Cloning of a Novel Inosine-Guanosine Transporter Gene fromLeishmania donovani by Functional Rescue of a Transport-deficient Mutant*

Purine transport is an indispensable nutritional function for protozoan parasites, since they are incapable of purine biosynthesis and must, therefore, acquire purines from the host milieu. Exploiting a mutant cell line (FBD5) of Leishmania donovanideficient in inosine and guanosine transport activity, the gene encoding this transporter (LdNT2) has been cloned by functional rescue of the mutant phenotype. LdNT2 encodes a polypeptide of 499 amino acids that shows substantial homology to other members of the equilibrative nucleoside transporter family. Molecular analysis revealed that LdNT2 is present as a single gene copy within the leishmanial genome and encodes a single transcript of 3 kilobase pairs. Transfection of FBD5 parasites with LdNT2re-established their ability to take up inosine and guanosine with a concurrent restoration of sensitivity to the inosine analog formycin B. Kinetic analyses reveal that LdNT2 is highly specific for inosine (K m = 0.3 μm) and guanosine (K m = 1.7 μm) and does not recognize other naturally occurring nucleosides. Expression ofLdNT2 cRNA in Xenopus oocytes significantly augmented their ability to take up inosine and guanosine, establishing that LdNT2 by itself suffices to mediate nucleoside transport. These results authenticate genetically and biochemically that LdNT2 is a novel nucleoside transporter with an unusual and strict specificity for inosine and guanosine.

Purine transport is an indispensable nutritional function for protozoan parasites, since they are incapable of purine biosynthesis and must, therefore, acquire purines from the host milieu. Exploiting a mutant cell line (FBD5) of Leishmania donovani deficient in inosine and guanosine transport activity, the gene encoding this transporter (LdNT2) has been cloned by functional rescue of the mutant phenotype. LdNT2 encodes a polypeptide of 499 amino acids that shows substantial homology to other members of the equilibrative nucleoside transporter family. Molecular analysis revealed that LdNT2 is present as a single gene copy within the leishmanial genome and encodes a single transcript of 3 kilobase pairs. Transfection of FBD5 parasites with LdNT2 reestablished their ability to take up inosine and guanosine with a concurrent restoration of sensitivity to the inosine analog formycin B. Kinetic analyses reveal that LdNT2 is highly specific for inosine (K m ‫؍‬ 0.3 M) and guanosine (K m ‫؍‬ 1.7 M) and does not recognize other naturally occurring nucleosides. Expression of LdNT2 cRNA in Xenopus oocytes significantly augmented their ability to take up inosine and guanosine, establishing that LdNT2 by itself suffices to mediate nucleoside transport. These results authenticate genetically and biochemically that LdNT2 is a novel nucleoside transporter with an unusual and strict specificity for inosine and guanosine.
Leishmania donovani is a protozoan parasite and the etiologic agent of visceral leishmaniasis, a devastating and invariably fatal disease if untreated. The parasite exhibits an intricate life cycle in which the extracellular, flagellated promastigote exists in the phlebotomine sandfly vector, and the intracellular amastigote resides in the phagolysosome of macrophages and other reticuloendothelial cells of the mammalian host. Drugs are the only defense against visceral leishmaniasis, but the efficacy of these empirically derived agents is compromised both by drug toxicity and resistance (1). Thus, it is increasingly imperative to identify new and unique biochemical targets in the parasite for potential therapeutic exploitation.
Among the most conspicuous metabolic differences between parasites and their mammalian hosts is the purine pathway. Whereas animal cells synthesize purine nucleotides de novo, all protozoan parasites are incapable of synthesizing purines and depend upon purine acquisition from their hosts to survive and proliferate (2). Hence, each genus of parasite has evolved a unique complement of purine salvage enzymes in order to scavenge purines from the host milieu (2). The first step in this salvage process involves the translocation of purines across the parasite plasma membrane, a process mediated by membrane permeases. These permeases also initiate the uptake of pyrazolopyrimidine nucleobase and nucleoside analogs of hypoxanthine and inosine that are selectively toxic to Leishmania spp. (3,4). Thus, purine transporters play vital roles in both purine nutrition and antiparasitic drug targeting intimating that these membrane proteins could be targets for either inhibitor or cytotoxic ligand development.
Genetic and biochemical investigations have demonstrated that L. donovani promastigotes express two nucleoside transport activities of nonoverlapping ligand specificity (5). The first, LdNT1, transports adenosine, pyrimidine nucleosides, and the cytotoxic adenosine analog tubercidin, and the second, LdNT2, recognizes inosine, guanosine, and the cytotoxic inosine isomer formycin B (FoB) 1 (5,6). Mutant L. donovani clones deficient in LdNT1 or LdNT2 activity have been isolated by virtue of their resistance to either tubercidin or FoB, respectively (5). The availability of these transport-deficient mutants and the ability to transfect Leishmania with cosmid expression libraries (7) provided an avenue for cloning the genes encoding nucleoside transporter proteins by selecting for functional recovery of the wild type drug sensitivity phenotype and, thus, nucleoside transport capability. This functional rescue scheme was previously employed to clone the LdNT1 locus (8).
Functional rescue has now been exploited to isolate LdNT2. LdNT2 is present as a single copy gene within the leishmanial genome and encodes a transcript of ϳ3 kb. Functional expression of LdNT2 in nucleoside transport-deficient L. donovani and in Xenopus laevis oocytes revealed LdNT2 to be a novel high affinity inosine-guanosine transporter with a singular predicted membrane topology. Parasite Cell Culture-L. donovani wild type (DI700) and LdNT2deficient (FBD5) parasites (5) were propagated at 26°C in Dulbecco's modified Eagle's-Leishmania (DME-L) medium (9) containing 100 M xanthine as a purine source. The FBD5 cells were maintained continuously in DME-L supplemented with 1 M FoB to ensure that the population did not harbor any wild type revertants. Transfectants were propagated in DME-L supplemented with 10% fetal bovine serum (Life Technologies, Inc.) and 50 g ml Ϫ1 of hygromycin B (DME-L/FBS/HYG).
Transfection and Isolation of Cosmids-To isolate cosmids containing LdNT2, 30 independent transfections were performed, using previously reported parameters (10), on exponentially growing FBD5 cells resuspended at 1 ϫ 10 8 cells ml Ϫ1 in electroporation buffer (10). FBD5 promastigotes were transfected with 10 g of DNA prepared from a cosmid library of L. donovani Ld4 strain DNA in the shuttle vector cLHYG, which encompasses the hygromycin phosphotransferase gene (7). Transfectants were maintained in the absence of hygromycin B for 24 h post-transfection, after which time they were plated on semi-solid (1% Noble agar, Difco) DME-L/FBS/HYG and incubated at 26°C in 5% CO 2 . After 2-3 weeks, isolated colonies were picked and inoculated into individual wells of 96-well microtiter plates containing 200 l of DME-L/FBS/HYG and proliferated at 26°C in 5% CO 2 until individual wells contained at least 5 ϫ 10 6 cells ml Ϫ1 . To screen for FoB-sensitive FBD5 transfectants, aliquots of 10 l from individual wells of the master microtiter plate were transferred into wells of two replica 96-well microtiter plates, one plate containing DME-L/FBS/HYG supplemented with 1 M FoB and the other containing DME-L/FBS/HYG alone. Clones that were determined to be susceptible to 1 M FoB were expanded from the appropriate wells in the master microtiter plates, and the cosmids were rescued from these transfectants by alkaline lysis (10).
Isolation of LdNT2-Cosmids F30F12 and F44H9 were subjected to restriction digestion with BamHI, BglII, EcoRI, EcoRV, HindIII, NotI, SacII, and XbaI. To localize the LdNT2 gene within the F30F12 cosmid, various restriction fragments were subcloned into the leishmanial shuttle vector pSNAR, encompassing the neomycin phosphotransferase gene (11), and tested for restoration of FoB sensitivity after transfection into FBD5 cells and selection in 20 g ml Ϫ1 G418. An ϳ5-kb EcoRV fragment that conferred the appropriate phenotype was sequenced in its entirety on a Perkin-Elmer Applied Biosystems 377 DNA automated sequencer using dye-terminator cycle methodology by the Core Facility located within the Department of Molecular Microbiology and Immunology at the Oregon Health Sciences University. The LdNT2 open reading frame (ORF) was identified by BLAST searching of available data bases (12) and subsequently sequenced in both directions. Pairwise alignments with other members of the equilibrative nucleoside transporter (ENT) family (8,(13)(14)(15)(16) were performed using the algorithm of Needleman and Wunsch (17). Multisequence alignments were conducted using the Feng-Doolittle algorithm (18). Membrane spanning domains were deduced from hydropathy plots constructed from three independent algorithms (19 -21).
Southern Analysis-Genomic DNA was isolated from L. donovani promastigotes according to standard procedures (22). Restriction enzyme-digested DNA was blotted onto GeneScreen Plus® hybridization transfer membrane (NEN Life Science Products) and hybridized to a 2.2-kb HindIII fragment derived from the 5-kb EcoRV fragment in pSNAR that contains the LdNT2 ORF.
RNA Extraction and Northern Blotting-Total cellular RNA was isolated from ϳ5 ϫ 10 8 exponentially growing L. donovani promastigotes using the RNeasy Midi kit (Qiagen Inc., Valencia, CA). Poly(A) ϩ RNA, prepared from total RNA using the Oligotex mRNA mini kit (Qiagen Inc., Valencia, CA), was subjected to denaturing agarose electrophoresis, transferred to a GeneScreen Plus® hybridization transfer membrane (NEN Life Science Products), and probed with the 2.2-kb HindIII fragment described above. Signals were normalized by hybridization to probes corresponding to Leishmania enriettii ␣-tubulin gene (23).
Vectors for Transport Assays-LdNT2-mediated transport was measured both in FBD5 L. donovani and X. laevis oocytes. For expression in oocytes, a 2.2-kb HindIII fragment encompassing the entire LdNT2 ORF, ϳ100 bp of 5Ј-untranslated region, and 600 bp of 3Ј-untranslated region was excised from the 5-kb EcoRV fragment and subcloned into the pOG-1 oocyte expression vector (24), creating pLdNT2-OG-1. For expression in L. donovani, an ϳ2.5-kb BamHI-NotI fragment from pLdNT2-OG-1 encompassing the 2.2-kb HindIII LdNT2 fragment and ϳ300 bp of pOG-1 vector-derived sequence encoding the 3Ј-untranslated region of the X. laevis ␤-globin gene was subcloned into the appropriate sites within the leishmanial expression vector pALTNEO (25) generating pLdNT2-ALTNEO.
Nucleoside Uptake into FBD5 Cells Expressing LdNT2-FBD5 promastigotes were transfected with either pLdNT2-ALTNEO or pALT-NEO, and the resulting transfectants were maintained continuously in DME-L/FBS supplemented with 25 g ml Ϫ1 G418. Uptake of [ 3 H] inosine (0.31 Ci mmol Ϫ1 ) and [ 3 H]guanosine (0.05 Ci mmol Ϫ1 ) was measured in FBD5 cells harboring either pLdNT2-ALTNEO or pALT-NEO as described previously (6). Briefly, FBD5 promastigotes (4 ϫ 10 8 cells ml Ϫ1 ), resuspended in phosphate-buffered saline, pH 7.4, and supplemented with 10 mM D-glucose, were mixed with radiolabel for various times. Uptake was terminated by a modified oil-stop technique using a dibutyl phthalate cushion (6). All transport measurements on the LdNT2 transfectants were performed on stationary phase parasites, since inosine transport into exponentially growing parasites was nonlinear at concentrations proximal to the K m value even after a few seconds. Technical limitations of the transport assay precluded the use of fewer cells. Initial rates for each nucleoside concentration were determined by linear regression analysis over the linear portions of the assay, and K m values were determined by Hanes analysis. Competition experiments were performed in the same buffer containing 1 M [ 3 H] inosine (0.31 Ci mmol Ϫ1 ) and 100 M unlabeled inhibitor.
Nucleoside Uptake into X. laevis Oocytes-Oocytes were dissected and defolliculated as described previously (26,27) and maintained at 16°C in frog Ringer's solution supplemented with 2.5 mM pyruvate, 0.5 mM theophylline, and 50 g/ml gentamycin (Life Technologies, Inc.). The pLdNT2-OG-1 plasmid was linearized with NotI and capped cRNA synthesized in the presence of the cap analog m 7 GpppG by T7 polymerase (28). Stage V-VI oocytes were microinjected with 5-20 ng of cRNA 1 day after defolliculation. Control oocytes were injected with equivalent volumes of water. [ 3 H]Inosine (31.3 Ci mmol Ϫ1 ) and [ 3 H]guanosine (5.0 Ci mmol Ϫ1 ) uptake was assayed after 3 days of cRNA expression. Each oocyte was dissolved in 0.25 ml of 5% sodium dodecyl sulfate, and oocyte-associated radiolabel incorporation was quantitated by liquid scintillation spectrometry.

RESULTS
Cloning of LdNT2 by Functional Rescue of the FoB Sensitivity Phenotype-The cloning of LdNT2 was predicated on the functional restoration of a wild type phenotype (FoB-sensitive, nucleoside transport-competent) in a FoB-resistant, nucleoside transport-deficient (FBD5) background (5). Six thousand independent hygromycin B-resistant colonies representing more than five genome equivalents of leishmanial DNA (7) were picked after transfection with a leishmanial cosmid library and tested for sensitivity to 1 M FoB. Of these, only two colonies exhibited a wild type FoB-sensitive phenotype, and both were subsequently determined to be inosine transport-proficient (data not shown). The cosmids, designated F30F12 and F44H9, were recovered from the two FoB-sensitive transfectants and subjected to restriction endonuclease analysis with the enzymes EcoRI, NotI, HindIII, XbaI, SacII, EcoRV, BglII, and BamHI. Restriction mapping revealed that both cosmids were distinct, but both contained a common ϳ11-kb EcoRI fragment. The cosmid F30F12 (Fig. 1A) was selected for further analysis. The location of the LdNT2 gene was determined by subcloning fragments of F30F12 DNA into the leishmanial transfection vector pSNAR (11), transfecting them back into FBD5 cells, and testing for concomitant restoration of FoB sensitivity and inosine transport capability (Fig. 1). A 5-kb EcoRV fragment conferring the appropriate phenotype was sequenced in its entirety, and an ORF (LdNT2) was identified that upon conceptual translation encoded a hydrophobic polypeptide with significant homology to other parasite and mammalian equilibrative nucleoside transporters (8, 13-16, 26, 29).
Molecular Characterization of LdNT2 in Wild Type and FBD5 Cells-Southern blot analysis of L. donovani genomic DNA digested with a battery of restriction enzymes that cut either within (PstI, PvuI, and SalI) or outside (EcoRI and HindIII) the ORF indicated that LdNT2 was a single copy gene (Fig. 4A). This result is compatible with the nucleotide sequence of the 5-kb EcoRV fragment that encompasses LdNT2. The Southern blot of FBD5 genomic DNA hybridized with the LdNT2 ORF was identical to that of wild type DNA (Fig. 4A), demonstrating that neither a gross deletion nor rearrangement of the LdNT2 locus confers the nucleoside transport-deficient phenotype on FBD5 cells.
Northern analysis of wild type poly(A) ϩ RNA revealed a major LdNT2 transcript of ϳ3-kb (Fig. 4B). A fainter hybridizing band at ϳ5 kb was also observed, which could conceivably be an unprocessed mRNA. FBD5 cells, as well as FBD5 cells transfected with either pALTNEO or pLdNT2-ALTNEO, also expressed both the major and minor transcripts, indicating that loss of LdNT2 function in FBD5 cells cannot be attributed to a lack of LdNT2 transcription (Fig. 4B). As expected, LdNT2 is overexpressed in the FBD5 strain transfected with pLdNT2-ALTNEO. The size of this transcript, which arises from splice acceptors within the vector and insert, is also ϳ3 kb. Levels of poly(A) ϩ RNA were normalized for each cell line using the L. enriettii ␣-tubulin gene (23). There is an additional band that hybridizes to the ␣-tubulin probe in the pLdNT2-ALTNEO lane, suggestive of an alternatively spliced transcript, the reason for which is unclear.
Functional Characterization of LdNT2 in L. donovani-To establish that LdNT2 is a functional nucleoside transporter, LdNT2 was subcloned into the pALTNEO leishmanial expression vector and transfected into FBD5 cells. Uptake assays using 5 M [ 3 H]inosine confirmed that transfection with LdNT2 bestows a robust inosine transport phenotype on FBD5 cells (71 pmol/s/10 8 cells) (Fig. 5). FBD5 cells transfected with pALTNEO alone displayed only minimal inosine transport capability (1 pmol/s/10 8 cells) (Fig. 5). Substrate saturation curves with FBD5 pLdNT2-ALTNEO cells revealed that [ 3 H] inosine and [ 3 H]guanosine transport displayed Michaelis-Menten kinetics with an apparent K m value of 0.3 Ϯ 0.1 (n ϭ 4) and 1.7 Ϯ 0.5 M (n ϭ 3), respectively. Representative experiments for both inosine and guanosine are displayed in Fig. 6. These values are comparable to those determined for wild type parasites (data not shown and Ref. 6).
The phenotypic consequence of restored nucleoside transport proficiency in FBD5 cells was also assessed by growth phenotype in FoB. The effective concentration of FoB that inhibited growth of the LdNT2 transfectants by 50% (EC 50 value) was 4.5 Ϯ 3.9 nM (n ϭ 4), a value similar to that determined for wild type promatigotes (5.1 Ϯ 3.5 nM) (n ϭ 4) (5). In contrast, FBD5 cells transfected with the empty pALTNEO vector exhibited an EC 50 value of 5.7 Ϯ 0.9 M (n ϭ 4). Functional Expression of LdNT2 in X. laevis Oocytes-To confirm that LdNT2 by itself encodes a functional inosineguanosine transporter, LdNT2 cRNA was expressed in Xenopus oocytes. Oocytes injected with the LdNT2 cRNA transported [ 3 H]inosine and [ 3 H]guanosine 10 -20-fold more efficiently than water-injected control oocytes (Fig. 8). Rates of uptake were 0.36 and 0.32 pmol min Ϫ1 (oocyte) Ϫ1 for inosine and guanosine, respectively.

DISCUSSION
LdNT2 encoding the L. donovani inosine-guanosine transporter was identified and cloned after screening for restoration of a wild type phenotype in inosine-guanosine transport-deficient FBD5 cells following transfection with an LdNT2-con-taining cosmid. A similar functional rescue strategy has also been exploited to pinpoint and isolate the LdNT1 locus encoding the L. donovani adenosine-pyrimidine nucleoside transporters (8) and offers a powerful genetic approach toward the isolation of any gene for which a selection or screen can be devised. Other leishmanial genes that have been isolated by complementation of mutant phenotypes include several involved in the biosynthesis of lipophosphoglycan (31,32), an important cell surface glycoconjugate and one crucial for biogenesis of the glycosome (33), a unique kinetoplastid peroxisomal-like microbody that accommodates glycolytic and other nutritional enzymes (34).
The predicted amino acid sequence indicates that LdNT2 is a member of the ENT family. This family includes transporters FIG. 2. Protein sequence alignments with LdNT2. The deduced protein sequence of LdNT2 was aligned with hENT1 (13), hENT2 (14,15), TbNT2 (16), and LdNT1 (8), by the method of Feng and Doolittle (18). Amino acids identical among all five transporters are shaded black, and conserved amino acids are shaded gray. The predicted membranespanning domains for hENT1 and LdNT2 are indicated by the solid lines above and below the aligned proteins and are numbered sequentially.

FIG. 3. Predicted topology of LdNT2.
LdNT2 topology was predicted by three hydropathy algorithms (19 -21), which predict nine membrane-spanning domains and an NH 2 terminus on the opposite side of the membrane to the large hydrophilic loop between membranespanning domains 5 and 6. Open circles indicate both hydrophobic and polar residues; open circles with either ϩ or Ϫ indicate charged residues; gray circles with the letter N indicate potential N-linked glycosylation sites; and black circles with white letters indicate residues that are invariant between LdNT2 and the four other transporters aligned in Fig. 3. from mammalian cells (13)(14)(15), a variety of protozoan parasites (8,16,26,29,35), and as yet many functionally uncharacterized ORFs from eukaryotic cells uncovered among a variety of genome sequencing projects (13). ENTs are distinct from bacterial nucleoside transporters (36,37), as well as from mammalian concentrative nucleoside transporters (38,39) both in their primary sequences and topological profiles. Multisequence alignment of LdNT2 with other members of this family reveal that LdNT2 shares a number of common residues, most of which reside in predicted membrane-spanning domains, including the charged residues aspartate 389 and arginine 393 within predicted TM 7 (Fig. 2). Hydropathy predictions suggest that LdNT2 exhibits an unusual nine membranespanning topology with a large hydrophilic loop between TMs 5 and 6, whereas other members of the ENT family are conjectured to possess 11 membrane-spanning domains. The inferred model for LdNT2 lacks a membrane-spanning domain equivalent to TM 2 of other members of the ENT family and predicts that TM 8 encompasses two of the carboxyl-terminal membrane-spanning domains of the other transporters. If this model is correct, it suggests that the large hydrophilic loop would be on the opposite side of the membrane from the amino terminus, unlike the other ENTs. Whether this loop is exo-or endofacial in any of these transporters, however, is unknown, since these topological predictions have yet to be confirmed experimentally for any ENT member. The topological predictions for LdNT2 were made by three independent algorithms (19 -21) in which individual parameters for length of membrane-spanning regions, hydrophobicity, and surface probability were considered. These same algorithms predict 11 membrane-spanning regions for other ENT members.
Functional characterization of LdNT2 indicates that it is a novel high affinity transport system for both inosine and guanosine (Fig. 6) that excludes other purine and pyrimidine nucleosides and bases (Fig. 7). This unusual ligand specificity and affinity are distinct from all previously characterized ENTs from both mammalian cells and parasites. For instance, hENT1 and hENT2 both exhibit a broad ligand specificity for all naturally occurring purine and pyrimidine D-nucleosides (30), but the affinities of the human transporters for these nucleoside ligands are much lower than for either LdNT1 (8) or LdNT2. It is worth noting, however, a sodium-dependent nucleoside transport activity that appears to be specific for guanosine has been described in human acute promyelocytic leukemia cells (40).
LdNT2 also recognizes a variety of cytotoxic inosine and guanosine analogs, although large excesses of 2-aminopurine riboside and 9-deazainosine failed to impede inosine entry (Fig.  7). The inability of LdNT2 to recognize efficiently either 2-aminopurine riboside or adenosine implies that the exocyclic oxygen on C-6 of the purine ring is a critical determinant for ligand recognition by LdNT2. The inosine analog NBMPR, in which the C-6 hydroxyl is replaced by an S-nitrobenzyl moiety, does not inhibit LdNT2 at 100 M, a concentration 3 orders of magnitude greater than that required to inhibit hENT1 (30).
Southern analysis revealed that LdNT2 is present as a single copy within the leishmanial genome and that there are no gross anomalies in the LdNT2 locus of the mutant FBD5 line. Moreover, LdNT2-specific transcripts of a comparable size and intensity were observed in both wild type and mutant parasites. These results indicate that loss of LdNT2 function in FBD5 cells is most likely due to minor deletions or point mutations within the LdNT2 ORF. Whether this loss of function is due to mutations within both LdNT2 alleles or to mutations within one allele with a concomitant loss of heterozygosity, as has been previously observed for Leishmania (41,42), is unknown. Isolation and sequence analysis of the mutant LdNT2 locus from FBD5 cells should differentiate between these two models.
The availability of a molecular clone encoding LdNT2, a mutant strain deficient in LdNT2 activity, and both homologous and heterologous expression systems in which to assess LdNT2 function provide a foundation for a thorough analysis of nucleoside transport activity in Leishmania. LdNT2-deficient mutants, the ability to create further mutants, and the identification of conserved residues among nucleoside transporters provide an avenue for the genetic dissection of nucleoside transport both by forward and reverse genetic techniques and suggest LdNT2 as a paradigm for the study of nucleoside transport in higher eukaryotes. Particularly noteworthy is the ability to generate nonlethal loss-of-function mutants with facility.
Finally, it is worth noting that functional differences between parasite and mammalian transporters might be exploited therapeutically. Indeed, the selective toxicity of the drugs melarsoprol and pentamidine, both currently employed in the treatment of African trypanosomiasis, is mediated by their uptake on a novel adenine-adenosine transporter (P2) (43,44). Since LdNT2 differs from its human counterparts in terms of ligand specificity, ligand affinities, and inhibitor profiles, these discrepancies might ultimately be availed of pharmacologically.