unpolymerized actin, the pool of monomeric actin is largely sequestered by actin monomer-binding proteins, principally thymosin and profilin, such that it is not free. In addition, an actin monomer that dissociates from a filament is bound by ADP, which must be exchanged for ATP before it is competent to bind a filament barbed end, and, as noted above, ADP-ATP exchange on actin monomers is slow.

Addition or dissociation of monomers from filaments can be blocked by proteins that bind specifically to a filament end (capping proteins). Filaments can be broken into short filaments by severing proteins. Thus, modulation of actin monomer-binding proteins, capping proteins, and severing proteins enables the cell to regulate the rate of actin assembly or dissociation in localized regions of the cell. Because of its ability to rapidly polymerize into filaments, to rapidly dissociate into monomers, or to be severed into short filaments, actin provides the cell with a dynamic structural scaffold that can be rearranged quickly in response to appropriate stimuli.

Actin polymerization is commonly measured in vitro by a spectrofluorometric assay in which fluorescently tagged actin (pyrene-actin) gives a wavelength-specific signal on polymerization (30, 80). Polymerization of pure actin requires monovalent and divalent cations, typically potassium and magnesium. An example of the readout from the spetrophotometric assay is shown in Fig. 2B. In a typical experiment, potassium and magnesium are added to an experimental sample that contains monomeric actin and ATP and polymerization is monitored spectrophotometrically over several minutes. Initially there is a lag phase, which represents the time required for actin nucleation, the process in which actin nuclei (trimers), which serve as the starters for actin filaments, are formed. The lag represents a kinetic barrier to nucleation that is due to the metastable character of nuclei; only some will go on to form filaments. The lag phase is followed by a rapid polymerization phase, during which short filaments elongate, which is seen as a steep rise in the curve. The slope of the curve represents the rate of actin polymerization. Following the rapid polymerization phase, a steady-state equilibrium is reached as the growth of the filaments due to monomer addition is balanced by the loss of monomer from the filaments; the equilibrium is seen as a leveling off of the curve. Thus, the spectrophotometric assay permits the analysis of the extent of the kinetic barrier to polymerization, the rate of polymerization, and the relative steady-state amount of polymerized actin under a given set of experimental conditions. The dissociation of actin filaments can be monitored in a similar assay system.

A tool that is commonly used in the study of actin assembly in vitro is cytoplasmic extracts isolated from living cells. Extracts allow the investigator to study the motility of objects that are not able to enter intact mammalian cells, such as Sepharose beads coated with a protein of interest (18) or Escherichia coli cells that express a protein of interest (56, 78). Extracts also allow the investigator to deplete specific factors or add purified proteins, antibodies, or peptides. Most commonly used are extracts from Xenopus laevis oocytes, human platelets, and bovine brain. Variations in the methods of preparation exist; depending on the details of preparation, extracts may contain somewhat different concentrations of various cytoskeletal proteins.

Mammalian cells have the capacity to form several types of cell extensions that are dependent on de novo actin polymerization. These extensions are mediated by local rearrangements of the actin cytoskeleton in the cell cortex, the region just beneath the plasma membrane. Actin filaments assemble near the leading edge of the membrane and thereby push it forward (109); they disassemble at some distance away from the membrane (174). Lamellipodia are sheet-like extensions and filopodia (microspikes) are long thin extensions of the cortex and plasma membrane. Formation of these extensions is regulated by a complex set of signaling pathways involving the small GTPases Cdc42 and Rac (discussed below and summarized in Fig. 3). As will become apparent in this review, microbes that utilize actin-based motility do so by means of these preexisting cellular actin rearrangement signaling pathways.

Rho GTPases are a family of small molecules that function as molecular switches in the regulation of a variety of cellular processes, including actin cytoskeletal dynamics, transcription, cell adhesion, and cell cycle progression (reviewed in references 11 and 71). Ten members of the family, many with multiple isoforms, have been identified in mammalian systems. Rho, Rac, and Cdc42 are the most extensively characterized and appear to be most important in actin cytoskeletal signaling pathways. Rho GTPases cycle between an active GTP-bound state and an inactive GDP-bound state. The GTP-bound form is able to interact with downstream effector molecules, while an intrinsic GTPase activity returns the molecule to its inactive state.

The principal function of Rho GTPases is regulation of the assembly and organization of the actin cytoskeleton (Fig. 3). Extracellular factors induce cytoskeletal changes through these switches. Filopodium formation, which can be induced by bradykinin, is thought to be regulated by Cdc42, since expression of a dominant negative Cdc42 inhibits this response (81). Lamellipodium formation, which can be induced by several growth factors (platelet-derived growth factor, insulin, or epidermal growth factor) is thought to be regulated by Rac, since expression of a dominant negative Rac specifically inhibits this response (141). Stress fiber and focal-adhesion formation are thought to be regulated by Rho, since exoenzyme C3 transferase, which inactivates Rho, blocks these responses (140). Thus, three relatively distinct signal transduction pathways are thought to be involved in the regulation of the actin cytoskeleton.

View larger version (23K): [in this window] [in a new window]   FIG. 3.   Signaling pathways involving the small GTPases Cdc42 and Rac. Shown is an overview of the Cdc42 and Rac signal transduction pathways that lead to actin rearrangements in the form of lamellipodia (membrane ruffles) and filopodia (microspikes). (Adapted from reference 11 with permission of the publisher.)

Since the microbial pathogens discussed in this review utilize the Cdc42 pathway for actin-based motility, components of this pathway are described in detail in the following sections.

Each of the microbial pathogens that utilizes actin-based motility activates actin assembly through the Arp2/3 complex. The Arp2/3 complex is localized to areas of dynamic actin assembly in the cell (97, 116, 118, 173, 185). Data from the last several years suggest that the Arp2/3 complex is probably essential to the actin rearrangements that mediate the formation of lamellipodia and filopodia, as well as the formation of microbial actin tails.

Arp2/3 is a stable complex of seven polypeptides, present in 1:1 stoichiometry with each other (93, 97, 118, 185, 189). The complex is present in all eukaryotes examined, and each of the subunits is evolutionarily highly conserved. That the complex is functionally important has been demonstrated in yeast, where deletion of subunits is lethal or severely deleterious to cell growth (6, 88, 103, 112, 113, 155, 189, 190).

Two of the seven subunits can be modeled to fold similarly to conventional actin (73) and have therefore been designated Arp2 and Arp3 (for "actin-related protein"). The surface of the Arp2 and Arp3 subunits that corresponds to the barbed end of conventional actin is relatively similar to conventional actin, while the surface that corresponds to the pointed end of conventional actin is relatively divergent (73). The similarities at the barbed ends suggest that the barbed ends might function similarly to the barbed ends of conventional actin. The other five polypeptides in the complex are novel, and their designations vary among species (94).

The Arp2/3 complex is the only known cellular factor that stimulates the nucleation of actin filaments that grow from their barbed ends. It increases the rate of nucleation of new actin filaments, both in vitro (115, 187) and on the surface of Listeria monocytogenes (186). It is thought that the barbed ends of the Arp2 and Arp3 subunits serve as the surface on which nucleation is initiated.

Arp2 and Arp3 are not known to interact within an unactivated Arp2/3 complex (reviewed in reference 117). A model for the mechanism of Arp2/3 complex stimulation of actin nucleation suggests that on activation, the conformation of the complex is altered such that Arp2 and Arp3 are brought together to form a stable dimer within the complex. The barbed ends of Arp2 and Arp3 are thought to be exposed, so that polymerization of a new actin filament can be initiated on them. The filament would then extend by polymerization off the barbed end of this nucleus. This model predicts that the Arp2/3 complex caps the pointed end of the growing filament, to which it has been shown to bind with high (nanomolar) affinity in vitro (14, 115).

In addition to nucleating new actin filaments, the Arp2/3 complex links filaments at 70° angles (12, 115, 124). Branched structures similar to those observed in vitro are seen at the leading edge of motile cells (Fig. 4A to C) (173). Arp3 and one of the smaller subunits, p35 (or its human homolog p34), appear to bind actin directly (118). The 70° angle and the actin filaments themselves are quite stiff (12); as a result, the polymerizing, cross-linked network could generate force against the cell membrane, thereby pushing it forward. Extension of the filaments is halted by the binding of capping protein to the free barbed ends, which probably occurs at a distance from the leading edge or the microbial surface.

View larger version (95K): [in this window] [in a new window]   FIG. 4.   Dendritic network at the leading edge of a lamellopodium of a motile cell. (A) Branching of actin filaments at approximately 70° angles, consistent with Arp2/3-mediated cross-linking. (B and C) Higher magnification of boxes b and c, respectively, in panel A. (D) Dendritic nucleation model. The proposed model of Arp2/3 complex-mediated nucleation and branching of actin filaments is shown. The Arp2/3 complex binds to the side of an actin filament and is bound by a WASP family member (in this diagram, Scar). Both binding to the side of the filament and binding by a WASP family member are thought to play a role in activation of Arp2/3 complex nucleation of the new filament on the side of the mother filament. (E) Proposed model for the formation of branched networks of actin filaments in lamellipodia at the leading edge of motile cells. The Arp2/3 complex, in conjunction with a WASP family member (not shown), nucleates actin filaments on the sides of existing filaments. Cofilin severs filaments to generate additional uncapped barbed ends that can polymerize quickly. These processes lead to the rapid extension of an actin filament network that generates the force to push the cell membrane forward. (Panels A, B, C, and E reprinted from reference 173 with permission of the publisher. Panel D reprinted from reference 96 with permission of the publisher.

Branching of filaments by the Arp2/3 complex appears to occur by binding of the Arp2/3 complex to the sides of filaments. The bulk of evidence supports a model in which the Arp2/3 complex binds the side of an actin filament with direct interactions between the Arp3 and p34 subunits and the exposed regions of actin subunits within the mother filament (2, 12, 115), although an alternative model has been proposed (124). Binding of the Arp2/3 complex to the side of actin filaments stimulates actin nucleation (2, 5). Side binding is thought to activate the Arp2/3 complex so that a new filament is initiated from the barbed ends of the Arp2 and Arp3 subunits. This mechanism has been designated the dendritic nucleation model of Arp2/3-mediated actin assembly (Fig. 4D) (115).

In isolation, the Arp2/3 complex activates actin nucleation only weakly (187). Arp2/3 complex activation of actin nucleation is stimulated (i) by other cellular proteins, in particular members of the Wiskott-Aldrich syndrome protein (WASP) family (194), and (ii) by filamentous actin (2, 5). As described in detail below (see "Listeria"), L. monocytogenes has evolved to directly stimulate Arp2/3 complex-activated actin nucleation by mimicking WASP, while Shigella and vaccinia virus have evolved to indirectly stimulate Arp2/3 complex-activated actin nucleation by activating WASP family members on their surfaces. Mimicking of WASP or activation of the WASP family member leads to stimulation of the Arp2/3 complex and rapid actin assembly.

The WASP family includes WASP, which is expressed only in hematopoeitic cells and is mutated in Wiskott-Aldrich syndrome (35); N-WASP, which is expressed ubiquitously and particularly enriched in the brain (104); Scar (WAVE), which is expressed widely (95); and the yeast isoform Las17p/Bee1p (90). Wiskott-Aldrich syndrome is a rare inherited immunodeficiency (reviewed in reference 163). Patients suffer from impaired immunity, thrombocytopenia, eczema, and hematopoietic malignancies. Cytoskeletal abnormalities have been demonstrated in lymphocytes. However, the wide range of clinical abnormalities have not yet been explained at the molecular level.

Members of the WASP family are structurally similar, sharing particular functional domains and domain organization (Fig. 5). Near the carboxy terminus of all of these proteins is one or two verprolin homology (V) domains (also known as WASP homology 2 [WH2] domains) and an acidic (A) domain. Between these two domains is a region that has been referred to as a cofilin homology (C) domain. In WASP and N-WASP, the verprolin homology, cofilin homology, and acidic domains are together referred to as the VCA domain. In Scar, they are referred to as the WA domain. The VCA (or WA) domain stimulates Arp2/3 complex activation of actin nucleation (96, 145, 188). The verprolin homology domain binds monomeric actin, and the acidic domain binds the Arp2/3 complex (95, 100, 107, 145). The cofilin homology domain is involved in intramolecular interactions with the amino-terminal regulatory region of the protein (see below). Despite its name, the structure of this domain appears to be dissimilar from that of the related sequences in cofilin, and no cofilin-like activity has been demonstrated for it or the intact proteins (39, 63). Experimental data suggest that this region participates in actin binding (100, 107).

View larger version (20K): [in this window] [in a new window]   FIG. 5.   Domain structure of WASP family members. Black, WH1 domain; blue, GBD; yellow, proline-rich region (number indicates the number of motifs containing a stretch of five or more prolines); green, WH2 or verprolin homology domain; pink, acidic domain (numbers indicate the numbers of acidic/basic residues). The cofilin homology domain lies between the WH2 and acidic domains and is not specifically indicated. Numbers to the right indicate length in amino acids. (Adapted from reference 63 with permission of the publisher.)

In all WASP family members, amino-terminal to the WH2 domain(s) is a proline-rich region that binds the actin monomer-binding protein profilin (106, 169). This proline-rich region enhances stimulation of the Arp2/3 complex by the WH2 and acidic domains (96). However, the role of profilin in Arp2/3 complex activity is uncertain. Deletion of the proline-rich region of Scar has minimal effects on actin assembly (106), and profilin inhibits rather than stimulates polymerization in the presence of a Scar fragment that contains the proline-rich, WH2, and acidic domains (96). However, in the presence of N-WASP and the Arp2/3 complex, Cdc42-stimulated nucleation of actin is enhanced by profilin (193). Of note, if the concentration of free monomeric actin is held constant, the stimulation of actin assembly by both Cdc42 and a carboxy-terminal fragment of N-WASP is enhanced by profilin, even though the carboxy-terminal fragment of N-WASP does not contain the proline-rich sequences that bind profilin (193), suggesting that a part of the enhancement that is mediated by profilin may be independent of binding to N-WASP.

In WASP and N-WASP, amino-terminal to the proline-rich region is a consensus binding site for the GTPase Cdc42 (105), known as the GTPase-binding domain (GBD). Within this domain is a 16-residue conserved motif, known as the Cdc42/Rac-interactive binding (CRIB) motif, which is also present in unrelated Cdc42 ligands. Near the amino terminus of WASP and N-WASP is a pleckstrin homology (PH) or WASP homology 1 (WH1) domain, which is involved in the binding of phosphatidylinositol-4,5-bisphosphate (PI(4,5)P₂) (104). Adjacent to the PH domain in N-WASP but not in WASP is a calmodulin-binding (IQ) motif (Fig. 5). Recently, a lysine-rich basic motif that lies just amino-terminal of the GBD has also been shown to specifically bind PI(4,5)P₂ (131).

In the inactive state of N-WASP or WASP, the GBD forms an intramolecular interaction with the so-called cofilin homology domain (74, 105). The carboxy-terminal end of the GBD (C-terminal to the CRIB motif) and the cofilin homology domain are each both necessary and sufficient for this autoinhibitory interaction (74). Binding of Cdc42 to WASP, and by extension to N-WASP, disrupts this intramolecular linkage and autoinhibition (74). One of the GBD residues involved in autoinhibition is known to be mutated (Leu²⁷⁰ to Pro) in the naturally occurring human disease X-linked severe congenital neutropenia, which is characterized by severe recurrent bacterial infections (36).

The GBD, to which Cdc42 binds, and the basic motif, to which PI(4,5)P₂ binds, together constitute the essential regulatory region of N-WASP and, by extension, WASP (131, 144). Whereas N-WASP that is not bound by Cdc42 and PI(4,5)P₂ or that is bound by only one of these molecules stimulates Arp2/3 complex-mediated actin nucleation only weakly, binding of both molecules synergistically stimulates N-WASP activation of Arp2/3-mediated actin nucleation (39, 131, 145). Consistent with these observations, in the absence of Cdc42 and PI(4,5)P₂, an N-WASP fragment that consists only of the WH2, cofilin, and acidic domains, thereby lacking the regulatory region, stimulates actin polymerization markedly more than full-length N-WASP does (39, 145).

It has been suggested that in their folded autoinhibited conformations, WASP and N-WASP are unable to bind the Arp2/3 complex. Structural studies further suggest that residues that are involved in binding to the Arp2/3 complex may be inaccessible (74). However, data addressing whether the Arp2/3 complex binds autoinhibited N-WASP are conflicting (131, 144).

Cofilins are actin-associated proteins that have been described to have both actin-severing and -depolymerizing activities (21, 23, 67). The family consists of highly homologous members that have essentially identical activities. Depending on the species, cofilins have been designated cofilin, actin-depolymerizing factor (ADF), actophorin, or destrin. In motile cells, cofilin contributes to actin polymerization at the leading edge by severing actin filaments (23, 196), thereby generating a pool of uncapped barbed ends that can rapidly extend. In its phosphorylated state, cofilin is inactive. Inactivation of cofilin by overexpression of the kinase that phosphorylates it completely inhibits the generation of actin barbed ends and abolishes cell motility (196). Thus, cofilin is essential to the motility of eukaryotic cells.

Arp2/3 complex-mediated linkages between filaments are metastable. At a distance from the leading edge, the actin filament branches dissociate from the mother filaments and depolymerize. Dissociation of the Arp2/3 complex from the mother filaments is accelerated by the conversion of ATP-actin to ADP-actin in the filaments, a process that is enhanced by cofilin (14). The Arp2/3 complex is thought to dissociate from the pointed ends of the filaments, either before or after debranching, allowing rapid depolymerization of actin from the free pointed ends.

The WASP family of proteins and the Arp2/3 complex appear to be essential to the pathway of de novo actin filament nucleation that induces lamellipodium and filopodium formation, as well as microbial motility (described in detail below). At the cell cortex, activated Arp2/3 complex nucleates new actin filaments by stabilizing and capping the barbed end of actin oligomers. It does so after binding to the sides of existing actin filaments and/or by forming branches at the uncapped barbed ends of existing filaments. Actin filaments extend rapidly by polymerization on uncapped barbed ends. Also at the cell cortex, cofilin severs existing and/or newly polymerized filaments, thereby generating additional uncapped barbed ends off which polymerization occurs. As a result, branched networks, which provide the force to push the cell membrane forward, are rapidly formed (Fig. 4E). The Arp2/3 complex is activated by WASP family members. Autoinhibition of WASP family members, which is mediated by an intramolecular interaction, is relieved by the binding of Cdc42 and PI(4,5)P₂.

L. monocytogenes and the related L. ivanovii are facultative intracellular pathogens that move through cells and spread directly from one cell into an adjacent cell by means of actin-based motility. This section of the review will focus on actin assembly induced by L. monocytogenes; relatively little is known about that induced by L. ivanovii (22, 50, 57, 72, 82). L. monocytogenes is widespread in nature and causes disease in a variety of herd animals as well as in humans. In humans it can cause meningitis, fetal death, and diarrhea. The ability to spread directly between cells allows Listeria to avoid many components of the host immune response. In the murine model of infection, the organism colonizes and replicates in the liver and spleen. The 50% lethal dose for mice of an L. monocytogenes mutant defective in actin assembly is 4 log units higher than that of wild-type L. monocytogenes (76).

L. monocytogenes is able to enter and assemble actin in all types of adherent cells that have been tested. In the cytoplasm of infected cells, it has been documented to move at rates of 13.2 to 22 µm/min (34, 58), which are approximately the same rates of movement as those documented for Shigella (5.4 to 26 µm/min [58, 158, 197]), threefold higher than the rates of movement documented for Rickettsia (4.8 to 8 µm/min [58, 59]), and five- to eightfold higher than the rates of movement documented for vaccinia virus (2.8 µm/min [32]) (Table 1). Dabiri et al. reported individual L. monocytogenes cells moving at 88 µm/min in J774 macrophages (34), but such speeds have not been documented by other investigators. In cytoplasmic extracts, L. monocytogenes has been documented to move at 6.4 to 13 µm/min, which is about half the rate of movement of intracellular L. monocytogenes (58, 78, 99, 176).

When visualized by electron microscopy, Listeria actin tails in the cytoplasm of infected cells consist of multiple short actin filaments that are arranged in a disorganized-appearing nonparallel network (Fig. 6) (177, 180). Decoration of tails with the S1 subfragment of myosin, which binds to subunits within actin filaments in an asymmetric fashion, demonstrates that the barbed ends of the filaments within the tails are oriented toward the bacterial body (177). Actin tails formed on the surface of beads coated with the Listeria actin assembly protein ActA have the dendritic appearance of actin filaments near the leading edge of lamellipodia (Fig. 4A) (20).

View larger version (108K): [in this window] [in a new window]   FIG. 6.   Thin-section electron microscopy of an actin tail formed by an intracellular L. monocytogenes bacterium. Note that the tail consists of short actin filaments that are bundled in a nonparallel fashion. (Reprinted from reference 180 with permission of the publisher.)

In Vero cells, Listeria actin tails average 1 µm in diameter and 5 µm in length, compared to 0.7 µm in diameter and 7 µm in length for Shigella and 1.5 µm in diameter and 5 to 17 µm in length for Rickettsia (58, 75) (Table 2). Of note, the length and diameter of actin tails for any particular organism vary among different cell lines, such that comparisons can not be made reliably across cell lines.

Near to the bacterium, filaments in the Listeria tail are longer than those at a distance from the bacterium and appear to lie parallel to the sides of the bacterial body, with relatively few filaments immediately adjacent to the pole (58, 78). This differs from the tails of Shigella, where the filaments appear short throughout the tail and, near to the bacterium, are adjacent to the pole and not the sides of the bacterial body. These differences in patterns of actin filaments on Listeria and Shigella mirror the differences in localization of the actin-polymerizing proteins of the two organisms, ActA and IcsA, as described below. The filaments within the tails of Listeria have been reported to be short, generally about 0.1 µm in length (58), although other data suggest that they may be several fold longer (M. Coughlin and T. J. Mitchison, unpublished data). The filaments within the tails of Shigella have also been reported to be about 0.1 µm in length, whereas those within the tails of Rickettsia and vaccinia virus have been reported to be longer, 0.3 to 3 µm (or >1 µm) and 0.74 µm, respectively (33, 58, 75).

To spread from one cell into an adjacent cell, Listeria pushes against the cell membrane to form a finger-like protrusion that contains the bacterium at its tip. The protrusion, along with the bacterium within it, is then taken up into the adjacent cell by an endocytic process. The actin tails that trail bacteria within protrusions differ in morphology from the tails within the body of the cell. Those within the protrusion consist predominantly of extremely long filaments of actin that are bundled in a parallel array (Fig. 7) (156). They contain relatively few short actin filaments of the type seen in actin tails within the body of the cell; the few short filaments that are present are oriented at angles to the long filaments (156). Near to the bacterial body within protrusions, the filaments splay out and are less tightly bundled (156). The reason that tails within the protrusions differ in morphology from those within the cell body is uncertain, although a possible explanation is that factors involved in generating the branched array or in capping the growing filaments near to the bacterial body may be partially excluded from the protrusions.

View larger version (139K): [in this window] [in a new window]   FIG. 7.   Negative-staining electron microscopy of an actin tail formed by an L. monocytogenes bacterium within a cell surface protrusion. Note that the tail consists of long actin filaments that are bundled in a parallel fashion. (Reprinted from reference 156 with permission of the publisher)

The L. monocytogenes surface protein ActA mediates actin assembly. L. monocytogenes strains that carry a gene disruption in ActA are unable to assemble actin tails or spread from cell to cell in a tissue culture monolayer (16, 37, 76) and are 3 log units less virulent in mice (16). Heterologous expression of ActA in L. innocua, a nonpathogenic Listeria species that does not normally polymerize actin, enables this organism to assemble actin tails (78). Purified ActA that is linked to the surface of streptococci (161) or to the surfaces of beads (18) enable the streptococci or beads to form actin tails, demonstrating that ActA is sufficient to mediate actin-based motility in the absence of other Listeria factors.

ActA is a 639-amino-acid protein that is anchored in the bacterial membrane by the carboxy-terminal 26 amino acids, such that following cleavage of the 29-amino-acid amino-terminal secretion signal, the 584-amino-acid residue mature amino terminus is exposed on the bacterial surface (76). ActA has a calculated molecular mass of 67 kDa, although it migrates on sodium dodecyl sulfate-polyacrylamide gel electrophoresis at 90 kDa (76). ActA expression is markedly increased inside mammalian cells (16, 43, 110). Several strains that constitutively overexpress ActA have facilitated its purification and been used in motility assays (160). ActA from Listeria-infected cells migrates as three distinct polypeptides, as a result of its phosphorylation during intracellular growth (16). Phosphorylation appears to occur on residues within the proline-rich repeat region (see below), since an ActA derivative with this region deleted is not phosphorylated (85). The role of ActA phosphorylation in Listeria motility is unknown.

The functions of sequences within the surface-exposed portion of ActA have been characterized by several groups, largely through extensive deletion analysis and site-directed mutagenesis. Native and recombinant ActA are readily purified, facilitating the analysis of mutant ActA proteins in vitro. Removal of the carboxy-terminal transmembrane domain of ActA leads to its secretion, allowing its easy purification from culture supernatants. Note that in the published literature, some authors designate as amino acid 1 the amino-terminal residue of mature ActA (after cleavage of the signal peptide); in this review, we designate as amino acid 1 the amino-terminal residue of full-length immature ActA, which includes the signal peptide.

The entire amino-terminal portion of the mature protein, amino acids 30 to 263, is essential for actin polymerization in Listeria-infected cells. Deletion of amino acids 50 to 260 leads to loss of the ability to assemble actin filaments in Listeria-infected mammalian cells (85), and deletion of amino acids 30 to 265 leads to loss of association of actin with a derivative of ActA that anchors on the mitochondria of transfected cells (126). A fragment that consists of ActA amino acids 30 to 263 is sufficient for motility in cytoplasmic extracts when artificially bound to the bacterial surface (86).

ActA-mediated stimulation of actin polymerization requires the mammalian host protein complex Arp2/3 (186, 187). As discussed in detail above, the known biochemical activities of the Arp2/3 complex are (i) cross-linking or branching of actin filaments at 70° angles (12, 115, 124) and (ii) weak nucleating activity (116, 118). By immunofluorescence, the Arp2/3 complex is located throughout the actin tail of motile Listeria (185, 186). Sequestration of Arp2/3 complex by addition to cytoplasmic extracts of the Arp2/3-binding domain of Scar1 (WAVE), a WASP family member, or depletion of the Arp2/3 complex from platelet or bovine brain extracts inhibits Listeria motility (39, 101, 194). Similar inhibition of Listeria motility is seen on overexpression of the same domain of Scar1 in infected cells (101). Motility is rescued by the addition of purified Arp2/3 complex (39, 101, 194), specifically indicating that sequestration of the Arp2/3 complex interferes with Listeria actin tail formation. Reconstitution of Listeria motility using purified cytoskeletal proteins requires the Arp2/3 complex (91). Using pure proteins in vitro, in the absence of the Arp2/3 complex the mature surface-exposed portion of ActA (amino acids 29 to 613) or ActA amino acids 29 to 263 has no effect on the nucleation of actin polymerization, but in the presence of Arp2/3 complex each has a marked stimulatory effect (187) (Fig. 8). Finally, the actin filaments in the tail of intracytoplasmic Listeria are branched at acute angles, similar to the acute angles at which the Arp2/3 complex branches actin in vitro (12), and, by electron microscopy, the Arp2/3 complex is localized to the Y junctions of the dendritic network of tails formed by ActA-coated beads in extracts (20). These data suggest that the Arp2/3 complex mediates some or all cross-linking or branching of actin filaments in the Listeria tail.

View larger version (25K): [in this window] [in a new window]   FIG. 8.   ActA stimulation of Arp2/3 complex-mediated actin polymerization. Fluorescence intensity is plotted as a function of time in pyrene-actin polymerization assays. 1, actin; 2, actin and Arp2/3 complex; 3, actin and ActA (amino acids 29 to 263); 4, actin, Arp2/3 complex, and ActA (amino acids 29 to 263). (Reprinted from reference 187 with permission of the publisher.)

After entry into the cell cytoplasm or placement in cytoplasmic extracts, the association of Listeria with actin that is first visualized is the assembly of a "cloud" of filamentous actin that surrounds the bacterium. This cloud evolves into an actin tail that extends exclusively from one pole of the bacterium. An area of uncertainty at present is whether the Arp2/3 complex or another factor(s) is responsible for the conversion of actin clouds into actin tails. Whereas actin tails consist of filaments that have been shown to be cross-linked (20), it is not known whether the filaments within the actin cloud are cross-linked. The data are confusing. In vitro, in the presence of the Arp2/3 complex and actin, wild-type Listeria forms clouds of polymerized actin on its surface but rarely forms actin tails (187). In addition, sequestration of the Arp2/3 complex by overexpression of the Arp2/3-binding domain of Scar1 eliminates actin tail assembly but not the formation of actin clouds on the bacterial surface (101). These data suggest that either (i) the affinity of ActA for the Arp2/3 complex is adequately high that, in the second set of experiments, sufficient Arp2/3 complex remains associated with ActA to nucleate the polymerization of filaments on the Listeria surface but inadequate amounts are present to branch or cross-link filaments into a tail, or (ii) as pertains to the first set of experiments, while the Arp2/3 complex is required for cross-linking or branching of the filaments into a tail, other host factors must also be present.

Alanine mutagenesis shows that an extensive region of ActA, stretching from amino acids 166 to 256, appears to be involved in the ability of ActA to convert actin clouds into actin tails. At 4 h of infection, wild-type Listeria is associated with an actin tail about twice as often as it is associated with an actin cloud. Mutation of clustered charged amino acids within amino acids 166 to 256 leads to an inversion of this ratio as well as up to a 50% decrease in rates of movement, by a mechanism that is unknown (86a). Whether these residues in ActA are involved in modifying the interaction with the Arp2/3 complex or in binding another host factor is not known.

Since the WH2 and acidic domains of WASP family members are required for WASP-mediated Arp2/3 complex activation of actin nucleation, investigators have looked to identify and characterize homologous domains in ActA, as well as other domains that might function in actin assembly. A summary of the results of these works is shown in a schematic diagram of the functional domains of ActA in Fig. 9A.

View larger version (18K): [in this window] [in a new window]   FIG. 9.   Functional domains of L. monocytogenes ActA. (A) Schematic of ActA. SP, signal peptide; A, WASP-like acidic domain, possibly involved in Arp2/3 complex binding; WH2, WASP homology 2 domain; PI, phosphoinositide binding region; Pro-rich repeats, proline-rich repeats; TM, transmembrane anchor; AB, monomeric actin-binding region, amino acids 60 to 100; dimer, region required for ActA dimerization, amino acids 126 to 155; clouds/tails, region important for formation of actin tails, rather than actin clouds, amino acids 166 to 256. (B) Alignment of ActA amino acids 85 to 99 with the first of two WH2 domains of N-WASP and of ActA amino acids 121 to 170 with the WH2 and Arp2/3 complex-binding domains of WASP family members. (C) Alignment of ActA amino acids 32 to 45 with the acidic domains of WASP family members. Sequences used for alignment: mouse WASP, human N-WASP, and human Scar1. (Panel B adapted from reference 195 with permission of the publisher. Panel C adapted from reference 159 with permission of the publisher.)

The ability of ActA amino acids 29 to 263 to stimulate Arp2/3 activation of actin assembly in vitro suggested that this region of ActA could interact with both Arp2/3 and monomeric actin. ActA amino acids 30 to 263 fused to a linker and expressed on the bacterial surface mediate actin tail assembly in cytoplasmic extracts (86), which demonstrates that the region is sufficient for actin tail assembly. When ActA is purified and bound to beads, the same region of ActA polymerizes actin in cytoplasmic extracts (28).

ActA binds the Arp2/3 complex with moderately high affinity (K_d, 0.7 ± 0.2 µM) (195). ActA inhibition of the spontaneous polymerization of actin in a pyrene-actin assay best fits a model in which ActA binds two actin monomers (195). One of these actin-binding sites appears to lie between ActA amino acids 60 and 101. Deletion of amino acids 60 to 101 leads to a marked decrease in the binding of ActA to monomeric actin, as measured by the loss of ability of the construct to inhibit the spontaneous polymerization of actin in both a pyrene-actin assay and an actin-pelleting assay (in which monomeric actin remains soluble whereas polymerized actin pellets) (159). A similar effect is seen using ActA lacking the entire amino-terminal region (amino acids 31 to 262) (159). An ActA fragment that consists of amino acids 60 to 100 binds actin, as measured in the same pyrene-actin assay, and a glutathione S-transferase (GST) fusion to ActA amino acids 62 to 103 binds fluorescently labeled monomeric actin (28, 195). Mutagenesis of charged residues to alanine within amino acids 60 to 66, 70 to 77, or 80 to 88 decreases the binding of the mutant ActA to monomeric actin (86a), indicating that the entire region between amino acids 60 and 88 is involved in binding to monomeric actin (86a). Amino acids 85 to 99 align with the first of two WH2 domains of N-WASP (195). WH2 domains are known to bind monomeric actin.

A second actin-binding domain in ActA has been identified at amino acids 121 to 138, immediately amino-terminal to the Arp2/3-binding domain (see below) (195). An ActA fragment that consists of amino acids 121 to 152 binds actin, as measured in the pyrene-actin assay (195). However, alanine mutagenesis of clustered charged residues within this region does not affect binding to monomeric actin in the same assay (86a), raising the possibility that actin binding of this region is context specific. Within this region, amino acids 121 to 138 align with the actin-binding WH2 domain of WASP family members, most closely with the second of the two WH2 domains of N-WASP (Fig. 9B) (195).

Actin polymerization assays of ActA derivatives in the presence of the Arp2/3 complex suggest that ActA amino acid residues 121 to 170 contain an Arp2/3-binding and activation domain. Amino acids 121 to 170 of ActA activate Arp2/3-mediated actin polymerization (195), and deletion of amino acids 136 to 165 from an active fragment of ActA leads to loss of this activity (159). Furthermore, deletion of amino acids 136 to 165 leads to loss of ActA-mediated actin assembly by Listeria in the cytoplasm of mammalian cells (159). Within this region, ActA amino acids 143 to 169 align with the Arp2/3-binding acidic domains of WASP family members (Fig. 9B) (195), suggesting that these residues are responsible for the observed Arp2/3-activating activity. In fact, deletion of five basic amino acids within this region (amino acids 146 to 150, KKRRK) leads to loss of actin tail assembly on Listeria in cytoplasmic extracts (86) and to a marked reduction in activation of Arp2/3-mediated actin polymerization (159). When all of these five amino acids are changed to alanine, similar results are obtained (86a). Moreover, mutation of either Arg¹⁴⁸ to Lys or Arg¹⁴⁹ to Ser leads to loss of actin polymerization on the bacterial surface (128). Of note, the Arp2/3 complex still colocalizes with the bacterium, as determined by immunfluorescence of the 21-kDa subunit, suggesting that Arg¹⁴⁸ and Arg¹⁴⁹ are required for activation of the Arp2/3 complex but not for its binding (128). Alanine mutagenesis of the charged residues within the overlapping stretch of amino acids 140 to 147 leads to a more modest effect, whereas alanine mutagenesis of amino acids 156 to 160 leads to a moderate effect, as seen in actin polymerization assays (86a). Taken together, these data suggest that Arg¹⁴⁸ and Arg¹⁴⁹ are absolutely critical and that amino acids 156 to 160 are important for the activation of Arp2/3-mediated actin polymerization.

The entire amino terminus of ActA (amino acids 30 to 612) binds Arp2/3 with moderately high affinity (K_d, 0.5 ± 0.2 uM). Deletion of ActA amino acids 136 to 165 does not eliminate binding to the Arp2/3 complex (159), suggesting that the stretch between ActA amino acids 143 to 169 is not the sole region of ActA that binds Arp2/3. The amino terminus of mature ActA (amino acids 31 to 58) contains a stretch of acidic amino acids that aligns with a carboxy-terminal acidic domain of WASP family proteins (Fig. 9C). Deletion of this domain leads to a decrease in ActA activation of Arp2/3-mediated actin nucleation (159). In addition, alanine mutagenesis of charged residues within amino acids 32 to 42 or 44 to 54 results in significantly decreased rates of Listeria movement (86a). Mutations in amino acids 44 to 54 result in an approximately twofold decrease in activation of Arp2/3-mediated actin polymerization, whereas mutations in amino acids 32 to 42 do not alter the activation of Arp2/3-mediated actin polymerization (86a). In WASP family members, the carboxy-terminal acidic domain is known to be involved in binding to the Arp2/3 complex. Taken together, these data suggest that in ActA, amino acids 31 to 58 may be a second region, in addition to amino acids 143 to 169, that is involved in Arp2/3 binding, although the exact limits of the binding region are still poorly defined.

ActA amino acids 121 to 170 chemically cross-link with a zero-length cross-linker to the Arp2/3 complex subunits p40, Arp2, and Arp3 (195). The WASP family member Scar1 chemically cross-links to the same Arp2/3 complex subunits (195). Furthermore, Scar1 and N-WASP each compete with the binding of ActA to the Arp2/3 complex in vitro (195).

ActA binding to the Arp2/3 complex competes with the binding of profilin but not cofilin (actophorin) (195). The protein sequence of Arp2 is 50% identical to that of conventional actin, and the profilin- and cofilin-binding domains of conventional actin have been identified (13, 53, 154). These data, along with the chemical cross-linking studies described above, have permitted the proposal of a binding site of ActA on Arp2 within the Arp2/3 complex (Fig. 10) (195). This binding site would place ActA or other nucleation-promoting factors (i.e., WASP family members) at the site on Arp2 that is analogous to the barbed end of conventional actin (195). This may enable ActA or another nucleating factor to assist in the addition of actin monomers to the Arp2/3 complex.

View larger version (29K): [in this window] [in a new window]   FIG. 10.   Proposed location of the ActA-binding site on the Arp2/3 complex and on the Arp2 subunit. (A) Proposed site for binding of ActA, Scar, and N-WASP to the nearest-neighbor model of the Arp2/3 complex. Shading indicates subunits to which ActA, Scar, and N-WASP cross-link. Open circles and dark lines indicate subunits that cross-link within the complex (118). Ovals indicate interactions demonstrated by two-hybrid analysis (95). (B) Profilin- and actophorin (cofilin)-binding sites on conventional actin (13, 53, 154). (C) Proposed structure of the Arp2 subunit, based on homology to conventional actin (73). The binding site of ActA that has been proposed on the basis of competition experiments (see the text) is indicated. (Reprinted from reference 195 with permission of the publisher.)

The data on ActA interactions with the Arp2/3 complex and monomeric actin provide compelling evidence that ActA is structurally and functionally similar to members of the WASP family of nucleation-promoting factors. This is the first example of several in this review of a microbial protein that has evolved to mimic a mammalian one.

The Arp2/3 complex is distributed throughout the length of the actin tail on intracellular Listeria (185, 186). The Arp2/3 complex, bound to ActA on the bacterial surface, forms the pointed end of newly nucleated actin filaments, which rapidly polymerize on the free barbed ends. Recent data indicate that the Arp2/3 complex dissociates from ActA by the time approximately 10 actin monomers have added to the barbed end of the new filament. As described above, the Arp2/3 complex also links filaments at 70° angles, most probably by initiating new filaments after binding to the side of an existing filament or by cross-linking the Arp2/3-capped pointed end to the side of an existing filament (2, 115). Thus, it appears that in conjunction with its dissociation from ActA, the Arp2/3 complex forms a cross-linked or branched network of rapidly growing actin filaments at the bacterial surface. The 70° angle and the actin filaments themselves are quite stiff (12); as a result, the polymerizing, cross-linked network can generate force against the bacterial surface, thereby pushing the bacterium forward. Extension of the filaments is halted by the binding of capping protein to the free barbed ends (see below). Free Arp2/3 complex probably rapidly binds the available Arp2/3 binding site on ActA, where it is activated and initiates further de novo actin nucleation.

The central region of ActA (amino acids 293 to 390) contains four 11-amino-acid proline-rich repeats, the first three of which contain the motif FPPPP and the fourth of which contains the motif FPPIP (37, 76). These proline-rich repeats are separated by three 24- to 33-amino-acid sequences, the first two of which are almost identical. Proline-rich sequences similar to those found in ActA are also present in the mammalian cytoskeletal proteins zyxin and vinculin (138). Zyxin is present in focal contacts, microfilaments, and dynamic regions of the membrane (138), and vinculin is present in focal contacts (17, 47) and intercellular adherens junctions (48, 49).

The vasodilator-stimulated phosphoprotein (VASP) and the related protein Mena are present on the surface of intracytoplasmic Listeria (22, 52). On motile bacteria, VASP and Mena localize to the bacterial pole at the junction of the bacterium with the actin tail (22, 52), in the same distribution as ActA. Consistent with this localization, purified VASP binds ActA (22).

VASP and Mena are members of the Ena/VASP family of proteins, which also includes Ena and Evl. VASP, Mena, and Evl are mammalian proteins, while Ena is the Drosophila protein Enabled (51). These proteins localize to focal contacts and regions of dynamic actin rearrangements, such as the leading edge of motile cells (52, 137). The rate of extension of the membrane at the leading edge correlates with the local concentration of VASP (147).

The polypeptides of members of this family contain three domains: highly homologous amino- and carboxy-terminal domains, called Ena/VASP homology domains 1 and 2 (EVH1 and EVH2), and a central more divergent proline-rich region. The amino-terminal EVH1 domain, which is similar to the WH1 domain of WASP family members (see "Actin cytoskeleton: dynamics and function" above), mediates interactions with focal contact proteins, including zyxin and vinculin (15, 52, 119, 138, 139). The proline-rich region mediates interactions with profilin, SH3 domains, and WW domains (1, 41, 52, 136). The carboxy-terminal EVH2 domain mediates tetramerization of the protein and interactions with actin filaments (4, 87). Of note, the Ena/VASP proteins and the WASP family of proteins have proline-rich motifs that contain the consensus (G/A/L/S)PPPPP (132), that are distinct in both structure and function from the proline-rich motifs of ActA, and that do not bind VASP (119). VASP is enriched in platelets, and Mena and Evl are enriched in brain extracts. In Listeria motility assays, VASP, Mena, and Evl functionally complement one another, although not to wild-type rates of movement (87).

The proline-rich repeats of ActA are required for VASP binding. Deletions or site-directed mutagenesis of these repeats leads to loss of detectable VASP by immunofluorescence microscopy (119, 127, 162). Peptide scanning of ActA identified only the proline-rich motifs as VASP ligands (119). The binding site of the ActA proline-rich motifs on VASP, Mena, and Evl, is the EVH1 domain. ActA binds the EVH1 polypeptide of Mena or Evl fused to GST in a solid-phase binding assay (119). Preincubation of Listeria with the EVH1 polypeptide of VASP, Mena, or Evl fused to GST inhibits bacterial actin assembly on addition to extracts (87). A single tryptophan residue within the EVH1 domain (Trp²³ of Mena and Ena) is critical for binding to the ActA proline-rich region (42, 130).

ActA-derived peptides that contain residues flanking the FPPPPT core bind with higher affinity than do truncated subsequences (119). The ActA dodecapeptide ³³³FEFPPPPTEDEL³⁴⁴ binds with the highest affinity of all peptides tested (K_d, 19 µM) and with significantly higher affinity than zyxin or vinculin peptides, indicating that determinants of both specificity and affinity reside in these flanking residues (7). This peptide binds the VASP EVH1 domain with significantly higher affinity than does a zyxin-derived peptide (7), and a similar ActA-derived peptide binds VASP with higher affinity than does intact zyxin or vinculin (119). Moreover, a variety of other proline-rich peptides do not bind VASP (119). The phenylalanine at position 333, the glutamic acid at position 343, the leucine at position 344, and the prolines at positions 336 and 339 are essential for the observed high-affinity binding to EVH1 (7). The extremely high affinity of ActA for VASP suggests that intracytoplasmic Listeria is able both to readily recruit VASP from its mammalian ligands zyxin and vinculin and to remain tightly bound to it.

Several roles for the interaction of Ena/VASP proteins with ActA in Listeria actin-based motility have been proposed and are supported to different extents by the existing data. VASP is known to bind the actin monomer-binding protein profilin (136), leading to the suggestion that VASP serves to increase the local pool of actin monomers in the vicinity of ActA. Profilin is present on the surface of intracytoplasmic Listeria, and the amount of profilin associated with the bacteria decreases proportionally with the amount of associated VASP (162). Consistent with this, in depletion/add-back experiments, VASP is required for profilin to have an effect on Listeria movement (87). However, data on the requirement of profilin for wild-type rates of Listeria motility are inconsistent. Two groups report differing results in assays of Listeria motility in extracts depleted of profilin (87, 99, 176). When Listeria motility is reconstituted using purified protein components, profilin increases the speed of motile Listeria but is not absolutely required for movement (91). Fairly consistent among these studies is that profilin accelerates Listeria movement modestly, that it accelerates movement less significantly than VASP, and that it requires VASP for its effect.

A second role proposed for VASP in Listeria motility is the bundling of actin filaments into a tail. In VASP-depleted extracts, actin polymerizes on the surface of Listeria into loosely organized networks but does not bundle to form tails (87). Also, in vitro, VASP can bundle actin filaments (87). However, since ActA amino acids 166 to 256, which lie outside of the proline-rich repeats, appear to also be required for the conversion of actin clouds into actin tails (see "Requirement of Host Arp2/3 complex for ActA-mediated actin assembly" above), if VASP is important in this process, it is probably not the only factor.

A third role proposed for VASP in Listeria motility is the initiation of actin polymerization on the bacterial surface. The data on this are conflicting. Deletion of the ActA proline-rich repeats, to which VASP binds, leads to decreased speeds of Listeria motility, with a linear relationship between the number of repeats and the speed (162). The frequency of actin tail formation is also lower in these strains, although in a nonlinear relationship that is suggestive of a role for the long repeats located between the proline-rich motifs in initiation of movement (162). In the complete absence of any proline-rich motifs, Listeria forms actin tails, albeit at low frequencies (162). Deletion of the proline-rich repeats also leads to a slowing of bacterial spread between cells, as measured by the plaque assay, and an increase of the 50% lethal dose in a mouse model of listeriosis (162). In the absence of the proline-rich repeats, no VASP or profilin is detected on the bacterial surface by immunofluorescence (162). When Listeria motility is reconstituted using purified protein components, Listeria is able to assemble actin tails in the absence of VASP, but the presence of VASP increases the speed of moving bacteria about 10-fold (91). In parallel experiments, in the presence of VASP, profilin increases the rate of movement about twofold (91).

In contrast, microinjection of a synthetic peptide that matches the second ActA proline-rich repeat leads to a complete halt of bacterial actin tail assembly and motility, not simply a decrease in speed (165). In addition, immunodepletion of VASP from platelet extracts or Evl from brain extracts of Mena^/ mice leads to a complete loss of Listeria actin tail formation that is reconstituted by the addition of recombinant VASP (87).

In extracts, the affinity of intact VASP for ActA appears to be higher than that of the VASP EVH1 domain alone, suggesting that VASP domains outside of the EVH1 domain interact with other regions of ActA or with other molecules in the assembly complex (87). In addition, as mentioned above, a large region of ActA (amino acids 166 to 256), which is just amino-terminal to the proline-rich repeats (amino acids 263 and 390), appears to be important in the conversion of actin clouds to actin tails (86a).

Whether VASP is required for the initiation of actin polymerization on the bacterial surface remains unresolved. Possible explanations for the observed discrepancies in the data include the possibility that in addition to the proline-rich region, VASP binds another site on ActA and the possibility that another factor that binds VASP and is essential to VASP-mediated activity is titrated out of the extracts or cytosol on immunodepletion of VASP. Such a factor might interact within ActA amino acids 166 and 256. Evidence against VASP binding to another site on ActA is that no ActA peptides outside of the proline-rich region bind VASP (119) and no VASP is detectable by immunofluorescence on the surface of ActA derivatives lacking the proline-rich region (119), although this latter assay could miss small amounts of VASP. Evidence against an essential factor being immunodepleted along with VASP is that repletion with micromolar amounts of recombinant VASP restores motility to wild-type rates (87).

It has been thought that Ena/VASP proteins serve to accelerate actin assembly by recruiting monomeric actin to sites of actin reorganization and perhaps also by bundling actin filaments. Recent data on the effect of overexpression and deletion of Ena/VASP proteins on cell motility have raised questions about the exact mechanism of action of these proteins in Listeria actin-based motility (9). Overexpression of Mena inhibits the motility of fibroblasts in a dose-dependent manner (9). Cells that genetically lack all Ena/VASP proteins move more quickly than do the same cells complemented with Mena (9). Also, constitutive targeting of Ena/VASP proteins to the cell membrane decreases cell motility and the rates of membrane protrusion and retraction (9). Taken together, these data indicate that Ena/VASP proteins negatively regulate fibroblast movement (9). However, this study addresses the effect of Ena/VASP proteins on fibroblast motility and not on actin dynamics per se. Fibroblast motility is mediated by membrane protrusion and retraction, which requires a delicate balance of factors. The precise way in which this balance is disrupted by alteration in Ena/VASP protein concentration is unclear. Reconciliation of these data with the observed effect of Ena/VASP proteins in Listeria motility will require further work, including the analysis of Listeria motility in cells lacking all Ena/VASP proteins.

Cofilin (ADF) depolymerizes actin. While its mode of action has been controversial, the data suggest that it (i) severs actin, thereby increasing the number of filament ends and consequently the number of sites at which dissociation of monomers can occur (67), and (ii) increases the dissociation of monomers from the pointed end (21). Possibly important to the controversy is the observation that the ability of cofilin to sever actin filaments is dependent on the method of its preparation (67).

In highly diluted platelet extracts, addition of cofilin leads to an increase in the rate of movement of Listeria and a shortening of the Listeria actin tail (21). These effects probably result at least in part from the generation of an increased pool of monomeric actin under conditions in which it is limiting. Reconstitution of Listeria motility using purified cytoskeletal proteins also requires cofilin, perhaps for similar reasons, since the reconstitution experiments were conducted using filamentous actin and not monomeric actin (91). Thus, whether cofilin is important in Listeria motility other than to contribute to the maintenance of an intracytoplasmic pool of monomeric actin remains uncertain. Of note, the rates of movement achieved in the reconstitution experiments are three- to fivefold lower than those achieved in intact cytoplasmic extracts (78, 99, 176), suggesting that the reconstitutive mixture used in these experiments may be incomplete.

Capping protein binds to the barbed ends of actin filaments and blocks actin assembly. Marchand et al. (99) suggested that some factor might maintain the filaments near the bacterial surface uncapped to allow their rapid elongation, although such a factor has not been identified. Recent evidence suggests that the regulated capping of these growing filaments is essential to Listeria motility. Capping protein is required for reconstitution of Listeria motility with purified components in the presence of filamentous actin (91). In the absence of capping protein, unregulated extension of the branched network of actin filaments generated at the bacterial surface may occur. Such unregulated polymerization can lead to the extension of filaments beyond the vicinity of the bacterial body to form a fishbone-like web (124), which would be inefficient at providing force behind the bacterial body. Unregulated polymerization would also lead to depletion of the local stores of monomeric actin, further impinging on bacterial movement.

Phosphoinositides bind to and play a role in the regulation of several actin-associated proteins, including capping protein, vinculin, profilin, and gelsolin. Phosphoinositides also bind ActA in vitro (28, 166). Binding of phosphoinositides to ActA induces a conformational change in the amino-terminal portion of the protein (amino acids 29 to 267) (28). The binding site appears to be located toward the carboxyl end of this region, between amino acids 217 and 266 (28). Within this binding site is a stretch of amino acids with significant homology to a domain found within cecropins. Cecropins are peptide antibiotics, and the cecropin domain that is homologous to ActA forms an amphipathic -helix that is thought to associate with membranes (28). Several phosphoinositides bind ActA with different affinities (28). It has been proposed that phosphoinositide bound to ActA might inhibit the capping activity of capping protein near the bacterial surface (99). The data are not convincing, however, since ActA binds the D-3 phosphoinositides better than PI(4,5)P₂ (28), and the D-3 phosphoinositides that have been directly tested do not inhibit capping protein activity (152). Moreover, ActA