Second-site suppression of a nonfunctional mutation within the Leishmania donovani inosine-guanosine transporter.

LdNT2 is a member of the equilibrative nucleoside transporter family, which possesses several conserved residues located mainly within transmembrane domains. One of these residues, Asp(389) within LdNT2, was shown previously to be critical for transporter function without affecting ligand affinity or plasma membrane targeting. To further delineate the role of Asp(389) in LdNT2 function, second-site suppressors of the ldnt2-D389N null mutation were selected in yeast deficient in purine nucleoside transport and incapable of purine biosynthesis. A library of random mutants within the ldnt2-D389N background was screened in yeast for restoration of growth on inosine. Twelve different clones were obtained, each containing secondary mutations enabling inosine transport. One mutation, N175I, occurred in four clones and conferred augmented inosine transport capability compared with LdNT2 in yeast. N175I was subsequently introduced into an ldnt2-D389N construct tagged with green fluorescent protein and transfected into a Deltaldnt1/Deltaldnt2 Leishmania donovani knockout. GFP-N175I/D389N significantly suppressed the D389N phenotype and targeted properly to the plasma membrane and flagellum. Most interestingly, N175I increased the inosine K(m) by 10-fold within the D389N background relative to wild type GFP-LdNT2. Additional substitutions introduced at Asn(175) established that only large, nonpolar amino acids suppressed the D389N phenotype, indicating that suppression by Asn(175) has a specific size and charge requirement. Because multiple suppressor mutations alleviate the constraint imparted by the D389N mutation, these data suggest that Asp(389) is a conformationally sensitive residue. To impart spatial information to the clustering of second-site mutations, a three-dimensional model was constructed based upon members of the major facilitator superfamily using threading analysis. The model indicates that Asn(175) and Asp(389) lie in close proximity and that the second-site suppressor mutations cluster to one region of the transporter.

Leishmania donovani is a protozoan parasite that causes visceral leishmaniasis, a disease that is invariably fatal if untreated. The genus is digenetic, existing as the extracellular promastigote in the phlebotomine sandfly vector and as the intracellular amastigote within the phagolysosome of macrophages in the infected mammalian host. The current arsenal of drugs employed to treat leishmaniasis is far from ideal, and drug efficacy is compromised by both toxicity and therapeutic failure. The need for additional drugs is therefore acute.
Rational approaches to drug development dictate exploitation of novel targets within the parasite. Perhaps the most striking metabolic disparity between protozoan parasites and their mammalian hosts is the inability of the former to synthesize purine nucleotides de novo (1). Thus, purine acquisition from the host is an indispensable nutritional function for all protozoan parasites, and many unique purine salvage enzymes have been identified and characterized (1). The initial component of purine uptake involves the translocation of purines into the parasite, a process that is mediated by nucleoside and nucleobase transporters.
Nucleoside permeation into L. donovani is mediated by two high affinity transporters with nonoverlapping ligand specificity, LdNT1 and LdNT2, both of which are members of the equilibrative nucleoside transporter (ENT) 1 family. LdNT1 transports adenosine and pyrimidine nucleosides, whereas LdNT2 is selective for inosine and guanosine (2,3). Essentially nothing is known about the permeation mechanism of the ENTs, and only a few amino acids that govern ligand recognition have been identified. Conserved Gly residues within transmembrane domains (TM) 4 and 5 of human ENT1 (hENT1) and TM5 of LdNT1 have been shown to be critical for transport function and ligand selectivity (4,5), and a Leu within TM2 of hENT1 has also been revealed as a ligand specificity determinant (6). Additional studies have supported a role for TMs 3-6 (7) and Met 33 within TM1 of hENT1 in the binding of dipyridamole and dilazep (8). Dipyridamole and dilazep are potent inhibitors of hENT1 but do not significantly affect the majority of parasite nucleoside transporters (9).
Multisequence alignments of ENT family members reveal a number of conserved or "signature" residues, located predominantly in predicted TMs (10). The most striking among these signatures are two charged residues, an Asp and an Arg, that are located within TM8. The evolutionary conservation of these charged residues within a TM implies their significance in transporter structure or function. Site-directed mutagenesis studies revealed that although Asp 389 mutants of LdNT2 are produced and localize normally to the cell surface, they are severely compromised in function. Thus, Asp 389 is essential for both inosine and guanosine permeation supporting a functional role of this residue in nucleoside translocation (11).
To elucidate further the role of Asp 389 in the translocation mechanism of LdNT2, a screen for second-site suppressors was implemented using the nonfunctional ldnt2-D389N mutant. The screen was performed with ade2 Saccharomyces cerevisiae that are mutationally deficient in purine biosynthesis and naturally incapable of purine nucleoside transport, although they maintain the capacity to transport purine nucleobases and can therefore survive with exogenous adenine (6). A number of second-site suppressor mutations were identified. However, one mutation, N175I, occurred with considerable frequency and was subsequently shown to significantly suppress the D389N phenotype in yeast and Leishmania. The distribution of mutations throughout the TMs 1-8 underscores the premise that Asp 389 is a conformationally sensitive residue, because diverse mutations suppress the D389N null phenotype. Finally, a tertiary topology prediction of LdNT2 is presented, a model that is supported by previously reported biochemical and genetic data. This model indicates that the second-site suppressor mutations cluster in one region of the transporter and, perhaps most significantly, that Asn 175 and Asp 389 are located proximal to each other.

EXPERIMENTAL PROCEDURES
Strains and Culture Methods-The S. cerevisiae strain YPH499 (Mata ura3-52 lys2-801 ade2-101 trp1-⌬63 his3-⌬200 leu2⌬1) was constructed by Sikorski and Hieter (12). Rich (YPD) and minimal (SC) medium plates were prepared as described, and standard methods were used for genetic manipulation of yeast (13). All Escherichia coli transformations were performed in the DH5␣ strain (Invitrogen) by using usual methodologies (14). L. donovani promastigotes were cultured at 26°C in DME-L medium (Invitrogen) as described (15). The construction and phenotypic characterization of the ⌬ldnt1/⌬ldnt2 L. donovani null mutant in which both copies of LdNT1 and LdNT2 were eliminated by targeted gene replacement followed by loss-of-heterozygosity will be described elsewhere. The ⌬ldnt1/⌬ldnt2 strain was cultured continuously in 50 g/ml hygromycin (Roche Applied Science) and 50 g/ml phleomycin (Research Products International, Mount Prospect, IL) for which the selective markers hygromycin phosphotransferase and phleomycin-binding protein, used in the gene replacement strategy, confer resistance. Cell lines generated by transfection of ⌬ldnt1/⌬ldnt2 with pXG-GFPϩ2Ј (16) were selected and maintained in 25 g/ml geneticin (BioWhittaker, Inc., Walkersville, MD) as well as 50 g/ml hygromycin and 50 g/ml phleomycin.
Generation of pRS424-Cu-ldnt2-D389N for the Second-site Suppressor Screen-LdNT2 was subcloned into the pRS424-Cu yeast-E. coli shuttle vector (17). This vector contains the TRP1 gene for selection in yeast, an ampicillin resistance gene for selection in E. coli, and a polylinker flanked by the CUP1 copper-inducible promoter and CYC1 transcriptional terminator. LdNT2 was amplified by PCR using primers that introduced a BamHI site immediately upstream of the translational start site, a His 6 tag immediately downstream of the translational start site, and an EcoRI site downstream of the stop codon. The PCR product was then ligated into the pCR2.1-TOPO® pCR2.1-TOPO vector (Invitrogen). LdNT2 was subsequently excised with BamHI/ EcoRI from the pCRII-TOPO construct and subcloned into pRS424-Cu to generate pRS424-Cu-LdNT2. The D389N mutation was introduced into the pRS424-Cu-LdNT2 construct by site-directed mutagenesis to generate pRS424-Cu-ldnt2-D389N using the QuikChange® (Stratagene, La Jolla, CA) protocol (11).
Mutagenic Library Construction for Second-site Suppressor Screen-A library of mutant ldnt2 constructs within pRS424-Cu-ldnt2-D389N was prepared by mutagenic PCR of pXG-GFPϩ2Ј-ldnt2-D389N (11), followed by in vivo reconstitution of the yeast expression vector encompassing the PCR amplicons in the YPH499 strain (18). The open reading frame of ldnt2-D389N was amplified from pXG-GFPϩ2Ј-ldnt2-D389N, a leishmanial expression construct, to provide a PCR template devoid of yeast sequences. Mutagenic PCR was conducted using the Diversify PCR random mutagenesis kit (BD Biosciences) and pXG-GFPϩ2Ј-ldnt2-D389N as template under conditions that were expected to produce ϳ2.0 base changes per open reading frame. Two independent PCRs were pooled, fractionated on agarose gel, and the PCR amplicons isolated.
To construct the mutagenic library in the yeast shuttle vector, two unique restriction sites were engineered into pRS424-Cu-ldnt2-D389N. Silent mutations were introduced at Pro 16 and Trp 17 to generate an NcoI site and Arg 470 and Ser 471 to generate a BglII site. The engineered construct was subsequently digested with NcoI and BglII, and the resulting linearized DNA, designated pRS424-Cu-ldnt2-D389N⌬1200 bp, containing only 45 bp of 5Ј and 60 bp of 3Ј LdNT2 coding sequence, was gel-purified. DNA isolation was performed with a QIAquick gel extraction kit (Qiagen), and the concentration of DNA was determined on a DU® series 640 Spectrophotometer (Beckman Instruments, Seattle, WA).
Second-site Suppressor Screen in Yeast-To screen the library of random mutants by in vivo recombination in yeast, YPH499 cells were transformed with 0.25-1.0 g of the PCR-mutagenized pXG-GFPϩ2Ј-ldnt2-D389N and 1 g RS424-Cu-ldnt2-D389N⌬1200 bp using the TRAFO method (19). Transformants were plated on SC trpϪ adeϪ ϩ 100 M inosine ϩ 100 M CuSO 4 media. An aliquot of cells was plated on SC trpϪ to allow quantitation of the transformation efficiency. The plates were incubated at 30°C for 3-5 days and colonies capable of growing in inosine were isolated. Colonies were restreaked on SC trpϪ adeϪ ϩ 100 M inosine ϩ 100 M CuSO 4 media to verify the phenotype.
Confirmation and Sequencing of Candidate Plasmids-Plasmids containing mutant ldnt2 genes were recovered from yeast colonies that grew on SC trpϪ ϩ 100 M inosine ϩ 100 M CuSO 4 plates by the smash and grab method (14). The recovered DNA was electroporated into ElectroMAX TM DH5␣-E TM cells (Invitrogen) in 0.2-cm cuvettes with a Gene Pulser TM (Bio-Rad) with electroporation parameters of 1.8 kV, 25 microfarads, and 200 ohms. The amplified plasmids in E. coli were harvested and retransformed into YPH499 cells using the lithium acetate method. Transformants were plated on SC adeϪ trpϪ plates containing 100 M inosine and 100 M CuSO 4 to confirm the contribution of the plasmid to the transport phenotype. The plasmids were sequenced in both directions at the Vollum Institute sequencing core facility at the Oregon Health and Science University. All transport data reported in yeast were obtained with retransformed strains.
Transport Capabilities of the Second-site Suppressor Clones-Nucleoside transport measurements in yeast were performed by a previously described oil-stop method with some modifications (20). YPH499 transformants were grown overnight in SC trpϪ media after which they were diluted to an absorbance at 600 nm (A 600 ) of 0.25 per ml. CuSO 4 to a concentration of 100 M was added to induce transporter expression, and cells were grown to mid-log phase (A 600 ϳ1). For each time point, 5.0 ml of culture at an A 600 ϳ1 were sedimented by centrifugation and resuspended in 100 l of yeast nitrogen base supplemented with 0.5% glucose. Transport capability was then assessed at 10 M [ 3 H]inosine (0.2 Ci/mmol) (Moravek Biochemicals, Inc., Brea, CA).
Nucleoside transport measurements in L. donovani promastigotes were performed by the oil-stop method described previously (20). Transport was measured using 10 M [ 3 H]inosine (0.2 Ci/mmol) and 10 M [ 3 H]guanosine (0.15 Ci/mmol), and kinetic data were obtained by measuring rates of inosine transport at concentrations between 0.2 and 15 M for the N175I mutant and 1 and 50 M for the N175I/D389N mutant. [ 3 H]Inosine (19.5 Ci/mmol) and [ 3 H]guanosine (15 Ci/mmol) were purchased from Moravek Biochemicals, Inc. Transport rates were calculated by linear regression analysis and kinetic parameters determined by the method of Hanes (21).
Introduction of Site-directed Mutations into ldnt2-Mutations within LdNT2 and D389N-ldnt2 were introduced by the QuikChange® method described above. Mutations were inserted within the LdNT2 open reading frame in the previously described pXG-GFPϩ2Ј-LdNT2 vector (11), confirmed by nucleotide sequencing, and the wild type and mutant constructs were then transfected into the ⌬ldnt1/⌬ldnt2 cell line using standard electroporation parameters (22).
Integral Membrane Protein Preparations-Crude membrane fractions were prepared as reported previously (11). The pattern of expression of mutant ldnt2 proteins in L. donovani was determined by immunoblotting (11).
Cell Surface Labeling-Cell surface biotinylation was performed as described (11), with the exception that the beads were ImmunoPure® immobilized streptavidin beads (Pierce). For analysis, all samples were fractionated on 10% SDS gel by electrophoresis, and biotinylated green fluorescent protein (GFP)-tagged LdNT2 protein was detected by immunoblotting using mouse anti-GFP (living colors A.V. monoclonal JL-8, BD Biosciences) as reported (11). To determine the relative amount of GFP-tagged LdNT2/ldnt2 at the cell surface, the immunoblots were scanned by using an Epson Perfection 3200 Photo Scanner (Epson America, Inc., Long Beach, CA), and the intensity of each band corresponding to GFP-tagged transporter was assessed by the UN-SCAN-IT gel TM version 4.3 (Silk Scientific, Inc., Orem, UT). The percentage of relative cell surface expression for each mutant GFP-tagged ldnt2 transporter to wild type GFP-tagged LdNT2 was determined as described (11). Equal loading in each lane was ensured using polyclonal antisera to the leishmanial myo-inositol transporter (MIT) (11,23). MIT antibody was a generous gift from Dr. Scott M. Landfear of the Oregon Health and Science University.
Fluorescence Microscopy-Leishmania cells expressing transporter tagged with GFP were prepared and affixed to slides for microscopy by a method related previously (11). Images were acquired on a Zeiss Axiovert 200 inverted microscope, and deconvolution was carried out using Axiovision 3.1 software (Carl Zeiss Optical, Chesterfield, VA).
Threading Analyses and Tertiary Topology Prediction-Initially, the LdNT2 amino acid sequence (2) was submitted to PSI-BLAST (www. ncbi.nlm.nih.gov/BLAST/) to identify other related sequences. The top 48 hits from the PSI-BLAST query of LdNT2, all of which were ENT family members, were incorporated into a text file and submitted to ModWeb (alto.compbio.ucsf.edu/modweb-cgi/main.cgi), a web-adapted version of the MODELLER software developed for protein modeling by Sali and Blundell (24) and available as freeware through the University of California, San Francisco. In addition, the LdNT2 sequence was also submitted to the web-based programs mGenThreader (bioinf.cs.ucl. ac.uk/psipred/) and 3DPSSM (www.sbg.bio.ic.ac.uk/ϳ3dpssm/). The mGenThreader, 3DPSSM, and ModWeb programs each generated a consensus secondary structure profile of the ENT family, and this profile was used by these programs to screen the available protein data bases in order to identify proteins whose three-dimensional structures have been determined and that best fit the predicted secondary structure profile. Each program then threads the consensus secondary structure profile onto the three-dimensional structure of the proteins that best fit the predicted consensus secondary structure for the ENT family.

Isolation of Suppressor Mutants-Previous data have
shown that the D389N mutation in LdNT2 abolishes both inosine and guanosine transport in parasites and imply that this residue is part of a generalized ENT translocation mechanism (11). To examine further the role of Asp 389 in LdNT2 function, a screen for second-site suppressor mutations was developed for ldnt2-D389N. This screen was implemented in the YPH499 strain of S. cerevisiae, which is incapable of both purine biosynthesis and purine nucleoside transport. In initial experiments, YPH499 cells transformed with pRS424-Cu-LdNT2 were shown to be capable of forming colonies on plates containing 100 M inosine as the sole purine source, whereas YPH499 cells transformed with pRS424-Cu-ldnt2-D389N did not grow even after 7 days of incubation (data not shown). A library of random mutants within ldnt2-D389N was generated as described under "Experimental Procedures," and a total of 5 ϫ 10 5 transformants was screened for growth on inosine. A total of 24 colonies were obtained after 5 days on inosine plates; however, plasmids from only 16 clones restored the ability of growth on inosine following retransformation in YPH499 cells.
Sequence of the Second-site Suppressor Mutations-Sequence analysis of the 16 clones revealed that two had reverted back to Asp 389 from the D389N mutation, and three were identical, each containing a single N175I second-site mutation. Of the 12 unique clones (Table I), three harbored single mutations that suppressed the D389N phenotype (S50T, Y213C, and N175I), whereas the remaining nine contained either two or three secondary mutations, as well as the starting D389N alteration. These "suppressor" mutations spanned TMs 1-8, excluding TM5 ( Fig. 1 and Table I).
Transport Capability of the Second-site Suppressor Clones-To verify the transport phenotype of each positive clone that grew on inosine, transport measurements were conducted at 10 M [ 3 H]inosine (Fig. 2). As shown in Fig. 2, all 12 clones that suppressed the D389N growth phenotype were capable of transporting inosine, and 6 (shaded gray in Fig. 2) exhibited a transport capacity significantly greater than YPH499 cells transformed with wild type LdNT2. Because of the frequency of N175I within the 12 clones and the robust transport capacity that it conferred to the transformants, it was chosen for further investigation.
Creation and Evaluation of ldnt2 Mutants in Leishmania-To ensure that the suppression of the D389N mutation could be reproduced in other expression systems, the N175I mutation was inserted into both pXG-GFP2Јϩ-ldnt2-D389N and pXG-GFP2ϩЈ-LdNT2, and the resulting constructs were transfected into nucleoside transport-deficient ⌬ldnt1/⌬ldnt2 L. donovani. These lines were designated ⌬ldnt1/⌬ldnt2[GFP-  Table II), close to the 1.3 Ϯ 0.6 M value determined previously for wild type GFP-LdNT2 (11). Most interestingly, the K m value obtained for the GFP-ldnt2-N175I/D389N second-site suppressor mutant with inosine was significantly higher, 11 Ϯ 3.6 M (Fig. 4B and Table II).
To compare the levels of LdNT2/ldnt2 protein at the plasma membrane and flagellum among the transfectants, the GFP-LdNT2/GFP-ldnt2 on the cell surface was quantitated using a membrane-impermeable biotin probe (Fig. 6). Equal loading of  1  N175I  2  N175I, I345V  3  N175I, L123Q  4  N175I, D205G, W339R  5  Y213C  6  Y213C, R393Q  7  D159N, I220T  8  P52H, I209V  9  V124G, I218T, W339R  10  N84S, I118P, F214C  11  S50T  12 I327T, G365V, M378T the cell surface biotinylation membrane fractions was verified by probing with an independent antibody to MIT. Comparable amounts of GFP-LdNT2/GFP-ldnt2 were detected at the cell surface in all transfectants except ⌬ldnt1/⌬ldnt2[GFP-N175I] in which surface membrane ldnt2 protein was ϳ50% of the LdNT2 level observed in the ⌬ldnt1/⌬ldnt2 [GFP-LdNT2] line. As expected, no cell surface biotinylation was observed in ⌬ldnt1/⌬ldnt2 [GFP] parasites. The results of the biotinylation experiments were mirrored by parallel analyses of fractionated integral membrane proteins from each of the GFP transfectants (Fig. 6). These analyses confirmed the reduced level of ldnt2 protein in the ⌬ldnt1/⌬ldnt2 [GFP-N175I] parasites. Substitutions at Asn 175 -To assess the range of secondary mutations at Asn 175 that would suppress the D389N phenotype, a series of mutations were created at Asn 175 within pRS424-Cu-ldnt2-D389N, and the mutant alleles were transformed into yeast and screened for growth on 100 M inosine. Of the substitutions introduced (Gly, Ala, Ser, Thr, Gln, Asp, Lys, Phe, Val, or Leu) at Asn 175 , only the branched chain amino acids, Val and Leu, as well as Ile, at residue 175 enabled growth on inosine plates (Table III).
Threading Analyses and Tertiary Topology Prediction-To  (37). The location of the specific residues depicted in the figure can be obtained from Table I. Asp 389 , the site at which the original null mutant D389N was constructed (11), is shaded black. The Asn 175 , Tyr 213 , and Ser 50 residues at which single mutations were obtained that were capable of suppressing the D389N phenotype are shaded gray.  Table I. Table II. spatially orient the second-site suppressor mutants, a tertiary topology model of LdNT2 was constructed. A suitable template for constructing the ENT model was ascertained by threading analyses, which exploit web-based algorithms to identify existing macromolecular crystal structures that exhibit similarity at the secondary structure level to secondary topology profiles of ENTs without the necessity of significant sequence identity (25,26). Three independent threading algorithms (mGenThreader, 3DPSSM, and ModWeb) all selected a major facilitator superfamily (MFS) member as a template upon which to construct an ENT model. Alignment of selected members of the ENT family with the identified MFS members, despite low primary sequence homology, showed good agreement in their secondary topologies (see Supplemental Material). The three-dimensional model predicted by ModWeb was considered the best because it was the most consistent with the secondary topology of the ENTs. Specifically, ModWeb identified the crystal structure of the E. coli glycerol 3-phosphate transporter, an MFS member, with a high probability (expected exponential value of Ϫ56 ). In this structure, the helix packing for LdNT2 was similar to that of MFS members, excluding, of course, TM12 of the latter (Fig. 7A). Furthermore, all the residues on various ENTs that have been identified through biochemical and genetic data to be either solvent-accessible (27,28) or important ligand selectivity determinants (4 -8) were oriented toward the putative ligand-binding pore (Fig. 7A). Finally, the majority of the second-site suppressor mutations obtained in the screen are localized on TMs that cluster in the tertiary topology model (Fig. 7B). DISCUSSION All ENT family members that have been functionally characterized to date contain a conserved DXXXR pentapeptide within TM8 (10). Site-directed mutants of Asp 389 within LdNT2 cripple transport capability, which cannot be ascribed to changes in ligand affinity or protein expression at the parasite plasma membrane and flagellum (11). Thus, Asp 389 is a key component of the LdNT2 translocation mechanism. Because structural information on ENTs is limited, a forward genetic strategy was exploited to select for second-site suppressor mutations within LdNT2 that would further our understanding of the role that Asp 389 plays in nucleoside transport. Second-site suppression strategies have proven useful in the elucidation of the functional role of cryptic first-site mutations within a variety of membrane proteins, revealing information about both tertiary structure and critical functional domains  FIG. 6. Cell surface expression of wild type and mutant ldnt2 in ⌬ldnt1/⌬ldnt2 transfectants. A, detection of GFP in integral membrane protein fractions prepared from GFP-LdNT2 and GFP-ldnt2 mutants using GFP monoclonal mouse antibody. B, live parasites were subjected to cell surface biotinylation, lysis, immunoprecipitation, and Western analysis using GFP antibody as described under "Experimental Procedures." C, detection of MIT using polyclonal rabbit antibody.  29 -31). In this study, second-site mutations that suppress the D389N null-transport phenotype were selected in S. cerevisiae purine auxotrophs by virtue of their ability to confer growth on inosine. A diverse number of mutations that occurred throughout the first 8 TMs (except TM5) and their interconnecting loops were identified (Fig. 1). Of the 16 positive clones, 2 were revertants to the wild type allele. No other substitution at position 389 was obtained in our screen, substantiating our previous observation that Asp 389 is a crucial signature residue in the ENT permeation mechanism. Because Asp 389 is a negatively charged residue within a TM helix, we conjectured that it might participate in an electrostatic interaction with a positively charged counterpart within the LdNT2 structure. The only alteration of a positive charge among the suppressors was R393Q that appeared in context with a Y213C mutation, a mutation that suppressed the D389N phenotype by itself (see Table I and Fig. 2). Subsequent analysis of a site-directed R393Q/D389N double mutant indicated that R393Q was not able to rescue the D389N null phenotype (data not shown). These data indicate that the critical role of Asp 389 in the translocation mechanism is not mediated through electrostatic interactions, although it is conceivable that a charge partner to Asp 389 is an irreplaceable residue and thus would not be selected in this screen. It is also conceivable that the spectrum of mutations that could be theoretically generated by the random PCR-based mutagenesis strategy may be insufficiently expansive to identify a charge alteration.
The most prevalent second-site mutation, N175I, occurred in 4 of the 12 individual suppressor clones, and possibly in 6 of 14 clones, if the three identical N175I/D389N suppressor mutants were independent. Asn 175 is predicted to be located at the COOH terminus of the loop between TMs 4 and 5 (Fig. 1). Unlike Asp 389 , Asn 175 is not a conserved residue among ENT family members (Fig. 8), although it is located within an evolutionarily conserved region of ENTs (32). A detailed kinetic characterization of N175I suggested that the primary defect in D389N is likely a minor perturbation in protein structure, because N175I by itself is not an inhibitory mutation and does not alter ligand affinity suggesting that Asp 389 and Asn 175 do not interact within LdNT2 ( Fig. 4 and Table II). In contrast, N175I in the context of D389N does alter ligand affinity but not protein expression or targeting (Figs. 4 -6 and Table II). Rather N175I in the context of D389N is a "bypass" suppressor, which does not completely reverse the incapacitating constraints imposed by the D389N mutation. This presumed modification in protein structure of N175I/D389N did not alter ligand selectivity, because no uptake of adenosine and uridine was detectable (data not shown).
The only amino acid substitutions at Asn 175 able to suppress the D389N phenotype were the branched chain amino acids, Ile, Val, and Leu (Table III). This preference for large nonpolar aliphatic residues as suppressor mutations at position 175 implies a local volume compensation that might shift the position of TM5 relative to TM8. The predilection for Ile mutations, rather than genetic alterations to Val and Leu, at position 175 in the suppressor screen (Table I), would be expected from the higher number of nucleotide changes required for the Val and Leu missense mutations to occur. Most interestingly, the Ile, Val, and Leu substitutions at Asn 175 are all predicted to shift the local secondary structure probability from loop to helix according to the Protein Sequence Analysis server (bmercwww.bu.edu/psa/), suggesting that perhaps these mutations cause an extension of one helical turn in TM5. No local secondary structure shifts were predicted for the D389N mutation by using this algorithm.
The diversity of second-site mutations obtained in the screen implied that the D389N mutation likely confers a constraint that is alleviated by other mutations within the ldnt2 structure. The second-site mutations grouped to three regions of LdNT2: TMs 1-3, TM6, and the COOH cusp of the large intracellular loop between TMs 6 and 7 (Fig. 1). Mutations within the second cluster were found in almost 50% of the suppressor mutants. The third clustering in the highly divergent intracellular loop between TMs 6 and 7 suggests that this loop may also contribute to the conformational changes that accompany translocation (Fig. 1). A clustering of second-site suppressor mutations was also observed with the lactose permease, which portended potential interactions between residues that were later confirmed when the lactose permease crystal structure was solved (31,33). The lack of a crystal structure for any ENT family member is obviously an impediment to interpreting the significance of the clustering of the second-site suppressor mutants within the first 8 TMs of the D389N ldnt2 null mutant.
In the absence of structural information on ENTs, a tertiary structure prediction for all ENTs was computationally gener- FIG. 7. Tertiary topology prediction. A, the schematic of the predicted helical arrangement is shown with the residues that have been found to be critical in ligand binding (4 -8) highlighted in magenta. These include Met 33 and Leu 92 within hENT1 (6,8), Gly 179 and Gly 183 within hENT1 (5) and LdNT1 (4), respectively, Asp 389 and Arg 393 within TM8 of LdNT2 (10), Cys 337 in LdNT1 (4), and an exofacial Cys 140 in rENT1 (28). The magenta rectangle on TM5 denotes the face of TM5 in LdNT1 that was found to be solvent-accessible (27). The roles of these residues in transport function and ligand recognition are described under "Discussion." B, the schematic depicts the clustering of the helices accommodating the majority of the second-site suppressors and is shown in orange. TM8 is shown is teal. ated by threading analysis in order to provide a structural framework in which to inspect the D389N and suppressor mutations. Threading analyses utilize algorithms that generate a consensus secondary structure profile for a protein family, which is then used to search available protein structure data bases to identify suitable templates upon which a three-dimensional model can be constructed. Such threading analyses have become more reliable recently because of the accuracy of the more sophisticated secondary structure algorithms (25). Although the tertiary topology prediction was constructed by the ModWeb algorithm, which elected to build a tertiary model for the ENT family upon the template of the structurally resolved glycerol 3-phosphate transporter, it is notable that all three independent threading analyses identified MFS members as suitable templates. Most interestingly, Baldwin et al. (9) have previously hypothesized that the overall three-dimensional structures of ENTs would be similar to those of the human glucose transporters, which are members of the MFS. Crystal structures of MFS members encompass 12 TMs with cytosolic NH 2 and COOH termini and a large intracellular loop between TMs 6 -7. ENTs are similar to MFS members in their predicted membrane topology accommodating a large loop between TMs 6 and 7, although they are conjectured to have 11 TMs with an extracellular COOH terminus (34). Although the exclusion of TM12 in the ENT prediction from the MFS template is a concern, an extensive mutagenesis study of TM12 of the lactose permease, an MFS member, indicates that residues in this domain do not participate in ligand binding or translocation (35). The validity of the predicted helix packing in the ENT model (Fig. 7A) is supported by the location of key residues within TMs 1, 2, 4, and 5, all of which map toward the putative aqueous pore in the model, that have been identified by structure-function studies to be critical for ligand selectivity or to be solvent-accessible (4 -6, 8, 27, 28, 36). Specifically these include the following: Met 33 and Leu 92 within hENT1, which have been implicated in the interaction with the vasodilators, dipyridamole and dilazep (Met 33 and Ref. 8), as well as with the ligands, inosine, guanosine, S-nitrobenzyl-4-thioinosine and dilazep (Leu 92 and Ref. 6); Gly 179 and Gly 183 within hENT1 (5) and LdNT1 (4), respectively, that are important determinants for uridine transport as well as for dilazep interaction in the case of Gly 179 ; Asp 389 and Arg 393 within TM8 of LdNT2, which constitute a conserved signature motif (DXXXR) within the majority of ENTs (10) and are important determinants for transporter function and targeting (11); Cys 337 in LdNT1,which when mutated to a bulkier residue results in a transporter with impaired transport capability (4); residues mapped to the solvent-accessible face of TM5 in LdNT1 (27); and an exofacial Cys 140 in rENT1 (28). Because of the low sequence identity (Ͻ14%) between the MFS and ENTs and the inherent difficulties in modeling unstructured loop domains, this ENT structural prediction will, of course, require extensive computational refinement and experimental validation to evolve. When the second-site mutations that suppress the D389N null phenotype are mapped onto this model, they cluster across TMs 1, 3, and 6, and all map to one region of the protein, suggesting that alterations in the tertiary structure in this part of the protein are able to suppress the deleterious effects of the D389N mutation (Fig. 7B). Thus, we have developed a second-site suppressor screen in yeast for LdNT2, an ENT family member. This strategy, which has proven instru-mental in probing the transport mechanism of the lactose permease, provides a useful approach for future structure-function studies of members of this ENT family. The second-site suppressor data confirm our previous hypothesis that Asp 389 induces or stabilizes a conformation of the transporter that is vital to the LdNT2 permeation mechanism (11). Furthermore, we have generated a preliminary tertiary prediction for the ENT family that can serve as a framework in which to understand the structural role of specific residues and domains in nucleoside translocation and/or recognition.