An novel dynamin-related protein has been recruited for apicoplast fission in Toxoplasma gondii

Abstract

Background

Apicomplexan parasites cause numerous important human diseases including malaria and toxoplasmosis. Apicomplexa belong to the Alveolata, a group that also includes ciliates and dinoflagellates. Apicomplexa retain a plastid organelle (the apicoplast) that was derived from an endosymbiotic relationship between the alveolate ancestor and a red alga. Apicoplasts are essential for parasite growth and must correctly divide and segregate into daughter cells upon cytokinesis. Apicoplast division depends on association with the mitotic spindle, although little is known about the molecular machinery involved in this process. Apicoplasts lack the conserved machinery that divides chloroplasts in plants and red algae, suggesting that these mechanisms are unique.

Results

Here we demonstrate that a dynamin-related protein in Toxoplasma gondii (TgDrpA) localizes to punctate regions on the apicoplast surface. We generate a conditional dominant-negative TgDrpA cell line to disrupt TgDrpA functions and demonstrate that TgDrpA is essential for parasite growth and apicoplast biogenesis. Fluorescence recovery after photobleaching and time-lapse imaging studies provide evidence for a direct role for TgDrpA in apicoplast fission.

Conclusions

Our data suggests DrpA was likely recruited from the alveolate ancestor to function in fission of the symbiont, and ultimately replaced the conserved division machinery of that symbiont.

Introduction

Plastid organelles trace their evolutionary origins to cyanobacteria that were incorporated into eukaryotic cells by a process of endosymbiosis. This evolutionary history dictates that they cannot be formed de novo. Instead, existing plastids divide to give rise to daughter organelles that partition into daughter cells upon cell division. Previously studied plastids contain an FtsZ-based division apparatus, retained from the cyanobacterial endosymbiont [1]. In addition, plant and red algal plastid division involves a dynamin-like protein called ARC5 (also known as DRP5B [2, 3].

Apicoplasts, the non-photosynthetic plastids of apicomplexan parasites, must correctly divide and segregate into daughter cells for parasites to remain viable [4]. Surprisingly, apicomplexan genomes lack homologues to both ARC5 and FtsZ [5], suggesting that apicoplast division is mechanistically different to that in previously studied plastids. One striking difference is the association of the apicoplast with the centrosomes of the mitotic spindle [6, 7]. This association is thought to ensure proper segregation during cytokinesis, parcelling out apicoplasts to a highly variable number of daughter cells formed in the complex apicomplexan budding process [8]. While centrosome association provides a unifying model for segregation, it remains unclear how apicoplast fission occurs. One model suggests that fission depends on force generated by daughter cell budding [6], while electron microscopic studies identify apparent plastid division rings [9, 10], suggesting that protein components may mediate fission.

Dynamins are large GTPase proteins that function in a range of contractile processes, including the scission of endocytic vesicles, cytokinesis, nuclear remodelling and the fission of mitochondria, chloroplasts and peroxisome organelles [11-13], and we were interested in whether dynamins had a role in apicoplast division. Apicomplexan genomes encode three dynamin-related proteins that are phylogenetically distinct from ARC5. In this study, we characterise dynamin-related protein A (DrpA) in the apicomplexan T. gondii. We demonstrate that TgDrpA is required for apicoplast fission, and we present a detailed model for how TgDrpA functions in this process.

Results

T. gondii contains three dynamin-like proteins that are phylogenetically distinct from ARC5 dynamins

Using previously characterised proteins from yeast, plants and red algae, we performed homology searching for dynamin-related proteins in apicomplexan parasites. We examined the genomes of T. gondii, Plasmodium vivax, P. falciparum, Theileria parva and Cryptosporidium parvum. In each organism, we identified two characteristic dynamin-related proteins. To ascertain the evolutionary history of these dynamins, we performed phylogenetic analyses on a multiple sequence alignment of a broad spectrum of dynamin-related proteins. These indicated that ARC5/Drp5B dynamins from plants, diatoms, green and red algae cluster together with strong bootstrap support, and are sister to a group of dynamins involved in cytokinesis [12]. These ARC5 and related proteins are clearly distinct from apicomplexan dynamins (Figure 1). One apicomplexan dynamin group (that we term Dynamin-related protein B or DrpB) forms a well-supported clade that includes dynamins from ciliates, a phylum of alveolates related to Apicomplexa (Figure 1; ref. [14], Breinich et al, this issue). The other apicomplexan dynamin (which we term DrpA) forms a clade with ciliate dynamins that is not well supported by bootstrap analysis. Removing ARC5 and related sequences enables us to incorporate more characters in our analysis and this additional data suggests that DrpA also forms an alveolate-specific clade (Figure S1A in Supplemental Data). A third protein with some similarity to dynamins is also present in apicomplexan genomes. These so-called DrpC proteins match only to the GTPase domain of dynamins and phylogenetic analyses indicate that DrpCs are distinct from ARC5 dynamins (Figure S1B). We conclude that ARC5 is not a conserved component in apicoplast division.

Figure 1. Phylogenetic analyses of dynamins.

We generated a multiple sequence alignment of the conserved region of a range of dynamin related proteins. The analysis included 449 residues and 39 taxa. We generated phylogenetic trees using PHYLIP, performing bootstrapping with 400 replicates. Bootstrap values based on neighbour-joining (NJ), maximum parsimony (MP) and maximum likelihood (ML) analyses were determined. In this figure we depict the consensus maximum likelihood tree. Dynamin homologues included in the analysis are from Drosophila melanogaster (DmShibire), human (HsDNM1 and HsDNM1L), Caenorhabditis elegans (CeDyn1 and CeDrp1), Cyanidioschyzon merolae (CmDnm1 and CmDRP5B), Dictyostelium discoideum (DdDymA, DdDlpA and DdDlpB), the diatoms Thalassiosira pseudonana (TpDYN1 and TpDRP5B) and Phaeodactylum tricornutum (PtrDYN1 and PtrDRP5B), Saccharomyces cerevisiae (ScVps1 and ScDnm1), Arabidopsis thaliana (AtADL2b, AtDRP5A and AtDRP5B), Physcomitrella patens (PpDRP5A and PpDRP5B), Chlamydomonas reinhardtii (CrDRP5A and CrDRP5B), the ciliates Tetrahymena thermophila (TtDrp1 and TtDrp7) and Paramecium tetraurelia (PteDRP1 and PteDRP2), the apicomplexans Plasmodium falciparum (PfDYN2 and PfDYN1), Plasmodium vivax (PvDYN2 and PvDYN1), Toxoplasma gondii (TgDrpA and TgDrpB), Theileria annulata (TaDrpA and TaDrpB) and Cryptosporidium parvum (CpDrpA and CpDrpB), and the trypanosomatids Trypanosoma brucei (TbDLP) and Leishmania mexicana (LmDLP).

Tg DrpA localises to the periphery of the apicoplast

To characterise the function of DrpA in Toxoplasma gondii we examined its localisation. We generated parasites expressing the entire open reading frame of TgDrpA fused to an N-terminal HA tag, expressed from the native TgDrpA promoter, and performed immunofluorescence assays. TgDrpA localises in many small patches throughout the cytosol of T. gondii, while a major component of TgDrpA fluorescence clusters at the apical end of the cell (Figure 2A). Co-localisation with the apicoplast stromal marker acyl carrier protein (ACP) indicates that this cluster occurs around the periphery of the apicoplast (Figure 2A). During division, apicoplasts form a distinctive U-shape, with the base of the “U” being the point of organelle fission [6, 10]. In dividing apicoplasts, we typically observed TgDrpA localising to this point of fission, as well as to the ends of the organelle (Figure 2B-C, arrows). Later in apicoplast fission, when the base of the apicoplast becomes more constricted, the punctate dot of TgDrpA observed early in the process appears as a more tubule-like structure between the dividing lobes of the apicoplast (Figure 2C, arrowheads). We found that TgDrpA does not localise to the Golgi (Figure S2A), and only occasionally to the mitochondrion (Figure S2B).

Figure 2. TgDrpA localises to the apicoplast.

(A-B) Immunofluorescence assays of a cell line expressing HA-tagged TgDrpA, labelled with anti-HA (green) and the apicoplast stromal marker anti-acyl carrier protein (ACP; red) antibodies. Arrows point to sites of apicoplast fission. Scale bars are 2 μm.

(C) Immunofluorescence assay of the HA-TgDrpA cell line, labelled with anti-HA (green), anti-IMC (red) and anti-ACP (blue) antibodies. Arrows point to sites of apicoplast fission, where HA-TgDrpA forms a punctate dot. Arrowheads point to the tubule-like structure adopted by HA-TgDrpA later in apicoplast fission, when the apicoplast has become further constricted. Scale bar is 2 μm.

(D) Anti-HA Western blot of the HA-TgDrpA cell line, labelling a protein band at approximately 90 kDa.

We performed an anti-HA Western blot on proteins extracted from the HA-DrpA cell line. This revealed the presence of a band of approximately 90 kDa, conforming to the expected size of HA-tagged DrpA (Figure 2D). We next performed protease protection assays in conditions where cytosolic but not apicoplast stromal markers were accessible to thermolysin. We found that TgDrpA was sensitive to thermolysin (Figure S2C), consistent with TgDrpA localising to the cytosol. In the absence of clear markers for the four membranes that surround the apicoplast, our data cannot rule out the possibility that DrpA might localise to one or more of these intermembrane spaces.

Tg DrpA is essential for parasite growth

Having established the localisation of TgDrpA, we next wanted to determine its function. Dynamin proteins are self-assembling GTPases containing an N-terminal GTPase domain, a middle domain and a C-terminal GTPase effector domain (GED; Figure 3A, top). Expression of dynamins with mutations in the GTP-binding site has been shown in other systems to specifically disrupt dynamin function in a dominant-negative fashion (eg. [15, 16]). We generated a dominant-negative DrpA where a lysine in the GTP-binding site was changed to an alanine (DrpA^K42A). To generate stable cell lines inducibly expressing dominant-negative DrpA, we fused a destabilisation domain (DD) tag to the N-terminus of DrpA^K42A (Figure 3A, bottom). DD-tagging promotes proteosomal degradation of the protein, with degradation prevented by the small molecule Shield-1 [17, 18]. To determine whether we could regulate expression of DD-DrpA^K42A, we grew parasites for 0 to 20 hours on 0.1 μM Shield-1, extracted proteins and performed Western blotting. In the absence of Shield-1, we detected low levels of DD-DrpA^K42A (Figure 3B). Levels increased 3 hours after the addition of Shield-1, and were maximal after about 9 hours (Figure 3B).

Figure 3. TgDrpA is essential for parasite growth.

(A) Dynamin-related proteins such as TgDrpA consist of three conserved domains: an N-terminal GTPase domain, a middle domain and a C-terminal GTPase effector domain (GED). To generate dominant-negative TgDrpA mutants, we mutated the lysine in the GTP-binding motif of the GTPase domain to alanine (K42A). We fused an N-terminal destabilisation domain (DD) and HA-tag to this construct and generated clonal cell lines.

(B) To demonstrate regulated expression of DD-DrpA^K42A, we performed Western blotting of cells grown for 0 to 20 hours in 0.1 μM Shield-1, probing blots with anti-HA antibodies and anti-GRA8 antibodies as a loading control. Asterisk represents a probable degradation product of DD-DrpA^K42A.

(C) We performed fluorescence growth assays on wild-type (top) and DD-DrpA^K42A (bottom) parasites expressing tandem-YFP. Parasites were grown in the absence of Shield-1 (green diamonds), in the presence of Shield-1 (blue squares) or pre-incubated for three days in the presence of Shield-1 before the assay, and continued to grow in Shield-1 (red triangles). Error bars represent one standard deviation from the mean.

To determine whether DrpA is essential for parasite growth, we expressed tandem-yellow fluorescent protein (YFP) in DD-DrpA^K42A mutant parasites, and monitored growth using a fluorescence growth assay [19]. DD-DrpA^K42A mutant parasites grew robustly in the absence of Shield-1 (Figure 3C, bottom). However, compared to wild-type parasites (Figure 3C, top), growth of DD-DrpA^K42A parasites in 0.1 μm Shield-1 was slowed after about 5 days. Preincubation of DD-DrpA^K42A mutant parasites in Shield-1 for three days resulted in negligible growth of the parasites.

As a second measure for parasite growth, we performed plaque assays. In the presence of Shield-1 we saw a severe reduction in plaque size compared to the no-Shield-1 control (Figure S3A), consistent with the importance of TgDrpA for parasite growth. We also preincubated parasites for 12 hours in the presence of Shield-1, washed out the drug for a further 12 hours, and set up plaque assays in fresh flasks in the absence or presence of Shield-1. As expected, parasites grown in the presence of Shield-1 exhibited severe defects in growth. Interestingly, cultures preincubated with Shield-1 for 12 hours then grown in the absence of Shield-1 showed plaques of a similar size to parasites grown entirely in the absence of Shield-1, but the number of plaques was reduced 63% when compared to the no Shield-1 control. This suggests that approximately 40% of parasites are no longer viable after a 12-hour incubation in Shield-1. We conclude that TgDrpA is essential for parasite growth, and that ablation of TgDrpA function results in a delayed death effect that is typical of processes affecting the apicoplast [4, 20].

Tg DrpA is essential for normal apicoplast morphology and biogenesis

To directly test the impact of loss of DrpA function on the apicoplast, we generated a DD-DrpA^K42A mutant cell line that targeted red fluorescent protein (RFP) to the apicoplast. In the absence of Shield-1, apicoplast morphology appeared normal, with a single apicoplast organelle localising to the apical end of each parasite (Figure 4A). Upon incubation in Shield-1, we observed severe defects in apicoplast biogenesis (Figure 4B-D). Apicoplasts frequently occurred as branched tubules that appeared to connect several cells within a vacuole (Figure 4C-D). We also observed apicoplasts mis-localised to the basal end of parasites (Figure 4B) or entirely missing from one or more parasites within a vacuole (Figure 4C-D). To quantify these defects, we grew DD-DrpA^K42A mutant parasites for 0, 3, 6, 9, 12 and 20 hours in Shield-1. We imaged 100 four cell vacuoles and scored apicoplast morphologies into four categories: normal apicoplasts (as in Figure 4a), basal stunted (Figure 4B), basal elongated (Figure 4C-D), and cells where apicoplasts were absent (Figure 4C-D). In the absence of Shield-1, most apicoplasts appeared normal (Figure 4E). After six hours of growth on Shield-1, most apicoplasts localised to the basal end of the cell and were stunted in appearance. After around 12 hours an increasing number of the basally-localised apicoplasts were elongated, while approximately 40% of parasites had lost their apicoplast (Figure 4E). This value correlates to the loss of viability in ~40% of parasites after 12 hour incubation in Shield-1 (Figure S3A), and we hypothesise that the growth defects we observe in the TgDrpA^K42A mutant results from loss of this essential organelle.

Figure 4. TgDrpA is essential for apicoplast biogenesis.

(A-D) Immunofluorescence assays of DD-DrpA^K42A parasites co-expressing apicoplast-targeted RFP (red), co-labelled with anti-IMC antibodies (green). In the absence of Shield-1, every parasite contains a single apically-localised apicoplast (A). In the presence of Shield-1, apicoplasts localise to the basal end of the parasite (B), frequently elongating towards the apical end of the cell and being absent in some cells (C-D). Scale bars are 2 μm.

(E) Quantification of the apicoplast defect in DD-DrpA^K42A parasites. We grew parasites for 0-20 hours on Shield-1 and imaged 100 four-cell parasite vacuoles. We classified apicoplasts into four categories: normal apicoplasts (purple), basal stunted apicoplasts (green), basal elongated apicoplasts (orange), and cells lacking apicoplasts (blue).

(F) Time-lapse imaging of apicoplasts in DD-DrpA^K42A parasites co-expressing cytosolic YFP and apicoplast-targeted RFP, grown in the presence of Shield-1. Scale bar is 2 μm.

To gain a dynamic understanding of the observed phenotypes, we performed time-lapse imaging of mutant parasites expressing cytosolic YFP and apicoplast-targeted RFP. We added Shield-1 to parasites 6 hours before commencing imaging. Initially there were no obvious defects in apicoplast morphology, with both cells in the two-cell vacuole containing a single, apically-localised apicoplast (Figure 4F; Supplemental Movie 1). After approximately 150 minutes, apicoplasts from both cells formed a “U” shape, typical of apicoplast immediately preceding fission [6, 10]. Subsequent imaging revealed that apicoplasts are unable to divide. Approximately 20-30 minutes later, cytokinesis commenced, with apicoplasts not dividing and becoming localised to the basal end of the cell (Figure 4F). We followed this vacuole for a further 6 hours. Apicoplasts remained localised to the basal end of each cell in the vacuole, remaining in a small “stumpy” form. After about 5 hours, apicoplast began elongating from the basal end of the cell. These data suggest that apicoplast morphology is normal up to the point of apicoplast division. Unable to divide, apicoplasts localise to the basal end of the cell and elongate before the next round of cell division.

We sought to quantify the basal localisation of apicoplasts in the DD-DrpA^K42A mutant. We generated a DD-DrpA^K42A mutant cell line expressing YFP-MORN1 and apicoplast RFP. MORN1 forms a contractile, basal complex in parasites, (Figure 5A, arrowheads; [21, 22]), in addition to labelling the centrocone and growing daughter bud. In the presence of Shield-1, apicoplasts exit parasites through the YFP-MORN1 labelled basal complex (Figure 5B, arrowheads), with the apicoplast constricting at the basal complex (Figure 5B, inset). We grew parasites for 24 hours on Shield-1 and measured the distance between the basal complex to the apicoplast of the same cell. In the absence of Shield-1, the average distance of the apicoplast from the basal complex is 2.710 μm (standard deviation 0.546), while the value drops to 0.535 μm (standard deviation 0.672) in the presence of Shield-1 (Figure 5B). We next examined the DrpA mutant phenotype by electron microscopy. We observed apicoplasts localising to the basal end of the parasite, and in some cases exiting the parasite (Figure 5C-D, pink arrows).

Figure 5. Apicoplasts localise to the basal end of parasites upon overexpression of dominant-negative TgDrpA.

(A-B) Live cell imaging of DD-DrpA^K42A parasites grown in the absence (A) or presence (B) of Shield-1, co-expressing FNR-RFP and YFP-MORN1. Arrow depicts the basal complex of a parasite, co-localising in (B) with a point of constriction in the apicoplast (inset). Scale bars are 2 μm.

(C) Quantification of the distance between the YFP-MORN1 labelled basal complex and the nearest point of the apicoplast. DD-DrpA^K42A parasites were grown in the absence (left) or presence (right) of Shield-1, with distance in μm shown on the y-axis. The blue circle represents the mean value for each data set.

(D-E) Electron micrograph images of DD-DrpA^K42A parasites grown on Shield-1. The basal end of parasites is marked by an electron dense area that likely corresponds to the contractile, MORN1-containing basal complex (black arrowheads). Apicoplasts (A) localise to the basal end of the parasites (pink arrowheads). The mitochondrion (M) and nucleus (N) are also shown. Scale bar is 1500 nm in (D) and 1000 nm in (E).

We conclude that incubation of the DD-DrpA^K42A mutant in Shield-1 results in rapid and severe defects in apicoplast biogenesis and division. The DrpA^K42A mutant is predicted to act in a dominant-negative way to disrupt DrpA function. However, the data presented in Figures 4 and 5 do not rule out the possibility that the observed defects in apicoplast fission are a consequence of DrpA overexpression. To test this, we overexpressed wild-type DrpA, the DrpA^K42A mutant and DrpA where the entire GTPase domain was deleted. We observed no effects on apicoplast biogenesis in cells overexpressing wild-type DrpA, while deletion of the entire GTPase domain resulted in apicoplast biogenesis defects identical to the DrpA^K42A point mutant (Figure S2). We conclude that the DrpA^K42A mutant acts in a dominant-negative way to disrupt native TgDrpA functions. We examined the effects of disrupting TgDrpA function on other cellular functions. We found that DrpA has no role in protein targeting to the apicoplast or secretory pathways, or in biogenesis of micronemes and rhoptries, specialised secretory organelles in Apicomplexa (Figure S4A-C). We found that although there were no consistent defects, we could not entirely rule out a minor role for DrpA in mitochondrial biogenesis (Figure S4D-E).

Tg DrpA mutants are incapable of apicoplast fission

Our data indicate that TgDrpA is essential for apicoplast biogenesis. To elucidate the mechanism of TgDrpA function in this process, we examined whether TgDrpA has a role in apicoplast fission. In the absence of apicoplast fission, we predict that apicoplasts from adjoining cells would remain connected. To experimentally test this, we performed fluorescence recovery after photobleaching (FRAP). We imaged DrpA^K42A parasites expressing apicoplast-targeted RFP, grown in the absence or presence of Shield-1. We laser bleached apicoplast fluorescence from one parasite within a vacuole and measured recovery over two minutes. In the absence of Shield-1, we observed no fluorescence recovery of apicoplast fluorescence (Figure 6A-B; Supplemental Movie 2). Average recovery in fluorescence after two minutes was 0.8 % of relative fluorescence units with a standard deviation of 0.9%. For parasites grown in Shield-1, we saw a consistent and significant recovery of fluorescence (Figure 6C-D; Supplemental Movie 3). Average recovery in fluorescence after two minutes was 22.3% with a standard deviation of 6.2%. Our data suggest that apicoplasts of adjoining cells in a single vacuole maintain a physical connection in the DrpA mutant, consistent with a defect in apicoplast fission. We performed similar FRAP analysis on mitochondrial fluorescence and found negligible recovery (Figure S4F-I).

Figure 6. Disruption of TgDrpA function results in defects in apicoplast fission.

(A-D) Fluorescence recovery after photobleaching in apicoplasts of DD-DrpA^K42A parasites grown in the presence or absence of Shield-1. We imaged parasites at 5 second intervals over 2 minutes, bleaching a diffraction limited region of the field-of-view (at the position indicated by the laser symbol) after 10 seconds. (A) and (C) show images from single experiments, imaged before (left, 0”), directly after (middle, 15”) and at the end of the experiment (right, 120”). (B) and (D) show quantifications of fluorescence recovery over time in five (B) or ten (D) independent experiments. DD-DrpA^K42A parasites co-expressing apicoplast-targeted RFP were grown in the absence (A-B) or presence (C-D) of Shield-1.

Having established that TgDrpA is required for apicoplast fission, we sought to elucidate the mechanistic role of TgDrpA in this process. T. gondii daughter cells form within mother cells by internal budding (Figure 7A, [8]). The scaffold of the daughter buds consists of subpellicular microtubules and an inner membrane complex (IMC), flattened membrane sacs that are stabilised by a network of IMC proteins. Daughter buds form near the centrosomes and extend towards the basal end of the mother cell, incorporating the nucleus and various organelles, before contracting at the base to enclose the newly formed daughter (Figure 7A; [22, 23]). MORN1 is a recently identified protein that localises to a ring at the growing end of the daughter bud and ultimately forms the basal complex (Figure 7A; [8, 21, 22, 24]). In addition, MORN1 localises to the centrocone, an elaboration of the nuclear envelope that contains the mitotic spindle and localises adjacent to the centrosome. Immediately before apicoplast fission, the apicoplast typically adopts a “U”-shape, with the ends of the daughter bud localising to the base of the “U” (Figure 7B; Figure 2C; [6]). We have previously hypothesised that the growing daughter bud functions in apicoplast division, possibly generating the force necessary for fission [5, 6]. To examine this, we performed time-lapse imaging on the apicoplast RFP/DrpA^K42A mutant cell line grown in the absence of Shield-1 using YFP-MORN1 as a dynamic marker for the growth of the daughter bud (Figure 7D). After 35 minutes of imaging, each cell contains a single apicoplast and two MORN1 rings (Figure 7D; Supplemental Movie 4). After 70 minutes, the MORN1 rings have moved towards the basal ends of the parasites. The apicoplast in the bottom parasite has adopted a “U” shape, with the two MORN1 rings localised at the base of the “U”. As daughter budding proceeds, the “U”-shape of the apicoplast elongates, and after 90 minutes, the MORN1 rings have moved further towards the basal end of the mother cell, and the apicoplast in this parasite has divided. After 100 minutes, the MORN1 ring has extended further still towards the basal end of the cell, while the apicoplasts remain at the apical end of the forming daughters, associated with the centrosomes/centrocones. Soon after, the cells undergo cytokinesis, with the MORN1 rings becoming the basal complex of the newly formed daughter cells (Supplemental Movie 4).

Figure 7. TgDrpA has a direct role in apicoplast fission.

(A) Schematic of a dividing T. gondii parasite. The daughter bud (blue) comprises of subpellicular microtubules and the inner membrane complex (IMC). Subpellicular microtubules and an IMC are also present in the mother cell. MORN1 (lilac) labels the basal end of the mother cell (basal complex), the growing ends of the daughter buds (MORN1-ring), and the centrocone, an extension of the nuclear envelope that localises near the centrosomes (red). A pre-divided, “U”-shaped apicoplast (green) and a pre-divided nucleus (grey) are also depicted.

(B-C) Immunofluorescence assay of DD-DrpA^K42A parasites grown in the absence (B) or presence (C) of Shield-1, co-expressing apicoplast RFP (red) and labelled with an anti-IMC antibody (green). The inner membrane complex of daughter cell buds can be seen inside the mother cells, with the ends of the daughter IMC localising at the base of the dividing, “U”-shaped apicoplasts.

(D) Time-lapse imaging of DD-DrpA^K42A parasites grown in the absence of Shield-1, co-expressing FNR-RFP and YFP-MORN1. YFP-MORN1 labels the basal end of the parasites, the centrocone (an extension of the nuclear envelope that localises adjacent to the centrosome) and the growing bud of the daughter cell.

(E) Time-lapse imaging of DD-DrpA^K42A parasites grown in the presence of Shield-1, co-expressing FNR-RFP and YFP-MORN1. Purple arrows in 90 and 110 minute samples represent the direction of daughter cell budding. Note that between the 90 and 110 minute time points, this cell rotates almost 180°. Arrowheads at 125 minutes depict the MORN1-labelled basal complex, and white arrows at 125 and 130 minutes label a point of attachment of the apicoplast to the centrosome/centrocone that is released 5 minutes later. Scale bars are 2 μm.

These data suggest that growth of the daughter bud is involved in generating the “U”-shaped apicoplast, and that apicoplast division occurs when the MORN1 ring localises to the base of the “U”. This raises two hypotheses for the role of DrpA in apicoplast fission: DrpA may function in formation of the daughter bud, which in turn is necessary for apicoplast fission, or DrpA functions directly in apicoplast fission and has no effect on daughter bud formation. To test this, we visualised the daughter bud of DrpA^K42A mutants co-expressing apicoplast RFP by immunofluorescence assays with an anti-IMC antibody, growing parasites in the presence of Shield-1. In parasites where apicoplasts are unable to divide, the daughter bud appears normal (Figure 7C). To examine daughter bud formation and apicoplast fission in a more dynamic way we performed time-lapse imaging on the previously described YFP-MORN1 and apicoplast-targeted RFP cell line grown in the presence of Shield-1 (Figure 7E). We examined a four-cell vacuole grown for 6 hours in Shield-1 where apicoplast morphology initially appeared normal. After 40 minutes, we see the development of MORN1 rings at the apical end of the parasites, near the apicoplast (Figure 7E; Supplemental Movie 5). After 90 minutes we see the formation of “U”-shaped apicoplasts in each parasite, with the two MORN1 rings for each parasite localising at the base of the “U”. Purple arrows indicate the direction of daughter cell budding. Twenty minutes later, the “U”-shaped apicoplasts have stretched out further, but have not divided. After 125 minutes, cytokinesis has begun. In the top cell, the MORN1 rings have contracted to close off the newly formed daughter cells (arrowheads), but the apicoplast is clearly not divided. The apicoplast ends remain attached to centrosomes/centrocones (white arrow). Five minutes later, one apicoplast branch has released from the centrosome/centrocone, and appears to localise to the basal end of the daughter cell (white arrow). After 195 minutes of imaging, the apicoplasts of all 8 newly formed daughters localise to the basal end of the cell, and are no longer connected to the centrosomes. We conclude that after ablation of DrpA, daughter cell budding is normal. Furthermore, apicoplast division appears normal up to the point of organellar fission. Together with the localisation of native DrpA to the base of the “U”-shaped apicoplast (Figure 2B-C), these data suggest a direct role for DrpA in apicoplast fission. Although daughter budding has a role in extension of the apicoplast into an elongated “U” shape [5, 6], this extension is not sufficient to mediate apicoplast fission.

Discussion

The data presented in this study indicate that TgDrpA functions in apicoplast fission. Curiously, apicomplexan DrpA proteins are phylogenetically distinct from ARC5 dynamins that play a similar role in chloroplast division in plants [2]. This suggests that DrpA evolved from a host cytoplasmic dynamin that was recruited to endosymbiont division independently of ARC5, a remarkable example of convergent evolution. Dynamins appear to be promiscuous membrane-modifying enzymes, whose cellular function is largely dependent on the membranes to which they are recruited. It is therefore of considerable interest to identify the mechanisms of TgDrpA recruitment to the apicoplast. The apicoplast progenitor was a red alga [25, 26] that likely had an ARC5 and FtsZ-based chloroplast division apparatus. Why was it necessary to evolve a second dynamin to replace the function of these ubiquitous plastid division proteins? A crucial step in the establishment of a successful endosymbiotic organelle is a mechanism to correctly divide and partition within the host cell [27]. Compared to their red algal precursors, apicoplasts are surrounded by two additional membranes, the outermost of which is an endosomal membrane. During the early phase in the endosymbiotic relationship, ARC5 and FtsZ were encoded by the red algal genome and functioned in division of the two innermost membranes. A separate mechanism was required to divide the outer membranes, and our data suggest that DrpA may have been recruited for this role (possibly from an original role in the endosomal pathway). It is not clear why ARC5 and FtsZ were subsequently lost, but we speculate that the general reduction in size of the apicoplast that occurred upon loss of photosynthesis may have simplified this division process to the extent where DrpA alone was sufficient to mediate fission. It is noteworthy that DrpA homologues are present in Cryptosporidium species (Figure 1), Apicomplexa that are thought to have lost their apicoplast. It is conceivable that Cryptosporidium DrpA has acquired a novel function, but equally possible that the role of DrpA in apicoplast fission evolved more recently. Examining the functions of DrpA homologues in Cryptosporidium and other alveolates should provide clues to how and when its role in apicoplast fission evolved.

We demonstrate that ablation of DrpA function results in specific defects in apicoplast fission. We also demonstrate a role for the extension of daughter buds in generating the “U”-shape of apicoplasts that immediately precedes fission. Likely this process is mediated by growth of subpellicular microtubules, and we have previously shown that treatment of T. gondii with the microtubule disrupting agent oryzalin, which disrupts subpellicular microtubules, inhibits apicoplast fission [6]. Based on these observations, we have argued that the force generated by daughter budding is required for apicoplast fission [5, 6]. We now extend this model to include a role for DrpA. Although strong DrpA labelling is apparent at the site of apicoplast fission, DrpA associates with the apicoplast at all points in the cell cycle (Figure 2A-C). Why, then, does DrpA-mediated fission only occur during daughter-cell budding? In Figure 8 we present a model for how fission and budding are mechanistically coordinated. Apicoplast ends become associated with centrosomes, which anchors them to the apical end of the cell. Soon after, daughter budding commences, with the ends of the forming daughter cells (as marked by the MORN-1 ring) stretching and consequently constricting the apicoplast. In yeast, dynamin-mediated mitochondrial fission requires the assembly of dynamin spirals around the organelle at the site of fission [28]. It is thought that these spirals form at sites where mitochondria are already constricted [28, 29]. A recent study of the DrpA homologue in the apicomplexan Plasmodium falciparum (PfDYN2) demonstrated that PfDYN2 is capable of self-association and GTP hydrolysis [30], suggesting that DrpA likely functions in a similar way to other characterised dynamin-related proteins. We propose that stretching of “U”-shaped apicoplasts by forming daughter cells constricts the organelle to the extent where DrpA can assemble in spirals. As in other dynamin-based constriction models, GTP hydrolysis causes extension of the DrpA spiral (Figure 2C, arrowheads) and results in further constriction of the apicoplast, until organellar fission is complete. In such a way, parasites can coordinate apicoplast fission with cytokinesis. Under this model, daughter cell budding is responsible for initial constriction of the apicoplast, DrpA functions in the actual fission process, and centrosome attachment mediates correct segregation of the apicoplast into daughter cells. Ongoing studies seek to experimentally test this model.

Figure 8. Model for the role of TgDrpA in apicoplast fission.

Before daughter cell budding, centrosomes (red) anchor the apicoplast (green) to the apical end of the cell. Extension of the daughter bud (lilac) results in formation of a “U”-shaped apicoplast, constricting the apicoplast at the base of the “U”. DrpA is then able to assemble into functional, multimeric spirals (red) that hydrolyse GTP and function in the actual fission of the apicoplast. When functional DrpA is present (top right), apicoplasts divide and properly segregate with centrosomes into daughter parasites. In the absence of functional DrpA, apicoplast fission is unable to occur. Extension of the daughter bud generates the force necessary to release apicoplasts from centrosomes, with apicoplasts then localising to the basal end of the newly formed daughter parasites (bottom right). Apicoplasts in the two newly formed daughter cells remain connected and elongate before the next round of cell division.

Experimental Procedures

Parasite culture and manipulation

Parasites were grown in human foreskin fibroblasts as previously described [31]. We grew DrpA^K42A mutant parasites in 0.1 – 0.2 μM Shield-1 (a kind gift from Tom Wandless, Stanford U.) where applicable. Fluorescence growth assays were performed as previously described [19]. Cloning and plasmid construction are described in the Supplemental Data.

Phylogenetic analyses

We generated multiple sequence alignments of dynamin homologues from a range of organisms using ClustalX. Sequences used for alignments were identified on publicly available databases. Phylogenetic analyses were performed using PHYLIP as previously described [32]. The GenBank accession number for TgDrpA is FJ264918. Accession numbers for other proteins used in the alignment are listed in the Supplemental Data. Alignments are available from the authors upon request.

Protein analyses

Western blotting and pulse-chase analyses were performed as described previously [33]. For Western blotting, we used anti-HA antibodies (Roche) at a dilution of 1:100 and anti-GRA8 (a kind gift from Gary Ward, U. Vermont) at 1:200,000.

Microscopy

Fluorescence and live cell images were acquired using both a DM IRBE inverted epifluorescence microscope (Leica) fitted with a 100X oil immersion objective lens (PL APO 1.40 NA) and an IX71 inverted epifluorescence microscope (Olympus) with a 100X oil immersion lens (UPlanApo 1.35 NA). Images on the Leica microscope were recorded using a Hamamatsu C4742-95 digital camera, and adjusted for brightness and contrast using Openlab software (Improvision). Images on the Olympus microscope were recorded using a Photometrics Coolsnap HQ camera and processed using SoftWoRx software (Applied Precision). Time-lapse imaging was performed in a humidified chamber heated to 37°C with 5% CO₂, with cells grown in Mattek glass-bottom culture dishes. Images were processed to account for cell drifting. Photobleaching of RFP in the apicoplast was performed on the Olympus microscope, using ten 300 ms pulses with a 488 nm laser on a specified, diffraction-limited region. Conditions for bleaching of RFP in the mitochondrion was identical, except that we used 1 second pulses. Immunofluorescence assays were performed as previously described [33]. We used anti-ACP antibodies at a dilution of 1:2000, anti-HA at 1:50 to 1:100, anti c-myc (Roche) at 1:500, anti-MIC5 at 1:500, anti-ROP4 (a kind gift from Gary Ward, U. Vermont) at 1:500 and anti-IMC (Mab 45.36; a kind gift from Gary Ward, U. Vermont) at 1:500 to 1:2000. For electron microscopy, cells were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 2 hours at room temperature, followed by fixation with 1% osmium tetroxide in 0.1 M cacodylate buffer (pH 7.4) for 2 hours on ice. Afterwards the samples were brought through a graded series of ethanol and subsequently infiltrated with increasing concentrations of Epon : ethanol (3:1; 1:1; 1:3 for 2 hours each and finally pure Epon overnight). Following change for fresh Epon, samples were polymerized at 60°C for 48 hours and sectioned as monolayers. Sections (60 nm) were collected on Formvar-coated, carbon-stabilized hexagonal 100 mesh copper grids and post-stained for 4 minutes with 20% (w/v) uranyl acetate in 70% (v/v) methanol/water followed by 2 minutes Reynolds's lead citrate staining [34]. The grids were examined in a transmission electron microscope Tecnai 12 (FEI Company, Eindhoven, The Netherlands) at 120kV. Images were recorded using a CCD camera (MegaView II, Olympus Soft Imaging Solutions GmbH, Münster, Germany). Image processing was done with Analysis 3.2 (Soft Imaging Systems GmbH, Münster, Germany).