Molecular & Biochemical Parasitology 118 (2001) 11–21

Analysis of apicoplast targeting and transit peptide processing inToxoplasma gondii by deletional and insertional mutagenesis

Sunny Yung, Thomas R. Unnasch *, Naomi Lang-UnnaschDi�ision of Geographic Medicine, BBRB 203, Uni�ersity of Alabama at Birmingham, 1530 3rd A�enue South, Birmingham,

AL 35294-2170, USA

Received 22 June 2001; accepted in revised form 14 August 2001

Abstract

Deletion and insertion mutagenesis was used to analyze the targeting sequence of the nuclear encoded apicoplast protein, theribosomal protein small subunit 9 of Toxoplasma gondii. Previous studies have shown that nuclear encoded apicoplast proteinspossess bipartite leaders having characteristic signal sequences followed by serine/threonine rich transit sequences. Deletionanalysis demonstrated that the first 55 amino acids of the rps9 leader were sufficient for apicoplast targeting. Insertionalmutagenesis tagging the leader sequence with a hemagglutinin (HA) tag was used to study the events involved in the targetingpathway. Transfectants with insertions near the N-terminus of the transit displayed HA tagged precursors outside of theapicoplast, in the perinuclear region. In contrast, transfectants with the HA tag inserted near the carboxyl end of the transit-likeregion had apicoplast labeling. Western blot analysis of HA tagged stable isolates suggested that processing of the HA taggedleaders was a multi-step process, with processing occurring both outside of and at or within the apicoplast. © 2001 ElsevierScience B.V. All rights reserved.

Keywords: Toxoplasma gondii ; Apicoplast; Protein targeting; Ribosomal protein

www.parasitology-online.com.

1. Introduction

The apicomplexan apicoplast is a non-photosyntheticplastid that is essential for cell survival [1]. First iden-tified in Plasmodium and Toxoplasma [2–4], the api-coplast now appears widespread throughout thephylum [5]. Ultrastructural and molecular data suggestthis organelle originated by secondary endosymbiosis,whereby a heterotrophic eukaryote engulfed and re-tained a photosynthetic eukaryote, resulting in a plastidwith more than two membranes [6]. Electron mi-croscopy studies have shown the apicoplast in Toxo-plasma gondii to be surrounded by four membranes [4],two inner membranes from the plastid membranes ofthe primary endosymbiont, a third membrane hypothe-

sized to be the plasma membrane of the secondaryendosymbiont and an outermost membrane that ap-pears to be connected to the host endomembrane sys-tem. Protein trafficking to secondary or complexplastids is complicated by these extra membranes. Thus,nuclear encoded proteins destined for the complex plas-tid require a signal peptide to direct the product to theendomembrane system and a transit peptide to directthe protein into the plastid, unlike plastid-targetedproducts in higher plants, which require only a transitpeptide [7].

The model of complex plastid targeting [8] proposesthat a signal sequence initially directs a nuclear encodedcomplex plastid protein to the endoplasmic reticulum(ER). Once in the ER, the signal sequence is cleaved,exposing the putative transit domain. The transit pep-tide then traffics the pre-protein to the plastid via theER, golgi, and vesicular trafficking pathways. Uponentry into the complex plastid, the putative transitsequence is processed, producing the mature protein.However, previous studies have only used reporter

Abbre�iations: GFP, green fluorescent protein; HA tag, hemagglu-tinin tag; rps9, ribosomal protein small subunit 9.

* Corresponding author. Tel.: +1-205-975-7601; fax: +1-205-933-5671.

E-mail address: trunnasch@geomed.dom.uab.edu (T.R. Unnasch).

0166-6851/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 6 -6851 (01 )00359 -0

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–2112

genes tagged with apicoplast leaders to study this pro-cess. As these studies only identified the final locationof the targeted reporter protein, they do not providemuch insight into the pathway involved in the targetingprocess itself.

The ribosomal small subunit 9 (rps9) in T. gondiiis a nuclear encoded protein targeted to the apicoplast.The rps9 gene, absent in the apicoplast 35 kbgenome, was recovered from polyA+ cDNA. Its ge-nomic sequence harbors spliceosomal intronsindicative of nuclear encoded genes and encodes abipartite leader characteristic of nuclear encoded api-coplast proteins. Antibodies to rps9 detected theprotein in the apicoplast by immunofluorescence assay[6]. We have employed deletion analyses to map theimportant domains involved in targeting nuclear en-coded proteins to the apicoplast. Our resultssuggest that redundant signals exist within the leadersequence. We have also employed constructs with epi-tope tags inserted within the leader to explore thepathway involved in the targeting process. These resultssuggest that the targeting pathway does proceedthrough the ER as predicted by the model, and thatprocessing of the leader is a multi-step process, withprocessing steps occurring both outside and at orwithin the apicoplast.

2. Materials and methods

2.1. Cell culture and parasite growth

The RH strains with an HXGPRT knock-out mutant(NIH AIDS reference and Reagent Repository,Bethesda MD, http://www.niaid.nih.gov/reagent) weremaintained by serial passage in human foreskin fibrob-lasts (HFF) or Vero cells. Host cells were grown inDulbecco’s modified eagle medium (Gibco BRL) sup-plemented with 5% fetal bovine serum (FBS, Collabo-rative Biomedical Products).

2.2. Chemicals, antibodies, stains, enzymes

Mycophenolic acids, xanthine, and 4�,6�-diamidino-2-phenylindole (DAPI) were purchased from SigmaChemical Corporation (St. Louis, MO). Rabbit anti-bodies to green fluorescence protein (GFP) were a giftfrom Pamela Silver [9]. Anti-GFP monoclonal antibod-ies and MitoTracker Red CMXRos were obtained fromMolecular Probes (Eugene, OR). Antiserum to hemag-glutinin tag (HA tag) was purchased from Roche (Indi-anapolis, IN). Antiserum to the ER chaperone BiP waskindly provided by Dr. Jay Bangs [10]. Restrictionenzymes were purchased either from Promega(Madison, WI), or New England Biolabs (Beverly,MA).

2.3. GFP expression plasmids

The parental pGRA-GFP plasmid [11] was createdfrom GRA1/GFP5/GRA2-SK, which was kindly pro-vided by Dr Kami Kim [12]. The plasmid pGFA-GFPcontains the m-gfp5 gene from the thermostable foldingmutant GFP5 [13]. The pGRA-GFP vector has thehxgprt minigene as a selectable marker [14].

The rps9 leader sequence, which includes the putativesignal and transit sequence of rps9, was amplified froma clone with the full rps9 cDNA sequence using Taqpolymerase (Promega) and 30 cycles of amplificationwith primers rps9.F1-Nsi1 (sense) and rps9.R156-Nsi1(antisense) (Table 1). The resulting PCR product wasligated into the pCR2.1-TOPO vector (Invitrogen),transformed, and its sequence confirmed. The rps9(1–156)-TOPO plasmid having the correct rps9 leader se-quence was digested with NsiI and cloned into a NsiI,digested and phosphatase treated pGRA-GFP. A simi-lar procedure was used to create all deletion constructs.Sequences of all primers are listed in Table 1.

2.4. Hemagluttinin tag expression plasmids

For each construct, the HA tag was inserted at adifferent region of the putative rps9 transit sequence.The HA tag was constructed by annealing oligosHA.linker1 with HA.linker2 (Table 1) at 260 �M con-centration in 50 mM NaCl, 10 mM Tris pH 8.0, 1 mMEDTA. The annealing reaction was heated to 94 °C for2 min and gradually cooled to create a double strandHA linker with ends compatible with a ClaI and SpeIdigested vector.

A PCR-mediated, linker-scanning mutagenesismethod [15] was used to create the HA tag constructs.The rps9(1–156) insert was ligated to an EcoR1 cutand phosphatase treated pGEM3Z vector (Promega).The resulting rps9(1–156)-GEM3Z plasmid which lacksSpeI and ClaI restriction sites was used as the templatefor PCR. Using outward directing primers with ClaI orSpeI sites and Pfu polymerase (Stratagene), linear PCRproducts with ClaI ad SpeI sites at each end weregenerated. Depending on the primers, ClaI and SpeIsites could be introduced anywhere along the length ofthe rps9 leader sequence. Table 1 lists all primer se-quences used for these HA tagged constructs. Eachamplified linear DNA was gel purified, digested withClaI and SpeI, and ligated with the HA linker toproduce plasmids having and HA tagged rps9 leader inpGEM3Z. Digestion of each plasmid with NsiI andligation of each HA tag insert into a NsiI digested andphosphatase treated pGRA-GFP vector created plas-mids HA38-GFP, HA55-GFP, HA84-GFP, HA109-GFP, HA127-GFP, and HA143-GFP.

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–21 13

Table 1PCR primers and linkers

Name Oligonucleotide sequencea

5�-ATGCATATGGCCCTCGAACG-TTGGTrps9.F1-NsiIG-3�

rps9 transF34-NsiI 5�-ATGCATGCC-TTCATAGTCCCTCAA-3�5�-ATGCATGACT-ATGAAGGCAGTTCCrps9.R37-NsiI-3�5�-ATGCATGTTGCGTTGAGGGAC-3�rps9.R41-NsiI

rps9.R49-NsiI 5�-ATGCATGAAATTCCTGGTAAAGCT-3�5�-ATGCATAACCAAGGGAAGGACACGGrps9.R55-NsiIA-3�5�-ATGCATTGGGAGTGACCGAGGGGAGrps9.R84-NsiIA-3�5�-ATGCATATCTTCGAAAGGACATCCrps9.R104-NsiI-3�

rps9.R127-NsiI 5�-ATGCATATCCGTCCCTAGAGCTGTG-3�5�-ATGCATCGGCGTTCCAGGCTCAGGrps9.R143-NsiI-3�

rps9.R156-NsiI 5�-ATGCATATAAGACCTTTTCCGGG-3�rps9.R169-NsiI 5�-ATGCATCGTCAGCTTTCCAGAACCAG

-3�5�-ATGCATAAGATAGTCGGCTGCGTCTCrps9.R179-NsiI-3�

rps9.R208-NsiI 5�-ATGCATGGCCTCCGCAATGATATCAAATTC-3�

rps9.R220-NsiI 5�-ATGCATGATTGCTCCAGACTGCCCG-3�5�-ATGCATTTTGCTGTACTGTTCCTTCrps9.R270(full)

-NsiI -3�5�-CATCACCTTCACC-CTCTC-3�GFP.R2 primer

HA.linker1 5�-CGATTACCCATACGATGTTCCAGATTACGCTA-3�

HA.linker2 5�-CTAGTAGCGTAATCTGGAACATCGTATGGGTAAT-3�5�-GACTAGTCAACGCAACCTGCATAGCTrps9.38F-SpeIT-3�

rps9.38R-ClaI 5�-CCATCGATA-GGGACTATGAAGGCAGTTCCG-3�

rps9.49R-SpeI 5�-GACTAGTGAAATTCCTGGTAAAGCTATG-3�5�-GACTAGTAAGCCGATTGGTTTGCAGGrps9.55F-SpeIAGA-3�

rps9.55R-ClaI 5�-CCATCGATAACCAAGGGAAGGACACGGA-3�5�-GACTAGTGGGAGTGTCGGCCAGGTGrps9.84F-SpeIAC-3�

rps9.84R-ClaI 5�-CGATCGATTGGGAGTGACCGAGGGGAGA-3�5�-GACTAGTTCGCTTCCTAGTGACTGGGrps9.109F-SpeIT-3�

rps9.109R-ClaI 5�-GGATCGATAGTTTCATGGACTGCATCTTC-3�5�-GACTAGTGCGACACTCTCAAGCATTCrps9.127F-SpeI-3�

rps9.127R-ClaI 5�-CCATCGATATCCGTCCCTAGAGCTGTG-3�

rps9.143F-SpeI 5�-GACTAGTAACAGCTTCTCATGGGGCAC-3�5�-CCATCGATCGGCGTTCCAGGCTCAGGrps9.143R-ClaIT-3�

a Restriction endonuclease recognition sites are underlined.

2.5. Electroporation and selection of stabletransfectants

HXGPRT deficient tachyzoites were harvested andresuspended (107 parasites total) in 300 �l of cytomixbuffer (120 mM KCl, 0.15 mM CaCl2, 5 mM MgCl2, 2mM EDTA, 10 mM KH2PO4, 25 mM HEPES-KOHpH 7.6, 2 mM ATP, 5 mM reduced glutathione) with30–40 �g of sterilized plasmid DNA in a 2 mm gapcuvette (BioRad) and electroporated under previouslydescribed protocols [14]. After electroporation, para-sites were left undisturbed for 15 min before inoclutat-ing human foreskin fibroblasts (HFF) grown on glasscoverslips (for fluorescent microscopy), T-25 cm2 flasks(for stable selection or for FACS), or 100×20 mmtissue dishes with half confluent Vero cells (for cloningGFP positive plaques). Stable transfectants were se-lected using medium having 25 �g ml−1 mycophenolicacid and 50 �g ml−1 xanthine. For cloning directlyfrom a 100×20 mm tissue plate, each plate was inocu-lated with 1 �l of the transfection solution. Afterovernight incubation, soft agar (0.9% w/v) was pouredonto each plate [16]. Around 5–7 days later, plaqueswere screened under an inverted fluorescent microscopefor GFP fluorescence. Sterile Pasteur pipette tips wereused to pick positive plaques, which were inoculatedinto individual wells of a 24-well plate, or into a T-25cm2 flask.

Alternatively, cloning was accomplished by first inoc-ulating 250 �l of transfection mixture into a T-25 cm2

flask. When parasites lysed the Vero cell monolayer, thepolyclonal population was sterilely sorted for GFPfluorescence using a Becton Dickinson FACStarPlus tocollect the brightest 2% of the parasites. Thirty thou-sand sorted positives were used to infect a 10 cm platewith a Vero cell monolayer for soft agar cloning. Mostparasites expressing GFP fluorescence could be clonedand passaged stably for weeks without losing fluores-cence intensity.

2.6. Florescence microscopy

Intracellular parasites were grown in a monolayer ofHFF adhered to coverslips. Extracellular parasites wereresuspended in phosphate buffered saline (PBS) andsettled onto poly-lysine coated coverslips. The cover-slips were washed with PBS, fixed with 3% para-formal-dehyde, permeablized with 0.5% Triton X-100, andincubated with the appropriate primary and secondaryantibodies (1:1000 dilution). 4�,6�-diamidino-2-phenylin-dole (DAPI) was used as a DNA specific dye that stainsthe nucleus and apicoplast but not the mitochondria ofT. gondii [4]. Between each change of reagent, thecoverslips were washed three times with PBS. Thecoverslips were mounted with Mowiol onto glass slidesand allowed to dry overnight before analysis.

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–2114

Immunofluorescence images were taken at the UABimaging Facility with a Zeiss Axiovert microscopeequipped with an ultraviolet filter set (Chroma Tech-nology; Inc in Brattleboro, VT). Images were capturedby a Photometrics Sensys cooled CCD, high resolution,monochromatic camera (Roper Scientific; Tucson, AZ).Image Acquisition Software was purchased from IPLabSpectrum from Scanalytics (Fairfax, VA).

2.7. Western blot analysis

Parasites were lysed in SDS sample buffer and imme-diately boiled for 5 min. SDS gel electrophoresis andelectroblotting onto immobilon membranes were per-formed according to standard protocols [17]. Afterblocking in 5% fat-free dry milk in TBS–Tween buffer(20 mM Tris–HCl pH 7.6, 137 mM NaCl, 0.1% w/vTween 20), blots were incubated with anti-GFP poly-clonal antibodies (1:3000 dilution) or anti-HA antibod-ies (1:3000 dilution). Bound antibodies were detectedwith the appropriate secondary antibodies conjugatedto horseradish peroxidase. Supersignal West picochemiluminescent substrate (Pierce) was used for detec-tion according to the manufacturer’s protocol.

3. Results

As a first step in mapping the domains involved inplastid targeting, a plasmid was constructed consistingof the full length rps9 leader sequence (designatedrps9(1–156)) fused to a green fluorescent protein (GFP)reporter. The rps9(1–156) parental plasmid was thenused to prepare a series of nested deletions of the rps9leader. Stable transfectants containing the full lengthconstruct and the deletions were examined to determinethe location of the GFP reporter protein. Three differ-ent cellular localization patterns were found (Fig. 1).Some transfectants (e.g. rps9(1–37)-GFP) displayedperinuclear labeling (Fig. 1A). This pattern of localiza-tion was identical to what was seen in parasites stainedwith antibodies to the endoplasmic reticulum (ER)chaperone BiP [18] (Fig. 1, panels B and C) suggestingthat parasites exhibiting this pattern of fluorescence hadthe GFP reporter protein localized to the ER. In con-trast, the constructs containing only the transit likedomain of rps9 and lacking the putative ER entrysignal sequence (amino acids 1–34, e.g. rps9(34–156)-GFP) displayed a different fluorescence pattern (Fig. 1,panels D–F). This pattern was identical to that seen incells stained with the mitochondrial specific dye Mito-Tracker Red CMXRos [19], (Fig. 1, panels E and F),suggesting a mitochondrial location of the GFP. Fi-nally, some of the constructs (e.g. the full length con-struct rps9(1–156)-GFP) produced a third pattern offluorescence (Fig. 1, panels G–I) that co-localized with

the apicoplast marker, acyl carrier protein [6] (Fig. 1,panels H and I).

Studies involving the various deletions in the leaderare summarized in Fig. 2. As a first step in mapping thefunctional domains of the rps9 leader, deletions of thecarboxyl terminus of rps9 were constructed to deter-mine the minimal amount of the rps9 leader sufficientfor apicoplast targeting. The construct rps9(1–55)-GFPsuccessfully targeted the GFP reporter protein to theapicoplast (Fig. 2). In contrast, a construct containingthe first 49 amino acids of the leader (rps6(1–49)-GFP)displayed perinuclear staining with anti-GFP antibodiesin an immunofluorescence assay (Fig. 2). These resultssuggest that the first 55 amino acids of the transitsequence are sufficient to successfully target the re-porter to the apicoplast, and that at least part of thetransit signal resides in the region encompassing aminoacids 49–55.

As mentioned in Section 1, earlier studies have sug-gested that the apicoplast leader is bipartitate in nature,with the amino terminal end encoding an ER entry likesignal and the remaining portion of the leader encodinga mitochondrial like transit signal. To further explorethis hypothesis, a series of amino terminal deletionswere constructed in the leader sequence. As predictedby the hypothesis proposing that the leader is biparti-tate in nature, the N-terminus deletion plasmidrps9(34–156)-GFP, which lacked the putative signalsequence but encodes the remaining portion of theleader sequence exhibited mitochondrial fluorescence(Fig. 2). Similarly, a construct containing amino acids34–55 (rps9(34–55)-GFP) was also found to target theGFP reporter protein to the mitochondrion (Fig. 2). Incontrast, the reporter protein produced from constructrps9(34–49) GFP was localized to the cytoplasm. To-gether, these results suggest that amino acids 34–55 aresufficient for transit targeting, and that at least part ofthe information essential for targeting resides in aminoacids 49–55.

Both the carboxyl and amino terminal deletion stud-ies suggested that amino acids 49–55 were important intargeting the GFP reporter to the apicoplast. In orderto determine if these residues were essential for thetargeting process, a plasmid was constructed consistingof the intact leader sequence with amino acids 49–55deleted. Surprisingly, this construct (rps9(D49–55)-GFP targeted the encoded GFP reporter protein to theapicoplast (Fig. 2). This result suggests that some re-dundancy might exist in the targeting information en-coded by the rps9 leader.

The deletion experiments described above suggestedthat the first 55 amino acids of the leader are sufficientto correctly target the GFP reporter protein to theapicoplast. However, the deletion studies resulted inleaders that were severely modified relative to the fulllength leader sequence. Furthermore, the deletion stud-

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–21 15

Fig. 1. Localization of GFP in constructs containing deletions of the rps9 leader. Panels (A–C) localization of GFP in cells transfected withrps9(1–37)-GFP. Panel (A) Reactivity of anti-GFP antibodies in transfected cells. Panel (B) Reactivity of antibodies of the ER marker proteinBIP in transfected cells. Panel (C) Merged image of panels A and B. Panels (D–F) Localization of GFP in cells transfected with rps9(34–156)-GFP. Panel (D) Reactivity of anti-GFP antibodies in transfected cells. Panel (E) Cells stained with the mitochondrial specific dye MitoTrackerRed CMXRos. Panel (F) Merged image of panels D and E. Panels (G–I) Localization of GFP in cells transfected with rps9(1–156)-GFP. Panel(G) Reactivity of anti-GFP antibodies in transfected cells. Panel (H) Reactivity of antibodies to the apicoplast protein acyl carrier protein. Panel(I) Merged image of panels G and H.

ies only provided information as to the localization of thereporter protein and did not shed any light on the stepsinvolved in processing the leader itself. To address thesequestions, linker scanner mutagenesis was used to inserta hemagglutinin (HA) epitope tag at different locationsin the transit domain of the rps9 leader. The insertionsites of HA tags within the transit domain are shownschematically in Fig. 3. Constructs were numbered ac-

cording to the insertional site. For example, HA38-GFPhas the HA tag inserted after amino acid 38 of the rps9leader. Stable lines containing the HA tagged rps9-GFPconstructs were obtained for all of the tagged constructswith the exception of HA55-GFP. No stable transfec-tants containing this construct were obtained after nu-merous attempts. Therefore, studies of this constructwere carried out on transiently transfected lines.

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–2116

A summary of the results obtained in the localizationstudies involving the HA tagged constructs is providedin Fig. 3, and examples of the immunolocalizationpatterns obtained are shown in Fig. 4. Since the HA tagis relatively small (13 amino acids), we predicted thattargeting of the HA tagged constructs would less dis-rupted than in the deletion constructs described above.In support of this hypothesis, HA tagging did nottotally disrupt targeting in any of the tagged constructs,since GFP fluorescence was seen at the apicoplast in allof the insertional mutants (Figs. 3 and 4). However,targeting was somewhat affected in these parasites, asGFP was also detectable in the perinuclear space ofconstructs tagged in the amino terminal half of theleader (HA38-GFP and HA55-GFP) (Figs. 3 and 4).Thus, although these constructs were able to target thereporter to its correct location, this process was not asefficient as with the parental leader construct, whereGFP was only detectable in the apicoplast (e.g. Fig. 1,panel C). In contrast to the results obtained withparasites tagged in the amino terminal portion of theleader, the three constructs tagged in the carboxylterminal portion of the leader only contained GFP in

the apicoplast (Figs. 3 and 4). This suggested thatprocessing in these lines was not affected by the incor-poration of the HA tag.

The presence of the HA tag in the leader of thetagged lines made it possible to localize the leadersequence in the transfected parasites, and gain someinsight into the processing pathway involved in matura-tion of apicoplast targeted nuclear encoded proteins. Asdiscussed above, constructs containing the HA tag in-serted in the amino terminal half of the transit sequence(HA-38-GFP and HA55-GFP) all contained GFP inboth the perinuclear space and in the apicoplast. How-ever, when lines containing these constructs wereprobed with the anti HA tag antibody, protein was onlydetected in the perinuclear space (Figs. 3 and 4). Incontrast, the three constructs containing the HA taginserted in the carboxyl terminal half of the leader(HA109-GFP, HA127-GFP and HA148-GFP) exhib-ited a different localization pattern. In these lines, theprotein that reacted with the anti-HA tag antibodylocalized only to the apicoplast (Figs. 3 and 4). Theseresults, when taken together, suggested that the first 55amino acids of the leader sequence are removed prior

Fig. 2. Summary of targeting of rps9-GFP deletion constructs. The boundaries of the deletion constructs are schematically indicated. The deletionconstructs contained varying portions of the signal sequence (white boxes) the transit sequence (grey boxes) and the mature rps9 sequence (blackboxes). All of the deletion constructs contained the complete GFP open reading frame (Hatched box). The location of the GFP protein is givento the right of each deletion diagram. P, perinuclear; A, apicoplast; M, mitochondria; C, cytoplasm.

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–21 17

Fig. 3. Localization of GFP and the HA tag in parasites transfected with HA tagged leaders. HA tagged constructs were generated by linkermediated insertion into rps9(1–156) as described in Section 2. The white boxes indicate the putative signal sequence of rps9. Gray boxes representthe putative transit sequence. Black boxes represent the mature portion of rps9. In each panel, ex, extracellular; in, intracellular; P, perinuclear;A, apicoplast localization.

to or during import of the protein into the apicoplast,while the carboxyl terminal domain of the leader (fromamino acids 109–158) is retained upon import.

To further investigate the pathway involved in pro-cessing of the leader, western blots were prepared withextracts derived from two HA tagged constructs repre-senting the two different patterns of HA localizationshown in Figs. 3 and 4. The blots were developed withantibodies to GFP and with antibodies recognizing theHA tag. Anti-GFP antibodies detected three proteinswith molecular weights of 47, 41 and 39 kDa in cellstransfected with HA38-GFP (Fig. 5, panel A). In cellstransfected with HA109-GFP, bands of 47 and 41 kDawere detected. In parasites transfected with the un-tagged rps9(1–156)-GFP construct, bands of 46 and 39kDa were detected. The 46 kDa band is approximately1 kDa smaller than the largest band detected in the twotagged constructs. This difference roughly correspondsto the predicted increase in mass that would result fromthe addition of the HA tag to the leader in the taggedconstructs (1.4 kDa).

In the Western blot developed using the anti HA tagantibody, a different pattern was found (Fig. 5, panelB). Here, the anti-HA tag antibody recognized the 47kDa band but not the 39 kDa band in parasites trans-fected with HA38-GFP. In parasites transfected withHA109-GFP, two bands of 47 and 40 kDa were recog-nized, identical to the pattern seen in the western blot

developed with the anti-GFP antibody. As expected,transfectants of the untagged leader construct rps9(1–156)-GFP did not react with anti-HA antibodies (Fig.5, panel B).

4. Discussion

As described in Section 1, it has been hypothesizedthat the leader of apicoplast targeted proteins is bipar-tite in nature. Nuclear encoded apicoplast proteins arethought to first be directed to enter the secretory path-way via a signal sequence present at the amino terminalend of the peptide. Once directed to the secretorypathway, the signal sequence is cleaved, exposing asecond domain. This second domain directs the proteinto the apicoplast. After import into the apicoplast, thetransit-like domain is removed, resulting in the produc-tion of the mature protein. This model is supported byprevious studies employing reporter proteins taggedwith leader sequences derived from T. gondii for acylcarrier protein [6] and rps9 [20]. However, because thesestudies relied on detection of the reporter proteinsalone, they were unable to detect the intermediates inthe targeting pathway. In the current study, we haveemployed deletion and HA tagged constructs of therps9 leader attached to the amino terminal end of aGFP reporter. The deletion constructs allowed us to

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–2118

map the domains important in the targeting process,while the HA tagged constructs have allowed us tofollow intermediates in the targeting pathway, sheddinglight on the steps involved in targeting and in process-ing the leader sequence.

Data derived from studies of the carboxyl terminaldeletions demonstrated that large C-terminal portionsof the rps9 leader could be deleted without affectingapicoplast targeting of GFP. These results suggestedthat the carboxyl terminal 102 amino acids of the leaderwere not necessary for correct targeting. Furthermore,the first 55 amino acids of the rps9 leader were found tobe sufficient for apicoplast targeting. In contrast, thefirst 49 amino acids were not sufficient to target thereporter to the apicoplast. These results suggested thatamino acids 49–55 contain important information forthe correct targeting to the apicoplast. However, addi-tional studies suggest that some redundancy may existin signals encoded in the leader. For example, we notedthat while constructs containing only the first 55 aminoacids of the rps9 leader were capable of targeting theGFP reporter to the apicoplast, these transfectants didnot display as strong an in vivo GFP fluorescence asdid transfectants containing constructs with the fulllength leader (data not shown). This suggests that whilethe first 55 amino acids were capable of targeting the

reporter to the apicoplast that this process might not beas efficient as it was with constructs containing the fulllength leader. Furthermore, although the nested dele-tion studies suggested that amino acids 49–55 con-tained information essential to direct the targeting tothe apicoplast, a construct deleting only amino acids49–55 was targeted to the apicoplast. These results,when taken together, suggest that while the first 55amino acids are sufficient to target a reporter to theapicoplast, some redundancy exists in the targetingsignal encoded in the leader. This finding is in line withprevious studies that have demonstrated that plantplastid and mitochondrial targeting signals generallycontain redundant information and that small deletionsor insertions, therefore, do not generally alter targeting[21,22].

A positively charged amphiphilic alpha helix struc-ture could be predicted between amino acids 40–52 ofthe rps9 leader, and the PSORT [23] and TargetP [24]programs predicted the rps9 putative transit-like do-main to be a mitochondrial targeting peptide. Theseobservations may explain the mitochondrial targetingability of the transit domain of rps9 when expressed inthe absence of the putative signal sequence encoded inamino acids 1–34. This finding suggests that the api-coplast transit signal of rps9 may have much in com-mon with the mitochondrial transit signal.

Fig. 4. Immunolocalization assays of GFP and HA tag sequence in hemagglutinin tag insertion transfectants. Arrows point to the apicoplasts. Notall apicoplasts were in the photographed focal plane.

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–21 19

Fig. 5. Western blot analysis of HA tagged constructs. Extracts ofuntagged (rps9 (1–156)) and HA tagged extracellular parasites wereused to prepare duplicate western blots which were probed withanti-GFP and anti-HA tag antibodies. HA tagged constructs weregenerated by linker mediated insertion into rps9(1–156) as describedin Section 2. Panel (A) Western blot probed with anti-GFP antibod-ies. Panel (B) Western blot probed with anti-HA tag antibodies.

near the carboxyl terminus of the transit domain, there-fore, did not appear to affect apicoplast targeting.

Western blot and immunolocalization studies utiliz-ing anti-GFP and anti-HA tag antibodies suggest thatprocessing of the rps9 leader is a multi-step process. Inparasites transfected with HA38-GFP and HA-55-GFP,GFP was localized to both the perinuclear space andthe apicoplast, while the HA tag was present only in theperinuclear space. Furthermore, in western blot analysisof these constructs only the largest protein recognizedby the GFP antibodies was recognized by the HA tagantibodies. In contrast, in parasites transfected with theHA109-GFP, HA tagged protein was located in theapicoplast, and the peptides recognized by both theGFP and HA tag antibodies were identical. Theseresults, when taken together, suggest that some sort ofcleavage event is occurring between amino acids 55 and109 of the leader, as HA tags inserted at or beforeposition 55 are not imported into the apicoplast, whileHA tags inserted at position 109 are imported into theapicoplast.

The 39 kDa peptide common to all the constructscannot represent a fully processed product, as it is 9kDa larger than mature GFP expressed without aleader. It is possible that the recognition site for thefinal cleavage event resulting in the removal of the lastportion of the rps9 leader is encoded in the mature rps9protein itself. If this was the case, this site would beabsent in the GFP chimeras, and the final cleavageevent would not occur. However, western blots detectthe 39 kDa peptide in rps9(1–169) GFP transfectedparasites. Nothing of the size predicted for cleavage atthe amino terminal end of the mature rps9 protein isseen (data not shown). Since rps9(1–169) GFP containsroughly the first 25 amino acids of the mature rps9protein, this result suggests that if a signal for the finalcleavage event is encoded in rps9, it is not localizedwithin the first 25 amino acids of the mature protein. Itis possible that, as has been documented in the case ofribosomal protein L-18 of Chlamydomonas reinharditii[26], final processing may occur during or after riboso-mal assembly. Further work will be needed to addressthis possibility.

These results, when taken together, suggest a modelfor the pathway involved in the trafficking of the rps9leader tagged GFP constructs (Fig. 6). The peptidecontaining the full length bipartitate leader is co-trans-lationally introduced into the ER. At this point thesignal sequence (roughly amino acids 1–34) is removed,exposing the transit sequence. The peptide is thentranslocated to the apicoplast through the perinuclearspace. Trafficking to the apicoplast involves a recogni-tion site encoded in amino acids 49–55 of the leadersequence, as well as a redundant site or sites encodedelsewhere in the leader. During the trafficking process,the transit peptide is cleaved between amino acids 55

As mentioned above, small insertions in the leadersequence do not severely affect targeting in mitochon-dria and plant plastids [21,22]. This also seems to be thecase in apicoplast targeting in T. gondii, as all of theHA tagged constructs appeared to target the GFPreporter to the apicoplast. However, some effects ontargeting were seen in constructs containing the HAtags. For example, the HA38-GFP and HA55-GFPconstructs contained GFP protein both in the api-coplast and in the perinuclear region. The GFP in theperinuclear region co-localized with anti-BiP antibodies(data not shown). Since the perinuclear nuclear envel-ope has been proposed as the intermediary between theER and Golgi [25], this result supports the model inwhich the nuclear encoded complex plastid protein firsttransverses the secretory system. Furthermore, likerps9(1–55)-GFP, the fluorescence intensity of HA38-GFP was noticeably dimmer than that seen in theuntagged full length construct (data not shown). Theseresults suggest that the amino terminal HA taggedconstructs were targeted correctly, but that the target-ing process was less efficient than in constructs contain-ing the unmutated leader. As a result, precursors mayhave either been backed up in the secretory system,requiring more time to be imported in these mutants, orsome proportion of the products may be mistargeted.In contrast to the amino terminal tagged constructs,transfectants with the HA tag inserted in the carboxylterminal half of the leader contained all of the GFPreporter protein in the apicoplast. Having the HA tag

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–2120

Fig. 6. Proposed model for rps9 processing. Details of the model are given in the text. The proposed steps in the model are as follows. Step 1,cleavage and removal of the signal sequence (ca. amino acids 1–34). Steps 2 and 3, cleavage of the leader sequence at a site between amino acids55 and 109. Step 4, import of the protein into the apicoplast and removal of the remainder of the leader. The bracket surrounding steps 2 and3 indicates that the cellular compartment where the cleavage of the leader sequence at the site between amino acids 55 and 109 occurs is notknown.

and 109. The 41 kDa peptide detected by the anti-GFPantibodies in extracts prepared from HA38-GFP para-sites may represent an intermediate in this process, ormay represent an aberrantly processed version of thepeptide derived from mistargeted proteins, as discussedabove. Finally, once in the apicoplast, the remainingportion of the leader is removed, resulting in the pro-duction of the mature protein. This final processingstep may occur concurrently with ribosome assembly.

In summary, the data reported above demonstratethat the amino terminal 55 amino acids of the rps9leader are sufficient to target a reporter gene to theapicoplast of T. gondii, but that other sites encoded inthe leader may also be involved in this process. Experi-ments employing the HA tagged versions of the leaderdemonstrate that, as hypothesized, trafficking of thenuclear encoded apicoplast proteins proceeds throughthe ER. Finally, these data suggest that processing ofthe HA tagged leaders is a multi-step process, involvingboth signal sequence removal and a cleavage event

between amino acids 55 and 109 of the leader. Furtherexperiments will be necessary to map the redundanttargeting signals in the rps9 leader, to demonstrate theprecursor product relationship between the 47 kDaproduct and the 39 kDa product, to determine exactlywhere the cleavage event that gives rise to the 39 kDaproduct occurs, and to study the final cleavage eventsthat result in the production of fully mature RPS9protein.

Acknowledgements

The authors thank Con Beckers and members of theBeckers, Lang–Unnasch, and Unnasch labs for helpfulsuggestions and discussions. We thank the technicalhelp of Albert Tousson and Marion Spell for mi-croscopy and FACS. This work was supported by agrant from the US National Institutes of Health (RO1Al 48737).

S. Yung et al. / Molecular & Biochemical Parasitology 118 (2001) 11–21 21

References

[1] McFadden GI, Roos DS. Apicomplexan plastid as drug targets.Trends Microbiol 1999;7:328–33.

[2] McFadden GI, Reith ME, Mulholland J, Lang-Unnasch N.Plastid in human parasites. Nature 1996;381:482.

[3] Wilson RJM, Denny PW, Preiser PR, Roberts K, Roy A, WhyteA, Strath M, Moore DJ, Williamson DH. Complete gene map ofthe plastid-like DNA of the malaria parasite Plasmodium falci-parum. J Mol Biol 1996;261:155–72.

[4] Kohler S, Delwiche CF, Denny PW, Tilney LG, Webster P,Wilson RJM, Palmer JD, Roos DS. A plastid of probable greenalgal origin in apicomplexan parasites. Science 1997;275:1485–9.

[5] Lang-Unnasch N, Reith ME, Munholland J, Barta JR. Plastidsare widespread and ancient in parasites of the phylum apicom-plexa. Int J Parasitol 1998;28:1743–54.

[6] Waller RF, Keeling PJ, Donald RGK, Striepen B, Handman E,Lang-Unnasch N, Cowman AF, Besra GS, Roos DS, McFaddenGI. Nuclear-encoded proteins target to the plastid in Toxo-plasma gondii and Plasmodium falciparum. Proc Nat Acad SciUSA 1998;95:12352–7.

[7] McFadeen GI. Plastids and protein targeting. J Euk Microbiol1999;46:339–46.

[8] Schwartzbach SD, Osafune T, Loffelhardt W. Protein importinto cyanelles and complex chloroplasts. Plant Mol Biol1998;38:247–63.

[9] Seedorf M, Damelin M, Kahana J, Taura T, Siliver P. Interac-tion between a nuclear transporter and a subset of nuclear porecomplex proteins depends on Ran GTPase. Mol Cell Biol1999;19:1547–57.

[10] Bangs JD, Uyetake L, Brickman MJ, Balber AE, Boothroyd JC.Molecular cloning and cellular localization of a BiP homologuein Trypanosoma brucei. J Cell Sci 1993;105:1101–13.

[11] Yung S, Lang-Unnasch N. Targeting of a nuclear encodedprotein to the apicoplast to Toxoplasma gondii. J Euk Microbiol1999;46:79–80.

[12] Kim K, Eaton MS, Schubert W, Wu S, Tang J. Optimizedexpression of green fluorescent protein in Toxoplasma gondiiusing thermostable green fluorescent protein mutants. MolBiochem Parasitol 2001;113:309–13.

[13] Siemering KR, Golbik R, Sever R, Haseloff J. Mutations thatsuppress the thermosensitivity of green fluorescent protein. CurrBiol 1996;6:1653–63.

[14] Roos DS, Donald RG, Morrissette NS, Moulton AL. Moleculartools for genetic dissection of the protozoan parasite Toxoplasmagondii (review). Methods Cell Biol 1994;45:27–63.

[15] Barnhart KM. Simplified PCR-mediated, linker-scanning muta-genesis. BioTechniques 1999;26:624–6.

[16] Karsten V, Huilin Q, Beckers CJ, Joiner KA. Targeting thesecretory pathway of Toxoplasma gondii. Methods 1997;13:103–11.

[17] Sambrook J, Fritsch EF, Manistis T. Molecular Cloning: ALaboratory Manual. Cold Spring Harbor, NY: Cold SpringHarbor Laboratory Press, 1989.

[18] Haas IG, Wabl M. Immunoglobulin heavy chain bindingprotein. Nature 1983;306:287–9.

[19] Melo EJ, Attias M, De Souza W. The single mitochondrion oftachyzoites of Toxoplasma gondii. J Struct Biol 2000;130:27–33.

[20] DeRocher A, Hagen CB, Froehlich JE, Feagin JE, Parsons M.Analysis of targeting sequences demonstrates that trafficking tothe Toxoplasma gondii plastid branches off the secretory system.J Cell Sci 2000;113:3936–77.

[21] Reiss B, Wasmann C, Bohnert H. Regions in the transit peptideof SSU essential for transport into chloroplasts. Mol Gen Genet1987;209:161–221.

[22] Hurt E, Pesold-Hurt B, Suda K, Opligger W, Schatz G. The firsttwelve amino-acids (less than half of the pre-sequence) of animported mitochondrial protein can direct mouse cytosolic dihy-drofolate reductase into the yeast mitochondrial matrix. EMBOJ 1985;4:2061–8.

[23] Nakai K, Horton P. PSORT: a program for detecting sortingsignals in proteins and predicting their subcellular localization.TIBS 1999;24:34–5.

[24] Emanuelsson O, Nielsen H, Brunak S, von Heijne G. Predictingsubcellular localization of proteins based on their N-terminalamino acid sequence. J Mol Biol 2000;300:1005–16.

[25] Hager KM, Striepen B, Tilney LG, Roos DS. The nuclearenvelope serves as an intermediary between the ER and Golgicomplex in the intracellular parasite Toxoplasma gondii. J CellSci 1999;112:2631–8.

[26] Liu XQ, Gillham NW, Boynton JE. Chloroplast ribosomalprotein L-18 in Chlamydomonas reinhardtii is processed duringribosome assembly. Mol Gen Genet 1988;214:588–91.