Toxoplasma gondii Rab6 Mediates a Retrograde Pathway for Sorting of Constitutively Secreted Proteins to the Golgi Complex*

Toxoplasma gondii relies on protein secretion from specialized organelles for invasion of host cells and establishment of a parasitophorous vacuole. We identifyT. gondii Rab6 as a regulator of protein transport between post-Golgi dense granule organelles and the Golgi.Toxoplasma Rab6 was localized to cisternal rims of the late Golgi and trans-Golgi network, associated transport vesicles, and microdomains of dense granule and endosomal membranes. Overexpression of wild-type Rab6 or GTP-activated Rab6(Q70L) rerouted soluble dense granule secretory proteins to the Golgi and endoplasmic reticulum and augmented the effect of brefeldin A on Golgi resorption to the endoplasmic reticulum. Parasites expressing a nucleotide-free (Rab6(N124I)) or a GDP-bound (Rab6(T25N)) mutant accumulated dense granule proteins in the Golgi and associated transport vesicles and displayed reduced secretion of GRA4 and a delay in glycosylation of GRA2. Activated Rab6 on Golgi membranes colocalized with centrin during mitosis, and parasite clones expressing Rab6 mutants displayed a partial shift in cytokinesis from endodyogeny (formation of two daughter cells) to endopolygeny (multiple daughter cells). We propose that Toxoplasma Rab6 regulates retrograde transport from post-Golgi secretory granules to the parasite Golgi.

The Golgi complex coordinates secretory protein maturation and sorting and is a central intermediary of bidirectional transport between exocytic and endocytic pathways. In most protozoa, the early secretory pathway is well conserved in that of other eukaryotes, performing the critical function of biosynthetic transport to organelles, the cell surface, and the extracellular environment. Members of Apicomplexa, a diverse phylum of obligate intracellular parasites, are distinct from other eukaryotes in harboring three unique polarized secretory organelles, termed micronemes, rhoptries, and dense granules. As exemplified in Toxoplasma gondii, sequential secretion from these organelles is essential for host cell invasion and the concomitant formation of an intracellular parasitophorous vac-uole (PV) 1 enveloping the parasite (1). Micronemes and rhoptries are apically tethered and secreted during attachment and penetration of a host cell. Dense granules in Toxoplasma secrete aggregates of soluble and transmembrane proteins constitutively at a basal level and are stimulated for enhanced release following parasite invasion, suggesting a regulated, but Ca 2ϩ -independent, component (2).
Immunohistochemical analysis indicates that the sorting of dense granule proteins from those targeted to micronemes and rhoptries occurs at the late Golgi cisternae or trans-Golgi network (TGN). Although rhoptries originate from a precursor organelle formed by the Golgi during parasite division, the process regulating biogenesis of Toxoplasma dense granules is completely unknown. Two hypotheses are proposed to account for the biogenesis of regulated secretory vesicles in specialized mammalian cells (reviewed in Ref. 3). In the selective aggregation model, secretory proteins aggregate in the TGN and are sorted for entry into secretory granules by receptors in the TGN membrane. Proteins that fail to aggregate may exit the TGN independently in constitutive secretory vesicles. In the sortingby-retention model, regulated proteins are retained and sorted from constitutive secretory cargo after formation of immature secretory granules. In this model, immature secretory granules budding from the TGN contain the bulk of biosynthetic cargo and undergo maturation by selective budding of constitutive secretory vesicles containing non-aggregate proteins. Toxoplasma dense granule protein aggregates may dissociate and insert post-translationally into target membranes only after secretion into the PV (4 -6). Understanding the biogenesis of Toxoplasma dense granules and the mechanisms involved in regulating dense granule protein transport will provide fundamental insight into the minimal requirements for sorting between regulated and constitutive secretory pathways in eukaryotes.
Molecular evidence and genomic sequencing efforts indicate that the molecular machinery regulating vesicular transport in Apicomplexa is partially conserved in other eukaryotes, particularly early in the secretory pathway. Morphologically, these protozoa possess an endoplasmic reticulum (ER) that is contiguous with the nuclear envelope and a Golgi apparatus that ranges from single dispersed cisternae in malarial parasites (genus Plasmodium) to a stacked apically oriented Golgi in Toxoplasma. Components of the vesicle budding, transport, and fusion machinery have been cloned in Toxoplasma, including N-ethylmaleimide-sensitive fusion protein, multiple Rab proteins (7), ARF1 (8), and subunits of the coatomer (9) and adapter complexes (10). Evolutionarily conserved sorting motifs function in the transport of proteins to rhoptries and micronemes (10,11). In contrast, targeting motifs have not been found in dense granule proteins, and heterologous soluble proteins expressed with an N-terminal signal sequence are constitutively secreted into the PV through dense granules (12).
To further define regulation of dense granule secretion, we sought to identify molecular effectors of dense granule protein transport. In conjunction with effector proteins, monomeric GTPases of the Rab family localize to distinct intracellular compartments and confer one level of specificity on vesicle transport and fusion by mediating tethering or docking of opposing membranes prior to pairing of SNAREs. The small GTPase Rab6 appears to function in intra-Golgi transport (13,14) and COPI/ARF1-independent retrograde transport from the Golgi to the ER in mammalian cells (15,16). The yeast homolog of Rab6, Ypt6p, was initially reported to function in early Golgi transport (17,18). Current evidence now indicates that Ypt6p aids recycling of endosome-derived vesicles with late Golgi membranes and is dependent upon a guanine nucleotide exchange factor complex, SNAREs, and the VPS (vacuolar protein sorting) complex (19 -21).
In this report, we show that Toxoplasma Rab6 regulates protein transport at a step between dense granules and the late Golgi cisternae and that overstimulation or inhibition of Rab6 function partially blocks constitutive secretion of proteins by disregulating their transport to dense granules. The effect of Rab6 overstimulation appears to enhance the brefeldin A-induced resorption of constitutively secreted cargo and Golgi cisternae to the ER. Furthermore, expression of either GTPactivated or dominant-interfering Rab6 mutants altered parasite cytokinesis, leading to a partial switch in daughter parasite budding to endopolygeny. These results suggest that Toxoplasma Rab6 mediates a retrograde pathway from post-Golgi secretory organelles to the late Golgi and may provide a link between Rab6 function and coupled mitosis and cytokinesis.

EXPERIMENTAL PROCEDURES
T. gondii Growth and Isolation-T. gondii RH strain cells were maintained in vitro by infection of Vero cell monolayers as previously described (22). Hypoxanthine-xanthine-guanine phosphoribosyltransferase (HXGPRT)-deficient T. gondii cells were obtained through the National Institutes of Health AIDS Research Reference Reagent Program (23).
T. gondii Nucleic Acid Isolation, cDNA Synthesis, and PCR Amplifications-T. gondii RNA was extracted with TRIzol (Invitrogen). cDNA synthesis was performed with Superscript II reverse transcriptase (Invitrogen) and primer oligo(dT 17 )XX. PCR amplification of cDNA was performed with Taq DNA polymerase (Roche Molecular Biochemicals) using degenerate oligonucleotides and targeting domains conserved either among Rab proteins in general (primer URABF, encoding the conserved PM3-binding site QLWDTAGQE) or Rab6 proteins in particular (primers AVVYDIT and AQEYNT). PCR cycling conditions were as follows: 94°C for 45 s, 52.5°C for 1 min, and 72°C for 1 min for 35 rounds of amplification. Amplification products from the URABF/ oligo(dT 17 )XX primer set were used as template for nested 3Ј-RACE/ PCR with the AIVVYDIT/oligo(dT 17 )XX primer set. Template for 5Ј-RACE/PCR was generated by reverse transcription with Rab6R and dCTP tailing with terminal deoxynucleotide transferase (New England Biolabs Inc.) and used for PCR with Rab6R and the RACE anchor primer (Invitrogen). Nested 5Ј-RACE/PCR was performed with Rab6I and the anchor primer. Nested PCR products were cloned into pGEM-T (Promega) and screened by colony hybridization. Sequence analysis of independent RACE clones confirmed the 5Ј-and 3Ј-ends of T. gondii rab6 cDNA, including an overlapping region of 133 bp. The full-length open reading frame (ORF) of T. gondii rab6 was amplified from cDNA using the Rab6F/Rab6R primer set, overlapping the initiation and stop codons, respectively. Sequences of independent clones confirmed the identity of the full-length 597-bp rab6 ORF to the 5Ј-and 3Ј-RACE clones, and Rab6 homology was confirmed by TBLASTN analysis. The T. gondii rab6 cDNA sequence has been deposited in the GenBank TM /EBI Data Bank (accession number AF228419). The sequences of the oligonucleotide primers used in the isolation of rab6 cDNA are as follows (where positions of deoxynucleotide base degeneracy are in parentheses and I is inosine): oligo(dT 17 Rab6R, 5Ј-AAC  AGA TCT TTC AGC AAG AGC AGG ATG-3Ј; Rab6I, 5Ј-GAT CCA CTT  GGT TGT ATT GAG-3Ј; Rab6HANsi, 5Ј-GTC AAC ATG CAT TAC CCA  TAC GAC GTC CCA GAC TAC GCG GAG GCG ACA GTT GCG-3Ј; and  R6rPacI, 5Ј-TCT TTA ATT AAC AAG AGC AGG ATG AAG TC-3Ј. Colony and Northern and Southern Blot Hybridization Analyses-Northern blot analysis was performed after electrophoresis of 20 g of total RNA on denaturing formaldehyde-agarose gels (1.2% Sea Kem GTG agarose (FMC Corp.), 20 mM HEPES, pH 7.8, 0.2 mM EDTA, and 6% formaldehyde) in HEPES/Formalin running buffer (20 mM HEPES, pH 7.8, 0.2 mM EDTA, and 0.1% formaldehyde). RNA samples were transferred to Zetaprobe membranes (Bio-Rad) in sodium phosphate transfer buffer (25 mM Na 2 HPO 4 /NaH 2 PO 4 ), pH 6.5.
Southern and colony blot hybridization analyses were performed as described (24). A rab6 cDNA fragment generated by PCR primers AIVVYDIT and AQEYNT was agarose gel-purified and labeled by random oligonucleotide priming with [␣-32 P]dCTP and the Rediprime II kit (Amersham Biosciences) to a specific activity of Ͼ10 8 cpm/g. Hybridizations were performed in high stringency buffer.
Phylogenetic Analysis-Sequences homologous to Rab6 were aligned using the Genetics Computer Group Wisconsin package. Phylogenetic reconstruction was performed using the PAUP program (phylogenetic analysis underln]using parsimony). Bootstrap analysis was carried out with 1000 replicates. The consensus tree derived from bootstrap analysis was topologically identical to the single best parsimony tree. Graphical output from the tree file was provided using TreeView (R. Page, University of Glasgow, Glasgow, Scotland).
Construction of Expression Vectors-A cytosolic T. gondii HA-Rab6 expression vector was developed by amplifying the rab6 ORF using the Rab6HANsi/R6rPacI primer set. The product was subcloned into the NsiI-PacI sites of pSAG, a SAG1 expression vector (25), modified by removal of the SAG1 signal sequence for cytosolic protein expression, creating vector pSAGRab6HA. HA-Rab6(Q70L) and HA-Rab6(N124I) were generated by amplification of the rab6 ORF with mutagenic primers and cloned by triple ligation into the NsiI-PacI sites of pSAG. Expression vectors encoding HA-Rab6(T25N) were created by site-directed mutagenesis of the corresponding HA-Rab6 vectors with complementary mutagenic primers by amplification using Pfu Turbo polymerase (Stratagene).
For bacterial expression, a 1.7-kilobase pair NcoI-BamHI fragment was excised from pNTPRab6 and subcloned into NcoI-BamHI-digested pET15b. HA-Rab6(Q70L) and HA-Rab6(N124I) were subcloned into the same sites after amplification of pSAGRab6 with appropriate mutagenic primers. The rab6 ORF in all vectors was fully sequenced for confirmation of the site-directed mutations and exclusion of additional mutations. All restriction enzymes were purchased from New England Biolabs Inc.
Dual Secretory Reporter Constructs-Plasmid vector pminCATBAP was modified for coexpression with HA-Rab6 or HA-Rab6 point mutants. This vector contains the heterologous secretory reporter Escherichia coli alkaline phosphatase (BAP), expressed from the T. gondii nucleoside-triphosphate hydrolase-3 promoter and fused to the nucleoside-triphosphate hydrolase-3 signal sequence (12). The rab6 ORFs were excised as HindIII-BamHI fragments from pSAGRab6HA and subcloned by replacement of the dihydrofolate reductase-chloramphenicol acetyltransferase fusion cassette in pminCATBAP, generating the pHXBAPRab6HA series plasmids.
Expression Analysis in T. gondii and Selection of Stable Lines-Parasites were transfected by electroporation according to published protocols (22). For selection of stable lines, hypoxanthine-xanthineguanine phosphoribosyltransferase-deficient parasites were transfected, passaged twice in human foreskin fibroblast cultures under selection with 25 M mycophenolic acid and 50 M xanthine, and cloned by limiting dilution on 96-well plates (22).
Immunoblot Analysis of Endogenous and Recombinant T. gondii Rab6 -Bacterial expression plasmids pET15HARab6, pET15HARab6 (Q70L), and pET15HARab6(N124I) were transformed into E. coli BL21(DE3) pLysS (Novagen). Crude extracts from isopropyl-␤-D-thiogalactopyranoside-induced cultures and extracts from T. gondii and Vero cells were subjected to 12% SDS-PAGE and prepared for immunoblotting or GTP binding by electrophoretic transfer to nitrocellulose membranes. Immunoblot analysis was performed with the ECL detection system (Amersham Biosciences) using, as primary antibody, rabbit serum (Zymed Laboratories, Inc.) immunized with a synthetic peptide encompassing T. gondii  and displaying strong enzymelinked immunosorbent assay activity. Membranes were exposed on X-Omat AR autoradiography film (Eastman Kodak Co.). Scanned images were semiquantitatively analyzed with Scion Image software.
GTP Overlay Assay of Recombinant T. gondii Rab6-Nitrocellulose transfer blots containing crude recombinant T. gondii HA-Rab6, HA-Rab6(Q70L), and HA-Rab6(N124I) proteins were analyzed for GTPbinding activity as previously described (26), except that phosphate buffer was substituted for Tris-HCl in the binding buffer (50 mM Na 2 HPO 4 /NaH 2 PO 4 , pH 7.0, 5 mM MgCl 2 , 1 mM EGTA, and 0.3% Tween 20). Blots were preincubated in binding buffer supplemented with 1 mM ATP for 30 min. [␣-32 P]GTP was added to a concentration of 1 Ci/ml and incubated for 2 h, and the blots were washed with several changes of binding buffer for 2 h. Gels were dried and exposed at Ϫ70°C on X-Omat AR film.
Metabolic Labeling and Quantitative Immunoprecipitation-Parasite cultures were washed three times with cysteine/methionine-free minimal essential medium (Invitrogen) and metabolically labeled in 1 ml of minimal essential medium containing 5% fetal bovine serum and 100 Ci of [ 35 S]Met/Cys (Promix, Amersham). For analysis of BAP and GRA2 secretion, parasites were labeled for 30 min. For pulse-chase analysis of GRA4, parasites were labeled for 10 min and chased after three washes for 0, 15, and 30 min. Parasites were freed from host cells by syringe passage in 1 ml of phosphate-buffered saline containing proteinase inhibitors (2) and centrifuged at 1400 ϫ g for 10 min at 4°C. Cell pellets were lysed in radioimmune precipitation assay buffer (24) containing Complete protease inhibitor mixture (Roche Molecular Biochemicals). Supernatants following the 1400 ϫ g spin (containing T. gondii soluble secreted proteins of the vacuolar space) were used for immunoprecipitation after addition of Nonidet P-40 to 1% as previously described (28) with antibody to BAP, GRA4 (mAb T8-4B9), or GRA2 (mAb T4-1F5) as described above. Gels were dried and exposed to Biomax film (Kodak). Images were scanned and quantitated using Scion Image software.

Identification and Phylogeny of the T. gondii Rab6
Homolog-Because Rab6 functions in Golgi transport in yeast and mammalian cells, we sought to identify this protein in Toxoplasma as a potential regulatory target for secretory protein trafficking. Combinatorial RACE/PCR amplification was used to clone a T. gondii rab6 cDNA. In genomic digests, a rab6 cDNA probe strongly hybridized to a single T. gondii restriction fragment in each case (Fig. 1A), indicating that rab6 is a single-copy gene in T. gondii. Northern blot analysis indicated a major RNA transcript size of ϳ1.3 kb, along with a larger, possibly unprocessed transcript (Fig. 1B). In neither case did the rab6 probe hybridize to nucleic acid controls derived from host cell cultures, confirming that Rab6 is of parasite origin.
The Toxoplasma rab6 cDNA encodes a 209-amino acid protein (Fig. 1D) with a predicted molecular mass of 23.7 kDa. Of the available sequences, the Toxoplasma Rab6 protein is most homologous to Rab6 of the Apicomplexa malarial parasites Plasmodium falciparum and Plasmodium berghei (76 -77% similarity and ϳ72% identity using BESTFIT analysis). The phylogenetic relationship of T. gondii Rab6 to other Rab6 homologs was examined by alignment (Fig. 1D) and parsimony analysis using the PAUP program (Fig. 1C). The Rab family domains are well conserved, including the nucleotide-binding pocket domains and switch regions, which confer nucleotidedependent conformational changes and interaction with Rab effectors (Fig. 1D). Distinct from Saccharomyces and Plasmodium, Toxoplasma Rab6 encodes a conservative Ser 85 residue within a motif essential for interaction with a cytoskeletonassociated Rab6 effector, Rabkinesin-6 (29). The hypervariable C-terminal region and extreme N terminus of Rab6, important in conferring functional specificity and intracellular distribution between Rab proteins, are divergent among Rab6 homologs. A single most parsimonious phylogenetic tree, well supported by bootstrap analysis, was generated with representatives of animal, plant, and fungal Rab6 homologs (Fig. 1C).
Toxoplasma Rab6 Localizes to the Golgi in Both Parasite and CHO Cells-In transiently or stably transfected Toxoplasma tachyzoites, both HA-Rab6 (Fig. 3A) and HA-Rab6(Q70L) (Fig.  3C) proteins localized, as determined by indirect immunofluorescence assay, predominantly to a discrete irregular structure apical to the parasite nucleus. HA-Rab6 signal was detected in lower amounts throughout the parasite cytoplasm and in one or more dense granules (Fig. 3, A and C). We reasoned that Rab6 motifs necessary for Golgi localization would be functionally conserved among eukaryotes, given the general ability to functionally complement subsets of Rab proteins between yeast and mammals. To compare the distribution of Toxoplasma Rab6 within a higher organism, T. gondii Rab6 was transiently expressed in CHO-K1 cells and localized with endogenous markers, including the Golgi-associated tethering proteins giantin and GRASP65, the ER protein calnexin, and mannose 6-phosphate receptor protein, a marker principally of late endosomes. In addition to labeling a cytosolic pool, T. gondii HA-Rab6 localized in a juxtanuclear pattern typical of the Golgi (Fig. 3, E and H) and colocalized in this region with giantin (Fig. 3,  E-G). Likewise, HA-Rab6 signal overlapped GRASP65 on the Golgi (Fig. 3, H-J). In contrast, HA-Rab6 exhibited minimal colocalization with the mannose 6-phosphate receptor and calnexin (data not shown).
To more precisely define Rab6 localization in Toxoplasma, HA-Rab6(Q70L) parasite clones were subjected to immunoelectron microscopy (Fig. 4, A-E). Activated Rab6 was most heavily concentrated on late Toxoplasma Golgi cisternae (Fig. 4, A and  F). Specifically, Rab6 was distributed on the apical (trans) cytosolic face of the late Golgi cisternae and the TGN, on the cisternal rims of the trans-and medial-Golgi cisternae, and on peripheral vesicles closely associated with the cisternal rims. Labeling of the cis-Golgi, the intermediate compartment between the Golgi and the nuclear envelope, and the transitional ER of the nuclear envelope was minimal. Rab6 also strongly labeled microdomains on membranes of a subset (ϳ42%, n ϭ 38) of dense granule sections (Fig. 4, B, C, and E, arrows). Rab6 was sporadically associated with microdomains on membranes of large electron lucent vesicles, putatively of endosomal function, but was not associated with membranes of mature rhoptries, micronemes, apicoplasts, or mitochondria (Fig. 4, A  and E). This distribution implicates activated Rab6 in a transport process to the late Golgi in Toxoplasma, possibly involving dense granules.
Localization of Toxoplasma Rab6 in both the parasite (Fig. 5) and transiently transfected CHO cells (data not shown) was also sensitive to the fungal metabolite brefeldin A (BFA), an inhibitor of the ARF1 GTPase exchange protein. Following addition of 5 g/ml BFA, transiently expressed HA-Rab6 (Fig.  5C) collapsed to the perinuclear envelope, along with a reticular and cytosolic labeling pattern throughout the parasite (Fig.  5, F and I). Activated HA-Rab6(Q70L) was likewise dispersed by BFA treatment to a prominent perinuclear and reticular pattern, suggesting a complete collapse of the parasite Golgi into the ER, similar to tubulovesicular retrograde Golgi-derived elements in mammalian cells. In total, these results indicate that Toxoplasma Rab6 contains evolutionarily conserved Golgi-targeting signals, despite significant C-terminal divergence of Rab6 proteins.  C, and E), on lucent vacuole membranes (E), and associated with cytoplasmic centrosomes (Ce) (D). No labeling was observed in rhoptries (Rh), micronemes (Mi), apicoplasts (Ap), and the nucleus (Nu). Control cryosections of wild-type parasites with equivalent antibody dilutions were completely free of immunogold label (data not shown). Also shown is the frequency distribution of HA-Rab6(Q70L) antigen (Ag) on cryosections (F). 19 random cryosections encompassing 34 parasites were quantitated for total gold distribution. DG, dense granules; cytoplϩsv, labeling of the cytoplasm and small vesicle (these were co-quantitated because vesicle membranes were frequently difficult to differentiate); ERϩtER, labeling of the ER and perinuclear envelope; ERGIC, intermediate compartment between the Golgi and nuclear envelope; endosome, electron lucent vacuoles; mic, micronemes; rhop, rhoptries; mito, mitochondrion; apicopl, apicoplast; IMC/PM, inner membrane complex and plasma membrane.
(N124I), or GDP-restricted HA-Rab6(T25N) were coexpressed along with the secretory reporter protein BAP. This protein, fused to the nucleoside-triphosphate hydrolase-3 signal sequence, behaves as a soluble secreted dense granule protein in Toxoplasma tachyzoites (12). Secretory BAP localized to dense granules and the PV upon secretion (Fig. 5, A and D). In the presence of transiently overexpressed HA-Rab6 (Fig. 5B) or BFA (Fig. 5, D and G), BAP secretion into the PV and distribution in dense granules were partially inhibited, with signal accumulating in the Golgi and the ER-associated perinuclear envelope. Rab6 overexpression augmented the BFA-induced retrograde appearance of BAP in the parasite ER within the first 15 min of treatment (Fig. 5J).
The role of activated and inhibitory Rab6 mutants in secretory transport was assessed by transient overexpression in transfected parasites. GTP-activated mutant HA-Rab6(Q70L) (Fig. 6A) induced a similar but more pronounced accumulation of BAP in the Golgi and the ER-associated perinuclear envelope (Fig. 6B) in comparison with overexpressed wild-type Rab6. HA-Rab6(N124I) was cytosolic in distribution (Fig. 6D), whereas HA-Rab6(T25N) was partially Golgi-localized (Fig.  6G). Each of these mutants induced dispersal of BAP within the parasite in a small vesicular and reticular pattern independent of dense granules (Fig. 6, E and H), with BAP accumulating at or near the late Golgi in the case of HA-Rab6(T25N) (Fig. 6H), although secreted BAP was still detectable in parasite vacuoles (Fig. 7A). Transient overexpression of HA-Rab6(N124I) was toxic because parasite vacuoles often contained rounded degenerative bodies of atypical parasite morphology (data not shown).
Parasites expressing Rab6 mutants were further analyzed for transport of endogenous dense granule proteins GRA1-4. In the case of HA-Rab6(T25N), GRA1 accumulated in the same Golgi-proximal anterior compartment (Fig. 7B, arrows), colocalizing with BAP (Fig. 7C). Although BAP was dispersed from dense granules, these remained weakly positive for GRA1; both secreted proteins were still apparent in the PV (Fig. 7, B and C, near the asterisks). Identical results were obtained when parasites were labeled with endogenous GRA2-4, suggesting an attenuated effect on transport of dense granule protein aggregates in comparison with soluble secretory BAP. In contrast, transport of glycosylphosphatidylinositol-anchored proteins to the cell surface and microneme or rhoptry proteins from the ER and Golgi was uninhibited in the presence of the Rab6 mutants (data not shown). Altogether, these results indicate that alteration of Rab6 function by overexpression leads to partial disruption in the transport of constitutively secreted proteins between dense granules and the Golgi.
Dense Granule Secretion Is Differentially Inhibited in Transient and Stable Rab6 Mutants-To determine whether bulk secretion into the PV was altered by Rab6 levels, secretion of BAP was quantitatively assessed in intracellular parasites by metabolic pulse labeling. BAP secreted from parasites transiently overexpressing HA-Rab6 or HA-Rab6(Q70L) was ϳ50 -60% of that secreted from parasites expressing BAP alone, whereas expression of HA-Rab6(N124I) reduced BAP secretion by ϳ10% (Fig. 8A), despite the loss in labeling of dense granules. This result suggests that constitutive secretion may not require transport through dense granules, although this is the default route in wild-type parasites (2,12). We then examined FIG. 5. Overexpressed Rab6 augments the brefeldin A-induced retrograde transport of dense granule proteins. A, in untreated parasites, BAP was secreted through dense granules and accumulated in the PV. In D-I, treatment with BFA induced intracellular retention, ultimately accumulating BAP in the perinuclear ER. In parasites transiently overexpressing HA-Rab6 (C, arrows), BAP secretion through dense granules was partially inhibited, accumulating in the Golgi and perinuclear ER (B, arrows and asterisks). Both BAP (E and H) and transiently expressed HA-Rab6 (F and I) were partially collapsed to the perinuclear ER following treatment with 5 g/ml BFA for 15 or 60 min. In J, shown is the quantitation of the appearance of BAP in the parasite ER following BFA (5 or 20 g/ml) addition. Perinuclear and reticular labeling was more pronounced and rapidly induced in cells overexpressing HA-Rab6. the effect of low level expression of Rab6 mutants in tachyzoites by isolating stable parasite clones coexpressing HA-Rab6(Q70L) or Rab6(N124I) along with BAP. These clones retained endogenous Rab6 along with a single integrated copy of the mutant proteins expressed from the low level tachyzoite Sag1 promoter (data not shown). In contrast to parasites transiently overexpressing HA-Rab6(Q70L), BAP secretion was essentially unaltered in HA-Rab6(Q70L) clones, whereas HA-Rab6(N124I) parasite clones exhibited a similar attenuation upon secretion of BAP compared with those transiently expressing the mutant (n ϭ four independent experiments) (Fig. 8B).
To confirm these results for endogenous dense granule proteins, we examined the secretion of GRA4 by quantitative immunoprecipitation in parasite clones expressing HA-Rab6 (Q70L) or HA-Rab6(T25N). Once again, the activated Rab6 mutant had little effect on dense granule secretion compared with untransfected parasites, whereas stable expression of the dominant inhibitory mutant was attenuated upon secretion of GRA4 (Fig. 8C). To address the possibility that protein modification in the parasite Golgi could be altered in the presence of the Rab6 mutants due to retrograde imbalance of glycosylation enzymes, the O-linked glycosylation of GRA2 was examined by metabolic pulse-chase. Within minutes of synthesis, GRA2 is modified by O-linked glycosylation from an ϳ26-kDa precursor to an ϳ28-kDa mature protein (27,32). At 15 min of chase, a kinetic lag was observed in the mobility of GRA2 in parasites stably expressing HA-Rab6(T25N), whereas GRA2 mobility in parasites expressing HA-Rab6(Q70L) was unaltered compared with wild-type parasites (Fig. 8D). By 30 min of chase, nearly 100% of GRA2 was modified in all parasite clones (Fig. 8D). In summary, elevated overexpression of wild-type and GTP-activated Rab6 blocks constitutive protein secretion, whereas GDP-bound and nucleotide-free Rab6 mutants appear to attenuate dense granule protein transport and glycosylation. Consequently, Rab6 appears to regulate a retrograde pathway involving transport of dense granule proteins to the Golgi.
Rab6 Marks Golgi Fission, and Rab6 Mutants Alter Cytokinesis in Toxoplasma Tachyzoites-Upon both plaque assay and uracil incorporation analysis, growth or invasiveness of activated and inhibitory Rab6 clones was reduced in comparison with wild-type parasites (Fig. 9). Although Rab6 mutants did not alter transport of rhoptry proteins to the post-Golgi organelle, HA-Rab6(Q70L) and HA-Rab6(N124I) parasites were enriched in tubulovesicular precursor rhoptry organelles (data not shown). Because rhoptry biogenesis is itself developmentally regulated, we examined the parasite clones for developmental alterations. Daughter Toxoplasma merozoites form in pairs within the maternal parasite by a process of internal budding termed endodyogeny. This process is initiated through a pair of cytoplasmic centrosomes, which form the intranuclear spindle poles and behave as a microtubule-organizing center (33). During mitosis, the Golgi stacks partition in two near the centrosomes. This was easily visualized in parasites expressing Golgi-localized HA-Rab6(Q70L). The medial fission of the Golgi stacks occurred in close juxtaposition with the duplicated centrosomes and appeared to be complete prior to nuclear division (Fig. 10,  K-N). During cytokinesis, activated Rab6 and centrin were perfectly colocalized (Fig. 10M, arrow with asterisk), suggesting a tethering of the late Golgi with the centrosomes.
Concurrently with Golgi partitioning, a scaffold for daughter cell assembly is formed from subpellicular microtubules and an associated network of flattened membranes, the IMC. During cytokinesis, the IMC migrates posteriorly toward and around each nuclear pole, partitioning the replicated nucleus into the two forming daughters. Strikingly, some HA-Rab6(Q70L) and HA-Rab6(N124I) parasite vacuoles exhibited unusual patterns of division, forming multiple daughters (Fig. 10A), indicative of replication by endopolygeny rather than endodyogeny. The IMC of each daughter enveloped a complete set of organelles, including Golgi, apicoplasts, and pre-rhoptries, indicating that appropriate organellar replication and segregation were occurring concurrent with nuclear division. The frequency of endopolygeny was quantitated using antibody to IMC1, an antigen of the IMC in both adult and forming daughter parasites, thus serving as a hallmark of daughter cell formation by microscopy (34). In nearly all untransfected tachyzoites, parasites budded by endodyogeny, with two forming daughters distinguishable (Fig. 10B), although on rare occasions (ϳ2% of all vacuoles containing dividing parasites) (Fig. 10F), multiple daughter parasites could be found in a single parent. However, in HA-Rab6(Q70L) (data not shown) and HA-Rab6(N124I) (Fig. 10, E-J, arrows) clones, vacuoles harboring parasites replicating by endopolygeny were 5 and 10 times more common, respectively (Fig. 10F). These results suggest that altered Rab6 function enhances uncoupling of cytokinesis and karyokinesis in developing parasites.

DISCUSSION
The highly polarized secretory pathway of T. gondii, complete with a single stack of apically oriented Golgi cisternae, provides a simple model for protein transport in eukaryotes. We have identified the T. gondii homolog of the highly conserved GTPase Rab6 as the first marker of the parasite Golgi and as a potential mediator of retrograde transport from dense granule secretory organelles. Whereas Rab6 paralogs function in retrograde transport from post-Golgi-, intra-Golgi-, and Golgi-to-ER transport pathways in mammalian cells, and Ypt6p functions in recycling of Golgi proteins from the endosomal pre-vacuole in yeast, Toxoplasma Rab6 appears to regulate in part post-Golgi transport of constitutively secreted proteins to the parasite Golgi. Our results further suggest that Rab6 function may be required for normal cytokinesis or that parasite cell division may be sensitive to perturbation of the Rab6mediated retrograde pathway.
We propose that the role of Rab6 in Toxoplasma dense gran-  ule protein transport is retrograde from a post-Golgi compartment, based upon several observations. Generally, activated forms of Rab proteins concentrate on the target membrane for vesicle transport. Both wild-type and GTP-activated Rab6 localize to the late Golgi cisternae and inhibit constitutive secretion of dense granule proteins only when highly overexpressed, resulting in Golgi and ER retention of the dense granule proteins. In this instance, Rab6 augments the effect of BFA on retrograde transport to the ER, resulting in resorption of the Golgi cisternae and associated cargo. BFA inhibits GTPase exchange proteins mediating ARF1 recruitment that is essential for COPI-regulated vesicle budding (35). Toxoplasma ARF1 mutants induce a similar block in the early transport of dense granule proteins (8), indicating that both COPI-mediated and Rab6-mediated pathways function in dense granule protein transport.
In contrast to activated Rab6, dominant inhibitors Rab6(N124I) and Rab6(T25N) alter post-Golgi transport of dense granule proteins, accumulating these proteins in the Golgi and altering glycosylation rates, indicative of an imbalance in Golgi modification enzymes. However, the viability of the mutants and the relatively modest effects on constitutive secretion in stable clones indicate that Rab6 function is dispensable for secretion in Toxoplasma, in common with Ypt6p in yeast. Nonetheless, the specific association of Rab6 with dense granule membranes and small vesicles near the trans-face of the Golgi complex, in addition to the entrapment of dense granule proteins in the Golgi induced by dominant-negative mutants, implicates a transport pathway linking the two organelles. This pathway may serve a recycling function from mature dense granules or alternatively play a direct role in dense granule biogenesis or maturation through protein sorting.
A novel Rab6-mediated Golgi-to-ER retrograde transport in mammalian cells was initially proposed based upon the BFAlike phenotypes of activated Rab6 on Golgi enzymes (14). This pathway was further defined by following transport of Shiga and Shiga-like toxins to the ER (15) and the finding that FIG. 10. Rab6 mutants are altered in cytokinesis, and activated Rab6 colocalizes with centrin during cytokinesis. A, transmission electron microscopy of a HA-Rab6(Q70L) parasite clone in late endopolygeny, with three forming daughter parasites displaying nuclei (N), apicoplasts (A), and rhoptries (P), enveloped within an IMC. B, immunofluorescence analysis of wild-type parasites undergoing endodyogeny (arrows) with IMC1. In daughter cells, IMC1 tracked the IMC plane as it migrated posteriorly toward the nucleus (B). C, 4,6-diamidino-2phenylindole (DAPI) staining of DNA; D and E, merge and phase-contrast images, respectively. F, rates of endopolygeny in Toxoplasma hypoxanthine-xanthine-guanine phosphoribosyltransferase (HXGPRT)-deficient, HA-Rab6(Q70L), and HA-Rab6(N124I) clones. Occurrence of vacuoles with parasites undergoing endopolygeny (more than two daughter parasites) is defined by nascent IMC1 labeling and expressed as a fraction of all vacuoles with daughter parasites. 1000 vacuoles with dividing parasites were scored for each clone. G, HA-Rab6(N124I) parasites undergoing endopolygeny (arrows) with anti-IMC1 antibody. H, 4,6-diamidino-2-phenylindole staining of DNA; I and J, merge and phase-contrast images, respectively. In one cell, a daughter parasite is emerging from the maternal body (G-I, asterisks). K-N, Golgi membranes labeled with Rab6(Q70L) (K) undergoing medial fission (lower parasite vacuole, horizontal and vertical arrows) and during cytokinesis (arrows with asterisks). Centrin (L) is juxtaposed with Golgi membranes during mitosis (M), but only colocalized during daughter cell budding (M, arrow with asterisk). In the phase-contrast image (N), a forming daughter is delimited by a phase-contrast lucent zone (arrow with asterisk). fluorescent protein-Rab6 transits along microtubule tracks between the Golgi and ER sites at the mammalian cell periphery (16). It is thought that this retrograde pathway may regulate recycling of proteins lacking ER retrieval motifs, recycling of lipids, and the slow continuous cycling of resident Golgi proteins. A post-Golgi function for mammalian Rab6 in recycling between early or recycling endosomes and the TGN has now distinguished the highly homologous Rab6AЈ isoform from Rab6A, which is implicated in Golgi-to-ER transport (36). Evidence for Rab6 association with post-Golgi secretory vesicles is less clear, although Rab6 has been postulated to play a role in transport of rhodopsin on post-Golgi vesicles in retinal rod photoreceptor cells (37) and to function at an early step in the biogenesis of small synaptic vesicles of embryonic neurons (38) and was found to be associated with secretory granules in rat atrial myocytes (39).
The role of the yeast homolog of Rab6, Ypt6p, is increasingly well understood. Yeast late Golgi membrane proteins cycle through a pre-vacuolar endocytic compartment and are retrieved to the Golgi. Ypt6p mutants disrupt the cycling of the endoprotease Kex2p and Vps10p, the sorting receptor for the vacuolar hydrolase carboxypeptidase Y, which is consequently partially missorted to the plasma membrane (40). Recent identification of effector and binding proteins acting on Ypt6p has elucidated a mechanistic role for GTP-activated Ypt6p, the trimeric VPS52-VPS53-VPS54 complex, and the SNARE Tlg1p in sorting and docking of transport vesicles to the late Golgi. The Ypt6p GTP exchange factor has been identified as a stable heteromeric complex of Ric1p-Rgp1p peripherally associated with Golgi membranes (20). After GDP/GTP exchange, activated Ypt6p recruits the VPS52-VPS53-VPS54 complex to the Golgi membrane, which in turn binds the endosomal/late Golgi SNARE Tlg1p on endosome-derived vesicles and allows fusion upon SNARE pairing (19).
It is possible that an intermediate endosomal compartment may exist between dense granules and the Golgi. Toxoplasma Rab5A localizes to a tubulovesicular compartment closely opposed to, but distinct from, Golgi cisternae, and Rab5 mutants alter cholesterol ester-rich lipid body formation (41). The yeast Rab5 homolog, Ypt51p, controls transport of Golgi-derived vesicles to the yeast pre-vacuole and vacuole, opposite in nature to Ypt6p-mediated transport (42). Toxoplasma Rab5 and Rab6 may act as opposing partners in similar pathways. For now, identification of endosomal and recycling compartments in Toxoplasma awaits characterization with biochemical and molecular markers in addition to Rab5.
A novel role for Rab6 function was suggested by the appearance of Rab6 mutant parasites altered in cytokinesis. Paired internal budding (endodyogeny) is the predominant mechanism of cytokinesis in T. gondii. Although karyokinesis and cytokinesis can be uncoupled in Toxoplasma tachyzoites, as demonstrated through the use of microtubule inhibitors (33), organellar and nuclear divisions are coupled through association with centrioles. Endopolygeny, a derivative of endodyogeny, occurs as a consequence of multiple nuclear and organellar divisions paired with new apical complex and IMC formation anywhere in the cytoplasm of the parent cell. This differs from schizogony, a common form of asexual division in Apicomplexa in which multiple daughter nuclei are ordered along the maternal cell periphery before cytokinesis initiates. Using the IMC as a hallmark of nascent daughter formation, HA-Rab6(N124I) and HA-Rab6(Q70L) parasite clones partially shifted to endopolygeny. Furthermore, activated Rab6 on Golgi membranes colocalized with centrin specifically during cytokinesis, suggesting a role for Rab6 in coordinating cytokinesis with the nuclear cycle.
Recently, effectors of Rab6 have been identified that may function directly in cytokinesis. A Rab6-binding kinesin, alternately named Rabkinesin-6, RB6K, and Rab6-KIFL, was first identified through a Rab6 protein two-hybrid screen (43). This motor protein interacts with the effector domain of GTP-activated Rab6 and binds to microtubules both in vitro and in vivo. Rab6 may regulate the association and dissociation of Rabkinesin-6 with the microtubule cytoskeleton and, as a complex, control movement of Golgi membranes and transport vesicles along microtubules (43). However, Rab6-KIFL is also proposed to function in cell division during cytokinesis (44,45). Rab6-KIFL was shown to localize to the mid-zone of the mitotic spindle and to the cleavage furrow and midbody during late stages of mitosis, and overexpression of the protein induced a defect in cleavage furrow formation, blocking cytokinesis. A second effector, GAPCenA, has been identified by a yeast twohybrid screen and serves as a Rab6-specific GTPase-activating protein, with the dual capacity to transiently interact with centrosomes (46). GAPCenA bears homology to spindle checkpoint proteins in yeast, can form complexes with ␥-tubulin, and has a role in microtubule nucleation. Intriguingly, assembly of the Toxoplasma conoid, an initial step in parasite cytokinesis, involves ␥-tubulin. Although they remain to be identified, Rab6 effectors similar to Rabkinesin-6/Rab6-KIFL and GAPCenA may be operative in Toxoplasma, and alterations in parasite cytokinesis might be due to a dominant-interfering effect of Rab6 mutants on these effectors. In total, it appears that Rab6 or its effectors may play bimodal roles, functioning not only in protein transport, but also in cell division.