The actin stubs model for parasite actomyosin motility shown in Figure 1 can explain the observations that were puzzling in the light of a conventional microfilament system. But what are the components needed to make motility based on F-actin stubs work? Homology searches of apicomplexan genomes for microfilament proteins return a strikingly low number of orthologs of proteins that one might expect in highly motile cells expressing one or two different actin isoforms (Table 1). Many protein families that are considered key actin regulators in yeast and higher eukaryotes are absent in Apicomplexa, and where orthologs do exist they often display prominent heterologies. For example, in the expanding leading edge of motile mammalian cells, the creation of transient-free barbed ends that support actin polymerization is associated with the severing activity of gelsolin, while filament nucleation and branching is mediated by the Arp2/3 complex (20). Neither gelsolins nor components of the Arp2/3 complex are encoded in apicomplexan genomes (12,14). Other examples of actin-modulating proteins apparently missing in Apicomplexa include most families of filament cappers, filament cross-linking proteins like α-actinin or fimbrin, actin depolymerizing factor (ADF)-interacting and F-actin depolymerizing capping protein (CP), actin interacting protein-1 (Aip1) and regulatory classes like Mena/Ena/VASP proteins, WASP homology (WH)-1-domain containing proteins (verprolin) and WH2-domain containing proteins [Wiskott-Aldrich syndrome protein (WASP)/WASP family verprolin homology protein (WAVE)/suppressor of cAMP receptor (Scar)].
Actin depolymerizing factor/cofilins are among the most important proteins catalyzing microfilament turnover. They serve as actin polymers and enhance the rate of dissociation at the pointed ends (21). In contrast to Toxoplasma and Cryptosporidium, Plasmodium species express two ADFs, one of which has recently been characterized. The Plasmodium ortholog (PfADF1; PFE0165w) is abundantly expressed in all motile stages and binds monomeric actin with a preference for ADP-bound over ATP-bound monomers, but does not interact with actin polymers (22). Moreover, while other ADFs inhibit nucleotide exchange on monomeric actin, this essential Plasmodium ortholog stimulates nucleotide exchange. Thus, the biochemical properties of recombinant PfADF1 suggest that the protein may spur F-actin formation by interacting with ADP-actin monomers in order to generate polymerizable ATP-actin monomers.
The PfADF1 as well as the ADF sequences of Cryptosporidium and Toxoplasma all lack the established actin-polymer-binding motifs (22,23). In the light of these new findings, the reported F-actin-binding activity of the recombinant Toxoplasma protein (24) may need to be studied in greater detail. Based on structural models, it has been suggested that non-homologous replacements in the apicomplexan actin sequences may explain the intrinsic instability of parasite microfilaments (16,18). In such a scenario, the filament depolymerizing activity of ADF/cofilins may not be required. Notably, the second Plasmodium ADF isoform, PfADF2 (PF13_0326), resembles conventional ADFs and apparently contains the F-actin-binding sites that are lacking in PfADF1. However, this protein appears to serve a specialized function that may be restricted to life cycle stages that include neither the pathogenic merozoites nor the transmissible sporozoites (22).
The regulation of Plasmodium ADF activity remains to be characterized. Typically, ADF activity is under tight regulation by a phosphoregulatory switch mechanism: Phosphorylation/dephosphorylation at an amino-terminal serine results in deactivation/activation of ADF and, as a consequence, in F-actin stability or F-actin disassembly and severing, respectively (25). Apicomplexan ADFs all retain a serine in position 3, but the parasite genomes encode none of the two ADF-selective kinase families that phosphorylate and inactivate ADF, LIM kinases (LIMK) (26) and TES kinases (TESK) (27). However, with 99 putative protein kinases in the parasite genome (28), phosphorylation may well be exerted by functional equivalents of LIMK and TESK, which remain to be identified. In support of the existence of an ADF kinase, a paralog of 14-3-3ξ, a class of proteins that stabilizes phosphorylated, inactive ADF (29) is present. Of the corresponding ADF-specific phosphatases, Slingshot phosphatase (30) is missing while a potential paralog of chronophin (31), a member of the haloacid dehalogenase family of phosphatases, is present, at least in the genomes of Plasmodium and Toxoplasma species.
The major actin-sequestering protein in non-muscle cells is profilin, which keeps actin monomers in an unpolymerized yet polymerization-competent state (32). Polymerization from profilin–actin complexes is mediated by WASP homology domains, apparently lacking in the phylum Apicomplexa, and by formin proteins (discussed below). Apicomplexan profilins are unusually long, containing at least an extra stretch of >25 amino acids inserted into the consensus at a site that, by structure-based sequence comparison, appears to be outside the actin- and polyproline-binding sites. The recombinant Plasmodium protein binds both monomeric actin and polyproline (H.S., unpublished data) showing that this Apicomplexa-specific sequence does not disturb interaction with these two ligands. Therefore, as profilin is also implicated in other functions, this sequence may relate to aspects other than actin regulation. Apicomplexan parasites probably express a single profilin isoform, which may be expected to exert vital functions, possibly also during parasite locomotion and host cell entry. We expect that in the absence of readily polymerizable actin, the role of profilin in formin-mediated polymerization will be crucial for the parasites, while the monomer sequestering function will be less significant. Likewise, the role of profilin as an actin nucleotide exchange factor will be less important in the presence of an ADF isoform with similar properties.
Adenylate cyclase-associated protein [CAP (33)] is a ubiquitous eukaryotic protein that binds monomeric ADP-actin. Interestingly, apicomplexan CAP-like orthologs exist that correspond only to the C-terminal portion of conventional CAPs, thus comprising the actin-binding domain, while lacking the regulatory adenylate cyclase-interacting domain. The latter domain has been shown to also mediate a ternary interaction with F-actin and ADF (34), a function that may be redundant in the absence of an F-actin-binding ADF in motile Plasmodium stages (22). The CAP-like ortholog from Cryptosporidium parvum displays the same twin-lobed extended β-sheet structure as the actin-binding domain of the yeast CAP ortholog, Srv2 (35,36).
A striking example for divergence of protein function within this phylum is the actin modulator toxofilin (15,37) from Toxoplasma, which lacks altogether in Plasmodium, Cryptosporidium and Theileria. Toxofilin sequesters actin monomers and caps filament ends, and green fluorescent protein-toxofilin impedes actin turnover in mammalian non-muscle cells. Toxofilin contains a cleavable signal peptide and was recently found in secretory organelles using a proteomics approach (38). Thus, toxofilin is likely secreted by the parasite and may perform regulatory functions in the host cell and possibly also in the parasite.
Actin polymerization machineries
Formin/diaphanous family proteins trigger monomer addition from profilin–actin complexes at the barbed end while capping that end (39). The barbed-end capping activity is mediated by the ring-like structures of the formin homology-2 (FH2) domains, while the profilin-binding activity resides in the adjacent FH1 domain. Notably, the apicomplexan genomes encode two to three FH2-domain containing proteins indicating that this mode of regulation of profilin-mediated actin polymerization may be conserved. Apart from the FH2 domain, Plasmodium formin-like protein-1 (FLP1; PFL0925w) also contains a rudimentary FH1 domain, whereas FLP2 (PFE1545c) contains two oligo-proline stretches upstream of the FH2 domain. These sequences may well be sufficient for recruiting profilin–actin complexes (40). Other domains typically found in formin/diaphanous proteins, such as the Rho GTPase-binding domain, the self-regulatory Diaphanous autoinhibitory domain (DAD) and the dimerization motif (39) cannot be found in the parasite orthologs, posing the questions of whether and how they dimerize and how they are regulated. Although apicomplexan FH2-containing proteins remain uncharacterized so far, expression profiling revealed a characteristic upregulation of PfFLP2 in invasive stages (41) supporting a role in transient rapid formation of microfilaments during parasite locomotion.
The Arp2/3 complex is an important filament nucleator in the leading lamella of motile cells. Its components Arp2 and Arp3 form a polymer end template on which actin monomers can add to build a filament (42). None of the principal Arp2/3 complex components is encoded in the genomes of Apicomplexa; however, there are a number of actin-related sequences in apicomplexan genomes the evolutionary relationship of which has been analyzed recently (14). These authors suggest that actin-like protein-1 (ALP1; Pf11_0114), which is most similar in sequence to Arp2 and Arp3, may perform Arp2/3-like functions. This is an interesting hypothesis especially in the light of poorly polymerizing Plasmodium actin. It evokes the questions of whether ALP1 would need to form a homo- or a hetero-dimer to nucleate actin polymers, and whether a putative ALP1 nucleator complex would need additional regulatory components. Also other Arps or ALPs might be able to interact with only one end of an actin polymer to nucleate or restrict polymer growth. At least several of the nine actin-related proteins are likely involved in the regulation of actin turnover. Of these, Arp1 (PFA0190c) may be part of the dynactin complex as in other organisms (14).
Actin filament-binding proteins
Filamentous actin capping is a means of preventing uncontrolled filament elongation, while simultaneously stabilizing filament nuclei for elongation at the free end. Given the low intrinsic stability of parasite F-actin (16–18), this nuclei stabilizing function is likely of particular importance for Apicomplexa.
In the absence of gelsolin genes, the putative barbed-end capping protein up-regulated in infective sporozoites-17 (UIS17) PfCPβ (PFE0880c), with homology to the β-subunit of mammalian CP, and its corresponding partner PfCPα (PFE1420w) are candidates for microfilament nucleating and stabilizing proteins in Plasmodium. The UIS17/PfCPβ was first identified in a screen for genes that are specifically upregulated in motile infectious sporozoites (43). It is still unclear whether these CPs are identical with the members of a previously identified heterotrimeric protein complex with F-actin-binding and capping activity, which also contained hsp 70 (44). Notably, CP is not necessarily implicated in actin regulation, as it is also a candidate component of the dynactin complex, the additional components of which are well conserved in apicomplexan genomes (14).
Coronins are F-actin-binding and bundling proteins implemented in microfilament regulation, actin–tubulin cross-regulation and microfilament-linked membrane trafficking (45). A coronin gene (PFL2460w) has been identified in Plasmodium and the protein shown to be associated with filamentous actin (46). This coronin does not retain the extremely basic N-terminal signature sequence, but it contains three WD40 repeats and a C-terminal coiled-coil region, thereby resembling the short orthologs that have been linked to actin regulation. The Plasmodium ortholog shares sequence similarity with the C-terminal part of yeast coronin, but lacks the MAP1B-homology region implicated in microtubule binding. It also lacks PST-rich sequences found, for example, in human coronin-7, a protein that does not interact with actin and is linked to Golgi trafficking.
Apicomplexa possess unconventional myosins with considerable divergence from the myosins of other species (47). The Plasmodium genome encodes probably six different myosins (48) which all belong to the phylum-specific class XIV, with single heads and relatively short tails. The Plasmodium myosins remain largely uncharacterized except for Myosin A (MyoA; Pf13_0233), which, like its paralog TgMyoA in Toxoplasma (49,50), is likely the motor that drives parasites forward and powers invasion of host cells. The PfMyoA localizes to the periphery of invasive stages (51,52) and is fixed to the inner membrane complex through its binding partner, Myosin A tail domain interacting protein (MTIP) [PFL2225w (53)], an interaction that has been reconstituted in vitro to produce actomyosin motility (54). Apart from their role in force generation, myosins may induce actin filament formation and participate in F-actin turnover regulation. Therefore, in the presence of only a few classes of actin regulators, it will be interesting to test if the apicomplexan myosins also classify as regulators of actin turnover. An attractive hypothesis is that the otherwise unstable Plasmodium actin polymer stubs assemble at the tip of myosin heads (Figure 1), a stabilizing interaction that would bring them into their correct cellular context.