Proteomic Analysis of Rhoptry Organelles Reveals Many Novel Constituents for Host-Parasite Interactions in Toxoplasma gondii*

Rhoptries are specialized secretory organelles that are uniquely present within protozoan parasites of the phylum Apicomplexa. These obligate intracellular parasites comprise some of the most important parasites of humans and animals, including the causative agents of malaria (Plasmodium spp.) and chicken coccidiosis (Eimeria spp.). The contents of the rhoptries are released into the nascent parasitophorous vacuole during invasion into the host cell, and the resulting proteins often represent the literal interface between host and pathogen. We have developed a method for highly efficient purification of rhoptries from one of the best studied Apicomplexa, Toxoplasma gondii, and we carried out a detailed proteomic analysis using mass spectrometry that has identified 38 novel proteins. To confirm their rhoptry origin, antibodies were raised to synthetic peptides and/or recombinant protein. Eleven of 12 of these yielded antibody that showed strong rhoptry staining by immunofluorescence within the rhoptry necks and/or their bulbous base. Hemagglutinin epitope tagging confirmed one additional novel protein as from the rhoptry bulb. Previously identified rhoptry proteins from Toxoplasma and Plasmodium were unique to one or the other organism, but our elucidation of the Toxoplasma rhoptry proteome revealed homologues that are common to both. This study also identified the first Toxoplasma genes encoding rhoptry neck proteins, which we named RONs, demonstrated that toxofilin and Rab11 are rhoptry proteins, and identified novel kinases, phosphatases, and proteases that are likely to play a key role in the ability of the parasite to invade and co-opt the host cell for its own survival and growth.

Rhoptries are specialized secretory organelles that are uniquely present within protozoan parasites of the phylum Apicomplexa. These obligate intracellular parasites comprise some of the most important parasites of humans and animals, including the causative agents of malaria (Plasmodium spp.) and chicken coccidiosis (Eimeria spp.). The contents of the rhoptries are released into the nascent parasitophorous vacuole during invasion into the host cell, and the resulting proteins often represent the literal interface between host and pathogen. We have developed a method for highly efficient purification of rhoptries from one of the best studied Apicomplexa, Toxoplasma gondii, and we carried out a detailed proteomic analysis using mass spectrometry that has identified 38 novel proteins. To confirm their rhoptry origin, antibodies were raised to synthetic peptides and/or recombinant protein. Eleven of 12 of these yielded antibody that showed strong rhoptry staining by immunofluorescence within the rhoptry necks and/or their bulbous base. Hemagglutinin epitope tagging confirmed one additional novel protein as from the rhoptry bulb. Previously identified rhoptry proteins from Toxoplasma and Plasmodium were unique to one or the other organism, but our elucidation of the Toxoplasma rhoptry proteome revealed homologues that are common to both. This study also identified the first Toxoplasma genes encoding rhoptry neck proteins, which we named RONs, demonstrated that toxofilin and Rab11 are rhoptry proteins, and identified novel kinases, phosphatases, and proteases that are likely to play a key role in the ability of the parasite to invade and co-opt the host cell for its own survival and growth.
Apicomplexan parasites are protozoan parasites defined by the presence of a complex of specialized organelles at their apical end (1,2). The phylum Apicomplexa contains many important pathogens of humans and animals, including the causative agents of malaria (Plasmodium spp.) and chicken coccidiosis (Eimeria spp.), as well as some of the opportunistic infections associated with AIDS patients such as Toxoplasma gondii and Cryptosporidium parvum (3). Most Apicomplexa have the capacity to invade and replicate within cells of their vertebrate hosts (1). These host cells range from red blood cells in the case of Plasmodium spp. to virtually any nucleated cells in the case of Toxoplasma. Indeed, T. gondii is capable of infecting almost any warmblooded animal and is arguably one of the most successful protozoan parasites, being of worldwide distribution with a seroprevalence that exceeds 50% for many of its host species (4).
The apical complex that defines these protozoa has been implicated in attachment, invasion, formation of the parasitophorous vacuole, and subsequent growth of the parasites within the various cell types that are invaded (1,5). The complex consists of a microtubular spiral (the conoid) and two sets of apical secretory organelles called the micronemes and rhoptries (6,7). The micronemes are small, tubularshaped (ϳ40 ϫ 100 nm) organelles that contain adhesins that appear to facilitate attachment and invasion (5). These organelles release their contents upon contact with a host cell and include several molecules that span the plasma membrane of the parasite and provide a connection between the host cell and the actin/myosin-based motor of the parasite (8).
The rhoptries are perhaps the most unusual organelles found in the Apicomplexa (9,10). They are large, club-shaped organelles that consist of a bulbous body and a narrow electron-dense neck that extends through the conoid at the apical tip of the parasite. The rhoptry necks serve as ducts through which the contents of the rhoptries are secreted during invasion. The neck portion of the rhoptry may also fuse with the micronemes to facilitate release of these organelles (11). The rhoptries release their contents after attachment has been completed and at the commencement of invasion (12). The secreted contents of the rhoptries include proteins and lipids that are packaged into a crystalline array in the organelle.
The currently known components of the rhoptries of Toxoplasma have been designated ROP proteins (ROP1, ROP2, ROP4, ROP8, and ROP9) (13)(14)(15)(16). In addition, there are two rhoptry proteases (a subtilisin-like protein TgSUB2 and a cathepsin B-like protein, Toxopain-1) and a sodium-hydrogen exchanger TgNHE2 (17)(18)(19). Based on cDNA and genomic sequencing, the predicted amino acid sequence for all eight known rhoptry components includes a typical signal peptide characteristic of eukaryotic secreted proteins. In many cases, rhoptry proteins are initially synthesized as "pre-pro-proteins" (13). Removal of the pro-domain occurs late during the process of transport to the rhoptries, probably within the rhoptries themselves (20 -22). In the case of the ROP1 protein, the pro-domain includes information sufficient for targeting artificial fusion constructs to the rhoptries (23,24). Each of the proteins known so far to be associated with the rhoptries has been localized within the bulbous body of the rhoptry rather than in the constricted necks (15)(16)(17)(18)(19)25). Only a single monoclonal antibody has been found specifically to stain the rhoptry neck region, but the identity of the target of this antibody is unknown (26).
ROP1 and ROP9 are soluble rhoptry proteins whose function is as yet unknown (14,16). ROP2, ROP4, and ROP8 are members of a family of membrane-associated proteins at least one of which (ROP2) has been implicated in recruitment of host cell mitochondria to the exterior face of the parasitophorous vacuole membrane (PVM) 3 (13,15,25,27). This is apparently accomplished by the protein having its processed N terminus exposed to the host cell cytosol where it is "mistakenly" recognized as a mitochondrial import signal (27). Host mitochondria recognize this sequence and in trying to import the ROP2 protein, which is firmly anchored in the PVM, are brought into close proximity to the PVM. The TgSUB2 and Toxopain-1 proteases have been implicated in processing of rhoptry proteins, whereas TgNHE2 is suggested to function in organellar osmotolerance (17)(18)(19).
To understand the contribution of rhoptries to the processes of invasion and establishment of the parasitophorous vacuole for intracellular replication, we have undertaken a comprehensive proteomic analysis of their major protein constituents. We report here 38 novel proteins identified by mass spectrometry of a purified rhoptry fraction from T. gondii. HA epitope tagging and/or immunofluorescence using antibodies to peptides and recombinant proteins corresponding to 12 of these molecules confirms their rhoptry origin. Four antibodies react with the ductlike neck portion of the organelle and establish the first genes identified that encode rhoptry neck proteins in Toxoplasma.

MATERIALS AND METHODS
Parasite and Host Cell Culture-Human foreskin fibroblasts (HFFs) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen) and 2 mM glutamine. T. gondii tachyzoites (RH strain) were maintained in confluent monolayers of HFFs.
Purification of Rhoptries-Purified rhoptries were isolated using a Percoll gradient centrifugation protocol (modified from Refs. 28 and 29) followed by a sucrose floatation gradient as follows. 1.3 ϫ 10 10 tachyzoites (freshly lysed out of host cells) were collected by centrifugation at 1300 ϫ g for 10 min at 4°C. All subsequent steps were carried out at 4°C. The parasites were washed once in PBS and once in R buffer (250 mM sucrose, 10 mM MOPS, pH 7.2, 2 mM dithiothreitol, 1 mM EDTA, Complete protease inhibitor mixture (Roche Applied Science)). The washed parasites were resuspended in R buffer at a concentration of 5 ϫ 10 8 parasites/ml and then disrupted by passage through the French press as described (28)). Intact parasites and large debris were removed by centrifugation at 1300 ϫ g for 20 min. The supernatant was then centrifuged at 25,000 ϫ g for 25 min to obtain a high speed organellar pellet. The pellet was resuspended in R buffer and Percoll added to 30% and centrifuged for 25 min at 61,500 ϫ g. The brownish band (containing a mixed fraction of rhoptries, dense granules, mitochondria, and apicoplasts) near the bottom of the gradient was collected, diluted in R buffer, and pelleted again at 100,000 ϫ g for 90 min to pellet the organelles and remove the Percoll. For the sucrose floatation gradient, the organelles were gently resuspended in 300 l of R buffer and mixed with 2 ml of 60%(w/v) sucrose in S buffer (10 mM MOPS, 2 mM dithiothreitol, 1 mM EDTA, Complete protease inhibitor mixture). This mixture was overlaid with steps of sucrose (48,45,42,39, and 36%) in S buffer and centrifuged in the SW41 rotor at 150,000 ϫ g for 18 h. Fractions were collected, and peak fractions containing rhoptry proteins were pooled, diluted in R buffer, and pelleted at 100,000 ϫ g for 90 min. The purified rhoptry fraction was stored in R buffer at Ϫ80°C.
Carbonate Analysis of Membrane-bound Versus Soluble Proteins-Membrane and soluble fractions from the rhoptry/dense granule fraction were isolated using sodium carbonate extraction as described (31). A freshly prepared organellar R/DG fraction (200 g) was pelleted and resuspended in 10 ml of ice-cold 100 mM sodium carbonate, pH 11.5. The samples were incubated on ice for 30 min, and the membrane fraction was pelleted at 100,000 ϫ g for 90 min at 4°C, resuspended in R buffer, and stored at Ϫ80°C. The soluble fraction was trichloroacetic acid-precipitated, washed in ether, resuspended in 2% SDS, and stored at Ϫ80°C. Equivalent fractions of the soluble and membrane fractions were separated by SDS-PAGE, stained with Coomassie Brilliant Blue R-250 (Coomassie Blue), and analyzed by Western blot. Western blots were probed with antibodies directed against known soluble (GRA3) (28) and membrane-associated (ROP 2/3/4) (13) proteins to confirm the effectiveness of the fractionation.
Mass Spectrometry of Rhoptry Proteins Resolved by SDS-PAGE-Rhoptry proteins were separated on a 10% SDS-polyacrylamide gel and visualized by Coomassie Blue staining. The entire protein gel profile was cut into 51 contiguous gel slices (Fig. 3). Gel slices were cut to include prominent protein bands, but slices were also taken from regions of the gel where there was no visible staining. Each slice was placed into a separate tube and processed for mass spectrometry independently. Gel slices were washed in 100 mM ammonium bicarbonate (1 h) followed by 50% acetonitrile, 100 mM ammonium bicarbonate (30 min), and finally 100% acetonitrile (10 min) before drying in a vacuum centrifuge. In gel digestion of proteins was then performed by incubation of the gel slice in a trypsin solution overnight at 37°C with shaking. The resulting peptides were desalted and concentrated using ZipTip TM pipette tips containing C 18 reverse phase media. The resulting concentrated desalted sample was then analyzed by tandem mass spectrometry (MS/MS) in an AB Q-star Pulsar. Tandem mass spectra were acquired in the following way, The time-of-flight-MS scan was run for 10 s to detect charged ions with a mass to charge ratio between 400 and 1000. Mass spectra were then acquired using a data-dependent acquisition mode in which each time-of-flight-ms full scan mass spectrum was followed by collisioninduced dissociation of the eight most intense parent ions. Fragmentation was performed in positive polarity, and mass spectra were acquired for 15 s in the range of 50 -2,000 m/z. Data Base Searching and Bioinformatics Analysis-The resulting MS/MS spectra were interpreted using the MASCOT software by querying locally installed versions of the 10ϫ Toxoplasma data base (ToxoDB, www.toxodb.org/). Partial carboxymethylation and oxidation of methionine residues were considered in the search. Our local data base containing a total of 881,411 T. gondii sequences was made up of the following: 790,769 genomic open reading frames Ͼ50 amino acids long; 69,764 EST open reading frame Ͼ50 amino acids; 4,954 Glimmer HMM protein predictions; 8,336 TigrScan protein predictions; 7,588 TwinScan protein predictions. For each identified protein, the exact position of each peptide was checked manually against alignments of genome contigs, EST positions, and protein prediction models using a generic gene model organism data base construction set (GBrowse www.gmod.org/) available for T. gondii at www.toxodb.org/. Where possible, each protein was assigned a TwinScan number. In cases where a matching TwinScan was either not available or ambiguous, contig coordinates were assigned for future identification of the protein.
Generation of Anti-peptide Antibodies-Antibodies were raised against synthetic peptides deduced from sequences ROP10 (TgTwin-scan_5042) (CEVDVFNPPLATKSD) and TgTwinscan_2001 (CVKLV-LYDDKDTALR) that contain an N-terminal cysteine to aid in coupling to carrier proteins. The peptides were coupled to Imject-activated keyhole limpet hemocyanin (Pierce) according to the manufacturer's instructions. The peptide-keyhole limpet hemocyanin conjugates were then dialyzed against PBS, and 100 g was injected into BALB/c mice in RIBI adjuvant (Corixa). Identical boosts were given at 21-day intervals. Polyclonal antisera was collected after the second boost and screened for reactivity by immunofluorescence analysis (IFA).
Generation of Antibodies to Recombinant Protein-For expression in bacteria, portions of the genes encoding predicted rhoptry proteins were PCR-amplified and subcloned into either the pET28a (Novagen) vector or the pET161GW-D-TOPO (Invitrogen) vector that encode additional N-and/or C-terminal (His) 6 tags for rapid purification. Plasmids were sequenced to verify the junctions of the vector and insert and then transformed into Escherichia coli BL21DE3 cells for expression. Bacteria containing the expression constructs were grown to an A 600 of 0.6 and induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside for 5 h. His-tagged proteins were purified using nickel-nitrilotriacetic acidagarose (Qiagen) under denaturing conditions and eluted using low pH according to the manufacturer's instructions. Eluted proteins were dialyzed against PBS and 10 -50 g injected per immunization into BALB/c mice. Polyclonal antiserum was collected after the second boost and screened by IFA and Western blot analysis. The protein name, region expressed (in amino acids), and oligonucleotide primers used for amplification are indicated in TABLE ONE.
Localization of RON4 to the Rhoptry Necks by Immunoelectron Microscopy-Lung and peritoneal exudate from T. gondii acutely infected mice containing tachyzoites and brain from chronically infected mice containing bradyzoites were processed for immunoelectron microscopy as described previously (32). Thin sections were cut and mounted on Formvar-coated nickel grids. The grids were floated on drops of 1% BSA in PBS to reduce nonspecific staining. Sections were then floated on drops of rabbit anti-RON4, and after washing the antibodies were localized using goat anti-rabbit Ig conjugated to 10-nm gold particles. The rabbit anti-RON4 serum used for immunoelectron microscopy was raised to the bacterially expressed RON4. Sections were stained with uranyl acetate prior to examination in the electron microscope.
Fluorescence Microscopy and Co-localization Studies-For IFA, HFFs on glass coverslips were infected with RH strain parasites for 30 h before fixation in either 2.5-3.5% formaldehyde in PBS for 15 min or 100% ice-cold methanol for 3 min. The formaldehyde fixative was quenched in PBS, 100 mM glycine for 5 min, and then the coverslips were washed in PBS and blocked in PBS, 3% BSA. The samples were permeabilized in PBT (PBS, 3% BSA, 0.1% Triton X-100) for 20 min and incubated in primary antibodies diluted in PBT for 1 h. The primary antibodies used were mouse polyclonal antisera generated against peptides for ROP10 (diluted 1:500) and TgTwinscan_2001 (1:200) and against His 6 -tagged proteins for ROP5,11-15 (1:200), RON1-4 (1:200), toxofilin (1:500), and Rab11 (1:300). Rabbit anti-ROP2 antiserum diluted 1:1000 (a gift from Jean-Francois Dubremetz (15)) was used for co-localization to the rhoptries. The samples were then washed in PBS and incubated in secondary antibodies (Alexa 488-conjugated goat anti-mouse and Alexa 594-conjugated goat anti-rabbit; Molecular Probes) diluted 1:2000 in PBT for 1 h. Following washing in PBS, the samples were mounted in Vectashield and viewed by fluorescence microscopy. The microscopes used were an Olympus BX60 (100ϫ oil immersion objective) with images collected with a Hamamtsu Orca digital CCD camera using Image Pro Plus 4.0 software and a Zeiss Axioskop (100ϫ oil immersion objective) with images collected with Zeiss Axiovision software.
HA Epitope Tagging-ROP16 was PCR-amplified from RH genomic DNA using the forward primer TGCCTCAACTGTAAACGTCTT and reverse primer TTACGCGTAGTCCGGGACGTCGTACGGGTAC-4 J. Kronegg and D. Buloz, personal communication.

Oligonucleotide primers used for amplifying sequences encoding regions of proteins for bacterial expression
The protein name, region expressed, and oligonucleotide primers used (5Ј-3Ј) for amplification are listed in the table.
ATCCGATGTGAAGAAAGTTC. The forward primer was designed to anneal ϳ2 kbp upstream of the ATG start codon as predicted by EST sequences; the resulting region should include all of the 5Ј-untranslated region and most or all of the ROP16 promoter. The reverse primer contains the sequence for a single HA tag at the C-terminal end of the protein. The amplified PCR product was cloned into the pHEX2 plasmid containing the hypoxanthine-xanthine-guanine phosphoribosyltransferase (HXGPRT) gene for use in selecting transformants (37). The resulting vector was transfected into RH parasites lacking HXGPRT expression, and stable clones were derived using mycophenolic acid selection (37). The resulting HA-tagged ROP16 was localized by immunofluorescence microscopy using the fluorescein-conjugated 3F10 anti-HA antibody (Roche Applied Science). The same approach was used for HA-tagging toxofilin except the amplifying oligonucleotides were forward primer GTGCTCGGTTCCATATGTTAG and reverse primer (encoding the HA tag fused to the 3Ј-end (C terminus) of the toxofilin coding region) TTACGCGTAGTCCGGGACGTCGTACG-GGTACGACGAGGGCATAGCGCCTAC.

RESULTS
To understand the function of rhoptries, we embarked on a comprehensive proteomic analysis of their contents in T. gondii tachyzoites. This first required the development of a protocol for obtaining highly purified preparations of these organelles in quantities sufficient for MS analysis (Fig. 1A). To do this, we began with the Percoll gradient protocol of Dubremetz and co-workers (28,29), which isolates a fraction enriched for R/DG but also contains contaminating mitochondria and plastids. To purify further the rhoptries for our study, this R/DG fraction was then subjected to sucrose gradient floatation using a gradient of 36 -60% sucrose. Fractions were analyzed by immunoblotting using antibodies to markers with known locations, and those that were enriched for rhoptry proteins were pooled. The pool of purified rhoptry fractions was compared with the R/DG preparation by Western blot analysis, using antibodies to known proteins from the rhoptries (anti-ROP2/3/4) (13), dense granules (anti-GRA3) (28), and mitochondria (anti-HSP60, Stressgen Biotechnologies). As shown in Fig. 1B, sucrose floatation enriches for the rhoptries and efficiently removes the vast majority of dense granules and contaminating mitochondria.
In preliminary experiments using the R/DG fraction, the organellar material was subjected to two-dimensional SDS-PAGE, and prominent spots were excised for MS analysis (data not shown). These revealed a large number of dense granule components (GRA proteins), but the representation of rhoptry proteins was poor, presumably because of the hydrophobic nature of many rhoptry proteins making them less amenable to solubilization in the isoelectric focusing step of two-dimensional SDS-PAGE. Thus, although ROP9 was detected in the two-dimensional SDS-PAGE separation, other rhoptry proteins were not, including the ROP2-8 family of proteins. To investigate the hydrophobic nature of rhoptry proteins further and to determine what fraction of rhoptry proteins are embedded in membranes, we fractionated the R/DG material by exposure to sodium carbonate, pH 11.5 (Fig. 2) (31). This treatment releases proteins that are soluble or associated with FIGURE 1. Purification of rhoptries. A, schematic of the isolation of rhoptries from T. gondii. A rhoptry/dense granule (R/DG) preparation was isolated by Percoll gradient essentially as described previously (see "Materials and Methods") (13,29). The R/DG fraction was then subjected to a sucrose floatation gradient to isolate the purified rhoptry fraction that corresponds to the top band in the gradient. B, comparison of the Percoll gradient generated R/DG fraction versus the subsequently purified (sucrose-gradient) rhoptry fraction by Western blot analysis. Sucrose floatation enriches for rhoptries as assessed by ROP2/3/4 staining while removing most of the dense granules (GRA3) and contaminating mitochondria (HSP60). membranes through ionic interactions, leaving behind the proteins that are anchored in the membranes by transmembrane domains or other strong interactions that might interfere with the isoelectric focusing (such as glycosylphosphatidylinositol anchors). The fractionation was successful, as shown by the partitioning of known soluble (NTPase) and membrane-associated proteins (ROP2/3/4) in the R/DG preparation (Fig. 2). Most surprisingly, most of the R/DG proteins detectable by Coomassie Blue staining are found within the membrane fraction. Thus, we concluded that many novel rhoptry proteins were likely to be membrane-bound and not amenable to analysis by two-dimensional SDS-PAGE.
In response to the above experiments, we elected to use conventional one-dimensional SDS-PAGE for resolving the rhoptry proteome into fractions based on gel mobility (i.e. approximate size). The purified rhoptry fraction was separated by 10% SDS-PAGE, the gel was stained with Coomassie Blue stain, and 51 contiguous gel slices excised (Fig. 3). Each of the gel slices was subjected to in-gel trypsin digestion followed by tandem MS to obtain peptide fragmentation data suitable for data base searching. The resulting MS/MS data were used to search the most current version of ToxoDB (www.toxodb.org, version 3.0), a data base of Toxoplasma genomic cDNA sequences and protein predictions models. The genomic sequence for T. gondii is not yet annotated but consists of compiled sequences representing a total of ϳ10 times the size of the Toxoplasma genome (now estimated at ϳ65 ϫ 10 6 base pairs) and is thus likely to contain virtually the entire genome. The cDNA sequences (ESTs) were from ϳ69,000 cDNA clones from a variety of developmental stages but predominantly tachyzoites. In addition, we searched our MS peptide fragmentation data against 4,954 Glimmer HMM protein predictions, 8,336 TigrScan protein predictions and 7,588 TwinScan protein predictions for Toxoplasma. Peptides detected by MS often hit a combination of genomic sequences, EST, and protein prediction sequences that corresponded to the same gene. For each identified protein, the exact position of each peptide was checked manually against alignments of the genome with ESTs and protein prediction models using a generic gene model organism data base construction set (GBrowse www.gmod.org/) available for T. gondii at www.toxodb.org.
TABLE TWO shows the resulting data and includes only unambiguous statistically significant hits in which each protein sequence was identified by two or more distinct peptides. Where possible, each protein was assigned a TwinScan number (TABLE TWO, column 1). In cases where a matching TwinScan was either not available or ambiguous, contig coordinates were assigned for future identification of the protein. The corresponding MASCOT probability score and number of peptides identified are given in column 2 of TABLE TWO. Probability scores over 50 were considered as significant, but in most cases scores were considerably higher (up to 977). The numbers of peptides identified for each protein was correspondingly high (up to 19), with an average of 6.0 peptides per protein hit (n ϭ 51). In many cases the same protein was detected in multiple gel slices (TABLE TWO, column 2), presumably because of carry-over of high abundance proteins from one band to another during SDS-PAGE or partial degradation during sample preparation; in some cases, it may also be due to processing of the protein in vivo. Where a protein was detected in multiple gel slices and multiple MS data sets were obtained, the best probability score with the  . SDS-PAGE for liquid chromatography/tandem mass spectrometry analysis. Rhoptry proteins were separated on a 10% SDS-polyacrylamide gel and visualized by Coomassie Blue staining. The entire protein gel profile was cut into 51 contiguous gel slices. Gel slices were cut to include prominent protein bands, but slices were also taken from regions of the gel where there was no visible staining. Each slice was placed into a separate tube and processed for mass spectrometry independently for subsequent liquid chromatography/tandem mass spectrometry analyses. A diagram of the gel is shown at the right to aid in determining the presence of bands because the relative abundance of the proteins in the fraction made imaging difficult. Each gel slice was given a unique number as shown. The appended a, b, or c indicate regions where a large, densely stained band was subdivided into multiple slices for tryptic digest and MS analysis. OCTOBER 7, 2005 • VOLUME 280 • NUMBER 40

Identification of rhoptry proteins by MS
Each protein identified by MS is listed either by Twinscan number or by TGG sequence coordinates (column 1). Column 2 gives the MASCOT probability score and number of peptides hit for each match; probility scores over 50 were considered as significant. Where a protein was detected in multiple gel slices and multiple MS data sets were obtained, the best probability score with the greatest number of peptide hits is given. The number of the gel slice in which the protein was detected is indicated in column 3. Names for each protein are shown in column 4 where they have been previously assigned (not boldface). For novel rhoptry proteins that we have confirmed by immunolocalization or epotope tagging, new designations (ROP5, ROP11-16, and RON1-4) have been assigned (boldface). Immunolocalization data is shown in column 5 for this study (boldface) or previous studies (italics). Colmn 6 gives the BLAST score probability of the top sequence matches (protein versus protein) for each protein sequence identified by MS, as well as peptide tags representing the actual first and last peptides detected by MS for each protein species. Although our analysis is not strictly quantitative for any of the proteins detected, proteins that appear in multiple gel slices and are also identified by a large number of peptides are likely to represent those of greatest abundance in our rhoptry preparation. Thus ROP2 (TgTwin-Scan_1315) was detected in six gel slices with the best match being hit by 13 peptides. Interestingly, several of the novel proteins were similarly predicted to be of high abundance. Most notably, the protein encoded by TgTwinScan_2103 (later designated ROP5) appeared in 17 gel slices and was hit by 17 peptides. Where peptides identified a previously characterized T. gondii protein, the identity of this protein is given in TABLE TWO, column 4. Protein sequences were subjected to BLAST analysis (protein versus protein), and matches and homologues were recorded along with the probability score in TABLE TWO, column 6. Also shown in TABLE TWO, column 6, are the first and last peptides identified by MS/MS analysis that unambiguously identifies each protein and can also be used to determine the exact position of the relevant gene in the genome.
As expected, several of the known rhoptry proteins were readily detected in our proteomic analyses, including known members of the ROP2/4/8 family and ROP9 (TABLE TWO). Most surprisingly, however, ROP1, TgNHE2, TgSUB2, and Toxopain-1 were not detected in this experiment. The TgNHE2 protein contains 12 transmembrane domains that may have impeded entry of the protein into SDS-polyacrylamide gels (17). The mature form of Toxopain-1 migrates at ϳ30 kDa by SDS-PAGE, which would have migrated at the bottom of the gel used for mass spectrometry analysis and thus may have not been recovered (19). In a subsequent experiment, a second high percentage SDSpolyacrylamide gel (15%) was run, and material was taken for MS analysis as before. In this second experiment, ROP1 was successfully detected by MS/MS analysis (best score 228/4 peptides), but TgNHE2, TgSUB2, and Toxopain-1 were still not detected (data not shown). The apparent absence of these known rhoptry proteins in our fraction cannot currently be explained but is likely due to low abundance and/or co-migration on the gels with a highly abundant protein that masks the signal in MS, a known problem with MS analysis of complex mixtures.
In total, 38 previously unidentified proteins were detected in our proteomic analysis. To characterize further these novel proteins and to verify their localization, a total of 19 synthetic peptides were made corresponding to 12 of the novel polypeptides. Although we have successfully used this strategy in the past for studying the surface antigens of the parasite where the tertiary structure of the protein is known (38,39), this approach proved less successful here. Only two peptides yielded antibody that gave a signal above background in immunofluorescence microscopy. One of these antisera, raised against a peptide from TgTwinScan_5042, gave a localization pattern that was clearly rhoptry based on co-localization with antibodies against the known rhoptry protein ROP2 (Fig. 4A). We designated this confirmed novel rhoptry protein ROP10. The second anti-peptide antibody (corresponding to TgTwinScan_2001) localized to the plastid as shown by co-localization using Hoechst staining of the 35-kb plastid DNA (Fig. 4B) (40). Peptides have the advantage and disadvantage of representing only a single portion of the protein. On the one hand, the peptide can be chosen from a region of unique sequence, and the probability of cross-reaction to a related protein can be reduced. On the other hand, peptides can adopt many shapes and so are more likely to generate nonspecific antibody. In such cases, if the antibodies fortuitously cross-react with some portion of an abundant protein, the resulting signal can be extremely strong. The discrepancy between the identification of TgTwinScan_2001 as a rhoptry protein by fractionation but a plastid protein by the anti-peptide antibody is further discussed below.
To overcome the limitations of the anti-peptide approach and to increase the sensitivity of detection, His 6 -tagged recombinant protein was generated corresponding to 10 of the previously uncharacterized proteins that were identified in this analysis. These proteins were used to raise polyclonal antisera in mice, which were then used in IFA. Fig. 5 shows the results for all 10 that yielded a signal by IFA. The patterns clearly indicate that by comparison to the known rhoptry protein ROP2, all 10 antibodies stain the rhoptries, although interesting variations in the exact pattern of localization are seen. Antibodies raised against 6 of the 10 proteins stain the bulbous bodies of the rhoptries as shown by co-localization with ROP2 (15), which is localized to this subdomain of the organelle (Fig. 5A). One of these, TgTwinscan_2103, has simultaneously been shown by Dubremetz and c-workers (28) to be the ROP5

Proteomic Discovery of Novel Rhoptry Proteins in Toxoplasma
protein, which was previously named by detection with the monoclonal antibody T53E2. 5 We thus propose to retain the name ROP5 for this protein and designate the additional confirmed rhoptry body proteins ROP11-15, respectively (TABLE TWO and Fig. 5A).
As an alternative approach and to facilitate future studies, we also used a strategy involving a C-terminal HA epitope tag to examine the TgTwinscan_6965 gene product. After introducing such a construct into the parasites, IFA with anti-HA antibodies confirmed rhoptry localization for this protein, which we therefore designated ROP16.
Antibodies to the four remaining proteins stain adjacent to but slightly anterior of the ROP2 staining in each parasite. This staining pattern is characteristic of the neck portion of the organelle (24,26). These are the first genes identified that encode rhoptry neck proteins in Toxoplasma, and we thus propose they be annotated as RON1-4 (Rhoptry Neck 1-4) as shown in Fig. 5B. RON4 is unique in that it is also detected in between parasites within the parasitophorous vacuole (Fig.  5B, arrow). In some parasite vacuoles, this signal is also apparent at the posterior end of the parasites, within a matrix of material that may correspond to the spaghetti-like network that is found within vacuoles and/or the residual body, a structure that is an anucleate remnant of the unusual division that these intracellular parasites undergo (data not shown). Similar staining of both the rhoptries and the PV has been seen previously for the (rhoptry body) cathepsin-like protease Toxopain-1 (19). The means by which some proteins are detected while others disappear from the vacuole following invasion is currently unknown.
To confirm localization to the rhoptry necks, antisera raised against RON4 was used in immunoelectron microscopy of T. gondii (Fig. 5C). Gold particles were observed specifically in the rhoptry necks of the parasites and not seen in the rhoptry bodies or other Toxoplasma organelles. Rhoptry neck labeling with anti-RON4 antibodies was most prominent at the junction of the body and neck portion of the organelle and was present in samples from both the tachyzoite and bradyzoite forms of the parasite. In agreement with the IFA results, staining could also be seen in the PV where gold particles were observed at membranous structures, possibly the tubulovesicular network (data not shown). Staining with anti-ROP15, although predominantly rhoptry bulb, also extends somewhat into the duct-like rhoptry necks (Fig. 5A, short  arrow). This is the only example of staining of both subcompartments; with the sensitivity of the reagents used here, all others appear to localize either to the rhoptry bodies or necks but not both.
To understand more about these novel rhoptry proteins, the complete open reading frame was determined for each of their respective genes. This required direct analysis by cDNA sequencing because Toxoplasma genes are frequently interrupted by many introns, and there is no absolutely reliable algorithm currently available for gene prediction in the genomic sequence of this parasite. With the exception of ROP14, all of the proteins contain a predicted signal peptide as expected for rhoptry proteins targeted to an organelle via the secretory pathway (Fig.  5D). Although lacking an apparent signal peptide, ROP14 does contain multiple predicted transmembrane domains, a feature common to other Toxoplasma proteins demonstrated to enter the secretory pathway in the absence of a signal peptide (41). Many of the other signal peptide-containing proteins also contain at least one additional hydrophobic region as follows: ROP5, ROP11, ROP13, ROP16, RON2, and RON3. In some cases, this hydrophobic region could function as a transmembrane domain, and so at least some of these proteins might be integral membrane proteins. It should be noted, however, that hydrophobic regions such as these could serve functions other than to span a membrane. Resolution of the true function of these domains in the proteins described here will require further experimentation. The C terminus of RON1 contains a predicted GPI anchor addition sequence that also suggests membrane association of this protein. GPI anchors have been identified previously in the Plasmodium rhoptry protein RAMA (42) and predicted for the Toxoplasma microneme protein TgSUB1 (43), but the mechanism by which they are directed to these locations (as opposed to the parasite surface) is unknown. The high frequency of hydrophobic domains or predicted GPI anchors agrees with the results of the carbonate extraction (Fig. 2), which indicated many rhoptry proteins are membrane-associated.
As expected for these unusual organelles, BLAST homology searches of proteins identified in the fraction revealed that many proteins were unique to either Toxoplasma or the Apicomplexa as a whole (TABLE  TWO). A number of the novel proteins identified, however, contain domains suggesting protein activities that are clues to their potential function in Toxoplasma infections. Twelve of the proteins identified contain putative protein kinase domains. At least nine of these have some homology to the ROP2 family of rhoptry proteins that have been reported previously (44) to have a kinase domain. We also identified proteins with homology to protein phosphatase 2C (TgTwinscan_7301) and serine and insulinase-type proteases (TgTwinscan_6982 and TgTwinscan_7264).
One of the features of many parasite antigens is short, tandemly repeated sequences within the protein (16,42,45,46). This is also the case here for several of the novel proteins identified in the rhoptry fraction (TABLE THREE). The tandem repeats vary in number (5)(6)(7)(8)(9)(10)(11)(12)(13) and length (8 -18 amino acids) but have some common features. They are not random sequence but are particularly rich in proline and charged residues. Two of the repeats have a pair of acidic residues (Asp or Glu), which have also been found in the repeats in the Plasmodium RAMA protein. The RON4 protein also contains a perfect 44-amino acid repeat (QPPTAAPRTSRSVDTGSGSDASTEQQAGGQKVVTP-IPASKGIYP) that is not repeated in tandem, but separated by 85 amino acids. These repeating sequences may have particular importance to the function of these proteins and/or their trafficking to the rhoptries.
Unexpectedly, our proteomic analysis of the rhoptry fraction also identified the known Toxoplasma proteins Rab11 and toxofilin. To determine whether these proteins localize to the rhoptries or are contaminants of the preparation, we expressed portions of these proteins as  His 6 -tagged fusion proteins and raised antibodies to the purified proteins. Toxofilin is a novel protein that has been shown to interact with parasite actin and a protein phosphatase 2C (47,48). It has been localized previously primarily to an "apical-cytoplasmic" location but with some posterior staining as well. This apical location in Toxoplasma and the presence of a predicted signal peptide at the N terminus of the protein (48) would suggest the rhoptries as a candidate location for toxofilin. Fig. 6A shows that antibodies raised against recombinant toxofilin do indeed stain the rhoptries as demonstrated by co-localization with ROP2. Although difficult to resolve, staining with anti-toxofi-lin may also extend somewhat into the narrow rhoptry necks at the apical portion of each organelle. No other staining was detected in the apical cytoplasm or in posterior locations, contrary to what has been reported previously (48). This rhoptry localization for toxofilin was confirmed using IFA and parasites engineered to express a C-terminal HAtagged version of this protein (data not shown).
Although the intracellular location of Rab11 in Toxoplasma has not been reported, its potential localization to the rhoptries is suggested by recent results in mammalian cells demonstrating that Rab11 co-localizes with the cholesterol-staining probe filipin and filipin also stains the FIGURE 5. Localization of novel rhoptry proteins using antibodies raised against recombinant protein or epitope-tagging. Gene sequences for 10 of the putative novel rhoptry proteins were used to express His 6 -tagged recombinant protein. The proteins were purified over nickel-agarose by virtue of the His 6 tag and were used to immunize mice. The resulting antisera were used for IFA to determine the true localization of the proteins they represent. Note that the overall appearance of the phase images differ because many antibodies only worked on cultures fixed with methanol, which produces a rounding of the parasites within the vacuole (ROP5, ROP11, ROP12, and RON1-4), whereas the remainder were fixed with the more conventional formaldehyde. A, rhoptry proteins detected in the rhoptry bodies as shown by ROP1 or ROP2 co-localization. Staining for ROP15 appears to extend into the rhoptry necks at the apical end of each rhoptry (short arrow). Staining for ROP16 was with fluorescein-conjugated anti-HA antibody and a parasite line expressing a C-terminal HA-tagged version of ROP16. Co-localization was performed using 594-conjugated anti-ROP1 antibodies. B, rhoptry neck (RON) staining occurs at the apical end of each rhoptry identified by ROP2 staining. RON4 is also detected in between the parasites in the PV. C, immunoelectron microscopy with anti-RON4 antibodies demonstrates that RON4 is localized to the neck portion (arrow) of the rhoptries (R) and is not present in the bulbous bodies of the organelle. Lines also point to the conoid (C) and the micronemes (M). D, schematic of confirmed novel ROPs and RONs. The positions of predicted signal peptides, hydrophobic domains (TMpred score Ͼ800), and GPI anchor addition sequences are shown. Also shown are tandem repeats present in RON1 and a 44-amino acid repeat in RON4.

Tandem repeats identified in rhoptry fraction proteins
The consensus sequence, length and number of tandem repeats are shown in the table. The repeats in RON1 and TgTwinscan_7264 contains "DD" or "EE" as shown previously in tandem repeats in both rhoptry and non-rhoptry protein from Plasmodium as well as other parasites. (49 -51). By using antibodies produced against recombinant Rab11 protein, we verified the rhoptry localization for Toxoplasma Rab11 as assessed by ROP2 co-localization (Fig. 5B). Although the majority of the staining was found to co-localize to the rhoptries, small amounts of staining were reproducibly detected in more posterior, punctate spots within the cytoplasm of the parasite. We cannot determine whether this signal is because of a small amount of Rab11 in this location or because of a cross-reacting species, perhaps another Rab protein. In common with Rab proteins that have been studied in other systems, Toxoplasma Rab11 is likely to associate with the cytoplasmic face of the rhoptry membrane by a C-terminal lipid modification (52,53). The sequence at the C terminus of Toxoplasma Rab11 (CCSA) agrees perfectly with the CCXX motif that indicates membrane association mediated by geranyl-geranyl modification of the two cysteines.

cholesterol-rich rhoptries in Toxoplasma
We also detected some of the highly abundant proteins from the dense granules (NTPases, GRA3, and GRA7) and the surface antigen SAG1 in our rhoptry fraction (TABLE TWO). This is consistent with the fact that these proteins are among the most abundant of all Toxoplasma proteins (e.g. it has been reported that NTPase represents ϳ8% (54) and SAG1 (p30) ϳ5% of total parasite protein (55)). Obtaining large scale preparations that are completely devoid of such abundant proteins is not feasible, given the extreme sensitivity of the MS analysis used here. It should also be noted that the trafficking of highly abundant proteins through the secretory pathway is unlikely to be 100% accurate, and so it is possible that small amounts of SAG1 and/or GRA proteins may leak into the rhoptries and/or that the contamination may originate from intermediate compartments in the secretory pathway through which these proteins transit on their way to their final destination. We also detected the known proteins tgd057 and glucose-6-phosphate dehydrogenase. The tgd057 sequence contains a predicted signal peptide, but no localization data have been reported. The glucose-6-phosphate dehydrogenase has homology to plastidic versions of the protein, and thus we suspect that it is the result of a contamination of the rhoptry preparation with some amount of apicoplast material. This result may indicate that the anti-peptide results for TgTwinscan_2001 are correct in suggesting a plastid location for this protein (Fig. 4). The fact that other abundant proteins from these compartments were not detected in our MS analysis and 11 of the 12 novel proteins that we examined were confirmed as deriving from rhoptries strongly indicate that our rhoptry preparation was extremely enriched for these organelles.
Rhoptries are part of the defining feature, the so-called apical com-plex, in the phylum Apicomplexa (10). To address whether the proteins identified above are common to other Apicomplexan parasites, we took advantage of the near-complete genome sequence of Plasmodium falciparum and performed a bioinformatics analysis to search for close homologues of our novel Toxoplasma proteins. Although Plasmodium is a member of the Apicomplexa, it is only distantly related to Toxoplasma; therefore, proteins conserved between the two genera are likely to be found throughout the phylum (3,56). Nine of the newly identified rhoptry proteins in Toxoplasma were found to have homologues in Plasmodium and are not present in non-Apicomplexan organisms (ROP15, RON1, RON2, RON3, RON4, TgTwinscan_0556, TgTwin-scan_2579, TgTwinscan_4705, and Tgg_994550) (TABLE TWO).

DISCUSSION
By using MS and a new method for obtaining highly purified rhoptries, we have identified a large number of novel proteins that appear to be localized to these unusual organelles. Recently, others have completed a proteomic analysis of the excreted/secreted (ESA) material released by extracellular Toxoplasma tachyzoites stimulated with low concentrations of ethanol. 6 This fraction is believed to consist of proteins from both the dense granules and micronemes. Only 1 of the 38 novel proteins identified in our rhoptry fraction was also found in the excreted/secreted fraction of Zhou et al. (78) (TgTwinscan_7264). This may suggest that this protein is in fact a dense granule protein, although rhoptry localization cannot be excluded because the rhoptry protein ROP9 was also identified in the ESA material. Further experiments using antibody verification or epitope/GFP tagging will ultimately resolve the true localization for this protein. The extremely small overlap between the subproteome preparations of rhoptries and dense granules/micronemes strongly argues for the respective purity of both fractions.
Currently, there is no known signature sequence that can be used to predict which proteins are targeted to the rhoptries, although the data presented here should provide an excellent starting point for identifying such motifs. Others have reported that the ROP2/4/8 family are type I transmembrane proteins with a YXX motif (tyrosine followed three amino acids downstream by a hydrophobic residue) near the C terminus that is involved in rhoptry targeting (57). Of the 11 novel proteins we have verified, 3 (ROP5, ROP11, and ROP16) appear to be superficially similar in overall structure to ROP2/4/8 family proteins. All three have homology to protein kinases, but only two, ROP11 and ROP16, have a YXX motif near the C terminus. One possible explanation for the lack of such a sequence in ROP5 could be that this protein traffics to the rhoptries by interaction with other family members that contain the necessary targeting information. Similar "escorting" of proteins has been seen previously (58,59) in targeting of microneme proteins in Toxoplasma. Alternatively, the true structure and topology of this, and perhaps other members of this family, could be very different from that proposed for ROP2 with entirely distinct trafficking mechanisms. Having a more complete set of such proteins will greatly facilitate definitive identification of the trafficking signals. Likewise, having a set of four rhoptry neck proteins that have very distinct localizations from the classical rhoptry bulb proteins will allow even further refinement of the suborganellar trafficking of rhoptry proteins.
The discovery of toxofilin and Rab11 in the rhoptry fraction by MS led us to verify rhoptry localization for these proteins by antibody production and IFA analysis. For toxofilin, localization to the rhoptries explains the presence of a predicted N-terminal signal peptide in the sequence (48) but complicates the story as to how this protein may interact with the two Toxoplasma proteins to which it has been shown to bind, actin and a protein phosphatase 2C (47,48). There is no pre-FIGURE 6. Toxofilin and Rab11 localize to the rhoptries. Localization of toxofilin and Rab11 using antibodies against His 6 -tagged proteins. Top, antibodies against recombinant toxofilin stain the rhoptries. Rhoptry co-localization using anti-ROP1 antibodies is shown. Toxofilin staining extends from the rhoptry bodies into the rhoptry necks (arrowhead). Bottom, antibodies raised against recombinant Rab11 largely stains the rhoptries as shown by ROP2 co-localization. Additional staining in the posterior portion of the parasite is also reproducibly detected, which does not co-localize with the rhoptries. dicted transmembrane domain in the toxofilin sequence, so it should be located wholly within the luminal domain of rhoptries; actin and the protein phosphatase 2C, however, both lack signal peptides and thus are unlikely to be inside the rhoptries, strongly arguing against their possible interaction with toxofilin in vivo. One intriguing possibility is that toxofilin is secreted into the cytoplasm of the host (either before the PVM is formed, as with the so-called evacuoles (60), or later by traversing the PVM by an unknown mechanism) where it interacts with host actin and/or a protein phosphatase 2C. By using the antibody generated here, we see no evidence for secretion into the host, although detection may be problematic if small amounts of protein are distributed throughout the host cytosol. Determining the precise function of toxofilin will undoubtedly be aided by this more clearly defined localization to the rhoptries within the parasite.
Rab11 is a member of a family of small GTPases involved in the regulation of vesicular trafficking in eukaryotic cells (61). In mammalian cells, Rab11 is involved in trafficking of recycling endosomes, regulating exocytosis and cholesterol homeostasis (49,62,63). These features are common to the cholesterol-rich rhoptries of Toxoplasma, which are believed to be formed via the convergence of the secretory and endocytic pathways and thought to be related to secretory exosomes or lysosomes (49,51,64). The identification of Rab11 in our proteomic analysis and subsequent verification of Rab11 localization to the rhoptries indicate that Toxoplasma Rab11 is a regulator of trafficking of protein and/or lipids to the rhoptries. These results may also provide insight into the origin of these unique secretory organelles.
One common property of many rhoptry proteins described in this and previous studies is the presence of short, tandemly repeated peptide sequences (16,42,45,65,66). The repeats in Toxoplasma vary in length (8 -18 amino acids) and number (6 -13), but all contain a high proportion of acidic residues (often acidic dipeptides) and proline, lysine, and glutamine. Similar tandem repeats containing acidic dipeptides have been identified in Leishmania (67) and in secreted proteins (both rhoptry and non-rhoptry) in Plasmodium (42,45,65,66) where they have been speculated to be involved in protein-protein interactions during invasion and/or interfacing with the immune system of the host, although these hypotheses have yet to be directly tested. The presence of similar repeats in Plasmodium proteins that are not rhoptry-derived (e.g. the dense granule protein RESA (65,68)) indicates that it is likely that tandem repeats will be found in proteins from other compartments in Toxoplasma and cannot, in itself, be used as indicators of rhoptry localization.
Based on whether they have homologues in other genera, the proteins that have been identified here fall into several classes. First, there is the group that has no homologue currently identified in any other organism outside of genera that are extremely closely related to Toxoplasma (i.e. outside the family of Sarcocystidae that includes Neospora caninum and Sarcocystis spp. but notably does not include Eimeria). These proteins may have drifted to an evolutionary point where their homologues are no longer recognizable in other families, or they could be truly restricted to the Sarcocystidae. The evolutionary pressure for the emergence of these proteins may have derived from some specialized property of these organisms, such as their ability to form tissue cysts and their broad host range in mammals. Tissue cysts are important developmental stages for these organisms and allow them to be passed among intermediate hosts through predation, scavenging, or carnivorism (69).
Among the 38 novel Toxoplasma rhoptry constituents identified in this study, we failed to detect clear homologues for many known rhoptry proteins of Plasmodium, including RhopH2-3, RAP1-3, PF148, and RAMA1 (42,45,70,74). Consistent with this, homologues for these genes are not apparent in the near-complete Toxoplasma genome, and so they are likely to be truly restricted to Plasmodium and its close relatives. Similarly, the majority of verified rhoptry proteins in Toxoplasma (ROP1, ROP2 family proteins, ROP9, toxofilin, ROP12, ROP13, ROP15, and ROP16) do not have apparent homologues in Plasmodium spp. The fact that the Toxoplasma tachyzoite and Plasmodium merozoite rhoptry proteins are so unique with respect to each other argues for an exquisite adaptation of these organelles to their respective target cells (i.e. nucleated versus non-nucleated).
The second class of proteins revealed by this study are those that are conserved throughout the Apicomplexa, including Plasmodium spp., but not found in organisms outside this phylum. We report here nine such proteins, five of which have been verified as being derived from the rhoptries by antibody co-localization studies (ROP14, RON1, RON2, RON3, and RON4). These proteins are likely to be involved in rhoptry functions common to all Apicomplexans such as host cell invasion and/or creation of the PVM. It is intriguing that all four verified Toxoplasma rhoptry neck (RON) proteins have homologues in Plasmodium, indicating that rhoptry neck proteins are more highly conserved than other rhoptry proteins. In addition to rhoptry release, the neck portion of the organelle has been suggested to serve as a duct for the release of microneme proteins, and thus RON proteins may facilitate fusion of the compartments for secretion (11). Related paralogues for RON2 (TgTwinscan_3112, TgTwinscan_0430), RON3 (TgTwinscan_3931), and RON4 (TgTwinscan_2928) are detected within the Toxoplasma genome. None of these paralogues was identified in the proteomic data presented here, and thus their localization and function remain unknown. Most interestingly, RON2 and its paralogues are similar to the CLAG/RhopH1 family of proteins, which are also localized to the rhoptries of Plasmodium (71,72).
Recently, Sam-Yellowe et al. (75) have reported a proteomic analysis of a rhoptry-enriched fraction from Plasmodium merozoites. Thirty six candidate Plasmodium rhoptry proteins were identified, but not confirmed, by using a differential analysis of the rhoptry-enriched fraction compared with whole parasite lysates. Homologues for none of the novel proteins identified here, including homologues of ROP14 and RON1-4 that were verified as rhoptry proteins in our analyses, were identified as candidate rhoptry proteins in the rhoptry-enriched fraction of Plasmodium. However, several of the novel proteins identified in that study did share features with some of the proteins identified here, i.e. the presence of repeats (TABLE THREE) and weak homology to dentin proteins (RON1, Tg_Twinscan_6384). The homology to dentin in proteins in the rhoptry fraction is restricted to a large number of acidic residues present in these proteins and thus is of unclear significance. The absence in the Sam-Yellowe et al. study (75) of proteins corresponding to the Plasmodium homologues we report in TABLE TWO may reflect technical differences in our two approaches and/or different sensitivities or abundances, especially because neither study claims to have uncovered all the rhoptry contents. Alternatively, the Plasmodium homologues for the rhoptry proteins we identified may be expressed at a different stage of the Plasmodium life cycle (e.g. in sporozoites, for invasion of hepatocytes). Least likely is the possibility that the homologues localize to a different subcellular compartment in these two genera, but we cannot yet formally exclude that possibility.
The PVM is modified by parasite rhoptry proteins during invasion. Because many of the rhoptry proteins appear to be membrane-associated as indicated by the carbonate extraction procedure (Fig. 2), they may be secreted to the PVM and serve in the formation of the interface between the host and parasite. The PVM is expected to have several proteins that are able to span it in order to provide channels for exchange with the host cytosol. These channels have been demonstrated experimentally and shown to be permeable to molecules up to about 1200 daltons (76), but their composition has not been determined. Among the candidates for such a function is ROP14, which appears to be a multipass transmembrane protein, a common feature in membrane transporters.
The third broad class of proteins that we have observed here are those encoding functions that are common to most eukaryotic cells such as kinases, phosphatases, and proteases. These proteins are likely to have specialized functions in the rhoptries, in the parasitophorous vacuole, or on the PVM. Both ROP4 and toxofilin have been shown to be phosphoproteins and represent potential substrates for rhoptry-localized kinases and phosphatases (44,47). Molestina and Sinai (77) have shown recently that an unidentified parasite kinase is responsible for the phosphorylation of host IKB␣ localized at the PVM in infected cells. The putative kinases identified here represent good candidates for the parasite kinase activity described; of the 38 novel proteins identified, 12 have homology to protein kinases. These proteins typically have varying homology to the ROP2 family proteins and the same general structure, whereby the putative kinase domains could be exposed to the host cytoplasm following secretion and targeting to the PVM. The kinase domains of some ROP2 family members have been shown to be missing a conserved aspartic acid residue in the catalytic domain, which may suggest that only a subgroup of these proteins is the actual kinase (44). Although the key aspartate is present in some of the ROPs identified here, the conclusion that these are functional kinases will require direct experimentation. The presence of a predicted protein phosphatase 2C (TgTwinscan_7301) in the rhoptry fraction further indicates that phosphorylation will likely play an important role in rhoptry function, either within the rhoptries themselves or upon release into the host cell or nascent PV during invasion. The two proteases identified in the rhoptry fraction have homology to serine and metalloproteases (insulinase-like). As proteases, these proteins have potential therapeutic value, particularly given that serine protease inhibitors have been shown to block invasion of host cells by Toxoplasma (35). Ultimately, determining the role of each of these proteins will require the generation of mutant parasites that are deficient in the relevant gene, identification of the proteins with which they associate, and in cases where they are predicted to have enzymatic activity, the identity of their substrates. The work reported here, however, provides many important targets for such analyses.