Arginase plays a pivotal role in polyamine precursor metabolism in Leishmania. Characterization of gene deletion mutants.

The polyamine pathway of protozoan parasites has been successfully targeted in anti-parasitic therapies and is significantly different from that of the mammalian host. To gain knowledge into the metabolic routes by which parasites synthesize polyamines and their precursors, the arginase gene was cloned from Leishmania mexicana, and Deltaarg null mutants were created by double targeted gene replacement and characterized. The ARG sequence exhibited significant homology to ARG proteins from other organisms and predicted a peroxisomal targeting signal (PTS-1) that steers proteins to the glycosome, an organelle unique to Leishmania and related parasites. ARG was subsequently demonstrated to be present in the glycosome, whereas the polyamine biosynthetic enzymes, in contrast, were shown to be cytosolic. The Deltaarg knockouts expressed no ARG activity, lacked an intracellular ornithine pool, and were auxotrophic for ornithine or polyamines. The ability of the Deltaarg null mutants to proliferate could be restored by pharmacological supplementation, either with low putrescine or high ornithine or spermidine concentrations, or by complementation with an arginase episome. Transfection of an arg construct lacking the PTS-1 directed the synthesis of an arg that mislocalized to the cytosol and notably also complemented the genetic lesion and restored polyamine prototrophy to the Deltaarg parasites. This molecular, biochemical, and genetic dissection of ARG function in L. mexicana promastigotes establishes: (i) that the enzyme is essential for parasite viability; (ii) that Leishmania, unlike mammalian cells, expresses only one ARG activity; (iii) that the sole vital function of ARG is to provide polyamine precursors for the parasite; and (iv) that ARG is present in the glycosome, but this subcellular milieu is not essential for its role in polyamine biosynthesis.

Leishmania is a genus of protozoan parasite that is the causative agent of leishmaniasis, a spectrum of devastating and potentially deadly diseases that affects ϳ12 million people worldwide. The parasite exhibits a digenetic life cycle in which the extracellular promastigote form resides in the phlebotomine sandfly vector, whereas the intracellular amastigote inhabits the phagolysosomes of macrophages from the infected mammalian host. Because there are no vaccines available to prevent leishmaniasis, chemotherapy offers the only avenue to combat the disease. Unfortunately, the current arsenal of drugs used to treat leishmaniasis is far from ideal, mainly because of toxicity and therapeutic unresponsiveness. Thus, the identification, characterization, and validation of novel therapeutic targets are urgently needed.
One biochemical pathway that has been successfully exploited for the treatment of parasitic disease is that for polyamine biosynthesis. Polyamines are ubiquitous organic cations found in virtually every eukaryotic cell and play critical roles in key cellular processes such as growth, differentiation, and macromolecular biosynthesis (1,2). DL-␣-Difluoromethylornithine (DFMO), 1 an irreversible inhibitor of ornithine decarboxylase (ODC), the first enzyme in polyamine synthesis, is capable of eradicating Trypanosoma brucei infections in mice (3) and patients with late stage African sleeping sickness (4,5). DFMO is also active against many other parasites (6,7), including the promastigotes of Leishmania donovani (8). Interestingly, the selectivity of DFMO for the metabolic machinery of trypanosomes is not based on dissimilar DFMO binding affinities but rather to differences in ODC turnover rates between parasites and mammals (9 -11). In addition, inhibitors of S-adenosylmethionine decarboxylase (ADOMETDC), the enzyme that provides aminopropyl moieties for spermidine and spermine synthesis, are also effective anti-trypanosomal agents (12,13).
Leishmania sp. have served as particularly valuable model systems for dissecting metabolic pathways in protozoan parasites because of their ability to proliferate axenically in defined growth medium and the facility by which their genome is genetically tractable (14 -17). All the genes of the polyamine pathway, ODC, ADOMETDC, and spermidine synthase (SPDSYN), have been cloned from L. donovani, and the creation of ⌬odc, ⌬adometdc, and ⌬spdsyn knockouts by targeted gene replacement has demonstrated the essential role of each of these enzymes in L. donovani promastigote proliferation and revealed significant dissimilarities between the polyamine biosynthetic pathways of this genus of protozoan parasite and the mammalian host (18 -20).
Despite the plethora of molecular and biochemical studies on the polyamine biosynthetic pathway of T. brucei, L. donovani, and other parasites (21)(22)(23)(24)(25), little is known about the metabolic avenues by which polyamine precursors are produced. Ornithine, the amino acid from which polyamines are generated, is produced from arginine in mammalian cells by two genetically and biochemically distinct arginase (ARG) enzymes. An ARG activity has also been detected in Leishmania (26), and an arginase gene sequence from Leishmania amazonensis has been reported, although neither the gene nor protein was functionally characterized (27). ARG has also been touted as a potential antileishmanial drug target, because n-hydroxyarginine, an inhibitor of ARG that is produced by macrophages as an intermediate during the formation of nitric oxide, can reduce polyamine levels in Leishmania amastigotes and lowers parasite loads (28).
To initiate an investigation into the pathways by which polyamine precursors are synthesized and to begin a validation of ARG as a potential therapeutic target, we have cloned arginase from L. mexicana and generated ⌬arg knockouts by double targeted gene replacement. The characterization of these gene deletion mutants reveals that the arginase is essential for promastigote viability and that the lethality of the null mutation could be conditionally bypassed by either low concentrations of putrescine, high concentrations of ornithine or spermidine, or episomal complementation. Furthermore, we have established that the L. mexicana ARG is localized to the glycosome, a unique organelle found exclusively in Leishmania and close relatives (29,30), and that this localization is mediated by the peroxisomal targeting signal type 1 (PTS-1). However, genetic complementation demonstrated that this glycosomal venue is not essential for ARG function, at least in promastigotes. . Parasites were cultivated in DME-L, a completely defined Dulbecco's modified Eagle-based medium that was specifically developed for growing Leishmania promastigotes (33). Transfected parasites were maintained in a modified DME-L medium, DME-L-CS, in which the bovine serum albumin component of DME-L was replaced with 10% chicken serum to avert polyamine oxidasemediated polyamine toxicity (8). arginase/arg heterozygotes were maintained continually in 50 g/ml hygromycin, whereas ⌬arg homozygotes were grown in DME-L-CS supplemented with 50 g/ml phleomycin, 50 g/ml hygromycin, and either 200 M ornithine and/or 200 -500 M putrescine.

Materials
Parasite growth experiments were initiated at 1.0 ϫ 10 5 parasites/ ml, and parasites were enumerated either every 24 or 48 h by hemacytometer or using the vital dye alamarBlue TM (BioSource International, Camarillo, CA) technology (34). Reduction of alamarBlue TM was measured at 570 and 600 nm on a Multiskan Ascent plate reader (Thermo Labsystems, Vantaa, Finland). The percent reduction of dye was calculated according to a formula published in the manufacturer's brochure. The greatest reduction was expressed as maximal proliferation, and growth was then plotted as a function of nutrient concentration.
Cloning of L. mexicana ARG-The arginase sequence from L. amazonensis (GenBank TM accession number AF038409) was used to design primers to amplify the ARG from L. mexicana genomic DNA using the PCR. Genomic DNA was isolated by standard protocols. The sense primer, 5Ј-GGATCCATGGAGCACGTGCAGTACAAGTTC-3Ј, encompassed the initiation methionine codon (boldface type) and was preceded by a BamHI restriction site (underlined), whereas the antisense primer, 5Ј-TCTAGACTACAGCTTGGAGCTCGTATGCGGAGT-3Ј, encoded the termination codon (boldface type) to which a BglII site (underlined) was attached at the 5Ј-end. The arginase open reading frame (ORF) was amplified from 20 ng of L. mexicana DNA using a high fidelity polymerase (Advantage HF 2 DNA polymerase, BD Bioscience, Palo Alto, CA) and standard PCR conditions for amplifying sequences from genomic DNA (95°C for 5 min followed by 32 cycles of 95°C for 1 min, 55°C for 1 min, and 68°C for 2 min). The ϳ1.0-kb-amplified DNA fragment was subcloned using the pCR® 2.1-TOPO® vector of the TOPO TA Cloning TM kit (Invitrogen), and limited nucleotide sequencing confirmed the identity of the putative arginase fragment. The resulting plasmid was designated TOPO TA-arginase. The subcloned arginase fragment was then used to screen an L. mexicana cosmid library constructed by Dr. Scott M. Landfear (Oregon Health & Science University, Portland, OR) using high stringency conditions. Positive cosmids were subjected to two rounds of purification, and the cosmid DNA isolated by standard methods. The arginase locus within a positive cosmid was mapped by Southern blotting, and a ϳ7-kb BamHI fragment encompassing the arginase ORF was subcloned into pBluescript (Stratagene, La Jolla, CA). The resulting plasmid was designated pBluescript-arginase and submitted for nucleotide sequencing. The entire nucleotide sequences of the arginase ORF (both directions), and ϳ1 kb of each of the flanks was obtained.
Construction of pXG-GFPϩ2Ј-ARG-To localize ARG, the gene was inserted into pXG-GFPϩ2Ј, a leishmanial expression vector that confers resistance to G418 and synthesizes foreign protein as an NH 2terminal fusion with green fluorescent protein (GFP). The arginase ORF was excised from the TOPO TA-arginase vector with BamHI and ligated into the BamHI cut pXG-GFPϩ2Ј vector. Sequence analysis verified the correct orientation and reading frame of the arginase coding region. The resulting expression plasmid was designated pXG-GFPϩ2Јarginase. This plasmid was then transfected into wild type L. mexicana using standard electroporation conditions (17,18). Transfected parasites were selected in 20 g/ml G418 and used for localization studies.
Fluorescence Microscopy-Lab-Tek® II Chambered Coverglass slides (Fisher Scientific) were coated with a 1:10 dilution of poly-L-lysine (Sigma-Aldrich Corp.) for 15 min. The chambers were rinsed with double-deionized water to remove excess poly-L-lysine and allowed to dry at room temperature. 1 ϫ 10 7 Leishmania promastigotes were resuspended in 1.0 ml of phosphate-buffered saline (PBS), pipetted into the chamber, and allowed to attach for 15 min. Chambers were rinsed once and then overlaid with 500 l of PBS. Images were taken on a Zeiss Axiovert 200 inverted microscope and deconvolution performed using the constrained iterative method by Axiovision 3.1 software (Carl Zeiss Optical, Chesterfield, VA).
Immunofluorescence Microscopy-L. donovani promastigotes at a density of ϳ5 ϫ 10 6 /ml were pelleted by centrifugation at 3000 ϫ g for 10 min, rinsed with PBS, and affixed to 4-well chamber slides (Nalge Nunc. International, Rochester, NY) that had been pretreated with poly-L-lysine. Cells were fixed in a PBS solution containing 4% paraformaldehyde and 0.1% glutaraldehyde or PBS containing only 4% paraformaldehyde for 30 min at room temperature. Chamber slides were rinsed once with PBS to remove fixative and incubated in a blocking solution consisting of PBS supplemented with 2% goat serum, 0.1% Tween 20, and 0.1% Triton X-100, which also permeabilized the parasites, for 1 h at room temperature. Antibodies against the L. donovani ODC, SPDSYN, and ADOMETDC proteins (18 -20) and purified antibodies to the L. donovani hypoxanthine-guanine phospho-ribosyltransferase (HGPRT) (35) were diluted 1:500 with blocking solution and incubated with the chamber slides to which wild type L. donovani (for ODC, SPDSYN, and ADOMETDC localization) or L. mexicana transfected with pXG-GFPϩ2Ј-arginase (for HGPRT detection) cells were affixed.
After incubation of fixed parasites with primary antibody for 1 h, all chamber slides were rinsed six times for 5 min with a PBS wash solution containing 0.1% Tween 20. Fixed cells were then incubated with secondary anti-rabbit antibody conjugated to Texas Red dye (Molecular Probes, Eugene, OR) that had been diluted 1:1000 in blocking buffer. The incubation and all subsequent steps were performed in the dark. Secondary antibody was removed by soaking chamber slides six times for 5 min in wash buffer. Chamber slides were washed in PBS and mounted with either Vectashield® mounting medium (Vector Laboratories, Inc., Burlingame, CA) or a solution of 50 mM Tris, pH 8.0, 8% n-propylgalate (Sigma-Aldrich Corp.), and 90% glycerol. Images shown in Fig. 2 (E and F) were obtained by Aurelie Snyder of the Oregon Health & Science University-Molecular Microbiology and Immunology Research Core Facility (www.ohsu.edu/core) with the Applied Precision Deltavision® image restoration system. Deconvolution was performed using the iterative constrained algorithm of Agard et al. (36), and additional image processing was performed on an Silicon Graphics Octane workstation. Images for Fig. 2 (A-D and G) were obtained by using a Zeiss Axiovert 200 inverted microscope and deconvolution accomplished using Axiovision 3.1 software (Carl Zeiss Optical, Chesterfield, VA).
Molecular Constructs for the Replacement of the Arginase Alleles-The 5Ј-and 3Ј-flanking regions of the L. mexicana arginase ORF were identified from sequence analysis of the arginase locus and cloned into the appropriate sites within the pX63-HYG and pX63-PHLEO vectors. The 5Ј-flanking region was amplified by PCR with TaqDNA polymerase (Promega Corp., Madison, WI), pBluescript-arginase as a template, the sense primer, 5Ј-CCGCTGAAGCTTGGAGATACGCCCCCGAGG-3Ј (HindIII site underlined), and the antisense primer, 5Ј-GTTCAACAT-GTCGACCTTGCCAT-3Ј (SalI site underlined). The resulting ϳ720-bp PCR product was first subcloned into the pCR® 2.1-TOPO® vector and then excised with HindIII and SalI and inserted into pX63-HYG and pX63-PHLEO that had been digested with HindIII and SalI. The resulting plasmids were designated pX63-HYG-5ЈF and pX63-PHLEO-5ЈF. To generate the 3Ј arginase flank for subcloning into pX63-HYG-5ЈF, a sense primer, 5Ј-GTAAATCCCGGGAAGCTATAGACGCGTGT-GTG-3Ј, and an antisense primer, 5Ј-GGCATTCCCGGGGCGTTTA-CACTCCCTGG-3Ј (SmaI sites underlined), were synthesized, and pBluescript-arginase was used as the template for PCR amplification. The ϳ500-bp PCR product was subcloned into the pCR® 2.1-TOPO® vector, excised with SmaI, gel-purified, and ligated into SmaI-cut pX63-HYG-5ЈF vector to generate the allelic replacement vector pX63-HYG-⌬arg.
Because of the presence of a SmaI sites within the PHLEO coding region of pX63-PHLEO, a different cloning strategy was employed to insert the 3Ј-flanking region of arginase into pX63-PHLEO-5ЈF. A ϳ1.0-kb sequence from the 3Ј-flanking region of arginase was amplified by PCR using the sense primer, 5Ј-TGCGCACACACAGATCTATATT-TAT-3Ј, and the antisense primer, 5Ј-GCAACTCCGAAACCAGATCTC-CTCC-3Ј (BglII sites underlined). The 1.0-kb PCR fragment was subcloned into the pCR® 2.1-TOPO® vector, excised with BglII, and the fragment was cloned into pX63-PHLEO-5ЈF that had been digested with BamHI and BglII to generate pX63-PHLEO-⌬arg. The correct orientations of the 5Ј-and 3Ј-flanking regions within the gene-targeting Generation of ⌬arg Null Mutants-The ⌬arg knockouts were generated by double-targeted gene replacement starting with wild type MNYC/BZ/62/M379 L. mexicana. pX63-HYG-⌬arg and pX63-PHLEO-⌬arg were digested with HindIII and BglII to liberate linear fragments, designated HYG-⌬arg and PHLEO-⌬arg, respectively, containing the drug resistance marker and the arginase flanking regions. HYG-⌬arg and PHLEO-⌬arg were isolated from agarose gels and then transfected into parasites using standard electroporation conditions for transfection of Leishmania promastigotes (17,18). First, HYG-⌬arg was transfected into wild type L. mexicana to create the arginase/arg heterozygote, and colonies were isolated by selection on plates of semi-solid DME-L-CS medium containing 50 g/ml hygromycin. The genotype of the arginase/arg heterozygote was then confirmed by Southern blot analysis using arginase flanking regions as probes. A heterozygous clone was then subjected to a second round of transfection with PHLEO-⌬arg and ⌬arg null lines selected in liquid DME-L-CS media containing 50 g/ml hygromycin, 50 g/ml phleomycin, and 200 M ornithine to circumvent potential ornithine auxotrophy. The ⌬arg genotype was verified by Southern blotting, and clones were isolated on semi-solid DME-L-CS plates containing 50 g/ml hygromycin, 50 g/ml phleomycin, and 200 M ornithine. A total of 12 clones were picked and screened by Southern blotting using the arginase ORF as a probe. The gene replacements of two of the ⌬arg clones were evaluated in more detail by probing Southern blots with the arginase flanks.
ARG Assays-The ARG activities in wild type, knockout, and transfected parasites were determined using lysates prepared from 1 ϫ 10 8 exponentially growing parasites that had been resuspended in 150 l of 50 mM glycine/10 mM MnCl 2 /1 mM dithiothreitol, pH 9.5. A protease inhibitor mixture (Roche Applied Science) was added, and the cells were sonicated three times for 10 s each. Protein concentrations in the resultant cell lysates were determined according to the Bradford method (37). ARG assays were initiated by the addition of 20 l of parasite extract (2.5 mg of protein/ml) to an 80-l reaction mixture containing 50 mM glycine/10 mM MnCl 2 /1 mM dithiothreitol, pH 9.5, and 10 mM L-[guanidino-14 C]arginine (31.3 Ci/mmol). At various intervals, the enzyme reaction was terminated by the addition of 10-l aliquots of the reaction mixture to 5 l of glacial acetic acid. The [ 14 C]urea product was separated from unreacted L-[guanidino-14 C]arginine by paper chromatography using Whatman #1 chromatography paper as the matrix and n-butanol:acetic acid:water (8:1:2) as the mobile phase. R F values for urea and arginine were 0.31 and 0.00, respectively. The [ 14 C]urea formed during the assay was detected by autoradiography and quantitated using a Beckman model LS 6500 liquid scintillation counter.

RESULTS
Isolation of the L. mexicana ARG-The L. mexicana ARG was isolated from a cosmid library using a PCR product amplified from genomic DNA as a hybridization probe. Sequence analysis revealed a 987-bp ORF predicting a polypeptide of 329 amino acids and a molecular mass of ϳ36 kDa. The predicted amino acid sequence of the L. mexicana ARG exhibited 38.5%, 32.6%, and 55.8% amino acid sequence identities to the human ARG I, human ARG II, and Saccharomyces cerevisiae ARG proteins, respectively (Fig. 1A). The protein was virtually identical to its L. amazonensis equivalent, differing in only two amino acids proximal to the COOH terminus (V306D and C308R). A noteworthy feature of the leishmanial ARG protein is the deduced COOH-terminal tripeptide, Ser-Lys-Leu, the archetypal PTS-1 that can mediate the translocation of proteins into the glycosome (44 -48). Southern blot analysis of the ARG locus revealed the gene to be single copy (data not shown),  (18 -20) and visualized by complexing with Texas Red-conjugated antibody. and a restriction map was accordingly compiled (Fig. 1B).
Localization of ARG, ODC, ADOMETDC, and SPDSYN-To determine whether the L. mexicana ARG is indeed localized to the glycosomal compartment, a GFP-tagged arginase construct, pXG-GFPϩ2Ј-arginase, was transfected into wild type L. mexicana. Analysis of the transfected line by direct fluorescence demonstrated a punctate staining pattern indicating ARG was constrained to an intracellular organelle ( Fig. 2A). Co-localization experiments with antibodies against the L. donovani HGPRT, a known glycosomal marker (35), indicated that ARG and HGPRT inhabited the same subcellular milieu (Fig. 2, B and C). In contrast, immunofluorescence images obtained with L. donovani wild type parasites using monospecific antibodies to the ODC, ADOMETDC, and SPDSYN proteins (18 -20) showed uniform diffuse staining patterns indicating that all three polyamine biosynthetic enzymes were localized to the Leishmania cytosol (Fig. 2, E-G).
Construction and Molecular Characterization of ⌬arg Knockouts-To evaluate the physiological role of ARG in intact Leishmania, each gene copy was sequentially replaced in L. mexicana with a drug resistance cassette lacking arginase coding sequences. The first arginase copy was eliminated with HYG-⌬arg, and clonal arginase/arg heterozygotes were selected on plates containing 50 g/ml hygromycin. The allelic replacement within the arginase/arg heterozygote was confirmed by Southern blotting (Fig. 3, A-C). A second round of transfection using the PHLEO-⌬arg construct was then implemented to target the remaining wild type allele in the arginase/arg heterozygote so as to generate the ⌬arg knockout. Surprisingly, no parasites were recovered from semi-solid agar plates containing 50 g/ml hygromycin and 50 g/ml phleomycin, the selective agents for the drug resistance cassettes, and 200 M ornithine. However, a batch culture of arginase/arg parasites transfected with PHLEO-⌬arg, selected in liquid medium under identical conditions, was obtained. After confirming the loss of arginase in the liquid culture by Southern blotting, the uncloned population was plated on semi-solid media containing 50 g/ml phleomycin, 50 g/ml hygromycin, and 200 M ornithine, and six colonies were selected for further analysis. The presence of hygromycin on the plates after the second round of transfection ensured that the PHLEO-⌬arg construct had replaced the remaining wild type allele in the arginase/arg heterozygote (rather than eliminating the previously targeted allele), and ornithine was added to the selective medium to circumvent potential ornithine auxotrophy from arginase loss. After Southern blotting using the arginase ORF as a hybrid- ization probe confirmed the ⌬arg genotypes in all six clones, two of these clonal isolates, ⌬arg2 and ⌬arg5, were selected for further analysis.
Southern blot analyses of BamHI-and SacI-digested genomic DNA from wild type, arginase/arg, and ⌬arg2 and ⌬arg5 parasites hybridized to the arginase ORF revealed the expected ϳ5.0-kb band in the wild type and arginase/arg strains and clearly demonstrated the loss of all arginase coding sequences in the ⌬arg2 and ⌬arg5 null lines (Fig. 3A). Furthermore, blots hybridized to the arginase 3Ј and 5Ј flanking regions demonstrated the expected molecular rearrangements that occurred as a consequence of the homologous recombina-tion events that gave rise to the heterozygous and homozygous knockout lines (Fig. 3, B and C). Replacement of a wild type arginase copy with either HYG-⌬arg or PHLEO-⌬arg created additional 3.6-and 3.4-kb SacI-BamHI fragments that hybridized to the 3Ј-flanking probe and extra 6.0-kb BamHI and 6.5-kb BamHI-SacI fragments that hybridized to the 5Ј-flank, respectively (Fig. 3, B and C). The loss of the remaining wild type arginase allele from the heterozygote is apparent after the second round of transfection with PHLEO-⌬arg in both knockout lines. A schematic diagram of the expected sizes of the BamHI and SacI restriction fragments from the wild type and rearranged loci is also displayed (Fig. 3D).
ARG Activity Assays-To establish the phenotypic consequences of the gene replacements in ⌬arg parasites, ARG activities were measured in wild type, heterozygous, and knockout parasite extracts (Fig. 4). Whereas significant ARG activity was detected in both wild type and arginase/arg L. mexicana, the activity of ARG in the ⌬arg5 knockout was negligible. Somewhat surprisingly, the arginase/arg parasites had slightly higher ARG activity than their parental wild type strain, a trend that was consistently observed in several independent experiments.
Nutritional Requirements of the ⌬arg Null Mutants-To establish whether the ⌬arg knockouts required nutritional supplements to survive and proliferate, the abilities of wild type, arginase/arg, and ⌬arg L. mexicana to grow in medium lacking ornithine, the product of the ARG reaction, were evaluated. Although both wild type and arginase/arg heterozygotes grew at equivalent rates in unsupplemented medium, both the ⌬arg2 and ⌬arg5 knockout clones failed to thrive (Fig. 5A). In the absence of supplementation, the homozygous null mutants divided ϳ2-3 times after which they assumed a rounded morphology, arrested growth, and died within 2 weeks (data not shown).
To determine the nutritional requirements of the ⌬arg knockouts, parasites were initially incubated with 200 and 500 M ornithine. These concentrations of ornithine enabled survival and restored the ability of the ⌬arg knockouts to multiply, although the growth rate was much slower than that exhibited by either wild type or arginase/arg parasites (Fig. 5B). Indeed, maximal growth rate of the homozygous ⌬arg null mutants could only be rescued by concentrations of ornithine Ͼ 1.0 mM (Fig. 5C).
Because ornithine, the product of the ARG reaction, is both a precursor of glutamate and proline synthesis and also a metabolic product of these two nonessential amino acids in mammalian cells (49), the abilities of glutamate and proline to rescue the lethal ⌬arg mutation were evaluated. Neither glutamate nor proline, at concentrations up to 5 mM, could rescue the ⌬arg conditionally lethal phenotype or increase the rate of growth of the parasites in the absence or presence of equimolar quantities of ornithine (Fig. 5C).
Because ornithine is the immediate polyamine precursor, the effect of exogenous polyamine on the ⌬arg growth phenotype was also evaluated. Surprisingly, putrescine at a concentration of 200 M, regardless of whether ornithine was present, was much more effective than equivalent concentrations of ornithine in enabling optimal growth of the ⌬arg parasites (Fig.  5B). Because the ⌬arg knockouts grew much better in 200 M putrescine compared with the same concentration of ornithine, the levels of putrescine and ornithine that were required for optimal proliferation of ⌬arg parasites were determined (Fig. 5,  C and D). Although a concentration of Ͼ1.0 mM ornithine was required to effect maximum parasite proliferation, optimal growth of ⌬arg parasites was achieved at 5-10 M exogenous putrescine. Spermidine could also rescue the ⌬arg parasites (Fig. 5E). The concentrations of spermidine required to enable ⌬arg growth, however, were much higher than those needed for putrescine rescue.
Ligand Transport into Arginase/arg L. mexicana-One potential and trivial explanation for the two order of magnitude discrepancy between the concentrations of ornithine and putrescine that circumvented the growth auxotrophy of ⌬arg knockouts was a differential capacity of the parasites to take up the two arginine downstream products. To test this conjecture, the ability of L. mexicana to take up the two positively charged compounds, each at a concentration of 20 M was compared. At these ligand concentrations, the overall rate of putrescine uptake was ϳ2-fold greater than that of ornithine (Fig. 6A). The rates of putrescine and ornithine uptake into intact parasites were also compared as a function of ligand concentration (Fig. 6B). Apparent K m values for putrescine and ornithine uptake were 33.5 Ϯ 2.5 and 42.2 Ϯ 8.8 M, and V max values were 373.2 Ϯ 6.6 and 316.6 Ϯ 14.4 nmol/min/10 7 parasites, respectively.
Amino Acid, Polyamine, AdoMet, and dAdoMet Pool Analyses-To evaluate the metabolic consequences of an ARG deficiency in L. donovani promastigotes, ornithine and arginine pools were compared in wild type parasites and ⌬arg knockouts grown in the absence or presence of either putrescine or ornithine. Ornithine and arginine pools in wild type promastigotes were 130 and 57 nmol/10 7 parasites, respectively (Fig. 7A). As expected, no ornithine could be detected in the ⌬arg null mutant grown in 200 M putrescine, whereas arginine pools were 2-fold greater than in wild type parasites. Intracellular ornithine could be observed in knockouts supplemented with 200 M or 1 mM ornithine, although the pools were much lower than those detected in wild type promastigotes. Low levels of arginine and no ornithine were found in ⌬arg cells incubated without putrescine or ornithine for 4 days, although it should be noted that these cells were nutritionally compromised and not proliferating. Polyamine pools were also determined for these parasites. Although wild type parasites contained equivalent concentrations of both putrescine and spermidine, the knockouts grown in either 200 M putrescine or 1 mM ornithine exhibited relatively low putrescine pools, whereas spermidine levels were similar to those in wild type parasites (Fig. 7B). Both polyamines were found at low levels in unsupplemented ⌬arg knockouts or ⌬arg parasites grown in 200 M ornithine. Finally, AdoMet and dAdoMet pools were measured. AdoMet levels were not significantly different in the null mutants grown in either putrescine or ornithine than in wild type parasites, although dAdoMet levels were somewhat elevated over wild type pools (Fig. 7C). Unsupplemented ⌬arg cells displayed a low AdoMet content, but dAdoMet concentrations were still higher than those found in the wild type line.
Mislocalization Does Not Affect ARG Function in L. mexicana Promastigotes-The glycosomal localization of the L. mexicana ARG was established above using direct fluorescence (Fig. 2). To determine whether the glycosomal milieu is crucial to ARG function, the ⌬arg knockout was transfected with a mutant GFP-tagged arg construct, GFP-arg⌬skl, in which the PTS-1 was deleted. The ⌬arg[pGFP-ARG] and ⌬arg[pGFP-arg⌬skl] transfectants expressed similar levels of ARG/arg protein as judged by Western blotting with anti-GFP antibody, as well as comparable levels of ARG activity (Fig. 8, A and B). The mislocalization of the GFP-arg⌬skl fusion protein to the cytosol was then confirmed by direct fluorescence microscopy (Fig. 9). Regardless of the location of the ARG/arg activity in the ⌬arg transfectants, both the ⌬arg[pGFP-ARG] and ⌬arg[pGFP-arg⌬skl] lines were capable of growing in growth medium lacking ornithine and putrescine supplementation (Fig. 8C). DISCUSSION The creation and characterization of ⌬arg parasites by double-targeted gene replacement established that arginase is an essential gene in L. mexicana promastigotes. The ⌬arg line exhibited polyamine auxotrophy, and this auxotrophy could be pharmacologically rescued by either putrescine, spermidine, or ornithine supplementation or complemented with an episomal copy of arginase. The ability of exogenous putrescine or spermidine to circumvent the lethality of the ⌬arg null mutation demonstrates that the sole vital function of ARG, at least in Leishmania promastigotes, is to provide the ornithine precursor for polyamine synthesis. Whether ARG plays supplementary but nonessential metabolic roles in promastigotes or has additional vital functions in amastigotes remains to be determined.
In mammalian cells, ornithine is also both a precursor and product of proline and glutamate metabolism through a pyrroline 5-carboxylate intermediate (49). Neither proline nor glutamate, both of which are absent from DME-L medium and are considered nonessential amino acids in Leishmania (50), however, affected the conditionally lethal growth phenotype of the ⌬arg parasites in the absence or presence of ornithine. The failure of proline or glutamate to promote growth of the ⌬arg knockouts cannot be imputed to lack of amino acid entry into the parasite, because both proline (51,52) and glutamate (data not shown) are efficiently taken up by Leishmania promastigotes. Thus, the inability of proline and glutamate at 5 mM concentrations to rescue the ⌬arg knockouts demonstrates that ARG deficiency does not cause proline or glutamate auxotrophy. The fact that low concentrations of putrescine can completely rescue the conditionally lethal ⌬arg mutation establishes that ornithine is an unlikely precursor of either nonessential amino acid. Moreover, it appears that neither proline nor glutamate can substantially contribute to the ornithine pool for polyamine synthesis. Consistent with this latter premise is a report that radiolabeled proline is not converted to ornithine by L. donovani promastigotes (50). Cumulatively, these data strongly support a singular function for ARG in Leishmania promastigotes, that of supplying a source of polyamines.
That the concentration of putrescine required to fully rescue the growth impairment of the ⌬arg lesion was two orders of magnitude lower than that of ornithine, the product of the ARG reaction remains somewhat mysterious. This ϳ100-fold difference could not be ascribed to differential uptake rates of the two compounds, because the rates of putrescine incorporation were only ϳ2-fold greater than that of ornithine (Fig. 6). These uptake experiments were performed in parallel at 20 M ligand, a concentration of ornithine that does not enable growth of the ⌬arg knockout but a concentration of putrescine that fully circumvents the conditionally lethal mutation (Fig. 5). The kinetic parameters, i.e. K m and V max values for ornithine and putrescine uptake into intact L. mexicana promastigotes were also similar (Fig. 6), although it should be noted that the kinetic parameters for the uptake of either dibasic ligand may reflect more than one transport system. However, pool measurements revealed that ornithine levels in ⌬arg parasites propagated in either 200 M or 1 mM ornithine were only 1% and 15% of those in wild type cells with a concomitant diminution of the putrescine pools (Fig. 7, A and B). These data imply that ODC activity is reduced in ⌬arg parasites grown in ornithine. Using the experimentally determined value for cell volume of a Leishmania promastigote (53), ϳ11 fl per cell, the intracellular ornithine level in ⌬arg parasites grown at the suboptimal concentration of 200 M is still ϳ12 mM, a value much greater than the K m value of 0.42 mM calculated for the L. donovani ODC enzyme (54). Thus, the reduced apparent flux through ODC in the ⌬arg cells cannot be explained by the low levels of ornithine alone. It is possible that subcellular sequestering of the ornithine and putrescine pools in the parasite can somehow contribute to the drastic differences observed for the amounts of ornithine and putrescine that rescue the ⌬arg phenotype.
Interestingly, spermidine levels in ⌬arg cells grown in 1 mM ornithine were normal, although they were markedly reduced in ⌬arg parasites grown in 200 M exogenous ornithine, a suboptimal concentration for growth (Fig. 7B). Coupled with the observation that ⌬arg parasites in 200 M putrescine proliferate rapidly and demonstrate normal spermidine but diminished intracellular putrescine pools, these data support the previously drawn conclusion that spermidine is the vital polyamine for parasite viability and vitality (18 -20). It is noteworthy that incubation of the ⌬arg mutants in ornithine or putrescine concentrations that allowed optimal proliferation did not restore either corresponding intracellular pool to wild type levels. The ability to create a conditionally lethal phenotype at the arginase locus demonstrates that there exists only a single ARG activity in L. mexicana promastigotes. In contrast, mammalian cells express two genetically and biochemically distinct ARG isoforms, ARG I, a cytosolic enzyme that participates in the urea cycle, and ARG II, a mitochondrial protein that is speculated to be involved in proline, glutamate, and polyamine biosynthesis (49). The L. mexicana ARG appears to play a functional role more similar to ARG II, although the parasite protein exhibits a greater amino acid sequence identity to human ARG I than to ARG II (Fig. 1). Leishmania do not appear to express many of the urea cycle enzymes (55), and no sequences encoding urea cycle enzymes, other than for ARG and carbamoyl phosphate synthetase, an isoform of which is required for pyrimidine biosynthesis, have been deposited in the emerging leishmanial genome sequencing data base. Skeletal muscle and several other tissues also accommodate yet another protein that converts arginine to ornithine, arginine:glycine amidiniotransferase, an enzyme involved in creatine biosynthesis (56). Clearly, the conditionally lethal nature of the ⌬arg knockouts establishes that Leishmania promastigotes have only a single avenue for ornithine biosynthesis. Interestingly, Trypanosoma cruzi, the causative agent of Chagas' disease and a close evolutionary relative of Leishmania, appears to lack ARG (55). T. cruzi also lacks ODC (57) activity and is, therefore, an obligatory scavenger of host polyamines.
The L. mexicana ARG exhibits considerable homology to ARG enzymes from phylogenetically diverse organisms and contains all ten invariant sequences that are present in 21 different ARG proteins from phylogenetically diverse species (58,59). According to the recently determined crystal structure of the rat ARG, most of these conserved residues are involved either directly in divalent cation coordination or in positioning other highly conserved residues for interaction with Mn 2ϩ (60). The most striking structural feature of the L. mexicana ARG is the COOH-terminal tripeptide, Ser-Lys-Leu, which has now been genetically verified as the PTS-1 that serves as the topogenic signal for targeting ARG to the glycosome. No equivalent PTS-1 sequence is observed among 31 ARG family members that have been analyzed in detail (58,59). Moreover, the L. donovani ODC (61), ADOMETDC (20), and SPDSYN (19) polypeptides do not accommodate a PTS-1, and all three enzymes have now been immunolocalized to the cytosol (Fig. 2). Thus, the enzymes involved in the synthesis of polyamines, like those for purine and pyrimidine nucleotide synthesis (35,(62)(63)(64), are compartmentalized between the glycosome and cytosol in Leishmania. The rationale for the sequestering of ARG from the cytosolic ODC, SPDSYN, and ADOMETDC is not apparent, but the glycosomal milieu is not essential for ARG function in promastigotes because of the ability of cytosolic arg (arg⌬skl) to complement ⌬arg parasites (Fig. 8). Interestingly, arginine, the substrate of ARG, is an essential amino acid in Leishmania promastigotes (50). Thus, the polyamine pathway in Leishmania promastigotes takes a rather circuitous route, originating with external arginine, which is then translocated across the parasite membrane through some yet undefined permeation mechanism, through the cytosol to the glycosome, after which the ornithine product of ARG is then released back into the cytosol for conversion to putrescine and spermidine, the two polyamines found in Leishmania (18 -20, 26, 65, 66).
Although the polyamine auxotrophy of the ⌬arg knockouts indicates that Leishmania only accommodate a single ARG activity, there is a second gene within the L. major genome data base (AL121851) that encodes an ARG family member, one that exhibits high sequence homology with agmatinase (AGM) enzymes. Of note, this putative agmatinase sequence lacks an obvious PTS-1. AGM is present in prokaryotes and plants and is part of an alternative pathway by which polyamines are produced from arginine in these organisms. In the AGM pathway, arginine is decarboxylated by arginine decarboxylase to produce agmatine, which is then hydrolyzed to putrescine with the resultant liberation of a urea molecule. Recently, AGM has also been identified in mammalian cells (67,68), although the existence of an arginine decarboxylase is still unresolved. The AGM pathway does not appear to contribute to the putrescine pool in Leishmania promastigotes, because ⌬odc (18) and ⌬arg promastigotes are putrescine auxotrophs. Furthermore, although ARG and AGM catalyze similar reactions and belong to the same enzyme superfamily (about 24% identity), the putative AGM does not appear to be a functional ARG, because the ⌬arg parasites displayed no residual ARG activity. The putative L. mexicana AGM has been recently cloned in this laboratory to assess its biochemical and physiological function in vitro and in vivo.
These genetic studies have demonstrated that ARG is an essential protein in Leishmania promastigotes. The existence of ARG inhibitors that can pharmacologically replicate a genetic deficiency in the enzyme, the unusual subcellular milieu of ARG, and the previously established biochemical disparities between the mammalian and leishmanial polyamine biosynthesis pathways, also imply that ARG could be a potential target for therapeutic manipulation of certain parasitic diseases. In the future, it will be critical to evaluate the phenotypic consequences of an ARG deficiency on the amastigote form of the parasite. Because these studies were undertaken in L. mexicana, an infectious strain (69,70), infectivity studies in macrophages can now be initiated to determine the physiological role of ARG protein in the infectious form of the parasite. These studies will ultimately answer the fundamental issue of whether amastigotes depend upon endogenous polyamine synthesis or are able to scavenge polyamines within the phagolysosome where they reside.