A Conditional Mutant Deficient in Hypoxanthine-guanine Phosphoribosyltransferase and Xanthine Phosphoribosyltransferase Validates the Purine Salvage Pathway of Leishmania donovani*

Leishmania donovani cannot synthesize purines de novo and express a multiplicity of enzymes that enable them to salvage purines from their hosts. Previous efforts to generate an L. donovani strain deficient in both hypoxanthine-guanine phosphoribosyl-transferase (HGPRT) and xanthine phosphoribosyltransferase (XPRT) using gene replacement approaches were not successful, lending indirect support to the hypothesis that either HGPRT or XPRT is crucial for purine salvage by the parasite. We now report the genetic confirmation of this hypothesis through the construction of a conditional Δhgprt/Δxprt mutant strain that exhibits an absolute requirement for 2′-deoxycoformycin, an inhibitor of the leishmanial adenine aminohydrolase enzyme, and either adenine or adenosine as a source of purine. Unlike wild type parasites, the Δhgprt/Δxprt strain cannot proliferate indefinitely without 2′-deoxycoformycin or with hypoxanthine, guanine, xanthine, guanosine, inosine, or xanthosine as the sole purine nutrient. The Δhgprt/Δxprt mutant infects murine bone marrow-derived macrophages <5% as effectively as wild type parasites and cannot sustain an infection. These data establish genetically that either HGPRT or XPRT is absolutely essential for purine acquisition, parasite viability, and parasite infectivity of mouse macrophages, that all exogenous purines are funneled to hypoxanthine and/or xanthine by L. donovani, and that the purine sources within the macrophage to which the parasites have access are HGPRT or XPRT substrates.

line of L. donovani that was selected by targeted gene replacement in the presence of 2Ј-deoxycoformycin (dCF), an inhibitor of the leishmanial adenine aminohydrolase (AAH) enzyme (11), and adenine as a source of purine. The creation and characterization of this ⌬hgprt/⌬xprt mutant confirms our main supposition that either HGPRT or XPRT is absolutely essential for purine acquisition and parasite viability.
Targeting and Episomal Constructs-The flanking regions from the HGPRT, APRT, and XPRT loci and the oligonucleotides used for their amplification by the PCR have been described (5,15). The construction and authentication of the pX63-HYG-⌬hgprt, pX63-NEO-⌬xprt, and pX63-HYG-⌬xprt targeting vectors employed in the previous allelic replacements of the HGPRT and XPRT loci have also been described (5,15). The pX63-NEO-⌬hgprt/⌬xprt construct was created by replacing the 5Ј-untranslated region (UTR) of XPRT in the pX63-NEO-⌬xprt vector with the 5Ј-UTR of the HGPRT. The pX63-PHLEO-⌬xprt replacement plasmid was generated by excising the 5Ј-and 3Ј-UTRs of XPRT from pX63-HYG-⌬xprt and inserting them into the appropriate sites within pX63-PHLEO (19).
Gene Replacements and Complemented Lines-All genetic manipulations were conducted on LdBob promastigotes. Single ⌬hgprt, ⌬aprt, and ⌬xprt knock-outs were created in LdBob using the same targeting constructs, protocols, and gene replacement strategies employed previously for the generation of these mutants within an avirulent L. donovani background (5,15). The ⌬hgprt/⌬xprt double knock-out was generated after three sequential rounds of targeted gene replacement. The pX63-NEO-⌬hgprt/⌬xprt, pX63-HYG-⌬hgprt, and pX63-PHLEO-⌬xprt plasmids were linearized with HindIII and BglII and transfected into 5 ϫ 10 7 parasites using reported electroporation conditions (17). Homologous integrations were selected by plating parasites on semisolid medium containing selective concentrations of either Geneticin, hygromycin, or phleomycin (19), as appropriate for the drug resistance marker of the targeting cassette. The HGPRT/hgprt/XPRT/xprt and ⌬hgprt/XPRT/xprt lines were generated and maintained in the M199based medium, whereas the ⌬hgprt/⌬xprt null mutant was selected and maintained in the modified DME-L medium described above. The HGPRT/hgprt/XPRT/xprt, ⌬hgprt/XPRT/xprt, and ⌬hgprt/⌬xprt lines were all maintained continuously under selective pressure in the drugs for which they contained resistance markers.
Enzyme Assays-2 ϫ 10 9 parasites were washed two times in phosphatebuffered saline (PBS), resuspended in 50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 1 mM phosphoribosylpyrophosphate, and protease inhibitor mixture, lysed by sonication, and fractionated by centrifugation at 9,740 ϫ g at 4°C. Radiolabel incorporation assays were performed on cell-free lysates of promastigote extracts as described (22), whereas amastigote enzymatic assays were conducted with lysates that were not fractionated by centrifugation. These assays measure the rates of radiolabeled preformed purine nucleoside and nucleobases into phosphorylated, anionic metabolites, i.e. nucleotides and nucleic acids.
Purine Metabolism in Live Cells-The rates of conversion of radiolabeled purine into nucleotides were measured using the earlier described DE-81 filter disk method (22). Cells were washed with PBS and resuspended at a density of 1 ϫ 10 8 cells/ml in either promastigote or amastigote medium containing 2 M radiolabeled purine but lacking bovine serum albumin, FBS, and hemin. At each time point 1 ϫ 10 7 cells were removed, washed once in PBS, lysed in 1% Triton X-100, and spotted onto DE-81 filter disks. Parasite metabolism of various radiolabeled purines was quantified by liquid scintillation.
Growth Phenotypes-To assess the abilities of genetically manipulated strains to grow in various different purine sources, all parasites were washed several times with PBS, resuspended in modified DME-L medium lacking purine, and incubated at 26°C for 4 h before they were seeded at a density of 5 ϫ 10 4 cells/ml in 1.0-ml aliquots of modified DME-L containing 100 M purine and 5% dialyzed FBS. Amastigotes were seeded at a density of 5 ϫ 10 3 cells/ml into amastigote medium containing 20% FBS and 100 M purine, incubated for 7-10 days, and counted by hemacytometer.
Macrophage Infections-Stationary phase promastigotes were washed two times in purine-free promastigote medium and resuspended in Dulbecco's modified Eagle's medium supplemented with 4 mM L-glutamine, 1.5 g/liter sodium bicarbonate, 4.5 g/liter glucose, and 10% FBS. 2 ϫ 10 5 bone marrow-derived mouse macrophages, from Balb/c mice, and 2 ϫ 10 6 promastigotes were placed in 4-well Lab-TekII CHAMBER SLIDES (Nalge Nunc International Corp., Naperville, IL) containing 1.0 ml of macrophage growth medium and incubated at 37°C in a humidified 5% CO 2 incubator. After 16 h, adherent macrophages were washed 10 times in PBS to eliminate residual extracellular promastigotes after which fresh growth medium was added and then again 24 h later. After an additional 24-h incubation, the chambers were washed three times with PBS, and macrophages were stained using the Diff-Quik kit (International Medical Equipment Inc., San Marcos, CA). Parasites were visualized on a Zeiss Axiovert 200 M scope (Carl Zeiss Microimaging, Thornwood, NY) using 60ϫ oil immersion light and photographed with an AxioCam MRm camera (Zeiss), and parasites were enumerated. Color photographs of parasitized macrophages were visualized on a Zeiss Axiophot microscope using 40ϫ oil immersion and photographed with a Leica DC 300 camera (Leica Camera AG, Solms, Germany).

RESULTS
The ⌬hgprt/⌬xprt knock-out was created after three rounds of transfection with drug resistance cassettes carrying 5Ј-and 3Ј-UTRs of HGPRT and XPRT (Fig. 1). Because HGPRT and XPRT are colocalized within a 4359-bp EcoRI fragment in the L. donovani genome (5), the first copy of both genes was displaced with X63-NEO-⌬hgprt/⌬xprt (linearized pX63-NEO-⌬hgprt/⌬xprt), a construct containing the 5Ј-UTR of HGPRT and the 3Ј-UTR of XPRT, to create the HGPRT/hgprt/XPRT/ xprt double heterozygote. The heterozygote was then transfected with X63-HYG-⌬hgprt to generate the ⌬hgprt/XPRT/xprt line, and the latter was transfected with X63-PHLEO-⌬xprt to create the ⌬hgprt/⌬xprt double knock-out (Fig. 1). The last round of transfection was performed in medium containing 20 M dCF and 100 M adenine, whereas the HGPRT/hgprt/XPRT/xprt and ⌬hgprt/XPRT/xprt progenitors of the double knock-out were isolated in medium lacking dCF and containing 100 M adenine as a purine source. 50 -100 drug-resistant colonies were obtained within days after the first two cycles of transfections, but surprisingly only two barely visible colonies were obtained after 4 weeks following the last round of transfection in the adenine-dCF medium. Both colonies were picked and expanded in liquid culture medium containing adenine and dCF.
Southern blot analysis of the HGPRT/hgprt/XPRT/xprt, ⌬hgprt/ XPRT/xprt, and ⌬hgprt/⌬xprt strains divulged the new alleles that had been created by the homologous gene replacement events (Fig. 2). The digestion of genomic DNA with EcoRI and hybridization to the HGPRT and XPRT open reading frames revealed the presence of the common 4359-bp restriction fragment in the wild type and HGPRT/hgprt/XPRT/ xprt lines and its absence in the ⌬hgprt/⌬xprt null mutant. The ⌬hgprt/ XPRT/xprt strain, as expected, lacks a band for HGPRT but exhibits a hybridization signal at 5939 bp when probed with the XPRT open reading frame that reflects the appropriate integration of the hygromycin resistance marker within the X63-HYG-⌬hgprt cassette into the HGPRT locus. For comparison, a Southern blot analysis of previously isolated single knock-outs of HGPRT, APRT, and XPRT are also depicted, and the signals are appropriate for the expected homologous integrations (Fig. 2).
Western blot analysis of wild type, HGPRT/hgprt/XPRT/xprt, ⌬hgprt/XPRT/xprt, and ⌬hgprt/⌬xprt extracts confirmed the absence of HGPRT and XPRT protein in strains in which the corresponding gene had been eliminated (Fig. 3). The single ⌬hgprt, ⌬aprt, and ⌬xprt null mutants for which Southern blot data are shown in Fig. 2 also lacked the proteins corresponding to the deleted genes.
The abilities of wild type and ⌬hgprt/⌬xprt promastigote lysates to incorporate individual radiolabeled purines into nucleotides were   assessed over a 15-min time interval (Fig. 4). Whereas wild type promastigotes were capable of incorporating all purine bases and nucleosides tested, the ⌬hgprt/⌬xprt could not convert guanine, guanosine, hypoxanthine, or inosine to nucleotides. The double knock-out could, however, incorporate radiolabeled adenine and adenosine into phosphorylated products during the 15 min assay interval, whether or not dCF was added to the extracts (Fig. 4). [ 14 C]Xanthine conversion to nucleotides could not be measured in promastigote extracts for technical reasons.
The ability of live promastigotes to metabolize various [ 14 C] radiolabeled purine bases and nucleosides was also measured. These results corroborated the results from the enzymatic assays described above. Additionally, the metabolism of [ 14 C]xanthine could be measured in intact wild type cells but not in the double knock-out (Fig. 5). [ 14 C]Adenine and [ 14 C]adenosine metabolism could be measured in both wild type and mutant cells during the 2-h time course in the presence or absence of dCF.
The ability of ⌬hgprt/⌬xprt promastigotes to proliferate in various purine sources was assessed (Fig. 6). Whereas wild type parasites could grow in modified DME-L medium with any of the added purine nucleobases or nucleosides as a purine source, continuous and robust growth of ⌬hgprt/⌬xprt promastigotes could be maintained only in adenine or adenosine in the presence of 20 M dCF. A small number of mutant parasites were observed at the end of the experiment in adenine or adenosine alone (when the knock-out parasites in dCF had reached stationary phase growth), but these parasites were not capable of further proliferation in the absence of dCF and soon died (data not shown). The double mutant could not grow in xanthine, xanthosine, guanine, guanosine, inosine, or hypoxanthine, although a limited amount of growth was observed when inosine was added as the purine in the presence of dCF. The ⌬hgprt/⌬xprt parasites incubated in inosine and dCF were also not capable of long term survival after several additional weeks of incubation. The limited proliferation in inosine was further investigated with immucillin H, an iminoribitol analog of inosine that is a potent inhibitor of the leishmanial inosine-uridine nucleoside hydrolase activity (23). Although immucillin H at 20 M allowed the ⌬hgprt/⌬xprt double knock-out to replicate for several cell cycles in 100 M inosine, indefinite growth in the inosine-immucillin H combination could also not be sustained (data not shown).
Both wild type and ⌬hgprt/⌬xprt parasites were capable of transformation to axenic amastigotes, as assessed by their expression of A2 proteins, a family of amastigote-specific markers (24). No A2 was observed in the promastigotes. The growth phenotypes of both wild type and knock-out axenic amastigotes were identical to their promastigote counterparts (data not shown). The metabolic capacities of the axenic amastigotes toward sundry purines were also indistinguishable from the promastigote equivalent.
Wild type L. donovani promastigotes were capable of sustaining a robust infection of bone marrow-derived murine macrophages, whereas the ⌬hgprt/⌬xprt knock-out could not (Fig. 7). Parasitemia of the wild type strain was ϳ21 parasites/macrophage, whereas the double knock-out infectivity was ϳ1 parasite/macrophage, a 20-fold difference. The inability of ⌬hgprt/⌬xprt cells to proliferate inside macrophages could not be imputed to a failure to infect the mammalian cells, because similar numbers of intracellular parasites were observed for both wild type   and knock-out parasites 4 h postinfection (data not shown). Supplementation of the macrophage medium with either 100 M adenine or a combination of 100 M adenine plus 20 M dCF allowed the ⌬hgprt/ ⌬xprt mutant to reach a parasite load of 7.5 parasites/macrophage and 12 parasites/macrophage, respectively, indicating that it is possible for the ⌬hgprt/⌬xprt mutant to proliferate within macrophages if vital nutrients are provided to its host cells (Fig. 7D). Both complemented lines, ⌬hgprt/⌬xprt[pHGPRT] and ⌬hgprt/⌬xprt[pXPRT], also sustained robust infections in the macrophages, although the infectivity was lower than that of wild type parasites. Parasite loads of ϳ8 parasites/ macrophage and ϳ13 parasites/macrophage were observed for ⌬hgprt/ ⌬xprt[pHGPRT] and ⌬hgprt/⌬xprt[pXPRT], respectively (data not shown).

DISCUSSION
The purine salvage pathway of Leishmania is myriad and complex, although determining which of the many routes of purine salvage are functional has defied genetic dissection (1,25). Mutational and gene replacement schemes in L. donovani have demonstrated that none of the four known enzymes capable of converting host purine nucleobases or nucleosides to the nucleotide level, HGPRT, APRT, XPRT, or AK, is essential by itself (5,15,16,26). Furthermore, the ability to generate viable ⌬hgprt/⌬aprt/ak Ϫ (16) and ⌬xprt/⌬aprt/ak Ϫ4 L. donovani promastigotes reveals that the parasite can rely on either a functional XPRT or HGPRT activity for all of its purine nutritional requirements. Indeed, it has been possible to generate mutant parasites by targeted gene replacement in every conceivable combination except mutants accommodating a combined ⌬hgprt and ⌬xprt genotype. These results were the basis for our fundamental hypothesis that either HGPRT or XPRT is necessary and sufficient for Leishmania parasites to salvage purines, maintain viability, and sustain proliferation. The hypothesis was sustained by the inability to generate a ⌬hgprt/⌬xprt double knock-out even within a genetic background complemented with an episomal copy of either HGPRT or XPRT. 4 We now report genetic proof of our principal hypothesis by the creation and characterization of a conditional ⌬hgprt/⌬xprt double knockout. Taking advantage of the colocalization of the HGPRT and XPRT genes in the leishmanial genome (5), a ⌬hgprt/⌬xprt mutant was selected after three rounds of targeted gene replacement, with the final transfection being achieved in the presence of 100 M adenine and 20 M dCF, an inhibitor of the L. donovani AAH (11). The ⌬hgprt/⌬xprt mutant is absolutely reliant on the presence of dCF and either adenine or adenosine as a purine source for lasting survival and growth. No other naturally occurring purine tested could enable indefinite, long term proliferation of either stage of the parasite. However, several rounds of replication by the double knock-out could be sustained with adenine or adenosine in the absence of dCF, but this proliferation could not be perpetuated interminably (Fig. 6). The ability of the ⌬hgprt/⌬xprt to undergo several cell divisions in adenine or adenosine can be ascribed to the time interval required for AAH to convert the 6-aminopurine source to hypoxanthine. This contention is buttressed by short term radiolabel incorporation experiments demonstrating that ⌬hgprt/⌬xprt lysates (Fig. 4), as well as intact parasites (Fig. 5), were perfectly capable of converting adenine or adenosine to the nucleotide level via APRT. Eventually, however, the actions of AAH convert adenine or adenosine, which is cleaved to adenine by L. donovani promastigotes (11,22,26), to hypoxanthine, a dead end nutrient for ⌬hgprt/⌬xprt cells. The limited ability of the mutant promastigotes to grow sparingly in inosine plus dCF or immucillin H suggests that L. donovani express an activity capable of salvaging the nucleoside. Although inosine kinase activity has not been detected in Leishmania (1), a nucleoside phosphotransferase activity capable of phosphorylating the pyrazolopyrimidine nucleoside analogs allopurinol riboside and formycin B, has been observed in L. donovani promastigotes (27,28). Whether this phosphotransferase is also capable of recognizing inosine is not known.
To date, there are no effective vaccines to protect against visceral leishmaniasis (29). Generating a strain with intrinsic attenuating mutation(s) could theoretically be exploited as a live attenuated vaccine for immunizing against the disease is a valid alternative strategy to control   amastigotes (B, C). The scale bar represents 50 m. D, the average number of amastigotes/macrophage is depicted for wild type cells, the ⌬hgprt/⌬xprt mutant, and for the ⌬hgprt/⌬xprt mutant when supplemented with adenine or adenine plus dCF during the course of infection. leishmaniasis rather than the conventional paradigm of treatment with toxic drugs. Previous studies using noninfectious Leishmania tarentolae or attenuated strains of Leishmania major have demonstrated that these lines are capable of triggering a protective immune response against further challenge by virulent Leishmania in susceptible rodents (29,30). The inability of the ⌬hgprt/⌬xprt knock-out to sustain an infection of murine macrophages bolsters this mutant strain as a candidate for such a live vaccine strategy against visceral leishmaniasis. The ⌬hgprt/⌬xprt mutant infects macrophages at a level Ͻ5% of that of wild type parasites. Preliminary results indicate that the overall infection rate of the ⌬hgprt/ ⌬xprt mutant in macrophages can be increased by the addition of either adenine alone or, a combination of adenine and dCF, to the growth medium. Thus, it should be theoretically feasible to sustain an infection of the ⌬hgprt/⌬xprt strain in susceptible mammals by dietary supplementation with adenine and/or dCF until a protective immune response has been established. The strain could then be eliminated by withdrawal of the dietary additions. It is important, of course, to determine the stability of the mutant genotype and phenotype. Genetic studies are currently underway to determine whether the ⌬hgprt/⌬xprt double mutant is stable and does not revert by down-regulating its AAH activity thereby allowing the parasite to presumably grow on adenine or adenosine alone.