Unusual characteristics of ciliate actins

REVIEW ARTICLE Eduardo Villalobo áPilar Pe Ârez-Romero RocõÂoSa Ânchez-Silva áAntonio Torres Unusual characteristics of ciliate actins Received: 30 April 2001 / Accepted: 15 June 2001 / Published online: 10 November 2001 ÓSpringer-Verlag and SEM 2001 Abstract Actin is a cytoskeletal protein that is ubiquitous in eukaryotes, hence the corresponding genes and proteins have been isolated from numerous organisms as dierent as animals, plants, fungi and protozoa. Several atomic models are available for the monomeric as well as the ®lamentous form, and more than 70 proteins that bind actin and control ®lament dynamics have been isolated from diverse eukaryotes. Moreover, the function and dynamics of the actin cytoskeleton in several eukaryotic systems have been depicted in depth. Unlike other protozoa, such as amoeba, actin is not an abundant protein in ciliates, whose cytoskeleton is mainly composed of microtubular arrays. Ciliate actin has been studied in several species, and it was established early on that this ciliate protein is very dierent from that of other eukaryotes. Similarly, the actin-binding proteins studied in ciliates display great dierences with those of other eukaryotes. Consequently, ciliate actin has been considered as ``unconventional,'' and this review focuses on molecular data leading to this conclusion. Keywords Actin áActin-binding protein áCiliate á Cytoskeleton Introduction Actin is one of the most highly conserved and best studied eukaryotic proteins. It may be present as monomeric G-actin in the absence of salt, or as micro- ®laments (F-actin) in the presence of physiological concentrations of salt. Monomeric actin binds 1 mol of ATP and one divalent cation per mol of protein. It is a Mg 2+ -stimulated ATPase, but at low salt concentrations, in which it does not polymerise, this ATPase activity is low. Under polymerisation conditions, monomers are incorporated into ®laments in the ATP state and the nucleotide is hydrolysed to ADP, although ®lament assembly is not dependent on this hydrolysis. Actin protein is a major cellular component of both the cytoskeleton and the muscle sarcomere. In muscle cells, actin is involved in myo®brillar construction. In non-muscle cells, the actin cytoskeleton plays an essential role in multiple cellular processes, including cell elongation and shape determination, cytoplasmic streaming, cell motility, division-plane localisation, chromosome segregation, secretion, endocytosis, and organelle transport [45]. All of these processes rely on the capacity of the actin cytoskeleton to respond to cellular signals and reorganise spatially and temporally into a variety of speci®c structures. The 42-kDa G-actin monomer typically consists of 375 amino acid residues. Primary structures of actins from dierent eukaryotic species exhibit exceptionally high sequence identities. There are at least six dierent isoforms of actin in mammals; two cytoplasmic actins, b and c, and four muscle actins. The latter include two striated muscle (a-skeletal and a-cardiac) and two smooth muscle (a-aortic and c-enteric) actins. In most eukaryotes, actin is encoded by a multigene family [7], although organisms such as yeasts [19], Giardia [13], and several ¯agellated protista [2] contain only a single actin gene. The multigene families are likely to have arisen from gene duplications followed by divergence of the duplicated copies during evolution. The dierent actin isoforms produced by the members of the actin gene families are typically highly conserved at the amino acid sequence level, apparently due to their critical functional roles in cells. Crystallographic data [5, 29,48] indicate that actin monomer is a globular protein consisting of two domains connected by a hinge region. These two domains, originally termed large and small (although now they are Int Microbiol (2001) 4: 167±174 DOI 10.1007/s10123-001-0032-1 E. Villalobo áP. Pe Ârez-Romero R. Sa Ânchez-Silva áA. Torres (&) Departamento de MicrobiologõÂa, Facultad de BiologõÂa. Universidad de Sevilla. Apdo 1095. 41080 Seville, Spain E-mail: [email protected] Tel.: +34-954557115 Fax: +34-954557830 known to have almost identical sizes), are further divided into subdomains. The small domain is composed of subdomain 1 (amino acid residues 1±32, 70±144, 338± 375, rabbit a-actin numbering) and subdomain 2 (amino acid residues 33±69), whereas the large domain comprises subdomain 3 (amino acid residues 145±180 and 270±337) and subdomain 4 (amino acid residues 181± 269). The large subdomain has a structural core made up of residues from both subdomains 3 and 4, whereas the small-domain core is made up exclusively of residues from subdomain 1. Subdomains 1 and 3 are thought to have evolved by duplication of an ancestral gene coding for a polypeptide of about 150 amino acids that appears to be an ancient nucleotide-binding pocket also found in heat shock proteins, sugar kinases and several polypeptides known to regulate the prokaryote cell cycle [3]. Subdomains 2 and 4 were probably inserted subsequently into subdomains 1 and 3, respectively. The nucleotide/cation complex is bound at the bottom of the cleft between subdomains 1 and 3, with the adenosine base resting in a hydrophobic pocket formed between subdomains 3 and 4. There are 15 residues involved in nucleotide-binding sites, of which 13 belong to subdomains 1 and 3, and two are within subdomain 4. The region of contact with DNase I consists of hydrogen bonds, electrostatic and hydrophobic interactions involving subdomain 2 (amino acid residues 39±46; 60±64) and subdomain 4 (amino acid residues 202±204 and 207). Residues 40±50 within subdomain 2 are highly disordered and form the DNase I loop. Actin-actin interactions in the ®lament involve about 45 residues. Subdomains 2 and 4 of the actin monomer have actinactin interfaces interacting with subdomains 1 and 3 of other actin monomers. Several amino acid residues in subdomain 2, including a region of the sequence that is nearly identical to that of the DNase I contact, have been shown to be in contact with other actin monomers. The actin helix is stabilised through a loop of 11 amino acids (262±272) that includes a four-residue hydrophobic plug. This loop inserts into a hydrophobic pocket formed by subdomains 2 and 3 of two adjacent monomers on the opposing strand. In the cell, many actin-binding proteins (ABPs) regulate the nucleation and assembly-disassembly of actin into ®lament. Most families of ABPs have been widely conserved over phylogeny, in both primary structure and biochemical properties, and they can be found in organisms as diverse as humans and yeast. This suggests that these proteins already existed in a common eukaryotic ancestor and that the basic mechanisms regulating the dynamics of the actin cytoskeleton are conserved among diverse organisms and cell types. More than 70 ABPs that can control ®lament dynamics have been identi®ed [35,37]. Some affect monomers by controlling sequestration or nucleotide exchange; others control ®lament formation and stability by regulating capping, nucleating, crosslinking, bundling, and severing. Finally, cellular signalling molecules and small GTPases control actin remodelling by regulating the activities of ABPs and numerous direct and indirect eectors [36,44]. Ciliate actin This cytoskeletal protein has been studied in several species, although results are scarce, sometimes contradictory, and generally restricted to Tetrahymena and Paramecium. The following sections are dedicated to reviewing the most relevant knowledge on ciliate actin. The protein The ®rst attempts to demonstrate the presence of actin in ciliates were carried out by indirect techniques, such as heavy meromyosin (HMM) decoration or immunodetection. Thus, actin was localised in the cortex of Discophrya [23], Paramecium [49] and Tetrahymena [31], and in the oral apparatus of Pseudomicrothorax [25]. Lately, Paramecium actin was found in food vacuoles, around contractile vacuoles [9], and in the epiplasm [8]. However, it was not detected deeper in the cortex, as previously suggested. In Tetrahymena, actin was also found in the food and contractile vacuole, in the division furrow [38], and in the basal body-cage complex [28]. The link between ciliate actin and basal bodies is not an exception, since actin has also been reported to be associated with axonemes in algae [43] and birds [47], and with centrosomes in vertebrates [6]. Evidence of the association of ciliate actin with basal bodies has been obtained by us using an antibody raised against centractin, a centrosome-associated actin. In Paramecium, this anti-centractin antibody labelled basal bodies not only in the cortical rows but also in the oral apparatus (our unpublished data). Actin has been puri®ed in Climacostomum [18], Paramecium [49] and Tetrahymena [27]. In these ciliates, the puri®ed actin corresponds to a protein of about 43 kDa, if some confusing reports on Tetrahymena actin are not considered [39,41]. Conventional methods were used to purify actin in Paramecium and Tetrahymena, whereas DNase I-anity chromatography was used in Climacostomum. Using this last method, a 43-kDa protein was also isolated in Paramecium [8], but the authors questioned whether this protein was actually actin. In fact, although its peptide mapping did not coincide with that of rabbit muscle actin, an antibody raised against this 43-kDa protein [8] decorated the epiplasm, the same structure decorated by a monoclonal antibody against rat skeletal actin [34]. Far from shedding light on ciliate actin, these results pose new questions as to its biochemical nature. Nevertheless, to date a detailed biochemical characterisation of actin is available only in Tetrahymena [27]. In some respects, Tetrahymena actin shows biochemical properties similar to that of rabbit muscle actin, namely: (1) it polymerises in vitro into micro®laments in an 168 ion-dependent-manner; (2) it forms arrowhead structures with HMM; and (3) it activates the Mg +2 -ATPase of myosin S-1. However, Tetrahymena actin shows unusual properties in two respects: ®rst, it does not bind to phalloidin; second, it does not inhibit DNase I activity. This latter feature of Tetrahymena actin raises the possibility that the puri®cation of Paramecium and Climacostomum actin by DNase-I anity chromatography is artefactual. This is supported by the fact that the DNase-I-puri®ed actin of Climacostomun forms micro- ®laments in vitro but independently of salt concentration. All these unusual properties allow us to assume that, overall, ciliate actin can be considered unconventional. This would explain in part why many antibodies raised against conventional actin fail to recognise any protein in the 43 kDa range in total protein extracts of many ciliate species. For instance, in total protein extracts, the antibody of Lin [34] did not detect any polypeptide in Paramecium and Tetrahymena, although the antibody recognised a polypeptide of about 43 kDa in Climacostomum [18] and, surprisingly, a set of polypeptides ranging from about 110 to 97 kDa in Euplotes (our unpublished results). The genes Actin genes have been sequenced in many species, despite the fact that the protein has been isolated in only a few ciliate species. As in yeasts, the number of macronuclear actin genes in ciliates is low, ranging from one in Histriculus [42] and Tetrahymena [10,26] to three in Oxytricha [12,30] and Stylonychia [24]. The size of the actin macronuclear molecules in hypotrichs, in which the size of the whole gene unit can be obtained, ranges from about 1,200 bp to about 1,600 bp, excluding telomere repetitions. The micronuclear counterparts of the macronuclear actin genes have been sequenced only in Oxytricha and Urostyla. Interestingly, in the micronuclear DNA of Oxytricha trifallax, ten macronucleardestined sequences (MDSs) corresponding to actin I are dispersed and disordered in the chromosome, a phenomenon called ``scrambling'' [14]. Scrambled MDSs must be rearranged during macronuclear development to form a functional macronuclear molecule. Macronuclear actin genes are relatively A+T-rich (mean of about 56%), although this richness depends on the region considered and the species analysed. In general, the A+T content in non-coding regions (5¢- and 3¢- untranslated region and introns) is higher than in the coding regions. This A+T content is especially high (75± 91%) in the two short introns present in Paramecium [11], Vorticella and Opisthonecta actins [46], the only available ciliate actin genes known to be interrupted by introns. Note that, in Paramecium, one intron is located within subdomain 1 and the other one within subdomain 3, which supports the idea that both domains arose by duplication of an ancient polypeptide and/or by exon shuing. A promoter sequence motif (TATA-box) can be found in most but not all the available actin sequences, but whether this motif is directly implied in transcription is not known. In Histriculus cavicola [42], there are two potential TATA boxes (40 and 65 bp upstream of the initiation codon) in its unique actin gene, and the initiation of transcription occurs in the fourth nucleotide (A) of the ®rst motif (position ±40). Therefore, it is assumed that the second motif (position ±65) might be the actual promoter. This second motif (position ±65) is 22 bp upstream of the transcription initiation, resembling that described in Euplotes crassus actin (a potential TATAbox is situated 21 bp upstream of the transcription initiation, which is also an A [22]). Note that the 64 nucleotides upstream of the initiation codon are rather similar in H. cavicola,O. fallax, and O. trifallax actin sequences, and that a perfect alignment can be produced around the TATA box [42]. Related to the polyadenylation consensus sequence (AATAAA), it is uncommon to ®nd perfect matches in ciliate actin genes, but this is not surprising, as yeast and plants would not require this motif to add the poly(A) tract [50]. In fact, in E. crassus, where no perfect polyadenylation signal can be recognised in actin, two dierent polyadenylation sites have been described. These correspond to two dierent developmental stages, although whether polyadenylation addition to a particular site is related to a particular stage of development remains to be determined [22]. The expression level of actin has been examined in a few cases. In Tetrahymena, for instance, actin seems to be actively expressed, but with little ¯uctuation throughout the cell cycle [54]. In Sterkiella histriomuscorum, our previous results indicated that actin is also expressed actively, although to a lesser extent than a-tubulin. Moreover, actin is down-regulated during starvation (reaching a minimum in the cyst) and upregulated during excystment, probably until the level of the vegetative cell is reached. The primary amino acid sequence At this level, the primary amino acid sequences of ciliate actins shows a relatively low level of identity with those of actins deposited in databases. In some cases, ciliate actins are no more similar to each other than to those of other organisms. For example, Paramecium actin is approximately 71% and 61.3% identical to Arabidopsis thaliana and Oxytricha trifallax actins, respectively. In the same manner, Paramecium actin is more similar to actin of an early diverging protozoan (58.3% identity with Giardia lamblia actin) than to the actin of a member of the ciliates (56.5% identity with Euplotes crassus actin). This situation is accentuated in Oxytricha nova, which has three highly divergent actin genes (actin II and actin III are only 50% identical). These striking dierences among ciliate actin sequences show the high evolutionary rate of actin within the phylum Ciliophora. 169 As a consequence of this unusually rapid rate of evolution, ciliates appear to be polyphyletic, emerging quite low in the actin phylogenetic tree [13]. However, a wealth of morphological evidences and phylogenetic trees of rRNA and tubulins [1] unquestionably demonstrates that ciliates constitute a solid monophyletic group. Maybe actin, in contrast to tubulin, is poorly used in ciliates. If this is true, actin would have less functional constraints and, therefore, it could evolve at a high mutation rate. This assumption could explain, in part, why ciliate tubulins are so similar in their amino acid sequences, whereas ciliate actins are not. It could also explain the unusual characteristics of ciliate actin, which might have evolved new functions. The tertiary structure Since the X-ray structures of monomeric actin have become available [5, 29,48], it has been possible to determine precisely which amino acids are signi®cant for any given actin property, such as binding to divalent cation, to nucleotide, or to DNase I. Superimposition of ciliate actin sequences on the three-dimensional structure of rabbit actin has allowed us to analyse the degree of conservation of several ligand-binding sites in the actin of ciliates. Furthermore, we have analysed in more detail the structure of H. cavicola actin by in silico modelling. First, we aligned the actin amino acid sequences of several ciliates with that of rabbit a-actin and calculated the percentage of identity by subdomain (Table 1). The highest degree of conservation is found in subdomain 1, especially in the region of amino acids 70±144, and in the region of amino acids 145±180 of subdomain 3. These zones constitute the inner hydrophobic core of actin and they form the hydrophobic pocket involved in nucleotideand cation-binding. Accordingly, the nucleotidebinding sites are well-conserved motives in ciliate actins (Table 2). Subdomains 2 and 4, and the region of amino acids 270±337 of subdomain 3 are less conserved (Table 2), with the greatest divergence mainly located on the surface of the molecule (Fig. 1); therefore, the DNase loop (amino acids 40±50) of subdomain 2 and the peripheral regions of amino acids 194±203 and 223± 242 inside of subdomain 4 are not well-conserved (Table 2, Fig. 1). The DNase-I-binding sites of ciliate actins share 35.3±70.6% identity with the corresponding sequence of rabbit a-actin (Table 2). Since T. pyriformis actin, with 58.8% identity of the DNase-I-binding sites, lacks DNase-I-binding activity, it is likely that hypotrich, oligotrich and peritrich actins, with a DNasebinding motif less conserved than in Tetrahymena, also fail to bind DNase I. The actin-actin interactions in the micro®lament include a region that is almost identical to that associated with DNase I±actin contact, the DNase I loop (divergent in ciliate actins) being of primary importance for the stabilisation of the actin ®lament. Moreover, in rabbit a-actin, a loop (amino acids 262± 272) that includes a fourresidue hydrophobic plug inserted into a hydrophobic pocket formed by two adjacent monomers on the opposing strand is also crucial for stabilisation of the actin helix. In a conventional actin, such as that of yeast, a mutation (L 266 D) in the region of amino acids 262±272 produces disruptions in hydrophobic interactions, which results in the inhibition of actin polymerisation at low temperature [4]. The 262± 272 loop is not well-conserved in ciliate actins (Table 2). In this region, hypotrich and Halteria actins share only 27.3±45.5% identity with rabbit a-actin, and hydrophobic residues within the plug are replaced by hydrophilic ones. These modi®cations in ciliate actin Table 1 Percentages of identity between rabbit muscle actin (whole molecule and subdomains) and actins from dierent ciliates Whole molecule Subdomain 1 Subdomain 2 Subdomain 3 Subdomain 4 1a 1b 1c 3a 3b Amino acids 1±375 Amino acids 1±32 Amino acids 70±144 Amino acids 338±375 Amino acids 33±69 Amino acids 145±180 Amino acids 270±337 Amino acids 181±269 Hypotrichs Histriculus cavicola 65.5 72.4 85.3 67.5 48.6 69.4 44.1 65.2 Oxytricha nova I 65.8 64.5 85.3 67.5 45.9 69.4 52.9 64.0 Oxytricha nova II 67.2 71.9 75.7 85.0 58.3 80.0 56.1 58.0 Oxytricha fallax 66.2 73.3 69.3 67.5 45.9 69.4 50.0 65.2 Oxytricha trifallax 66.8 73.3 85.3 67.5 48.6 69.4 50.0 66.3 Euplotes crassus 61.9 59.4 65.3 67.5 54.1 72.2 60.3 58.4 Paramecium tetraurelia 73.6 78.9 83.1 83.7 70.2 83.3 73.1 61.3 Tetrahymena pyriformis 74.9 53.1 82.0 89.4 75.6 86.4 69.5 66.3 Peritrichs Vorticella microstoma 69.8 68.4 78.6 76.3 62.2 83.7 69.1 58.4 Opisthonecta matiensis 64.2 61.4 77.3 73.3 62.2 67.0 58.8 55.4 Olygotrichs Halteria sp. 66.2 68.7 81.3 65.7 48.6 81.3 48.5 67.4 170 sequences may indicate unusual polymerisation properties or an inability to polymerise. Second, we aligned the actin amino acid sequence of H. cavicola with that of rabbit a-muscle actin and then carried out in silico modelling, taking into account the tertiary structure of the complex actin±DNase I available from the Protein Data Bank (accession number 1ATN). These in silico structures have also served to calculate a molecular dynamics model and can be used to search for conformational dierences among them. The in silico structures of H. cavicola and rabbit a-muscle actin are similar, although small dierences can be seen in subdomains 2 and 4. These dierences can also be seen by comparing the average structures created with molecular dynamic simulations (Fig. 2), as well as by comparing the percentage of identity of these subdomains at the amino acid level (see the precedent paragraph). Moreover, during the simulation, subdomain 2 moved toward subdomain 4, the displacement being more pronounced in H. cavicola actin than in rabbit a-muscle actin. The tighter interaction between subdomains 2 and 4 in H. cavicola actin caused an allosteric hindrance that could inhibit DNase I binding. The actin-binding proteins As can be deduced from the preceding sections, knowledge on ciliate actin is still very scarce and ABPs are not an exception but rather an extreme case. In fact, data are Fig. 1 The three-dimensional structure of actin±DNaseI complex according to Kabsch et al. [30]. The ®rst and last amino acids residues in helices and sheet strands are speci®ed. Regions of subdomains 2 and 4 that are highly divergent in ciliate actins are shown in blue (amino acids 40±50, 194±203, 223±242 and 262±272) Table 2 Percentages of identities between nucleotide-binding sites, DNase-I-binding sites, DNase loop (amino acids 40±50), and amino acid 194±203, 223±242 and 262±272 loops of rabbit muscle actin and the corresponding regions in ciliate actin sequences Nucleotide binding DNase I binding DNase I loop 194±203 loop 223±242 loop 262±272 loop Histriculus cavicola 73.3 35.3 9.1 60 26.6 36.4 Oxytricha nova I 86.7 35.3 9.1 40 26.6 36.4 Oxytricha nova II 73.3 52.9 45.5 40 26.6 27.3 Oxytricha fallax 86.7 35.3 9.1 40 26.6 36.4 Oxytricha trifallax 93.3 35.3 9.1 40 26.6 36.4 Euplotes crassus 86.7 58.9 45.5 30 26.6 45.5 Paramecium tetraurelia 86.7 70.6 72.7 40 20.0 63.6 Tetrahymena pyriformis 80.0 58.9 63.6 30 26.6 54.4 Vorticella microstoma 86.7 41.2 45.5 30 20.0 45.5 Opisthonecta matiensis 86.7 47.0 45.5 30 13.3 54.5 Halteria sp. 86.7 35.3 9.1 60 20.0 27.3 171 restricted to Tetrahymena. It is generally believed that all eukaryotic cells contain actin and its ®lament-binding motor protein myosin. This is certainly true, but in ciliates the demonstration of myosin has been dicult, maybe because of the high divergence of ciliate myosin. A report on the presence of Tetrahymena myosin appeared in 1995 [21], i.e. more than a decade after the ®rst reports on Tetrahymena actin. The authors described the presence of two polypeptides of 180 and 15 kDa in the basal-body cage complex of Tetrahymena that could correspond, based on biochemical evidence, to the heavy and light chains of myosin. Furthermore, authors in the same laboratory described the cloning of a myosin heavy chain gene [20] which is indeed expressed in growing cells. The predicted amino acid sequence of this gene shows that all the signature motives for the head domain of known myosins are conserved. Nonetheless, a phylogenetic analysis shows that Tetrahymena myosin heavy chain belongs to a new myosin family. Moreover, disruption of this gene aects endocytosis and macronuclear elongation [53]. Pro®lin, an actin-sequestering protein that modulates actin polymerisation, has also been described in Tetrahymena. A conventional pro®lin of 12.8 kDa was isolated in T. pyriformis by poly (L-proline) anity column [16]. The same procedure was used to isolate pro®lin in T. thermophila, and the puri®ed protein was used to obtain its corresponding cDNA [52]. This cDNA sequence predicts a polypeptide of 16.7 kDa with little homology with the previously reported T. pyriformis pro®lin gene [15]. This last Tetrahymena pro®lin gene is also divergent when compared to mammalian pro®lins but its Nand C-terminal regions are relatively conserved. By immuno¯uorescence, pro®lin was detected in the division furrow of Tetrahymena [17]. Pro®lin seemed not to be the only ABP that co-localised with actin in the division furrow, since three other Tetrahymena co-localising proteins have been described: ®mbrim, elongation factor 1a(EF-1a) and calmodulin. Tetrahymena ®mbrin was isolated as a 61-kDa polypeptide that was shown to be the partial degradation product of a 71-kDa polypeptide. A cDNA corresponding to this ®mbrin was sequenced and predicted to be a protein of about 65.1 kDa with two actin-binding domains, but lacking the EF-hand-binding domain. This last feature and the low homology with other known ®mbrins suggests that this protein is a new member of the ®mbrin/plastin family. Tetrahymena ®mbrin co-localises with actin in dividing cells, but in interphase cells also at the oral apparatus and vacuole pores [51]. EF-1a is involved in protein synthesis in eukaryotes, although some reports show that it can perform other functions, usually related to cytoskeleton regulation. In fact, Tetrahymena EF-1aco-precipitates with F-actin and has F-actin bundling activity, as shown by electron microscopy [33]. Furthermore, it has been shown that Ca +2 / calmodulin directly interacts with EF-1ainhibiting its Factin bundling activity [32]. Both EF-1aand calmodulin have been detected in the oral apparatus and apical region of contractile vacuoles in interphase Tetrahymena cells [40], as has also been observed with ®mbrin. Summarising, the results described here show that the ciliate ABPs identi®ed thus far are somehow dierent from those of other eukaryotes. This is similar to the case of ciliate actin when compared to conventional actins, thus rearming the unconventional character of the ciliate actin cytoskeleton. Future directions It is clear that the function of ciliate actin is still unknown. We can assume that ciliate actin plays the same or a similar role that conventional actin does. This extrapolation, however, could be inexact, since the primary amino acid sequence and biochemical properties of ciliate actins dier from those of conventional actins. Indications on ciliate actin function come from the localisation of this protein in the cell. These data suggest that actin may be involved in some steps of phagocytosis, the contractility of the cortex and the division furrow. Nevertheless, implication of ciliate actin in these or other processes needs more experimental evidence, which can be obtained using current molecular approaches, such as transformation. This technique has been successfully applied to Tetrahymena,Paramecium, Stylonychia and Euplotes. Transformation would allow, for example, observation of actin dynamics in vivo through fusion with green ¯uorescent protein. It would Fig. 2 Computer-simulated three-dimensional structure of rabbit a-actin (A) and Histriculus cavicola actin (B). These representations correspond to the average structures of molecular dynamic simulations. White bars Distance between subdomains 2 and 4, which is shorter in H. cavicola actin than in rabbit a-actin 172 also be possible to use ``gene silencing'' to reduce actin expression, if this were not lethal for cells. This phenotype can be obtained whenever the role of actin becomes less essential for organelle movements, cell shape or chromosome rearrangements in ciliates than in other eukaryotes. Preliminary experiments developed in collaboration with A. Fleury (Universite ÂParis-XI, Orsay, France) have allowed us to obtain a non-lethal phenotype in actin-transformed Paramecium. However, as actin expression has not been analysed yet, we do not know whether the gene has actually been silenced. The role of ciliate actin as both regulator of cellular signalling pathways and co-ordinator of cellular behaviour should be exploited. Attention to ABPs, kinases, phosphatases, calmodulin, or small GTPases and to morphogenetic processes, such as conjugation, encystment/excystment, regeneration and polymorphic transformation, will no doubt lead to very promising lines of research. Acknowledgements This research was supported by DGICYT grant PB97±0710-C02±01. We thank M. A. de la Rosa and A. DõÂaz-Quintana for their help with modelling analysis. We also thank B. Pe Ârez Uz for critically reading the manuscript. References 1. Baroin Tourancheau A, Villalobo E, Tsao N, Torres A, Pearlman E (1998) Protein coding trees in ciliates: comparison with rRNA-based phylogenies. Mol Phylogenet Evol 10:299± 309 2. Bhattacharya D, Stickel S, Sogin M (1991) Molecular phylogenetic analysis of actin genic region from A. bisexualis and C. costata. J Mol Evol 33:525±536 3. Bork P, Sander C, Valencia A (1992) An ATPase domain common to prokaryotic cell cycle protein, sugar kinases, actin, and hsp70 heat shock proteins. Proc Natl Acad Sci USA 89:7290±7294 4. Chen X, Cook RK, Rubenstein PA (1993) Yeast actin with a mutation in the ``hydrophobic plug'' between subdomains 3 and 4 (L 266 D) displays a cold-sensitive polymerization defect. J Cell Biol 123:1185±1195 5. Chik JK, Lindberg U, Schutt CE (1996) The structure of an open state of b-actin at 2.65A resolution. J Mol Biol 263:607±623 6. Clark SW, Meyer DI (1992) Centractin is an actin homologue associated with the centrosome. Nature 359:246±250 7. Cleveland DW, Lopata MA, MacDonald RJ, Rutter WJ, Kirschner MW (1980) Number and evolutionary conservation of aand b-tubulin and band c-actin genes using speci®c cloned cDNA probes. Cell 20:95±105 8. Cohen J, Beisson J (1988) The cytoskeleton. In Go Èrtz H-D (ed) Paramecium. Springer, Berlin, Heidelberg New York, pp 363± 392 9. Cohen J, Garreau de Loubresse N, Beisson J (1984) Actin micro®laments in Paramecium: localization and role in intracellular movements. Cell Motil Cytoskeleton 4:443±468 10. Cupples CG, Pearlman RE (1986) Isolation and characterization of the actin gene from Tetrahymena thermophila. Proc Natl Acad Sci USA 83:5160±5164 11. Diaz-Ramos C, Villalobo E, Perez-Romero P, Torres A (1998) Paramecium tetraurelia encodes unconventional actin containing short introns. J Eukaryot Microbiol 45:507±511 12. Dizick SJ, Prescott DM (1999) Three macronuclear molecules encoding three highly divergent actin genes in Oxytricha nova. Eur J Protistol 35:375±377 13. Drouin G, Moniz de Sa ÂM, Zuker M (1995) The Giardia lamblia actin gene and the phylogeny of eukaryotes. J Mol Evol 41:841±849 14. DuBois M, Prescott DM (1995) Scrambling of the actin I gene in two Oxytricha species. Proc Natl Acad Sci USA 92:3888± 3892 15. Edamatsu M, Hirono M, Takemasa T, Watanabe Y (1991) The primary structure of Tetrahymena pro®lin. Biochem Biophys Res Commun 175:543±550 16. Edamatsu M, Hirono M, Watanabe Y (1990) Puri®cation and characterization of Tetrahymena pro®lin. Biochem Biophys Res Commun 170:957±962 17. Edamatsu M, Hirono M, Watanabe Y (1992) Tetrahymena pro®lin is localized in the division furrow. J Biochem 112:637± 642 18. Fahrni JF (1992) Actin in the ciliated protozoan Climacostomum virens: puri®cation by DNase I anity chromatography, electrophoretic characterization, and immunological analysis. Cell Motil Cytoskeleton 22:62±71 19. Gallwitz D, Sures I (1990) Structure of a split yeast gene: complete nucleotide sequence of the actin gene in Saccharomyces cerevisiae. Proc Natl Acad Sci USA 77:2546±2550 20. Garces J, Gavin RH (1998) A PCR screen identi®es a novel, unconventional myosin heavy chain gene (MYO1)inTetrahymena thermophila. J Eukaryot Microbiol 45:252±259 21. Garces JA, Hoey JG, Gavin RH (1995) Putative myosin heavy and light chains in Tetrahymena: co-localization to the basal body-cage complex and association of the heavy chain with skeletal muscle actin ®laments in vitro. J Cell Sci 108:869±881 22. Ghosh S, Jaraczewski JW, Klobutcher LA, Jahn CL (1994) Characterization of transcription initiation, translation initiation, and poly (A) addition sites in the gene-sized macronulcear DNA molecules of Euplotes. Nucleic Acid Res 22:214±221 23. Hackney CM, Butler RD (1981) Tentacle contraction in glycerinated Discophrya collini and the localization of HMMbinding ®laments. J Cell Sci 47:65±75 24. Harper DS, Jahn CL (1989) Actin, tubulin and H4 histone genes in three species of hypotrichous ciliated protozoa. Gene 75:93±107 25. Hauser M, Hausman K, Jockusch BM (1980) Demonstration of tubulin, actin and alpha-actinin by immuno¯uorescence in the microtubule-micro®lament complex of Pseudomicrothorax dubius. Exp Cell Res 125:265±274 26. Hirono M, Endoh H, Okada N, Numata O, Watanabe Y (1987) Cloning and sequencing of the Tetrahymena actin gene and identi®cation of its gene product. J Mol Biol 194:181±192 27. Hirono M, Kumagai Y, Numata O, Watanabe Y (1989) Puri- ®cation of Tetrahymena actin reveals some unusual properties. Proc Natl Acad Sci USA 86:75±79 28. Hoey JG, Gavin RH (1992) Localization of actin in the Tetrahymena basal body-cage complex. J Cell Sci 103:629±641 29. Kabsch W, Mannherz HG, Suck D, Pai EF, Holmes KC (1990) Atomic structure of the actin:DNase I complex. Nature 347:37± 44 30. Kaine BP, Spear BB (1980) Putative actin genes in the macronucleus of Oxytricha fallax. Proc Natl Acad Sci USA 77:5336±5340 31. Katsumaru H, Fukui Y (1982) In vivo identi®cation of Tetrahymena actin probed by DMSO induction of nuclear bundles. Exp Cell Res 137:353±363 32. Kurasawa Y, Hanyu K, Watanabe Y, Numata O (1996) Factin bundling activity of Tetrahymena elongation factor 1ais regulated by Ca +2 /calmodulin. J Biochem 119:791±798 33. Kurasawa Y, Watanabe Y, Numata O (1996) Characterization of F-actin bundling activity of Tetrahymena elongation factor 1ainvestigated with rabbit skeletal muscle actin. Zool Sci 13:371±375 34. Lin JJC (1981) Monoclonal antibodies against myo®brillar components of rat skeletal muscle decorate the intermediate ®laments of cultured cells. Proc Natl Acad Sci USA 78:2335± 2339 173 35. Maciver SK (1998) How ADF/co®lin depolymerizes actin ®laments. Curr Opin Cell Biol 10:140±144 36. Mackay DJ, Hall A (1998) Rho GTPases. J Biol Chem 273:20685±20688 37. McGough A (1998) F-actin-binding proteins. Curr Opin Struct Biol 8:166±176 38. Metenier G (1984) Actin in Tetrahymena paravorax: ultrastructural localization of HMM-binding ®laments in glycerinated cells. J Protozool 31:205±215 39. Mitchell EJ, Zimmerman AM (1985) Biochemical evidence for the presence of an actin protein in Tetrahymena pyriformis. J Cell Sci 73:279±297 40. Numata O, Kurasawa Y, Gonda K, Watanabe Y (2000) Tetrahymena elongation factor-1ais localized with calmodulin in the division furrow. J Biochem 127:51±56 41. Numata O, Watanabe Y (1982) In vitro assembly and disassembly of 14-nm ®lament from Tetrahymena pyriformis.The protein component of 14-nm ®lament is a 49,000-dalton protein. J Biochem 91:1563±1573 42. Perez-Romero P, Villalobo E, Diaz-Ramos C, Calvo P, Torres A (1999) Actin of Histriculus cavicola: Characteristic of the highly divergent hypotrich ciliate actins. J Eukaryot Microbiol 46:469±472 43. Piperno G, Luck D (1979) An actin-like protein is a component of axonemes from Chlamydomonas ¯agella. J Biol Chem 254:2187±2190 44. Pollard TD, Blanchoin L, Mullis RD (2000) Molecular mechanisms controlling actin ®lament dynamics in nonmuscle cells. Annu Rev Biophys Biomol Struct 29:545±76 45. Reisler E (1993) Actin molecular structure and function. Curr Opin Cell Biol 5:41±47 46. Sa Ânchez-Silva R, Ferna Ândez-Aliseda MC, Perez-Romero P, Calvo P, Torres A (2000) Actin of peritrich ciliates. J Eukaryot Microbiol (Abstracts from Section Meetings 2000) 47. Sandoz D, Gounon P, Karsenti E, Sauron E (1982) Immunocytochemical localization of tubulin, actin and myosin in axonemes of ciliated cells from quail oviduct. Proc Natl Acad Sci USA 79:3198±3202 48. Schutt CE, Myslik JC, Rozycki MD, Goonesekere NCW, Lindberg U (1993) The structure of crystalline pro®lin-b-actin. Nature 365:810±816 49. Tiggemann R, Plattner H (1981) Localization of actin in the cortex of Paramecium tetraurelia by immuno and anity ¯uorescence microscopy. Eur J Cell Biol 24:184±190 50. Wahle E, Keller W (1992) The biochemistry of 3¢-end cleavage and polyadenylation of messenger RNA precursors. Annu Rev Biochem 61:419±440 51. Watanabe A, Yonemura I, Gonda K, Numata O (2000) Cloning and sequencing of the gene for a Tetrahymena ®mbrim-like protein. J Biochem 127:85±94 52. Wilkes DE, Otto J (2000) Molecular cloning of pro®ling from Tetrahymena thermophila. Gene 246:295±301 53. Williams SA, Hosein RE, Garces JA, Gavin RH (2000) MYO1, a novel, unconventional myosin gene aects endocytosis and macronuclear elongation in Tetrahymena thermophila.J Eukaryot Microbiol 47:561±568 54. Zimmerman AM, Zimmerman S, Thomas J, Ginzburg I (1983) Control of tubulin and actin gene expression in Tetrahymena pyriformis during cell cycle. FEBS Lett 164:318±321 174