The Plasmodium falciparum Vps4 homolog mediates multivesicular body formation

Malarial parasites and Toxoplasma are both members of the apicomplexan family of protozoan parasites. These organisms contain three distinct secretory organelles, micronemes, rhoptries and dense granules, which discharge sequentially during the process of host cell attachment and invasion. Most unusual morphologically are the rhoptries, club-shaped structures containing a long duct through which the luminal contents are extruded at the time of host cell invasion. The rhoptry lumen has a honeycomb-appearance by transmission electron microscopy; it has been postulated that this appearance results from internal membranes and/or vesicles (Bannister et al., 1986), and also that these internal membranes contain trans-membrane proteins. Altogether, this suggests that rhoptries might be related to multivesicular bodies (MVB) in higher eukaryotic cells.

A multivesicular body (MVB) compartment is associated with a variety of functions in eukaryotic cells (Jiang et al., 2002; Piper and Luzio, 2001). Early endosomal MVBs function as endocytic carrier vesicles that deliver contents to late endosomal MVBs for degradation. Internal membranes of late endosomal MVB are enriched in cholesterol and lysobisphosphatidic acid, the latter of which is absent from the limiting membrane. Formation of internal membranes is suggested to involve either invagination of the outer membrane (Odorizzi et al., 1998) or autophagy, with engulfment of cytosol by the delimiting membrane (Teter and Klionsky, 2000). In either case, sorting of specific lipids and proteins must subsequently occur to generate the topologic features of the MVB. Of interest, MVB can participate in either degradation or storage. In the latter case, MVB function as secretory lysosomes, with release of internal membranes as exosomes (Blott and Griffiths, 2002; Stoorvogel et al., 2002). There may even be separate compartments for degradation and lytic functions in secretory lysosomes; plants have a lytic vacuole and a protein storage vacuole that are not clearly segregated in early seed development but subsequently become distinct structures.

Rhoptries are cholesterol enriched (Coppens and Joiner, 2003; Foussard et al., 1991), T. gondii rhoptry biogenesis occurs along the endocytic pathway (Hoppe et al., 2000), rhoptry discharge apparently releases internal membranes, and rhoptries may be a prototype of a lysosome-related organelle (Que et al., 2002). We therefore hypothesized that MVBs intersect with the rhoptry biogenesis pathway in T. gondii and perhaps more generally. To explore this hypothesis, we utilized the P. falciparum homolog (Kissinger et al., 2002) of the yeast gene Vps4 (Babst et al., 1997) (PfVps4). Vps4 is a class E gene in the endocytic pathway, and has an essential role in MVB formation (Odorizzi et al., 1998). The aberrant structures that accumulate when Vps4 function is disrupted sit at the interface between early and late endosomes, and are enriched in cholesterol and lysobisphosphatidic acid (Bishop and Woodman, 2000). Our results indicate that MVB formation occurs along the endocytic pathway in T. gondii and possibly in P. falciparum.

The RH strain of T. gondii was maintained by growth on monolayers of primary human foreskin fibroblasts (HFF). Transient transfection of plasmids into T. gondii by electroporation was carried out as described previously (Hoppe et al., 2000; Roos et al., 1994). Mammalian cell transfection was performed by calcium phosphate precipitation.

The 3D7 strain of P. falciparum provided by Malaria Reseach & Reference Reagent Resource Center (MR4/ATCC 10801 University Boulevard, Manassas, Virginia) was maintained in asynchouronous culture as described (Vielemeyer et al., 2004). Transfection of P. falciparum 3D7 with GFP-PfVps4/pHD22Y and GFP-PfVps4-E214Q/pHD22Y (pHD22Y from MR4) (Fidock and Wellems, 1987) was achieved by the RBC preloading method published previously (Deitsch et al., 2001) with minor modifications. To select for stably transfected parasites, WR99210 (a gift from Jacobus Pharmaceutical Company) was added to the culture at a final concentration of 20 nM, 72 hours following the first electroporation.

The PfVps4 gene was amplified by PCR from P. falciparum genomic DNA with following primers: (sense) CCC AAG CTT ATG GAC TCT GAA GAA ACA ATA AAT TTG and (antisense) ATA GGA TTC TTA TGT ACC GTT CAT TCC ATA TTG. The mutants, PfVps4-K160Q/PCR-Script and PfVps4-E214Q/PCR-Script that have a point mutation in the AAA domain at amino acid 160 (lysine to glutamine) and 214 (glutamic acid to glutamine), respectively, were created by site-directed mutagenesis (Stratagene Menasha, WI). Subcloning was performed using respective constructs (GFP-PfVps4/pEGFP, GFP-PfVps4-K160Q/pEGFP and GFP-PfVps4-E214Q/pEGFP) as templates for PCR and ligated into the AvrII/PacI sites of the T. gondii expression vector, pNTP (Karsten et al., 1997). To generate plasmids with a selectable marker for stable transfection in T. gondii, EGFP-PfVPS expression cassettes were excised with SpeI and subcloned into pC3W1 (Donald and Roos, 1993).

For construction of the PfVps4 stable expression vector, PCR amplified GFP-PfVps4 product was first ligated into XhoI site of the plasmid pHC3CAT (obtained from MR4) (Crabb et al., 1997), from which the CAT gene was excised. The resulting construct, GFP-PfVps4/pHC3, was used as the template to obtain a cassette that contained the CAM promoter, GFP-PfVps4 coding region (WT and E214Q mutant), and the HSP86 3′ UTR. The PCR-amplified cassettes were ligated into the plasmid pHD22Y KpnI to create the expression vectors GFP-PfVps4/pHD22Y and GFP-PfVps4-E214Q/pHD22Y.

The N-terminal fragment containing 102 amino acids and the C-terminal fragment containing 100 amino acids of PfVps4 were separately cloned into BamHI/HindIII sites of the E. coli expression vector pET28a (Novagen, Madison, WI) in which the peptides are fused to a 6-His tag. The fusion proteins were purified by the His.Bind Purification kit (Novage). A mixture of His/PfVps4p^NT and His/PfVps4p^CT antigens was used to immunize Super Clean New Zealand White rabbits. Purified N-terminal and C-terminal antigens were used separately to inoculate mice (Cocalico Biologicals, Reamstown, PA). Antisera were passed through an AminoLink Plus Coupling Gel column (Pierce Rockford, IL 61105) loaded with induced pET28a lysate before affinity purification. The pre-cleared antisera were then loaded onto the AminoLink column coupled to the purified specific antigens (His/PfVps4 peptides). After repeated washing, the antibodies were eluted with 100 mM glycine buffer (pH 3). The eluted antibodies were then desalted and stored in PBS containing 1% BSA.

Immunofluorescence microscopy with T. gondii-infected cells was conducted as previously described (Hoppe et al., 2000). For co-localization with GFP, a goat anti-mouse-Alexa Fluor 594 conjugate (Molecular Probes Eugene, OR) (1:1000 in dilution) was used. For P. falciparum, parasite-infected RBC (iRBC) were collected onto glass slides for observation by fluorescence microscopy.

For thin-section transmission electron microscopy (TEM), cells were fixed in 2.5% glutaraldehyde (Electron Microscopy Sciences; EMS) in 0.1 M sodium cacodylate buffer (pH 7.4) for 1 hour at room temperature, as previously reported (Coppens and Joiner, 2003). For immunoelectron microscopy, the cells were fixed for 2 days at 4°C with 8% paraformaldehyde (PFA, EMS) in 0.25M HEPES (pH 7.4). The cells were then infiltrated, frozen cryosectioned, and labeled as described (Folsch et al., 2001) with the difference that mouse anti-Vps4 antibodies were used (1:50 in PBS/1% fish skin gelatin), followed directly by 5 nm protein A-gold (Department of Cell Biology, Medical School, Utrecht University, the Netherlands).

For Plasmodium RBC infected with GFP-PfVps4-WT, stably transfected cells were disrupted by addition of 0.05% saponin. The iRBC lysate was centrifuged for 5 minutes, at 500 g, at 4°C and washed twice with PBS before being resuspended in the homogenization buffer (20 mM HEPES/KOH pH 6.8; 50 mM potassium acetate; 200 mM sorbitol; 1 mM EDTA; protease inhibitor cocktail; 1 mM PMSF). For Toxoplasma samples, HFF cells infected with GFP-PfVps4-WT and GFP-PfVps4-E214Q were harvested by scraping and disrupted by passing through a syringe needle (gauge 27). The disrupted cells were then centrifuged at 500 g for 10 minutes at 4°C before being resuspended in the homogenization buffer.

The homogenized samples (extracellular P. falciparum and T. gondii) were repeatedly frozen in liquid nitrogen and thawed in a 42°C water bath followed by a low-speed centrifugation to discard the debris. The supernatants were then centrifuged at 100,000 g for 1 hour at 4°C (Beckman TLX ultracentrifuge Rotor: TLA120.1). Both the pellets (total membrane fraction) and supernatants (cytosolic fraction) were collected. The membrane fractions were sonicated and resuspended in the homogenization buffer to a final volume equal to that of the supernatants before the high-speed centrifugation. Approximately 10⁷ of parasites in each sample were used to be resolved on a SDS-PAGE gel.

Lysates of parasites (10⁷ in number) or subcellular fractions (prepared as described above) were probed by immunoblot, as previously described (Hoppe et al., 2000). Rabbit anti-PfVps4 antibodies (1 μg/ml) and/or anti-GFP (1:5000 in dilution, a kind gift from Graham Warren's lab, Yale School of Medicine) were followed by protein A/G-HOURP (Pierce Rockford, IL) as a secondary detection reagent.

To visualize fluorescent cholesterol associated with GFP-PfVps4-E214Q and GFP-PfVps4-WT, infected cells were grown on coverslips in a 24-well plate and incubated in culture medium containing 10% lipoprotein deficient serum (LPDS). After 24 hours, cells were labeled with 0.1 mg of [NBD-Cholesterol]-LDL for 2 hours. Coverslips with live cells were directly observed with an epifluorescence microscope. For lipid droplet staining, GFP-PfVps4E214Q cells were cultured in medium containing 10% FBS, then washed and fixed in PFA followed by staining with oil red O (El-Jack et al., 1999). For localization of cholesterol, COS-7 cells on a coverslip were transiently transfected with wild type and/or mutated VPS constructs by calcium precipitation and allowed to grow in LPDS for 24 hours at 37°C. Two hours before harvest, the cells were boosted with medium containing 10% FBS plus 10% LDL. Following washing and fixing in 3.7% PFA, cells were stained with 10 μg/ml of filipin in PBS for 40 minutes at room temperature. After repeated washing, samples were mounted on a microscope slide and observed by fluorescence microscopy.

The P. falciparum genome encodes a single Vps4 homolog of 1260 nucleotides, which we term PfVps4. A search of the GenBank Databank (http://www.ncbi.nlm.nih.gov/) revealed two existing Plasmodium falciparum 3D7 putative ATPases (accession numbers NP_702437 and AAN37161) that are 100% identical to PfVps4. The open reading frame of PfVps4 encodes a protein of 419 amino acids with a predicted molecular weight of 48,228 Daltons. PfVps4 is found on chromosome 14 [PlasmoDB gene locus PF14_0548, The Plasmodium Genome Database (http://PlasmoDB.org)] of the P. falciparum genome. A Plasmodium yoelii yoelii homolog (GenBank accession number EAA17765) was 84% identical (92% similar) to PfVps4. Other Vps4 homologs are shown in Table 1. PfVps4 is most similar to mouse SKD1, human Vps4B, and Saccharomyces cerevisiae Vps4p. A Schizosaccharomyces pombe BEM1 BUD5 double mutant suppressor protein is also highly homologous to PfVps4.

All Vps4 proteins have a conserved ATPase domain (AAA) (PFAM ID pfam00004, ATPase family associated with various cellular activities). AAA family proteins often perform chaperone-like functions that assist in the assembly, operation, or disassembly of protein complexes. The ATPase domain of PfVps4 is highly homologous to Vps4 ATPase domains, and contains conserved residues required for ATP binding and hydrolysis.

Polyclonal antibodies to PfVps4 were generated for use in localizing the native protein in P. falciparum. Following affinity purification of the antisera either on solid phase or on nitrocellulose, immunoblot of P. falciparum revealed a single band of 48 kDa, consistent with the predicted molecular weight of the protein product (Fig. 1A).

PfVps4 localized to the cytosol of P. falciparum trophozoites, as demonstrated by immunocryoelectron microscopy (Fig. 1B). Morphometric analysis of 26 sections of different parasites showed that 87±6% of gold grains were associated with the parasite cytosol, while 5±2%, 4±3%, 2±1% and 2±1% were associated with the endoplasmic reticulum, unknown lucent vacuoles, nucleus and digestive vacuole, respectively.

We next generated a stable line of P. falciparum expressing a chimera of GFP fused to the N-terminus of PfVps4 (GFP-PfVps4-WT) (Fig. 2A, top construct) (Bishop and Woodman, 2000). Fusions of GFP to the N-terminus of yeast or human Vps4 are reported to localize and function normally (Bishop and Woodman, 2000). Confirming the results by immunocryoelectron microscopy, GFP-PfVps4-WT gave a cytosolic localization pattern by immunofluorescence (Fig. 2B).

The cytosolic location of GFP-PfVps4-WT was confirmed by subcellular fractionation. The predominant signal recognized by anti-GFP migrated at the expected position for GFP-PfVps4 (75 kDa) and was present in the total lysate and the 100,000 g supernatant but not the 100,000 g pellet (Fig. 3A). Anti-GFP did not recognize any bands in the total lysate from infected cells containing non-transfected parasites (3D7 lane). When the same blot was stripped and probed with anti-PfVps4 antiserum, an additional 48 kDa band was present in both transfected and non-transfected parasites (Fig. 3B). This almost certainly represents endogenous PfVps4, which again fractionates exclusively into the supernatant fraction. This latter result confirms and extends the data from immunoelectron microscopy (Fig. 1B).

Mutations were made in the PfVps4 ATP domain (Fig. 2A, middle and lower constructs), based on prior mutational analysis of yeast and human Vps4 (Babst et al., 1998; Bishop and Woodman, 2000). The K160Q mutation is predicted to interfere with ATP binding, while the E214Q mutation is predicted to interfere with ATP hydrolysis. In other systems, both mutants act as dominant negatives.

The GFP-PfVps4-E214Q mutant was transiently expressed in P. falciparum. In comparison with the situation with wild type PfVps4, the ATP hydrolysis mutant localized predominantly to discrete punctuate GFP-positive structures in P. falciparum trophozoites (Fig. 2C). Attempts to generate stable P. falciparum lines expressing GFP-PfVps4-E214Q were unsuccessful, presumably due to toxicity of the dominant negative PfVps4 homolog.

We therefore turned to the closely related organism, Toxoplasma gondii, to extend our studies. This general strategy is one we have used previously (Beckers et al., 1997). Both wild type and mutant PfVps4 were expressed in T. gondii (Fig. 4A). Diffuse staining of the cytosol was observed with GFP-PfVps4-WT. Occasional parasites had more discrete yet irregular areas of enhanced fluorescence. In contrast, the EQ mutant of GFP-PfVps4 typically revealed 2-5 discrete punctate GFP-positive structures, distributed throughout the cytoplasm, similar to the appearance in P. falciparum. Hence, the wild type and mutant GFP-PfVps4 behaved in T. gondii analogously to their behavior when expressed in P. falciparum.

Localization of wild type and mutant GFP-PfVps4 expressed in T. gondii was also assessed by subcellular fractionation (Fig. 3C,). With GFP-PfVps4-WT, the transgenic protein migrated as a single band at the expected molecular weight of 75 kDa (three right-hand lanes in Fig. 3C). The protein fractionated exclusively to the supernatant. For GFP-PfVps4-E214Q (three middle lanes) a portion of the transgenic protein fractionated into the pellet. An additional band at 66 kDa was also observed with the EQ mutant (12% of total), although this band was largely absent from the pellet fraction, even with longer exposure (not shown). This latter result suggests that only full length GFP-PfVps4-E214Q associates with the membrane of the aberrant structures visualized by immunofluorescence. No signal was detected in non-transfected T. gondii probed with anti-GFP (three left hand lanes).

We sought to determine whether mutant GFP-PfVps4 localized to known organelles in T. gondii. While mutant GFP-PfVps4-positive structures were reminiscent in shape to dense granules, no co-localization was observed with two dense granule markers GRA3 (Fig. 4B, GRA3 panel) or NTPase (not shown). Similarly, mutant GFP-PfVps4-positive structures did not co-localize with endogenous rhoptry (ROP2,3,4) or microneme (MIC2) proteins (Fig. 4B). Of note, rhoptry morphology as detected by ROP2,3,4 staining was preserved in the EQ mutant.

We assumed that the PfVps4 structures were most likely of endosomal origin. We therefore examined co-localization between mutant PfVps4 and T. gondii Rab5 (TgRab51), a marker of the early endosomal compartment that localizes to a post-Golgi structure in T. gondii (Robibaro et al., 2002). The PfVps4 structures were partially overlapping (Fig. 4C) with the early endocytic compartment in T. gondii, consistent with the notion that PfVps4-positive mutant compartments in mammalian cells can express both early and late endosomal markers.

We had previously demonstrated that T. gondii rhoptry biogenesis is mediated, in part, by endocytic targeting machinery within the parasite (Ngô et al., 2003) and endocytic motifs within ROP2 (Hoppe et al., 2000). We therefore asked whether compartments known to intersect with the rhoptry biogenesis pathway overlapped with PfVps4. Nearly perfect co-localization was observed between mutant GFP-PfVps4 and the ROP2 (ΔYEQL)-HA compartment (Fig. 4D). This latter structure is the compartment in which rhoptry proteins accumulate if deleted of endocytic sorting motifs within the cytoplasmic tail (Hoppe et al., 2000; Ngô et al., 2003), or if function of the AP-1 adaptor is impaired (Ngô et al., 2003). Although we assume that the TgRab51 and ROP2 (ΔYEQL)-HA are also overlapping, and may in fact be nearly identical, our current reagents do not allow a test of that hypothesis. Nonetheless, the results make clear that PfVps4 is on the endocytic pathway in T. gondii, and by inference in P. falciparum. This vesicular compartment accumulates rhoptry proteins in T. gondii when transport to the mature organelle is impaired.

Mutant GFP-PfVps4 structures with T. gondii were cholesterol enriched, as determined by staining with filipin as a fluorescent probe for sterol detection (not shown). In contrast, the structures did not co-localize with parasite lipid droplets (Fig. 5A), which contain predominantly cholesteryl esters and triglycerides (Coppens et al., unpublished), despite the similarity in size and shape between the two structures at the light microscopic level. When parasites were loaded with NBD-cholesterol, punctate structures were observed in Toxoplasma expressing mutant PfVps4 (Fig. 5B). These punctuate structures are likely to represent the multivesicular bodies generated by expression of mutant PfVps4, and known to be stained by filipin in other systems (Bishop and Woodman, 2000). NBD-cholesterol was more diffusely distributed in wild-type parasites. This suggests that PfVps4 is involved in the cholesterol trafficking pathway, at least when expressed in T. gondii.

TEM of T. gondii stably expressing GFP-PfVps4E214Q showed the presence of enlarged compartments (arrows) containing intraluminal vesicles surrounded by one membrane (Fig. 6A,B). Moreover, some parasites exhibited a stack of curved cisternal membranes (panel C, arrow), reminiscent of the class E compartment in yeast (Babst et al., 1997; Odorizzi et al., 1998). The multivesicular compartments were clearly labeled on cryosections with anti-GFP antibodies conjugated to 5 nm gold (panel D, arrowheads). The gold labeling was seen on the limiting compartment membrane and on the perimeter of the internal vesicles, suggesting that the latter may be generated by budding into the interior of the compartment. As was observed by immunofluorescence, rhoptry morphology was preserved in the EQ mutant when assessed by TEM studies.

While showing that PfVps4 localized to the endocytic and rhoptry biogenesis pathways in T. gondii and possibly in P. falciparum, the above data did not explicitly demonstrate that PfVps4 has a conserved function in higher eukaryotes. We therefore expressed PfVps4 in COS cells. The majority of the GFP-PfVps4-WT fluorescence was cytosolic (Fig. 7, GFP panels), with occasional discrete areas of more intense staining. In contrast, both the ATP binding and ATP hydrolysis mutants of PfVps4 generated large vesicular structures distributed throughout COS cells, which had GFP fluorescence in a ring-like pattern (Fig. 7, GFP panels). Some punctate areas of internal GFP fluorescence were also observed. These distinctive structures are indistinguishable in size, shape and staining characteristics to those previously reported in a variety of mammalian cell types with over-expression of dominant negative PfVps4 (Bishop and Woodman, 2000). These structures contain early and late endosomal markers, and the morphology of MVB. This result suggests that PfVps4 has a conserved function across wide phylogenetic boundaries, supporting the hypothesis that PfVps4 is on the endocytic pathway in T. gondii, and possibly in P. falciparum.

The distribution of cholesterol in cells expressing wild type and mutant PfVps4 was examined (Fig. 7, filipin panels). Filipin staining to detect cholesterol showed plasma membrane and faint tubulovesicular staining in COS cells transiently transfected with wild type GFP-PfVps4. In contrast, filipin staining of GFP-PfVps4 mutant cells illustrated intense staining of all enlarged endosomal structures marked by GFP-PfVps4. The filipin staining was distributed throughout the lumen of the compartment, probably reflecting accumulation of internal membranous structures. These results mirror previously reported findings in mammalian cells (Bishop and Woodman, 2000), and further support conservation of PfVps4 function in mediating cholesterol transport in the endocytic pathway.

Increasing evidence suggests that rhoptries are formed on the endocytic pathway (Hoppe et al., 2000; Joiner and Roos, 2002; Ngô et al., 2003). Rhoptry protein cytoplasmic tails interact with the μ chain of the adaptor complexes involved in formation of clathrin-coated vesicles, and dominant negative versions of T. gondii AP-1 or anti-sense down-regulation of AP-1 impair rhoptry biogenesis (Ngô et al., 2003). A T. gondii cathepsin B homolog localizes to rhoptries (Que et al., 2002). Mutations in a tyrosine-based or dileucine-based endocytic motif in the T. gondii rhoptry protein, ROP2, block delivery of the protein to mature rhoptries, and result in accumulation of the mutated protein in a compartment adjacent to the rhoptries. Our data illustrate that this structure in mature tachzyoites of T. gondii is almost perfectly overlapping with the mutant PfVps4 compartment, and hence a compartment with similarity to MVB. The data do not distinguish with certainty whether the MVB/ROP2 (ΔYEQL)-HA compartment is antecedent to the rhoptries, or instead whether ROP2 is diverted to the MVB only upon loss of AP-1 dependent recycling (Valdivia et al., 2002), such as would occur with the ROP2 (ΔYEQL) mutation.

Rhoptries in T. gondii are enriched in cholesterol (Coppens and Joiner, 2003; Foussard et al., 1991). In addition, dominant positive Tgrab51 increases cholesterol accumulation in the organism, probably by facilitating transport through the Golgi apparatus (Robibaro et al., 2002). In COS cells and in T. gondii, the enlarged dominant negative PfVps4 compartment is enriched in cholesterol, as is the case in other cells. While this argues that cholesterol transport occurs via the Vps4 pathway, the explanation for the high cholesterol:phospholipid ratio in rhoptries (1.5:1, higher than compatible with phospholipid bilayers) is not yet explained. Similarly, the presence in T. gondii of lipid droplets containing cholesterol esterified by fatty acids (Coppens and Joiner, unpublished) may or may not be connected to the Vps4 compartment.

Class E vps mutants were initially identified based on a defect in sorting of newly synthesized hydrolases to the yeast vacuole, coupled with an enlarged endosomal (class E) compartment (Raymond et al., 1992). Of the 15 members of this family in yeast, 11 were recently subdivided into three distinct endosome-associated complexes, ESCRT-I, II, and III (reviewed by Conibear, 2002). Ubiquitinated cell surface receptors are recognized by ESCRT-I, a 350 kDa cytosolic complex consisting of VPS23/28/37. Binding leads to activation of ESCRT-II, a 150 kDa cytosolic complex of VPS22/25/36. This complex then leads to the assembly of multiple copies of the core class E machinery, ESCRT-III, on the membrane, from a set of dimeric cytosolic complexes containing the class E genes VPS20/Snf7 and VPS2/24. Two additional class E genes, VPS60 and DID2, are either redundant with, or regulate, ESCRT-III activity. The final step involves interaction of Vps4 with ESCRT-III VPS2/24 complex to catalyze the release of the entire machinery from the membrane.

Of great interest, both the P. falciparum genome (Bahl et al., 2002) and the T. gondii genome encode all of the components of the ESCRT-III complex (VPS20/Snf7, VPS2/24, DID2 and VPS60), but no genes from ESCRT-I or ESCRT-II. This argues against the notion that ESCRT-I and ESCRT-II are necessary for ESCRT-III assembly and function, as suggested in yeast for internalization and degradation of ubiquitinated proteins (reviewed by Conibear, 2002). The results suggest that P. falciparum and T. gondii encode only select but different members of the class E gene family, consistent with the `stripped down' yet specialized nature of their secretory and endocytic pathways (Joiner and Roos, 2002).

We thank Marc Pypaert in the Yale Center for Cell and Molecular Imaging for excellent assistance and scientific input for electron microscopy. This work was supported by NIH grants AI48443, AI30060, TW0194, and by a Burroughs Wellcome Fund New Initiatives in Malaria award to K.A.J., and by an American Heart Association award to IC (SDG-0230079N).