Multiple Functionally Redundant Signals Mediate Targeting to the Apicoplast in the Apicomplexan Parasite Toxoplasma gondii

Most species of the protozoan phylum Apicomplexa harbor an endosymbioticorganelle—the apicoplast—acquired when an ancestralparasite engulfed a eukaryotic plastid-containing alga. Severalhundred proteins are encoded in the parasite nucleus and areposttranslationally targeted to the apicoplast by a distinctivebipartite signal. The N-terminal 20 to 30 amino acids of nucleus-encodedapicoplast targeted proteins function as a classical signalsequence, mediating entry into the secretory pathway. Cleavageof the signal sequence exposes a transit peptide of variablelength (50 to 200 amino acids) that is required for directingproteins to the apicoplast. Although these peptides are enrichedin basic amino acids, their structural and functional characteristicsare not well understood, which hampers the identification ofapicoplast proteins that may constitute novel chemotherapeutictargets. To identify functional domains for a model apicoplasttransit peptide, we generated more than 80 deletions and mutationsthroughout the transit peptide of Toxoplasma gondii ferredoxinNADP⁺ reductase (TgFNR) and examined the ability of these alteredtransit peptides to mediate proper targeting and processingof a fluorescent protein reporter. These studies revealed thepresence of numerous functional domains. Processing can takeplace at multiple sites in the protein sequence and may occuroutside of the apicoplast lumen. The TgFNR transit peptide containsat least two independent and functionally redundant targetingsignals, each of which contains a subdomain that is requiredfor release from or proper sorting within the endoplasmic reticulum.Certain deletion constructs traffic to multiple locations, includingthe apicoplast periphery, the rhoptries, and the parasitophorousvacuole, suggesting a common thread for targeting to these specializedcompartments.

INTRODUCTION

Top

Introduction

Plastid organelles are found in a diverse array of eukaryoticorganisms, including the chloroplasts of green algae and plants,the rhodoplasts associated with red algae, and a variety ofplastids in other taxa. For example, apicomplexan parasites,including Plasmodium species (which cause $~$ 500 million casesof human malaria annually, resulting in 2 to 3 million deaths[36, 58]), Toxoplasma gondii (which can lead to severe manifestationsin immunocompromised hosts or unborn fetuses [50]), and Eimeriaspecies (which are responsible for coccidiosis in poultry andother farm animals [19]), contain a plastid organelle calledthe apicoplast. In retrospect, the apicoplast was first notednearly 50 years ago in morphological studies of these parasites(34, 44) and was variously described as the Golgi adjunct, thespherical body, the organelle plurimembranaire, and the Hohlzylinder(45). It has recently become clear that this organelle is aplastid (13, 21, 26, 30) that is essential for parasite survival,raising the possibility that the apicoplast provides a parasite-specifictarget for chemotherapy (4, 14, 17).

All plastids are thought to be monophyletic in origin (31, 35),tracing their ancestry to a single endosymbiotic event involvingthe colonization of a eukaryote by a photosynthetic cyanobacterium.Subsequent divergence led to the establishment of red and greenalgal lineages. Secondary endosymbiotic colonizations of diverselineages by eukaryotic red or green algae have given rise toplastids in the brown algae, euglenoids, cryptomonads, chlorarachniophytes,apicomplexans, etc. (51). All of these plastids share severalcommon features. First, they retain a vestigial genome derivedfrom the cyanobacterial ancestor. The apicoplast genome of T.gondii and Plasmodium falciparum is $~$ 35 kb long and encodes fewerthan 30 proteins (21, 28, 56, 57; http://www.sas.upenn.edu/ $~$ jkissing/toxomap.html).Second, these organelles are surrounded by multiple membranes,typically two for primary plastids, but three or four for secondaryendosymbionts, betraying the complex origins of the latter (13,21, 41). Third, the vast majority of proteins responsible forplastid metabolic functions have been transferred to the nucleargenome during the course of eukaryotic evolution; these proteinsare translated on cytoplasmic ribosomes and must find theirway back into the plastid, using targeting information acquiredduring transfer to the nucleus (10, 21, 27, 41, 43).

In plants and green algae and in glaucocystophyte algae, thecytoplasmic synthesis of proteins destined for the plastid includesan N-terminal domain responsible for targeting mature proteinsto the plastid import machinery and transport across both chloroplastor cyanelle membranes, respectively (reviewed in references9 and 46). Targeting of proteins into complex plastids occursby a variety of mechanisms, all of which require a bipartiteN-terminal extension (reviewed in reference 51). The plastidsof Euglena and dinoflagellates are enclosed within three membranesand acquire their proteins from the endoplasmic reticulum (ER)via the Golgi complex (49). In contrast, diatom and cryptomonadplastids are enclosed by four membranes, the outermost of whichis studded with ribosomes, suggesting that the first step ofprotein import is cotranslational translocation across the outermembrane of these plastids (6).

Apicomplexan plastids are also enclosed by four membranes, andproteins destined for the apicoplast possess a bipartite leadersequence (53). The first 20 to 30 amino acids (aa) of nucleus-encodedapicoplast-targeted (NEAT) proteins function as a classicalsecretory signal sequence (SS). Cleavage by a signal peptidasewithin the ER exposes a 50- to 200-aa transit peptide (TP) thatis responsible for targeting the protein to the organelle (41,54). A detailed characterization of NEAT protein targeting signalsis likely to enhance our understanding of plastid evolutionand organellar targeting and to facilitate the identificationof apicoplast proteins, many of which are attractive as parasiticidaldrug targets that are unlikely to be found in mammalian hostspecies.

Previous studies of the TP of T. gondii RPS9 (ribosomal smallsubunit protein 9) suggested that targeting information is locatedwithin the N-terminal portion of this peptide and that the TPmay be processed multiple times during import into the apicoplast(11, 59). More recent evidence suggests that TgRPS9 containsredundant targeting domains (60). Analyses of NEAT proteinsfrom P. falciparum have revealed the importance of a chargedamino acid distribution in the TP (15), permitting the developmentof computational tools for the detection of candidate apicoplastproteins in malaria parasites (15, 61; http://plasmoDB.org).These tools fail to detect many NEAT proteins in T. gondii,however, despite proper targeting of P. falciparum NEAT proteinsin transgenic T. gondii and vice versa (unpublished observations).In summary, precisely which elements of the TP are necessaryand sufficient for targeting and translocation into the apicoplastremains unclear.

We have undertaken a systematic study of TP function, focusingon the NEAT protein ferredoxin NADP⁺ reductase (FNR) from T.gondii because the leader of this protein (SS plus TP) is relativelyshort, easily manipulated, and functions in both T. gondii andP. falciparum. Our results indicate that the TgFNR TP is processedat least twice and that processing is required for translocationinto the lumen of the apicoplast. Furthermore, this peptideincludes two functionally redundant TP domains, each containingsubdomains that facilitate exit from the ER. Several constructstarget proteins to multiple locations—the apicoplast periphery,rhoptries, and parasitophorous vacuoles—suggesting a possiblelink for targeting to these different subcellular compartments.

MATERIALS AND METHODS

Top

Materials and Methods

Plasmids and deletion constructs. The TP of TgFNR (52) was subcloned into the BglII and AvrIIsites of the T. gondii expression vector ptubYFP/sagCAT (47)to produce ptubFNR-YFP/sagCAT. High-fidelity PCR (with Herculaseenhanced DNA polymerase; Stratagene) was utilized to generatethe various deletion constructs presented in this study, usingthe primers listed in Table 1; the resulting constructs werenamed according to the predicted FNR peptide and are referredto by their corresponding numbers in Table 1. To generate sequential5-aa deletions from the C terminus of the TP, we subcloned PCRproducts, using sense primer 1 and one of the antisense primers(2 to 25), between the BglII and AvrII sites of ptubYFP/sagCAT.FNR-34 was generated by using primers 1 and 26. Constructs designedto define the FNR SS were generated by using primers 1 and 27,28, or 29. Deletion of the FNR SS was achieved by utilizingprimers 30 (containing a BglII site and an ATG initiation codon)and 31. For the generation of FNR ${Delta}$ 26-50, FNR ${Delta}$ 26-82, and FNR ${Delta}$ 26-86,primer 33 was utilized in combination with primers 34, 35, and36, respectively. The 5' end of primer 33 was designed to containnucleotide substitutions that generated a unique restriction(BsrGI) site without affecting the amino acid sequence. Primers34 and 35 each contained a BsrGI site at their 5' ends to facilitatecloning. Deletions designed to define the second FNR targetingdomain were generated by utilizing primer 1 and primes 12 to17, with either FNR ${Delta}$ 26-50 or FNR ${Delta}$ 26-82 as a template (see Results).The deletion of aa 27 to 30 was achieved with primers 37 and38, while the deletion of aa 83 to 86 was achieved with primers39 and 40. In order to assess the number of 25-aa targetingsubdomains present within the FNR TP, we used primer 41 in conjunctionwith each of the primers 42 to 61, followed by blunt-end ligation.The template for each of these reactions varied based on which25-aa stretch was being generated. For example, to generatethe construct FNR ${Delta}$ 26-30;56-150, we combined primer 41 with primer42 in a PCR, with the FNR55 construct as the template.

View this table:
[in this window]
[in a new window]
TABLE 1. PCR primers used for this study

Cell culture and fluorescence microscopy. T. gondii tachyzoites (RH strain) were maintained by serialpassaging in human foreskin fibroblast (HFF) cells as previouslydescribed (42). Uninfected HFF monolayers were maintained exactlyas previously described (42). The transfection of T. gondiiwas achieved by electroporation of 50 µg of DNA into 10⁷freshly harvested parasite tachyzoites as previously described(42). Parasites were inoculated onto confluent HFF monolayerson 22-mm-diameter glass coverslips in six-well plates and incubatedfor 24 or 48 h prior to processing.

For direct visualization of yellow fluorescent protein (YFP),infected coverslips were inverted onto glass slides and examinedimmediately. For indirect immunofluorescence, coverslips werefixed for 10 min in 4% paraformaldehyde, permeabilized for 10min in 0.25% Triton X-100, blocked for 1 h in phosphate-bufferedsaline (PBS) (pH 7.4) plus 3% bovine serum albumin fractionV (Fisher), incubated for 1 h with a primary antibody (in blockingsolution), washed, and incubated for 1 h with a secondary antibody.The primary antibodies used were as follows: rabbit polyclonalanti-acyl carrier protein (ACP; diluted 1:1,000) (53), mousemonoclonal anti-green fluorescent protein (GFP) (Clontech; 1:1,000),and mouse monoclonal anti-ROP2/3/4 (1:3,000) (3, 24). The followingwere used as secondary antibodies: Alexa 488-conjugated goatanti-mouse (Molecular Probes; 1:500) and Alexa 594-conjugatedgoat anti-rabbit (Molecular Probes; 1:500). Finally, apicoplastand nuclear DNAs were stained with 4',6'-diamidino-2-phenylindole(DAPI) for 5 min (in PBS), and samples were washed with PBSand mounted on glass slides by the use of Fluoromount-G (SouthernBiotechnology Associates, Inc.) for examination under eithera Zeiss Axiovert 35 microscope equipped with a heated stage(Biopteks) or a Leica DM IRBE microscope equipped with a motorizedfilter wheel. Both inverted microscopes were equipped with a100-W Hg-vapor lamp and an Orca-ER digital camera (Hamamatsu).Images were captured with Openlab 3.1 software (Improvision),and serial 0.1-µm-thick sections were subjected to iterativedeconvolution and three-dimensional reconstruction with Volocity2.0 software (Improvision).

Protein analysis. Parasites were harvested shortly after monolayer lysis, filteredthrough 3-µm-pore-size polycarbonate filters (Nuclepore),pelleted at 900 x g, and resuspended in PBS at 10⁶ parasites/µl.NuPAGE LDS sample buffer (Invitrogen) and 0.5 M dithiothreitolwere added to each sample, followed by denaturation at 70°Cfor 10 min. Samples of $~$ 10⁷ parasites were loaded into precastbis-Tris-4 to 12% polyacrylamide gels (Invitrogen) and wererun in NuPAGE morpholineethanesulfonic acid-sodium dodecyl sulfaterunning buffer at 200 V for 45 min. The transfer of proteinsto nitrocellulose was performed by using a Trans-Blot SemiDryapparatus (Bio-Rad) according to the manufacturer's instructions.After the transfer, the membrane was blocked for 1 h in PBSplus 5% nonfat dry milk, incubated for 1 h with a mouse anti-GFPantibody (1:1,000 in blocking solution) plus 0.2% Tween 20 (Sigma),washed twice in PBS plus Tween 20, and then incubated for 1h with a horseradish peroxidase-conjugated goat anti-mouse secondaryantibody (1:2,500) (Sigma) plus Tween 20. After further washesin PBS plus Tween 20, a chemiluminescence reaction was performedby using ECL Western blotting detection reagents according tothe manufacturer's (Amersham Biosciences) instructions, andblots were exposed on Kodak BioMax film.

RESULTS

Top

Results

Defining the boundaries of the bipartite FNR leader sequence (SS plus TP). Based on multiple sequence alignments of FNR proteins from variousspecies (52), the T. gondii FNR leader sequence is predictedto be $~$ 150 aa long. The fusion of either the full-length TgFNRgene (18) or the complete TgFNR leader (TgFNR-150, includingboth the presumed SS and the TP) (Fig. 1a) to a YFP reportertargets YFP to the apicoplast in T. gondii (Fig. 1b). The processingprofile for this fusion construct revealed that the mature form(Fig. 2b, lane 1) was slightly larger than the $~$ 27-kDa molecularmass expected for YFP alone (arrowhead), suggesting that cleavageoccurred upstream of the reporter protein. The removal of fiveor six additional amino acids from the C terminus of the FNRpeptide yielded a processed product of close to 27 kDa (TgFNR-145[Fig. 2b, lane 2] and TgFNR-144 [data not shown]), but furtherdeletions resulted in alternative processing (TgFNR-140 [Fig.2b, lane 3]; similar results were obtained for TgFNR-141 to-143 [not shown]). N-terminal sequencing of the purified proteinrevealed that the mature N terminus of FNR is Ala143.

View larger version (59K):
[in this window]
[in a new window]
FIG. 1. The bipartite apicoplast targeting signal. (a) Diagram representing the constructs used for the experiments shown in panels b, c, and d. The secretory signal is shown in blue, the plastid TP is in green, and YFP is in orange. (b) The fusion of the first 150 aa of FNR to YFP results in its targeting to the apicoplast, seen as oval structures in the apical juxtanuclear region of transfected T. gondii tachyzoites. (c) Deletion constructs containing only the FNR SS fused to YFP (FNR-31, –26, –24, and –22) result in secretion via dense granules (small dots within the parasites) into the parasitophorous vacuole surrounding the parasites. (d) Deletion of the SS (FNR- ${Delta}$ 1-24) results in YFP expression within the parasite cytoplasm only. FNR constructs diagrammed at the top are shown to scale, with summaries of the targeting results shown to the right. AL, apicoplast lumen; AP, apicoplast periphery; C, cytoplasmic; ER, endoplasmic reticulum; R, rhoptry; S, secreted; X, complex localization. Panels B, C, and D are representative images showing direct YFP fluorescence in live parasites. Bars = 5 µm.

View larger version (65K):
[in this window]
[in a new window]
FIG. 2. C-terminal deletions of the TP. (a) Diagram of FNR constructs depicting sequential 5-aa deletions between aa 105 and 150. (b) Western blot of deletion constructs FNR-150 through FNR-110, illustrating mature and processed forms of each fusion construct; note the alternative processing site in constructs shorter than TgFNR-145. Inefficient processing has been previously described (53). (c) Deconvolved fluorescence image of TgFNR-105 (representative of all constructs shown in panel a) illustrating targeting into the apicoplast lumen. The abbreviations are as described in the legend to Fig. 1. For microscopic images, green = YFP (anti-GFP), red = apicoplast (anti-ACP [53]), blue = nuclear and apicoplast DNA (DAPI [42]). A video of the images shown in panel c is available for download (http://roos.bio.upenn.edu/ $~$ oharb/Harb2c.mov).

In order to define the beginning (N terminus) of the TP domain,we identified potential SSs by using a variety of computationalalgorithms. The neural network-based algorithm SignalP-NN (33;http://www.cbs.dtu.dk/services/SignalP/) predicted a eukaryoticsignal cleavage site between aa 29 and 30; the SignalP hiddenMarkov model predicted cleavage between aa 21 and 22, whilePSORT (32; http://psort.ims.u-tokyo.ac.jp/) predicted cleavagebetween aa 23 and 24. To determine which prediction was accurate,we generated several deletion constructs (Fig. 1a) and examinedtheir ability to direct YFP to the secretory pathway. TgFNR-31,TgFNR-26, TgFNR-24, and TgFNR-22 were all sufficient for YFPsecretion (Fig. 1c), while shorter fragments (e.g., TgFNR-20)were unable to direct secretion (not shown). The sizes of theproteins observed on Western blots also suggested that cleavageoccurs close to aa 21, although we have not been able to purifysufficient amounts of incompletely processed protein to directlyexamine the signal peptidase cleavage site. Amino acids 1 to21 constitute a typical eukaryotic SS (33), containing a positivelycharged N-terminal region, a 10-aa h-region, and small residuesat the –1 and –3 positions relative to the cleavagesite.

The deletion of aa 1 to 24 (Fig. 1a) abolished entry into thesecretory pathway, resulting in the cytosolic expression ofYFP (Fig. 1d), as has previously been observed for the NEATprotein ACP in both P. falciparum and T. gondii (54). Studiesof the NEAT protein TgRPS9 have shown that deletion of the SScan result in the mistargeting of GFP to the mitochondria (11,59), but this was not observed for TgFNR. The replacement ofthe TgFNR SS with an alternative SS derived from the secretedprotein TgGRA8 (8) restored the targeting of YFP to the apicoplast(data not shown).

The C-terminal 45 aa of the TgFNR TP are not required for apicoplast targeting or translocation. Previous deletion studies of the TPs of T. gondii NEAT proteinshave been unsuccessful in clearly delineating specific domainsinvolved in targeting to, or translocation into, the apicoplast(11, 59), because truncated proteins are often expressed poorlyor are retained within the ER (presumably due to misfolding).Fortunately, this appears not to be the case for FNR proteindeletions. In order to identify regions of the TgFNR TP thataffect processing, targeting, and/or translocation, we generatednested 5-aa deletions from the C terminus of the 150-aa leaderdomain, as shown in Fig. 2.

Progressive deletions, from TgFNR-150 to TgFNR-105 (Fig. 2a),resulted in progressively smaller unprocessed preproteins (Fig.2b, upper bands). Alternative processing was observed beyondTgFNR-144, however, as indicated by the increased size of theprocessed (lower) bands in Fig. 2b (note that the processingof the TP was inefficient, as has previously been described[53]). This observation is consistent with an intermediate orcryptic processing site between aa 100 and 105 (multiple processingsites have also been reported for TgRPS9 [59]). Despite thealternative processing of the TP domain, deletions down to TgFNR-105had no apparent effect on the targeting of YFP to the apicoplast(Fig. 2c), consistent with previous results suggesting thatthe N-terminal portion of the TP is more important than theC-terminal portion for apicoplast targeting (11, 59). Theseresults also demonstrate that processing at the native TP cleavagesite is not necessary for import.

Separating targeting to the apicoplast from translocation into the organellar lumen. Further nested deletions down to TgFNR-40 (Fig. 3a) permittedthe targeting of YFP to the region of the apicoplast, but notinto the apicoplast lumen (Fig. 3c). This phenotype coincideswith the elimination of the second cleavage site (Fig. 3b, lanes6 to 11), suggesting that cleavage of the TP may be requiredfor import into the apicoplast lumen but not for targeting tothe organelle periphery. In summary, the region from aa $~$ 23 to40 contains sufficient information for apicoplast targeting(when combined with a SS).

View larger version (59K):
[in this window]
[in a new window]
FIG. 3. C-terminal deletions of the TP delineate between targeting and translocation. (a) Diagram of FNR constructs depicting sequential 5-aa deletions between aa 40 and 105. (b) Western blot of deletion constructs down to TgFNR-40. (c) Deconvolved fluorescence image of TgFNR-50 (representative of all constructs shown in panel a) illustrating targeting to, but not into, the apicoplast. The abbreviations are as described in the legend to Fig. 1. For microscopic images, green = YFP (anti-GFP), red = apicoplast (anti-ACP [53]), blue = nuclear and apicoplast DNA (DAPI [42]). Bars = 5 µm. A video of the images shown in panel c is available for download (http://roos.bio.upenn.edu/ $~$ oharb/Harb3c.mov).

Reducing this minimal TP by 6 aa (TgFNR-34) (Fig. 4a) resultedin the simultaneous targeting of YFP to the apicoplast periphery,the rhoptries, and the parasitophorous vacuole (Fig. 4b andc). Moreover, YFP fluorescence never colocalized with densegranule markers (not shown). These results suggest that proteinsdestined for the apicoplast and rhoptries may share a commontargeting pathway beyond the obvious fact that both pass throughthe ER. However, further experiments to confirm this observationare required. Interestingly, a similar phenotype was observedfor several other deletion constructs (see below).

View larger version (74K):
[in this window]
[in a new window]
FIG. 4. TgFNR-34 targets YFP to multiple compartments. (a) Diagram of TgFNR-34 and -150. (b) Direct YFP fluorescence image of TgFNR-34, showing complex targeting pattern within the parasite and secretion into the parasitophorous vacuole. (c) Fluorescence image of TgFNR-34 (fixed specimen), showing complex targeting pattern within the parasite, including colocalization with rhoptries. The insert (bottom right) is an enlarged view of the apical region of one of the parasites showing green fluorescence at the periphery of the apicoplast and colocalizing with the rhoptries. The abbreviations are as described in the legend to Fig. 1. For microscopic images, green = YFP (anti-GFP or native fluorescence in panel b), red = rhoptries (anti-ROP2/3/4 [3]), blue = nuclear and apicoplast DNA (DAPI [42]). Bars = 5 µm.

A second apicoplast targeting domain in the TgFNR TP. In order to assess the importance of the minimal TP definedabove for apicoplast targeting, a deletion construct lackingaa 26 to 50 of the FNR leader was generated, as shown in Fig.5a (TgFNR- ${Delta}$ 26-50). This construct remained able to traffic tothe apicoplast, indicating the presence of redundant targetingsignals (data not shown). A series of further deletions fromthe N terminus of the TP were therefore engineered (Fig. 5a);of these, only TgFNR- ${Delta}$ 26-82 and TgFNR- ${Delta}$ 26-86 failed to fully translocateinto the apicoplast lumen. A microscopic evaluation indicatedthat TgFNR- ${Delta}$ 26-82 was targeted to the apicoplast periphery only(Fig. 5c), but this protein was processed nevertheless (Fig.5b, lane 2), suggesting that TP maturation may not require completetranslocation into the organellar lumen (see Discussion). Furtherdeletion (TgFNR- ${Delta}$ 26-86) resulted in a combination of targetingto the periphery of the apicoplast and retention within theER (Fig. 5d), and Western blots showed no evidence of processingbeyond SS cleavage (Fig. 5b, lane 3).

View larger version (42K):
[in this window]
[in a new window]
FIG. 5. Redundant motif within apicoplast targeting signal. (a) Diagram depicting deletions that define a redundant domain with the FNR TP. (b) Western blot demonstrating processing of TgFNR- ${Delta}$ 26-82, but not TgFNR- ${Delta}$ 26-86. (c) Despite deletion of one apicoplast targeting signal, TgFNR- ${Delta}$ 26-82 still localizes to the organelle periphery. (d) Deletion of an additional 4 aa results in targeting to both the ER and apicoplast periphery (TgFNR- ${Delta}$ 26-86). The abbreviations are as described in the legend to Fig. 1. For microscopic images, green = YFP (anti-GFP), red = apicoplast (anti-ACP [53]), blue = nuclear and apicoplast DNA (DAPI [42]). Bars = 5 µm.

To identify potential motifs that may be associated with apicoplasttargeting signals, we compared patterns detected by the Teiresiasalgorithm (http://cbcsrv.watson.ibm.com/Tspd.html) with theobserved targeting patterns. Val-(Val/Ser)-Ser-Phe occurs atboth aa 27 to 30 and 83 to 86, which are closely linked to theredundant apicoplast targeting signals mapped above. The deletionof either or both motifs (Fig. 5a) had minor effects on theprocessing intermediates (not shown), but these mutations failedto affect apicoplast targeting. However, these motifs may beassociated with release from the ER (see below).

The second apicoplast targeting signal was mapped more preciselyvia C-terminal deletions of the TP domain in the TgFNR- ${Delta}$ 26-50and TgFNR- ${Delta}$ 26-82 backgrounds, as shown in Fig. 6a. The deletionof aa 106 to 150 had no effect on proper protein trafficking(in either background), but further deletions resulted in increasinglycompromised targeting to the apicoplast, as follows. A ${Delta}$ 101-150mutation in the TgFNR- ${Delta}$ 26-82 background produced targeting tothe apicoplast periphery, a ${Delta}$ 96-150 mutation produced stainingin the apicoplast periphery and the ER, and a ${Delta}$ 91-150 mutationyielded a secretory protein (Fig. 6c). Similarly, in the TgFNR- ${Delta}$ 26-50background, ${Delta}$ 96-150 and ${Delta}$ 91-150 mutations yielded localizationto the apicoplast periphery (Fig. 6b), while a ${Delta}$ 81-150 mutationyielded a complex pattern of localization to the apicoplastperiphery, the rhoptries, and the parasitophorous vacuole, asnoted previously for extreme deletions of the upstream apicoplasttargeting signal (TgFNR-34; Fig. 4b and c).

View larger version (82K):
[in this window]
[in a new window]
FIG. 6. Mapping the second redundant targeting motif within the apicoplast targeting signal. (a) Diagram depicting various deletions generated to map the second redundant targeting motif within the FNR TP. (b) TgFNR- ${Delta}$ 26-50,91-150 is targeted to the apicoplast periphery. (c) TgFNR- ${Delta}$ 26-82,91-150 is secreted into the parasitophorous vacuole. (d and e) Deleting both of the mapped apicoplast targeting signals results in a complex pattern of protein secretion targeting TgFNR- ${Delta}$ 26-50,83-105 to the apicoplast periphery (d), rhoptries (e), and the parasitophorous vacuole (not shown). The abbreviations are as described in the legend to Fig. 1. For microscopic images, green = YFP (anti-GFP or native fluorescence [c]), red = apicoplast (anti-ACP) or rhoptries (anti-ROP2/3/4 [e]). Bars = 5 µm.

In summary, aa 20 to 40 and 82 to 105 can each function as anindependent signal for targeting to (but not into) the apicoplast.Remarkably, a deletion construct lacking both of these domains(TgFNR- ${Delta}$ 26-50, ${Delta}$ 83-105) was still able to target YFP to the apicoplastperiphery (Fig. 6d) in addition to the rhoptries (Fig. 6e) andthe parasitophorous vacuole (not shown). This protein appearedunable to reach the apicoplast lumen, however, despite the retentionof the native protein-processing site, and neither aa 51 to82 nor aa 106 to 150 alone were able to function as an apicoplasttargeting signal.

To more directly test whether any other linear spans are capableof mediating apicoplast targeting, we generated peptide sequencesof 25 aa in length (sufficient to encompass either of the plastidtargeting domains defined above) every 5 aa throughout the TPand sandwiched them between the TgFNR SS and the YFP reporter(Fig. 7a). As shown in Fig. 7, four phenotypic groups were definedbased on the subcellular localization of YFP: (i) targetingto the apicoplast, associated with the two linear domains identifiedabove (Fig. 7b); (ii) retention within the ER, correlated withthe deletion of the V[V/S]SF motif from both targeting domains(Fig. 7c); (iii) secretion into the parasitophorous vacuolevia dense granules (Fig. 7d); and (iv) complex localizationwithin multiple secretory pathway compartments, including theapicoplast periphery, the rhoptries, and the parasitophorousvacuole (Fig. 7e).

View larger version (66K):
[in this window]
[in a new window]
FIG. 7. Scanning the apicoplast TP. (a) Diagram of overlapping sequential 25-aa segments scanning through the TgFNR TP inserted downstream of the SS and upstream of a YFP reporter. Four targeting patterns were observed, labeling the apicoplast (b), ER (c), parasitophorous vacuole (d), or rhoptries and parasitophorous vacuole (e). The abbreviations are as described in the legend to Fig. 1. All micrographs are direct fluorescence images of living parasites. Bars = 5 µm.

DISCUSSION

Top

Discussion

The identification of nucleus-encoded apicoplast proteins offersthe potential for understanding the function and evolution ofthis organelle, and secondary endosymbiotic plastids in general(40). In addition, the ability to recognize NEAT proteins inthe genomes of apicomplexan parasites suggests possible targetsfor parasite-specific chemotherapy (29). Signal peptide recognitionalgorithms perform acceptably in recognizing the first partof the bipartite apicoplast targeting signal. However, despiteadvances that facilitated the identification of NEAT proteinsin the genome of P. falciparum (15, 61), these algorithms remainimperfect, particularly across species boundaries, in contrastto the sequences themselves, which function surprisingly wellacross large phylogenetic distances.

In order to facilitate the identification of apicoplast targetingsignals, we performed a careful analysis of the T. gondii FNRTP, expressing a large series of deletion mutants in transgenicparasites and using fluorescent protein reporters to resolvesubcellular localization. The results of these studies are summarizedin Fig. 8. Deconvolution of high-resolution wide-field microscopicimages (Fig. 2C and 3C) enabled targeting to the apicoplastperiphery (designated AP) to be clearly distinguished from translocationinto the organellar lumen (designated AL). Further analysiswill be required to determine if these deletion constructs constitutetargeting and/or translocation intermediates.

View larger version (39K):
[in this window]
[in a new window]
FIG. 8. Summary of signal features within the TgFNR leader. The secretory SS is shown in magenta, and redundant apicoplast targeting domains are shown in green. Computationally predicted and experimentally determined cleavage sites are indicated by black and blue text, respectively. Red highlighting indicates the V-(V/S)-S-F motif that may be associated with release from the ER. The mature FNR gene is indicated, with conserved amino acids shown in red.

The elimination of the native TP processing site had no effecton luminal targeting, perhaps because of the presence of additional(or cryptic) processing sites (Fig. 2B), but further deletionof this second processing site (Fig. 3B) resulted in targetingto the periphery only. Targeting to the apicoplast peripherydoes not preclude TP processing, however (Fig. 5B). These resultssuggest that TP processing may be necessary for release intothe apicoplast lumen, but translocation is not essential forprocessing.

The processing of chloroplast TPs is carried out by metal-bindingstromal endopeptidases, and multiple processing sites have previouslybeen described in plants and algae (20, 39, 48), but a consensuscleavage motif remains elusive, possibly because several chloroplastpeptidases may recognize different cleavage sites (12, 22).The role of TP cleavage in plastid import is not clearly understood,but the down-regulation of a chloroplast-processing enzyme byantisense technology has been correlated with an inhibitionof protein import into the chloroplast (55), consistent withour observation that the abolition of processing is correlatedwith protein localization to the apicoplast periphery.

Two functionally redundant apicoplast targeting signals havebeen identified; both are $~$ 20 aa in length, and a series of scanningmutations (Fig. 7) argues that these are the only two linearepitopes within the FNR TP that are capable of mediating targetingto the apicoplast. The TPs of plant and algal chloroplasts arecomposed of multiple domains (2, 37, 38), each of which maybe involved in distinct steps of the targeting and import process.An additive effect of redundant targeting sequences has beenreported for the mitochondrial F1-ATPase beta subunit precursor(5). Functional redundancy in the TP may also reflect the evolutionaryhistory of these sequences, as domain shuffling has been shownto facilitate the acquisition of targeting information (7, 16,25).

Each of the two apicoplast targeting domains appears to containfunctional microdomains. The N-terminal region is implicatedin release from the ER (perhaps attributable to the consensusmotif V[V/S]SF). C-terminal truncations appear to lower thefidelity of apicoplast targeting, resulting in mislocalizationto the rhoptries and the parasitophorous vacuole in additionto apicoplast targeting. Other mutations within the TP alsoproduce this complex targeting pattern (designated "X"), suggestingthat targeting to these organelles branches off of a commonpathway.

Our findings on FNR are consistent with previous studies onthe T. gondii NEAT protein TgRPS9, which suggested the presenceof multiple processing sites and more than one apicoplast targetingpeptide (59, 60). These results are also consistent with indicationsthat the N-terminal portion of the TP is particularly important(59), although this region is not particularly basic or enrichedin lysine or asparagine, as found in P. falciparum (15). Proteintargeting into the apicoplast shares with mitochondria the requirementfor an N-terminal protein extension, but neither the FNR TPnor the ACP TP mediates mitochondrial targeting (Fig. 1). Theobservation that TgRPS9 SS deletions target to the mitochondria(11, 59) may be related to the high frequency with which randomlygenerated sequences function as mitochondrial targeting domains(1, 23).

Overall, NEAT protein sequences appear to be organized in aremarkably linear fashion. The extreme N terminus provides theSS (magenta segment in Fig. 8) required for entry into the secretorypathway itself (S in Fig. 1, 6, and 7); the removal of thisdomain results in cytosolic localization (Fig. 1, C). The penultimatedomain (green) is required for targeting to the apicoplast periphery(AP); the N-terminal portion (red) is associated with releasefrom the ER, while the C-terminal portion is required for efficientapicoplast targeting since truncations in this region resultin complex targeting (X). Finally, proteolytic cleavage sites(blue) are required for proper entry into the apicoplast lumen(AL). The ability to recognize apicoplast targeting sequences,combined with the rapidly expanding genome and expressed sequencetag databases for apicomplexan parasites, should facilitatestudies of the evolution of complex plastids, the relationshipsbetween apicomplexan parasites, the biological function(s) ofthe apicoplast, and the potential of this organelle as a parasiticidaldrug target.

ACKNOWLEDGMENTS

We acknowledge Boris Striepen and other past and present membersof the Roos lab for helpful comments and Paul Warfel for editingthe supplemental videos. We also thank Frank Seeber for theoriginal FNR construct, Con Beckers for the anti-ROP2/3/4 antibody,and Geoff McFadden for the anti-ACP antibody.

O.S.H. and M.J.C. are supported by National Research ServiceAwards F32 AI10654 and F32 AI10482, and this research was supportedby grants from the National Institutes of Health to D.S.R.,who is also a Burroughs-Wellcome Scholar in Molecular Parasitologyand an Ellison Foundation Senior Scholar in Global InfectiousDiseases.

FOOTNOTES

* Corresponding author Mailing address: Department of Biology, University of Pennsylvania, Philadelphia, PA 19104-6018. Phone: (215) 898-2118. Fax: (215) 746-6697. E-mail: droos{at}sas.upenn.edu.

Present address: The Rockefeller University, New York, NY 10021.

REFERENCES

Top