Plasmodium falciparum, the causative agent of malaria, relies on a sophisticated protein secretion system for host cell invasion and transformation. Although the parasite displays a secretory pathway similar to those of all eukaryotic organisms, a classical Golgi apparatus has never been described. We identified and characterised the putative Golgi matrix protein PfGRASP, a homologue of the Golgi re-assembly stacking protein (GRASP) family. We show that PfGRASP is expressed as a 70 kDa protein throughout the asexual life cycle of the parasite. We generated PfGRASP-GFP-expressing transgenic parasites and showed that this protein is localised to a single, juxtanuclear compartment in ring-stage parasites. The PfGRASP compartment is distinct from the ER, restricted within the boundaries of the parasite and colocalises with the cis-Golgi marker ERD2. Correct subcellular localisation of this Golgi matrix protein depends on a cross-species conserved functional myristoylation motif and is insensitive to Brefeldin A. Taken together our results define the Golgi apparatus in Plasmodium and depict the morphological organisation of the organelle throughout the asexual life cycle of the parasite.
The intracellular parasite Plasmodium falciparum is the causative agent of malaria and responsible for over two million deaths each year (Butler, 2002). After an initial multiplication step in liver cells, the parasite invades and multiplies within red blood cells. To survive, the parasite extensively modifies the host cell by exporting parasitic proteins to the host cell cytoplasm and to its cell surface. It has recently been shown that protein transport from the parasite via the surrounding vacuole and into the host cell involves a pentameric motif conserved across the genus Plasmodium (Marti et al., 2004; Hiller et al., 2004). Secreted proteins are synthesised in the parasite cytoplasm and translocated into the ER. This is mediated through a classical hydrophobic leader sequence, although some can be considerably recessed (Papakrivos et al., 2005). In addition, a functional retrieval mechanism for soluble proteins such as PfBiP via the KDEL receptor PfERD2 has been demonstrated in this parasite (Elmendorf and Haldar, 1993a; Van Dooren et al., 2005). However, morphological and functional evidence for a Golgi apparatus and post-Golgi transport pathways in Plasmodium is still rather sparse. The Golgi apparatus is the central hub of the eukaryotic secretory machinery and plays a pivotal role in protein modification, processing and sorting. In general, the Golgi is organised into three functionally distinct regions: the cis-Golgi network (entry face), the Golgi stack and the trans-Golgi network (exit face). The ordered structure of this organelle is believed to reflect the requirement for the processing machinery to be compartmentalised for a sequential series of modification and sorting events (Farquhar and Palade, 1998).
The Golgi of protists offers an interesting insight into variations of the structure and organisation of this organelle. For example, representatives of two early-diverged eukaryotic lineages exhibit entirely distinct Golgi morphology and function: the diplomonad parasite Giardia intestinalis appears to have no stable Golgi compartment and there is no evidence for abundant protein modifications (Marti et al., 2003), whereas Trypanosoma brucei (kinetoplastids), the causative agent of sleeping sickness in humans, shows a classical stacked Golgi and high abundance of N-glycosylation in surface proteins (He et al., 2004). Further, Toxoplasma gondii, an apicomplexan parasite related to Plasmodium, has a single stacked Golgi apparatus as part of a highly polarised secretory pathway (Hager et al., 1999; Joiner and Roos, 2002; Pelletier et al., 2002). By contrast, morphology, organisation, function and even localisation of the Golgi in Plasmodium are still controversially discussed (Lingelbach, 1993; Elmendorf and Haldar, 1993b; Benting et al., 1994; Mattei et al., 1999; Bannister et al., 2000). Immunofluorescence and electron microscopy have not as yet provided unambiguous results as to the nature of the Golgi (Aikawa, 1971; Elmendorf and Haldar, 1993a; Van Wye et al., 1996; Bannister et al., 1990; Bannister et al., 2000; Bannister et al., 2003). For instance, it was shown that PfERD2, a Golgi marker protein that in mammalian systems is concentrated in the cis-Golgi, is localised to the perinuclear region of the parasite (Elmendorf and Haldar, 1993a; Van Wye et al., 1996; Noe et al., 2000). It was also shown that the distribution of the trans-Golgi marker PfRab6 (Novick and Brennwald, 1993) in the parasite is distinct from PfERD2 (De Castro et al., 1996; Van Wye et al., 1996). It was concluded that in early blood stages the parasite displays a functional but primitive, unstacked Golgi with distinct compartments (Van Wye et al., 1996). This data was partially supported by electron microscopy. A typically stacked Golgi apparatus has not been identified in the parasite, but a discoid cisterna close to the nucleus has been described that was provisionally specified as a minimal Golgi apparatus (Bannister et al., 2000; Bannister et al., 2003). Previous fixation methods for Plasmodium have resulted in very poor preservation for all membranous structures and organelles, so the characterisation of the Golgi in fixed Plasmodium parasites must be treated with caution. To overcome the problem of poor organellar fixation, we used live-cell microscopy to visualise the Golgi apparatus.
In the present study, we characterised the P. falciparum orthologue of the Golgi re-assembly stacking protein (GRASP). GRASP proteins are peripheral membrane proteins involved in stacking of Golgi cisternae (Barr et al., 1997; Barr et al., 1998; Shorter et al., 1999). They are conserved from yeast to mammals and are Golgi-defining matrix proteins. Here we show that PfGRASP is expressed in blood stage parasites as a 70 kDa protein. It is localised to a compartment juxtaposed to the nucleus. Using green fluorescence protein (GFP)-tagged chimeric proteins we analysed morphology and development of the Golgi apparatus in vivo during the life cycle of the parasite. We also show that targeting of PfGRASP depends on a functional N-terminal myristoylation motif. In addition we used indirect immunofluorescence microscopy and specific antibodies against other Golgi marker proteins to analyse the spatial organisation and compartmentalisation of the Golgi apparatus in Plasmodium.
Materials and Methods
Cell culture and transfection of P. falciparum
Plasmodium falciparum asexual stages (3D7) were cultured in human 0+ erythrocytes according to standard procedures (Trager and Jensen, 1976). 3D7 parasites were transfected as described previously (Wu et al., 1996) with 100 μg purified plasmid DNA (Qiagen). Positive selection for transfectants was achieved using 15 nM WR99210, an antifolate that selects for the presence of the human dhfr gene (Fidock and Wellems, 1997).
Nucleic acids and constructs
PfGRASP was identified using the online BLAST tool implemented in the Plasmodium database (www.plasmodb.org). The cDNAs encoding PfGRASP and PfERD2 (PlasmoDB PF13_0280) were generated using total RNA of parasites in a reverse-transcription reaction (Invitrogen) with specific oligonucleotides. Subsequently, the genes were PCR amplified with Vent Polymerase (Stratagene) and sequenced to detect unwanted mutations. The following primers were used: PfGRASP-S, 5′-GCGCGGTACCATGGGAGCAGGACAAACG-3′ and PfGRASP-AS, 5′-GCGCCCTAGGCAATATGTTCTTTCTTAC-3′; PfERD2-S, 5′-GCGCGGTACCATGAATATATTTAGACTG-3′ and PfERD2-AS, 5′-GCGCCCTAGGTTTTACTTCACCATTAAATG-3′.
To generate transfection vectors expressing C-terminal GFP fusion proteins, PCR products were digested with KpnI and AvrII (bold) and cloned into pARL1a- (Crabb et al., 2004). GFP was previously inserted into the XhoI site of pARL1a- with an additional 5′ AvrII site. To alter the putative myristoylation site in PfGRASP we changed the amino acid glycine at position 2 to alanine by using the sense primer (PfGRASP-G2/A-GFP: 5′-GCGCGGTACCATGGCAGCAGGACAAACG-3′) and fused the fragment in-frame with GFP as described above.
Antisera and immunoblotting
Mouse antisera were raised against synthetic polypeptides of PfGRASP (N529SSKMDNITKGTYIN543) and PfRab6 (PlasmoDB, PF11_0461, N181EANVVDIQLTNNSNKND197). Other antibodies used in immunodetection were rabbit anti-PfERD2 (obtained through the Malaria Research and Reference Center, NIH, MRA-72; accession number NP705420) (Elmendorf and Haldar, 1993a), monoclonal anti-GFP (Roche) and anti-PfBiP (Malaria Research and Reference Center, MR-19; AAA29501) (Kumar et al., 1991). For immunoblots, parasite proteins from a synchronised culture were separated on 10% SDS-PAGE gels and transferred to nitrocellulose membranes (Schleicher & Schuell). Anti-PfGRASP was diluted 1:1500 in phosphate-buffered saline (PBS). The secondary antibody was sheep anti-rabbit IgG horseradish peroxidase (Sigma) and used at a 1:3000 dilution. The immunoblots were developed by chemiluminescence using ECL (Amersham).
Immunofluorescence and analysis of GFP-expressing parasites
Green fluorescence of GFP-expressing transfectant cell lines was observed and captured in live cells through the erythrocytic life cycle every 8 hours using a Leica Axioskop 2 and OpenLab software (Improvision) or a Leica confocal microscope (TCS SP2) and Leica confocal software (LCS). Immunofluorescence assays were performed on fixed parasites as previously described (Tonkin et al., 2004). The primary antibody dilutions used in 3% BSA: rabbit anti-PfERD2 (1:500), rabbit anti-PfBiP (1:1000), mouse anti-PfGRASP (1:1000) and mouse anti-PfRab6 (1:1000). The cells were incubated with Cy3 anti-mouse IgG antibodies (Molecular Probes), Alexa-594 goat anti-rabbit IgG antibodies (Molecular Probes) and DAPI (1:1000, Roche). Dual-colour fluorescence images were captured using a Leica Axioskop 2 microscope or a Leica confocal microscope.
Brefeldin A treatment
Synchronised parasites were incubated with 5 μM Brefeldin A from a 10 mM ethanol stock solution. Control parasites were incubated with ethanol to exclude any morphological alterations due to the presence of ethanol. Brefeldin A was removed after 24 hours and the parasites were cultured for another 24 hours to ensure viability after treatment.
Synchronised parasites were harvested at 8 hour intervals and total RNA was isolated using Trizol (Invitrogen). Total RNA (1 μg) was reverse transcribed and quantified in a LightCycler (Roche) using Quantitect SYBR Green RT-PCR Kit (Qiagen) and gene-specific primers. Reaction conditions were 50°C for 20 minutes, 95°C for 15 seconds, 55°C for 20 seconds and 72°C for 20 seconds. A stage-specific control was performed using the early upregulated etramp gene (early-transcribed membrane protein; PlasmoDB PF10_0019) (Spielmann et al., 2003). Serial dilutions of 3D7 total RNA were used as a standard reference control. The relative transcription ratios through the asexual life cycle of grasp and etramp compared with the reference gene actin (PlasmoDB PFL2215w) were calculated. Primers used were as follows: actin-S, 5′-TGCACCACCAGAGAGAAAAT-3′; actin-AS, 5′-ACTTGGTCCTGATTCATCGT-3′; etramp-S, 5′-ATGTTTTAGGAGGTAGTAGTGGATCAG-3′; etramp-AS, 5′-GCGAGAACAGAAGCTACGGAAG-3′; grasp-S, 5′-AAAGTGCAAATTCAACTTAATGATTCA-3′; grasp-AS, 5′-TCTTTAATTTGACAAGGTGGAATTTTATG-3′.
Identification of the Golgi-stacking protein PfGRASP and primary sequence analysis
A few intrinsic Golgi markers are conserved across the eukaryotic lineage, the most prominent ones being members of the GRASP family. We identified a single Pfgrasp homologue in the genome of the parasite (chr10.phat_190, www.plasmodb.org). The gene is located on chromosome 10, encompasses 1869 base pairs in length and encodes a 574 amino acid protein. Contrary to the annotation (one-exon structure; PlasmoDB PF10_0168), PfGRASP is encoded by a two-exon gene, whereas the first exon encodes the first 11 amino acids. This short N-terminal extension displays a cross-species conserved putative myristoylation site (NMT MYR Predictor) (Bologna et al., 2004) that might facilitate membrane attachment (Fig. 1A,B). Cloning and sequencing of the gene from both genomic DNA and cDNA isolated from infected red blood cells confirmed the gene structure (data not shown). The Plasmodium protein displays an overall identity of 22% with human and rat (Rattus norvegicus) GRASP55 and 19% identity with the GRASP protein identified in the related apicomplexan parasite Toxoplasma gondii (Pelletier et al., 2002) (Fig. 1B). The homology is restricted to the N-terminal 220 amino acids. This domain is implicated in dimerisation and trans-oligomerisation in mammalian GRASP proteins (Barr et al., 1998; Wang et al., 2003; Wang et al., 2005). PfGRASP has a Mr of 69×103 and is larger than all other GRASP proteins. This is due to an ∼120 amino acid extension of the unconserved C-terminal domain of PfGRASP (residues 219-573). Although these enlargements of proteins (either because of tandem repeat insertions and/or homopolymer runs) are well known in P. falciparum, the biological significance is unknown (Gardner et al., 1998; Pizzi and Frontali, 2000; Brocchieri, 2001). In the mammalian system this poorly conserved C-terminus is characterised by serine/proline richness (e.g. 33% in human GRASP55) and is a known substrate for mitotic phosphorylation (Shorter and Warren, 2002; Wang et al., 2005). By contrast, the C-terminal domain of PfGRASP is only slightly enriched in serine and proline (12%). Nevertheless the C-terminal domain does encompass 20 putative serine phosphorylation sites (NetPhos 2.0 server) (Blom et al., 1999).
PfGRASP is transcribed and expressed throughout the asexual life cycle
Conclusive microarray and proteomic data on Pfgrasp is not available (Bozdech et al., 2003; Le Roch et al., 2003; Florens et al., 2002). Therefore, we investigated Pfgrasp transcription and expression throughout the asexual life cycle. Pfgrasp gene transcription was analysed by real-time RT-PCR throughout the asexual blood stages of the parasite. A stage-specific control was performed using the early-transcribed gene etramp (Spielmann et al., 2003). Although transcription of etramp peaks in early stages (compared with levels of actin transcription), Pfgrasp is transcribed broadly across the asexual life cycle (Fig. 2A). This is reflected in the protein expression pattern of PfGRASP. Stage-specific immunoblots using PfGRASP-specific antibodies on synchronised parasite pellets show a major band at 70 kDa throughout the asexual life cycle (apparent Mr of PfGRASP is 69×103, Fig. 2B).
PfGRASP-GFP-expressing parasites and subcellular localisation
To morphologically characterise the Golgi in live parasites, a transgenic cell line expressing the intrinsic Golgi marker PfGRASP as a chimeric protein with GFP was generated. Full-length PfGRASP was fused to GFP and transfected into parasites. To confirm expression of the fusion protein, western blot analysis with anti-GFP antibodies were performed on transgenic parasites. A band corresponding to PfGRASP-GFP of ∼100 kDa was detected in the transgenic cell line but not in wild-type (WT) parasites (Fig. 3A, apparent Mr of PfGRASP-GFP is 97×103). Additionally, the antibodies recognized two smaller proteins, presumably GFP breakdown products previously described for GFP-expressing parasites (Waller et al., 2000). Antibodies specific to PfGRASP recognized the endogenous protein of ∼70 kDa in both the WT and transgenic cell line. In the transgenic parasite line an additional band of 100 kDa corresponding to the PfGRASP-GFP fusion protein was detected (Fig. 3B).
We investigated the localisation of PfGRASP-GFP by either fluorescence microscopy of live cells expressing a green fluorescent fusion protein or indirect immunofluorescence of fixed parasites. In trophozoites (<24 hours post invasion) PfGRASP-GFP was localised to two tightly defined compartments within the parasite juxtaposed to the nucleus with some cytoplasmic background fluorescence (Fig. 4A and supplementary material Movie 1). Importantly, PfGRASP specific antibodies used on either WT or PfGRASP-GFP expressing parasites reveal a similar staining pattern (Fig. 4B and Fig. 5A). Low levels of unspecific staining can be detected and may be a result of fixation.
Colocalisation of PfGRASP-GFP with ER and Golgi marker proteins
To establish PfGRASP-GFP as a marker for the Golgi and to generate more data with respect to Golgi architecture in Plasmodium we performed immunofluorescence assays using specific antibodies against previously described marker proteins for the ER and Golgi. Specifically, we used antibodies against the ER marker BiP (Elmendorf and Haldar, 1993a), the cis-Golgi marker ERD2 (Elmendorf and Haldar, 1993a; Van Wye et al., 1996) and the trans-Golgi marker Rab6 (De Castro et al., 1996; Van Wye et al., 1996).
ERD2 (ER-retention-defective complementation group 2), a seven-transmembrane-spanning receptor (Semenza et al., 1990; Lewis et al., 1990; Pelham et al., 1993) recognizes a C-terminal motif of X-aspartate-glutamate-leucine (XDEL) on certain soluble ER proteins and retrieves those proteins that have leaked to the Golgi complex back to the ER (Wilson et al., 1993; Lewis and Pelham, 1992). It is used as a cis-Golgi marker and can be localised to a single site adjacent to the nucleus in early (10 hours post-invasion) stages of the parasite, whereas in trophozoites two foci of PfERD2 are present (Elmendorf and Haldar, 1993a). Immunofluorescence assays with anti-PfERD2 antibodies show colocalisation to PfGRASP-GFP suggesting identity of these cellular compartments (Fig. 5B). Using antibodies directed against the ER marker protein PfBiP the ER could be visualised in young parasites as a ring of fluorescence encircling the nucleus of the parasite (Fig. 5C and supplementary material Movie 2). Parts of this membranous system, which are reminiscent of the tip of two protrusions extending from the nuclear envelope described recently for a PfBiP-GFP fusion in Plasmodium (Van Dooren et al., 2005), are in close proximity to the PfGRASP-GFP-defined compartment. Altogether, the close association of parts of the ER with the PfGRASP-defined compartment and the colocalisation of ERD2 validates PfGRASP as a Golgi marker.
Similar studies were performed using anti-PfRab6 antibodies. Rab6 are small GTPases used as markers of trans-Golgi compartments in various systems including Plasmodium (De Castro et al., 1996). Immunofluorescence patterns of early parasite stages (<24 hours post invasion) resulted in several (1-4) distinct fluorescent sites different to those defined by PfGRASP (Fig. 5D). In some cells partial association of PfRab6 with the PfGRASP-defined compartments could be observed (supplementary material Movie 3). This suggests that the trans-Golgi topology is distinct from those defined by PfGRASP.
Golgi dynamics during the asexual life cycle of Plasmodium
The Golgi is multiplied and distributed during mitosis and cellular division. Schizogony of P. falciparum requires duplication as well as multiplication of the Golgi within 48 hours. We used PfGRASP-GFP expressing parasites to visualise Golgi dynamics throughout the asexual life cycle of Plasmodium. In early parasite stages (8-16 hours post invasion) one single oval Golgi compartment in close proximity to the nucleus can be observed (Fig. 6a-b). As the parasite develops, but prior to nuclear division, a second Golgi is formed (24 hours post invasion, Fig. 6c). As nuclear division commences (32 hours), further Golgi multiplication occurs (Fig. 6d). This results in multiple Golgi compartments ensuring that each merozoite inherits one Golgi (Fig. 6e-g). These results were confirmed using a second transgenic cell line expressing the cis-Golgi marker ERD2 as a GFP fusion protein (supplementary material Figs S1 and S2). It is interesting to note that the expression of PfERD2-GFP also results in additional perinuclear staining reminiscent of the ER. This finding agrees with previous work showing the presence of low levels of PfERD2 in the ER (Van Wye et al., 1996). Owing to the additional perinuclear staining, (parts of) the ER can be visualised and followed during schizogony (supplementary material, Fig. S2).
Subcellular localisation of PfGRASP depends on an N-terminal putative myristoylation motif
Golgi localisation of GRASP proteins requires N-terminal myristoylation (Barr et al., 1997; Barr et al., 1998; Shorter et al., 1999). The covalent attachment of fatty acid chains facilitates localisation of water-soluble proteins to a membrane after its synthesis in the cytosol. This modification involves the attachment of myristic acid via an amide linkage to an N-terminal glycine. Phylogenetic analysis suggests this myristoylation site is conserved in PfGRASP (Fig. 1). To determine if targeting of PfGRASP depends on the putative N-terminal myristoylation site, this motif was disrupted by point mutation. A GFP-fusion construct was generated where the amino acid glycine was replaced by alanine at position 2 of the protein sequence (PfGRASP-G2/A-GFP) and transfected into parasites. Perturbation of the putative N-terminal myristoylation site resulted in the accumulation of the protein in the cytoplasm of the parasite (Fig. 7). This data suggests that targeting of PfGRASP requires glycine at position 2 presumably because it serves as a substrate for the N-myristoyltransferase (PlasmoDB PF14_0127) (Gunaratne et al., 2000) linking myristic acid via an amino bridge to the protein backbone and thereby facilitating membrane attachment.
PfGRASP distribution is insensitive to BFA treatment
Brefeldin A (BFA) is a fungal metabolite that specifically inhibits anterograde transport from ER to Golgi compartments (Lippincott-Schwartz et al., 1989; Orci et al., 1991). It allows fractionation of Golgi proteins. Proteins like ERD2 or N-acetylmannosidasetransferase are relocated to the ER upon BFA treatment (Elmendorf and Halder, 1993a; Pelletier et al., 2002). By contrast, structural Golgi components like the GRASP protein family are insensitive to BFA (Nakumura et al., 1997; Pelletier et al., 2002; Seemann et al., 2002). After BFA treatment of PfGRASP-GFP expressing parasites, we observed no relocation of the fusion protein to the ER (Fig. 8).
A GRASP homologue in P. falciparum
In recent years, Golgi matrix proteins have been characterised that define structure and identity of the Golgi complex (Short and Barr, 2000). GRASP proteins are implicated in cisternal stacking and signalling (Barr et al., 1997; Shorter at al., 1999; Seeman et al., 2000) and have been used for investigating Golgi development and architecture in protozoa (Pelletier et al., 2002; He et al., 2004). GRASP proteins interact with the golgin family of coiled-coil proteins providing an exoskeleton for this organelle. To date two GRASP paralogues are described in the mammalian system: the cis-Golgi marker GRASP65 interacting with GM130 and medial-Golgi marker GRASP55 interacting with Golgin45 (Barr et al., 1997; Barr et al., 2001). We identified a single grasp homologue in the genome of P.falciparum, established its expression profile and used this putative Golgi matrix protein to analyse Golgi organisation and dynamics during the asexual life cycle of the parasite. The homologue of GRASP in P. falciparum has an apparent Mr of 69×103 and is slightly more similar to mammalian GRASP55 than GRASP65 (34% vs 30% similarity). PfGRASP has a conserved domain structure compared with known GRASP proteins: (1) a N-terminal myristoylation motive; (2) a highly preserved N-terminal GRASP domain known to be both necessary and sufficient for dimerisation and trans oligomerisation (Wang et al., 2005); (3) a short middle so-called GM130 binding domain; and (4) a highly divergent C-terminus. The proline/serine-rich C-terminus of mammalian GRASP proteins is implicated in mitotic regulation via phosphorylation and is a substrate of caspases during apoptosis (Wang et al., 2005; Lane et al., 2002). Putative mitotic kinases like Cdc2-related enzymes are present and characterised in the parasite (Kappes et al., 1999; Le Roch et al., 2000) and might recognize phosphorylation sites in the C-terminal domain of PfGRASP. The region currently known to be important for GM130 binding is conserved in GRASP65 and GRASP55 and could be mapped to 194GYGXXHRI201 (Barr et al., 1998), although binding of GM130 to GRASP55 might not occur in vivo (Shorter et al., 1999). It is interesting to note that this highly conserved region in mammals and yeast is altered in P. falciparum to 205AYGXXHKI212 (Fig. 1) suggesting differences in the PfGRASP-interacting protein complex. Coincidentally, only coiled-coil proteins with very low similarity to GM130 and Golgin45 could be identified in the genome of the parasite.
Defining the Golgi
Cell lines expressing GFP fusion proteins allow localisation and tracking of the tagged protein in live cells during the cell cycle. This transgenic approach circumvents fixation of the cells and therefore might prevent ultrastructural disturbance as exemplified recently (Van Dooren et al., 2005). Using fluorescence microscopy on GFP-tagged PfGRASP we localised the protein within distinct foci in the parasite. Minor levels of background fluorescence (supplementary material Movie 1) point to a cytoplasmic pool of PfGRASP-GFP as described previously for mammalian GRASP65 (Ward et al., 2001). It was suggested that GRASP65 shuttles between a cytoplasmic pool and Golgi membrane association. Up to 30% of the endogenous GRASP65 (and up to 55% of GRASP65-GFP in the transgenic cell line) was distributed in the cytoplasm (Ward et al., 2001).
We validated the subcellular localisation of the fusion protein by using PfGRASP specific antibodies (Fig. 5A) and established PfGRASP as a Golgi marker. This is the first time that the Golgi apparatus in Plasmodium has been visualised in live cells. To further analyse the extent of Golgi compartmentalisation, the ER marker BiP, the cis-Golgi marker ERD2 and the trans-Golgi marker Rab6 were used in colocalisation experiments. Firstly, we analysed the spatial relationship between the ER and the PfGRASP-defined compartments by using PfBiP specific antibodies. We show that the ER, which forms a perinuclear envelop with 1-2 protrusions (>24 hours post invasion), lies in close proximity to the PfGRASP-defined compartment (Fig. 5C and supplementary material Movie 2). These ER protrusions might reflect ER exit sites and it will be interesting to further analyse the functional relationship between the ER exit sites and the Golgi.
ERD2 recognizes a C-terminal XDEL motif on certain soluble ER proteins in the cis-Golgi and retrieves them into the ER (Wilson et al., 1993; Lewis and Pelham, 1992). ERD2 was used as a cis-Golgi marker in Plasmodium and was localised to defined loci adjacent to the nucleus (Elemendorf and Haldar, 1993a). Further, these antibodies were used in immunofluorescence assays to analyse ERD2 distribution during schizogony, showing a close spatial relation between forming nuclei and ERD2 distribution in P. yoelii (Noe et al., 2000). Using the fluorescence of PfGRASP-GFP and anti-PfERD2-specific antibodies we showed that PfGRASP colocalises with the cis-Golgi marker ERD2 (Fig. 5B). As
PfGRASP appears to be more homologous to the mammalian medial-Golgi marker GRASP55, this colocalisation could argue for a limited compartmentalisation of the Golgi. It could also indicate that PfGRASP might be a functional equivalent of the cis-Golgi marker GRASP65 (therefore colocalising with ERD2) and the medial-Golgi is defined by an as yet unidentified PfGRASP homologue. In the absence of an appropriate medial-Golgi marker a conclusive answer remains elusive.
Another membrane-associated protein, the small GTPase PfRab6, was used to analyse Golgi architecture in the malaria parasite. Rab proteins are small, GTP-binding proteins that play a pivotal role in the regulation of vesicular trafficking in eukaryotic cells (Ward et al., 1997). Parasite-specific PfRab6 antibodies were used in indirect fluorescence microscopy assays and showed a localisation distinct from ERD2 in early stages (Van Wye et al., 1996). This is supported by our data showing a differential distribution of the PfGRASP-GFP-defined compartment and PfRab6 and in some cases some loose association (Fig. 5D and supplementary material Movie 3). This differential distribution was used as an indication for a functional but primitive and unstacked Golgi with distinct compartments in early stages of the parasite (Van Wye et al., 1996). Another explanation for the loose association of Rab6 and the Golgi might be the multifunctional role of Rab6 in vesicular transport. In the mammalian system it could be shown that Rab6 (with its two isoforms) does not only function in intra-Golgi transport (Martinez et al., 1994) but is also required for protein recycling between endosomes and the trans-Golgi network (Mallard et al., 2002; Stedman et al., 2003) and even to the ER (White et al., 1999). In yeast, the Rab6 homologue Ypt6p has also been implicated in multiple roles, for example the retrieval of proteins from endosomes to the trans-Golgi network, intra-Golgi retrograde and Golgi to ER transport (Siniossoglou et al., 2001; Bensen et al., 2001; Luo and Gallwitz, 2003). Therefore, the precise function of Rab6 and its associated compartments in the parasite have to be further investigated.
We could show that PfGRASP remains restricted within the parasite boundaries throughout the asexual life cycle, as shown previously for PfERD2 (Fig. 6, supplementary material Fig. S2) (Elemendorf and Haldar, 1993a; Noe et al., 2000). An interesting observation by other groups is the localisation of the Golgi markers sphingomyelin synthase and two homologues of the COPII protein complex PfSar1p, PfSec31p beyond the parasite boundaries in the erythrocyte cytosol and membranes (Elmendorf and Haldar, 1994; Albano et al., 1999; Adisa et al., 2001; Adisa et al., 2002). This differential localisation (compared with PfERD2) was used to argue for a rather unusual Golgi organisation in Plasmodium, which at least functionally expands into the host cell cytoplasm. In support of this theory it was shown that the export of PfSar1p and PfSec31p to the erythrocyte cytosol was inhibited by BFA treatment (Adisa et al., 2002). It will be interesting to further investigate the putative export of COPII proteins and vesicles to the host cell cytosol and to analyse the spatial relationship between Golgi and COPII vesicles.
PfGRASP depends on a functional myristoylation site at the N-terminus of the protein. We produced a cytoplasmic variant of PfGRASP-GFP by substituting glycine at position 2 of the amino acid sequence to alanine (Fig. 7). This suggests that PfGRASP, like its mammalian counterparts, is a substrate for N-myristoylation (Barr et al., 1997; Wang et al., 2003). Additional palmitoylation could also play an enhancing role in membrane anchoring of PfGRASP as reported for GRASP55 (Kuo et al., 2000).
Brefeldin A (BFA), a fungal metabolite, is an important tool to study protein trafficking in the endomembrane system (Lippincott-Schwartz et al., 1989; Lippincott-Schwartz et al., 1990). It inhibits secretion of vacuolar protein transport, specifically blocking secretory export to post-Golgi compartments and has been shown to reorganise the Golgi apparatus (Tamaki and Yamashina, 2002; Nebenfuhr et al., 2002). BFA relocates Golgi enzymes like N-acetylglucosaminotransferase (NAGTI) to the nuclear envelope but does not effect distribution of Golgi matrix proteins like GRASP (Nakamura et al., 1997; Pelletier et al., 2002; Seemann et al., 2002). BFA has been used extensively to investigate protein trafficking in Plasmodium, and it was shown that the cis-Golgi marker ERD2 is redistributed to the ER upon BFA treatment (Elemendorf and Haldar et al., 1993; Wickham et al., 2001). In contrast to ERD2, PfGRASP distribution in Plasmodium is not affected, stressing its function as a Golgi exoskeleton matrix protein. This also implies that Golgi proteins in Plasmodium can be fractionated upon BFA treatment.
Biogenesis of the Golgi
Golgi multiplication in T. gondii and T. brucei has been analysed in great detail using video microscopy. In T. gondii the new Golgi grows by lateral extension followed by medial fission (Pelletier et al., 2002). In T. brucei the Golgi appears de novo and was shown to be rather an independent entity (He et al., 2004). We used the PfGRASP-GFP expressing cell line to follow Golgi biogenesis in Plasmodium monitoring parasites throughout the asexual life cycle because single-cell video microscopy over longer time periods is not yet technically feasible in P. falciparum. Nevertheless, our data indicates that Golgi duplication starts between 16 and 24 hours post-invasion prior to nucleus division, matching the increased metabolic activity of the parasite. After schizogony is completed each merozoite displays one single Golgi (Fig. 6g). It is important to note that our data do not exclude the possibility of an excess Golgi during cell division as described for Toxoplasma and Trypanosoma. Most of these additional Golgi disappear before the mother cell undergoes cytokinesis (Pelletier et al., 2002; He et al., 2004). Owing to technical restrictions our data does not allow a conclusive prediction of the assembly of new Golgi cisternae in the parasite. The new Golgi might be formed de novo in close contact with the ER export sites or by fission of the old Golgi. In fact, both mechanisms could be operational in the biogenesis of a new Golgi as proposed for Pichia pastoris (Bevis et al., 2002).
Form follows function?
The biosynthetic-secretory pathway leads outwards from the ER towards the Golgi and the cell surface with the Golgi apparatus as the central hub (Warren and Malhotra, 1998). The major function of the Golgi is thought to be the processing and sorting of newly synthesised proteins and lipids. The ordered architecture of this organelle allows a functional compartmentalisation. In the mammalian system, two GRASP proteins, GRASP55 and GRASP65, are implicated in Golgi membrane stacking. GRASP65 is located in cis-Golgi membranes whereas GRASP55 is located more towards the medial Golgi stack. This differential localisation was used to argue that the GRASP protein family helps determine stacking of different cisternal layers (Wang et al., 2003). It was suggested that organisms with only one grasp homologue (e.g. Saccharomyces cerevisiae and T. brucei) might possess a simpler Golgi organisation compared with the mammalian cell system (He et al., 2004). In the Plasmodium genome only one grasp homologue could be identified. This simple architecture might be mirrored in a reduced range of biochemical processes taking place in this compartment. For instance, the Plasmodium genome encodes only a cryptic glycosylation capacity with no O-linked glycosylation pathway and reduced N-glycosylation (Templeton et al., 2004). In this light, a high degree of compartmentalisation of the Golgi to allow a spatial as well as biochemical separation of glycoprocessing seems to be unnecessary.
We thank Giel van Dooren, Stuart Ralph, Heinrich Hoppe, Carsten Wrenger and Till Voss for critically reading the manuscript and for helpful comments. We thank Christian Drosten and Jürgen Sievertsen for advice regarding Real-Time PCR. The sequence from PfGRASP was obtained from the PlasmoDB website at http://plasmodb.org. PfERD2 and PfBiP antibodies were kindly provided by the MR4. This work was supported by the Deutsche Forschungsgemeinschaft Grant GI312 and Deutscher Akademischer Austauschdienst (DAAD). N.S. performed this work as part of a doctoral study at the Faculty of Biology, University of Hamburg, Germany. M.M. is supported by a postdoctoral fellowship from the Swiss National Science Foundation, AFC is supported by HHMI International Research Scholar Grants. T.W.G. is in receipt of an Emmy-Noether fellowship (DFG).