Plastid segregation and cell division in the apicomplexan parasite Sarcocystis neurona

The phylum Apicomplexa represents a large and diverse group of protozoan parasites. Among these are the human pathogens that cause malaria, AIDS-related encephalitis (Toxoplasma) and severe enteritis (Cryptosporidium and Cyclospora). The phylum also contains many parasites of substantial veterinary importance such as Theileria, Babesia, Eimeria and Sarcocystis. Aside from their obvious medical and economic importance protozoan parasites have long fascinated cell biologists as model organisms. Recent cell biological work on Apicomplexa has been focused on the cellular structures and molecules involved in their ability to invade mammalian cells, their peculiar cell cycle and mechanisms of cell division, and the presence of several unique organelles. One of these organelles, which has received particular attention, is the apicomplexan plastid or apicoplast.

Given that the cells of the mammalian host lack plastids, the discovery of the apicoplast has provided an entire organelle of potential parasite-specific drug targets. A number of these targets have been validated including several metabolic pathways confined to the plastid, as well as the maintenance of the organelle and its genome (McFadden and Roos, 1999; Ralph et al., 2004). Our interest is focused on apicoplast division and segregation during parasite development. Plastid division is best understood in chloroplasts. Interestingly, chloroplasts still use many elements of their ancestral (cyano-) bacterial division machinery. All plant and algal genomes studied so far harbor two gene families homologous to the bacterial FtsZ gene (Miyagishima et al., 2004; Osteryoung et al., 1998; Stokes and Osteryoung, 2003). FtsZ is a prokaryotic homolog of tubulin and the main structural molecule in the bacterial division ring (Lowe and Amos, 1998; Lutkenhaus and Addinall, 1997). Plant FtsZ is targeted to the lumen of the chloroplast where it forms rings at the site of constriction, very much like its bacterial homolog (Kuroiwa et al., 2002; McAndrew et al., 2001; Vitha et al., 2001). Assembly and positioning of this ring is controlled and aided by the homologs of the bacterial cell division proteins MinD and MinE (Colletti et al., 2000; Dinkins et al., 2001; Itoh et al., 2001). In addition to these proteins found in all bacteria, higher plant chloroplasts also employ homologs of Ftn2 and Artemis, which are essential cell division proteins that are phylogenetically restricted to cyanobacteria (Fulgosi et al., 2002; Vitha et al., 2003). Recently, an eukaryotic element of the division machinery has been identified, a dynamin-like protein that forms a ring on the cytoplasmatic side and is required for proper organellar fission (Gao et al., 2003; Miyagishima et al., 2003). Surprisingly, our extensive comparative genomic analysis of the completed or nearly completed genomes of several plastid-harboring apicomplexan parasites (P. falciparum, P. yoelii, T. gondii, Eimeria tenella and Theileria annulata) did not identify a single homolog for any member of this conserved machinery. This is despite the fact that several proteins such as FtsZ are highly conserved and readily identified by BLAST search in the genomes of a diverse set of plants and algae (V.S. and B.S., unpublished data).

How is plastid division accomplished in apicomplexan parasites in the absence of the conserved machinery? Previous work on Toxoplasma has shown these organisms to employ a genuinely eukaryotic mechanism. Specifically, the dividing plastid associates with the centrosomes of the mitotic spindle (Striepen et al., 2000). As a result of this association the ends of the plastid are inserted into the budding daughter cells and segregated. Based on the observation that all Apicomplexa have lost the bacterial machinery, we hypothesized that centrosome association might be the conserved mechanism for apicoplast division. However, there is surprising morphological diversity among plastids in different species of Apicomplexa. While T. gondii maintains a small ovoid organelle, other apicomplexans harbor plastids that form complex highly branched structures [e.g. Plasmodium (Waller et al., 2000)] or long tubules (Sarcocystis, this study). We hypothesize that these morphologies are consistent with the centrosome association model and reflect the diversity of the apicomplexan cell division mechanisms. All lytic cycles in Apicomplexa start and end with a conserved motile stage, which is endowed with a set of organelles required for host cell attachment and invasion (Sibley, 2004). However, the development within the host cell and the underlying cell cycle model vary dramatically between closely related species and even between different life cycle stages of the same organism. In addition to more conventional models, Apicomplexa can forego cytokinesis and/or nuclear division for several cell cycles yielding multinucleate or polyploid stages that finally give rise to multiple daughter cells.

We have tested this hypothesis using the apicomplexan Sarcocystis neurona as a new cell biological model. S. neurona is the causative agent of equine protozoal myeloencephalitis (Dubey et al., 2001; Lindsay et al., 2004). This parasite can be maintained in tissue culture continuously, and its developmental stages are relatively large, permitting high resolution imaging in infected cells. Furthermore, the peculiar S. neurona cell division mechanism dissociates DNA replication, nuclear division and cytokinesis over an extended period, thereby providing a unique system to dissect the relationship of plastid maintenance, fission and segregation with mitosis and cytokinesis. In this study we demonstrate that S. neurona undergoes five cycles of chromosome replication without nuclear division. Interestingly, intranuclear spindles persist throughout the cell cycle providing a potential mechanism to organize chromosomes and organelles in an organism with a polyploid nucleus. By using molecular markers and pharmacological treatments we show that centrosomes and spindles are essential for plastid organization and faithful segregation. The timing and mechanism of organellar fission was also analyzed. Finally, a transfection system was established for this organism to validate the model in vivo through live cell microscopy and laser bleaching experiments.

Sarcocystis neurona strain SN3 was propagated in primary bovine turbinate (BT) cells. BT cells were grown in Dulbecco's modified Eagle's medium (DMEM, HyClone) supplemented with 10% heat-inactivated fetal bovine serum (HyClone), 2 mM L-glutamine, 5 U/ml penicillin, 10 μg/ml streptomycin, and 1 ml/l fungizone (Invitrogen). A continuous culture was maintained by weekly passage of 1 ml culture supernatant containing S. neurona merozoites into a fresh BT culture. We have noticed (as have others) that inefficient parasite egress from the host cell seems to limit the number of infective parasites that can be obtained (Ellison et al., 2001). In some experiments infected cultures were scraped after 72 hours and forced through a 26 1/2 gauge needle to break host cells. Free parasites were purified by filtration through 3 μm polycarbonate membrane (Costar) prior to infection of cell cultures.

A stable transgenic Toxoplasma gondii strain expressing the reporter ferredoxin NADH reductase-red fluorescent protein (FNR-RFP) was established and cultivated in human foreskin fibroblast (HFF) cells as previously described (Striepen et al., 2000).

Plasmids for S. neurona transfection experiments were initially derived from the T. gondii expression vector ptubYFP-YFP/sagCAT (Gubbels et al., 2003; Striepen et al., 2000) by replacement of the Toxoplasma tubulin promoter and the selectable marker cassette with the 5′ flanking region of the S. neurona SAG1 locus (Howe et al., 2004); R.Y.G., V.S., B.S. and D.K.H., unpublished data). To engineer the plastid reporter construct pSnsagFNR-RFP used in this study, the coding region of FNR-RFP along with the 3′ UTR derived from the T. gondii DHFR-TS gene was excised from plasmid ptubFNR-RFP (Striepen et al., 2000) with NotI and BglII and ligated into pSnsagYFP-YFP digested with NotI and BglII, placing the transgene under the control of the S. neurona promoter. The coding sequence of the T. gondii acyl carrier protein was excised for plasmid ptubACP-GFPsagCAT (Waller et al., 1998) using BglII and AvrII and was ligated into pSnsagYFP-YPF equally restricted to replace the first YFP cassette to yield pSnsagACP-YFP.

Freshly lysed S. neurona merozoites were purified by filtration through 3 μm polycarbonate filters and resuspended in cytomix (120 mM KCl, 0.15 mM CaCl₂, 10 mM K₂HPO₄/KH₂PO₄ pH 7.6, 25 mM Hepes pH 7.6, 2 mM EDTA, 5 mM MgCl₂, 2 mM ATP, 5 mM glutathione) to a density of 3×10⁷ parasites per ml. 50 μg plasmid DNA were ethanol precipitated, resuspended in 100 μl cytomix, mixed with 300 μl parasites in a 2 mm electroporation cuvette, and exposed to a 1800 V, 25 Ω and 25 μF pulse generated by a BTX ECM 630 apparatus (Genetronix).

For immunofluorescence, filter-purified S. neurona merozoites were allowed to settle for 10 minutes on polylysine-coated coverslips (coverslips were pretreated for 20 minutes with 10 mg/ml polylysine, washed and air dried). Alternatively, BT cells were grown to confluency on sterilized glass coverslips in six-well plates and subsequently infected with S. neurona merozoites. Cultures were fixed at various times post-infection for 10 minutes using 4% paraformaldehyde in PBS, followed by permeabilization with 0.25% (w/v) Triton X-100 in PBS. The following primary antibodies were used at the indicated dilutions: affinity purified polyclonal rabbit antiserum against T. gondii acyl carrier protein [ACP; 1:400; kindly provided by G. I. McFadden, University of Melbourne (Waller et al., 1998)], monoclonal antibody 12G10 raised against α-tubulin (1:25; kindly provided by J. Frankel, University of Iowa (Jerka-Dziadosz and Frankel, 1995; Jerka-Dziadosz et al., 1995)], monoclonal mouse and polyclonal rabbit anti-centrin [1:400; kindly provided by J. L. Salisbury, Mayo Clinic (Paoletti et al., 1996)], rat polyclonal anti-SnMic10 (Hoane et al., 2003) and rabbit polyclonal anti-IMC3 (Gubbels et al., 2004). All antibodies were diluted in PBS plus 1% BSA and detected using Alexa Fluor 488- or Alexa Fluor 546-conjugated goat anti-mouse, anti-rabbit or anti-rat antibodies (Molecular Probes). DNA was stained by incubation with 2 μg/ml 4′,6′-diamino-2-phenylindole (DAPI; Molecular Probes) for 10 minutes, excess dye was removed by washing, and coverslips were mounted with Gel/Mount (Biomeda Corporation). Fluorescence images were captured with a CCD camera (C4742-95; Hamamatsu) on a Leica DMIRBE microscope. Image acquisition, contrast adjustment and channel merging were performed using openlab software (Improvision).

To estimate the DNA content of individual schizont nuclei, the intensity of DAPI staining was measured in situ by image analysis. Fixed cultures were stained for tubulin and DNA as described above. Images were collected in the red and DAPI channel with constant exposure time and within the linear range of the CCD camera (the contrast was not adjusted). Nuclei were defined as objects by generating binary image masks for each image using a thresholding function. The mean (background-subtracted) pixel intensity was multiplied by the area of each nucleus to obtain a cumulative intensity measurement in arbitrary units. Host cell nuclei were measured as internal controls and provided results consistent with a confluent monolayer mostly arrested in G1 (the bulk of the measurements clustered around a mean of 7.5 10⁶ units Std 1.2 10⁶, n=20), and two cells with 12 and 16.5 10⁶ units, probably representing G2 and S phase, respectively.

For deconvolution microscopy and laser bleaching experiments confluent BT and HFF coverslip cultures were infected with transgenic FNR-RFP-expressing S. neurona and T. gondii, respectively. Cultures were observed on a Delta Vision Spectris deconvolution workstation (Applied Precision) with an Olympus 60× NA 1.2 water immersion lens 48 hours after transfection. Coverslips were mounted in medium (Secure-Seal, Molecular Probes) and maintained at 37°C during the observation. A 25 mW argon laser was used to bleach the sample, by exposing it to five 488 nm pulses for 300 mseconds each at 1.5 mW laser power. The size of the laser spot was nominally diffraction limited. Stacks of optical sections were taken before and after the bleach with z-increments of 0.3 μm and xy spacing of 0.1 μm. The pre- and the post-bleach images were set to identical contrast setting using the image-scaling function in the Softworx software (Applied Precision). Total projection images of sequential z-stacks were rendered using Volocity software (Improvision).

The dinitroaniline herbicide oryzalin was obtained from Riedel-deHaen (Seelze, Germany), and a 10 mM stock solution was prepared in DMSO. Confluent BT cell coverslip cultures were infected with S. neurona, allowed to develop for 36 hours, and then treated with 2.5 μg/ml oryzalin for either 24 hours or 48 hours prior to processing for immunofluorescence analysis.

BT cells were grown to confluency in polystyrene Petri dishes and infected with S. neurona merozoites. Cells were fixed in situ, 72 hours post infection, with 1% glutaraldehyde and 1% OsO₄ in 50 mM sodium phosphate buffer pH 7.2 for 1 hour on ice, followed by overnight staining with 0.5% aqueous uranyl acetate. Samples were dehydrated using a progressive ethanol series, ethanol was gradually exchanged for resin and samples were embedded in EPON (Electron Microscopy Sciences) directly in the dish. Blocks of polymer were cut from the dishes and ultrathin sections were taken parallel to the monolayer using a diamond knife. Sections were placed on Formvar-coated grids, stained with 4% uranyl acetate and 4% lead citrate and observed using a JEO JEM-100 CX II microscope.

For immunoelectron microscopy, infected cells were fixed with 2% paraformaldehyde and 0.1% glutaraldehyde in 0.5× PBS for 20-30 minutes on ice. Fixed cells were washed, scraped and centrifuged. The pellet was progressively dehydrated with ethanol and gradually infiltrated with LR-White on ice. Finally the sample was transferred to a gelatin capsule and polymerized at 50°C. Ultrathin sections cut from trimmed blocks and recovered on Formvar-coated grids were placed on 50 μl drops of the polyclonal anti-ACP antibody (1:400 in PBS 2% BSA) followed by 1:50 goat anti-rabbit IgG conjugated with 10 nm gold (BBinternational). After immunolabeling, sections were stained and analyzed as described above.

Sarcocystis is an obligate intracellular parasite. After invasion of the host cell, the parasite develops over 2-3 days during which the nucleus seems to grow continuously until it fragments into multiple nuclei, which are packaged into daughter cells (Speer and Dubey, 1999; Speer and Dubey, 2001). We have followed these events by fluorescence microscopy using S. neurona-infected tissue cultures as a model. The schizonts are lemon shaped and lie directly within the host cells cytoplasm. Their central large nuclei were intensely stained by the DNA intercalating dye DAPI (Fig. 1). Using a monoclonal antibody raised against α-tubulin from Tetrahymena thermophila (Jerka-Dziadosz and Frankel, 1995), we observed only moderate and diffuse staining in the parasite cytoplasm. This is in agreement with the previous reports that schizonts lack subpelicular microtubules, which are characteristic of the invasive stages of Apicomplexa (Morrissette and Sibley, 2002a). Within the nucleus however, two consistent and mutually exclusive staining patterns were observed: multiple intranuclear bundles of microtubules, or intensely stained dots right under the nuclear envelope (Fig. 1A,D). The microtubule bundles seemed to emanate from points close to the surface of the nucleus and coincided with apparent condensation of the DNA staining. This suggests that they represent mitotic spindles, segregating chromosomes (Speer and Dubey, 1999). To confirm this finding, infected cultures were double labeled with the anti-tubulin antibody and an anti-centrin serum. Centrin is a cytoplasmatic calcium binding protein that is concentrated in centrosomes surrounding the centrioles (Salisbury, 1995). This serum has previously been shown to label the centrosomes in the related parasite T. gondii (Striepen et al., 2000). As shown in Fig. 1C-F, the poles of these spindles clearly coincided with the centrin label. Note that the multiple centrosomes are linked by individual microtubules that span across the nucleus.

The staining patterns observed for the mitotic spindles were often too complex to count their numbers unambiguously. The dots in the second staining pattern however, could be easily counted and their number increased in geometric progression (2, 4, 8, up to 64; Fig. 2A-F, note that the dot often consists of two very close and poorly resolved structures). The increase in the number of these structures seemed to be directly correlated with the size of the nucleus. To quantify this relationship, the DNA content of individual nuclei was measured by image analysis. Schizont nuclei were defined using a binary mask (see Material and Methods for detail) and the cumulative intensity of their DAPI label was recorded for each mask. Fig. 2G shows such measurements for nuclei analyzed from 100 fields plotted against the number of tubulin dots recorded for each respective nucleus. The DNA content of schizont nuclei is clearly correlated with the number of dots and as might be expected doubles synchronously with the former indicating that these dots might be associated with or represent minute spindles. Upon double labeling with centrin antiserum, each red tubulin dot was flanked by two green centrosomes (Fig. 1D-F, see inset in F). To further investigate the nature of these structures, infected cultures were fixed and embedded for electron microscopy. Fig. 1G shows a section through a young schizont (the large central nucleolus and the homogenous chromatin structure suggests that this cell is in interphase). Consistent with immunofluorescence images, the electron micrographs demonstrate short baskets of microtubules within the nucleus (Fig. 2H,I, black arrowheads). These microtubules emanate from centrosomes (white arrowheads with black outline), which lie outside of the nuclear envelope in the cytoplasm. As many as 32 of these short spindles were observed associated with a single polyploid nucleus. In even later developmental stages, the nucleus is segregated into 64 daughter nuclei following the last chromosomal duplication event (Fig. 2F). Our analysis concludes that all stages in the developing schizont present either typical spindles (associated with mitosis and DNA condensation), or short interphase spindles. We failed to observe stages devoid of microtubular structures in the nucleus.

To identify the plastid in S. neurona and to follow its fate through development, parasites were analyzed by immunofluorescence using a polyclonal antibody raised against T. gondii acyl carrier protein (ACP). ACP is a nuclear-encoded protein that is posttranslationally targeted to the lumen of the apicoplast where it takes part in type II fatty acid synthesis. ACP has been previously used as a plastid marker in T. gondii and P. falciparum (Striepen et al., 2000; Waller et al., 1998; Waller et al., 2000). Extracellular S. neurona merozoites stained with this antibody yielded a pattern indistinguishable from T. gondii tachyzoites indicating that the antibody cross reacts and recognizes ACP from T. gondii as well as S. neurona (Fig. 3A-D). The antibody labeled a single small ovoid organelle located apically of the nucleus in extracellular merozoites of S. neurona, and staining coincided with a spot of extranuclear DAPI staining representing the plastid genome. This extranuclear DNA hybridized with DNA probes specific for the S. neurona plastid genome (V.S. and B.S., unpublished). In developing S. neurona schizonts, however, the ACP antibody labeled a single tubular structure (Fig. 3E). Costaining with DAPI revealed that this elongated tubular plastid was associated with the periphery of the nucleus (Fig. 3H). The tubular morphology of the plastid was maintained throughout the intracellular development of the parasite, and no differences were observed between parasites undergoing mitosis and those in interphase (data not shown).

The presence of a large continuous plastid is highly interesting as it mirrors the development of the nucleus in S. neurona and could lend additional support to the model that nucleus and plastid are organized and segregated by the same molecular machinery. Whether the plastid is indeed continuous could not be concluded with certainty based on immunofluorescence analysis, since some samples seemed to show small breaks in the plastid tubule depending on fixation conditions. To avoid fixation artifacts, we developed a transfection system for S.neurona to establish fluorescent-protein-expressing parasites for in vivo analysis (R.Y.G., V.S., B.S. and D.K.H., unpublished). Transfection vectors were constructed with the promoter region of the S. neurona SAG1 gene, driving the expression of a reporter gene. The gene for two well-characterized T. gondii plastid-targeted proteins (ACP and FNR) were translationally fused to YFP or RFP, respectively. Parasites transfected with these plasmids by electroporation were used to infect coverslip cultures. Images were taken 24-36 hours after transfection without fixation. Transfection with both of these independent reporters resulted in fluorescent labeling of a tubular organelle in close association with the nucleus (Fig. 3I-L).

S. neurona plastids, imaged in vivo, appeared continuous and showed no obvious breaks (see three-dimensional reconstruction derived from deconvolved serial optical z sections in Fig. 4E and Movie 1 in supplementary material). To directly test for organellar continuity, laser bleaching experiments were performed. Given the high rates of diffusion over the small cellular distances, exposure of the distal end of the organelle to a series of laser pulses should bleach fluorescence in the entire organelle. If the organelle is discontinuous, bleaching should remain partial (see Fig. 4A,B for a schematic outline of the experiment). A series of short exposures to the 488 nm laser line at modest power settings was applied to one end of a plastid labeled in vivo with FNR-RFP (a soluble luminal marker). Prior to and after bleaching, z-stacks of fluorescent images were collected. Sustained bleaching of a small portion of the S. neurona plastid resulted in bleaching of the entire organelle consistent with free marker diffusion through the organelle (Fig. 4C-H; E and H are total projections of the z-stacks, a rendered version of this data is available as a movie: see Movie 1 in supplementary material). To ensure that the imaging process itself did not cause excessive bleaching and that the laser provides sufficient spatial precision, T. gondii was used as a well-suited control. A T. gondii plastid labeled with FNR-RFP was bleached in a vacuole containing multiple tachyzoites. Bleaching was restricted to the single organelle targeted (Fig. 4I-N). Recovery of the fluorescent signal was observed in both S. neurona and T. gondii plastids over a period of 1 hour (data not shown). Signal recovery probably occurred by import and is consistent with the kinetics of apicoplast targeting in P. falciparum (van Dooren et al., 2002) and T. gondii (Derocher et al., 2005).

In the related parasite T. gondii, plastids are associated with the centrosomes during cell division (Striepen et al., 2000). To establish if and when such association occurs in the single tubular plastid in S. neurona, infected cultures were triple-labeled with anti-ACP (plastid, green), anti-α-tubulin (spindle, red) and DAPI (DNA, blue). The plastid was found to be in close contact with the nucleus throughout its development. The area of contact between nucleus and plastid coincided with the poles of spindles observed during mitosis and short spindles maintained through interphase (Fig. 3). Double labeling with antibodies against ACP and centrin revealed an equally tight association between plastid and centrosomes (Fig. 5A, arrowheads indicate individual centrosomes). To validate this association at higher magnification, S. neurona-infected cells were fixed and embedded directly in the tissue culture dish. Blocks were cut and trimmed to preserve the original orientation and sections were then taken parallel to the cell monolayer. Fig. 5B-D shows electron microscope images of serial sections taken through a developing S. neurona schizont. A tubular multi-membranous organelle was observed that wrapped around the upper right portion of the central nucleus. At higher magnifications, the four distinct membranes surrounding the plastid were apparent (Fig. 5D). Four centrioles can be identified in Fig. 5C (white arrowheads) that lie in intimate proximity to the plastid [also see Speer and Dubey (Speer and Dubey, 1999)]. To confirm the identification of the organelle as a plastid, S. neurona-infected cultures were also fixed and embedded for immunogold labeling. Again a tubular organelle was observed in close proximity to the nucleus. This organelle shows strong and specific labeling for the plastid protein ACP (Fig. 5E,F).

If spindles play an important role in plastid organization their disruption should affect plastid morphology. To test this, parasite microtubules were disrupted by treatment with oryzalin. Oryzalin is a dinitroaniline that has been shown to affect both subpellicular and spindle microtubules in Apicomplexa (Morrissette and Sibley, 2002a; Morrissette and Sibley, 2002b). Cultures infected with S. neurona were allowed to develop for 36 hours and then treated with 2.5 μg/ml oryzalin for 24 hours prior to processing for immunofluorescence. A profound loss of plastid organization was observed (Fig. 6C,D). In treated cells, the organelle coiled to the middle of the cell. Even though the organization of the plastid around the nucleus was lost in the drug-treated parasites (compare to untreated control in Fig. 6A,B), centrosome association was maintained. Upon double labeling of drug-treated parasites with anti-centrin and anti-ACP antibody multiple centrosomes were observed associated with the plastid (white arrowheads indicate individual centrosomes in Fig. 6E).

To endow each daughter cell with a plastid the tubular plastid has to undergo fission. In order to time this event precisely, a molecular marker was used. SnMIC10 is a secretory protein targeted to the micronemes, which probably plays a role in host cell invasion. Importantly, its expression shows tight developmental regulation (Hoane et al., 2003). Fig. 7A shows three consecutive developmental stages of S. neurona: a developing schizont (1), a late schizont during budding (2) and a terminal stage schizont showing completed daughter cell formation (3). Expression of SnMic10 is undetectable in the developing schizont (Fig. 5B), but initiates towards the end of the last replication cycle with staining first seen as a dot at mid-cell most probably representing the Golgi or a premicroneme compartment. After completion of daughter cell formation, SnMic10 shows a typical micronemal pattern underlying the apical surface of the merozoite. Plastids were always tubular in the absence of SnMIC10 staining (Fig. 7C,D1), and fission concurred with the beginning of Mic10 expression. This concurrence of plastid fission with the last cell cycle was also evident by triple labeling for plastid, microtubules and DNA (Fig. 7E-H).

To investigate the role of spindles in plastid fission, infected cultures were allowed to develop for 36 hours and then treated with oryzalin for 48 hours prior to processing for immunofluorescence. Plastid fission was evident in the absence of spindles in treated cultures (Fig. 6G). However, these plastids were of unequal size and had lost the nuclear association (Fig. 6H) observed in untreated parasites. We were interested to learn if the observed fission could be a consequence of daughter cell formation in the absence of spindles. For this purpose, infected cultures were treated with oryzalin as described above and labeled with an antibody to T. gondii IMC3. IMC3 is a component of the inner membrane complex underlying the parasites plasma membrane, and antibodies to this complex reveal newly forming daughter cells (Fig. 6J) (Gubbels et al., 2004). We observed that drug treatment did not prevent the formation of daughter cells (Fig. 6K,L), however, nuclear segregation and packaging was severely impaired, yielding daughter cells devoid of a nucleus.

Plant chloroplasts divide by binary fission and several constrictive rings have been described (Kuroiwa et al., 2002; McAndrew et al., 2001; Vitha et al., 2001). Many of the proteins that make up these rings, and the elements that regulate ring position, are conserved between plastids and their cyanobacterial ancestor (Osteryoung and Nunnari, 2003). By contrast, our previous work in T. gondii has indicated that the replication of parasite secondary plastids occurs in association with the centrosomes of the mitotic spindle (Striepen et al., 2000). The current study shows this to be a conserved mechanism among Coccidia and it might provide a model to understand plastid division in the malaria parasite Plasmodium. In Sarcocystis, we have found that there is tight association of the plastid with the mitotic spindle and its centrosomes using several molecular markers examined at the light and electron microscopic level. If the mechanism is conserved, why do plastids show such diverse morphology? An obvious difference between T. gondii and S. neurona are their divergent cell cycle models. In T. gondii, the segregation of chromosomes is immediately followed by nuclear division and cytokinesis, and parasites maintain a haploid genome (Radke et al., 2001; Sheffield and Melton, 1968). In S. neurona, however, the number of spindles and centrosomes that we observed in growing schizonts suggests that this organism completes six replication cycles prior to daughter cell formation (Fig. 2). The first five cycles involve a succession of chromosomal duplication and segregation accomplished by multiple synchronous spindles throughout the nucleus, but nuclear division and cytokinesis are postponed, resulting in a 32 N nucleus (see Fig. 8 for a schematic outline of this process). The sixth cycle concurs with nuclear segregation and cytokinesis. The surprising persistence of spindles throughout these six replication cycles (with a peculiar `mini spindle' during interphase) could provide a mechanism organizing chromosomes and organelles in later highly polyploid stages.

Indeed, pharmacological ablation of spindles results not only in a disorganized nucleus unable to properly segregate during budding, but also a disorganized plastid (Fig. 6). Molecular probes for the identification of centromeres and kinetochores are needed to fully test this model for chromosomes.

It is important to note how closely the development of the plastid mirrors the nuclear events. In T. gondii, an ovoid organelle is divided and segregated at the same time as the nucleus in every replication cycle. By contrast, in S. neurona we observed a continuous tubular plastid that grows alongside the nucleus (Figs 3, 4 and 5). Organelle continuity was experimentally confirmed in vivo using transgenic parasites (Fig. 5). Plastid fission into 64 individual daughter plastids occurs synchronously with the final nuclear division and segregation thus indicating that plastid fission is timed and regulated in a cell cycle-dependent way. This observation is consistent with a model in which the growing daughter cell pellicle, in conjunction with spindle attachment, could provide the constriction needed for fission (Striepen et al., 2000). An alternative model suggests the action of constrictive rings, similar to those seen in bacteria and plant chloroplasts (Matsuzaki et al., 2001). We have seen no evidence for plastid division rings in T. gondii or S. neurona, and we have not detected in apicomplexan genomes any of the genes for the conserved proteins associated with these structures (S.V. and B.S., unpublished data). We cannot exclude the possibility that such rings exist and are missed because they are extremely short lived, but they would have to consist of unique molecular components lacking homology to the cyanobacterial division apparatus. If constrictive rings are involved in plastid fission in S. neurona, their formation must be controlled in a cell cycle-dependent way and they have to form at multiple defined points along the tubular organelle in a spindle pole dependent-manner.

Spindles are clearly critical for the spatial organization of the plastid during parasite development, but are they required for fission? Ablation of spindles through pharmacological treatment did not prevent fission but broke the coordination of plastid and nuclear division (Fig. 6). Importantly, this treatment leaves intact plastid-centrosome association as well as daughter cell budding. This suggests that these two processes are sufficient for plastid fission and packaging, but that spindles are essential to coordinate this process with nuclear division. This model is in agreement with the observation that spindle ablation in T. gondii can lead to the formation of anucleate daughter cells, which in some cases still carry a plastid (Morrissette and Sibley, 2002b). The observation of sustained centrosome association under oryzalin treatment also suggests that this association is not mediated by microtubular structures and mechanisms.

The clarity of the subcellular structure in S. neurona when viewed by light microscopy is exceptional among apicomplexan species. This clarity combined with the parasite's peculiar cell cycle makes it a unique model for the cell biological analysis of cell division in Apicomplexa. Transfection technology, as demonstrated in this report, now opens the door to mechanistic studies through genetic manipulation and in vivo analysis.

Funding for this work was provided by grants from NIH-NIAID and additional support from Merck Research Laboratories to B.S., and Amerman Family Foundation and Fort Dodge Animal Health to D.K.H. We thank G. McFadden, S. Salisburry, J. Frankel and J. Gaertig for antibody reagents and Michael White for discussion and critical reading.