AP-1 in Toxoplasma gondii Mediates Biogenesis of the Rhoptry Secretory Organelle from a Post-Golgi Compartment*

We have previously demonstrated thatToxoplasma gondii has a tyrosine-based sorting system, which mediates protein targeting to the lysosome-like rhoptry secretory organelle. We now show that rhoptry protein targeting is also dependent on a dileucine motif and occurs from a post-Golgi endocytic organelle to mature rhoptries in an adaptin-dependent fashion. TheT. gondii AP-1 adaptin complex is implicated in this transport because the μ1 chain of T. gondii AP-1 (a) was localized to multivesicular endosomes and the limiting and luminal membranes of the rhoptries; (b) bound to endocytic tyrosine motifs in rhoptry proteins, but not in proteins from dense granule secretory organelles; (c) when mutated in predicted tyrosine-binding motifs, led to accumulation of the rhoptry protein ROP2 in a post-Golgi multivesicular compartment; and (d) when depleted via antisense mRNA, resulted in accumulation of multivesicular endosomes and immature rhoptries. These are the first results to implicate AP-1 in transport from a post-Golgi compartment to a mature secretory organelle and substantially expand the role for AP-1 in anterograde protein transport.

Obligate intracellular protozoa of the phylum Apicomplexa are highly sophisticated secretory cells. Although the host cell type and the nature of the intracellular vacuole differ when Plasmodium, Toxoplasma, Cryptosporidium, Eimeria, Theileria, and Babesia are compared, these organisms share common secretory organelles: micronemes, rhoptries, and dense granules. Differential secretion from these organelles directs invasion of host cells, formation of an intracellular parasitophorous vacuole, and subsequent modification of the vacuole for replication (reviewed in Ref. 1).
Most unique morphologically and biochemically are the rhoptries. These are flask-shaped secretory organelles that are assembled and positioned at the apical pole of the invasive forms of Apicomplexa (reviewed in Refs. [2][3][4]. At the onset of host cell invasion, prepackaged proteins and membranous materials from rhoptries are secreted through an apical opening. Rhoptry constituents are likely involved in a plethora of essential functions. These include directing host cell attachment and invasion (3), expansion and maintenance of the parasitophorous vacuole membrane surrounding the intracellular parasite (5), and attachment of host mitochondria and the endoplasmic reticulum to the parasitophorous vacuole membrane (6). Furthermore, rhoptry proteins are targets of the normal immune response; antibodies to rhoptry proteins block invasion; and immunization with selected rhoptry proteins provides partial protection against parasite challenge (2). Hence, the unique rhoptry organelle can serve as a potential target for blocking parasite invasion and growth. The biogenesis and phylogenetic relationship of rhoptries to secretory organelles in other cells are not well understood. Rhoptries and pre-rhoptries are described as the only acidified organelles in the parasite (7) and are packaged with specialized hydrolases (8) and cholesterol and membranes for secretion (9,10), characteristics of a secretory lysosome or lysosome-like organelle (11,12). We have recently demonstrated that an endocytic tyrosine-based motif (YXX⌽) in the cytoplasmic tail of the rhoptry protein ROP2, when mutated, inhibits delivery of the protein to rhoptries (13). The mutant ROP2 protein accumulates in a compartment of unknown origin, but distinct from the Golgi. Taken together, these results suggest that rhoptries may be analogous to secretory lysosomes.
Another implication of these results is that heterotetrameric adaptin complexes (reviewed in Refs. 9, 14, and 15) are potentially involved in protein sorting to rhoptries. In other eukaryotes, tyrosine-based (YXX⌽) or dileucine-based (LL) motifs in the cytoplasmic tail of transmembrane receptors are recognized by the medium or ␤ subunit, respectively, of heterotetrameric adaptin complexes. In turn, adaptins bind clathrin and induce the formation of clathrincoated vesicles (16 -18). The six AP (adaptor protein) complexes (AP-1A/AP-1B, AP-2, AP-3A/AP-3B, and AP-4) are distributed to different organelles in the cell, defining their specific function and cargo selection (15). AP-2 mediates endocytosis from the plasma membrane, whereas AP-1A/AP-1B, AP-3A/AP-3B, and AP-4 sort proteins at the trans-Golgi network (TGN) 1 and/or endosomes. Although AP-1 was originally suggested to function only in anterograde transport * This work was supported by National Research Service Awards 5T32AI07404 and 5F32AI10044 (to H. M. N.) and Grant RO1 AI30060 (to K. A. J.) from the National Institutes of Health and by a scholar award in molecular parasitology from the Burroughs Wellcome Fund (to K. A. J.). Work performed at the Institute for Genomic Research and Washington University was supported by National Institutes of Health Grants AI05093 and 1R01AI045806-01A1, respectively. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY117037.
§ Both authors contributed equally to this work. from the TGN to endosomes, recent evidence suggests that the complex mediates retrograde transport of mannose 6-phosphate receptors from endosomes to the TGN (19,20). In neuroendocrine, endocrine, and mammary epithelial cells, AP-1 is localized to immature secretory granules, where it functions in the removal of membranes and nonspecific cargo, presumably via a retrograde pathway, to facilitate maturation to mature secretory granules (21,22). AP-3 mediates direct transport of selected cargo from the TGN to lysosomelike secretory organelles and may also mediate transport from endosomes to lysosome-like organelles (reviewed in Ref. 12, 15, and 23). In contrast, there is no evidence supporting a role for AP-1 in anterograde transport from a post-Golgi compartment to another organelle and especially to a lysosome-like organelle.
On both a functional and genomic basis, it is increasingly clear that the Toxoplasma gondii secretory pathway can be considered a stripped-down version of the more complicated machinery present in higher eukaryotes (24,25). 2 Coupled with the presence of multiple distinct and easily recognized secretory organelles, the parasite permits insight into processes not readily discerned in other systems. We now show that rhoptry proteins are transported to mature rhoptries from a post-Golgi endocytic compartment in an AP-1-dependent fashion. Other adaptins and adaptors, including AP-3, GGA proteins, stonins, and ␤-arrestins (reviewed in Ref. 26), cannot participate because they are absent from the T. gondii genome. This result 2 Available at www.toxodb.org.
FIG. 1. Mutation of dileucine residues in the ROP2 cytoplasmic tail abolishes targeting to mature rhoptries. A: upper panels, ROP2-HA was expressed in T. gondii. When localized with anti-HA antibody, ROP2-HA precisely co-localized with the mature rhoptry compartment, as detected with anti-ROP2 polyclonal antibody. Lower panels, ROP2 with an L529A/L530A (LL529 -530AA) mutation in the cytoplasmic tail localized to a compartment (arrow) adjacent to, but largely distinct from, mature rhoptries. Although a vacuole showing minimal staining of the L529A/L530A compartment with anti-ROP2 polyclonal antibody is illustrated here for emphasis (and was also observed previously for ROP2⌬) (13), many vacuoles showed the expected partial staining of the mutant ROP2 compartment with anti-ROP2 polyclonal antibody. B: the L529A/L530A mutant of ROP2 did not co-localize with dense granules (GRA3), micronemes (MIC2), or the Golgi/TGN. substantially expands the role for AP-1 in anterograde transport and more specifically in transport to and biogenesis of lysosome-related organelles.

EXPERIMENTAL PROCEDURES
Cloning of the T. gondii 1 Chain (Tg1) Gene-A cDNA clone (zz38e01) with strong homology to the mouse 1A chain was located in the T. gondii expressed sequence tag data base (27). 3 The clone was found to contain the 117 C-terminal codons of a T. gondii 1 homolog, which we designated Tg1. The sequence of the clone was extended by PCR from parasite cDNA using degenerate primers to the conserved sequence YELLDE (amino acids 113-118 of the mouse 1A sequence), and full-length Tg1 was finally obtained by 5Ј-rapid amplification of cDNA ends from T. gondii cDNA. The Tg1 cDNA sequence has been deposited in the GenBank TM /EBI Data Bank under accession number AY117037. T. gondii and human foreskin fibroblast DNAs were isolated as previously described (13).
The GenBank TM /EBI accession numbers of adaptin sequences used for the sequence analysis in Fig. 2 are as follows: Arabidopsis thaliana 3 Available at www.tigr.org/tdb/t_gondii/.

FIG. 3. Tg1 is predominantly localized to the Golgi/TGN as indicated by immunofluorescence microscopy.
A, endogenous Tg1 was detected by an affinity-purified antiserum against Tg1, whereas overexpressed wild-type (Tg1-HA) and dominant-negative (Tg1-(D176A)-HA) Tg1 were detected with anti-HA monoclonal antibody. Staining of the putative Golgi (arrows) was partially disrupted by a 10-min treatment with 5 M brefeldin A (BFA). B, Tg1-HA (detected by anti-HA monoclonal antibody) partially co-localized with the Golgi/TGN as visualized in Tg1-HA parasites transiently expressing the Golgi/TGN marker bacterial alkaline phosphatase-low density lipoprotein receptor (LDLR) (13) or by staining for dynamin (polyclonal antibody MC63, provided by Mark McNiven, Mayo Clinic, Rochester, MD) or ADP-ribosylation factor-1 (ARF-1) (38) in Tg1-HA parasites. Tg1 also partially co-localized with structures labeled by anti-ROP2 polyclonal antibody. In any given microscopic field, Tg1-HA was concentrated to different extents in the Golgi relative to other structures in the organism, probably reflecting different levels of protein expression.
Mutation of the ROP2 Cytoplasmic Tail Dileucine Motif-The ROP2 cytoplasmic tail contains a dileucine motif at residues 529 and 530, 10 residues upstream of the YXX⌽ motif previously implicated in rhoptry protein transport. An L529A/L530A mutation was introduced into the cytoplasmic tail of ROP2 using the hemagglutinin (HA)-tagged ROP2 construct previously described (13). This construct was transiently transfected into T. gondii and localized as previously reported (13).
Production of Polyclonal Antibodies-The Tg1 coding region was subcloned into a pZT28a vector (Amersham Biosciences) to produce a His-Tg1 recombinant protein. His-Tg1 was purified on a column and sent to Cocalico Biologicals, Inc. for production of a rabbit polyclonal antiserum. The antiserum (anti-Tg1) was affinity-purified by adsorption and desorption to isolated recombinant proteins on nitrocellulose membrane as described previously (28). The affinity-purified antiserum recognized a polypeptide migrating at ϳ47 kDa upon immunoblotting.
Parasite Growth and Transfection-The growth conditions and transient and stable transfection of the RH strain of T. gondii grown on monolayers of Vero cells and human foreskin fibroblasts were as described previously (13).
Antisense RNA-targeted Depletion-The complete coding region of Tg1 was amplified by PCR and cloned in the antisense orientation into the AvrII and EcoRI sites of the ribozyme-histone cassette pNTPRZ (29). Transfection, selection, and cloning of antisense clones were done as previously described (29).
Mutagenesis of Tg1-An HA epitope tag was introduced into Tg1 between codons 231 and 232 by PCR, yielding Tg1-HA. (Addition of a c-Myc epitope tag at the N terminus of Tg1 resulted in a protein that was diffusely cytosolic and was presumed not to be incorporated into an adaptin complex.) D176A and W415A mutations were introduced in Tg1-HA by PCR, yielding Tg1(D176A)-HA and Tg1(W415A)-HA. These mutations have previously been shown to alter binding of rat 2 to tyrosine-containing motifs without affecting incorporation of the 2 chain into the AP-2 complex (30). Both Tg1-HA and Tg1(D176A)-HA were ligated into pNTP/Sec for transient transfection and into pDHFR-TS/ C3M2M3 for stable transfection as described previously (13). Sequences of all PCR-amplified fragments were verified by dideoxy sequencing at the W. M. Keck Sequencing Center of the Yale University School of Medicine.
For cryoimmunoelectron microscopy localization, monolayers of Vero cells or human foreskin fibroblasts were infected with wild-type RH or stably transfected parasite clones for 24 h, fixed in 8% paraformaldehyde in 0.25 M HEPES (pH 7.4) for 48 h, scraped off in phosphatebuffered saline, and pelleted in 10% fish skin gelatin. The gelatinembedded pellets were infiltrated overnight with 2.3 M sucrose and frozen in liquid nitrogen. Ultrathin cryosections were prepared with a Leica Ultracut microtome with cryoattachment and transferred to Formvar/carbon-coated specimen grids. Sections were incubated in phosphate-buffered saline containing affinity-purified rabbit anti-Tg1 antibody (1:20) or mouse anti-HA monoclonal antibody (1:100). Following phosphate-buffered saline washes, the sections were incubated with 1% fish skin gelatin containing 5-nm protein A-gold conjugate (1:70, from the laboratory of J. Slot) for 1 h. Sections were subsequently washed in phosphate-buffered saline, post-fixed in 1% glutaraldehyde, and contrasted with 0.5% uranyl acetate in 2% methyl cellulose. Sections were observed, and images were recorded with a Philips 410 FIG. 4. Localization of Tg1 by cryoimmunoelectron microscopy. Shown are electron microscopic images of Tg1-HA parasites demonstrating the gold localization of Tg1 (arrows) predominantly to the trans-most cisterna of the Golgi (G) and vesicles and tubules extending from or juxtaposed to the TGN (GVT) (panels A and B) and to membranes of microneme (M) and endosomes (E) and also to the membrane and lumen of aberrant rhoptries (R) (panel C). Endosomes are defined as vesicles 200 -300 nm in diameter that are clear (panel C, lower inset) or that contain membrane whorls or dense deposits and that are known to contain the early endosomal marker T. gondii Rab5 (44) or VPS4 (see Footnote 5). Similar localization was observed for endogenous Tg1 in RH, as detected by affinity-purified anti-Tg1 polyclonal antibody, and for Tg1(D176A)-HA in the stable transgenic parasites (data not shown). The organellar density of gold labeling (gold particles/M 2 ) for endogenous Tg1 in RH and Tg1(D176A)-HA was determined from a random sample of 20 -30 cell profiles (panel D). Density was calculated by estimating the number of gold particles divided by the estimated volume of organelles using a doublesquare test system. The density was expressed as a percentage of the sum of all organelles and cytoplasm estimates. Only well defined Golgi stacks were categorized as Golgi cisternae, whereas vesicles and tubules adjacent to the TGN were categorized as Golgi-associated vesicles and tubules. N, nucleus. Scale bars ϭ 200 nm.
transmission electron microscope. Specificity of localization was determined by stereological analysis of 35-50 random sections of whole cells. The total number of gold particles and organelle surface areas were estimated from positive prints of transmission electron microscope negatives using a double-square test system (33).
Transmission Electron Microscopy Quantitative Analysis of Organelles-Monolayers of Vero cells or human foreskin fibroblasts were infected with RH or stably transfected parasite clones for 24 h, fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate (pH 7.4) for 30 min, washed, and post-fixed in 1% osmium tetroxide in the same buffer. Fixed samples were processed for Epon embedding and thin-sectioned for transmission electron microscopy using standard protocols.
To assess the morphological differences in organelle biogenesis in wild-type RH-, Tg1-HA-, or Tg1(D176A)-HA-expressing parasites, 60 parasite sections from 20 -30 random vacuoles were photographed at low magnification (ϫ10,400) to examine gross cell morphology and at high magnification (ϫ21,000) to analyze organelle structure. Total organelle number and surface areas were calculated from prints of transmission electron microscope negatives using a double-square overlay test system (33).

Mutation of Dileucine Residues in Rop2 Abolishes
Targeting to Mature Rhoptries-Members of the Rop2 gene family contain dileucine residues upstream of an endocytic YXX⌽ motif previously shown to be involved in targeting to mature rhoptries. Because dileucine residues can also mediate endocytic targeting by binding to adaptin ␤ chains (see Refs. 34 and 35; reviewed in Ref. 11), these residues were mutated to di-alanine, and the ROP2 mutant was expressed in T. gondii. As illustrated in Fig. 1 (A), the L529A/L530A mutant localized to a compartment distinct from, but adjacent to, mature rhoptries, similar to the localization of ROP2 with a deletion of the YXX⌽ motif (ROP2⌬) (13). This compartment was distinct from the dense granules, micronemes, and Golgi/TGN of the parasite (Fig.  1B), as was the ROP2⌬ mutant (13). Importantly, both the ROP2⌬ mutant and the L529A/L530A mutant substantially colocalized with a multivesicular endosomal compartment marked by the endosomal marker VPS4 (vacuolar protein sorting-4). 4 Altogether, these data provide further evidence that rhoptry targeting occurs along the endocytic pathway and is mediated by adaptins. They also implicate adaptin machinery in mediating targeting from a post-Golgi compartment to a mature secretory organelle. We undertook experiments to further support and understand this observation. FIG. 5. Disruption of the YXX⌽-binding pocket of Tg1 abolishes its specific interactions with rhoptry proteins in a yeast two-hybrid growth assay. A, the T. gondii 1 homolog was fused to the Gal4 activation domain (AD), whereas the cytoplasmic tails of GRA4, GRA7, ROP2, and ROP4 were fused to the Gal4 binding domain (BD) as described under "Experimental Procedures." The strength of interaction between the activation and binding domains co-transformed into S. cerevisiae was measured by growth (A 600 nm ) in 5 mM 3-amino-1,2,4-triazole, a His3 inhibitor. Interactions between Tg1 and GRA4 and GRA7 were comparable to the negative control (yeast co-transformed with pGBT9 and pACT2, expressing the binding and activation domains alone). Interaction between Tg1 and ROP4 was 3-fold that between Tg1 and ROP2 and similar to that between a truncated ROP2 tail and a triplicated YEQL motif at the C terminus (ROP2.3Y) (13). B, either the ROP4 cytoplasmic tail or ROP2.3Y was fused to the activation domain, whereas wild-type Tg1 and dominant-negative Tg1 (described under "Results") were fused to the binding domain. Mutations in tyrosine-binding residues within Tg1 blocked binding to ROP4 and ROP2.3Y. Results shown are from a single experiment, representative of two experiments performed. Cloning of Tg1 of the AP-1 Adaptin Complex-Based on a cDNA sequence (clone zz38e01) from the T. gondii data base of expressed sequence tags, a full-length medium chain () of a clathrin adaptin was amplified from parasite cDNA. Sequence analysis of the cDNA clones predicted an open reading frame encoding a polypeptide of 430 amino acids with a predicted molecular mass of 47.6 kDa. Alignment and dendrogram analysis of all four known classes of chains indicated that this T. gondii homolog separates into the 1 cluster and shares the highest similarities (75-79%) and identities (58 -62%) with the adaptin 1 chain (Fig. 2) (data not shown). The sequence also contains the signature residues ( 173 VFLD 176 , Lys 203 , Val 399 , Leu 402 , and 414 PWVR 417 ) that are known to mediate binding to tyrosine and hydrophobic residues within the YXX⌽ sorting motif (16,18,36). We accordingly termed the T. gondii homolog Tg1.
Southern blotting of parasite and host genomic DNAs (data not shown) indicated that the Tg1 gene is parasite-specific and single-copy, as also suggested by the recently available T. gondii genome sequence. 2 Northern blotting of host and parasite total RNAs (data not shown) suggested that the Tg1 gene is translated as a precursor transcript and processed to a mature transcript, consistent with the presence of an intron.
Endogenous Tg1 and Tg1-HA Localize to the TGN-In mammalian cells, AP-1 is an essential regulator of membrane trafficking by mediating budding of clathrin-coated vesicles from the TGN, immature secretory granules, and endosomes. To define the specific roles of Tg1 in the multiple secretory and endocytic pathways in T. gondii, we first examined the localization of endogenous Tg1 using affinity-purified anti-Tg1 polyclonal antibody. As shown by immunofluorescence microscopy, endogenous Tg1 localized in RH parasites predominantly to a region anterior to the nucleus (Fig. 3), consistent with the Golgi/TGN of the parasite, although additional punctate and cytosolic staining was observed. Brefeldin A, which disrupts the T. gondii Golgi cisternae, but not the Golgi scaffold (25,37), only partially dispersed the Golgi/TGN staining of Tg1, suggesting that the protein is partially localized to a brefeldin A-resistant compartment(s) in addition to the Golgi.
We then stably expressed wild-type Tg1 tagged with the HA epitope (Tg1-HA). Putative Golgi/TGN localization and partial redistribution by brefeldin A (Fig. 3) were observed for the transgenic Tg1-HA reporter. This result suggests that the transgenic epitope-tagged Tg1 is incorporated into an AP-1 complex, which is appropriately localized in the parasite. Tg1-HA partially co-localized with the transgenic Golgi/TGN marker bacterial alkaline phosphatase-low density lipoprotein receptor (13) as well as with endogenous T. gondii proteins (dynamin and ADP-ribosylation factor-1) (38) with a predominant Golgi localization (Fig. 3B). Partial co-localization with the rhoptry compartment was also detected (Fig. 3B), although this was more accurately discerned by immunoelectron microscopy (see below). We analyzed the localization of Tg1 by qualitative (Fig. 4, A-C) and quantitative (Fig. 4D) immunolabeling of cryosections. Consistent with the immunofluorescence localization, endogenous Tg1 localized to the Golgi/TGN, to juxtaposed coated Golgi-associated vesicles and tubules and endosomes, and to both the rhoptry membrane and membranous lumen. No significant labeling was observed for dense granules, mitochondria, or the nucleus, although microneme labeling was also detected. Altogether, the results suggest that Tg1 is localized to multiple compartments in the secretory and endocytic pathways, including mature secretory organelles.
Mutation of Residues in Tg1 Responsible for Tyrosine Binding Does Not Alter Tg1 Localization-We next sought to disrupt Tg1 function to assess effects on transport through the secretory pathway and on organelle biogenesis. We have previously shown that Tg1 binds to the ROP2 cytoplasmic tail in a fashion dependent upon the YEQL motif in the ROP2 cytoplasmic tail (13). We therefore introduced mutations in Tg1 at residues previously shown in mammalian adaptins to bind to the YXX⌽ motif, specifically D176A and W415A (18,30), and stably expressed these mutants in T. gondii.
The Tg1(D176A)-HA mutant localized to the Golgi/TGN region and to punctate brefeldin A-resistant clusters (Fig. 3), suggesting that the protein was incorporated into an adaptin complex. In contrast, the W415A mutant was diffusely distributed in the cytosol (data not shown), suggesting that the protein failed to associate with other chains of the AP-1 complex. Quantification of immunostained cryosections indicated that Tg1(D176A)-HA was still associated predominantly with the Golgi/TGN, Golgi-associated vesicles and tubules, rhoptries, and micronemes (Fig. 4D). Overall, the relative association of Tg1(D176A)-HA with organelles was nearly identical to that of endogenous Tg1, further supporting a broad organellar distribution for AP-1 in T. gondii irrespective of interaction with cargo.

Binding of Tg1(D176A)-HA to Tyr-based Motifs Is Inhibited
Compared with That of Tg1-HA-We sought to confirm that introduction of the D176A and W415A mutations altered binding of Tg1 to tyrosine-based motifs. Binding of Tg1 to the cytoplasmic tails containing putative YXX⌽ motifs of dense granule (GRA4, YAEL; and GRA7, YRHF) and rhoptry (ROP2, YEQL; and ROP4, YREM) proteins was tested by yeast twohybrid analysis.
Tg1 bound to the cytoplasmic tail of ROP4 to a greater extent than to that of ROP2. In contrast, Tg1 did not bind to YXX⌽ motifs within the putative cytoplasmic tails of the dense granule proteins GRA4 and GRA7 (Fig. 5A). The avidity of Tg1 binding to the ROP2 tail was increased when a triplicated YEQL motif was added to the truncated tail (ROP2.3Y) (Fig. 5,  A and B), as previously reported (13). Importantly, both D176A and W415A mutations impaired binding to ROP2.3Y and ROP4 (Fig. 5B), confirming the functional roles of the conserved residues within Tg1 and the utility of the mutants as dominantnegative inhibitors of endogenous 1 function. Because the W415A mutant was not incorporated into an AP-1 complex, the remainder of the experiments were conducted with the D176A construct.
Dominant-negative Tg1(D176A) Alters Sorting of Rhoptry Proteins-We evaluated the effects of stably overexpressing the wild-type and D176A constructs on protein sorting and organelle biogenesis in T. gondii. The most striking phenotype as shown by immunofluorescence microscopy was an alteration in the pattern of staining with antibody to ROP2 (Fig. 6). In the majority of parasites overexpressing Tg1-HA, ROP2 labeling was no longer restricted to the defined apical cluster of mature rhoptries observed in wild-type parasites (upper panels), but was more fragmented into vesicles and tubules (middle panels). In parasites overexpressing dominant-interfering Tg1(D176A)-HA, the labeling of ROP2 decreased drastically in intensity and was detectable as thin tubules and vesicles FIG. 9. Targeted depletion of Tg1 by antisense RNA alters ROP2/3/4 localization and impairs parasite growth. Clones depleted of Tg1 by antisense RNA (1AS) isolated from two different experiments were immunoprecipitated from 10 6 RH and 1AS parasites using anti-Tg1 and anti-nucleoside-triphosphate hydrolase (NTPase) polyclonal antibodies, separated by SDS-PAGE, and blotted with the same antibodies (A). Up to 90% of Tg1, but not nucleoside-triphosphate hydrolase, was depleted in 1AS clones. Normal vacuoles in RH and distorted vacuoles commonly observed in 1AS clones were labeled by anti-ROP2/3/4 monoclonal antibody (B). Increased Tg1 depletion as quantitated by densitometry correlated directly with an increase in the percentage of vacuoles with distorted morphology (data not shown). extending nearly the length of the cell. In contrast, there were no detectable differences in the targeting of glycosylphosphatidylinositol-anchored proteins targeted to the cell surface (SAG1 (surface antigen glycoprotein-1)) or of dense granule proteins (GRA3 and GRA4) to mature dense granules (data not shown). Of note, there were no differences in the kinetics or extent of processing of precursor ROP2 to mature ROP2 when native and Tg1(D176A)-HA-expressing parasites were compared (data not shown). This indicates that ROP2 transport to a post-Golgi processing compartment is not substantially blocked by the dominant-negative Tg1(D176A) construct.
Disruption of Tg1 Function Alters Endosome and Rhoptry Structure-The above alterations in staining for ROP2 were defined more precisely and quantitatively by electron microscopy. In comparison with RH, rhoptries in Tg1(D176A)-HA parasites were thinner (Fig. 7B), with an overall 2.1-fold decrease in size (volumetric density of 0.013 for rhoptries in RH versus 0.006 for those in Tg1(D176A)-HA) (Fig. 7A). As shown by immunoelectron microscopy, the localization of ROP2 was substantially altered in the Tg1(D176A)-HA-expressing parasites (Fig. 8). Thin elongated rhoptries extending nearly the length of the cell were stained. Large tubular and multivesicular structures consistent with endosomes were most prominently labeled, in comparison with native parasites. ROP2 accumulated in electron-dense structures resembling the sites to which ROP2 with mutations in the YXX⌽ motif localized (Fig. 8, ROP2⌬ panel). On the other hand, minimal accumulation of ROP2 label was observed in the T. gondii Golgi, largely excluding the notion that AP-1 mediates direct transport from the Golgi to rhoptries. Overall, the results indicate that the AP-1 adaptin mediates transport of ROP2 from a post-Golgi endosomal compartment to the mature rhoptry organelle.
Targeted Depletion of Tg1 Alters Golgi, Endosome, and Rhoptry Structures as Well as Parasite Segregation-Finally, we investigated the consequences of lowering the expression level of Tg1 using the recently described antisense mRNA strategy (29). Immediately after isolation of Tg1 antisense stable clones, up to 90% depletion of Tg1 was detected by immunoblotting (Fig. 9A). In contrast, no alteration in expression of the dense granule protein nucleoside-triphosphate hydrolase was observed, confirming the specificity of the antisense effect.
Dramatic effects on parasite morphology were observed. As shown by differential interference contrast microscopy (Fig. 9B), Ͼ50% of the vacuoles contained large irregular cells suggestive of defective parasite division. This morphologic defect was shown by electron microscopy instead to reflect parasites that had completed division, but in which the plasma membranes of neighboring cells were still closely associated; hence, parasites were aggregated in the vacuole (data not shown). Staining for ROP2 was aberrant, clumped, and irregularly distributed.
Parasites depleted of Tg1 had swollen Golgi cisternae and an accumulation of lucent endocytic vesicles containing internal vesicles and membranes and also of immature rhoptries (Fig. 10A, IR), containing the honeycombed appearance characteristic of rhoptry luminal contents. There was a 7-8-fold increase in volumetric density of both endosome-like structures as well as immature rhoptries (Fig. 10B) in the 1AS6 clone. There was no significant alteration in the density of dense granules and micronemes. These results substantially extend the evidence that Tg1 mediates rhoptry biogenesis from a post-Golgi compartment with characteristics of multivesicular endosomes and immature rhoptries while also supporting an additional role for Tg1 in transport from the Golgi.

DISCUSSION
Delivery of ROP2 and ROP4 from the TGN to rhoptries can occur by three possible routes. The first pathway is direct transport from the TGN to mature rhoptries. This pathway is unlikely because ROP2 mutated to remove either tyrosinebased or dileucine motifs is not retained in the TGN, but rather in a post-Golgi organelle (Ref. 13 and this study). The second possibility is that rhoptry proteins are delivered first to the plasma membrane and then internalized to the endosomal/ lysosomal and rhoptry pathway. We do not favor this explanation because neither native nor mutant ROP2 is observed at the plasma membrane at steady state as shown by immunofluorescence (13) and immunoelectron (this study) microscopy localization. The third possibility is that rhoptry proteins are delivered from the TGN to an endosomal compartment and subsequently to the mature rhoptry. The data presented in this work strongly support this notion and directly implicate AP-1 in the process. The data contained herein (in particular, see Fig. 10) also support the expected involvement of AP-1 in transport from the Golgi. Our model for the role of T. gondii AP-1 in protein transport is illustrated in Fig. 11.
We had previously suggested (24,39) that rhoptries are lysosome-like organelles, analogous to melanosomes and platelet-dense granules (11,12). This raised the possibility that ROP2 was sorted by AP-3, especially because of the acidic "AP-3-like" YXX⌽ binding motif (EYEQL) (23) within the ROP2 cytoplasmic tail. Three lines of evidence argue against a role for AP-3 in rhoptry protein transport. First and by far the most important, no 3 or ␤3 chains are present in the T. gondii genome (of interest, the same is true in the closely related Plasmodium falciparum). Second, as shown by two-hybrid analysis, ROP2 (13) and ROP4 5 bind substantially less well to human 3 than to human 1. Finally, a chimera between the ROP2 ectodomain/transmembrane domain and the cytoplasmic tail of human tyrosinase (targeted to melanosomes in an AP-3-dependent fashion) (40) did not localize to rhoptries when expressed in T. gondii. 6 Taken together, these results implicate AP-1 rather than AP-3 in the transport of rhoptry proteins from a post-Golgi endosomal compartment to mature rhoptry organelles. The absence from the T. gondii genome of other adaptors that mediate endocytic transport, including GGA proteins, stonins, and ␤-arrestins (26), lends further credence to the argument that Tg1 is responsible for protein transport to the T. gondii rhoptry.
The crystal structure of a rat 2 fragment (residue 158 -435) bound to tyrosine-containing peptides resolved the critical residues that interact with the aromatic ring (Phe 174 and Trp 421 ) and phenolic hydroxyl group (Asp 176 , Lys 203 , and Arg 423 ) of the critical tyrosine (18). An independent mutational analysis of dominant-interfering 2 confirmed in vitro that Asp 176 and Trp 421 are critical for the internalization of TGN38 and the epidermal growth factor receptor (30), albeit no changes in the cell morphology or protein localization were detected. We adopted this mutational strategy to analyze the role of the adaptin 1 chain in mediating protein sorting in T. gondii. The partial phenotype of the dominant-interfering D176A mutant in T. gondii is almost certainly due to the presence of endogenous adaptin 1 chain and also to the capacity of other residues (Phe 174 , Lys 203 , and Arg 417 in T. gondii) within Tg1 to interact with the YXX⌽ motif. Although this partial phenotype might not have been apparent in other eukaryotic cells, the unique structure and polarized distribution of rhoptries allowed us to uncover the sorting defect.
It is now well established that the antisense RNA strategy inherently does not generate complete inhibition or complete loss of function of a gene and therefore is not suitable for the creation of a null mutant (41)(42)(43). Targeted depletion of the T. gondii adaptin 1 chain by antisense RNA allowed the stable clone to survive through at least three passages of the host cell prior to our isolation and analyses of single clones. At this 5 M. Yang, I. Coppens, S. Wormsley, H. C. Hoppe, and K. A. Joiner, unpublished data. 6 K. Paprotka and K. A. Joiner, unpublished data.
FIG. 11. Model for the role of T. gondii AP-1 in protein transport. Based on the data obtained during this study, previous publications, and submitted work, we present a model for the role of T. gondii AP-1 in protein transport. AP-1 mediates transport from the T. gondii Golgi to a post-Golgi endocytic compartment, to which the T. gondii early endocytic marker Rab5 (TgRab5) also localizes. Transport from the post-Golgi Rab5-positive compartment may then occur to a second endosomal compartment, bearing functional similarity to multivesicular endosomes (see Footnote 5). Experiments focusing on transport of ROP2 specifically (by mutating motifs in the protein necessary for interaction with adaptors) indicate that delivery of ROP2 from a post-Golgi compartment to the mature rhoptry is a key function for T. gondii AP-1. This does not obviate a predominate role for T. gondii AP-1 in transport of other cargo (that accumulate in the Golgi when AP-1-interacting motifs are mutated) from the Golgi to a post-Golgi compartment. IR, immature rhoptries; MVB, multivesicular body. stage, the mutants grew at an irregular and slow rate, and depletion of Tg1 eventually resulted in the substantial abortion of parasite vacuoles in the host cell. On the other hand, when maintained in continuous passage, 1 levels returned to those in wild-type parasites, and normal growth and morphology were restored, similar to what we observed in clones depleted of ROP2 by antisense mRNA. 6 Protein recovery that is accompanied by loss of phenotypes further supports the specificity of defects due to Tg1 depletion.
We have now identified two separate circumstances in which blocking transport of ROP2 to the mature rhoptry alters rhoptry morphology (this work). 7 These results are most consistent with a model in which transmembrane ROP2 (and perhaps other associated proteins) is required for establishing and maintaining structural integrity of the rhoptry. Whether components analogous to ROP2 exist in other lysosome-like organelles and whether their transport is AP-1-dependent remain to be determined.