Genetic Analysis of Purine Metabolism in Leishmania donovani *

To dissect the contributions of hypoxanthine-guanine phosphoribosyltransferase (HGPRT), adenine phosphoribosyltransferase (APRT), and adenosine kinase (AK) to purine salvage in Leishmania donovani, null mutants genetically deficient in HGPRT and/or APRT were generated by targeted gene replacement in wild type cells and preexisting mutant strains lacking either APRT or AK activity. These knockouts were obtained either by double targeted gene replacement or by single gene replacement followed by negative selection for loss-of-heterozygosity. Genotypes were confirmed by Southern blotting and the resultant phenotypes evaluated by enzymatic assay, resistance to cytotoxic drugs, ability to incorporate radiolabeled purine bases, and growth on various purine sources. All mutant strains could propagate in defined growth medium containing any single purine source and could metabolize exogenous [3H]hypoxanthine to the nucleotide level. The surprising ability of mutant L. donovani lacking HGPRT, APRT, and/or AK to incorporate and grow in hypoxanthine could be attributed to the ability of the parasite xanthine phosphoribosyltransferase enzyme to salvage hypoxanthine. These genetic studies indicate that HGPRT, APRT, and AK, individually or in any combination, are not essential for the survival and growth of the promastigote stage of L. donovani and intimate an important, if not crucial, role for xanthine phosphoribosyltransferase in purine salvage.

Protozoan parasites cause a variety of devastating and often fatal diseases in humans and their domestic animals. The treatment and control of parasitic diseases, however, is severely compromised by the dearth of effective and selective antiparasitic therapies. Many of the current antiparasitic drugs cause severe toxicity in the host, a predicament that can be attributed to lack of target specificity. Moreover, these drugs are potentially mutagenic and/or carcinogenic, they often require protracted courses with multiple drug administrations, and therapeutic unresponsiveness and drug resistance have exacerbated the necessity for new and improved antiparasitic agents.
The institution of a rational therapeutic regimen for the treatment and prevention of parasitic diseases hinges upon exploitation of fundamental biochemical disparities between parasite and host. Perhaps the most striking metabolic discrepancy between parasites and humans is the purine pathway.
Whereas mammalian cells can synthesize the purine heterocycle de novo, all protozoan parasites studied thus far are auxotrophic for purines (1). As a consequence, each genus of parasite has evolved a unique complement of purine salvage enzymes that enable the organism to scavenge host purines. Unique features of the purine salvage pathway of Leishmania and Trypanosoma constitute the basis for the susceptibility of these genera to several pyrazolopyrimidine analogs of naturally occurring purine bases and nucleosides (2,3). The intact parasites efficiently metabolize these analogs to the nucleotide level, whereas mammalian cells are essentially incapable of these metabolic transformations. One of these pyrazolopyrimidines, allopurinol (4-hydroxypyrazolo [3,4]pyrimidine, HPP), 1 a drug that is nontoxic to humans and is widely used in the treatment of hyperuricemia and gout (4), has demonstrated significant therapeutic efficacy in patients with either cutaneous leishmaniasis (5) or chronic Chagas disease (6).
Leishmania donovani, the causative agent of visceral leishmaniasis, is a digenetic parasite that exists as an extracellular promastigote within the insect vector, members of the phlebotomine sandfly family, and as a nonmotile intracellular amastigote within the phagolysosome of macrophages and other cells of the reticuloendothelial system of the mammalian host. L. donovani expresses a number of enzymes capable of converting preformed purines directly to nucleotides. These enzymes include hypoxanthine-guanine phosphoribosyltransferase (HGPRT; IMP:pyrophosphate phosphoribosyltransferase; EC 2.4.2.8), adenine phosphoribosyltransferase (APRT; AMP:pyrophosphate phosphoribosyltransferase; EC 2.4.2.7), xanthine phosphoribosyltransferase (XPRT; XMP:pyrophosphate phosphoribosyltransferase; EC 2.4.2.22), and adenosine kinase (AK). Leishmania also contain a plethora of purine interconversion enzymes including nucleosidases, phosphorylases, deaminases, and IMP branchpoint enzymes (1). Overall, the purine pathway is divagating and intricate, and this metabolic complexity and the apparent overall diploid nature of the parasite (7-9) has hindered a thorough characterization of the purine pathway in the parasite by straightforward biochemical and genetic approaches.
The ability of Leishmania to undergo a high rate of homologous gene replacement (7, 10 -12) and to take up foreign DNAs after transfection (13,14) now permits an assessment of specific gene function by targeted gene replacement and facilitates the genetic dissection of complex metabolic pathways such as that for purine acquisition. As both the HGPRT (15) and APRT (16) genes and their flanking regions have been isolated, we have examined the relative contributions of HGPRT, APRT, and AK to purine salvage in L. donovani promastigotes by sequentially eliminating each HGPRT and APRT allele from wild type cells and from preexisting mutants lacking either APRT (8) or AK (17) activity by homologous gene replacement and/or direct selection. These strains were created either by replacing each allele sequentially with independent drug resistance markers or by disrupting the first allele with one drug resistance marker and selecting for loss-of-heterozygosity (LOH) with drug pressure to obtain the homozygous null mutant. The new alleles in the heterozygotes and homozygotes were examined by Southern blotting, and the resultant phenotypes of the mutant strains have been evaluated for enzymatic activities, ability to take up purines, drug resistance profiles, and growth properties. Cell Culture-Promastigotes of the Sudanese 1S strain of L. donovani were grown in DME-L culture medium (18). Unless otherwise specified, DME-L contains 100 M xanthine as a purine source. Clonal lines of L. donovani were isolated as colonies on semi-solid DME-L medium containing appropriate selective agent as indicated (18). Initial recipient strains for the transfection experiments reported here included the wild type DI700 line and its two clonal derivatives, TUBA2 and APPB2A3. The isolation and biochemical characterizations of the AK-deficient (ak Ϫ ) TUBA2 (17) and the APRT-deficient (aprt Ϫ ) APPB2A3 (8) strains have been reported previously. TUBA2 and APPA2B3 cells and all of their progeny were continuously maintained in either 1 M tubercidin, a cytotoxic AK substrate (17), or 100 M 4-aminopyrazolopyrimidine, a subversive substrate of APRT (8), respectively, to ensure retention of the selected phenotype. For the genetic manipulations of the HGPRT and APRT loci described here, DI700, TUBA2, and APPB2A3 cell lines are designated DI700:H ϩ/ϩ A ϩ/ϩ , TUBA2:H ϩ/ϩ A ϩ/ϩ , and APPB2A3:H ϩ/ϩ , respectively, where H refers to HGPRT, A to APRT, and ϩ is the wild type allele (see Table I). No allelic nomenclature is given for the APRT locus of APPB2A3 cells, as the genetic basis for the APRT deficiency in this strain has not been characterized. Sensitivities of wild type and strains created by targeted gene replacement to HPP, 4-aminopyrazolo [3,4]pyrimidine (APP), 4-thiopurinol, and HPP riboside were performed as described (8,17).
DNA Manipulations-Protocols utilized by this laboratory for obtaining parasite DNA, for Southern blotting, and for nucleotide sequencing have all been described previously (19,20). The L. donovani HGPRT (15) and APRT (16) were isolated from cosmids as 3.5-kb EcoRI and 5.4-kb BamHI-XbaI restriction fragments, respectively. Flanking regions of the two genes were identified by restriction mapping and Southern blotting. An additional 5.2-kb SalI-EcoRI (21) DNA located immediately 5Ј to the 3.5-kb EcoRI DNA was also obtained from the HGPRT-containing cosmid to amplify the 5Ј-flanking region of HGPRT for construction of gene targeting vectors.
Construction of Knockout Vectors-The flanking regions from the HGPRT and APRT loci and the oligonucleotides used for their amplification by the polymerase chain reaction (PCR) have been reported previously (21). The construction and authentication of the pX63-NEO-⌬hgprt and pX63-HYG-⌬aprt plasmids employed in the allelic replacements of the HGPRT and APRT loci, respectively, have also been described (21). pX63-NEO-⌬hgprt contains a 1.7-kb 5Ј and a 1.8-kb 3Ј-flanking region from the L. donovani HGPRT encompassing the neomycin phosphotransferase (NEO) gene, while pX-HYG-⌬aprt includes a 1.4-kb 5Ј and a 1.5-kb 3Ј flank of the L. donovani APRT circumscribing the hygromycin phosphotransferase (HYG) marker. To generate a HYG construct for replacing the HGPRT locus, the 1.7-kb 5Ј and 1.8-kb 3Ј flanks employed for making pX63-NEO-⌬hgprt were reamplified using the PCR conditions described by Hwang et al. (21) and ligated into the appropriate restriction sites within pX63-HYG. The pX63-HYG-⌬hgprt construct was verified for identity and orientation by limited restriction mapping and nucleotide sequencing.
Transfection-Parasites were transfected using electroporation conditions reported previously (11,21). All targeting constructs were cleaved from their plasmids by digestion with HindIII and BglII just prior to electroporation, isolated on agarose gels, and purified using the GeneClean Kit (Intermountain Scientific Corp., Bountiful, UT). Designation of the targeting DNAs originated from the plasmid nomenclature without the initial letter, e.g. X63-HYG-⌬hgprt from pX63-HYG-⌬hgprt. All transfected cells were grown for 24 h under nonselective conditions prior to the initiation of drug pressure.
Gene Replacements-To generate lines heterozygous at either the HGPRT (hgprt/HGPRT) or APRT (aprt/APRT) locus, L. donovani promastigotes were transfected with the appropriate targeting construct, and heterozygotes were selected by the addition of either 50 g/ml Geneticin (G418) or hygromycin B to liquid medium or 16 g/ml G418 or 50 g/ml hygromycin B to semi-solid medium. Clonal populations were selected on agar plates as described (18). The genotypes of the heterozygotes were determined by Southern blotting using appropriate protein coding and flanking probes.
To generate homozygous null mutants (⌬hgprt or ⌬aprt) from the heterozygotes, two strategies were employed. The first required a second round of transfection with an independent drug resistance cassette, i.e. double targeted gene replacement. Using this approach, both G418 and hygromycin were added to the plates to select for parasites in which the wild type allele in the heterozygous recipient had been specifically disrupted by homologous recombination of the second targeting construct. The second protocol for selecting homozygous null mutants at either the HGPRT or APRT loci was to select for LOH by plating in semi-solid agarose to which a toxic substrate of the encoded gene product had been added. Specific selective conditions were 1-3 mM HPP to obtain ⌬hgprt clones and 100 M APP to obtain ⌬aprt clones. Selections for LOH at the HGPRT and APRT loci included G418 and/or hygromycin, as specified, in the medium to retain drug resistance cassettes that had been previously integrated at the appropriate locus/ loci by targeted gene replacement.
All L. donovani transformants that had integrated copies of either X63-NEO-⌬hgprt, X63-HYG-⌬hgprt, or X63-HYG-⌬aprt into the relevant loci were maintained continuously under selective pressure in the drugs for which they contained resistance markers. In addition, knockout cell lines created by single targeted gene replacement as described above were also grown perpetually in the agent in which they were selected, i.e. 2 mM HPP or 100 M APP as appropriate.
Radiolabel Incorporation Assays-The rate of [ 14 C]hypoxanthine incorporation into phosphorylated metabolites in the absence or presence of excess nonradiolabeled purines was ascertained at room temperature using a slight modification of the DE-81 filter disk method outlined by Iovannisci et al. (17). 10 8 parasites were harvested by centrifugation, washed with phosphate-buffered saline, and resuspended in 1.0 ml of modified DME-L growth medium lacking the hemin and xanthine components prior to the addition of radiolabel and purine additives. [ 14 C]Hypoxanthine and nonradiolabeled purines were present at 1.8 M and 100 M, respectively, in the uptake assays. The assay was terminated by removing 100-l aliquots of (10 7 ) cells, diluting them in ice cold phosphate-buffered saline, and subjecting them to centrifugation. The pelleted cells were washed twice in 1.0 ml of phosphate-buffered saline, lysed in 1% Triton X-100, and blotted onto a 1.5-cm 2 piece of DE-81 impregnated paper. The disks were washed as described (17).

Derivation of Mutant Strains by Gene
Replacement-A flow chart describing the derivation and lineages of all of the mutant strains that were created as a result of targeted gene replacement and/or direct selection in cytotoxic substrate is depicted in Fig. 1. Three parental recipient clonal cell lines were employed for the transfection experiments, DI700 wild type cells, TUBA2 ak Ϫ cells, and APPB2A3 aprt Ϫ cells. The three parental cell lines are designated DI700:H ϩ/ϩ A ϩ/ϩ , TUBA2:H ϩ/ϩ A ϩ/ϩ , and APPB2A3:H ϩ/n , respectively ( Fig. 1), as their HGPRT and APRT loci have not been targeted by gene replacement vectors. Heterozygotes (hgprt/HGPRT) in which a single wild type HGPRT allele was disrupted by targeted gene replacement approaches were obtained from DI700 (DI700:H ϩ/ n A ϩ/ϩ ) and TUBA2 (TUBA2:H ϩ/n A ϩ/ϩ ) cells with X63-NEO-⌬hgprt as the targeting vector and from APPA2A3 (APPB2A3: H ϩ/h ) cells with X63-HYG-⌬hgprt. Homozygous null mutants (⌬hgprt) were then obtained from the three hgprt/HGPRT lines by two independent strategies, either by targeting with the appropriate construct containing the second drug resistance marker (DI700:H n/h A ϩ/ϩ , TUBA2:H n/h A ϩ/ϩ , and APPB2A3:H h/n ) or by direct negative selection (DI700: H n/n A ϩ/ϩ and TUBA2:H n/n A ϩ/ϩ ) for LOH (see Fig. 1). Nomenclature for all the ⌬hgprt cell lines shown in Fig. 1 conforms to that employed for the heterozygotes.
The ability to generate ⌬hgprt null mutants after only a single round of transfection permitted the creation of L. donovani cell lines in which both wild type alleles of the HGPRT and APRT loci had been displaced, each with a single targeting construct. Cell lines heterozygous at the APRT locus (aprt/ APRT) were thus created from the DI700:H n/n A ϩ/ϩ and TUBA2:H n/n A ϩ/ϩ cell lines, as well as from the parental TUBA2:H ϩ/ϩ A ϩ/ϩ strain by targeted gene replacement using the X63-HYG-⌬aprt construct. These derivatives were designated DI700:H n/n A ϩ/h , TUBA2:H n/n A ϩ/h , and TUBA2: H ϩ/ϩ A ϩ/h , respectively (Fig. 1). ⌬aprt clonal progeny, DI700: H n/n A h/h , TUBA2:H n/n A h/h , and TUBA2:H ϩ/ϩ A h/h , were then selected for LOH from the three aprt/APRT heterozygotes in 100 M APP (Fig. 1).
Southern Blot Analysis-Southern blot analysis verified the existence of the new alleles that had been created after either double or single targeted gene replacement. The maps of the genomic loci for HGPRT and APRT and the novel alleles created by homologous recombination of X63-NEO-⌬hgprt into the HGPRT locus and X63-HYG-⌬aprt into the APRT locus have been reported previously (21). The X63-HYG-⌬hgprt construct and its rearranged allele after insertion into the chromosome are displayed in Fig. 2. The maps of the HGPRT and APRT loci are also included in Fig. 2 to show the location of the probes employed in the Southern blotting experiments. The expected size differences of the wild type and disrupted HGPRT alleles in strains that were generated or employed for these studies can be visualized in Fig. 3. The interrupted HGPRT alleles could be distinguished from the wild type allele by an altered EcoRI restriction pattern (Fig. 3). After cleavage of genomic DNA with EcoRI and hybridization to probe B derived from the 3Ј-flanking region of HGPRT (see Fig. 2), only the anticipated 4.6-kb EcoRI fragment from the wild type allele was discerned in the three parental, i.e, DI700:H ϩ/ϩ A ϩ/ϩ , APPB2A3:H ϩ/ϩ , and TUBA2:H ϩ/ϩ A ϩ/ϩ , strains in which the HGPRT locus was intact (see Fig. 3, probe B). Although the HGPRT was originally isolated as a 3.5-kb EcoRI fragment (see "Experimental Procedures"), the 4.6-kb EcoRI band is the expected size of the wild type HGPRT allele, as the 3Ј EcoRI restriction site of the cloned 3.5-kb fragment originates from the cosmid. In contrast, bands of 6.4 and 6.2 kb were observed in those strains in which an HGPRT allele had been targeted by either X63-NEO-⌬hgprt or X63-HYG-⌬hgprt, respectively (Fig. 3). Cell lines heterozygous for HGPRT (H ϩ/h or H ϩ/h ) retained the 4.6-kb EcoRI fragment derived from the wild type locus, while HGPRT null mutants (H n/h or H n/n ) lacked the wild type 4.6-kb EcoRI restriction fragment and exhibited only the expected 6.4-and 6.2-kb displacements by X63-NEO-⌬hgprt and X63-HYG-⌬hgprt, respectively (Fig. 3, probe B). The interruption of the wild type HGPRT allele was verified by hybridization of genomic DNA to probe A corresponding to the coding region of HGPRT (see Fig.  2). The 4.6-kb EcoRI fragment was detected only in cell lines that were wild type or heterozygous at the HGPRT locus, i.e. the H ϩ/ϩ , H ϩ/n , and H ϩ/h strains and not in ⌬hgprt null mutants that were derived by either double targeted (H n/h ) or single targeted (H n/n ) gene replacement (Fig. 3, probe A).
It should be noted that after targeting wild type cells with X63-NEO-⌬hgprt or X63-HYG-⌬hgprt, 23 of 24 clones isolated and analyzed by Southern blotting displayed only simple gene replacements, whereas 12 of 15 colonies isolated from hgprt/ HGPRT heterozygotes targeted with X63-HYG-⌬hgprt exhibited complex genetic events other than simple gene replacements. These enigmatic genetic alterations in the anomalous cell lines were not analyzed in detail, and the reasons for the large discrepancies in the frequency by which various genetic events were observed during each round of transfection is unknown. However, genetic events that do not involve simple allelic replacements after homologous recombination of an extrachromosomal fragment have been observed previously in Leishmania spp. (7, 12, 23).
Disruption of the wild type APRT locus in lines that were donovani strains derived for these investigations is depicted. Capital letter designations specify the locus, either HGPRT (H) or APRT (A), while superscripts specify the allele, either wild type (ϩ) or those replaced by X63-NEO-⌬hgprt ( n ), X63-HYG-⌬hgprt ( h ), or X63-HYG-⌬aprt ( h ). The three initial strains employed for these genetic analyses include the previously described wild type DI700, ak Ϫ TUBA2, and aprt Ϫ APPB2A3 strains (8,17), which are now designated for the purposes of these genetic studies DI700:H ϩ/ϩ A ϩ/ϩ , TUBA2:H ϩ/ϩ A ϩ/ϩ , and APPB2A3:H ϩ/ϩ A ϩ/ϩ , respectively. targeted by X63-HYG-⌬aprt was also validated by Southern blot analysis (Fig. 4). aprt/APRT heterozygotes and ⌬aprt homozygous knockouts were only derived from the DI700 and TUBA2 lines (see Fig. 1), as APPB2A3 cells already lacked APRT activity (8). Digestion of genomic DNA with SalI and BamHI and hybridization to probe D derived from the 5Јflanking region of APRT (Fig. 2) revealed the presence of the 3.5-kb SalI/BamHI restriction fragment from the wild type allele in all A ϩ/ϩ and A ϩ/h cells (Fig. 4, probe D). The aprt/ APRT heterozygotes contained an additional 1.4-kb SalI band (see Fig. 2) derived from the new allele created by integration of X63-HYG-⌬aprt into the APRT locus. Hybridization of DNA from all of the presumptive ⌬aprt null mutants to probe D revealed only the 1.4-kb band expected from the integration of the drug resistance cassette into both alleles of the APRT locus (Fig. 4, probe D). Genetic disruption of wild type APRT alleles was confirmed by hybridization of the same blot to probe C encompassing most of the APRT coding region (Fig. 4, probe C). APRT coding sequences existed only in cells containing one or two wild type APRT copies, whereas the knockout lines TUBA2:H ϩ/ϩ A h/h and TUBA2:H n/n A h/h failed to hybridize to probe C (Fig. 4, probe C). The Southern blots of aprt/APRT and ⌬aprt derivatives of DI700 cells hybridized to probes C and D have been published previously (21); thus Southern blots of only ak Ϫ TUBA2 and its derivatives are depicted in Fig. 4.
Enzyme Assays-The phenotypic consequences of HGPRT and APRT replacements were evaluated in most cell lines by direct enzymatic assay. These data paralleled those already reported for the DI700:H ϩ/ϩ A ϩ/ϩ , DI700:H ϩ/n A ϩ/ϩ , DI700: H n/n A ϩ/ϩ , DI700:H n/n A ϩ/h , and DI700:H n/n A h/h (21). A representative HGPRT enzyme assay is presented for the DI700:H ϩ/ ϩA ϩ/ϩ , DI700:H ϩ/n A ϩ/ϩ , DI700:H n/n A ϩ/ϩ , DI700:H n/h A ϩ/ϩ , TUBA2:H ϩ/ϩ A ϩ/ϩ , TUBA2:H n/n A ϩ/ϩ , TUBA2:H n/n A h/h , and APPB2A3:H h/n cell lines in Fig. 5A. All null mutants created by single and double gene targeting strategies expressed extremely low activities of HGPRT activity, as expected, whereas the DI700:H ϩ/ A ϩ/ϩ heterozygote expressed an intermediate level of HGPRT activity as compared with the strains that are wild type at the HGPRT locus. APRT deficiencies were established for the TUBA2:H n/n A h/h and APPB2A3:H h/n strains, whereas DI700:H ϩ/ϩ A ϩ/ϩ and TUBA2:H ϩ/ϩ A ϩ/ϩ expressed equivalent wild type APRT activity (Fig. 5B). An absolute deficiency in APRT could not be established definitively in any cell line expressing HGPRT activity because of the high rate of substrate deamination catalyzed by adenine deaminase activity present in Leishmania parasites (24). All TUBA2-derived  (21). The sizes of the fragments expected from EcoRI digestion of the wild type HGPRT allele or the rearranged hgprt allele after integration of the hygromycin resistance marker or from BamHI-SalI digestion of the wild type APRT locus are marked. The locations of probe A from the HGPRT coding region, probe B from the 3Ј flank of HGPRT, probe C from the APRT coding region, and probe D from the 5Ј flank of APRT are shown under the appropriate genomic loci. HGPRT and APRT coding regions are shown as black rectangles, whereas corresponding 5Ј (5Ј UTR) and 3Ј (3Ј UTR) untranslated regions of both loci that were amplified by PCR to create the gene targeting constructs are displayed as thin white rectangles. The antibiotic resistance marker HYG is indicated by the thick white box and is appropriately designated, and the corresponding pX-based flanking regions from the L. major dhfr-ts are indicated by the gray shaded boxes.
cell lines exhibited the expected AK deficiency of the TUBA2 clone (data not shown), while DI700 cells expressed ϳ6.2 nmol of AK activity/min/mg of protein.
Drug Resistance Phenotypes-As L. donovani HGPRT is known to catalyze the phosphoribosylation of HPP (2, 3, 15), a potential "lead" compound for the treatment of leishmaniasis, the effects of HGPRT deficiency on HPP sensitivity were evaluated in a variety of cell lines. All parental cell lines exhibited an EC 50 (effective concentration of drug that inhibits growth by 50%) for HPP of Ͻ10 M and all ⌬hgprt null mutants tested exhibited EC 50 values for HPP of Ͼ2 mM (data not shown). All ⌬hgprt lines exhibited cross-resistance to 4-thiopurinol, another pyrazolopyrimidine HGPRT substrate (2,3). Similarly, cells that were wild type at the APRT locus exhibited EC 50 values for APP of Ͻ1.0 M, whereas the EC 50 values for all ⌬aprt lines were greater than 100 M. All hgprt/HGPRT and aprt/APRT heterozygotes displayed intermediate EC 50 values for HPP (ϳ100 M) and APP (ϳ1.0 M), respectively. EC 50 values for wild type, heterozygous, and homozygous cell lines at both the HGPRT and APRT loci are similar to those published previously (21). No meaningful variations in HPP riboside sensitivity were observed among wild type and mutant cell lines (data not shown), consistent with the proposed non-phosphoribosylation-based mechanism for incorporation of this nucleoside analog into Leishmania parasites (25).

Growth of Mutant Cell Lines in Purines-
The ability of L. donovani to be propagated in completely defined growth medium enables an assessment of their ability to grow under conditions in which the purine source is varied. No differences in the rate or extent of growth were observed among L. donovani strains when the exclusive source of purine in the culture medium was hypoxanthine, guanine, adenine, xanthine, inosine, guanosine, or adenosine.
Radiolabeled Hypoxanthine Incorporation Assay-The capability of ⌬hgprt cells to grow in hypoxanthine was somewhat surprising in view of the fact that prokaryotic (26) and mammalian (27) cells lacking HGPRT activity cannot proliferate when exogenous hypoxanthine is the sole source of purine for nucleotide synthesis. As a consequence, the abilities of wild type and ⌬hgprt cells to incorporate [ 14 C]hypoxanthine into phosphorylated metabolites was compared. All cell lines tested, including DI700:H ϩ/ϩ A ϩ/ϩ , DI700:H n/h A ϩ/ϩ , DI700:H n/n A ϩ/ϩ , TUBA2:H n/n A ϩ/ϩ , TUBA2:H n/n A h/h , and APPB2A3:H h/n , could efficiently incorporate [ 14 C]hypoxanthine into their purine pool, intimating an alternative mechanism for hypoxan-  Fig. 2. The sizes of the restriction fragments hybridizing to the HGPRT and APRT probes are shown on the left in base pairs. thine conversion to the nucleotide level in ⌬hgprt cells (Fig. 6). The uptake of other purine bases was not evaluated for any of these cell lines, as adenine, the natural substrate of APRT, is funneled by Leishmania through hypoxanthine prior to incorporation into the nucleotide pool, whereas guanine is converted to xanthine, which is phosphoribosylated by XPRT, an enzyme whose function in intact parasites cannot yet be evaluated with genetic tools (8,28,29).
To further illuminate this hypoxanthine uptake into ⌬hgprt cells, the ability of high concentrations of nonradiolabeled naturally occurring purines to interfere with [ 14 C]hypoxanthine incorporation into these cell lines was compared. 100 M concentrations of either hypoxanthine, adenine, adenosine, inosine, or guanine inhibited the incorporation of 1.8 M [ 14 C]hypoxanthine into all cell lines tested by 80 -90%. Conversely, 100 M xanthine did not substantially alter the ability of wild type cells to phosphoribosylate [ 14 C]hypoxanthine, whereas all of the ⌬hgprt strains were affected noticeably. Representative [ 14 C]hypoxanthine incorporation data are presented in Fig. 7 for the wild type DI700:H ϩ/ϩ A ϩ/ϩ and triple mutant TUBA2: H n/n A h/h cell lines. Neither HPP or HPP riboside affected hypoxanthine incorporation rates into any of the L. donovani cell lines (data not shown).
As preliminary experiments (see Fig. 7) implied that xanthine in the culture medium selectively interfered with the ability of ⌬hgprt cells to incorporate [ 14 C]hypoxanthine, the effects of increasing concentrations of xanthine on [ 14 C]hypoxanthine uptake were compared for wild type DI700:H ϩ/ϩ A ϩ/ϩ and 2 ⌬hgprt strains, DI700:H n/h A ϩ/ϩ , and DI700:H n/n A ϩ/ϩ cells. As demonstrated in Fig. 8, xanthine could block [ 14 C]hypoxanthine incorporation into nucleotides by Ͼ95% at high concentrations in DI700:H n/h A ϩ/ϩ and DI700:H n/n A ϩ/ϩ cells, whereas wild type DI700:H ϩ/ϩ A ϩ/ϩ cells expressed a significant xanthine-refractory [ 14 C]hypoxanthine uptake component.

DISCUSSION
The ability of Leishmania to undergo efficient homologous recombination (7, 10) and the relatively recent development of methodologies to transfect these parasites with exogenous DNA (13,14) have enabled this genetic dissection of the purine salvage pathway by targeted gene replacement strategies. This genetic analysis has been greatly facilitated by the availability of mutationally derived cell lines deficient in either APRT (8) or AK (17) activity, thereby expediting the construction of cell lines with multiple mutations. Previous biochemical, genetic, and radiolabel incorporation studies have implied that HG-PRT, APRT, and AK are three enzymes that play primary roles in purine salvage by L. donovani (8,17,28). XPRT may play an important role in accessing host guanylate pools (28,29), but adenylate nucleotides are the predominant host purines. A nucleoside phosphotransferase activity has also been detected in L. donovani, albeit in low amounts, but is only known to recognize the purine analog, HPP riboside (25). As the L. donovani HGPRT (15) and APRT (16) and their respective flanking sequences have been isolated, a plethora of ⌬hgprt and ⌬aprt homozygous null mutants lines have been generated by targeted gene replacement within the wild type, aprt Ϫ , and ak Ϫ backgrounds. Strains constructed for this genetic evaluation of the purine pathway include: ⌬hgprt, ⌬aprt, ⌬hgprt/⌬aprt (as well as aprt Ϫ /⌬hgprt), ⌬hgprt/ak Ϫ , ⌬aprt/ak Ϫ , and ⌬hgprt/ ⌬aprt/ak Ϫ , and all of the obligatory heterozygous progenitors. From a biochemical perspective, targeted gene replacement allowed the creation of mutant null phenotype parasites defective in the three purine salvage enzymes in every conceivable combination, including three single mutants, hgprt Ϫ , aprt Ϫ , and ak Ϫ ; three double mutants, hgprt Ϫ /aprt Ϫ , hgprt Ϫ /ak Ϫ , and aprt Ϫ -/ak Ϫ ; and one triple mutant, hgprt Ϫ /aprt Ϫ /ak Ϫ (Fig. 1). All of these strains are theoretically syngeneic except for the rearranged HGPRT and APRT loci encompassing the drug resistance cassettes and for the uncharacterized mutations that conferred deficiencies in either AK or APRT activity.
Critical to the creation of L. donovani strains with multiple mutations was the ability to generate homozygous null mutants at either the HGPRT or APRT locus with only a single targeting construct. This preserved the other drug resistance marker employed in these studies for subsequent genetic manipulations, thereby allowing the creation of cells with more than one homozygous mutation, and resulted in a substantial savings both in time required for the construction of all of the mutant cell lines and in financial resources. Eliminating both alleles with only a single targeting vector required first the creation of the heterozygotes by homologous recombination of drug resistance cassettes containing 5Ј-and 3Ј-flanking sequences of HGPRT or APRT followed by selection for LOH in subversive substrate. LOH has been demonstrated previously in Leishmania at the HGPRT, APRT, and dihydrofolate reductase-thymidylate synthase loci (21,30), and these present studies embellish the previous findings by demonstrating that LOH at the HGPRT and APRT loci is neither marker-nor cell line-specific. The mechanism for LOH in Leishmania is not known but could include chromosome loss, chromosome nondisjunction, gene deletion, mitotic recombination, gene conversion, or possibly homologous recombination of stably maintained episomal DNAs originating either from the targeting construct or the rearranged chromosome locus. Simple chromosome loss (31), chromosome loss followed by duplication (31,32), and mitotic recombination (32,33) have all been demonstrated as mechanisms of LOH in mammalian cell systems. Overall, the ability to create null mutants with single targeting constructs greatly facilitates genetic investigations into complex and divagating metabolic pathways, such as that for purine salvage, and the availability of additional drug resistance markers for Leishmania parasites (34 -36) should allow the construction of knockout parasites bearing even more than the 2 homozygous mutations.
The genetic studies with HGPRT and APRT targeting constructs in L. donovani have revealed some unique and unexpected features of the purine salvage pathway in this protozoan parasite. Foremost, mutants defective in HGPRT, APRT, and/or AK activity retain the capacity to proliferate in completely defined medium in which the sole exogenous purine source is any of the four naturally occurring purine nucleobases, hypoxanthine, xanthine, guanine, or adenine, or the nucleosides adenosine, inosine, or guanosine. In contrast, mammalian cells rendered pharmacologically auxotrophic for purines with inhibitors of purine biosynthesis cannot salvage hypoxanthine through HGPRT (27), adenine through APRT (37), or adenosine through AK (38) when the germane enzyme is missing. Indeed, this pharmacologically induced purine deprivation has served as a valuable positive selectable marker for expression of these purine salvage enzymes in animal cell systems (27,(37)(38)(39). The viability of the mutant L. donovani strains in any single purine can be attributed to their ability to incorporate hypoxanthine efficiently into the purine pool. HGPRT-deficient mammalian cells cannot incorporate hypoxanthine (8,27,39). The uptake of other purine bases was not evaluated in these mutant strains, since it has been previously demonstrated that wild type L. donovani promastigotes funnel adenine through hypoxanthine prior to incorporation into the nucleotide pool and metabolize guanine and xanthine through XPRT (28,29). The channeling of other purines into hypoxanthine by L. donovani promastigotes is supported by our competition studies that demonstrate that excess adenosine, inosine, and adenine are as effective as equimolar concentrations of hypoxanthine in blocking [ 14 C]hypoxanthine incorporation into nucleotides (see Fig. 7). Guanine is a bit less effective at impeding [ 14 C]hypoxanthine incorporation, presumably because it is rapidly oxidatively deaminated to xanthine (28,29). The channeling of inosine, adenine, and adenosine into hypoxanthine can be respectively ascribed to the high levels of of nucleoside hydrolase (40), adenine deaminase (24), and adenosine hydrolase/phosphorylase (17) activities present in L. donovani promastigotes. None of these enzymes is expressed by mammalian cells.
The mechanism by which hypoxanthine, as well as the purines that are transformed into hypoxanthine, are assimilated into the nucleotide pool in ⌬hgprt/⌬aprt L. donovani is currently under study. The fact that high concentrations of xanthine only partially inhibit [ 14 C]hypoxanthine incorporation into wild type parasites but virtually obliterate [ 14 C]hypoxanthine uptake into ⌬hgprt strains (Fig. 8) implies that XPRT, an enzyme that is expressed in L. donovani (41), plays a central, but obviously not exclusive, role in hypoxanthine metabolism in this protozoan parasite. Two mechanisms could account for this inhibition of [ 14 C]hypoxanthine uptake by xanthine; either XPRT recognizes hypoxanthine as a substrate, or hypoxanthine is oxidized to xanthine prior to its phosphoribosylation. The latter hypothesis is unlikely, however, since L. donovani promastigotes lack detectable xanthine oxidase/dehydrogenase activity (41). The different extents by which xanthine inhibits [ 14 C]hypoxanthine incorporation into wild type and ⌬hgprt strains can be imputed to the fact that the former expresses two enzymes that phosphoribosylate hypoxanthine, whereas the ⌬hgprt mutants lacks the xanthine-refractory pathway, i.e.

HGPRT.
Little is known about XPRT from Leishmania, although mammalian cells lack the activity (41). However, it would appear that the enzyme might serve as a target for the selective incorporation of cytotoxic xanthine analogs, such as 6-thioxanthine, which is phosphoribosylated by the hypoxanthine-guanine-xanthine PRT (HGXPRT) of Toxoplasma gondii (42,43). Tuttle and Krenitsky have chromatographically distinguished the L. donovani XPRT and HGPRT activities, but the substrate specificity of XPRT could not be evaluated due to the inherent instability of the native enzyme in solution (41). The L. donovani HGPRT, however, does not recognize xanthine (15,41). Attempts are currently under way to isolate and overexpress the XPRT gene from L. donovani to provide sufficient recombinant protein for ultimate biochemical and structural characterization.
Enzyme and radiolabel incorporation studies with other protozoan parasites have revealed considerable differences in the purine salvage pathways among various genera (1). Thus, conclusions drawn from these studies on the purine pathway of L. donovani promastigotes are unlikely to have general applicability to other parasites, except perhaps to other parasites of the Trypanosomatidae family, e.g. Trypanosoma brucei, the etiologic agent of African sleeping sickness, and Trypanosoma cruzi, the parasite that causes Chagas disease. Moreover, there are differences between the purine salvage pathways of promastigotes and amastigotes, the latter being the infective form of the parasite. For instance, L. donovani amastigotes lack adenine deaminase and cannot convert, therefore, other purines into hypoxanthine (44). A similar genetic dissection of the purine pathway in amastigotes by targeted gene replacement will be undertaken with L. mexicana, a species that can be propagated as promastigotes, axenic amastigote-like forms (45), and amastigotes in macrophage lines (46), and the genes encoding L. mexicana HGPRT and APRT have already been isolated for these purposes. Extensive genetic studies on the purine pathway of other protozoan parasites have not been accomplished, although mutations in either AK (47) or HGX-PRT (42,43) have no apparent effect on survival of T. gondii. Preliminary experiments do suggest, however, that double knock-outs of both AK and HGXPRT activities may be lethal. 2 If these preliminary experiments are sustained, the HGXPRT activity plays a sufficient but not obligatory role in purine salvage in T. gondii. In contrast, neither HGPRT, APRT, nor AK, alone or in any combination, are essential to purine salvage by or survival of L. donovani promastigotes.
The purine auxotrophy of Leishmania and its unique purine salvage pathway offers a number of targets for therapeutic consideration. One therapeutic paradigm has been inhibitor design to block purine acquisition from the host. The channeling of extracellular purines through hypoxanthine (28,39) had suggested that HGPRT would be an appropriate target for inhibitor design, but our genetic investigations demonstrate, at least in promastigotes, that specific inhibitors of HGPRT alone would be therapeutically ineffective. Similar conclusions can be drawn about the potential effectiveness of APRT and AK inhibitors as a consequence of these studies. Indeed, targeting all three enzymes in combination may not be a valid therapeutic paradigm. However, if XPRT in conjunction with HGPRT proves indispensable to purine salvage by this protozoan parasite, one might envision that a common mechanism-based inhibitor might have thereapeutic potential or that targeting both enzymes with a combination of inhibitors might be a rational thereapeutic strategy. Targeting multiple steps in a metabolic pathway has considerable precedence in antiparasitic therapies (48,49). Testing XPRT function in both wild type and ⌬hgprt backgrounds can be accomplished by similar targeted gene replacement approaches, once the L. donovani XPRT is isolated.