A broad spectrum of actin paralogs in Paramecium tetraurelia cells display differential localization and function

Actin, an abundant cytoskeletal protein, is of paramount importance in forming the cell cortex, amoeboid movement, cyclosis, vesicle trafficking, cell division and for cell contraction etc. In the past few years, new aspects have emerged. This includes a contribution to docking of dense-core secretory vesicles (Gasman et al., 2004), endocytosis (Merrifield et al., 2002; Merrifield, 2004; Kirkham and Parton, 2005; Yarar et al., 2005; Kaksonen et al., 2006), arrangement of Golgi elements (Lin et al., 2005), formation of Golgi-derived vesicles (Carreno et al., 2004; Cao et al., 2005) and numerous interactions with phagocytic as well as endocytic and lysosomal vesicles of different stages (Damiani and Colombo, 2003; Drengk et al., 2003; Stoorvogel et al., 2004; Yam and Theriot, 2004; Kjeken et al., 2004; Hölttä-Vuori et al., 2005). A universal function of F-actin is the formation of the cleavage furrow in animal cells (Otegui et al., 2005). A plethora of proteins interacting with actin further contribute to the dynamics and specificity of actin interactions (Pollard et al., 2000). Such interactions allow for a broad diversification of the actin cytoskeleton, even though some lower eukaryotes possess only one actin gene (García-Salcedo et al., 2004), whereas humans have six (Pollard, 2001). The closely related ciliate Tetrahymena thermophila possesses four actin genes (Williams et al., 2006). In addition, a range of actin-like proteins (ALP) and actin-related proteins (ARP) exist (Goodson and Hawse, 2002; Kandasamy et al., 2004; Muller et al., 2005). Apicomplexan parasites (Plasmodium, Toxoplasma), close relatives of our non-parasitic ciliates, mostly contain a single conventional actin and several ARPs, of which some are unique to apicomplexans (Gordon and Sibley, 2005).

The ciliated protozoan Paramecium is a highly organized unicellular organism with an elaborate ultrastructure (Fig. 1). We now can show a widely diversified set of genes encoding actin and ARP in Paramecium (Table 1) (I.M.S., J.M., C.R., E. Wagner, H.P. and R.K., unpublished). Indirectly the question as to why this multiplicity has evolved, probably by gene duplications (Ruiz et al., 1998; Aury et al., 2006), and maintained during evolution, is addressed in this paper in the context of localization and functional analyses. These reveal differential positioning and functional engagement, in established vesicle trafficking pathways and beyond. Microfilaments of actin might play a role in transport to the cell cortex and the recognition of the docking sites by trichocysts (Beisson and Rossignol, 1975).

We used overexpression as green fluorescent protein (GFP) fusion proteins, eventually complemented by antibody localization and silencing of the respective genes by RNAi. Because of the abundance of many isoforms (Table 1), not all could be analyzed in detail.

In most species, actin is sensitive to drugs that stabilize either G-actin (cytochalasin B or D, latrunculin A) or F-actin (phalloidin, jasplakinolide) (Wieland and Faulstich, 1978; Visegrády et al., 2004). However, in lower eukaryotes the situation is rather complex. For instance, Apicomplexan parasites (Plasmodium, Toxoplasma), close relatives of our non-parasitic ciliates, contain abundant actin, mainly in monomeric or short polymeric forms (Dobrowolski et al., 1997; Poupel et al., 2000; Schmitz et al., 2005; Sahoo et al., 2006). Also surprising is the inability of any previous work to visualize in Paramecium a cleavage furrow by fluorescent phalloidin (Kersken et al., 1986a) whereas phagocytosis could easily be inhibited by cytochalasins (Cohen et al., 1984; Allen and Fok, 1985; Fok et al., 1985; Allen et al., 1995).

We now show that the specific localization of different actin isoforms in Paramecium is paralleled by functional diversification. Analysis of Paramecium cells appears to be rewarding as they display a most elaborate membrane-trafficking system (Fig. 1), with distinct, predictable pathways (Allen and Fok, 1983; Allen and Fok, 2000; Fok and Allen, 1988; Fok and Allen, 1990; Plattner and Kissmehl, 2003) in which the different actin isoforms participate, as we now show. In the present work we combined the different techniques to establish a functional and topological overview of actin diversification in a Paramecium cell.

Based on a Paramecium genome data base search (I.M.S., J.M., C.R., E. Wagner, H.P. and R.K., unpublished), Table 1 summarizes the actin isoforms considered in the present work. Essentially we found nine subfamilies of actin genes (ten members of subfamily 1, two or three members in the other subfamilies) and two subfamilies of ARP (with one or two members each), with identities on the amino acid level ranging between 100% and 63% (for members of the actin1 subfamily) and 60% to 18% for the other subfamilies, when referred to Paramecium actin1 (Table 1). In comparison with actins in other organisms, actins in P. tetraurelia are highly divergent, which makes an accurate classification difficult. Owing to contradictory hits within blast searches in the NCBI database and alignments with ARPAnno [an actin-related protein annotation server designed by Muller et al. (Muller et al., 2005)], some P. tetraurelia subfamilies could be assigned to both actins and ARPs. For example, NCBI Blast searches with actin9 leads to hits for both actins and ARP10 of other organisms, whereas ARPAnno alignment resulted in an equally low score for ARP2 and orphans (sequences lacking common defining characteristics), with the score being below the value which was determined as reliable by the designers. As other sequences could be clearly assigned to ARP2 (data not shown), we rely on the NCBI Blast search results for the classification of our sequences. To investigate the subcellular localization of the different actin isoforms, GFP constructs were made. For the isoforms act1-2, act1-6, act1-9, act2-1, act3-1, act5-1, act6-1 and act8-1, the open reading frame was cloned into pPXV-eGFP (Hauser et al., 2000), the Paramecium overexpression vector (Haynes et al., 1995). Previously, a GFP-actin construct was used to transform T. thermophila cells, which led to a failure of nuclear elongation and cytokinesis (Hosein et al., 2003). In contrast to the construct used in that work, we fused the GFP gene to the 5′ end of the actin gene with an 11 amino acid spacer interspersed to avoid any possible disturbance of folding and polymerization because of the GFP tag (Doyle and Botstein, 1996; Wetzel et al., 2003). None of our constructs led to any cell abnormalities. The plasmids (∼5 μg/μl) were introduced into postautogamous Paramecium cells by microinjection into the macronucleus and GFP fluorescence was analyzed in descendants of the injected cells. A summary of the subcellular localization of the actin isoforms investigated is given in Table 2.

To take into account the wide diversification of the actin1 subfamily (Table 1), the subcellular localization of the isoforms act1-2, act1-4, act1-6 and act1-9 were examined in more detail. GFP-act1-2 formed comet tails at food vacuoles, which propelled them through the cytoplasm (Fig. 2A-C). Alternatively it might be accumulated in patches at the surface of food vacuoles (Fig. 2D-F). Transformation with GFP-act1-4 resulted in diffuse fluorescence throughout the cytoplasm (data not shown), as with GFP-act1-6 (Fig. 2G,H). With GFP-act1-9, pronounced comet tails could be observed, which propelled food vacuoles even in opposite directions through the cytoplasm, with the tail always at the rear side (Fig. 2I-N and supplementary material, Movie 1). These GFP-act1-9 comet tails could not be observed on every food vacuole because they were very dynamic structures, which have rapidly polymerized and depolymerized. Furthermore, thin, very dynamic tails were also seen to propel smaller vesicles. At the posterior end of the oral cavity, GFP-act1-9 filaments could be observed which were catapulted into the cytoplasm (see supplementary material Movie 1).

Transformation of Paramecium cells with GFP-act3-1 resulted in a diffuse staining of the cytoplasm, although a specific labeling also could be observed (data not shown). A GFP signal was found at the surface of food vacuoles, but in this case the vacuoles were almost completely surrounded by a layer of GFP-act3-1 of variable thickness. To enhance the GFP fluorescence, anti-GFP antibodies were applied to transformed cells. In the cell cortex, GFP-act3-1 outlined the ridges of the egg-carton-like cell surface relief typical of Paramecium cells, as well as cilia (Fig. 3).

Localization of GFP-act4 was not possible because cells did not divide after microinjection and were too sensitive for observation under the microscope. Therefore, we raised a polyclonal antibody against a 160-residue peptide of act4-1 to obtain subfamily-specific antibodies. The region chosen has less than 25% identity to other subfamilies, which makes it unlikely that the antibodies will react with any of the other actins or ARPs. The antibodies were affinity purified not only against the peptide for immunization, but also against peptides used to raise antibodies against act1-1 (Kissmehl et al., 2004) and against act5-1, and were tested in western blots for their crossreactivity, which was negative (Fig. 4A). Immunolocalization studies with anti-act4antibodies showed labeling of the oral cavity, around nascent food vacuoles (Fig. 4B), in cilia and in the cell cortex (Fig. 4C). In dividing cells, a weak labeling of the micronucleus separation spindles could be observed (data not shown). The most striking labeling was that of the cleavage furrow, with which act4 was associated from the beginning of an early stage of division until fission was completed (Fig. 4D,F). For comparison, act5-1-specific antibodies did not label the cleavage furrow (Fig. 4G).

Transformation of Paramecium cells with GFP-act5-1 resulted in a discrete labeling around a sub-set of food vacuoles. In contrast to the labeling observed with other GFP constructs at food vacuoles, GFP-act5-1 formed neither patches nor tails but a fine ring on the surface of the respective food vacuoles. Indeed, there was no consistent covering of the whole food vacuole (arrowhead in Fig. 5A), and the label could not be found around all food vacuoles (Fig. 5A,B). At the oral cavity, the following structures were labeled: (1) Two half-moon-shaped labeled structures assigned to the peniculus (Fig. 5B); (2) a dynamic fiber system emanating into the cytoplasm (arrow in Fig. 5B) anchored by a wreath-like structure with pointed pattern, probably the quadrulus of the oral cavity. (Note that the cell shown in Fig. 5A contains numerous autofluorescent crystals.) Immunolocalization studies with affinity-purified antibodies against act5-1 showed similar localization to that observed with GFP-tagged act5-1 (Fig. 4G), thus excluding possible artifacts resulting from GFP overexpression or fixation/permeabilization during preparation for immunolocalization studies.

The fluorescent signals obtained with GFP-act8-1 were in many cases similar to those obtained with GFP-act5-1. Labeling around single food vacuoles could be observed, but in contrast to the incomplete envelopment with GFP-act5-1, the food vacuoles seem to be entirely covered (Fig. 6A,C). The two half-moon-shaped structures at the presumed peniculus and the curved structure of the presumed quadrulus are also labeled, and a dynamic fiber system that originates from the latter could be observed (Fig. 6A-C, supplementary material Movie 2 and Fig. S1). Careful comparison of cells transformed with GFP-act5-1 with cells transformed with GFP-act8-1 showed that one probably deals with two different fiber systems emanating from the oral cavity. Beyond that, focusing on the surface of transformed cells revealed regular rows of small dots underneath the plasma membrane, arranged ∼2 μm apart (Fig. 6D), which can easily be identified as `parasomal sacs'. They correspond to stationary sites of constitutive exocytosis and coated pit endocytosis (Allen and Fok, 2000; Plattner and Kissmehl, 2003).

Furthermore, small point-like to rod-shaped organelles could be observed (arrow in Fig. 6B) which, particularly in z-stacks (data not shown), showed some enrichment towards the cell surface. These organelles were smaller than 1.0 μm and did not move with the cyclosis stream. In electron microscope immunoanalysis with cells overexpressing GFP-act8, using anti-GFP antibodies followed by protein A coupled to 5 nm gold particles (pA-Au_5nm), we frequently found labeling at the periphery of the rough endoplasmic reticulum (ER) in association with smooth vesicular membrane-bounded elements (Fig. 6E,F). These may represent Golgi elements with some deformation, possibly as a result of the overexpression of actin (García-Salcedo et al., 2004). The fixation used preserves F-actin, its position in filaments and its antigenicity as shown in the literature.

GFP localization of act2-1 and act6-1 resulted in a diffuse staining all over the cytoplasm, with no specific highlighted structure. This is similar to the results obtained with GFP-act1-6. However, enhanced with anti-GFP antibodies, GFP-act2-1 is found in cilia (data not shown). The different subcellular localizations achieved by a GFP tag of specific isoforms or subfamily-specific antibodies are summarized in Table 2 and, with the inclusion of previous data, in the final schematic Fig. 11.

For functional analysis of the different actin subfamilies, gene-silencing experiments were performed using the RNAi approach by feeding. Paramecium cells were fed with E. coli expressing double-stranded RNA coding for a specific actin, which elicits a small interfering RNA-mediated silencing process of the target genes. As a negative control, the empty feeding vector (pPD) was used for mock silencing. To ascertain the reliability of the RNAi-based feeding procedure, we performed controls with the pPD-nd7 construct (Skouri and Cohen, 1997). Silencing of the nd7 ('nondischarge') gene successfully suppressed exocytosis of trichocysts. The effects of gene silencing on various behavioral and physiological aspects were examined. In detail, cell shape, cyclosis, exocytotic capacity, the contraction period of the contractile vacuoles, phagocytotic capacity, the general swimming behavior and the swimming reaction upon depolarization and hyperpolarization were analyzed. For each actin subfamily, one gene was cloned into the silencing vector. Within the respective subfamilies, the nucleotide identity is high enough to expect co-silencing (Ruiz et al., 1998). An exception is the large actin1 subfamily with ten members that vary up to 35%. Therefore, constructs for the specific isoforms act1-2, act1-6 and act1-9 were made. With none of these constructs could we observe any behavioral or functional effect (Table 3). Potentially, several genes of subfamily 1 are functionally redundant, as we could observe similar localization patterns (GFP-act1-2 similar to GFP-act1-9, and GFP-act1-6 to GFP-act1-4).

To confirm co-silencing within the subfamilies, and to exclude any functional compensation by the unsilenced paired isoform, we tested a construct where both isoforms of the actin3 subfamily, act3-1 and act3-2, were cloned in tandem. RNAi with this construct gave identical results compared with those obtained with the isoform act3-1 alone (Table 3). To exclude silencing variation resulting from the length of the introduced dsRNA or cross-silencing of other subfamilies, constructs of different lengths were designed for act2-1 and act3-1. Within both subfamilies, all constructs showed the same effects. A summary of the results obtained with the different gene silencing constructs is given in Table 3.

The number of cell fissions per day was calculated to see if silencing of any of the actin subfamilies affects the growth rate of Paramecium. In general, a slightly reduced division rate could be observed 24 hours after the start of the feeding experiment, probably because of slower division of cells recovering from autogamy and the change of the bacteria used for feeding. Cell growth was not influenced by most of the silencing constructs (Fig. 7). Only silencing of act4 or act9 resulted in a significant reduction of the division rate. Silencing of act9 led to half of the division rate, whereas cells silenced in act4 could not divide anymore and subsequently died.

We could recognize changes in the general cell shape after silencing of act4, act7 or act9. Act4-silenced cells were arrested during division and, according to their shape, named `boomerang' (Fig. 8A). Cells silenced in act7 were generally enlarged with a pointed anterior end, called `dolphin' (Fig. 8B). Paramecium cells fed with the act9 silencing construct obtained a triangular cell shape (Fig. 8C).

The general swimming behavior of Paramecium cells was also analyzed. Cells silenced in act2 or act3 showed similar phenotypes. Swimming was impaired as cells were stagnant and accumulated at the bottom of the wells. Cells silenced in act4 showing the boomerang phenotype and cells silenced in act9 with a triangular shape were swimming in narrow circles, probably because of their morphological changes. Other silencing constructs had no effect on the swimming behavior. The swimming reaction upon depolarization or hyperpolarization of the plasma membrane - pronounced backward swimming in the first and accelerated forward swimming in the latter case - was not impaired in any of the silencing assays (Table 3).

We considered that cytochalasin B treatment abolished trichocyst docking in Paramecium (Beisson and Rossignol, 1975) and the occurrence of anti-actin antibody labeling around trichocysts docked at the plasma membrane (Kissmehl et al., 2004). With ongoing cell divisions, every daughter cell has to produce a new arsenal of trichocysts and to transport them to the cell cortex. If any of the actin subfamilies play an essential role within this process, after 96 hours of feeding (approximately 16 fissions) a reduced number of docked trichocysts would be expected. None of the silencing constructs led to a reduction in exocytosis, except `boomerang' cells (silenced in act4) where the exocytotic capacity was reduced to 10% (Table 3).

To analyze the formation and fission of food vacuoles and the possible role of specific actin subfamilies in the process of phagocytosis, Paramecium cells were fed with Congo-Red-stained yeast to visualize newly formed food vacuoles. Cells silenced in act2, act3, act4, act7 or act9 were found to have a significantly reduced number of yeast cells taken up, compared with control cells (Fig. 9). Although silencing of act7 reduced the phagocytotic capacity of the cells to ∼70%, silencing of act2 or act3 reduced phagocytosis to 30%. With cells silenced in act4, phagocytosis was reduced to 30% and to 70% with cells silenced in act9, provided cells had retained normal morphology. In contrast, triangle cells (act9) had a phagocytotic capacity of only 10%, whereas in boomerang cells (act4) no new food vacuoles were generated at all (Fig. 9, Table 3).

In cells silenced in act9 with a triangle shape, the contractile vacuole complex was severely perturbed. The filling and expelling cycle was increased from a value of 10-15 seconds in control cells to more than 100 seconds (Fig. 10). It is noteworthy that no method has demonstrated the occurrence of any actin isoform in the osmoregulatory system. Therefore, any actin silencing effects are probably due to indirect effects outside the system itself. No other subfamily silencing impaired the contractile vacuole complex (Table 3). In triangular cells, the velocity of cyclosis was accelerated two to three times when compared with control cells (Table 3) - an aspect that is difficult to explain without more detailed studies being performed.

A variety of actin isoforms are expressed in Paramecium, frequently with a specific localization and concomitantly with a specific function. Many of the aspects we describe have remained undetected in previous affinity labeling studies with Paramecium. Some of our results are without precedent, also in other cells. The localization of isoforms from different families can overlap. Particularly, the interaction of actin isoforms with phagosomes is very complex. Some other isoforms have a distinct localization, e.g. in the cleavage furrow, in cilia or in some other prominent cytoskeletal aggregates. It should be stressed, however, that we do not claim that any of the prominent filamentous structures of Paramecium would be exclusively or predominantly made up of actin. However, intermingling with actin is a feature known from different cytoskeletal components in other cells (Manneville et al., 2003; Lin et al., 2005).

Previously we localized actin in Paramecium by using fluorescently tagged DNase, heavy meromyosin, phalloidin and anti-actin antibodies (Tiggemann and Plattner, 1981; Kersken et al., 1986a; Kersken et al., 1986b) and, more specifically, using anti-actin1-1 antibodies (Kissmehl et al., 2004). Altogether this resulted in labeling of cilia, basal bodies, oral cavity, surface of food vacuoles, cytoproct, infraciliary lattice, cell surface ridges, the complex formed by the cortical calcium stores (alveolar sacs) and the cell membrane, surroundings of trichocyst docking sites, between bundled microtubules and ⩽1.0 μm-sized vesicles.

Although these studies anticipated part of our current results, the abundance of actin paralogs found in the context of the Paramecium genome project suggested a more complicated pattern. By overexpression as GFP-fusion proteins and with subfamily-specific antibodies, a variety of new and detailed features for specific isoforms emerged.

The role of actin isoforms with predominantly or exclusively diffuse cytosolic localization (Table 2) is difficult to appreciate. In Apicomplexans the cause of the large percentage of G-actin is still unresolved, but physical properties unique to apicomplexan actins, as well as interaction with monomer-binding proteins, might play a role (Baum et al., 2006). Remarkably, maintenance of cytosolic actin is vital for Drosophila although the reasons for this remain unknown (Wagner et al., 2002).

Occurrence of actin in cilia has been ascertained already (Tiggemann and Plattner, 1981; Kissmehl et al., 2004). We now show specifically the presence of actin2-1, actin3-1, actin4-1 and actin5-1 in cilia. Actin might serve positioning of the inner dynein arms just as in Chlamydomonas flagella (Hayashi et al., 2001; Yanagisawa and Kamiya, 2001).

Swimming behavior is affected by silencing genes for actin2, actin3, actin4 and actin9, all but actin9 certainly occurring in cilia (Table 3). With actin4 and actin9, this could be due to a change in cell shape. Since phagocytosis is reduced, also after silencing of these genes, reduced food supply could compromise cells in their swimming activity. However, there was no effect on the swimming behavior upon de- or hyperpolarization.

GFP-actin3-1 outlines the ridges of the egg-box-shaped cell surface. A similar staining pattern is seen with anti-actin4-1 and anti-actin5-1 antibodies. This pattern resembles that of the outer lattice, a fibrous network sustaining the plasma membrane and delineating each cortical unit (Allen, 1971; Cohen and Beisson, 1988; Iftode et al., 1989). GFP-actin8-1 is localized, among others, to parasomal sacs (Fig. 6D), which are known to represent coated pits (Allen, 1988). Our identification relies on the small size of labeled dots and their very regular arrangement, both largely excluding any other surface structures. This corresponds to the complex role of actin in endosome formation at the cell membrane, as established for higher eukaryotes (Merrifield et al., 2002; Merrifield, 2004; Kirkham and Parton, 2005; Yarar et al., 2005).

Although actin was known to occur around the oral cavity of Paramecium (Cohen et al., 1984; Kersken et al., 1986b) we now can specify this for actin5 and actin8. They are associated with the bona fide peniculus and quadrulus structures, as defined by Allen (Allen, 1988), but also beyond, and as part of the oral cavity lining (Kissmehl et al., 2004) and the oral filament system. In this context the mobility of these fibers must be emphasized (see supplementary material Movie 2), which could be construed as an auxiliary mechanism to guide the processing food vacuole when leaving the buccal cavity. The identification of actin8-1 as part of the endosomal system, as described above, corresponds to the labeling of the quadrulus and peniculus. These oral elements contain a large number of parasomal sacs, arranged in several rows (Allen et al., 1992).

Involvement of F-actin in phagosome formation in general (Kjeken et al., 2004) and in Paramecium in particular has been repeatedly documented (see above). Beyond that, several functions in the lifespan of a phagosome may require specific actins.

The surface of some phagosomes displays a speckled appearance of GFP-actin. This can reflect a functional mosaic pattern whereby some vesicles fuse and others pinch off during circulation through the cell (Allen and Fok, 1983; Allen and Fok, 2000; Fok et al., 1985; Allen et al., 1995). For instance, vacuole fusion in yeast requires F-actin disassembly (Wang et al., 2003).

Fig. 9 shows inhibition of phagocytosis by silencing the actin genes of subfamily 2, 3, 4 or 9, particularly when the resulting cells are of aberrant morphology (Table 3). Since aberrant cells are also impaired in their swimming behavior, with the cells moving in small circles, reduced phagocytosis may be due to the effects of reduction of ciliary activity and further processing of phagosomes on engulfing bacteria. Remarkably, the actin isoforms investigated are localized to cilia by GFP or antibody labeling. However, indirect effects of an unknown kind might also play a role.

Cyclosis is an established actin-dependent process (Shimmen and Yokota, 2004) also occurring with phagosomes in Paramecium (Sikora et al., 1979). In Tetrahymena, an unconventional myosin is required for directed phagosome transport (Hosein et al., 2005), but no unilateral actin arrangement has been seen. By contrast, in Paramecium we surprisingly see GFP-actin isoforms (actin1-2 and actin1-9), attached unilaterally to the lee side, as if pushing phagosomes. This is reminiscent of the actin-based propulsion of Listeria and related pathogenic bacteria (Tilney and Portnoy, 1989).

Although the actin tails are short and not seen with immunostaining using anti-actin1-1 antibodies, their occurrence during overexpression may indicate a way of vesicle propulsion during cyclosis.

GFP-actin8-1 also stains rod-shaped, ⩽1.0-μm-long organelles scattered throughout the cytoplasm, with some enrichment towards the cell surface (Fig. 5B). According to size, form and position, as well as their number these structures may correspond to the numerous Golgi elements found in Paramecium (Estève, 1972). In immunogold EM analysis, label was enriched at the boundary of the rough ER (Fig. 6E,F), but overexpression might have distorted Golgi stacks, as observed in Trypanosoma cells (García-Salcedo et al., 2004), so they could not be identified unequivocally. The labeling resembles that achieved with the GFP fusion protein of the a8 subunit of the V-ATPase in Paramecium, which was shown to be associated with the Golgi complex (Wassmer et al., 2006).

Our observation that the micronuclear cytospindle contains actin4 according to antibody labeling is compatible with a role of actin in nuclear positioning (Starr and Han, 2003). The notorious failure to demonstrate F-actin in the cleavage furrow of Paramecium (Kersken et al., 1986a) is apparently due to the absence of phalloidin-binding motifs in the actin4 that forms this structure (our unpublished results). Using specific antibody labeling we were successful in visualizing this structure (Fig. 4B-F). Concomitantly, silencing of the actin4 genes inhibits cell division (Fig. 7A). In T. thermophila, silencing of the act1 gene impairs phagocytosis, motility and cell separation (Williams et al., 2006). The failure of twisting apart was followed by a progressive re-integration of the two daughter cells. Observing numerous single cells, we could not find this phenomenon in P. tetraurelia silenced in actin4. We found cells arrested in early `post-fission' stages, which neither grew nor divided anymore, and cells arrested during cell division. Note that the silencing effect may occur during different cell-cycle phases because P. tetraurelia cultures are not amenable to synchronization.

Localization of cortical F-actin and its effects can be intriguing. On the EM level, anti-actin1-1 antibodies labeled the narrow space between the plasma membrane and alveolar sacs (Kissmehl et al., 2004). In related Apicomplexans equivalent structures are also connected by short F-actin strands (Jewett and Sibley, 2003). A similar structure/function effect is discussed for mammalian cells whose cortical Ca²⁺ stores may be positioned by F-actin (Rosado and Sage, 2000; Wang et al., 2002; Roderick and Bootman, 2003; Turvey et al., 2005). Although we could not show such an effect in response to cytochalasin in Paramecium (Mohamed et al., 2003), this may be due to the involvement of drug-insensitive isoforms. Any disturbance, e.g. by gene silencing, may entail the considerable effects on cell form and performance, as we describe.

Some of the Paramecium actins may contribute to cell shaping, as seen by changes following silencing of actin7 or actin9 genes (Fig. 8B,C). Among them, only silencing of actin9 affected the division rate (Fig. 7).

In addition, silencing of actin9 considerably increased the time elapsed between expulsions of the contractile vacuole (Fig. 10). Since no actin isoform has been seen associated with the osmoregulatory system so far, the effects may be very indirect. One hypothetical possibility would be the interaction of actin with transport systems or ion channels (Fukatsu et al., 2004), which then might affect the performance of the contractile vacuole.

In the present work we show a specific localization and functional diversification of a plethora of actin isoforms in P. tetraurelia. The highly regular pattern of a Paramecium cell, with a membrane-trafficking system with distinct, predictable pathways, provides a worthwhile and interesting system to study the mechanisms of intracellular vesicle trafficking with the involvement of a highly differentiated actin multigene family.

Wild-type Paramecium cells derived from stock 51s were used, i.e. d4-2 cells for microinjection of GFP constructs and 7S cells for additional experiments. Cells were cultivated at 25°C in a decoction of dried lettuce inoculated with Enterobacter aerogenes, supplemented with 0.4 μg/ml β-sitosterol (Sonneborn, 1970).

For heterologous expression of peptides of actin4-1 and 5-1, we selected regions of the proteins with less than 25% identity to other subfamilies to obtain subfamily-specific antibodies. Within these subfamilies, the identities at these regions were more than 90% at the amino acid level (I.M.S., J.M., C.R., E. Wagner, H.P. and R.K., unpublished), ensuring that each member of one subfamily will be recognized by the antibodies. After changing all deviant Paramecium glutamine codons (TAA and TAG) into universal Glutamine codons (CAA and CAG) by PCR, the coding regions of either M1-A160 (actin 4-1) or H175-L290 (actin 5-1) were cloned into the NcoI-XhoI restriction sites of pRV11a vector, which contains a His₆ tag for purification of the recombinant peptides.

Recombinant actin4-1 and actin 5-1 peptides were purified by affinity chromatography on Ni²⁺-nitrilotriacetate agarose under denaturating conditions, as recommended by the manufacturer (Novagen, Darmstadt, Germany). The recombinant peptides were eluted with a pH gradient, pH 8 to pH 4.5, containing 8 M urea in 100 mM sodium phosphate buffer. The fractions collected were analyzed on SDS polyacrylamide gels, and those containing the recombinant peptide were pooled and dialyzed in phosphate-buffered saline (PBS).

Antibodies against the two recombinant actin peptides act4-1 and act5-1 were raised in rabbits. After several boosts, positive sera were taken at day 60 and affinity purified on a column loaded with the corresponding actin peptide. Finally, both sera were crosspurified against each other.

Protein samples were boiled and subjected to electrophoresis on 10% SDS polyacrylamide gels as previously described (Kissmehl et al., 2004). Gels were stained with Coomassie Blue R250 or prepared for electrophoretic protein transfer onto nitrocellulose membranes. Protein blotting was performed at 1 mA/cm² for 1 hour using the semidry blotter from Bio-Rad (Munich, Germany). Antibodies were diluted 1:1000 in 0.5% (w/v) non-fat dry milk and Tris-buffered saline, pH 7.5, and applied overnight at 4°C. Antibody binding was visualized by a second antibody coupled to peroxidase using an enhanced chemiluminescence detection (ECL) kit according to the manufacturer (Amersham Biosciences, Freiburg, Germany).

Cells were washed twice in 5 mM PIPES buffer, pH 7.0, containing 1 mM KCl and 1 mM CaCl₂, fixed in 4% (w/v) formaldehyde for 20 minutes at room temperature (RT) and then permeabilized and fixed in a mixture of 0.5% digitonin and 4% formaldehyde, dissolved in 5 mM PIPES buffer, pH 7.0, for 30 minutes. Cells were washed twice in PBS, twice for 10 minutes in PBS with 50 mM glycine added and 10 minutes in this solution with 1% bovine serum albumin (BSA) added. The rabbit anti-actin antibodies were applied at a dilution of 1:50 in PBS (+1% BSA) for 90 minutes at RT. After four 15-minute washes in PBS, FITC-conjugated anti-rabbit antibodies (Sigma-Aldrich, St Louis, MO), diluted 1:100 in PBS (+1% BSA), were applied for 90 minutes, followed by four 15-minute washes in PBS. Samples were shaken gently during all incubation and washing steps. Cells were mounted on coverslips with Mowiol supplemented with n-propylgallate to reduce fading and analyzed either in conventional Axiovert 100TV or 200M fluorescence microscopes or in a confocal laser-scanning microscope (CLSM 510) equipped with a Plan-Apochromat objective lens (all Carl Zeiss, Jena, Germany). Images acquired with the ProgRes C10plus camera and ProgRes Capture Basic software (Jenoptik, Jena, Germany) were processed with Photoshop software (Adobe Systems, San Jose, CA).

For overexpression of actin, the pPXV-eGFP vector or pPXV-eGFPmcs vector (Wassmer et al., 2006) were used (eGFP, enhanced green fluorescent protein) (Hauser et al., 2000). All genes were cloned between SpeI and XhoI sites in-frame with the gene encoding GFP, except act8-1 where StuI was used instead of SpeI because of an internal SpeI restriction site. PCR primers used were as follows: act1-2, 5′aA2, 5′-GCTCTAGAGAGGAACAGCATCAGCA-3′ and 3′aAX, 5′-CCGCTCGAGTCAGAAACACTTTCTGTGAACAATGG-3′; act1-4, 5′aA4 5′-GCTCTAGAGAGGAACAGCATCAGCAGGCCTATGAGTGAAGAACATCCAGC-3′ and 3′aA4X, 5′-CGGCTCGAGTCAGAAGCATTTTCTATGAACC-3′; act1-6, 5′SpeI-Act1.6, 5′-GCGACTAGTATGTAAGCTTAATATCCAGC-3′ and 3′Xho-Act1.6END, 5′-CCGCTCGAGTCAGAAACATTTTCTGTGAAC-3′; act1-9, 5′Spe-Act1-9, 5′-GGACTAGTATGAATGATGAAAAACCAGCAGTCG-3′ and 3′Xho-Act1-9, 5′-CCGCTCGAGTCAAGTGACTGTCTAACATTTTCTGTG-3′; act2-1, 5′bA1, 5′GCTCTAGAGAGGAACAGCATCAGCAGGCCTATGGACGACGTAATCCCAGTTG-3′ and 3′bAX, 5′-CCGCTCGAGTCAGAAGCATTTTCTGTGCACATAACC-3′; act4-1, 5′Spe-Act4-1,5′GGACTAGTATGAATAGCGATGAAATAATA-3′ and 3′Xho-Act4-1, 5′CCGCTCGAGTCAATTAGGACACTTTCTTTCTA-3′; act5-1, 5′SpeI-Act5.1, 5′GCGACTAGTATGGATAATGACATATTTGCTAATAACT-3′, 3′Xho-Act5.1, 5′CCGCTCGAGTCACAATTATTTTTTGATTAAAATG-3′; act6-1, 5′SpeI-Act6.1, 5′GCGACTAGTATGGAAAGTGAGTATGACTAAAAAG-3′, 3′Xho-Act6.1, 5′-CCGCTCGAAGTCAAAATGTTCTCTTATGAATAAG-3′; act8-1, 5′Stu-Act8.1, 5′-GAAGGCCTATGAATAATAATGATTCCACCTTCTATTA-3′, 3′Xho-Act8.1, 5′-CCGCTCGAGTCAAAAGCACTTTCTTTATACTA-3′.

PCR reactions and cloning were carried out according to standard procedures. To avoid possible disturbance of (de-)polymerization and localization, GFP was cloned at the N-terminus of the gene of interest, separated by an 11-amino-acid spacer (Doyle and Botstein, 1996; Verkhusha et al., 1999; Wetzel et al., 2003).

Plasmid DNA was prepared with a plasmid midi kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. 50 μg of plasmid DNA was linearized by digestion with 20 units of SfiI overnight at 50°C. The DNA was precipitated with 1/10 (v/v) 3 M sodium acetate pH 5.2, with 2.5 (v/v) ethanol added, and incubated for 60 minutes at -20°C. DNA was pelleted by centrifugation, washed with 70% (v/v) ethanol and dried. The pellet was resuspended in 10 μl Millipore-filtered water and centrifuged for 30 minutes at 4°C. Microinjection of the DNA into the macronucleus was carried out as previously described (Froissard et al., 2002).

Cells derived from clones transformed with GFP-act8-1 were fixed for 1 hour on ice with 8% formaldehyde + 0.1% glutaraldehyde in PBS, pH 7.4, followed by two washes in PBS, pH 7.4 at RT. Cells were dehydrated in an ethanol series followed by embedding in LR-Gold resin (London Resin, London, GB). Ultrathin sections were decorated with affinity-purified, polyclonal anti-GFP antibody (Wassmer et al., 2006), followed by protein A-gold (5 nm) conjugates (Dept. Cell Biol., Univ. Utrecht, Utrecht, NL). These and any further steps for electron microscope immuno-analysis were essentially as described (Kissmehl et al., 2004).

The double T7-promotor plasmid pPD129.36 described by Fraser et al. (Fraser et al., 2000) was used. The following primers were used.

act1-2, 5′aA2Xba1, 5′-GCTCTAGAGAGGAACAGCATCAGCA-3′, 3′aAX, 5′-CCGCTCGAGTCAGAAACACTTTCTGTGAACAATGG-3′; act1-6, 5′Xba-Act1.6, 5′-GCTCTAGAAAATAGGTATAGCGGGAGATGAT-3′ and 3′Xho-Act1.6, 5′-CCGCTCGAGCAAAAACAGGTACGCAATGAG-3′; act1-9, 5′Xba-Act1-9, 5′-GCTCTAGAATGAATGATGAAAAACCAGCAGTCG-3′ and 3′Xho-Act1-9, 5′-CCGCTCGAGTCAAGTGACTGTCTAACATTTTCTGTG-3′; act2-1, 5′Xba-Act2-1, 5′-GCTCTAGAAAGTAGCCTGGTCAAATGG-3′ and 3′Xho-Act2-1, 5′-CCGCTCGAGCATATCATCCCAACGAGTG-3′; act3-1, 5′Xba-Act3-1, 5′-GCTCTAGATACAGGATTCTAAATAAAACAA-3′, 3′Spe-Act3-1, 5′-GGACTAGTCTATTCCATAGCCTCCC-3′, 5′Stu-Act3-1, 5′-GAAGGCCTATGATAGAATCTCATCCTCCTGTTG-3′, 3′Xba-Act3-1, 5′GCTCTAGATCAAAAACATTTAATGTGAGCAATC-3′; act3-2, 5′Spe-Act3-2, 5′-GGACTAGTTACAAGAATCTTAAACGATTAA-3′, 3′Xho-Act3-2, 5′-CCGCTCGAGCTATCTCTATAGATTCCC-3′; act4-1, 5′Xba-Act4-1(518-535), 5′GCTCTAGAAAGCGCCAATCGGAGGAG-3′, 3′Xho-Act4-1(885-903), 5′-CCGCTCGAGTGGTGCCAAAGCAGACAAG-3′, act5-1, 5′Xba-Act5-1, 5′-GCTCTAGATATTGACTGAACCTCCTTATG-3′, 3′Xho-Act5-1; 5′-CCGCTCGAGGTACCTTGCTCTTCTCAAC-3′; act6-1, 5′Xba-Act6.1, 5′-GCTCTAGACTGCTGTTTTAAATAAGTCTG-3′ and 3′Xho-Act6.1, 5′-CCGCTCGAGTCAAAATGTTCTCTTATGAATAAG-3′; act7-1, 5′Xba-Act7-1, 5′-GCTCTAGAGGCTTACGAATTACCAGAC-3′ and 3′Xho-Act7-1, 5′-CCGCTCGAGAGCTCCCAACCAAGATGC-3′; act8-1, 5′Xba-Act8-1, 5′-GCTCTAGATTTCCAGTGGAAAAACAACAG-3′ and 3′Xho-Act8-1, 5′-CCGCTCGAGACCATCGGGCAAATCATACA-3′; act9-1, 5′Xba-Act9-1, 5′-GCTCTAGATTGGCAATGTACTTCCTC-3′ and 3′Xho-Act9-1, 5′-CCGCTCGAGTTCCAAAATATGTGTCAGTG-3′. PCR reactions and cloning were carried out according to standard procedures.

The RNaseIII-deficient E. coli strain HT115 (Timmons et al., 2001) was transformed with the gene-silencing plasmids. Overnight cultures in Luria-Broth (LB) medium supplemented with ampicillin (amp) and tetracycline were diluted with LB/amp medium 1:100 and the new cultures were grown to an OD_600nm between 0.2 and 0.4. The cultures were induced with 125 μg/ml isopropyl-thio-β-D-galactopyranoside (IPTG) for 3 hours, centrifuged and the pelleted bacteria resuspended in Paramecium culture medium. The OD_600nm was adjusted with medium to 0.25 and supplemented with 100 μg/ml ampicillin and 12 μg/ml IPTG.

Single Paramecium cells were isolated and grown for about twenty fissions, again isolated and grown for another twenty fissions before each clone was starved to induce autogamy (Berger, 1986). Autogamy was monitored by fluorescence microscopy after staining with Hoechst 33342 (Molecular Probes, Leiden, The Netherlands). Autogamous cells were fed, first with normal bacteria and used 3 days later for RNAi feeding experiments. Paramecium cells were washed twice in PIPES buffer and starved for at least 2 hours in PIPES at room temperature before use in feeding experiments. In single-cell experiments, one cell was added to 150 μl feeding solution in a depression well. Cells were cultured at 25°C during the experiment and transferred every 24 hours to a freshly prepared feeding solution. The phenotype was analyzed 96 hours after the start of the feeding experiment.

For evaluation of the phagocytotic capacity, cells were fed with Congo-Red-stained yeast, fixed with 4% (w/v) formaldehyde after 3, 5 and 10 minutes, and the number of yeast cells ingested was counted. The amount of yeast in control cells was set as 100% phagocytotic activity. This was evaluated in at least 20 cells per experiment, with five experiments per actin type. The decrease in phagocytotic capacity concerned the number of food vacuoles formed as well as the number of yeast cells per vacuole. Although a variation of up to 30 yeast cells per vacuole was observed in control cells, phagocytotically restricted Paramecium cells formed only small vacuoles with less than 10 yeast cells.

For statistical evaluation of changes in the division rate and in the phagocytotic capacity in RNAi experiments, one-way analysis of variance (ANOVA) was used to determine statistical significance.

We gratefully acknowledge the availability of genomic library from J. Cohen (CNRS Gif-sur-Yvette, France) and of an expression library from J. E. Schultz (University of Tübingen, Germany) as well as the access to the Paramecium genome database at an early stage of its development (J. Cohen and L. Sperling, CNRS, Gif-sur-Yvette, as well as Genoscope, Paris). We thank C. Stuermer and E. May for use of their confocal microscopes. Supported by Deutsche Forschungsgemeinschaft grants to H.P.