Regulated expression of the Leishmania major surfacevirulence factor lipophosphoglycan usingconditionally destabilized fusion proteinsLuciana Madeira da Silva, Katherine L. Owens, Silvane M. F. Murta1, and Stephen M. Beverley2

Department of Molecular Microbiology, Washington University School of Medicine, St. Louis, MO 63110

Edited by Thomas E. Wellems, National Institutes of Health, Bethesda, MD, and approved March 16, 2009 (received for review February 13, 2009)

Surface glycoconjugates play important roles in the infectious cycleof Leishmania major, including the abundant lipophosphoglycan(LPG) implicated in parasite survival in the sand fly vector and theinitial stages of establishment in the mammalian host macrophage.We describe a system for inducible expression of LPG, applying anovel protein-based system that allows controlled degradation ofa key LPG biosynthetic enzyme, UDP-galactopyranose mutase(UGM). This methodology relies on a mutated FK506-binding pro-tein (FKBP) destabilizing domain (dd) fused to the protein ofinterest; in the absence of rapamycin analogs, such as Shld1, the dddomain is destabilized, leading to proteasomal degradation,whereas drug treatment confers stabilization. Tests in L. majorusing dd fusions to a panel of reporters and cellular proteinsconfirmed its functionality, with a high degree of regulation andlow background, and we established the kinetics of protein acti-vation and/or loss. Two inexpensive and widely available ligands,FK506 and rapamycin, functioned similarly to Shld1, without effecton Leishmania growth or differentiation. We generated parasiteslacking UGM through deletion of the GLF gene and substitutionwith a ddGLF fusion construct, either as chromosomal knockins orthrough episomal complementation; these showed little or no LPGexpression in the absence of inducer, whereas in its presence, highlevels of LPG were attained rapidly. Complement lysis tests con-firmed the correct integrity of the Leishmania LPG coat. These datasuggest that the dd approach has great promise in the study of LPGand other pathways relevant to parasite survival and virulence.

inducible expression � FK506 � glycoconjugates � pathogen �trypanosomatid protozoa

Leishmaniasis is a parasitic disease infecting more than 12million people worldwide and constituting a significant pub-

lic health burden in affected areas (1). It is caused by protozoanparasites of the genus Leishmania that, depending on the species,cause a range of pathologies, from cutaneous or mucocutaneousto the fatal visceral leishmaniasis. Since the emergence ofreverse genetic approaches, including homologous gene replace-ment and heterologous expression, our ability to probe genefunction and expression related to parasite virulence has ad-vanced considerably (2, 3). Nonetheless, the Leishmania toolkitwould benefit from further development incorporating methodsbetter able to cope with challenges arising from the diploid natureof the Leishmania genome, as well as research into the limitationsof current regulatable systems, which are typically based on variousstrategies affecting transcriptional initiation (4). These face somechallenges because of the unique aspects of transcription and geneexpression in trypanosomatid protozoans (5, 6).

Protein-based regulatable systems offer some potential for over-coming limitations arising from regulatable transcription systems.Recently, Wandless and colleagues described a system in whichprotein levels are controlled through regulated degradation of acis-acting domain joined to the target protein of interest (7). Thedestabilizing protein domain consists of a modified, 108-aa FK506/rapamycin-binding protein (FKBP) engineered to bind selectivelyto the nontoxic FK506/rapamycin analog, Shld1. In the presence of

Shld1, the FKBP destabilization domain (dd) is properly folded,thereby conferring stability to the dd fusion protein, whereas in itsabsence, destabilization of the dd structure targets the protein fordegradation. This method was applied recently to the Apicompl-exan parasites Plasmodium falciparum and Toxoplasma gondii (8, 9),and here we report success with 2 species of Leishmania parasites.Of practical importance, the relative insensitivity of Leishmania toFKBP ligands, including rapamycin and FK506, allows them to beused in vitro, advantageous because of their lower cost and wide-spread availability.

We applied this to the development of regulated expression of akey Leishmania virulence molecule, lipophosphoglycan (LPG). Thecell surface of Leishmania is densely covered with a variety ofglycoconjugates playing major roles throughout the parasite’s lifecycle (10). One of the most intensely studied is the abundant LPG,consisting of a long phosphoglycan chain of galactose-mannose-phosphate-based repeating units attached to the plasma membranethrough a GPI anchor (11). In L. major, LPG plays a role in theestablishment of the infection of promastigotes in the sand flyvector, acting as an adhesin that promotes attachment of theparasites to the insect midgut wall through binding to a galectinreceptor (12). After inoculation of the parasite into the mammalianhost by the sand fly bite, LPG plays important roles in theestablishment of the infection by conferring resistance to lysis bycomplement, protection from oxidative damage, and remodeling ofthe initial phagolysome, including the induction of a transient delayin phagolysosomal fusion (13–15). The roles of LPG deduced by avariety of biochemical and cellular studies have been supported bystudies of mutant parasites lacking genes of the LPG biosyntheticpathway (13, 16–19).

LPG is assembled through a series of steps involving a diverse setof proteins located in cellular compartments, including the cytosol,glycosome, and secretory pathways (11, 20). A key linkage withinthe LPG core involves galactosylfuranose (Galf), an unusual par-asite sugar not made by the mammalian host (21, 22). The precursorfor Galf is UDP-Galf, which arises from the action of cytosolicUDP-galactopyranose mutase (UGM, encoded by the gene GLF)(16, 23), and glf� knockouts confirm the requirement for this genein Leishmania LPG biosynthesis (18). Although Galf occurs onother Leishmania glycoconjugates, including the abundant glyco-sylinositolphosphatidylinositols (GIPLs), previous studies establishthat GIPLs play little role in L. major infectivity (24) and, accord-ingly, glf� knockouts show phenotypes identical to lpg1� parasites

Author contributions: L.M.d.S., K.L.O., S.M.F.M., and S.M.B. designed research; L.M.d.S.,K.L.O., and S.M.F.M. performed research; L.M.d.S., K.L.O., S.M.F.M., and S.M.B. analyzeddata; and L.M.d.S. and S.M.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1Present address: Laboratorio de Parasitologia Celular e Molecular, Centro de Pesquisas ReneRachou, FIOCRUZ, Av. Augusto de Lima 1715, Caixa Postal 1743, CEP 30190-002, Belo Hori-zonte, MG, Brazil.

2To whom correspondence should be addressed. E-mail: beverley@borcim.wustl.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0901698106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0901698106 PNAS � May 5, 2009 � vol. 106 � no. 18 � 7583–7588

MIC

ROBI

OLO

GY

lacking LPG alone (18, 19, 25). In this work, we used the dd fusionapproach to regulate LPG expression through control of UGMlevels, yielding parasites that when placed in the ‘‘on’’ and ‘‘off’’states closely resemble WT and LPG-deficient mutants.

ResultsThe Modified FKBP dd Is Functional in L. major. Preliminary testssuggested that the stabilizing ligand Shld1 had little toxicity to L.major at 1 �M. We expressed a dd yellow fluorescent protein(ddYFP) in L. major after integration of constructs into the SSUribosomal locus (SSU::IR1PHLEO-ddYFP) and compared it toYFP expression in similar independent transfectants(SSU::IR1PHLEO-YFP). Parasites expressing unmodified YFPwere highly fluorescent [62 fluorescence units (FU)], whereas thoseexpressing ddYFP were only slightly more fluorescent than un-transfected controls (9 vs. 7 FU; Fig. 1 A and B). Notably, additionof 1 �M Shld1 to the ddYFP-expressing Leishmania for 24 h yieldedstrongly fluorescent parasites (�150 FU), 16-fold greater than seenin the absence of ligand (62-fold after correcting for backgroundfluorescence; Fig. 1 A and B). Similar results were obtained withddYFP expressed from multicopy episomal vectors.

We tested a cytosolic dd-Luciferase (ddLUC) construct afterexpression in L. major. In the absence of Shld1, ddLUC transfec-tants showed minimal luminescence, close to the WT background(1.0 � 104 vs. 1.2 � 104 photons per second; Fig. 1C). After

incubation in the presence of 1 �M Shld1 for 24 h, LUC activity rose350-fold (4.3 � 106 photons per second; Fig. 1C).

FK506 and Rapamycin Can Be Used as ddFKBP Stabilizing Compoundsin L. major. Shld1 was developed as a nontoxic analog of the FKBPligands FK506 or rapamycin, which show toxicity or deleteriouseffects on eukaryotic cells (26–28). Although the Leishmania sp.genomes reveal several TOR kinase homologs (http://tritrypdb.org/tritrypdb/), the action of these compounds on Leishmania parasiteshas received less attention (29). We established the growth inhibi-tion IC50 for Shld1, FK506 or rapamycin as 8.4 � 0.3, 7.7 � 0.2, or4.9 � 0.5 �M, respectively (Fig. 2A, Table S1). We then testedddYFP expression after growth of ddYFP-expressing L. major for24 h in the presence of 1 �M ligand (Fig. 1B). FK506 and rapamycininduced strong YFP fluorescence in ddYFP-expressing L. major,slightly higher than seen with Shld1 (Fig. 1B). Similarly, FK506 andrapamycin induced high levels of LUC expression in ddLUCexpressing L. major—200-fold and 250-fold that seen in the absenceof ligand (Fig. 1C).

Dose–response curves were performed for all 3 compounds byusing the ddYFP-expressing L. major and 24 h of induction (Fig.2B). Shld1 was the most potent, with an effective concentration(EC50) of 10 nM, followed by rapamycin (60 nM) and FK506 (200nM). As seen in previous studies, ddYFP expression was ‘‘tunable,’’in that variation in ligand concentration yielded variation in theaverage YFP fluorescence homogeneously within the cell popula-tion (Fig. S1A). The concentration required for maximal ddYFPexpression was 250 nM for Shld1 and 1 �M for FK506 andrapamycin (Fig. 2B). At 1 �M ligand, neither parasite growth nordifferentiation to the infective metacyclic form was affected (Fig.2A and Fig. S1B).

Kinetics of ddFKBP-Controlled Induction and Loss in L. major. Earlylog-phase promastigotes were treated with different concentrationsof stabilizing compounds, and ddYFP expression was assessed byflow cytometry for 24 h. Maximum ddYFP protein levels wereachieved within 8 h for all 3 drugs, with similar kinetics (Fig. 3 A–C).

To study decay, parasites were incubated 24 h with ligand, andthen were washed once and inoculated into fresh media. More than2 h after removal of Shld1 and FK506, ddYFP expression haddecreased to nearly basal levels, with a half-life of about 30 min (Fig.

Fig. 1. Functionality of the FKBP dd system in L. major. WT, YFP-expressing,ddYFP-expressing, or ddLUC-expressing promastigotes were grown in the pres-ence or absence of 1 �M Shld1, FK506, and rapamycin for 24 h, and fluorescencelevels were measured by flow cytometry (A and B) or luciferase activity assay (C).(A) Histogram plot representative of the distribution of fluorescence levels in thedifferent populations of cells. (B) Average mean fluorescence � SEM relative tothe maximum fluorescence intensity measured in the series. Experiments wereperformed in triplicate. (C) Average luciferase activity � SEM (p/s) for 3 indepen-dent clones. All transfectants shown express proteins from transgenes insertedinto the rRNA locus (SSU::IR1PHLEO-YFP, -ddYFP, or -LUC).

Fig. 2. Effects of Shld1 alternative ligands, FK506, and rapamycin on L. majorgrowth, ddYFP induction, and metacyclogenesis in L. major. Promastigotes(SSU::IR1PHLEO-ddYFP) were treated with different concentrations of drug orsolvent (negativecontrol) for24h,andcelldensity (A) andddYFPfluorescence (B)were then measured by Coulter and flow cytometry, respectively. Error barsrepresent SEM of 2 independent experiments performed in duplicate.

7584 � www.pnas.org�cgi�doi�10.1073�pnas.0901698106 Madeira da Silva et al.

3 D–F). The decay of ddYFP expression was slower after rapamycintreatment, reaching basal levels after more than 4 h and with adecay half-life of about 75 min (Fig. 3 D and E). Western blotanalysis with anti-GFP antibody, which also recognizes YFP, con-firmed these findings (Fig. S2). As summarized in Table S1, whenused at 1 �M, FK506 and rapamycin can be used effectively asFKBP dd stabilizing ligands in Leishmania.

Application of the ddFKBP System to Leishmania Proteins and inLeishmania braziliensis. We generated dd fusions for a variety ofL. major proteins, including DHCH1 (5,10-methylenetetrahy-drofolate dehydrogenase/5,10-methenyltetrahydrofolate cyclo-hydrolase), FTL (formate-tetrahydrofolate ligase), andDHFR-TS (dihydrofolate reductase thymidylate-synthase).Each fusion protein was expressed in L. major, and its levels weredetermined by Western blotting after 24 h in the presence orabsence of inducer. For each protein, strong regulation was seenin orders of magnitude similar to those seen for the reporterproteins. In the absence of inducer, expression of the fusionprotein was negligible at the conditions tested (Fig. S3A). Similarresults were obtained with ddYFP, ddLUC, ddFTL, orddDHCH1 proteins expressed in promastigotes of L. braziliensis,an early-diverging species belonging to the subgenus Viannia(Fig. S3 B and C). These results suggest that the ddFKBP fusionapproach will have broad applicability in Leishmania.

Generation of an L. major Mutant Lacking LPG Through Ablation ofGalf Synthesis. We generated a glf� null mutant (formally, glf�HYG/glf�PAC) by standard techniques of gene replacement (Fig. 4 andFig. S4A), which requires 2 successive rounds of targeting becauseLeishmania is predominantly diploid (30). Generation of the cor-rect planned replacements and loss of GLF were confirmed by PCR(Fig. S4B). Western blotting with the monoclonal antibody

WIC79.3, which recognizes the galactose-containing side chains ofLPG and other phosphoglycans, showed that the glf� parasiteslacked LPG (Fig. S4C) but retained expression of the high-molecular weight proteophosphoglycans (PPGs), which lack Galf(20). In contrast, control mutant lpg2�, lacking the Golgi GDP-mannose transporter required for the synthesis of all phosphogly-cans, showed loss of both LPG and PPG (Fig. S4C). These findingsagree with previous work using another L. major glf� strain (18).

Complementation of L. major glf� with ddUGM Confers RegulatableLPG Expression. A ddUGM fusion protein was expressed in the glf�mutant by transfection with a multicopy episomal vector,pIR1PHLEO-ddGLF (Fig. 4). glf�/�pIR1PHLEO-ddGLF trans-fectants were grown in the presence or absence of 1 �M FK506 for24 h, and the levels of ddUGM protein were assessed by Westernblotting. Some variations in basal LPG and/or ddUGM inductionwere seen in these transfectants that were attributed to differencesin copy number of the episomal vector (Fig. S5).

Dose–response studies for several clonal lines showing mini-mal basal LPG and ddUGM expression were performed (one isshown in Fig. 5A). In the absence of Shld1, a low level of LPGexpression was found, whereas addition of just 10 nM Shld1resulted in strong LPG expression without detectable expressionof ddUGM protein, suggesting that very low protein levels arerequired for LPG synthesis. WT levels of LPG were attained atjust 50 nM Shld1 (Fig. 5A).

Knockin Approaches Yield Improvements in ddGLF-Dependent LPGDown-Regulation. The utility of regulatable systems is ultimatelydependent on their ability to cross biologically relevant thresholds

Fig. 3. Characterization of the kinetics of expression and loss of ddYFP fluo-rescence upon addition or removal of Shld1, FK506, or rapamycin. (A–C) Drugswere added to logarithmic-phase cultures of L. major promastigotes(SSU::IR1PHLEO-ddYFP), and ddYFP fluorescence levels were measured by flowcytometry. (D–F) After 24 h in the presence of ligands, parasites were washed andthen suspended in fresh culture medium lacking drugs, and fluorescence levelswere measured periodically. (A and D) Shld1. (B and E) FK506. (C and F) Rapamy-cin. D–F Insets show the experiment extended to 24 hr. Data are presented as theaverage mean fluorescence � SEM relative to the maximum fluorescence inten-sity measured in the series. Two or 3 independent experiments were performedin triplicate.

Fig. 4. Generation of glf� mutants expressing ddUGM from multicopy epi-somes or chromosomal knockins. L. major LV39cl5 parasites were transfectedwith a construct replacing the GLF ORF of the first allele with that encoding ahygromycin B resistance marker (HYG). This heterozygous line (�/glf�HYG) wasused in 2 ways. As depicted on the left branch, the remaining GLF allele wasreplaced with that encoding a puromycin resistance marker (PAC), yielding ahomozygous glf� null mutant (formally, glf�HYG/glf�PAC). GLF was then re-stored by transfection with pIR1PHLEO-ddGLF, yielding glf�/pIR1PHLEO-ddGLFparasites. Alternatively, as depicted in the right branch, the remaining GLF allelein �/glf�HYG was replaced by transfection with a PHLEO-ddGLF knockin-targeting construct, yielding glf�HYG/ddGLF.

Madeira da Silva et al. PNAS � May 5, 2009 � vol. 106 � no. 18 � 7585

MIC

ROBI

OLO

GY

distinguishing WT and mutant phenotypes. However, expressionlevels from most common Leishmania vectors are considerablyhigher than those of chromosomal genes as a result of beingmulticopy episomes or from integration into the rRNA locus (31,32). Thus, overexpression of a destabilized protein might leaveprotein levels high enough to fulfill normal function, perhapsaccounting for the residual LPG expression evident in the studiesabove employing episomal ddUGM vectors. To test this, we gen-erated a chromosomal knockin line expressing a single copy ofddUGM from the normal GLF locus.

We first created a generic knockin cassette that consists of aphleomycin resistance ORF (PHLEO) followed by a Leishmaniaintergenic region driving expression of the dd domain (Fig. S6A).This was then flanked on the 5� side with 915 bp of the 5� GLFflanking region and on the 3� side with the 1.5-kb GLF ORF (Fig.

S6B). This fragment was transfected into a heterozygous linebearing an HYG replacement of one GLF allele, followed byselection for HYG and PHLEO (Fig. 4). Several clones glf�HYG/ddGLF were obtained with integration of the PHLEO-ddGLFcassette in the correct genomic locus (Fig. S6C), referred to here asddGLF knockin lines.

Importantly, LPG levels were undetectable in the ddGLFknockin line cells grown in the absence of FK506 for 48 h (Fig.5B, lane 3), in contrast to residual basal levels found in theglf�/�pIR1PHLEO-ddGLF transfectants (Fig. 5A, lane 3, andFig. S6D). Western blotting revealed that in the presence of 1 �MFK506, ddUGM levels were �10-fold lower in the ddGLFknockin lines compared with the glf�/�pIR1PHLEO-ddGLFtransfectants (Fig. S6D), as predicted. Thus, reduction in thelevels of ‘‘destabilized’’ UGM by the knockin strategy resulted inimproved down-regulation of LPG levels. Although some clonalvariability was evident, as seen before, in 5 clones the LPG levelsafter induction were fully restored to WT levels (Fig. 5B, WTlane 1 vs. lane 6 or 7).

Kinetics of ddUGM-Dependent LPG Induction and Loss. Similar tomany eukaryotic glycoconjugates, LPG synthesis occurs in theGolgi, followed by trafficking to the cell surface, where it iscontinuously ‘‘shed’’ with a half-life of about 3–7 h (33, 34). Thus,we examined the effects of regulated UGM expression on thekinetics of LPG acquisition and loss because this will affect its utilityin functional studies of LPG. The kinetics of Shld1 stabilization ofddUGM appeared to be somewhat less rapid than seen with ddYFPreporter, with maximum ddUGM expression requiring 24 h ofincubation with FK506. The appearance of LPG appeared to mirrorddUGM accumulation (Fig. 5C). The ddUGM levels dropped tonear-basal levels 4 h after removal of FK506. In contrast, LPG levelsfell more slowly, with a 3-fold decrease after 8 h, and complete lossof LPG was achieved 24 h after FK506 withdrawal (Fig. 5 B and C).Thus, after ddUGM destabilization, LPG levels are lost from a cellmore slowly than ddUGM, at rates similar to the rate of sheddingestablished in previous studies.

ddUGM-Dependent LPG Synthesis Recapitulates Expected Comple-ment Sensitivities. The Leishmania surface is highly sensitive toperturbations involving loss and/or overexpression of many surfacemolecules, including LPG, gp46 and gp63, and the SHERP/HASPfamily, as judged by increased sensitivity to lysis by complement (19,35–37). We used this assay to probe the faithfulness of the ddUGM-dependent LPG regulatory system to modulate the parasite surface.glf� mutants were highly susceptible to lysis by complement (Fig.6A), as seen in other LPG-deficient mutants (19).

The glf�/�pIR1PHLEO-ddGLF transfectants grown in the pres-ence of FK506 were completely protected from lysis by complement(Fig. 6B), as expected given their strong LPG expression (Fig. 5A).However, in the absence of ligand most parasites (64%) remainedcomplement-resistant, suggesting that the low LPG levels present inthese cells were nonetheless sufficient to preclude lysis. This mayreflect heterogeneity in the cellular population, most likely arisingfrom variation in plasmid copy number.

Consistent with their high LPG levels, the knockin ddGLFparasites were highly resistant to complement in the presence ofFK506 inducer. Importantly, in the absence of FK506 they becamehighly susceptible to complement lysis, similar to the glf� control(Fig. 6C) and consistent with the absence of LPG (Fig. 5B). Asnoted earlier, clonal variability was evident, with many lines show-ing an intermediate phenotype.

DiscussionHere, we investigated the functionality of a recently developedsystem to regulate protein levels that consists of tagging theprotein of interest with an FKBP dd, which in the absence of astabilizing ligand (Shld1, rapamycin, or FK506) is targeted to

Fig. 5. Regulatable LPG and ddUGM expression in L. major. LPG, ddUGM, andhistone 2A (loading control) levels were determined by Western blotting withappropriate antisera. Sample controls in all panels are WT L. major LV39c5 andglf�, the homozygous GLF null mutant. (A) Analysis of a representative glf�/�pIR1PHLEO-ddGLF line. Parasites propagated in the absence of inducer weregrown 24 h in the presence of Shld1 before harvest. (B) Analysis of a represen-tative ddGLF knockin line. In lanes 3–6, FK506 (1 �M) was added to parasitesgrown without drug, and samples were harvested immediately (0) or at 4, 8, or24 h thereafter. In lanes 7–10, parasites grown 24 h in the presence of FK506 (1�M)werewashedandsuspendedinfreshmedia lackingdrugandwereharvestedimmediately (0) or at 4, 8, or 24 h thereafter. (C) Quantitation of ddUGM (}) andLPG (bars) levels seen in the experiment shown in B.

7586 � www.pnas.org�cgi�doi�10.1073�pnas.0901698106 Madeira da Silva et al.

degradation by the proteasome (7). We show that this systemperforms as well as, and in some respects better than, thatdescribed in mammalian cells or Apicomplexan parasites (8, 9).The FKBP/Shld1 system functioned well in 2 Leishmania speciesrepresenting different subgenera, and it conferred strong ligand-dependent expression for all 4 parasite proteins tested (Fig. 1 andFig. S3). The kinetics of induction and degradation were alsofavorable, with full induction occurring within 8 h and lossoccurring within 2 h of drug removal (Fig. 3 and Table S1).Lastly, the utility of the system was shown by its ability to regulateexpression of the important surface glycoconjugate LPG (Fig. 5).

Hence, the ddFKBP system constitutes a promising moleculartool for the regulation of protein levels in L. major, with relativelyfast effects on protein levels upon addition and removal of rapa-mycin analogs. A second advantage arises from the property termed‘‘tunability,’’ in that the level of expression can be adjusted quan-titatively and homogeneously across the cell population by titrationof varying amounts of ligand (Fig. S1A) (7).

Rapamycin and FK506 Are Cost-Effective Ligands for the dd-InducibleSystem in Leishmania. Although Shld1 is nontoxic to cells from manyspecies tested in vitro thus far, and to mice as well, it is relativelyexpensive and of limited availability. Shld1 was 10-fold more potentin stabilization of the ddYFP fusion protein in L. major thanreported for mammalian cells (Fig. 2B and Table S1) (7). Althoughalternate FKBP ligands, such as rapamycin and FK506, are rela-tively inexpensive and available, in many species they show toxicitydue to inhibition of critical signaling pathways involving calcineurinand TOR kinases (38–40). However, Leishmania were relativelyinsensitive to FK506 and rapamycin (IC50: 7.7 and 4.9 �M), and fullinduction of ddYFP or ddLUC expression was achieved by 1 �M(Figs. 1 and 2 and Table S1).

The potent, efficacious properties of FK506 and rapamycin withthe ddFKBP system in Leishmania, in combination with theirrelative lack of toxicity in vitro, establish these as attractive alter-

natives to the more costly Shld1. Potentially, these ligands may beused in other organisms, depending on the relative potency ofddFKBP induction vs. toxicity. Although FK506 is highly toxic tomammalian cells, rapamycin has been used productively in mice invivo (41). However, it seems likely that their immunosuppressiveproperties may prove problematic because of the complex interac-tions of Leishmania parasites and their host immune system (42).

An Inducible System for LPG Expression. Expression of a ddUGMconstruct within the glf� mutant resulted in low levels of LPG in theabsence of ligand, but full LPG expression upon induction (Fig. 5).Robust LPG expression was induced at the lowest concentration ofShld1 tested (10 nM), at least 20-fold lower than required to getmaximal ddUGM expression (Fig. 5A). Remarkably, the low re-sidual levels of LPG expression in the uninduced glf�/�pIR1PHLEO-ddGLF line were sufficient to confer considerableresistance to lysis by complement, a sensitive, functional readout.Thus, we tested an alternative strategy, generating a ddGLFknockin at the normal GLF locus. As expected, the ddGLFknockins showed 10-fold lower levels of ddUGM expression in thepresence of ligand compared with the ddGLF episomal comple-mented line. Importantly, these showed undetectable ddUGM andLPG levels in the absence of ligand and were completely susceptibleto complement lysis. In contrast, in the presence of ligand, LPGlevels and complement resistance closely resembled that of WTparasites (Figs. 5 and 6). Together, these data confirm the bio-chemical and phenotypic ‘‘faithfulness’’ of the inducible system inits on and off states.

We anticipate that this inducible system will prove valuable infurther studies of the diverse roles of LPG in survival of theLeishmania parasite. These studies will need to take into accountthe kinetic properties of both ddUGM-dependent LPG synthesisand LPG shedding described here. LPG-dependent Leishmaniainteractions with its mammalian host take place rapidly afterdeposition by the sand fly vector, within a few minutes (complementlysis, macrophage uptake) to hours (inhibition of phagolysosomalfusion, differentiation), and are completed within 24 h, when theparasites have successfully established infection as amastigotes,which lack LPG (15, 43). The kinetics of ddUGM-dependent LPGappearance and loss, requiring upwards of 4 h for the significantgain or loss, may render its use in these settings problematic.Nonetheless, the tunability of the ddFKBP system would permitstudies of the quantitative dependency of LPG levels under thesecircumstances. In contrast, the role of LPG in the sand fly vectorshould be amenable to study by the ddUGM system, because theseLPG-dependent interactions span days, including initial survivalwithin the peritrophic matrix-enclosed blood meal, binding ofprocyclics to the parasite midgut lectin PpGalec, and release asinfective metacyclics (43).

We anticipate more generally that the properties of the ddFKBPregulatory system described here will be applicable to the study ofmany other parasite proteins in a variety of circumstances relevantto the parasite infectious cycle. Our studies identify several impor-tant variables to consider in the design of such strategies. First, asseen in other systems, there can be significant variation amongclonal lines, the origins of which are not always evident, suggestingthat screening at both the biochemical and biological levels for linesbehaving properly will be required. Second, it remains to bedetermined whether the destabilization module will function ef-fectively in cellular compartments other than the cytoplasm; forexample, the secretory pathway or organelles, such as the glyco-some or mitochondrion. Lastly, it is clear that genetic strategies thatminimize the level of expression of the protein in the uninducedstate—for example, using knockin rather than episomal or high-level expression vectors—may have the highest probability ofcrossing biologically relevant thresholds in the off state.

Fig. 6. Susceptibility to complement lysis. (A) Shown are WT (Left) and glf�

(Right). (B) glf�/pIR1PHLEO-ddGLF transfectants. (Left) Parasites grown 24 h in 1�M FK506. (Right) Parasites grown without ligand. (C) ddGLF knockin transfec-tants grown with or without drug as described for B. Complement lysis wasmeasured by flow cytometry, with increased fluorescence corresponding to PIuptake by lysed cells. Results are representative of 2 independent experiments.

Madeira da Silva et al. PNAS � May 5, 2009 � vol. 106 � no. 18 � 7587

MIC

ROBI

OLO

GY

Materials and MethodsFKBP dd Ligands. Shld1 was generously provided by Thomas Wandless (Stan-ford University, Stanford, CA). Rapamycin and FK506 were purchased from LCLaboratories.

Parasites. L. major Friedlin clone V1 (MHOM/IL/81/Friedlin) and LV39 clone 5(Rho/SU/59/P) were grown in M199-based medium (31). L. braziliensis (MHOM/BR/75/M2903) was grown in Schneider’s Insect Medium (Sigma–Aldrich) sup-plemented with 10% heat-inactivated FBS, 2 mM L-glutamine, 500 units ofpenicillin, and 50 �g/mL streptomycin. Metacyclics were isolated by negativeselection using peanut-agglutinin (PNA) (44).

Unless otherwise indicated, all experiments were performed with promas-tigotes during early to mid log phase of exponential growth (typically �8 �106 parasites per milliliter) because some proteins are extensively degraded instationary phase because of the activation of a lysosomal pathway (45). Theseinclude YFP and GFP but not DsRed2 or Luciferase.

DNA Constructs and Genetic Manipulation of Parasites. The description of themolecular constructs used in this work is presented in Figs. S5C and S6A, TableS2, and Table S3. DNAs were introduced directly or after digestion withappropriate restriction enzymes, followed by dephosphorylation. Transfec-tion was performed by using high-voltage electroporation (46), and clonaltransfectants were recovered after plating on semisolid media containing 1%agar, which contained the appropriate concentrations of selective drugs.

Flow Cytometry, Luciferase Assay, and Western Blotting. YFP expression wasdetermined by flow cytometry using a FACSCalibur cytometer (Becton

Dickinson) and quantitated by using CellQuest software (Becton-Dickinson). Luciferase activity was determined by suspending 106 cells in200 �L of M199 medium in a 96-well plate containing 1 �L of luciferin (30�g/mL). After 10 min of incubation, the plate was imaged by using aXenogen IVIS Photoimager, and luciferase activity was quantified as pho-tons per second (p/s). Preparation of samples for Western blotting and theantisera used are described in SI Text.

Complement Lysis Assay. Lysis by complement was assayed as described previ-ously (19).Briefly, late log-phasepromastigoteswerewashedonce inDMEM,and1 � 106 washed cells were then incubated at room temperature for 30 min with4% human serum in DMEM containing 0.4 �g/mL propidium iodide (PI), followedby flow cytometry.

ACKNOWLEDGMENTS. We thank T. J. Wandless for providing ddFKBPconstructs and Shld1 along with insightful advice and discussions, C. Arm-strong and D. Goldberg for discussions (Washington University School ofMedicine), S. J. Turco for LPG advice, D. McMahon-Pratt (Yale School ofPublic Health) for the L. braziliensis strain M2903, I. Wong (WashingtonUniversity School of Medicine) for providing the anti-histone H2A anti-body, K. Tinkum for help with initial constructs of ddYFP, T. Vickers(Washington University School of Medicine) and A. Li (Washington Uni-versity School of Medicine) for pXG-LmDHFR-TS, and L.-F. Lye and B.Anderson for comments on this manuscript. L.M.d.S. received a fellowshipfrom the Washington University Infectious Diseases Scholar Program. Thiswork was supported by National Institutes of Health Grants R01 AI29646,AI31078, and AI21903.

1. World Health Organization (2002) Urbanization: An increasing risk factor for leishmani-asis. Wkly Epidemiol Rec 77:365–370.

2. Beattie L, Evans KJ, Kaye PM, Smith DF (2008) Transgenic Leishmania and the immuneresponse to infection. Parasite Immunol 30:255–266.

3. Beverley SM (2003) Protozomics: Trypanosomatid parasite genetics comes of age. Nat RevGenet 4:11–19.

4. Yan S, Myler PJ, Stuart K (2001) Tetracycline regulated gene expression in Leishmaniadonovani. Mol Biochem Parasitol 112:61–69.

5. Campbell DA, Thomas S, Sturm NR (2003) Transcription in kinetoplastid protozoa: Why benormal? Microbes Infect 5:1231–1240.

6. Clayton CE (2002) Life without transcriptional control? From fly to man and back again.EMBO J 21:1881–1888.

7. Banaszynski LA, Chen LC, Maynard-Smith LA, Ooi AG, Wandless TJ (2006) A rapid, revers-ible, and tunable method to regulate protein function in living cells using synthetic smallmolecules. Cell 126:995–1004.

8. Armstrong CM, Goldberg DE (2007) An FKBP destabilization domain modulates proteinlevels in Plasmodium falciparum. Nat Methods 4:1007–1009.

9. Herm-Gotz A, et al. (2007) Rapid control of protein level in the apicomplexan Toxoplasmagondii. Nat Methods 4:1003–1005.

10. Naderer T, Vince JE, McConville MJ (2004) Surface determinants of Leishmania parasitesand their role in infectivity in the mammalian host. Curr Mol Med 4:649–665.

11. Beverley SM, Turco SJ (1998) Lipophosphoglycan (LPG) and the identification of virulencegenes in the protozoan parasite Leishmania. Trends Microbiol 6:35–40.

12. Kamhawi S, et al. (2004) A role for insect galectins in parasite survival. Cell 119:329–341.13. Spath GF, et al. (2003) Persistence without pathology in phosphoglycan-deficient Leish-

mania major. Science 301:1241–1243.14. Lodge R, Diallo TO, Descoteaux A (2006) Leishmania donovani lipophosphoglycan blocks

NADPH oxidase assembly at the phagosome membrane. Cell Microbiol 8:1922–1931.15. Lodge R, Descoteaux A (2005) Modulation of phagolysosome biogenesis by the lipophos-

phoglycan of Leishmania. Clin Immunol 114:256–265.16. Beverley SM, et al. (2005) Eukaryotic UDP-galactopyranose mutase (GLF gene) in microbial

and metazoal pathogens. Eukaryot Cell 4:1147–1154.17. Capul AA, Barron T, Dobson DE, Turco SJ, Beverley SM (2007) Two functionally divergent

UDP-Gal nucleotide sugar transporters participate in phosphoglycan synthesis in Leish-mania major. J Biol Chem 282:14006–14017.

18. Kleczka B, et al. (2007) Targeted gene deletion of Leishmania major UDP-galactopyranosemutase leads to attenuated virulence. J Biol Chem 282:10498–10505.

19. Spath GF, Garraway LA, Turco SJ, Beverley SM (2003) The role(s) of lipophosphoglycan(LPG) in the establishment of Leishmania major infections in mammalian hosts. Proc NatlAcad Sci USA 100:9536–9541.

20. Ilgoutz SC, McConville MJ (2001) Function and assembly of the Leishmania surface coat. IntJ Parasitol 31:899–908.

21. Turco SJ, et al. (1989) Structure of the phosphosaccharide-inositol core of the Leishmaniadonovani lipophosphoglycan. J Biol Chem 264:6711–6715.

22. Pedersen LL, Turco SJ (2003) Galactofuranose metabolism: A potential target for antimi-crobial chemotherapy. Cell Mol Life Sci 60:259–266.

23. Bakker H, Kleczka B, Gerardy-Schahn R, Routier FH (2005) Identification and partialcharacterization of two eukaryotic UDP-galactopyranose mutases. Biol Chem 386:657–661.

24. Zufferey R, et al. (2003) Ether phospholipids and glycosylinositolphospholipids are notrequired for amastigote virulence or for inhibition of macrophage activation by Leishma-nia major. J Biol Chem 278:44708–44718.

25. Spath GF, et al. (2000) Lipophosphoglycan is a virulence factor distinct from relatedglycoconjugates in the protozoan parasite Leishmania major. Proc Natl Acad Sci USA97:9258–9263.

26. Cardenas ME, et al. (1999) Antifungal activities of antineoplastic agents: Saccharomycescerevisiae as a model system to study drug action. Clin Microbiol Rev 12:583–611.

27. Cruz MC, et al. (1999) Rapamycin antifungal action is mediated via conserved complexeswith FKBP12 and TOR kinase homologs in Cryptococcus neoformans. Mol Cell Biol19:4101–4112.

28. Dumont FJ, Su Q (1996) Mechanism of action of the immunosuppressant rapamycin. LifeSci 58:373–395.

29. Meissner U, Juttner S, Rollinghoff M, Gessner A (2003) Cyclosporin A-mediated killing ofLeishmania major by macrophages is independent of reactive nitrogen and endogenousTNF-alpha and is not inhibited by IL-10 and 13. Parasitol Res 89:221–227.

30. Cruz A, Coburn CM, Beverley SM (1991) Double targeted gene replacement for creatingnull mutants. Proc Natl Acad Sci USA 88:7170–7174.

31. Kapler GM, Coburn CM, Beverley SM (1990) Stable transfection of the human parasiteLeishmania major delineates a 30-kilobase region sufficient for extrachromosomal repli-cation and expression. Mol Cell Biol 10:1084–1094.

32. Misslitz A, Mottram JC, Overath P, Aebischer T (2000) Targeted integration into a rRNAlocusresults inuniformandhigh levelexpressionoftransgenes inLeishmaniaamastigotes.Mol Biochem Parasitol 107:251–261.

33. Proudfoot L, Schneider P, Ferguson MA, McConville MJ (1995) Biosynthesis of the glyco-lipid anchor of lipophosphoglycan and the structurally related glycoinositolphospholipidsfrom Leishmania major. Biochem J 308:45–55.

34. King DL, Chang YD, Turco SJ (1987) Cell surface lipophosphoglycan of Leishmania dono-vani. Mol Biochem Parasitol 24:47–53.

35. McKean PG, Denny PW, Knuepfer E, Keen JK, Smith DF (2001) Phenotypic changesassociated with deletion and overexpression of a stage-regulated gene family in Leish-mania. Cell Microbiol 3:511–523.

36. Lincoln LM, Ozaki M, Donelson JE, Beetham JK (2004) Genetic complementation ofLeishmania deficient in PSA (GP46) restores their resistance to lysis by complement. MolBiochem Parasitol 137:185–189.

37. Joshi PB, Kelly BL, Kamhawi S, Sacks DL, McMaster WR (2002) Targeted gene deletion inLeishmania major identifies leishmanolysin (GP63) as a virulence factor. Mol BiochemParasitol 120:33–40.

38. Liu J, et al. (1991) Calcineurin is a common target of cyclophilin-cyclosporin A and FKBP-FK506 complexes. Cell 66:807–815.

39. Heitman J, Movva NR, Hall MN (1991) Targets for cell cycle arrest by the immunosuppres-sant rapamycin in yeast. Science 253:905–909.

40. Sabers CJ, et al. (1995) Isolation of a protein target of the FKBP12-rapamycin complex inmammalian cells. J Biol Chem 270:815–822.

41. Luker KE, et al. (2004) Kinetics of regulated protein-protein interactions revealed withfirefly luciferase complementation imaging in cells and living animals. Proc Natl Acad SciUSA 101:12288–12293.

42. Sacks D, Noben-Trauth N (2002) The immunology of susceptibility and resistance toLeishmania major in mice. Nat Rev Immunol 2:845–858.

43. Sacks DL, et al. (2000) The role of phosphoglycans in Leishmania-sand fly interactions. ProcNatl Acad Sci USA 97:406–411.

44. Sacks DL, Perkins PV (1984) Identification of an infective stage of Leishmania promastig-otes. Science 223:1417–1419.

45. Waller RF, McConville MJ (2002) Developmental changes in lysosome morphology andfunction in Leishmania. Int J Parasitol 32:1435–1445.

46. RobinsonKA,BeverleySM(2003) Improvements in transfectionefficiencyandtestsofRNAinterference (RNAi) approaches in the protozoan parasite Leishmania. Mol BiochemParasitol 128:217–228.

7588 � www.pnas.org�cgi�doi�10.1073�pnas.0901698106 Madeira da Silva et al.

Supporting InformationMadeira da Silva et al. 10.1073/pnas.0901698106SI TextWestern Blot Analysis. Promastigote lysates were prepared bysuspending parasites rapidly in 1� Laemmli loading buffer andimmediately boiling for 5 min. Samples (2 � 106 cell equivalentsper lane) were subjected to SDS/PAGE and electrotransferredto nitrocellulose membranes (Hybond-ECL; Amersham Bio-sciences). Membranes were blocked with 5% nonfat dry milk inTBST (20 mM Tris-base, pH 7.6; 140 mM sodium chloride; and0.05% Tween 20) for 1 h at room temperature. Protein loadingwas assessed with a rabbit anti-L. major histone H2A polyclonalantibody at 1:100,000 dilution. Polyclonal rabbit antibody anti-GFP (Abcam) was used at 1:2,500 dilution, and anti-FKBP12(Affinity BioReagents) was used at 1:1,000 dilution; monoclonalantibody WIC79.3 (1) was used at 1:2,000 dilution. Goat anti-rabbit IgG HRP-conjugated secondary antibody (Jackson Im-munoResearch) was used at 1:10,000; anti-mouse IgG HRP-conjugated secondary antibody (Amersham Biosciences) wasused at 1:5,000. ECL reactions were revealed with a PerkinElmerLife Sciences chemiluminescence kit. Quantitative Western blot-ting with WIC79.3 was performed with the Odissey InfraredImaging System (LI-COR Biosciences) using the manufacturer’sblocking buffer and goat anti-mouse IRDye 800CW secondaryantibody (Li-Cor Biosciences). Quantitation was done with theMulti Gauge V3.0 software (Fujifilm).

ddFKBP Constructs. The oligonucleotides, templates, vectors, andconstructs used or developed are described briefly below andwith more detail in Table S2 and Table S3. All relevant regionsof each construct were confirmed by sequencing and/or func-tional assays.

ORFs of interest were amplified by PCR and inserted intopGEM-T-Easy (Promega) by TA cloning according to themanufacturer’s instructions. The YFP and ddYFP ORFs werethen inserted into the Leishmania expression vectorpIR1PHLEO (strain B4054) directly after digestion withBamHI, yielding pIR1PHLEO-ddYFP (strain B6145) andpIR1PHLEO-YFP (strain B6146).

To generate a general plasmid as an entry expression vectorto fuse other genes of interest with the destabilizing domain, weamplified the ddFKBP segment with primers SMB3250 andSMB3251 and inserted the fragment in pGEM-T-Easy to gen-erate pGEM-ddFKBP (strain B6177). The ddFKBP fragmentwas released by digestion with BclI plus BglII (by using a plasmidpreparation obtained with a dam-Escherichia coli strain,SCS110) and ligated in the BamHI site of pIR1PHLEO, gener-ating pIR1PHLEO-ddFKBP (strain B6182), which has conve-nient unique BamHI, SpeI, and XbaI sites into which ORFs canbe inserted in frame (Fig. S5C). ORFs from the reporterluciferase (amplified from pGL3 Basic from Promega, whichlacks the peroxisome-targeting C-terminal tripeptide) and sev-eral L. major genes (FTL: LmjF30.2600; DHCH1: LmjF26.0320;DHFRTS: LmjF06.0860; GLF: LmjF18.0200) were inserted intothe BamHI site of B6182 to create pIR1PHLEO-ddFTL (strainB6247), pIR1PHLEO-ddDHCH (strain B6227), pIR1PHLEO-ddDHFRTS (strain B6332), and pIR1PHLEO-ddGLF (strainB6308) (Table S3).

Targeted Gene Replacement of L. major GLF. The 5� GLF f lankingregion was amplified by PCR from L. major LV39 genomic DNAwith primers SMB2188 and SMB2189, and the amplicon wasdigested by SalI plus SpeI and inserted into similarly digestedpT7Blue (Novagen), generating plasmid B5327. The 3� GLF

f lanking region was amplified by PCR with oligonucleotidesSMB2190 and SMB2191, digested with SpeI plus KpnI, andinserted into similarly digested B5327, generating plasmidB5352. The HYG and PAC ORFs were then inserted between theinternal SpeI/BamHI sites of B5352, yielding the desired GLFreplacement constructs (B5356 and B5357, respectively). Theseconstructs were digested with HindIII plus KpnI to yield theappropriate targeting fragment before transfection.

The first GLF allele was inactivated by using the GLF::HYGtargeting construct. L. major LV39 clone 5 (Rho/SU/59/P)promastigotes were transfected by electroporation (2), andclonal transfectant lines were recovered after plating on 50�g/mL hygromycin B. Multiple clonal lines showing the expectedreplacement by appropriate PCR tests were identified(GLF/GLF::�HYG). Several of these were infected into suscep-tible BALB/c mice at high inoculums (107). These showed lesiondevelopment typical of WT parasites, and after 1 month para-sites were recovered. Clone H1 was chosen for targeting of theremaining GLF allele with the GLF::PAC replacement con-struct, and transfectants were recovered after plating on semi-solid media containing 20 �g/mL puromycin and 50 �g/mLhygromycin B. Several clones were analyzed with regard tocorrect integration of the replacement cassette and the lack ofthe GLF ORF by PCR. Several homozygous replacements wereobtained, and one of these was chosen for future work(�GLF::HYG/�GLF::PAC clone H1P1, referred to as glf� here-after). This line was then transfected with pIR1PHLEO-ddGLF,yielding glf�/�pIR1PHLEO-ddGLF, which expresses theddGLF from the multicopy episomal pIR1PHLEO vector.

Generation of Knockin ddGLF L. major. To facilitate construction ofddFKBP knockin constructs, we created a generic ddFKBPknockin cassette with 3 key elements: a drug resistance marker(PHLEO), the DST gene intergenic region (DST IR), which hasbeen extensively used to drive expression of downstream genesin Leishmania (3), and the ddFKBP sequence (strain B6323). Wefirst digested pGEM-ddFKBP (strain B6177) with BclI plus BglIIto release the ddFKBP fragment and ligated it in the BamHI siteof pXGPHLEO (strain B3324), generating pXGPHLEO-ddFKBP (strain B6319). This plasmid was then used as templatein PCR amplification with primers SMB2563 and SMB3251 toobtain the PHLEO-DST-ddFKBP fragment, which was cloned inpGEM-T-Easy, yielding pGEM-PHLEO-DST IR-dd (strainB6323; Fig. S6A). This is a generic knockin cassette; for anygiven gene of interest (GOI), it is necessary to insert the 5�f lanking sequence of the GOI ORF upstream of the PHLEOresistance marker (site A) and insert the ORF of the GOI inframe with the ddFKBP sequence (site B) (Fig. S6A).

To create the ddGLF knockin fragment, we released the 5�f lank of GLF used previously in the knockout cassette (strainB5327) with SalI plus SpeI digestion and inserted it into pGEM-PHLEO-DST IR-dd (strain B6323), which had been digestedsimilarly, generating pGEM-5�UTR GLF-PHLEO-DST IR-dd(strain B6331). Finally, we inserted the ORF of GLF in theBamHI of B6331 in frame with the ddFKBP sequence, creatingpGEM-kiddGLF (strain B6336). This targeting construct wasreleased after digestion with NdeI plus SacII and was dephos-phoryated before use.

The heterozygous L. major �/glf�HYG mutant clone H1above was transfected with the knockin cassette by electropo-ration, and parasites were plated on semisolid M199 mediumagar plates containing 50 �g/mL hygromycin, 10 �g/mL phleo-

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 1 of 11

mycin, and 2 �M FK506 to maintain ddGLF activity. Severalcolonies were picked and grown in liquid medium with 50 �g/mLhygromycin, 10 �g/mL phleomycin, and 1 �M FK506; correct

integration of the cassette in the transfectants was verified byPCR. Parasites were cultivated in the presence of 1 �M FK506at all times unless otherwise indicated.

1. Kelleher M, Bacic A, Handman E (1992) Identification of a macrophage-binding de-terminant on lipophosphoglycan from Leishmania major promastigotes. Proc NatlAcad Sci USA 89:6–10.

2. Robinson KA, Beverley SM (2003) Improvements in transfection efficiency and tests ofRNA interference (RNAi) approaches in the protozoan parasite Leishmania. Mol Bio-chem Parasitol 128:217–228.

3. Ha DS, Schwarz JK, Turco SJ, Beverley SM (1996) Use of the green fluorescent proteinas a marker in transfected Leishmania. Mol Biochem Parasitol 77:57–64.

4. Banaszynski LA, Chen LC, Maynard-Smith LA, Ooi AG, Wandless TJ (2006) A rapid,reversible and tunable method to regulate protein function in living cells usingsynthetic small molecules. Cell 126:995–1004.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 2 of 11

A

B

Fig. S1. Exploration of parameters relevant to the use of the ddFKBP regulatory system in L. major. (A) Tunability of the ddFKBP system. L. major transfectantsexpressing ddYFP (SSU::IR1PHLEO-ddYFP) were treated with different concentrations of Shld1 (5 nM, 15 nM, 50 nM, 250 nM, and 1 �M) for 24 h, and YFP fluorescencelevels were measured by flow cytometry. Solvent (ethanol or EtOH)-treated cells were used as negative control. The figure shows strong, homogeneous expressionincreasing with increasing ligand concentration, the signature of tunability (4). (B) Effect of Shld1, FK506, and rapamycin on metacyclogenesis of L. major. To determinethe effect of these compounds on metacyclogenesis, promastigotes were incubated with different concentrations of Shld1, FK506, or rapamycin until stationary day2, and metacyclics were isolated by PNA methodology. Error bars represent SEM of 2 independent experiments performed in duplicate.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 3 of 11

Rap.

t1 t2 t8 t24

t1 t2 t8 t24

t0

DMSO

time(h)a

FK506

t1 t2 t8 t24

t1 t2 t8 t24

t1 t2 t8 t24

t0 YFP

EtOHShld1

anti-histone

anti-GFPddYFPYFP

a after addition of drug

FK506

t0 t0.5

t1 t2 t4 t8 t24

t0 t0.5

t1 t2 t4 t8 t24

EtOH

t0 t0.5

t1 t2 t4 t8 t24

WT

YFPtime(h)b

Shld1 Rap.

t0 t0.5

t1 t2 t4 t8 t24

t0 t0.5

t1 t2 t4 t8 t24

DMSO

anti-histone

anti-GFP

b after removal of drug

A

B

Fig. S2. Western blotting with anti-GFP antibodies confirms results obtained by flow cytometry (Fig. 3). (A) Stabilizing agents Shld1, FK506, or rapamycin wereadded to log phase L. major promastigote cultures, and samples were withdrawn at the time points indicated for Western blotting. (B) After a 24-h incubationwith the indicated agents, parasites were washed and inoculated into media lacking drugs. Samples were collected at the indicated times and analyzed byWestern blotting.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 4 of 11

ddFTL

WT L. braziliensis M2903

ddFTL

30 -

KDa

FK506:ddYFP

ddYFP

anti-FKBP12

- + - + - +

57 -

ddDHCH1

ddDHCH1

B

A

C

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

WT L

bEtO

H

FK-506

Lu

cif

era

se

ac

tiv

ity

(p

/s)

WT L. braziliensis ddLUC

ddDHCH175 -50 -37 -

25 -

anti-histone

150 -

anti-FKBP12

ddDHFR-TSddFTL

WT L. major FV1

FK506: - + - + - +ddDHCH1

ddFTL

ddDHFR-TS

Fig. S3. Regulation of Leishmania proteins fused to the ddFKBP. (A) L. major transfectants: ddDHCH1, SSU::IR1PHLEO-ddDHCH1; ddFTL, SSU::IR1PHLEO-ddFTL; ddDHFR-TS,SSU::IR1PHLEO-ddDHFR-TS. Parasites were grown in logarithmic growth phase for 24 h in the presence/absence of 1 �M FK506 and were subjected to Western blotting withanti-FKBP antisera (Upper) or anti-histone H2B (loading control; Lower). The arrows mark the expected mobilities of the various dd fusion proteins. (B) Western blotting of L.braziliensispromastigotetransfectantstreatedwithFK506asdescribedinA.ddDHCH1,SSU::IR1PHLEO-ddDHCH1;ddFTL,SSU::IR1PHLEO-ddFTL;ddYFP,SSU::IR1PHLEO-ddYFP.(C) Luciferase activity of L. braziliensis SSU::IR1PHLEO-ddLUC transfectant promastigotes incubated 24 h with 1 �M FK506. We note that the luciferase cassette used here lacksthe peroximal targeting signal, which in trypanosomatids targets proteins to the related organelle termed the glycosome.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 5 of 11

Fig. S4. Generation of L. major glf � null mutants. (A) Representation of the replacement targeting fragments used. (B) PCR analysis of a representative glf �

mutant. Primers 1 and 2 together and primers 3 and 4 together establish the 5� and 3� sides of the PAC replacement; primers 1 and 5 together and 6 and 4 togetherestablish the 5� and 3� sides of the HYG replacement; and primers 7 and 8 together confirm the presence or absence of the GLF ORF. Primers: 1, SMB3541; 2,SMB2889; 3, SMB2888; 4, SMB3542; 5, SMB2566; 6, SMB2565; 7, SMB2192; and 8, SMB3495 (Table S1). (C) LPG expression. (Upper) L. major phosphoglycans werevisualized by Western blotting with WIC79.3 of whole-cell lysates of logarithmic-phase promastigotes (2 � 106 cell equivalents per lane). (Lower) shows a Westernblot of these extracts with anti-L. major histone H2A as a loading control. Parasite lines were WT, a representative glf � mutant, and the lpg2� mutant describedpreviously, which lacks all phosphoglycans [Spath GF, Garraway LA, Turco SJ, Beverley SM (2003) Proc Natl Acad Sci USA 100:9536–9541].

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 6 of 11

37 -50 -75 -

25 -20 -

150 -

8 9 10 11 22 23

glf-/+pIR1PHLEO-ddGLF

8 9 10 11 22 23EtOH FK-506

KDa

anti-histone

anti

-FK

BP

ddUGM

AW

IC79

.3

150 -75 -50 -37 -

25 -20 -

8 9 10 11 22 23

glf-/+pIR1PHLEO-ddGLF

8 9 10 11 22 23EtOH FK-506

1313

anti-histone

LPG

PPG

B

Swa I

BamHISpe IXba I

ISwa

L. major SSU'

DST IR

dd'

CYS2

LPG1 IR

PhleoR

1.7K IR

L. major 'SSU

ampR

Unique sites for cloning genes of interest

SwaI sites are digested for integration into the SSU locus

pIR1-PHLEO-dd(B6182)

C

Fig. S5. Complementation of glf� knockout with episomal ddUGM expression. (A and B) Levels of ddUGM and LPG in 6 clonal glf�/�pIR1PHLEO-ddGLFtransfectants. Western blotting was performed with 3 � 106 cell equivalents per lane and using anti-FKBP12 (A) or WIC79.3 (B). The bars to the right of B markreactivity associated with PPG or LPG, whereas the arrow in A marks the mobility of the ddUGM fusion protein. In each panel, the lower portion shows a Westernblot with anti-L. major histone H2A polyclonal antibodies as a protein loading control. EtOH, cells treated with diluent (ethanol); FK506, cells treated with 1 �MFK506. Numbers 8, 9, 10, 11, 22, and 23 correspond to the identification numbers of the transfectants, which were chosen randomly. (C) Map of pIR1-Phleo-ddFKBP (B6182), a general Leishmania expression vector that can be used to tag GOIs with an N-terminal ddFKBP. This vector may be used to generate episomaltransfectants using the intact plasmid, or it may be integrated by homologous recombination into the SSU rRNA locus after digestion with SwaI to exposehomologous ends. Red arrows represent ORFs of genes coding for selectable markers (resistance to ampicillin and phleomycin) and the FKBP dd. Blue depictsintergenic regions that drive expression of the dd-tagged protein and phleomycin resistance drug marker. Black arrows represent fragments of SSU locus thatare used for homologous recombination to generate integrated transfectants with high protein expression levels.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 7 of 11

Site B:Insert the ORF of your GOI in frame with the ddFKBP sequence

INdeISalINot

RIEcoISpe

HIBamIIBglRIEcoIISacINot

amp R

S6 RNA Pol promoter

PHLEO

5' DST IRdd

Site A:Insert the 5’flank

of your GOI

pGEM-Phleo-DST IR-dd(B#6323)

A

Fig. S6. Knockin strategy for in situ tagging of genes of interest (GOI) in L. major and its application to GLF. (A) Map of pGEM-Phleo-DST IR-ddFKBP (strain B6323), sourceof a useful general ddFKBP knockin cassette facilitating the tagging of genomic copies of GOIs with an N-terminal ddFKBP. To customize the cassette for a specific GOI, it isnecessary to insert the 5� flanking sequence of the GOI at the left position of the Phleo resistance marker (site A) and the ORF of the GOI in frame with the ddFKBP sequence(site B). (B–D) Generation of ddGLF knockin L. major. (B) Schematic representation of the knockin cassette and WT GLF locus. (C) PCR confirmation of successful integrationoftheddGLFknockincassette.Primers1and2togetherestablishthepresenceofthePHLEOmarker;primers2and3togetherandprimers4and5togetherconfirmthecorrectintegration of the knockin cassette from the 5� and 3� sides, respectively; and primers 6 and 5 together confirm the presence of the GLF ORF. Primers: 1, SMB2563; 2, SMB2564;3, SMB3541; 4, SMB3250; 5, SMB3542; and 6, SMB2186 (see Table S2 for sequences). (D) Expression of ddUGM fusion proteins (Upper) or LPG (Lower) by Western blotting withanti-FKBP antisera or monoclonal antibody WIC79.3, respectively. Parasite lines include WT, glf� and lpg2� controls, 1 representative glf�/�pIR1PHLEO-ddGLF transfectant,and 2 representative ddUGM knockins. Samples are whole-cell lysates of logarithmic-phase promastigotes (3 � 106 cell equivalents per lane). A minus or plus sign indicatesgrowth for 24 h in the presence of 1 �M FK506. A constant cross-reacting band with a mobility slightly below that of the ddUGM fusion protein is evident in the anti-FKBP12blots and serves as a useful loading control; this band is seen in some but not all experiments (for example, compare D with Fig. S5A).

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 8 of 11

Table S1. Summary comparing the effects of Shld1, FK506, and rapamycin on the growth and expression of a ddYFP reporter inSSU::IR1PHLEO-ddYFP transfectant L. major promastigotes

CompoundEC50 half-maximum effective

concentration, nMEC100 maximum effective

concentration, �MTime for full

induction, h (ON)Time to decay 50%, h

(OFF)IC50 growth

inhibition, �M

Shld1 10 0.25 8 0.5 8.4 � 0.3*FK506 200 1 8 0.5 7.7 � 0.2Rapamycin 60 1 8 1.25 4.9 � 0.5

*The highest concentration of Shld1 was 1 �M because of limited availability and expense.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 9 of 11

Table S2. Sequences of oligonucleotides

Primer no. Sequence (5� to 3�)

Primers used in destabilizing domain constructs

SMB3225 GGATCC GCCACCATGGGAGSMB3226 GGATCCCCACCATGGTGAGCAAGGGCGAGGAGSMB3227 GGATCCGAGTTAGTCGAGTGCGTAGTCTGGTACSMB3250 TGATCACCACCATGGGAGTGCAGGSMB3251 AGATCTGGATCCTTCCGGTTTTAGAAGCTCCACATCSMB3543 AGATCTATGGAAGACGCCAAAAACATSMB3130 TTAGATCTTTACACGGCGATCTTTCCGCSMB3573 GGATCCATGGCCACCCGGAAGCTGCASMB3574 GGATCCTTACGAAAGACCCACAATCCSMB3575 GGATCCATGCCGTCTGCTCAGATCATSMB3576 GGATCCCTATGATACGCCCAACGCAGSMB3494 GGATCCACCATGtacccatacgatgttccagattacgctAGCGCTGACAAGGTGGTCATAATCSMB2180 GCATGCGGATCCTACGAGGCCGTCGACGACCATGTGCASMB3464 GGATCCATGTCCAGGGCAGCTGCGAGGTTSMB3465 GGATCCCTATACGGCCATCTCCATCTTGA1

Primers used to construct the gene replacement knockout cassette for L. major GLF

SMB2188 ATAGGCGTCGACGATGAGCGCAGGGGACGAGCASMB2189 ACTGACTAGTGATGGATTTGCTGCGTGTGCCTGCGTSMB2190 ATTATACTAGTGCGGCCGCGGATCCAAGGTCGGCAGCCATGGACGASMB2191 GGGGAAGGTACCGCGGCTACATGCAGATCGTC

Primers used to construct the knockin cassette for L. major GLF

SMB2563 GGTAACGGTGCGGGCTGACGCCACCATGGCCAAGTTGACCAGTGCPrimers used to confirm integration of GLF gene replacement knockout and dd knockin cassettes

SMB3541 TGTTACGGTGATGCGCTAAGSMB3542 CTCGTCCCCATTTCATCAAGSMB2888 ACGTCGAGTGCCCGAAGGACSMB2889 ACCGTGGGCTTGTACTCGGSMB2565 AATACGAGGTCGCCAACATCSMB2566 GAAAGCACGAGATTCTTCGCSMB2192 CCGCTGTGAACGTGGAGCGCATCASMB3495 GGATCCCGAGGCCGTCGACGACCATGSMB2564 CGAGATCCCACGTAAGGTGCTCAGTCCTGCTCCTCGGCCACGAAGSMB 2186 CGCGTAACCAAGGTGAACCCGATCACGAAGA

Underlining indicates restriction sites. Boldface indicates Kozak consensus sequence. Lowercase, HA tag.

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 10 of 11

Table S3. Constructs used in this work

Gene Template Oligos used for PCR pGEM-T-easy intermediateLeishmania expression

construct

YFP pBMN-FKBP(L106P)-YFP-HA(4) SMB3226 and SMB3227 pGEM-YFP (strain B6143) pIR1PHLEO-YFP (strainB6146)

ddYFP pBMN-FKBP(L106P)-YFP-HA (4) SMB3225 and SMB322 pGEM-ddYFP (strain B6142 pIR1PHLEO-ddYFP(strain B6145)

ddLUC pGL3 Basic (Promega) SMB3543 and SMB3130 pGEM-LUC (strain B6298) pIR1PHLEO-ddLUC(strain B6299)

ddDHCH1 L. major FV1 genomic DNA SMB3575 and SMB3576 pGEM-DHCH1 (strain B6353) pIR1PHLEO-ddDHCH1(strain B6227)

ddFTL L. major FV1 genomic DNA SMB3573 and SMB3574 pGEM-FTL (strain B6246) pIR1-PHLEO-ddFTL(strain B6247)

ddDHFR-TS pXG-LmDHFR-TS (B6098) (providedby Angela Li and Tim Vickers)

Plasmid was digested withBamHI for subcloninginto pIR1PHLEO-ddFKBP(B6182)

none pIR1PHLEO-ddDHFR-TS(strain B6332)

ddGLF L. major FV1 genomic DNA SMB3494 and SMB2180 pGEM-HA-GLF (strain B6307) pIR1PHLEO-ddGLF(strain B6308)

Madeira da Silva et al. www.pnas.org/cgi/content/short/0901698106 11 of 11