Isolation and Characterization of TgVP1, a Type I Vacuolar H (cid:1) -translocating Pyrophosphatase from Toxoplasma gondii THE DYNAMICS OF ITS SUBCELLULAR LOCALIZATION AND THE CELLULAR EFFECTS OF A DIPHOSPHONATE INHIBITOR*

Here we report the isolation and characterization of a type I vacuolar-type H (cid:1) -pyrophosphatase (V-PPase), TgVP1, from an apicomplexan, Toxoplasma gondii , a parasitic protist that is particularly amenable to molecular and genetic manipulation. The 816-amino acid TgVP1 polypeptide is 50% sequence-identical (65% similar) to the prototypical type I V-PPase from Arabidopsis thaliana , AVP1, and contains all the sequence motifs characteristic of this pump category. Unlike AVP1 and other known type I enzymes, however, TgVP1 contains a 74-residue N-terminal extension encompassing a 42-res-idue N-terminal signal peptide sequence, sufficient for targeting proteins to the secretory pathway of T. gondii . Providing that the coding sequence for the entire N-terminal extension is omitted from the plasmid, trans-formation of Saccharomyces cerevisiae with plasmid-borne TgVP1 yields a stable and functional translation product that is competent in aminomethylenediphos-phonate (AMDP)-inhibitable K (cid:1) -activated pyrophosphate (PP i ) hydrolysis and PP i -energized H (cid:1) transloca- tion. Immunofluorescence

Vacuolar-type H ϩ -translocating inorganic pyrophosphatases (V-PPases) 1 are primary electrogenic proton pumps that derive their energy from the hydrolysis of inorganic pyrophosphate (PP i ) (1). Long considered to be restricted to plants and certain photosynthetic bacteria, V-PPases have recently been identified in a wide range of organisms, including prokaryotic extremophiles and the kinetoplastid protists Trypanosoma and Leishmania, the causative agents of Chagas' disease and leishmaniasis, and the apicomplexan protists Plasmodium and Toxoplasma, the causative agents of malaria and toxoplasmosis (reviewed in Ref. 2). The discovery of V-PPases in these parasitic protists has attracted much attention. The seemingly complete absence of V-PPases from their animal hosts has given rise to the exciting possibility that this enzyme might serve as an effective drug target for a number of the diseases caused by these pathogens.
All characterized V-PPases are constituted of a single 75-82-kDa intrinsic membrane protein species that is now known to fall into two structurally and functionally distinct types, I and II. Type I V-PPases, as exemplified by the molecular prototype, AVP1 from Arabidopsis (3), exhibit a near obligate requirement for millimolar K ϩ for activity. Type II V-PPases, as exemplified by Arabidopsis AVP2 (4), by contrast, share only ϳ36% sequence identity with their type I counterparts and are insensitive to K ϩ .
A property of all V-PPases characterized to date, regardless of whether they are type I or type II enzymes, which distinguishes them from soluble PPases, is their high sensitivity to competitive inhibition by the 1,1-diphosphonate, aminomethylenediphosphonate (AMDP), and their relative insensitivity to irreversible inhibition by fluoride (5,6). Originally identified in screens for PP i analogs capable of functionally distinguishing V-PPases from soluble PPases and other phosphohydrolases, diphosphonates such as AMDP that contain a heteroatom (NH 2 or OH) on the bridge carbon are exquisitely potent V-PPase inhibitors (5). Taking their cue from the finding that protozoal V-PPases are as sensitive to inhibition by AMDP as their plant and photosynthetic bacterial counterparts, several investigators have shown that this compound inhibits the growth of Plasmodium falciparum (7), Toxoplasma gondii (8), and Trypanosoma cruzi (9). Although in all cases the apparent efficacy of AMDP in vivo is markedly lower than its efficacy in vitro, the potential for the development of derivatives of this compound or alternative V-PPase-specific agents for drug purposes is nevertheless evident.
Those plant V-PPases that have been characterized in detail are found primarily in vacuolar and Golgi membranes, where their activity contributes to the transmembrane H ϩ gradient that drives H ϩ -and/or electrically coupled secondary transport processes (1). By analogy, parallel biochemical and immunological investigations of the V-PPases of trypanosomatid and apicomplexan protists indicate that they are most closely associated with a vacuole-like organelle, the acidocalcisome (8, 10 -13). Acidocalcisomes are small, electron-dense vacuolysosomelike acidic compartments, replete with polyphosphates complexed with Ca 2ϩ , Mg 2ϩ , and other mineral ions, that are suspected to play a dominant role in Ca 2ϩ storage and signaling (reviewed in Ref. 14). The recent demonstration of V-PPases in the membranes bounding the contractile vacuoles of Chlamydomonas (15) and Dictyostelium (16) and the remarkable equivalence of the composition of these organelles with acidocalcisomes may also be pertinent to these considerations. Having made this point, it should be stressed that the association of V-PPases with acidocalcisome-like membranes does not preclude their association with other membranes. As indicated by the results of a number of studies of trypanosomatids and apicomplexans, V-PPase-like immunoreactivity is also to be found in the plasma membrane (7,8,12,13,17) and Golgi system (17), at least under some circumstances.
Several publications have described the preliminary in vitro biochemical characterization of PPase activities associated with membranes prepared from trypanosomatid and apicomplexan protists (reviewed in Ref. 14). However, the molecular basis for these activities has been less well defined. Only a single type I V-PPase gene has been isolated from T. cruzi (18), and in the apicomplexa, although genes for both type I and type II enzymes have been cloned from P. falciparum, PfVP1 and PfVP2, respectively, all attempts to elucidate the functional properties of these gene products by heterologous expression have been unsuccessful (7).
In the following we report the isolation and characterization of a type I V-PPase, TgVP1, from the apicomplexan protist, T. gondii, and rigorous analyses of the subcellular localization of the enzyme and of the effects of the V-PPase inhibitor AMDP on parasite morphology both during and after host cell invasion. We demonstrate that, upon heterologous expression in yeast, TgVP1 encodes an intrinsic membrane protein competent in PP i -dependent H ϩ transport that is unique among V-PPases in containing an N-terminal signal sequence sufficient for targeting proteins to the secretory pathway in T. gondii. Furthermore, using affinity-purified V-PPase-specific antibodies, we demonstrate a dynamic pattern of distribution of the V-PPase in invading parasites. Under most conditions, immunofluorescence microscopy of the V-PPase reveals a punctate apical distribution. However, during invasion of the host cell, this immunofluorescence undergoes a dramatic redistribution to assume a collar-like structure at the periphery of the parasite that migrates in synchrony with the penetration furrow as the parasite enters the host cell. Given this association of the V-PPase with the invasion apparatus, it is perhaps surprising that application of even high doses of AMDP to invading parasites has no significant effect on their establishment in the host cell, despite the facility with which lower doses impair intracellular parasite division. These results demonstrate the effects of AMDP at doses much lower than those reported previously and suggest a function for acidocalcisomes in host cell invasion in T. gondii.

MATERIALS AND METHODS
Host Cells and Parasites-T. gondii RH strain tachyzoites were maintained by serial passage in primary human foreskin fibroblast (HFF) cultures grown at 5% CO 2 in bicarbonate-buffered Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% heatinactivated newborn bovine serum (HyClone) and an antibiotic mixture of penicillin, streptomycin, and gentamycin as described (19). The culture medium was replaced with modified Eagle's medium containing 1% dialyzed fetal bovine serum (Invitrogen) immediately before infecting the fibroblasts with parasites.
Cloning Reagents-The PCR primers used for cloning, plasmid construction, RT-PCR, and 5Ј-RACE are listed in Table I. The T. gondii cDNA pools were prepared from and the 5Ј-RACE reactions were performed on total RNA and poly(A) mRNA purified from freshly lysed tachyzoites using a SMART RACE Kit (Clontech). DNA sequencing was by dye terminator chemistry using nested oligonucleotide primers.
Isolation of Genomic and cDNA Clones of TgVP1-The genomic and cDNA clones of TgVP1 were isolated by a combination of PCR and standard oligonucleotide hybridization screens of genomic and cDNA libraries. Initially, degenerate primers corresponding to the conserved "universal" V-PPase sequences DNAGGIAE and WDNAKKYI (primers "Universal1" and "Universal2" in Table I) were used in PCRs in which RH genomic DNA or tachyzoite cDNA pools were used as templates. The resulting unique 1028-bp genomic and 593-bp cDNA PCR products were cloned into pGEM-T vector (Promega), sequenced, and used as probes for subsequent parallel screens of an RH genomic library constructed in DASH-II (Stratagene, La Jolla, CA) and a tachyzoite cDNA library (from the AIDS Reference and Reagent Repository, National Institutes of Health, Bethesda).
The longest cDNA clone isolated in the hybridization screens was 1989 bp in length and, by comparison with the published sequences of V-PPases from other sources, appeared to lack the coding sequence for the first ϳ240 amino acid residues of the mature polypeptide. The largest of three overlapping clones from the hybridization screens of the genomic library yielded two restriction fragments after EcoRI-HindIII digestion, one of 6.2 and another of 3.4 kb, that hybridized at high stringency in Southern analyses with the 1989-bp partial cDNA sequence. These restriction fragments were cloned into the EcoRI-HindIII sites of Bluescript plasmid pKS ϩ (Stratagene) and sequenced. In order to deduce the translation start site of TgVP1 from its genomic sequence, several 5Ј-RACE and nested RT-PCRs were performed using the tachyzoite cDNA pools as template, and the primer combinations listed in Table I.
Plasmid Construction-The full-length TgVP1 cDNA (GenBank TM accession number AF320281) was cloned by RT-PCR of T. gondii tachyzoite cDNA pools using Pfu DNA polymerase (Stratagene) and sense and antisense primers TgVP.1 and TgVP.3ЈY (Table I) corresponding to positions 1-20 and 2430 -2451, respectively, of the ORF predicted from the TgVP1 genomic sequence (GenBank TM accession number AF320282). After digestion with XbaI the resulting 2451-bp PCR product was ligated into the multicloning site of PvuII-XbaI double-digested yeast expression vector pYES2 (Invitrogen) to generate pYMD23. Yeast expression vectors encoding truncated versions of TpVP1 were generated by PCR amplification of TgVP1 from pYMD23 using Pfu DNA polymerase and sense primers corresponding to the sequence immediately downstream of the predicted signal peptide (residues 121-138 of the TgVP1 cDNA; primer TgVP1.2, Table I) to generate pYMD24, and with sense primers corresponding to the sequence further downstream of this (residues 223-240; primer TgVP1.3, Table I) to generate pYMD25.
The T. gondii fluorescent protein and overexpression plasmids used in this study were constructed in ptubP30-YFP/sag-chloramphenicol acetyltransferase vector (20), in which a BglII site separates the 5Јuntranslated region of ␣-tubulin (21) from the P30 signal sequence (22), and an AvrII site separates the P30 sequence from the yellow fluorescent protein-coding sequence (YFP, a derivative of Aequoria victoria green fluorescent protein, GFP). TgVP1 coding sequences were cloned as BamHI-XbaI fragments into BglII (BamHI-compatible)-AvrII (XbaIcompatible) double-digested vector. For the construction of plasmids pYMD26-27, which contained coding sequences corresponding to the first 84 and 232 amino acid residues of TgVP1 in-frame with the YFP sequence, the TgVP1 coding sequences were amplified by PCR from YMD23 using the TgVP5ЈT sense primer in combination with the Nterm.1 and N-term.2 antisense primers (Table I). Plasmid pYMD28 containing the full-length TgVP1 cDNA fused to YFP was similarly generated by PCR using primers TgVP5ЈT and TgVP3ЈT (Table I). Plasmids pYMD31 and pYMD32, encoding D550N-substituted TgVP1, were constructed from T. gondii and yeast expression plasmids pYMD28 and pYMD25, respectively, using the QuickChange Site-directed Mutagenesis Kit (Stratagene) and the mutagenic primer "D550N" (Table I).
T. gondii expression plasmids pYMD29-30 and pYMD33-34 were engineered to contain a stop codon upstream of YFP/GFP in the expression vector. Plasmids pYMD29 and -33 were generated by PCR from pYMD23 and pYMD31, respectively, using primers TgVP5ЈT and TgVP.3ЈY (Table I). Plasmids pYMD30 and -34 were generated by ligation of the BamHI-XbaI fragments from pYMD25 and pYMD32, respectively, to BglII-AvrII double-digested vector.
Antibody Purification-Polyclonal V-PPase antisera PAB TK (324), PAB HK (326), and their respective preimmune sera were affinity-purified against the AVP1 polypeptide of vacuolar membrane-enriched vesicles purified from pYES2-AVP1-transformed yeast BJ5459 cells. Membrane protein (100 g) was separated and electroblotted as described below, and the band corresponding to the PAB TK -reactive 81-kDa AVP1 polypeptide was cut from the blot and blocked overnight in PBS ϩ 3% BSA. After a brief wash in PBS containing 0.1% BSA, the excised strips of nitrocellulose filter were placed in a 1:4 dilution of serum in PBS ϩ 3% BSA and incubated for 16 h at 4°C with shaking. Upon completion of the incubations, the blots were washed three times with TBS ϩ 0.1% BSA, twice with TBS ϩ 0.1% BSA ϩ 0.1% Nonidet P-40, and 3 more times with TBS ϩ 0.1% BSA, before elution of the antibodies into 150 l of 0.2 M glycine HCl buffer (pH 2.5) and immediate neutralization of the eluate with 75 l ice-cold 1 M K 4 PO 4 (pH 9.0) containing 3% BSA. Elution and neutralization were repeated three more times, and the eluates were pooled and desalted with PBS in a Centricon-30 filtration unit (Millipore Corp.). The final antibody preparations were stored in PBS containing 0.05% (w/v) NaN 3 at 4°C.
Measurement of PP i Hydrolysis and H ϩ Translocation-PP i hydrolytic activity was assayed as described (24) except that imidazole-based, rather than Tris-or BisTris-propane-based, buffers were used throughout to preclude competition with K ϩ and other monovalent cations (4). PPase activities are expressed as mol of PP i hydrolyzed/mg of membrane protein/min (ϭ mol/mg/min). PP i -dependent intravesicular acidification was monitored fluorimetrically using acridine orange (2.5 M) as indicator as described (4).
Western Analyses-For Western analysis, vacuolar membrane-enriched vesicles purified from the yeast transformants or different membrane fractions from T. gondii were subjected to denaturation, SDS-PAGE on 10% (w/v) acrylamide gels, electrotransfer, and immunoreaction with antibodies PAB HK or PAB TK , as described (24). Immunoreactive bands were visualized by ECL (Amersham Biosciences).
Protein Assays-Protein was estimated by the method of Bradford (25).
Light Microscopy-For light microscopy, HFF cells were grown to confluence on sterilized coverslips in 6-well plates. The confluent cultures were infected with 5 ϫ 10 5 parasites and examined at the times indicated. For the visualization of native fluorescent proteins, the coverslips were mounted in PBS. For the immunofluorescence analyses, parasite-infected cells were fixed in 3% paraformaldehyde and permeabilized with 0.25% Triton X-100 in PBS. The purified anti-V-PPase sera were used at dilutions of 1:200, and immunoreaction was detected using FITC-conjugated anti-rabbit immunoglobulin (1:1000) (Molecular Probes, Inc.). The anti-MIC3 antibodies (26) in the double-immunolabeling experiments were used at a dilution of 1:1000 and detected using Texas Red-conjugated anti-mouse immunoglobulin (1:1000) (Molecular Probes, Inc.).
Electron Microscopy-For electron microscopy, infected cells were fixed in situ with a freshly prepared mixture containing 1% glutaraldehyde (made from an 8% stock; Electron Microscopy Sciences, Fort Washington, PA) and 1% osmium tetroxide in 50 mM phosphate buffer (pH 6.2). After adding the fixative at room temperature, the specimens were incubated at 4°C for 45 min. The samples were rinsed with distilled water to remove excess phosphate before being gently scraped off the Petri dishes with a beveled scraper. Staining was with 0.5% aqueous uranyl acetate for 6 -16 h at 4°C. The samples were dehydrated with acetone and embedded in an Epon-Araldite resin mixture. Ultrathin (50 -70-nm thick) sections were cut and stained with uranyl acetate and lead citrate and examined using a Philips 200 electron microscope.
Assays of Parasite AMDP Sensitivity-The effects of AMDP on parasite growth were examined in HFF cells grown on coverslips and AAACCCggatccATGGCAACACTCCGTATCGAT

Isolation and Characterization of TgVP1
infected with tachyzoites at a concentration of 10 5 parasites per ml. The parasites were allowed to invade for 5 min at 37°C before aspirating the medium and replacing it with fresh medium to remove all free parasites. AMDP was added to the medium at the concentrations indicated after allowing 20 -30 min at 37°C for the parasites to become established in the host cells. The cultures were incubated in a humidified atmosphere containing 5% CO 2 for 24 h at 37°C, after which time the infected cells were fixed on coverslips in methanol at Ϫ20°C. The infected cells remaining in the dishes were processed for electron microscopy as described above. Parasite replication was assessed by counting the numbers of parasites per parasitophorous vacuole by direct visualization using a Zeiss microscope with phase objectives. To ensure random counting, fields from all regions of the coverslip were counted without prior microscopic examination, and all vacuoles within each field were counted. In all experiments the numbers of parasites per vacuole were determined for between 500 and 700 vacuoles from at least five separate experiments. To examine the effects of AMDP on host cell invasion, freshly emerged tachyzoites were incubated with AMDP for 5-10 min prior to infection. Infected cells were incubated and counted as described above.
Computer Programs-For measurements of the susceptibility of the V-PPase to inhibition by Ca 2ϩ , the concentration of free Ca 2ϩ ([Ca 2ϩ ] free ) was estimated by substitution of the appropriate stability constants into the SOLCON program (a kind gift from Dr. Yale Goldman, Department of Physiology, University of Pennsylvania). The stability constants were obtained from Martell and Smith (27) and Smith and Martell (28) and deployed as described (29). Sequences were aligned using ClustalW 1.7 (30). The putative membrane topology of TgVP1 was modeled using TopPred II, version 1.3 (31), as described for AVP1 (24), PVP (32), and AVP2 (4).

RESULTS
Isolation and Sequence Characteristics of TgVP1-To screen for genes encoding V-PPases in T. gondii, degenerate primers corresponding to the sequences, DNAGGIAE and WDNAKKYI (positions 511-518 and 694 -701 in AVP1) (32), conserved among all known V-PPases (2), were used as primers for PCR amplification of strain RH genomic DNA and tachyzoite cDNA pools. In so doing a unique 1028-bp genomic product and unique 593-bp cDNA product were isolated, each of which was cloned, sequenced, and determined to be capable of encoding a V-PPase. Both isolates were used as probes for hybridization screens of the RH genomic and tachyzoite cDNA libraries. Because the cDNA library screens yielded only truncated and incompletely spliced cDNAs but the screens of the RH genomic library yielded three distinct but overlapping clones, the complete sequence of this gene (GenBank TM accession number AF320281) was derived ultimately from detailed analyses of a 6985-bp portion of the longest of the three genomic clones isolated (Fig. 1).
Given that successive rounds of 5Ј-RACE and RT-PCR consistently failed to indicate an in-frame initiator codon 5Ј of the ATG at position 1350 of the genomic sequence (a site encompassed by a sequence similar to the consensus Kozak translation initiation sequence (TCCRTCATGG) (33)), this site was inferred to serve as the start site for the expression of this gene in tachyzoites (Fig. 1). Accordingly, RT-PCR of T. gondii total RNA and poly(A) mRNA using primers designed to span the entire predicted ORF yielded a 2451-bp product encoding an 816-amino acid (85 kDa) polypeptide (Fig. 1). Designated TgVP1, this deduced polypeptide is 50% sequence-identical (65% similar) to the prototypical type I V-PPase from AVP1 (3) and 57% identical (72% similar) to the type I V-PPase from P. falciparum (7) but only 37 and 33% identical (53 and 50% similar) to the type II enzymes from these organisms, respectively (4,7).
All of the sequence motifs and residues known to be characteristic of type I V-PPases are conserved in TgVP1 (Fig. 1). These include the following: (i) the putative "universal PPase catalytic" motif DX 7 KXE found in both soluble and membraneassociated PPases (34); (ii) Cys 677 , corresponding to Cys 634 of AVP1, whose substrate-protectable alkylation by maleimides irreversibly abolishes catalytic activity (35,36); (iii) residues Glu 345 and Asp 550 , corresponding to Glu 305 and Asp 504 in AVP1, which contribute to the susceptibility of the enzyme to inhibition by the hydrophobic carbodiimide N,NЈ-dicyclohexylcarbodiimide (24); and (iv) Glu 473 , corresponding to Glu 427 of AVP1, which has been inferred to participate directly in H ϩ translocation (24) but which is substituted by K in type II V-PPases (2,4). Of strategic significance is the conservation of the sequences TKAADVGADL(VS)GK(IN)E and HK(A-N)AV(IT)GDT(IV)GDPLK in TgVP1. Because these are the sequences recognized by peptide-specific antibodies PAB TK and PAB HK (23) deployed in all previous investigations of V-PPases in protists, confirmation of their conservation substantiates previous reports (8, 10) of V-PPase-associated immunoreactivity in T. gondii membranes.
The overall membrane topology predicted for TgVP1 using the TopPred II program (31) is similar to those predicted for AVP1 (24) and PfVP1 (7) with the exception of an additional putative N-terminal transmembrane span encompassing residues 14 -34 (Fig. 1). Sequence alignments between TgVP1, other type I V-PPases from higher plants, and PfVP1 indicate that the additional N-terminal transmembrane span of TgVP1 resides within a unique 74-residue N-terminal extension. Analyses of this N-terminal extension using SignalP (37) and PSORTII (38 -40) reveal a putative signal cleavage site between residues 39 and 40 (Fig. 1).
The results of Southern analyses (data not shown) indicate that the gene encoding TgVP1 exists as a single copy in the genome of T. gondii RH.
Heterologous Expression of TgVP1 in S. cerevisiae-To determine whether it encodes a V-PPase with properties similar to those of AVP1, TgVP1 was heterologously expressed in yeast. For this purpose, S. cerevisiae strain BJ5459 was transformed with vector pYES2 containing the entire coding sequence of TgVP1 (pYMD23) or truncated forms of TgVP1 lacking the coding sequences for either the 40-residue signal sequence (pYMD24) or the entire 74-residue N-terminal extension (pYMD25). For comparative purposes, yeast were also transformed with pYES2-AVP1 vector.
Expression of TgVP1, measured as the presence of a polypeptide of the appropriate size and immunoreactivity in the vacuolar membrane-enriched fraction from the yeast transformants, is contingent on deletion of the coding sequence for the entire 74-residue N-terminal extension. Vacuolar membraneenriched vesicles purified from pYMD25-transformed (BJ5459/ pYMD25) cells contain an intense PAB TK (and PAB HK )-reactive M r 75,000 band that reacts with antibody with similar intensity to the corresponding band in the equivalent membrane fraction from pYES-AVP1-transformed BJ5459 cells but which is absent from the corresponding fractions from pYMD23-and pYMD24-transformed BJ5459 cells (Fig. 2).
TgVP1-mediated PP i Hydrolysis and PP i -dependent H ϩ Translocation-The PAB TK -reactive polypeptide detected in the vacuolar membrane-enriched fraction from BJ5459/ pYMD25 cells catalyzes both PP i hydrolysis and PP i -dependent H ϩ -translocation. When assayed in reaction buffer containing 250 M NaF to abolish any contribution from contaminating yeast soluble PPase (23), the kinetics of TgVP1-catalyzed PP i hydrolysis are indistinguishable from those of AVP1. The K m values, PP i concentrations required for maximal activity, and maximal activities of TgVP1 and AVP1 are 34 and 38 M total PP i , 0.3 mM total PP i , and 0.5-0.6 and 0.6 -0.7 mol/mg/min, respectively, when PP i hydrolysis is assayed in reaction media containing 1.3 mM Mg 2ϩ (data not shown). As would be predicted for a type I V-PPase, the hydrolytic activity of heterologously expressed TgVP1, like AVP1, is K ϩ -activated (Fig. 3). TgVP1-mediated PP i hydrolysis is measurable in media containing choline chloride, albeit at a low level, but replacement of this salt with KCl or potassium gluconate increases activity by 2-3-fold. This behavior is similar to that seen with heterologously expressed AVP1 except that in the latter case K ϩ increases activity by about 8-fold versus choline (Fig. 3). A similar pattern is seen with PP i -dependent H ϩ -translocation except that the nature of the counter-anion is also important. Substitution of the permeant anion Cl Ϫ with the less permeant anion gluconate decreases the extent of both TgVP1-and AVP1-mediated intravesicular acidification by at least 60%, whereas substitution of K ϩ with choline decreases the extent of intravesicular acidification by at least 90% (Fig. 3). Previous investigations of both endogenous and heterologously expressed type I V-PPases have established that the permeant anion Cl Ϫ maximizes PP i -dependent intravesicular acidification by diminishing the magnitude and therefore the stalling action of the inside-positive membrane potential that would otherwise be generated by electrogenic H ϩ translocation (41).
Sensitivity of TgVP1 to AMDP and Ca 2ϩ -TgVP1 and AVP1 are similarly sensitive to inhibition by AMDP (5) but differentially sensitive to inhibition by Ca 2ϩ (Fig. 4). Although the concentration dependences for inhibition of TgVP1-and AVP1mediated PP i hydrolysis by AMDP superimpose to yield I 50 values of 0.9 and 3.0 M, respectively (Fig. 4A), TgVP1 is more than 8-fold more sensitive to inhibition by free Ca 2ϩ than AVP1 (Fig. 4B). A total concentration of 1.4 M, equivalent to a free Ca 2ϩ concentration of 0.15 M, is sufficient to inhibit TgVP1mediated PP i hydrolysis by 50%, whereas concentrations of greater than 70 and 1.2 M, respectively, are required to inhibit AVP1 to the same extent (Fig. 4B).
TgVP1 N-terminal Signal Sequence-To examine whether the putative signal sequence encompassed by the N-terminal extension of TgVP1 is capable of targeting peptides to the secretory pathway in T. gondii, YFP fusion plasmids containing the coding sequences for either the first N-terminal 84 amino acid residues (plasmid pYMD26), which encompass the entire N-terminal signal sequence and cleavage site, or the first 232 residues (pYMD27), which encompass the N terminus, cleavage site, and first predicted transmembrane span of the mature protein, or the entire coding region of TgVP1 inclusive of the N-terminal extension (pYMD28) were constructed and transfected into tachyzoites (Fig. 5).
In all tachyzoites transfected with pYMD26, YFP fluorescence is observed within dense granules and within the lumen of the parasitophorous vacuole (Fig. 5). In tachyzoites transfected with pYMD27, ϳ75% of the transfectants exhibit sequestration of the YFP fluorescence in the endoplasmic reticulum, whereas the remaining 25% exhibit a punctate fluorescence distribution indicative of the incorporation of YFP into inclusion bodies (Fig. 5). In tachyzoites transfected with pYMD28, the cells divide only rarely and undergo extensive inflation of the vacuole (Fig. 5).
To determine whether the vacuolate, non-dividing phenotype of pYMD28 transfectants might be attributable to the YFP fusion, two constructs, pYMD29 and pYMD30, that either contained or lacked the N-terminal extension but contained a stop codon between the TgVP1 and YFP coding sequences were engineered (Fig. 6). To determine whether the YMD28 transfectant phenotype might be because of PPase activity associated with the expression product, two of the overexpression constructs, pYMD33 and pYMD34, were engineered to contain a D550N substitution (Fig. 6). It has been established that the same substitution at the equivalent position, residue 504, in AVP1 abolishes catalytic activity whether it is measured as PP i hydrolysis or PP i -dependent H ϩ translocation (24). Analogous assays of heterologously expressed TgVP1 D550N (construct pYMD32) yield the same result (data not shown). Expression of the TgVP1 sequences in the tachyzoite transfectants was monitored in these experiments by indirect immunofluorescence microscopy using purified peptide-specific antibody PAB TK (Fig. 6).
Whereas removal of the YFP tag yields transfectants capable of division at higher frequency, the extents and rates of division of these cells are nevertheless markedly decreased. Moreover, whereas expression of both of the full-length constructs (pYMD29 and -33) results in intense immunostaining that appears to be membrane-associated (Fig. 6), expression of the constructs lacking the N-terminal extension (pYMD30 and -34) results in diffuse immunostaining throughout the cell (Fig. 6). Similar experiments were performed using T. gondii expression plasmid pDHFR-P30-GFP/sag-chloramphenicol acetyltransferase vector (42,48), which contains the dihydrofolate reductase promoter sequence instead of tubulin, but these yielded no observable expression. In no case, did expression of non-functional, D550N-substituted TgVP1 (constructs pYMD33 and pYMD34) abrogate the highly vacuolated, low division frequency phenotype of the transfectants (Fig. 6). Evidently the N-terminal extension of TgVP1 contains a signal sequence sufficient for directing transport to the secretory pathway of T. gondii, but overexpression of the coding sequence of TgVP1, regardless of whether or not the translation product is catalytically active, interferes with cell division and elicits increased vacuolation.
Subcellular Localization of Endogenous V-PPase-By having determined the disruptive effects of protein fusion and overexpression techniques in this context, the subcellular localization of TgVP1-related translation products was assessed by using the same purified peptide-specific antibody, PAB TK , as used in the experiments presented in Fig. 6. However, as is apparent from the immunofluorescence micrographs shown in Fig. 7, purification of this antibody before use is crucial if the results are to be intelligible. Immunofluorescence microscopy of both free and intracellular tachyzoites using crude preimmune and immune sera and FITC-conjugated secondary antibodies yields remarkably similar results, high intensity fluorescence throughout the tachyzoites (Fig. 7). In striking contrast, when both antisera are affinity-purified against heterologously expressed AVP1 and the specificities of the purification products verified by Western analyses of membranes isolated from S. cerevisiae transformed with either pYES2-AVP1 or empty pYES2 vector, the results are very different. Whereas microscopy using immunopurified preimmune serum discloses little or no immunofluorescence with either intracellular or free tachyzoites, incubation of the same preparations with immunopurified PAB TK clearly demonstrates a punctate anterior apical distribution of the antigen (Fig. 7).
Phase-dependent Changes in Subcellular Distribution-A striking property of the membrane system with which the V-PPase is associated in the intact parasite is the degree to which it is subject to dynamic phase-dependent changes in organization. Deployment of the same immunopurified antibodies as those used for free and intracellular tachyzoites for studies of trophozoites during host cell invasion reveals that FIG. 4. Sensitivities of TgVP1-and AVP1-mediated PP i hydrolysis to inhibition by AMDP (A) or Ca 2؉ (B). PP i hydrolysis was measured as described in Fig. 3 except that AMDP or Ca 2ϩ (CaCl 2 ) were added at the concentrations indicated. Free Ca 2ϩ ([Ca 2ϩ ] free ) was estimated using the SOLCON program as described (29). Values shown are means Ϯ S.E. (n ϭ 3).
FIG. 5. T. gondii tachyzoites expressing TgVP1-YFP protein fusions. T. gondii tachyzoites were transiently transfected with expression constructs containing the following: coding sequence for the first N-terminal 84 amino acid residues of TgVP1 (plasmid pYMD26), which encompass the entire N-terminal signal sequence and cleavage site; the first 232 residues, encompassing the N terminus, cleavage site and first predicted transmembrane span of the mature protein (plasmid pYMD27); or the entire TgVP1 ORF (plasmid pYMD28). In all cases the sequences encoded were C-terminally fused with YFP. The cells were examined under an Axiovert microscope equipped with a single emission filter and specific YFP filter ("Materials and Methods"). most of the V-PPase-specific staining assumes a transverse radial distribution soon after the parasite has made contact with the host cell. A collar-like structure is generated that migrates along the length of the parasite in synchrony with and immediately anterior to the apicobasally propagating penetration furrow (Fig. 8). Upon completion of infection, the V-PPaseassociated fluorescence disperses before reappearing again at the anterior apex of the intracellular tachyzoite (Figs. 8 and 9). Although this pattern is reminiscent of the dynamics of microneme protein redistribution during invasion (43), colocalization experiments on these and non-invading parasites using the microneme-specific antibody raised against the secreted microneme protein MIC3 (26) in parallel with reaction with the V-PPase-specific antibody, demonstrate segregation, sometimes diametric segregation, of the two antigens from each other (Fig. 9). MIC3 also propagates as a collar-like structure along the length of the trophozoite during infection, but its distribution is not coincident with that of the V-PPase immediately before and after host cell invasion (Fig. 9).
Effects of AMDP on Cell Invasion and Replication-Knowing that TgVP1 and AVP1, when heterologously expressed in yeast, are similarly sensitive to inhibition by AMDP (Fig. 4) and that reaction of V-PPase-specific antibody with trophozoites during host cell invasion is with a collar-like structure that appears to be associated with the penetration furrow (Fig. 8), we were interested to determine the efficacy of this 1,1-diphosphonate as an inhibitor of cell invasion and/or intracellular parasite replication.
The results of these screens were surprising in that AMDP impairs intracellular parasite replication but exerts little or no effect on host cell invasion. On the one hand, treatment of extracellular trophozoites with AMDP at concentrations as high as 100 M exerts no discernible effect on either the efficiency of host cell invasion or integration (data not shown). On the other hand, AMDP concentrations as low as 5 M are sufficient to interfere with intracellular parasite division. Treatment of infected cells with AMDP at concentrations of 5 M or greater decreases the number of parasites per vacuole after 24 h concomitant with the appearance of irregular parasite masses (Fig. 10). Examination of parasites treated with 10 M AMDP for 24 h indicates that in many cases daughter cell budding is stalled, leading to the appearance of large irregularly shaped parasites (Fig. 11A). No obvious swelling of the endoplasmic reticulum, nuclear envelope, or Golgi complex occurs, and no large vacuolar spaces are seen in the parasites. Indeed, these parasites contain a full array of normal secretory organelles indicating little or no perturbation of the secretory pathway. Furthermore, the parasite mitochondrion and apicoplast look similar to control parasites. However, in a small number of the parasites (ϳ5%), certain vesicular structures reminiscent of the acidocalcisomes described in trypanosomes (structures typified by the loss of most of their luminal electron density upon double-fixation (14)) lose most of their electron density and undergo swelling and disruption after exposure of the cells to AMDP (Fig. 11B).

DISCUSSION
The findings reported here constitute the first molecular and functional characterization of TgVP1, a type I V-PPase from the apicomplexan protist T. gondii. It is established that TgVP1 encodes a functional K ϩ -activated, PP i -dependent H ϩ pump that bears a close resemblance to the canonical type I V-PPases from plants. In addition, it is shown that the V-PPase FIG. 6. Immunofluorescence localization of TgVP1 in transiently transfected T. gondii tachyzoites. The tachyzoites were transiently transfected with expression constructs containing the TgVP1 cDNA either with (plasmids pYMD29 or -33) or without the coding sequence for the N-terminal extension (plasmids pYMD30 or -34) or containing the coding sequence for D550N-mutated TgVP1 (plasmids pYMD33 or -34). The D550N substitution yielded catalytically inactive enzyme. The transfected tachyzoites were fixed, permeabilized, and reacted with purified anti-V-PPase serum PAB TK before reaction with the FITC-linked secondary antibody and epifluorescence microscopy.

FIG. 7. Immunofluorescence detection of V-PPase in intracellular tachyzoites.
A, after reaction with crude preimmune and PAB TK sera. B, after reaction with purified preimmune and PAB TK sera. The infected monolayers were fixed, permeabilized, and reacted with crude or purified preimmune serum or V-PPase antiserum PAB TK before reaction with FITC-linked secondary antibody. has a punctate apical distribution in both free and intracellular tachyzoites at steady state but a dynamic distribution during host cell invasion. Upon initiation of infection of the host cell by T. gondii, a collar-like structure with which most of the immunodetectable V-PPase is associated is formed. This structure, or the V-PPase that is associated with it, moves from the anterior to the posterior of the parasite coincident with propagation of the penetration furrow.
Of the V-PPases that have been defined molecularly to date, TgVP1 is unusual in its possession of an N-terminal signal peptide. With the exception of the pump from T. cruzi which like TgVP1 does have one (18), a survey of all available V-PPase sequences from archaea, eubacteria, and plants reveals none, not even the type I V-PPase from P. falciparum (7), that contain a putative N-terminal signal peptide. The precise role of this sequence and why protists like T. gondii and T. cruzi have it and others not is not known, although in the case of TgVP1 it has been shown by the fusion experiments that the N-terminal extension is sufficient to direct polypeptides to the secretory pathway of T. gondii.
More sequences are needed to decide this issue, but these findings are consistent with the notion that different parasitic protists have different intracellular V-PPase distributions and that these are determined by the 5Ј-coding sequences of their genes, namely whether they do or do not specify an N-terminal extension. In the kinetoplastid protist T. cruzi, low to intermediate resolution immunological methods localize the V-PPase to acidocalcisomes (11,44), whereas the results of more recent immunogold electron microscopic analyses of T. gondii have been interpreted in terms of localization of the V-PPase to both acidocalcisomes and the plasma membrane (10). In Plasmodium, by contrast, in which neither of the V-PPase genes, PfVP1 nor PfVP2 (7), encode a polypeptide containing an Nterminal extension, V-PPase immunostaining is predominantly associated with the plasma membrane, an observation supported by the finding that PfVP1 fusions preferentially target to the plasma membrane in trophozoites (7). On the basis of these findings, the sufficiency of the N-terminal extension of TgVP1 for entry of this polypeptide into the secretory pathway, and the presence of this extension not only in TgVP1 but also in its cognate from T. cruzi, there is a possibility that targeting of the V-PPase to the acidocalcisome is contingent on initial entry of the translation product into the secretory pathway.
The pronounced punctate apical distribution of the V-PPase described here clearly indicates localization of the V-PPase of T. gondii to the membranes bounding acidocalcisomes or similar vesicular structures. Moreover, from what we have determined to be necessary to maximize monospecificity, the differences between the results presented here and those published by other investigators are explicable in terms of the different cross-reactivities of the antibody preparations employed, namely that unless the antisera used (antisera generated by our laboratory (23) in all cases) are not first carefully immunopurified, they react relatively indiscriminately with parasitic protists. We, like Rodrigues et al. (8), observe plasma membrane immunostaining as well as a punctate and diffuse staining throughout the cell when crude antiserum is employed. However, if the monospecificity of the antiserum is maximized by immunoaffinity purifying it against heterologously expressed AVP1, only the structure-specific focused staining we describe here is observed. The implication is clear, the antisera raised against peptide sequences TKAADVGADLVGKIE and HKAAVIGDTIGDPLK from AVP1, which have since proven to contain universal V-PPase antibodies (2), also contain antibodies that react with other parasite antigens. Because these antibodies were originally intended to be used in plants and yeast, they were raised in New Zealand White rabbits whose preimmune and immune sera were prescreened against these organisms, not parasitic protists.
What cannot be determined from the investigations we describe here is whether the immunofluorescence seen is solely attributable to TgVP1 or is inclusive of other V-PPases. The antisera used in these studies were raised against peptides that are conserved among all V-PPases identified to date and, as such, do not distinguish type I from type II V-PPases (2). It was disappointing but none of the strategies applied in an attempt to circumvent this limitation were fruitful. Overexpression of TgVP1 fusions, for instance YFP fusions, consistently yielded aberrant nondividing cells, and all of our attempts at isolating a type II sequence from T. gondii were negative. There is a possibility that T. gondii does not contain a type II equivalent, in which case the immunolocalizations obtained are indeed exclusively attributable to TgVP1, but it is more likely, as is the case for P. falciparum (7), that the levels of the type II V-PPase transcript are low, perhaps undetectable in the tachyzoite and bradyzoite stages that were screened. Because of this it is not known if the dynamic distribution of immunostaining is specific to a particular V-PPase or is a property of acidocalcisomes in general. Confirmation or refutation of these two alternatives will have to await the results of further screens, completion of the Toxoplasma genome sequencing project, and/or expansion of the corresponding EST data base.
The functional significance of the collar-like structure to which the V-PPase immunolocalizes during host cell invasion is intriguing. Despite the seeming similarity of the distribution of the V-PPase antigen with that of micronemes, the lack of colocalization of this enzyme with the microneme MIC3 antigen dispels the notion of a micronemal localization. Analogously, equivalent experiments using antibodies raised against rhoptry and dense granule antigens show little or no colocalization with the V-PPase (data not shown). The most logical  (N). B, the acidocalcisomes (A), identified by dense material associated with their bounding membranes (arrows), ruptured in some cases. Note that ribosomes and cytosolic materials seep into the ruptured organelles resulting in areas of low density cytoplasm. C and G denote conoid and Golgi structures, respectively. explanation of these findings is therefore that the V-PPase immunolocalization detected is primarily to acidocalcisomal membranes and that in T. gondii these structures redistribute to assume a collar-like configuration during host cell infection.
A speculation that warrants further investigation is that acidocalcisomes associate at least transitorily with the cytoskeleton and micronemes at the anterior end of tachyzoites and redistribute with the penetration furrow during parasite host cell invasion. Previous studies (45) demonstrating an aligned organization of electron-dense, acidocalcisome-like organelles at the cell periphery of amastigotes and in close proximity to the flagellum in trypomastigotes of Trypanosoma imply acidocalcisome-cytoskeleton interactions. Moreover, as would be expected if this were the case, subcellular fractionation of these electron-dense vacuoles often results in their adherence to and copurification with cell ghosts from Trypanosoma, Leishmania, Toxoplasma, and Plasmodium (reviewed in Ref. 14). It is therefore noteworthy that recent immunofluorescence studies of the calmodulin-actomyosin complex in intracellular T. gondii disclose an apically focused pattern very similar to that found for the V-PPase in this study (46) and that earlier studies demonstrated an anteriorally displaced concentration of intracellular Ca 2ϩ within unidentified compartments, possibly acidocalcisomes, in the same organism (47). Knowing that microneme discharge, one of the first events in host cell invasion, requires the mobilization of intracellular Ca 2ϩ stores (48), it is conceivable that acidocalcisomes serve as a source of this Ca 2ϩ . Acidocalcisomes may thereby provide the Ca 2ϩ required for activating the calmodulin-dependent myosin light chain kinases that regulate the actomyosin motor that governs parasite motility and host cell invasion (46). If this were the case, the V-PPase would provide the motive force for acidocalcisomal Ca 2ϩ accumulation by H ϩ /Ca 2ϩ antiport.
The finding that even high levels of AMDP do not block host cell invasion by T. gondii despite the dynamic pattern of V-PPase immunofluorescence during this process was unexpected. In retrospect, however, it might be expected in that biochemical investigations of the acidocalcisome-like intracellular structures in trypanosomatids and P. falciparum, which react with the same antisera as used in this study, have shown these structures to release Ca 2ϩ upon treatment with V-ATPase inhibitors and/or AMDP (12,49). Thus, the inactivity, not the activity, of acidocalcisome-associated H ϩ -pumps like the V-PPase may promote Ca 2ϩ release from these organelles, in which case AMDP would be expected not to interfere with host cell invasion. It may be instructive to consider, given how sensitive heterologously expressed TgVP1 is to inhibition by Ca 2ϩ , that Ca 2ϩ release itself may abolish V-PPase activity, so diminishing or eliminating futile PP i consumption and H ϩ gradient-dependent organellar Ca 2ϩ uptake during host cell invasion.
As an extension of our studies of the in vitro sensitivity of heterologously expressed TgVP1 to inhibition by AMDP and the studies made by ourselves and others (7-9) of the in vivo sensitivity of parasitic protists to this and other 1,1-diphosphonates, we have explored the morphological consequences to T. gondii of exposure to this agent. As a result, we have determined that the toxic action of AMDP is evident at concentrations lower than reported previously. Rodrigues et al. (8), for instance, by employing a global drug assay based on measurements of [ 3 H]hypoxanthine incorporation, reported that concentrations of AMDP as high as 50 M did not inhibit tachyzoite proliferation; indeed, 200 M was not sufficient for 50% inhibition. We, in contrast, by counting the number of parasites per vacuole and examining their ultrastructure, have determined that AMDP concentrations as low as 5 M (concentrations commensurate with the I 50 for the in vitro inhibition of TgVP1) interfere with parasite replication. AMDP concentrations of 5-10 M and above block or stall cell replication but not cell growth such that the formation of highly enlarged tachyzoites containing multiple nascent daughter cells is apparent. The effects of AMDP are accompanied by extensive enlargement of the parasite nucleus which becomes highly lobed and the inflation and/or disruption of acidocalcisome-like vesicular structures.
The apparent absence of V-PPases from vertebrates and their likely involvement in energy conservation and membrane transport make these enzymes potential targets for the development of antiprotozoal agents. Notwithstanding these hopes and needs, it should be appreciated that care is required when interpreting the antiparasitic effects of AMDP. Two considerations are crucial. First, the efficacy AMDP in vivo is usually less than its efficacy in vitro, which likely reflects its bulky anionic character and slow permeation of membranes. Second, the greater intrinsic sensitivity of V-PPases versus soluble PPases and other phosphohydrolases to AMDP is in part attributable to the K m values of V-PPases, which are at least an order of magnitude greater those of soluble PPases (5,41). Therefore, the specificity of AMDP and other PP i analogs as competitive inhibitors will be critically dependent on the PP i concentration prevailing in the compartment in which the inhibitor exerts its effects.