Cryptosporidium parvum Mitochondrial-Type HSP70 Targets Homologous and Heterologous Mitochondria

Abstract

A mitochondrial HSP70 gene (Cp-mtHSP70) is described for the apicomplexan Cryptosporidium parvum, an agent of diarrhea in humans and animals. Mitochondrial HSP70 is known to have been acquired from the proto-mitochondrial endosymbiont. The amino acid sequence of Cp-mtHSP70 shares common domains with mitochondrial and proteobacterial homologues, including 34 amino acids of an NH₂-terminal mitochondrion-like targeting presequence. Phylogenetic reconstruction places Cp-mtHSP70 within the mitochondrial clade of HSP70 homologues. Using reverse transcription-PCR, Cp-mtHSP70 mRNA was observed in C. parvum intracellular stages cultured in HCT-8 cells. Polyclonal antibodies to Cp-mtHSP70 recognize a ∼70-kDa protein in Western blot analysis of sporozoite extracts. Both fluorescein- and immunogold-labeled anti-Cp-mtHSP70 localize to a single mitochondrial compartment in close apposition to the nucleus. Furthermore, the NH₂-terminal presequence of Cp-mtHSP70 can correctly target green fluorescent protein to the single mitochondrion of the apicomplexan Toxoplasma gondii and the mitochondrial network of the yeast Saccharomyces cerevisiae. When this presequence was truncated, the predicted amphiphilic α-helix was shown to be essential for import into the yeast mitochondrion. These data further support the presence of a secondarily reduced relict mitochondrion in C. parvum.

The apicomplexan parasite Cryptosporidium parvum infects humans and other animals worldwide, causing self-limiting disease in healthy hosts but life-threatening disease in immunocompromised individuals. Humans become infected with C. parvum when they ingest resistant oocysts in water, soil, or food. After ingestion, the oocyst releases four infectious sporozoites that give rise to intracellular (but extracytoplasmic), asexually multiplying merozoites within enterocytes of the small intestine. Finally, sexual development leads to the development either of thin-walled oocysts that may excyst within an immunocompromised host and prolong infection or thick-walled oocysts that are expelled to the environment (15). Species of Cryptosporidium belong to the obligate parasitic phylum Apicomplexa, which includes other agents of medical and veterinary importance, i.e., toxoplasmosis (Toxoplasma gondii), malaria (Plasmodium spp.), babesiosis (Babesia microti), and avian coccidiosis (Eimeria spp.). The phylum is characterized by the presence of an apical complex. Unlike the case with other members of the Apicomplexa (Coccidia, Hematozoa, and Gregarina), morphological evidence for a mitochondrion in Cryptosporidium spp. has been limited (4, 7, 41, 48), and some thought the genus was amitochondriate (15, 44). However, recent data indicate that the Cpn60-containing, ribosome-studded organelle of C. parvum is indeed a mitochondrial remnant (40).

The mitochondrion is a unique organelle of eukaryotes, the acquisition of which is explained by endosymbiosis of an α-proteobacterial proto-mitochondrion into the host cell (9, 32). Improved phylogenetic analyses, as well as the discovery of mitochondrial homologues in Giardia, Entamoeba, parabasalia, and microsporidia (22, 38, 47, 50), suggest that there are no ancestrally branching, amitochondriate eukaryotes, i.e., that the mitochondrial compartment is essential.

According to robust morphological evidence and phylogenetic reconstruction, the common ancestor for all members of the Apicomplexa (Alveolata), including species of Cryptosporidium, must have possessed a fully functional mitochondrion. Among extant Apicomplexa, Cryptosporidium spp. are unique in the apparent absence of a well-developed and respiring mitochondrion.

Like other eukaryotes, most apicomplexans possess a mitochondrial genome, but the majority of mitochondrial proteins are encoded by the nucleus, synthesized as precursors in the cytosol, and imported into mitochondria (19, 29, 31). Unlike its apicomplexan relatives, C. parvum appears to lack both a mitochondrial genome (C. E. Riordan and J. Keithly, unpublished data) and apicoplast DNA (52). For example, Plasmodium falciparum has a 23-Mb nuclear genome that encodes 5,282 proteins, 466 of which (8.8%) are predicted to target the apicoplast (18) and 246 (4.7%) of which are predicted to target the mitochondrion (21). Initial analyses of the partial C. parvum nuclear genome indicate a putative existence of several nucleus-encoded mitochondrial proteins, but the existence of nucleus-encoded apicoplast proteins remains elusive (6).

The majority of mitochondrial matrix proteins have an NH₂-terminal mitochondrial presequence that is removed by a specific matrix protease upon translocation across the mitochondrial membrane. Although presequences do not share a common primary amino acid structure, they all do have the ability to form a positively charged amphiphilic α-helix (36). Moreover, the binding of the presequence helix into an apolar groove in Tom20 chiefly depends upon hydrophobic, rather than hydrophilic, amino acids (1, 39). Although mitochondrial protein import is thought to be evolutionarily conserved in animals and plants, some differences exist. For example, plant mitochondrial targeting presequences are longer and have a higher serine content (13). Because plants also contain nucleus-encoded genes for chloroplast proteins, it is thought that a more stringent organellar targeting system occurs in them and that programs analyzing plant genomes are subsequently less accurate in predicting mitochondrial presequences (30).

Ubiquitous chaperones belonging to the 70-kDa class are known to bind immature proteins or preproteins and to facilitate their maturation and translocation across membranes into several subcellular compartments. Eukaryotes possess at least three types. Those of the cytosol and endoplasmic reticulum (ER) result from an ancient gene duplication in eukaryotic lineages, whereas that of the mitochondrion results from endosymbiosis of a DnaK-containing α-proteobacterium together with numerous other mitochondrial proteins (8, 29). The mitochondrial 70-kDa chaperone (mtHSP70) is an essential component of the Tim (translocase of the inner membrane) mitochondrial import complex, which binds preproteins on the matrix side of the inner mitochondrial membrane, and serves as an import motor for matrix proteins (39). Although a cytosolic HSP70 had been known for C. parvum (26), neither an mtHSP70 nor a component of the mitochondrial translocase machinery had been observed.

Here we describe a C. parvum mtHSP70 (Cp-mtHSP70) that (i) is a nuclear gene with clear proto-mitochondrial origins, (ii) possesses a mitochondrial targeting sequence, and (iii) is part of the mitochondrial protein import machinery. A reporter gene (green fluorescent protein [GFP]) is used to show that the C. parvum presequence targets mitochondria in the apicomplexan T. gondii and the yeast Saccharomyces cerevisiae. Analyses of truncated C. parvum targeting presequences revealed the amino acids within the predicted helix of the presequence to be essential for import. Fluorescence and immunogold microscopy showed the localization of labeled polyclonal antibodies within the relict mitochondrion of C. parvum.

MATERIALS AND METHODS

Parasites and cell cultures.

Cryptosporidium parvum (IOWA strain; bovine origin) oocysts were obtained and treated as previously described (51). Human colonorectal adenocarcinoma (HCT-8) cell cultures were maintained in 25-ml flasks or in Transwell (diameter, 12 mm; pore size, 0.4 μm; Costar) plates at 37°C in 5% CO₂ until confluent. Clorox-treated and phosphate-buffered saline (PBS)-washed C. parvum oocysts were added to HCT-8 cultures in a 1:1 ratio. Mock-infected cultures were treated identically, except that C. parvum oocysts were not introduced into the culture. Total genomic DNA and RNA were isolated from excysted C. parvum sporozoites and host cells with the QIAamp DNA Mini kit (Qiagen) and RNeasy Mini kit (Qiagen) with the RNase-free DNase set (Qiagen) according to the manufacturer's instructions. T. gondii tachyzoites (RH strain) were grown in human foreskin fibroblast cells (a generous gift of K. Kim, Albert Einstein College of Medicine) following published protocols (27).

Cloning of Cp-mtHSP70.

A partial, unfinished gene was retrieved using a BLAST search of finished and unfinished databases at the National Center for Biotechnology Information server (10). A contig (cparvum_Contig1821) with high scores for both bacterial DnaK and eukaryotic mitochondrial homologues was retrieved. The sequence was confirmed by amplifying the gene using primers starting 21 nucleotides upstream and 48 nucleotides downstream from the predicted start and stop codon: primers F-21 (5′-CAC TGA CTT GTT CTT TCG AG-3′) and R-48 (5′-ACT AGG AGA GAG AAA TCT CTA GG-3′). The predicted gene was amplified using the following primers (start and stop codons are underlined): forward (5′-ATG TCT ATG ATA ATT AAT AGT AG C-3′) and reverse (R; 5′-TTA AGA ATC TGA ATC TTG AGA CTC-3′). All PCR amplifications were performed using PfuULTRA or PfuTURBO DNA polymerases (Stratagene). For verification the gene was cloned into pCR-BluntII-TOPO (Invitrogen) and sequenced from three independent PCRs.

RT-PCR and semiquantitative RT-PCR of C. parvum intracellular stages.

Differential mRNA display using reverse transcription (RT)-PCRs was performed essentially according to the method of Abrahamsen and Schroeder (2). To rule out DNA contamination of DNase-treated RNA, conventional PCR was performed using primers specific for C. parvum small subunit (SSU) ribosomal DNA and human β-globin. The cDNA was synthesized from total RNA using random hexamer primers from the SuperScript First-Strand Synthesis system for RT-PCR (Invitrogen). Subsequent PCR with 1 U of TaqDNA polymerase (Promega) contained 25 nM (each) deoxynucleoside triphosphates, 0.1 μCi of [α-³²P]dATP (3,000 Ci mmol⁻¹; Amersham), and 100 ng of each forward and reverse primer. Amplification was performed using a GeneAmp 9700 (Perkin-Elmer) cycler as follows: initial denaturation at 95°C for 5 min, cycling at 94°C for 50 s, 55°C for 30 s, 74°C for 60 s (total of 24 and 30 cycles), and a final extension at 74°C for 10 min. Amplicons were separated on 5% nondenaturing acrylamide gels and were measured by phosphor-imaging (STORM860) using ImageQuant 5.1 software (Molecular Dynamics). For identification of Cp-mtHSP70 in cDNA, the forward primer (5′-ATG CTT GGG TTG AAG CTA GAG-3′) and reverse primer (5′-GAA GTG TTA CCA TTA GTT GC-3′) yielded a 360-bp amplicon. To determine the ratio of C. parvum signal to growing parasites in infected cultures, the mRNA signal (cDNA dilution 1:50) was normalized to the C. parvum SSU rRNA signal detected in each of the samples. Three independent experiments were analyzed, including controls of RNA isolated from parallel cultures prior to infection, samples without RNA, and C. parvum DNA.

Generation of S. cerevisiae constructs and transformation.

The Escherichia coli/yeast shuttle GFP vector, pYX122-mtGFP, was digested with EcoRI and BamHI to exclude the Su9 mitochondrial targeting presequence (49). The NH₂-terminal 46 amino acids of Cp-mtHSP70 (¹GVVNSSGIAA RILKRSLPLV FSRY↓MSSKCE GKSSN⁴⁶) were amplified and cloned in frame, using EcoRI and BglII sites, into the pYX122-GFP vector as an NH₂-terminal GFP extension to yield pYX122(mtHSP)GFP. Cp-mtHSP70 nucleotide position 42 was primer mutated (C→T) to mask the EcoRI site, but to retain Asn. The truncated forms of the presequence (see Fig. 4C) were generated from pYX122(mtHSP)GFP using a PCR approach with primers spanning the mtHSP70 presequence plus GFP and were ligated into EcoRI- and HindIII-digested pYX122. As a positive control, an NH₂-terminal sequence of S. cerevisiae mtHSP70 (SSC1) (43) (¹MLAAKNILNR SSLSSSFRIA TRL↓QSTKVQG SVIGI³⁵) was amplified using primers with EcoRI and BamHI, and these were cloned into respective sites in frame to pYX122-GFP as an NH₂-terminal GFP extension. As a negative control, a vector overexpressing only GFP (with no ability to target mitochondria) was used. The identity of the vectors was verified by sequencing. Standard methods for yeast growth and transformation were used (49). Yeast cultures grown in selective media were examined with a Zeiss Axioskop 2 motplus fluorescence microscope. The MX1-4C yeast strain was a gift from R. Morse (34).

FIG. 4.

Presequence of C. parvum mtHSP70. (A) Secondary structure prediction. Line 1 (SSpro8): H, alpha-helix; G, 310-helix; I, pi-helix; E, extended strand; B, beta-bridge; T, turn; S, bend; C, the rest. Line 2 (Psi-Pred): H, helix; E, strand; C, coil. Line 3 (Predator): H, helix; E, extended sheet; _, coil. Line 4: amino acid sequence of the Cp-mtHSP70 NH₂ terminus. Eisenberg's hydrophobic moment plot is immediately beneath and aligned with the amino acid sequence above it. The arrow indicates the R-2 motif, xRx↓x(S/x). Numbering denotes amino acid residues. (B) Helical wheel plot of residues 15 to 32 with putative amphiphilic α-helix demonstrating partitioning of charged and hydrophobic amino acids (gray and black circles, respectively). The helical wheel plot was generated with the aid of Java-applet at http://cti.itc.virginia.edu/∼cmg/Demo/wheel/wheelApp.html. (C) Truncated NH₂-terminal amino acid sequences of Cp-mtHSP70 cloned into the E. coli/yeast shuttle vector pYX122-GFP, including the yeast SSC1 control (sequence in italic). Numbering above denotes amino acid residues. Constructs are named on the right. (D) Graph of the import efficiency of different GFP constructs based on Western blotting of subcellular fractions. Import efficiency is the ratio between mitochondrially imported GFP and total GFP (mitochondrial and cytosolic) relative to the control construct SSC1. Results are calculated from three independent experiments. Numbering of the constructs corresponds to that in panel 4C.

Generation of the T. gondii construct and transfection.

To evaluate the ability of the Cp-mtHSP70 presequence to target the intracellular compartment of T. gondii tachyzoites, the NH₂-terminal 46 amino acids (¹MSMIINSSFN GVVNSSGIAA RILKRSLPLV FSRYMSSKCE GKKSSN⁴⁶) were introduced in-frame between GRA1 and GFP using gene splicing by overlap extension (11). The HindIII/NsiI construct was generated using a fusion GRA1 plus mtHSP70 primer (5′-ACT ATT AAT TAT CAT AGA CAT CTT GCT TGA TTT CTT CAA AG-3′). The amplified product was digested with HindIII and NsiI and then cloned into the plasmid GRA1/GFP/GRA2-SK (provided by K. Kim) to create pTgGRA1-pre-GFP. Ten million freshly lysed T. gondii tachyzoites were transfected by electroporation (GenePulser, Bio-Rad) in a 4-mm cuvette (25 μF; 1.5 kV) using 80 μg of a respective plasmid DNA as previously described (27). After electroporation, parasites were inoculated into human foreskin fibroblast cells. Transient transformation was detected 24 to 48 h postinoculation by using an epifluorescence microscope (Olympus). Mitochondria of T. gondii tachyzoites were labeled with 10 nM MitoTracker Red CM-H₂XRos (Molecular Probes).

Antibody production.

The Cp-mtHsp70Δ1-34 (excluding the targeting presequence ¹MSMIINSSFN GVVNSSGIAA RILKRSLPLV FSRY³⁴) was amplified with overhanging BamHI and SalI, ligated in-frame into the polylinker cloning site of pMAL-c2X (New England Biolabs), and verified on an automated sequencer. The verified construct was expressed in E. coli Epicurian BL21-RIL CodonPlus (Stratagene). Purification was done according to the standard protocols (New England Biolabs), and the cells were induced at an A₆₀₀ of 0.5 with a final concentration of 0.3 mM isopropyl-β-d-thiogalactoside for 3 h at 20°C. The soluble maltose binding protein (MBP) fusion (MBP:Cp-mtHsp70Δ1-34) was purified using an amylose resin column and was cleaved using Factor Xa protease to separate MBP from Cp-mtHsp70Δ1-34. Five consecutive doses of polyacrylamide gel excised bands containing ∼200 μg of Cp-mtHsp70Δ1-34 were washed once overnight in 50 ml of H₂O and once overnight in 50 ml of PBS (pH = 7.4) and then homogenized in a final volume of 1 ml of PBS (pH = 7.4) and injected into a female rabbit (Y910) at 2-week intervals. Preimmune serum was bled prior to the first dose. The total serum of this rabbit was purified 2 weeks after the fifth dose over a SulfoLink gel column coupled with MBP:Cp-mtHsp70Δ1-34 using a SulfoLink kit (Pierce). For Western analysis, a working dilution of 1:1,000 was used for purified anti-Cp-mtHSP70.

Western analyses and measurement of the import of GFP.

Yeast mitochondrial extracts were prepared from yeast spheroplasts, and subcellular extracts (including purified mitochondria) were prepared as described by Diekert et al. (12). Identical volumes of proteins from different extracts were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to an Immun-Blot polyvinyl difluoride membrane (Bio-Rad). Nonspecific binding was blocked using 10% nonfat dry milk-3% bovine serum albumin (BSA) in PBS with 0.05% Tween-20 (PBS-T) for 1 h. Antibodies were diluted in 3% BSA-PBS-T and incubated at room temperature for 60 min.

For detection of yeast GFP import, rabbit polyclonal antibodies and purified anti-GFP (Invitrogen) were used at a dilution of 1:5,000. Mouse monoclonal anti-yeast cytochrome oxidase subunit III (DA3) was used at a concentration of 2.5 μg/ml (Molecular Probes). Secondary goat anti-rabbit horseradish peroxidase conjugate (1:10,000; BioSource) and goat anti-mouse horseradish peroxidase conjugate (1:2,500, Promega), respectively, were incubated for 60 min. Results were observed using Chemiluminescence Reagent Plus (NEN). The signals were quantified using ImageQuant 5.1 (Molecular Dynamics). The GFP signal was detected for mitochondrial and cytosolic fractions prepared as described by Diekert et al. (12). To take into account the variability in the fragmentation of mitochondria during homogenization, values were corrected by the ratio obtained for the endogenous mitochondrial cytochrome oxidase subunit III. For each construction, GFP import was expressed as a ratio between GFP detected in the prepared mitochondrial fraction to the cytosolic and mitochondrial fraction together. The resulting number was then expressed as a percentage of the ratio obtained for the yeast SSC1 control ± standard deviation.

Immunofluorescent antibody staining.

All manipulations were carried out at room temperature. Freshly excysted C. parvum sporozoites attached to poly-l-lysine (Sigma)-coated slides and in vivo-cultured intracellular parasite stages in HCT-8 cells attached to the Transwell (Costar) membranes were washed with PBS (pH = 7.4), fixed in fresh 4% paraformaldehyde in PBS for 5 min, and rinsed in 0.1 M glycerin-PBS. Cells were permeabilized using 0.2% Triton X-100 in 3% BSA-PBS-T for 3 min and were then washed with PBS. Slides were incubated for 120 min with purified anti-Cp-mtHSp70 antibodies in 3% BSA-PBS-T (dilution, 1:50), washed, and incubated further for 60 min in 1:80 green fluorescein isothiocyanate (FITC)-labeled goat anti-rabbit immunoglobulin G (Sigma) diluted in 3% BSA-PBS-T. The slides were counterstained for 5 min with 0.1 μg of the blue fluorescent nuclear stain DAPI (Sigma) ml⁻¹. Finally, slides were mounted using the SlowFade Light Antifade kit (Molecular Probes). The slides were examined under a Zeiss Axioskop 2 motplus microscope; black-and-white images were recorded with the ORCA-ER (Hamamatsu) digital camera and then superimposed and pseudo-colored by using OpenLab 3.1. software (Improvision). Controls were processed in parallel. These included preimmune serum as the primary antibody, with those without primary antibody containing only secondary antibodies.

Electron microscopy.

Purified sporozoites were fixed and embedded in epon or LR White as previously reported (40, 41). Briefly, sporozoites for epon embedding were fixed in osmium and electron microscopic (EM)-grade glutaraldehyde, whereas those for LR White were fixed in EM-grade methanol-free formaldehyde, with additional fixation in 4% formaldehyde-0.1% EM-grade glutaraldehyde. Thin LR White sections (0.1 μm) were incubated with purified anti-Cp-mtHSP70 or a preimmune serum as a primary antibody and a 1:50 dilution of 10-nm-gold-conjugated goat anti-rabbit immunoglobulin G as a secondary antibody and examined under a Zeiss electron microscope.

Prediction analyses.

Plant and nonplant algorithms were used with TargetP 1.0 to predict protein localizations (14). Apicoplast targeting was predicted using PlasmoAP (18). Secondary structures were predicted using SSPro8, Psi-Pred, and Predator (42).

Phylogenetic analyses.

Multiple sequence alignments included 53 sequences from DnaK and eukaryotic homologues. Initially the National Center for Biotechnology Information database was searched (blastp), using DnaK and its known homologues to retrieve a broad spectrum of sequences (3). To determine the origin of Cp-mtHSP70, only a limited number of diverse representatives from bacterial and eukaryotic taxa were finally selected because of the extensive computational analyses required. The existing homologues of T. gondii HSP70 were not used in the final phylogenetic analyses due to their origin within the unfinished T. gondii genome project. Sequences were aligned using Clustal X 1.81, PAM 250 matrix (45), and those ambiguously aligned were excluded to finally yield a total of 398 residues. These are available from the authors upon request.

Two methods were used for tree reconstruction based on amino acid sequence alignment. Protein maximum-likelihood analysis employing the JTT model of amino acid evolution (alpha parameter = 2, with eight rate categories of amino acid changes) and bootstrapped using 100 replicates was performed with ProML, SeqBoot, and Consense programs of PHYLIP 3.6a3 (16). A Bayesian phylogenetic search for tree space used a variant of the Markov chain Monte Carlo and was performed in the MrBayes 3.01 program (24). Metropolis-coupled Markov chain Monte Carlo analysis used the JTT model of amino acid evolution. The Markov chain was started from a random tree and run for 500,000 generations, sampled every 100 generations with four chains. The first 30,000 generations were finally discarded for calculation of posterior probabilities (PP).

Nucleotide sequence accession number.

The complete sequence for Cp-mtHSP70 has been deposited in the GenBank database under accession no. AY235430, along with the conceptually translated peptide under accession no. AAP59793.

RESULTS

C. parvum possesses mtHSP70.

A 2,052-bp intronless gene encoding a protein of 683 amino acids with a predicted molecular mass of 74.6 kDa (excluding the targeting presequence, 70.9 kDa) and having significant homology to mitochondrial homologues of the 70-kDa heat shock protein family (HSP70) was cloned and sequenced from C. parvum (Cp-mtHSP70). The AT content is 62.8%. A BLAST (blastp) search yielded highest scores with eukaryotic mitochondrial HSP70 homologues and prokaryotic DnaK. The sequence of Cp-mtHSP70 (accession no. AAP59793) contains 46% identities, 64% conservative substitutions, and 3% gaps compared (BLAST2) to C. parvum cytosolic HSP70 (accession no. AAC25925) (26). Unlike the case with cytosolic HSP70, no amino acid repeats of GGMP were found near the COOH terminus of Cp-mtHSP70. The identity (similarity) scores for Cp-mtHSP70 and HSP70 from completely sequenced genomes (P. falciparum and S. cerevisiae) using BLOSUM62 are as follows: for mitochondrial P. falciparum, 69% (79%) sequences, and for S. cerevisiae, 55% (72%) sequences; for cytosolic P. falciparum, 40% (58%) sequences, and for S. cerevisiae, 42% (59%) sequences. The amino acid sequence analyses indicate that Cp-mtHSP70 is a mitochondrial-type HSP70 (Fig. 1).

FIG. 1.

Multiple protein sequence alignment of cytosolic and mitochondrial HSP70. Conceptually translated full-length amino acid sequences of C. parvum together with S. cerevisiae and P. falciparum were aligned using the program Clustal X and the BLOSUM protein weight matrix. The upper three sequences are mitochondrial; the lower three are cytosolic. Black shading indicates that identical amino acids are conserved, whereas gray indicates conservation of similar amino acids. The threshold for shading was set to 50%. A solid line denotes the predicted Cp-mtHSP70 mitochondrial presequence; a dashed underline denotes the SSC1 yeast mitochondrial presequence (43). The two mitochondrial/proteobacterial sequence signatures motifs, GDAWV and YSPSQI, are denoted with an asterisk above the alignment. The species name and GenBank accession numbers are indicated at the right of the alignment.

Total RNA from C. parvum-infected HCT-8 cells isolated at 6, 12, 24, and 72 h postinfection (p.i.) and from freshly excysted and purified sporozoites was analyzed by semiquantitative RT-PCR to determine whether Cp-mtHSP70 is differentially expressed. Amplification through 24 cycles using specific primers yielded a distinct 360-bp RT-PCR product in HCT-8 cells at all h p.i., as well as in sporozoites (Fig. 2A). The overall pattern of amplification was identical in three independent experiments. Minor differences were noted in product abundance: peaks were highest at 12 and 72 h p.i. and decreased at 48 h p.i. (Fig. 2B). Although the signal for sporozoites is greater than those for all intracellular time points, interpretation is limited by the fact that these samples were prepared independently from those of HCT-8 cultures. Furthermore, and in congruence with previous data (2), C. parvum-infected HCT-8 controls began transcribing actin 12 h p.i., which gradually decreased over time, whereas transcription of the Cryptosporidium oocyst wall protein occurred during gametocyte (sexual) development at 48 to 72 h p.i. (data not shown). Together, these results suggest that Cp-mtHSP70 is constitutively expressed during the life cycle.

FIG. 2.

Transcription and expression of C. parvum mtHSP70. (A) RT-PCR analysis of expression during in vitro development in HCT-8 cells and sporozoites (spor.). Polyacrylamide gel signals (Cp-mtHSP70, SSU rRNA) were analyzed using ImageQuantNT (Molecular Dynamics) after 24 cycles. (B) Graph indicating amount of signal after 24 PCR cycles normalized to the amount of SSU rRNA signal detected in each of the infected samples. Three independent experiments were analyzed. (C) Western blot analysis of a crude extract of 10⁷ C. parvum sporozoites. Preimmune sera (lane 1) and immune anti-Cp-mtHSP70 sera (lane 2) are shown. The antisera recognize a single band (arrowhead) of ∼70 kDa in size.

Phylogeny of mtHSP70 and DnaK homologues.

For initial analyses, databases were searched (blastp), using known HSP70 and DnaK homologues to retrieve sequences over a wide taxonomic spectrum (3). To determine the origin of Cp-mtHSP70, 53 sequences from representative bacterial groups and eukaryotic taxa were selected (Fig. 3). The alignment consisted of 398 residues. Protein maximum-likelihood analysis with bootstrapping replicates (BP) and a Bayesian phylogenetic search with the PP were calculated (see Materials and Methods). Cytosolic HSP70 types are clearly monophyletic, with a PP value of 1.00 and BP value of 82%. The resolution within the ER HSP70 cluster is imprecise due to the less reliable alignment for them, but overall ER types of HSP70 are a sister group to those of the cytosol. The monophyly of cytosolic and ER HSP70 is supported with 1.00 PP and 100% BP. Mitochondrial HSP70 types cluster within α-proteobacterial DnaKs and the other bacteria as a monophyletic mitochondrial branch. Although the best-reconstructed trees support the monophyly of the mitochondrial branch, there is only 0.69 PP and <50% BP support. The sister relationship of α-proteobacterial sequences and the mitochondrial clade is also less supported (0.63 PP and <50% BP). This is due to an alternative branching of Erlichia plus Rickettsia and Rhodopseudomonas plus Agrobacterium plus Sinorhizobium with mitochondrial HSP70, i.e., maximum-likelihood monophyly of Erlichia plus Rickettsia with mitochondrial HSP70. The topology of the reconstructed trees is essentially identical to those recently reported for the analyses of homologues from Giardia, Entamoeba, parabasalia, and microsporidia (5, 22, 35).

FIG. 3.

Phylogenetic tree reconstruction of DnaK and HSP70 homologues. The tree was calculated with the MrBayes program using the JTT model and rooted on Methanothermobacter and Thermoplasma sequences. The numbers on branches are PP MrBayes/bootstrap support ProML, with only values of >0.5/50 shown. GenBank accession numbers for bacterial and archeal DnaK are as follows: Aquifex aeolicus, AAC07071; Borrelia burgdorferi, AAC66887; Chlamydia trachomatis, AAC3683; Ehrlichia sennetsu, AAC27487; Escherichia coli, AAC73125; Francisella tularensis, AAA69561; Haemophilus influenzae, AAC22889; Halobacterium salinarum, AAC41461; Leptospira interrogans, AAC35416; Methanothermobacter thermautotrophicus, CAA14651; Agrobacterium tumefaciens, CAA60592; Mycobacterium avium paratuberculosis, CAA42063; Rhodopseudomonas sp., BAA19796; Rickettsia prowazekii, CAA14651; Sinorhizobium meliloti, AAA64925; Thermoplasma acidophilum, AAC41460. GenBank accession numbers for eukaryote homologues, cytosolic, are as follows: Babesia bovis, AAF14194; Blastocladiella emersonii, AAA65099; Caenorhabditis elegans, AAA28078; Cryptosporidium parvum, AAC25925; Drosophila melanogaster, AAA28625; Giardia lamblia, EAA38588; Leishmania major, P14834; Mus musculus, AAA37864; Pisum sativum, CAA67867; Plasmodium falciparum, NP_704366; Saccharomyces cerevisiae, AAC37398; Schizosaccharomyces pombe, BAA25322; Theileria annulata, A44985; Trichomonas vaginalis, AAB52423; Trypanosoma cruzi, AAA30205. GenBank accession numbers for endoplasmatic reticulum (ER) HSP70s are as follows: Drosophila melanogaster, AAA28626; Giardia lamblia, EAA41481; Plasmodium falciparum, NP_704718; Trypanosoma brucei, AAC37174; Zea mays, AAC49900. GenBank accession numbers for mitochondrial HSP70s are as follows: Arabidopsis thaliana, AAO00750; Caenorhabditis elegans, AAB42371; Cryptosporidium parvum, AAP59793; Drosophila melanogaster, AAA28628; Eimeria tenella, CAA87086; Encephalitozoon cuniculi, CAA10035; Entamoeba histolytica, AAG16651; Giardia intestinalis, BAB84357; Homo sapiens, AAA67526; Leishmania major, P12076; Nosema locustae, AAC47660; Plasmodium falciparum, BAB17688; Saccharomyces cerevisiae, AAA63792; Schizosaccharomyces pombe, AAA35314; Solanum tuberosum, Q08276; Trichomonas vaginalis, AAB09772; Trypanosoma cruzi, AAA30215.

Because the full genome sequence for P. falciparum is known (21), all available sequences for its HSP70 complex have been retrieved. There is a single copy each of cytosolic, ER, and mitochondrial HSP70 in the genome of P. falciparum. As expected, phylogenetic reconstruction of P. falciparum, C. parvum, and other apicomplexan sequences shows that they cluster together (Fig. 3). Mitochondrial homologues from C. parvum, P. falciparum, and Eimeria tenella form a monophyletic clade within eukaryotic mitochondrial DnaK homologues, whereas cytosolic HSP70 types of C. parvum, P. falciparum, and Theileria plus Babesia are monophyletic within the eukaryotic cytosolic HSP70 clade (Fig. 3). This monophyly is strongly supported (1.00 PP and 99 to 100% BP). Thus, the phylogenetic distinctions between apicomplexan cytosolic and mitochondrial HSP70 strongly suggest different evolutionary origins for them: cytosolic HSP70 from a common ancestor with eubacteria and archaea, and mitochondrial HSP70 from an endosymbiotic event. Moreover, the ER type (also known as BiP) required to power ER posttranslational translocation by pulling the polypeptide into the ER membrane (8) has been identified in the completed C. parvum genome (www.CryptoDB.org).

Presequence of Cp-mtHSP70.

TargetP was used to predict the in silico localization (Table 1) of apicomplexan and yeast protein sequences that cluster within the mitochondrial and cystosolic branches of the phylogenetic tree, i.e., P. falciparum, S. cerevisiae, and C. parvum (Fig. 3). Although both plant and nonplant algorithms were used, a specific organellar compartment for the putative P. falciparum mtHSP70 homologue could not be assigned. For Cp-mtHSP70, on the other hand, the nonplant algorithm predicts a mitochondrial localization (0.74), but using a plant algorithm, compartmentalization into a chloroplast or apicoplast (0.79) instead of a mitochondrion (0.05) is predicted. PlasmoAP was used to further test the potential of Cp-mtHSP70 sequences for apicoplast targeting (18). No signal sequence of the bipartite apicoplast targeting presequence was detected, and apicoplast targeting was rejected.

TABLE 1.

Summary of in silico predicted protein localizations^a

Acc. no.	Prediction					RC
	mTP	cTP	SP	Other	Loc.
Cp AAP59793mit	0.74/0.50	0.79	0.02/0.13	0.40/0.47	M/C	4/2
Cp AAC25925cyt	0.041/0.09	0.23	0.11/0.12	0.92/0.81		1/3
Pf NP_701211mit	0.41/0.05	0.27	0.03/0.19	0.76/0.40		4/5
Pf NP_704366cyt	0.11/0.09	0.16	0.06/0.16	0.89/0.74		2/3
Sc AAA63792mit	0.93/0.62	0.23	0.01/0.04	0.15/0.28	M/M	2/4
Sc AAC37398cyt	0.15/0.16	0.10	0.07/0.21	0.78/0.72		2/3

^aValues were calculated by TargetP 1.0 using nonplant/plant algorithms: mTP and cTP, mitochondrion and chloroplast targeting peptides; SP, signal peptide; other, cytosolic localization or undetermined; Loc, localization based on “winner takes all” (either M = mitochondrion or C = chloroplast); RC, confidence index (1 = excellent; >5 = poor). mit, mitochondrion homologue; cyt, cytosolic based on phylogenetic analyses. Acc. no., accession number of protein sequence. Organism is indicated at left of accession no.: Cp, C. parvum; Pf, P. falciparum; Sc, S. cerevisiae.

Based upon SSPro8, Psi-Pred, and Predator programs for predicting secondary structures, the C. parvum presequence was predicted to form α-helices among amino acid residues 15 to 26, 17 to 38, and 15 to 29 (Fig. 4A). Using Eisenberg's plot, a significant hydrophobic moment was observed for residues 15 to 26 (Fig. 4A), suggesting the possibility of a bifacial amphiphilic α-helix between these amino acids (Fig. 4B).

Several motifs are known to be specific cleavage sites for a mitochondrial protein peptidase (20): motif R-2 [xRx↓x(S/x)]; motif R-3 [xRx(Y/x) ↓ (S/A/x)x]; or R-none motif [xx↓x(S/x)]. The primary amino acid sequence of Cp-mtHSP70 contains an R-2 motif, ³¹FSRY↓MSSK³⁸ (signature amino acids in bold), and predicts a mitochondrial presequence of 34 amino acids (¹MSMIINSSFN GVVNSSGIAA RILKRSLPLV FSRY³⁴). Although there are two potential initiation codons at the 5′ end of this peptide, based upon consensus nucleotide start codon sequences, the first ATG was chosen as the initial start of translation.

Heterologous targeting of Cp-mtHSP70 presequence to the mitochondrion of S. cerevisiae and T. gondii.

To compare the in silico predictions for the C. parvum presequence with in vivo targeting to heterologous mitochondria, the presequence was tested for its ability to deliver GFP into the mitochondrion of yeast and the genetically well-characterized apicomplexan T. gondii. Initially, the entire Cp-mtHSP70 presequence was cloned onto the NH₂-terminal end of GFP in the yeast expression vector pYX122(mtHSP)GFP (Fig. 4C). The tubular mitochondrial network of pYX122(mtHSP)GFP-transformed yeast cells exhibited a high-intensity green fluorescence (Fig. 5A, Cp) that was identical to that of the control yeast mtHSP70 homologue, SSC1 (Fig. 5A, Sc). These data indicate that the complete Cp-mtHSP70 presequence is capable of delivering GFP into the mitochondrial network of heterologous mitochondria. The GFP import was quantified by Western blot analysis of mitochondrial and cytosolic fractions from transformed yeast. The variability of fragmentation of mitochondria during homogenization and the release of GFP into the cytosolic fraction was corrected using immunodetection of the COXII mitochondrial marker. The complete C. parvum presequence delivered a GFP signal that was 96% ± 8% of that exhibited by isolated mitochondria of the SSC1 control. In contrast, the fluorescence of a negative control, vector Δ1-45(mtHSP)GFP lacking the presequence, diffusely stained the cytosol and not the mitochondrial network (Fig. 5A, cGFP). To further analyze the properties of the predicted helical structure of the presequence, truncated constructs were cloned into the pYX122-GFP vector (Fig. 4C). First, the targeting sequence was truncated consecutively at the NH₂ terminus, Δ1-10, Δ1-20, and Δ1-30. Yeast transformed with the Δ1-10 and Δ1-20 constructs predominantly targeted the yeast mitochondrial network (Fig. 5A, Δ1-10 and Δ1-20), yielding a GFP fluorescence signal in the tubular mitochondrial network of 92% ± 1% and 79% ± 3%, respectively (Fig. 4D). The Δ1-30 construct showed a diffuse cytosolic fluorescence (Fig. 5A, Δ1-30) and only 5% ± 5% of the GFP signal was localized within the mitochondrion, i.e., the removal of 21 to 30 amino acids appears to remove targeting capability. Furthermore, when various internal parts of the presequence were truncated, but leaving the first 10 NH₂-terminal amino acids in place, only the Δ11-20 construct was still able to specifically target the yeast mitochondrial tubular network (Fig. 5A, Δ11-20), delivering 75% ± 6% of GFP signal to the organelle. Constructs that eliminated most or the rest of the presequence showed a diffuse cytosolic fluorescence (Fig. 5A, Δ11-30 and Δ11-40), delivering only 13% ± 5% and 2% ± 2% GFP to the yeast mitochondrion, respectively (Fig. 4D). These data suggest that amino acid residues 11 to 30 are essential for mitochondrial targeting, with residues 21 to 30 being the most critical, while NH₂-terminal residues 1 to 10 appear to be insignificant. The data essentially agree with the predicted importance of the bifacial amphiphilic α-helix in the C. parvum presequence (Fig. 4A).

FIG. 5.

Targeting of C. parvum mtHSP70 presequence to heterologous mitochondria. (A) Targeting of the yeast S. cerevisiae using the Cp-mtHSP70 presequence at the NH₂ terminus of GFP vectors (see Fig. 4C). Each panel shows paired GFP and differential interference contrast (DIC) images. Full C. parvum presequence (Cp), negative control lacking the presequence (targeting the cytosol; cGFP), positive control—the yeast mtHSP70 homologue SSC1 presequence, targeting the yeast mitochondrial network (Sc), and the C. parvum truncated forms, Δ1-10, Δ1-20, Δ1-30, Δ10-20, Δ10-30, and Δ10-40, indicating the regions of the presequence excised, are shown. Bar, 5 μm. (B) Targeting of GFP using the Cp-mtHSP70 presequence at the NH₂ terminus of GFP (upper panels, GRA1-[pre]-GFP) or no-presequence control (lower panels, GRA1-GFP) using transfected T. gondii RH cultured in human foreskin fibroblast cells. A released single tachyzoite is shown. The mitochondrion is also labeled with MitoTracker Red CM-H₂XRos (MTR). In the upper panels, note targeting of GFP to the mitochondrion in the merged double-labeled image (merge). A composite of the merged image with DIC is shown on the right (DIC). Bar, 3 μm.

Because of the in silico ambiguity for organellar targeting (Table 1), the genetically well-characterized apicomplexan T. gondii was tested by using the C. parvum complete mitochondrial presequence. T. gondii possesses both a mitochondrion and an apicoplast (17). As expected, tachyzoites transfected with pTgGRA1-pre-GFP (C. parvum presequence as an NH₂ terminus of the GFP) correctly target the structurally well-defined T. gondii mitochondrion (33), whereas the plasmid lacking the C. parvum targeting presequence shows a diffuse cytosolic green fluorescence (Fig. 5B). Localization into the single T. gondii mitochondrion was confirmed by double labeling using the mitochondrion-specific dye MitoTracker Red CM-H₂XRos (Fig. 5B, merge). This is the first time T. gondii has been used as a surrogate host for mitochondrial targeting of C. parvum sequences.

Cp-mtHSP70 localizes to the relict mitochondrion of C. parvum.

To further investigate the expression of Cp-mtHSP70, a rabbit anti-Cp-mtHSP70 (see Materials and Methods) was tested by Western blotting. Polyclonal antibodies to a recombinant Cp-mtHSP70 recognized a single band in C. parvum extracts with an approximate molecular mass of 70 kDa (Fig. 2C).

Fluorescein (FITC)-labeled anti-Cp-mtHSP70 serum localized to a single compartment in freshly purified sporozoites of C. parvum (Fig. 6A), whereas preimmune control serum did not (data not shown). In sporozoites, the compartment localized by FITC-anti-Cp-mtHSP70 is in close apposition to the nucleus, as visualized by DAPI counterstaining, and the green fluorescence always appears as a single ovoid spot posterior to the nucleus (Fig. 6A, arrowheads).

FIG. 6.

Fluorescence microscopy localization of Cp-mtHSP70 in homologous mitochondria. (A) Immunofluorescent antibody staining of purified C. parvum sporozoites with FITC-labeled anti-Cp-mtHSP70. The three-panel composite shows FITC, FITC plus DAPI double labeling, and a merged image with DIC. Note the single labeling (white arrowhead) in each of the three sporozoites at the posterior end and near the nucleus. Bar, 2.5 μm. (B) Immunofluorescent antibody staining of 24, 48, and 72 h C. parvum-infected HCT-8 cultured cells labeled with FITC anti-Cp-mtHSP70. Composite DIC and DAPI images are shown. Note the presence of DAPI nuclear staining in close proximity to FITC-labeled compartments. At 48 h, the host cell nucleus (N) is visible in the DAPI+FITC panel. At 72 h, globular nuclear material is indicated by an arrow; FITC labeling is indicated by the large arrowhead, and FITC-satellite labeling is indicated by the small arrowhead. Bar, 2.5 μm.

C. parvum-infected HCT-8 cells cultured on Transwell membranes were processed for immunofluorescence microscopy at 24, 48, and 72 h p.i. with anti-Cp-mtHSP70 (Fig. 6B). At 24 h p.i., most of the intracellular life cycle stages are meronts that contain four merozoites (Fig. 6B, 24 h). Each merozoite shows a distinct DAPI-stained nucleus next to which a FITC-immunofluorescent spot can be seen, and nearly all of the HCT-8 cells were infected with several meronts. At 48 h p.i., C. parvum developmental stages containing 4 to 8 nuclei, and adjacent FITC-labeled oval dots could be observed (Fig. 6B, 48 h). Some developmental stages at 48 h p.i. did not appear to have DAPI-stained dividing nuclei, although FITC-labeled organelles were still apparent. These parasites more nearly resembled stages seen at 72 h p.i. The 72-h p.i. developmental stages were larger and often contained a DAPI-stained sphere to which several FITC-anti-Cp-mtHSP70 ovals were attached (Fig. 6B, 72 h). These stages are probably gametocytes of C. parvum which will eventually produce zygotes upon fertilization and subsequently yield either thin- or thick-walled oocysts. Some developmental stages stained with FITC-anti-Cp-mtHSP70 indicated not only a single round compartment, as seen in C. parvum sporozoites, but also a smaller, less distinct spot (Fig. 6B, arrowheads). Such staining might represent multiplication of both the organism and the organelles. However, the explicit interpretation of these two compartments could not be clearly resolved by conventional microscopy.

Immunoelectron microscopy of sporozoites was also used to determine the subcellular localization of Cp-mtHSP70 (Fig. 7). Previously we have shown that the relict mitochondrion, to which chaperone CpCpn60 has been localized (40), is posterior to the central nucleus in close apposition to the crystalloid body. Here we further confirm the presence of a mitochondrion in C. parvum by showing a full-length sporozoite (Fig. 7A) fixed in osmium and glutaraldehyde and stained with uranyl acetate. The membranes of the apical organelles, including the micronemes, rhoptry, and dense granules, as well as those of the posterior nucleus, relict mitochondrion, and crystalloid body, are clearly delineated. Three formaldehyde- and glutaraldehyde-fixed sections (which do not reveal membranes) embedded in LR White (Fig. 7B) clearly show the localization of 10-nm Cp-mtHSP70-specific immunogold particles to an organelle posterior to the nucleus and next to the crystalloid body. This compartment corresponds both to the relict mitochondrion observed in the epon-embedded section (Fig. 7A) and to that previously described for CpCpn60 (40).

FIG. 7.

Transmission electron microscopy of C. parvum sporozoites showing the relict mitochondrion, its relationship to other organelles, and Cp-mtHSP70-specific immunogold localization within this organelle. (A) Longitudinal section of an epon-embedded freshly excysted sporozoite fixed in buffered 2% glutaraldehyde, with additional fixation in 2% osmium tetroxide and 0.5% uranyl acetate. This fixation clearly shows the organelles and their membranes. The ribosome-studded mitochondrion (*) is between the nucleus (N) and the crystalloid body (CB). The apical organelles shown include the micronemes (M) and dense granules (D) for entry into host enterocytes, as well as the plant-type storage granule amylopectin (A). Bar = 0.5 μm. (B) A series of three representative LR White-embedded sporozoites showing two, six, and three Cp-mtHSP70-specific immunogold particles labeling the relict mitochondrion (*). The localization of the compartment is identical to that seen in the osmium-fixed, epon-embedded sporozoites (Fig. 7A), i.e., the double-membrane-bounded organelle wrapped in ribosomes posterior to the nucleus in close apposition to the crystalloid body and previously identified as the C. parvum relict mitochondrion (40). LR White-embedded sporozoites were fixed in EM-grade methanol-free formaldehyde, with additional fixation in 4% formaldehyde-0.1% EM-grade glutaraldehyde. This fixative does not clearly reveal organellar membranes but is excellent for localization of immunogold antibody particles. Bar, 0.2 μm. Note the identical localization to this organelle using immunofluorescence (see Fig. 6A and B).

DISCUSSION

The 70-kDa heat shock protein family is one that assists in protein folding in a variety of cellular compartments and that is broadly conserved among prokaryotes and eukaryotes. The HSP70 proteins consist of a highly conserved NH₂-terminal 44-kDa ATPase domain and a C-terminal region subdivided into a 15-kDa conserved substrate-binding domain and a 10-kDa putative substrate-stabilization domain that is less well conserved (8). Although Cp-mtHSP70 has several regions with significant similarity to all HSP70 sequences, the greatest scores were obtained to those of proteobacterial and eukaryotic mitochondrial HSP70. Moreover, alignments confirm the presence of two sequence signatures (22) in Cp-mtHSP70: a ¹⁴⁸GDAWV¹⁵² and a ¹⁵⁹YSPSQI¹⁶⁴ motif shared by mitochondria and proteobacteria. The phylogenetic analyses of diverse HSP70 sequences show monophyly of mitochondrial HSP70, including Cp-mtHSP70, within α-proteobacterial DnaK sequences. However, these analyses failed to resolve the relationships between lineages that are either deep branching or fast evolving, consistent with previous HSP70 analyses (25, 35). Importantly, the position of Cp-mtHSP70 within the mitochondrial clade was unambiguous, as was the position of the C. parvum cytosolic form. These data further confirm the presence of a mitochondrion in C. parvum.

As expected, the Cp-mtHSP70 presequence is rich in arginine (total of 3), alanine (total of 2), and serine (total of 7), while acidic amino acids are absent. Like the iron sulfur cluster protein IscS (28), Cp-mtHSP70 has an R-2 motif that is preferred by the mitochondrial matrix processing peptidase (20, 36). This motif is not found in mitochondrial CpCpn60 or CpIscU (28, 40), but these observations are congruent with data showing that only about two-thirds of all targeting peptides have R in position −3 or −2 (20). Although the subcellular localization of Cp-mtHSP70 was predicted in silico to be mitochondrial, the algorithms for in silico predictions have been developed using sequences from organisms that are evolutionarily distant from protists (14). For example, TargetP was unable to identify a presequence in the P. falciparum mtHSP70 homologue even though it clusters unambiguously within mitochondrial HSP70 types. Therefore, in cases like these, in silico results must be viewed with caution.

Initially, we elucidated the in vivo ability of Cp-mtHSP70 presequence to target GFP to heterologous yeast mitochondria. When the Cp-mtHSP70 presequence was attached as an NH₂ extension to a GFP reporter, this construct had the ability to direct the reporter to the mitochondrial network of yeast cells. Moreover, the targeting was as efficient as that for the yeast mtHSP70 homologue (43). Truncation of Cp-mtHSP70 at the extreme NH₂ terminus indicates that residues beyond amino acid 20 of the presequence are essential for mitochondrial targeting, thus supporting the in silico prediction that these amino acids probably do form a bifacial amphiphilic α-helix crucial for mitochondrial import (1, 39). This is consistent with the observation that Δ1-10 constructs were 92% as efficient as the SSC1 control in targeting GFP to yeast mitochondria and that even Δ1-20 and Δ10-20 constructs targeted mitochondria with 79 and 75% efficiency, respectively.

Although yeasts are suitable for demonstrating mitochondrial targeting, as mentioned previously, most Apicomplexa (E. tenella, P. falciparum, and T. gondii) also have a chloroplast remnant—the apicoplast (17). Therefore, T. gondii was used as a surrogate host to test whether the Cp-mtHSP70 presequence would target the single mitochondrion, and not the apicoplast, of this apicomplexan, which contains both organelles. As expected, only the single mitochondrion of T. gondii was targeted by GFP when the C. parvum presequence was used. As mentioned previously, this is the first time T. gondii has been used as a surrogate for organelle targeting in C. parvum. Because there is no transient or stable transfection system for C. parvum, directly targeting the C. parvum relict mitochondrion is precluded.

Nevertheless, evidence for the presence of a relict mitochondrion was obtained using FITC-labeled anti-Cp-mtHSP70. Here, for the first time, immunofluorescence of the mitochondrion was observed both in C. parvum sporozoites and in intracellular stages. These immunofluorescence data indicate not only that a relict mitochondrion is present in sporozoites but that this organelle occurs in developmental stages that are multiplying within host cells. Previously, only immunogold labeling of transmission electron micrograph sections by anti-CpCpn60 was able to clearly delineate the relict mitochondrion (40). Here we also show immunogold localization of anti-Cp-mtHSP70 to a compartment between the nucleus and crystalloid body identical with that for anti-CpCpn60 (40), further confirming that this organelle is a mitochondrion.

Transmission electron micrographs indicate an intimate association of this organelle with the nuclear membrane and the rough ER which yields an appearance of a “ribosome-studded” mitochondrion (41). In many eukaryotes the outer nuclear membrane is continuous with the rough ER, contributing to both protein synthesis and secretion (37). Interestingly, the nuclear envelope of the apicomplexan T. gondii appears to be an intermediate compartment for secretory trafficking from the ER to the Golgi apparatus (23). It is likely that studies of protein trafficking in C. parvum, particularly the association of the relict mitochondrion with the rough ER, will also yield new insights into the unique biology and evolution of this apicomplexan.

As previously suggested, the HSP70 and Cpn60 mitochondrial-type chaperones in C. parvum probably serve as a part of the fundamental elements for the import and maturation of many proteins, including Fe-S clusters (28). Similarly, these mitochondrial chaperones have also been observed in modified mitochondria of other protists, i.e., hydrogenosomes of Trichomonas vaginalis and mitosome/crypton of Entamoeba histolytica and Trachipleistophora hominis (46, 50). Although recent experimental evidence suggests that the secondarily reduced mitochondrial compartment is essential for Fe-S cluster biosynthesis in Giardia (47), whether this is also a critical function of the C. parvum relict mitochondrion is still an open question (19, 28).