Residues 334 and 338 in Transmembrane Segment 8 of Human Equilibrative Nucleoside Transporter 1 Are Important Determinants of Inhibitor Sensitivity, Protein Folding, and Catalytic Turnover*

Equilibrative nucleoside transporters (ENTs) are important for the metabolic salvage of nucleosides and the cellular uptake of antineoplastic and antiviral nucleoside analogs. Human equilibrative nucleoside transporter 1 (hENT1) is inhibited by nanomolar concentrations of structurally diverse compounds, including dipyridamole, dilazep, nitrobenzylmercaptopurine ribonucleoside (NBMPR), draflazine, and soluflazine. Random mutagenesis and screening by functional complementation for inhibitor-resistant mutants in yeast revealed mutations at Phe-334 and Asn-338. Both residues are predicted to lie in transmembrane segment 8 (TM 8), which contains residues that are highly conserved in the ENT family. F334Y displayed increased Vmax values that were attributed to increased rates of catalytic turnover, and N338Q and N338C displayed altered membrane distributions that appeared to be because of protein folding defects. Mutations of Phe-334 or Asn-338 impaired interactions with dilazep and dipyridamole, whereas mutations of Asn-338 impaired interactions with draflazine and soluflazine. A helical wheel projection of TM 8 predicted that Phe-334 and Asn-338 lie in close proximity to other highly conserved and/or hydrophilic residues, suggesting that they form part of a structurally important region that influences interactions with inhibitors, protein folding, and rates of conformational change during the transport cycle.

Nucleosides, which are hydrophilic molecules that do not readily diffuse across biological membranes, are the metabolic precursors of fundamental biological molecules such as DNA, RNA, and ATP. Nucleoside analogs are antimetabolite agents used in the treatment of various cancers and viral diseases (1). Members of the equilibrative nucleoside transporter (ENT) 6 family mediate uptake and release of nucleosides across cellular plasma or organellar membranes via either facilitated diffusion or proton-coupled mechanisms (2). Four human ENTs (hENT1-4) have been identified in humans and are thought to share a common 11-transmembrane segment (TM) topology (3). The two widely distributed, major plasma membrane transporters, hENT1 and hENT2, are designated equilibrative sensitive or insensitive (es or ei), respectively, based on their differential sensitivities to inhibition by nitrobenzylmercaptopurine ribonucleoside (NBMPR) (4 -6). The presence of hENT1 protein in pancreas adenocarcinoma has recently been implicated in the therapeutic response of patients treated with the nucleoside drug, gemcitabine; patients whose cancers have high abundance hENT1 have longer survival times than patients whose cancers have low abundance hENT1 (7). hENT3 mediates nucleoside transport across lysosomal membranes, whereas the more distantly related hENT4, also known as the plasma membrane monoamine transporter, transports monoamine neurotransmitters and adenosine (at acidic pH) in brain and heart tissue (8 -11).
Adenosine is a signaling molecule that, when released from cells as a result of ischemia or hypoxia-induced ATP depletion, binds to cell-surface adenosine receptors to induce a wide range of protective mechanisms, including vasodilation and inhibition of platelet aggregation, excitatory neurotransmitter release, and inflammatory responses (12). The primary mecha-nism by which the protective effects of adenosine are terminated is by its reuptake through ENTs that are present in high abundance on vascular endothelial cells (13). A group of highly potent inhibitors of nucleoside transport, draflazine, soluflazine, dilazep, dipyridamole, and NBMPR, block ENT-mediated adenosine uptake and thus potentiate protective effects of adenosine (14). These inhibitors, with the exceptions of draflazine and soluflazine, do not share any structural similarities with each other (Fig. 1A) and inhibit hENT1-and hENT2-mediated nucleoside transport, respectively, at nanomolar and micromolar concentrations.
The mechanisms of inhibition of dipyridamole, dilazep, NBMPR, and draflazine have been addressed in several studies. Dipyridamole inhibits mammalian es transporters competitively when present at nanomolar concentrations, whereas dilazep, which has two ionizable groups with pK a values of ϳ5 and 8, inhibits either allosterically or competitively depending on the pH (12,(15)(16)(17)(18)(19)(20)(21). More recently, residue 33, predicted to lie at the extracellular end of TM 1 of hENT1 and hENT2, was shown to be a functionally important component of the binding sites for dilazep, dipyridamole, and nucleosides for both transporters (22,23). Ile-429 in TM 11 of Caenorhabditis elegans ENT1 (CeENT1) plays a major role in dipyridamole interactions, whereas its counterpart in hENT1 (Leu-442) influences dipyridamole sensitivity when Met-33 is first mutated to Ile (21). Depending on experimental conditions, NBMPR binding has been shown to be either competitive or allosterically linked to the binding site for nucleoside permeants (15,16,20), and recent molecular evidence suggests that residues in TM 2 (Met-89 and Leu-92), TM 4 (Gly-154 and Ser-160), and TM 5 (Gly-179) of hENT1 are involved in both permeant translocation and NBMPR binding (Fig. 1B) (24 -27). Molecular determinants of binding of draflazine, which is a mixed type inhibitor of both nucleoside transport and NBMPR binding, and soluflazine to hENT1 have not yet been identified (12,28,29).
The aims of the work described in this study were to identify additional amino acid residues of hENT1 involved in binding of dilazep, dipyridamole, and/or NBMPR as well as residues involved in binding of draflazine and/or soluflazine. The random mutagenesis and functional complementation screening methods that were previously employed in the identification of Met-33 and Leu-442 of hENT1 (21,23) identified Phe-334 and Asn-338 of hENT1 as previously unrecognized residues involved in inhibitor binding. Mutants, largely based on their homologies with other ENT family members, were generated at each residue and tested for adenosine transport activities and inhibitor sensitivities by functional expression in yeast. The cell-surface abundance of each mutant was quantitatively determined using FTH-SAENTA, a novel membrane-impermeable NBMPR analog. Modeling of the TM 8 region as an ␣-helix was performed to assess the possibility that the identified residues line the permeant translocation pathway and/or inhibitor binding site(s). The results of this study suggest that residues in question are critical molecular determinants of coronary vasodilator binding, rates of catalytic turnover, and protein folding.
Plasmid Construction and Random Mutagenesis-The hENT1 reading frame was subcloned into pYPGE15 to generate pYPhENT1 as described previously (23). The rENT1 open reading frame was amplified by PCR using the primers (restriction sites underlined) XbaIrENT1 (5Ј-CCC TCT AGA ATG ACA ACC AGT CAC CAG-3Ј) and KpnIrENT1 (5Ј-CCC GGT ACC TCA CAC AAG TGC CCT TAA-3Ј) and inserted into pYPGE15 to generate pYPrENT1. Random mutagenesis of pYPhENT1 was performed by propagating the plasmid in the XL-1 RED mutator strain of E. coli (Stratagene, La Jolla, CA) for 30 -45 generations to obtain ϳ1 mutation per cDNA, according to the procedure of Greener et al. (34). Point mutations resulting in various amino acid changes of the identified residues were generated using the QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA).
Phenotypic Complementation and Screening of Mutants-The complementation assay was based on the ability of recombinant hENT1 produced in yeast to salvage exogenously supplied thymidine under conditions of dTMP starvation (35). In brief, KTK cells transformed with pYPhENT1 using a lithium acetate procedure (33) were plated directly onto CMM/GLU plates containing methotrexate (MTX) at 50 g/ml and sulfanilamide (SAA) at 6 mg/ml (CMM/GLU/MTX/SAA) at a pH of 7.0. Colonies formed with an efficiency of ϳ10 5 transformants/g DNA after incubation at 30°C for 3.5 days in the presence of 10 M thymidine, and complementation was prevented when either 100 M dipyridamole or 1 M draflazine was also present. Mutated pYPhENT1 (20 g) was transformed into KTK cells, which were then plated onto CMM/GLU/MTX/SAA with 10 M thymidine and either 100 M dipyridamole or 1 M draflazine and incubated at 30°C for 3.5 days. Colonies with apparent resistance to dipyridamole inhibition of complementation were isolated, grown in 1 ml of liquid CMM/GLU for 2 days, and restreaked onto CMM/GLU/MTX/SAA plates with 10 M thymidine and either 100 M dipyridamole, 10 M dilazep, or 1 M draflazine. The mutant hENT1 cDNAs were amplified from the yeast colonies by PCR, subcloned back into nonmutated pYPGE15, and sequenced.
Adenosine Transport in S. cerevisiae-Yeast cells containing pYPhENT1, pYPrENT1, or one of the various mutant trans-porters were grown in CMM/GLU media to an A 600 ϭ 0.5-1.2, washed twice in fresh medium, and resuspended to A 600 ϭ 4.0. Transport assays were performed at room temperature and at pH 7.4. Unlabeled adenosine, dipyridamole, dilazep, and NBMPR were obtained from Sigma, and [ 3 H]adenosine was obtained from Moravek Biochemicals (Brea, CA). Draflazine and soluflazine were kind gifts from the Janssen Research Foundation, Beerse, Belgium. The transport assays were performed in 96-well plates as described previously (36,37). Briefly, 50-l portions of yeast suspensions were added to 50-l portions of 2ϫ concentrated [ 3 H]adenosine in 96-well microtiter plates. Because initial rates of adenosine transport into yeast cells producing recombinant ENTs persist for intervals of 60 min (22), single time points of 10 min were used to estimate initial rates of transport. Yeast cells were collected on filter mats using a Micro96 Cell Harvester (Skatron Instruments, Lier, Norway) and rapidly washed with deionized water. The individual filter circles corresponding to wells of the microtiter plates were removed from filter mats with forceps and transferred to vials for quantification of radioactivity by scintillation counting. For inhibitor concentration-effect relationships, the yeast suspensions were first incubated for 15-30 min with the inhibitor to allow equilibration of the inhibitor with its binding sites before the addition of radiolabeled permeant as described previously (23).
The sulfhydryl modification experiments were performed essentially as described previously (38). Yeast cells producing hENT1, F334C, or N338C were incubated with 25 mM methanethiosulfonate ethylammonium (MTSEA), a specific and highly reactive membrane-permeable probe for free sulfhydryl groups. The cells were rapidly washed three times with fresh media, resuspended to A 600 ϭ 4.0, and subjected to adenosine transport assays.
Characterization of S. cerevisiae Membranes Containing Mutant hENT1 Proteins-Membranes were prepared from yeast cells harboring pYPhENT1 or mutants derived from this plasmid as described previously (35) in the presence of protease inhibitors (4-(2-aminoethyl)benzenesulfonyl fluoride, pepstatin A, E-64, and 1,10-phenanthroline) and protease inhibitor mixture for use with fungal and yeast extracts (Sigma). Their total protein content was measured using the bicinchoninic acid assay (Pierce). To quantify the abundance of hENT1 proteins by immunoblotting, samples were resolved by SDS, 12% (w/v) PAGE, electroblotted onto nitrocellulose membranes, blocked by incubation with 5% (w/v) skimmed milk powder, and then incubated with primary antibodies at 4°C overnight (1 g/ml affinity-purified rabbit antibodies raised against a synthetic peptide corresponding to residues 254 -272 of hENT1). Following incubation with a horseradish peroxidase conjugate of goat anti-rabbit IgG (1/50,000; Jackson ImmunoResearch, West Grove, PA), antigens were visualized using Supersignal West Pico chemiluminescent substrate (Pierce). Staining intensities were quantified using a Bio-Rad Fluor-S gel documentation system and multianalyst software (Bio-Rad).
Intact yeast cells and total membrane fractions prepared as described above were assayed for their ability to bind [ 3 H]NBMPR (Moravek, Brea, CA) using a filtration assay described previously (35). In brief, two to four samples of cells (10 6 ) or membranes (25 g of protein) were incubated for 45 min with the desired concentrations of [ 3 H]NBMPR (0.01-25 nM) alone or in the presence of either 0.5 M unlabeled NBMPR (to quantify total specifically bound NBMPR) or FTH-SAENTA, a tight binding impermeant inhibitor of hENT1 that is structurally related to NBMPR (to quantify cellsurface specifically bound NBMPR). All experiments were conducted at 20°C in binding buffer (100 mM KCl, 10 mM Tris-HCl, 0.1 mM MgCl 2 , 1 mM CaCl 2 , pH 7.4). The cells or membranes were then collected on Whatman GF/B filters (Whatman) under vacuum and washed three times with cold binding buffer before quantification of radioactivity by liquid scintillation counting.

Random Mutagenesis and Screening-Growth of yeast cells
in the presence of methotrexate and sulfanilamide results in the depletion of TTP pools and growth arrest (40). Although yeast cells lack endogenous thymidine transport systems (41), production of recombinant hENT1 allows for salvage from the external medium of thymidine, which is then phosphorylated to TMP by recombinant herpes simplex thymidine kinase in KTK yeast cells (35). When an inhibitor of hENT1 is added, thymidine salvage is blocked, providing a functional complementation assay for hENT1 mutants with reduced affinities for inhibitors. The yeast expression plasmid pYPhENT1 was randomly mutated by propagation in the E. coli mutator strain XL-1 RED and transformed into KTK yeast under complementation conditions in the presence of inhibitory concentrations of dipyridamole or draflazine. Soluflazine was not tested because high concentrations were required to inhibit complementation, and there were insufficient quantities available for screening. All of the clones capable of growth in the presence of inhibitory concentrations of dipyridamole contained mutations at Met-33, with the exception of two, which contained mutations resulting in the conversion of Phe at position 334 to Ser (F334S). All of the clones capable of growth in the presence of inhibitory concentrations of draflazine were identical and contained mutations that converted Asn at residue 338 to Ser (N338S). Both Phe-334 and Asn-338 are predicted to reside in TM 8 of hENT1 (Fig. 1B) (3).
Multiple Sequence Alignments of TM 8-To assess the level of conservation of the identified residues, multiple sequence alignments of the TM 8 region from 43 different ENT family members from a variety of different organisms was performed (data not shown). A portion of the multiple sequence alignments containing a few select sequences is presented in Fig. 1C. Phe-334 was conserved in 42% of the sequences analyzed and in 88% of the mammalian and nematode sequences. Many other ENTs from plants and protozoa contain bulky aliphatic side chains at positions corresponding to Phe-334, suggesting that hydrophobic residues are preferred. Asn-338 was conserved in 93% of the sequences analyzed, suggesting that this residue plays an important role in ENT structure and/or function.
Effects of Mutations at Phe-334 and Asn-338 in hENT1 on the Concentration Dependence of Adenosine Transport-To determine the functional role of Phe-334 and Asn-338, a series of mutations at each position was generated based, in part, on the identities of corresponding residues in other ENT family members. Adenosine was selected as the test permeant because it is transported by more ENT family members than any other nucleoside and, in mammals, is an important physiologically active molecule whose activity is affected by transport inhibitors (42). Furthermore, yeast cells do not possess endogenous adenosine transport systems, do not metabolize extracellular adenosine (43), and, when producing recombinant ENTs, display robust initial rates of adenosine uptake for time intervals of Ն60 min (22). Yeast cells producing recombinant "wild type" or mutated hENT1 were incubated in the presence of increasing concentrations of [ 3 H]adenosine for 10 min in the absence (total uptake) or presence (background) of 10 mM uridine. The apparent K m and V max values and the V max :K m ratios (transport efficiencies) are presented in Table 1. hENT1 displayed apparent K m and V max values of 18.6 Ϯ 2.3 M and 1280 Ϯ 30 pmol/ mg/min, respectively, indicating that recombinant hENT1 displayed a similar affinity for adenosine when produced in yeast as has been observed in transfected mammalian cells (44).
The apparent K m values of the Phe-334 mutants of hENT1 were similar to those of wild type hENT1, whereas most of the V max values of the Phe-334 mutants were significantly reduced, with the greatest reductions observed for F334C, F334V, and F334I (Table 1). In contrast, substitution of Phe with Tyr resulted in a 4.8-fold higher apparent V max value ( Fig. 2A). When the corresponding residue in rat ENT1 (rENT1) (Phe-335) was also mutated to Tyr, there was a 3.4-fold increase in the apparent V max value, suggesting that the hydroxyl group of Tyr enhanced the transport capacity of these ENT1 mutants ( Fig. 2 and Table 1). Furthermore, the introduction of a hydrophilic residue in place of an aromatic residue (i.e. F334S) was surprisingly well tolerated, indicating that the presence of a bulky aromatic residue at this position was not critical, despite the pattern of residues found at the corre-  (24 -27, 47, 49). C, protein sequences of 11 ENT family members were obtained from GenBank TM using the accession numbers AAC51103 (human; hENT1), AAF78452 (mouse; mENT1), AAB88049 (rat; rENT1), AAC39526 (hENT2), AAB88050 (rENT2), AF326987 (hENT3), AF326986 (mENT3), NP_694979 (hENT4), AAH25599 (mENT4), AAM46663 (Caenorhabditis elegans; CeENT1), CAB01882 (CeENT2) (Leishmania donovani; LdNT1.1), and AAF74264 (LdNT2) and subjected to multiple sequence alignments with DNAMAN version 4.03 software using the BLOSUM 62 substitution matrix. Residues that were identical in Ն50% of sequences are shaded.
sponding positions in other members of the ENT family (Fig. 1C).
For the Asn-338 mutants, significantly increased K m values were observed for the Ser and Ala substitutions, and most also displayed significantly decreased V max values, with the Asp and Gln substitutions having the most notable effects (Table 1). Both the bulky hydrophobic side chain of Met and the hydrophilic side chain of Asp were tolerated at position 338, suggesting that the extremely high level of conservation of Asn-338 was not predictive of its importance in transporter function.
Quantification of Functional Cell-surface Transporters-The results of Fig. 2 and Table 1 suggested that the Phe-334 and Asn-338 mutations altered the catalytic turnover and/or cellsurface abundance of functional hENT1 molecules. We therefore performed equilibrium binding experiments with [ 3 H]NBMPR at a saturating concentration of 25 nM using intact yeast cells producing either hENT1 or one of the various mutants to quantify the total number of "functional" hENT1 molecules. To distinguish total specifically bound [ 3 H]NBMPR from that which was localized to the cell surface, values obtained in the presence of either 0.5 M unlabeled NBMPR or a membrane-impermeant compound, FTH-SAENTA, which is structurally similar to NBMPR and SAENTA-x2-fluorescein (45), were subtracted from values obtained in their absence. hENT1 displays a similar affinity for FTH-SAENTA as for NBMPR, but FTH-SAENTA, which has fluorescein attached via a linker arm at the 5Ј-position of the ribosyl moiety, is impermeant and thus displaces only cell-surface site-bound [ 3 H]NBMPR from intact cells. 7 The fraction of the [ 3 H]NBMPR-binding sites that was sensitive to FTH-SAENTA was 50 -100% of that which was sensitive to unlabeled NBMPR, suggesting that hENT1 molecules were present in both cell surface and intracellular membranes in most samples (Fig. 3). The exceptions, N338C and N338Q, displayed cell-surface abundance that was similar to that of the total binding, suggesting that intracellular transporters, if present, were nonfunctional and/or misfolded. hENT1-F334C, F334V, and N338Q displayed significantly lower cell-surface abundance than that of wild type, whereas the other mutants displayed levels of cell-surface abundance that were not significantly different from wild type ( Table 1).
The apparent V max values and cell-surface NBMPR abundance values were used to calculate maximal turnover rates for hENT1 and each of the mutants (Table 1). hENT1 and most of the mutant transporters displayed turnover rates of 5-8 adenosine molecules/s. In contrast, F33I and N338D displayed turnover rates of 2 and 3 molecules/s, respectively, suggesting that the transport function of these mutants was modestly impaired. In contrast, F334Y displayed an enhanced rate of 16 adenosine molecules/s that was 3.2-fold faster than that of wild type hENT1. Taken together, these results indicated that, with the exception of F334Y, the observed differences in apparent V max for the Phe-334 and Asn-338 mutants could be attributed to alterations in cell-surface abundance of functional transporters.
Effects of Mutations in hENT1 on Sensitivity to Transport Inhibitors-Concentration-effect relationships for inhibition of adenosine transport were determined with dipyridamole, dilazep, NBMPR, draflazine, and soluflazine to determine the contributions of the identified residues to interactions of hENT1 with the inhibitors. The resulting IC 50 values are presented in Fig. 4.
The Phe-334 mutants displayed reduced sensitivities to dipyridamole and dilazep, with large volume side chains being preferred. The most notable effects were observed for F334S, which displayed IC 50 values that were 77-and 26-fold higher than wild type for dipyridamole and dilazep, respectively (Fig.  4A). F334Y was well tolerated for dipyridamole interactions and favored for dilazep interactions, displaying an IC 50 value that was 25% of the wild type value. Phe-334 mutations had

Kinetic properties of adenosine transport by hENT1 and its Phe-334 and Asn-338 mutants
Yeast cells producing hENT1, its various Phe-334 and Asn-338 mutants, rENT1, or its Phe-335 mutant were incubated for 10 min in the presence of increasing concentrations of ͓ 3 H͔adenosine. Average K m and V max values (mean Ϯ S.E., n ϭ 3) were determined by nonlinear regression analysis using GraphPad Prism version 4.0 software. The cell-surface binding was determined as the site-bound ͓ 3 H͔NBMPR that was sensitive to displacement by FTH-SAENTA. The turnover rates of hENT1 or one of its mutants were determined by dividing the molecules/s of adenosine transported at the V max by the corresponding number of cell surface-bound NBMPR molecules. The p values were obtained using two-tailed, unpaired t tests of the values obtained for the various mutant transporters against values obtained for the corresponding wild type transporters. little or no effect on the IC 50 values for NBMPR, draflazine, or soluflazine (data not shown). Asn-338 mutations resulted in increased IC 50 values for dipyridamole and dilazep with N338A, N338Q, and N338D displaying values that were Ն10-fold higher than the wild type values (Fig. 4B). N338C displayed an IC 50 value for dipyridamole that was similar to the wild type value. The Asn-338 mutants displayed IC 50 values for NBMPR ranging from 2.8-to 10-fold higher than the wild type value, suggesting that mutating this residue affected a relatively weak interaction point between hENT1 and NBMPR. In contrast, several Asn-338 mutations had dramatic effects on sensitivities to draflazine and soluflazine. N338M yielded IC 50 values that were 180-and 93-fold higher than wild type values for draflazine and soluflazine, respectively. The small and/or hydrophilic side chains of Cys, Asp, and Ser were well tolerated at position 338, and the wild type Asn residue yielded the lowest IC 50 values.

Protein
The IC 50 values observed for dipyridamole inhibition of adenosine transport were consistently ϳ50% of the values obtained previously for inhibition of uridine transport, even though dipyridamole is a competitive inhibitor of hENT1 transport of both adenosine and uridine (21). The reasons for this small discrepancy are unclear, but the relative differences in dipyridamole sensitivity for each of the mutants were highly consistent with the data obtained using uridine as the test permeant.
The F334Y and N338C mutations of hENT1 displayed similar or reduced IC 50 values for dilazep and dipyridamole, respectively. rENT1 shares 78% identity with hENT1 and displays IC 50 values for dilazep and dipyridamole inhibition that are several orders of magnitude higher than those of hENT1 (21,46). The corresponding mutations in rENT1 (F335Y and N339C) were generated and tested for inhibitor sensitivities. The F335Y mutation reduced the IC 50 value of rENT1 for dilazep to 11% of the wild type value (Fig. 4A, right panel). As a control, Ile-33 of rENT1, which corresponds to the previously identified Met-33 of hENT1 (23), was mutated to generate rENT1-I33M, which displayed an IC 50 value that was 1.4% of the wild type value and similar to that of hENT1. The N339C mutation reduced the IC 50 value of rENT1 for dipyridamole to 5.3% of the wild type value, and the I33M mutation yielded a similar decrease to 8.4% of the wild type value (Fig. 4B, bottom-right panel). These results suggested that these TM 8 residues are important molecular determinants of dipyridamole and dilazep interactions with hENT1 and rENT1.
Effects of N338C and N338Q Mutations on Cellular Distribution of Functional hENT1-The relatively minor effects on adenosine transport activity resulting from mutation of Asn-338 suggested that this highly conserved residue was not important for permeant recognition and translocation mechanisms of hENT1. However, N338C and N338Q displayed an altered distribution of functional transporters, which appeared to be absent from intracellular membranes, and in  Table 1. Each point is presented as the mean Ϯ S.E. (n ϭ 4), and where the size of the point is larger than the S.E., the latter is not shown. the case of N338Q, reduced cell-surface abundance (see Fig.  3). These observations suggested that N338C and N338Q may exhibit altered synthesis and/or trafficking. Further-more, the results of Fig. 4 suggested that these mutants also displayed reduced sensitivity to NBMPR. To further investigate these possibilities, immunoblotting and concentration dependence of [ 3 H]NBMPR binding was performed with wild type hENT1, hENT-N338C, and hENT1-N338Q (Fig. 5). Immunoblotting showed that the recombinant proteins migrated on SDS-polyacrylamide gels as both monomers and dimers, together with a small amount of putative degradation product, as reported previously (Fig. 5A) (35). The proportions of these species were not different for the wild type and mutant proteins, and quantification of the bands revealed that the hENT1-N338C and hENT1-N338Q mutant proteins were present at 75 and 67% of wild type levels, respectively (Fig. 5A). The K d values for NBMPR binding to the mutant and wild type proteins were similar, indicating that Asn-338 did not play an important role in NBMPR binding ( Fig. 5B and Table 2). In contrast, the B max values were substantially lower for the mutants, consistent with what was shown in Fig. 3. These results indicated that although the amount of total protein was only partially reduced by the N338C and N338Q mutations, the amount of functional protein was dramatically reduced. Furthermore, the functional protein was exclusively localized to the cell surface, suggesting impaired protein folding and/or altered trafficking of the N338C and N338Q mutants.
As expected, dilazep and dipyridamole inhibited NBMPR binding to the wild type protein ( Fig. 6; Table 2). The apparent K i values were similar to those reported previously (35). Mirroring the results obtained in transport experiments (Fig. 4), mutation of Asn-338 to Cys or Gln increased the apparent K i values for dilazep 4.6-and 10-fold, respectively. The apparent K i values for dipyridamole were not significantly different for N338C compared with wild type but 7.4-fold higher for N338Q ( Fig. 6; Table 2).
Sensitivity of hENT1, F334C, and N338C to MTSEA-Yeast cells producing hENT1, F334C, or N338C were incubated in the presence or absence of the sulfhydryl-specific membranepermeable reagent MTSEA, washed, and subjected to adenosine transport assays (Fig. 7). hENT1 was insensitive to MTSEA, but F334C and N338C were partially inhibited, suggesting that these residues were accessible to MTSEA and that the reaction impaired adenosine transport. MTSEA reactions performed in the presence of excess nucleosides or inhibitors did not protect F334C or N338C from modification (data not shown). These  . Comparison of site-specific binding of NBMPR to hENT1 and two of the Asn-338 mutants. A, immunoblots of membranes prepared from yeast cells producing hENT1, hENT1-N338C, or hENT1-N338Q, or harboring pYPGE15 (no insert), were stained with rabbit polyclonal antibodies against residues 254 -272 of hENT1. The mobilities of molecular mass markers are shown at the left. Numbers in parentheses indicate the relative amounts of hENT1 protein present in each lane, estimated by scanning densitometry. B, equilibrium, site-specific binding of [ 3 H]NBMPR to membranes from yeast producing wild type hENT1 (F), hENT1-N338C (‚), or hENT1-N338Q (OE), or harboring pYPGE15 (control, E) was measured as described under "Materials and Methods." Each point is the mean of duplicate determinations that differed from the mean by less than 10%. K d and B max values extracted from these data are given in Table 2.
data suggested that Phe-334 and Asn-338 either do not interact directly with nucleosides or inhibitors, do so in a manner that does not occlude them, or the reaction occurs too quickly for substrate protection to be observed.

DISCUSSION
Screening by random mutagenesis for residues of hENT1 involved in inhibitor interactions led to the identification of two previously unrecognized residues, Phe-334 and Asn-338. In this study, these residues were subjected to detailed analyses. Multiple sequence alignments of the TM 8 regions of ENT family members revealed that Asn-338 was highly conserved, and the bulky hydrophobic character of Phe-334 was also conserved. Furthermore, the identified residues were within regions previously identified as being highly conserved, and therefore likely to be structurally and/or functionally important in ENT proteins (2,9).
The experiments of Table 1 and Figs. 2 and 3 indicated that although most Phe-334 and Asn-338 mutations reduced the apparent V max values for adenosine transport by hENT1, these effects could in large part be attributed to reduced cell-surface abundance. The apparent rates of catalytic turnover for hENT1 and most of its mutants ranged from 5 to 8 molecules/s. In contrast, the increased V max values observed for F334Y appeared to be because of an increased rate of catalytic turnover ( Fig. 2; Table 1). Early studies on in situ hENT1 in human erythrocytes indicated that the rate of conformational change is significantly faster for the permeant-bound carrier than for the empty carrier (14); the implication for our assay is that the ratelimiting step in the transport cycle was the change from the nucleoside-free inward-facing conformation to the outwardfacing conformation. In F334Y, this step may occur more rapidly than in wild type hENT1, effectively increasing the population of transporters in the outward-facing conformation on the plasma membrane relative to that of wild type. There is currently insufficient structural information regarding ENT proteins to make confident predictions regarding the specific molecular interactions that would account for this mechanism. However, one possibility may be that in the empty, inward-facing conformation, the side chain hydroxyl of Tyr may interact with another residue in a manner that mimics an interaction that occurs in the nucleoside-bound conformation. Taken together, our results suggested that Phe-334 and Asn-338 were not critical residues for adenosine transport, despite their high level of conservation in the ENT family, although Phe-334 may influence the rate of catalytic turnover.   Table 2  Although functional hENT1 molecules were localized to both plasma and intracellular membranes, N338C and N338Q displayed an altered distribution, with functional molecules present only at the cell surface even though total protein abundance was relatively unaffected (Figs. 3 and 5). These results indicated that a significant portion of these mutant proteins may have misfolded and become nonfunctional during synthesis. Nonetheless, the remaining active fraction was functional and displayed similar turnover rates to wild type hENT1, suggesting that the functional mutant proteins were correctly folded. The helical wheel projection of TM 8 depicted in Fig. 8 shows that Asn-338 forms part of a hydrophilic patch together with the highly conserved Arg-341 and Arg-345, the corresponding residues of which have been demonstrated to be critical for function and plasma membrane targeting (Asp-389 and Arg-393, respectively) of LdNT2 from Leishmania donovani (Fig. 1C) (47). These findings are consistent with the findings of this study in which N338C and N338Q appeared to impair folding of the hENT1 protein into a functional transporter molecule. Future studies will address the potentially interesting role for this residue in protein folding and/or trafficking in more detail and in a model system more amenable to such studies. Fig. 8 suggests that Phe-334 and Asn-338 are located close together and on the same side of the transmembrane helix as other highly conserved and/or hydrophilic residues, suggesting that they are more likely to be located within the bundle of transmembrane helices, facing away from hydrophobic lipid bilayer. Furthermore, both the F334C and N338C mutants were sensitive to inhibition by the sulfhydryl-specific probe MTSEA, although the effects were only partial and not protectable by nucleosides. These data suggested that the effects observed in this study may have been a result of indirect effects on the nucleoside and inhibitor binding pockets.
The effects of Phe-334 mutations on dipyridamole and dilazep interactions suggested that large hydrophobic side chains were favored, whereas the effects of Asn-338 mutations on soluflazine and draflazine interactions favored hydrophilic side chains (Fig. 4). One explanation for these results are possible hydrophobic contacts between Phe-334 and dilazep/dipyridamole and hydrogen bonds between Asn-338 and draflazine/ soluflazine, although indirect mechanisms cannot be ruled out. If direct interactions occur, they may do so in a manner that does not occlude Phe-334 and Asn-338 or that the MTSEA reaction is much more rapid than the resident time of the permeants in their respective binding pockets. Nonetheless, the results of these studies suggest that the respective identities of residues 334 and 338 have important structural influences on the conformation of the hENT1-binding pockets for dilazep/ dipyridamole and draflazine/soluflazine, respectively.
Although its properties were not addressed in detail in this study, draflazine has been suggested to be a mixed type inhibitor of es transporters (12,28,29). Although Asn-338 may be accessible from the extracellular side, its location toward the cytoplasmic end of TM 8 suggests that draflazine is sufficiently hydrophobic to diffuse across the plasma membrane and thus can access its binding site from either side of the membrane with differing affinities, providing an explanation for its mixed type inhibition behavior (48). Further experimentation will be required to validate this model. This is the first study to address the molecular determinants of draflazine and soluflazine interactions with recombinant ENTs.
Both hydrophobic and hydrophilic residues at position 338 equally impacted dipyridamole, dilazep, and NBMPR interactions, suggesting that the hydrophilic character of Asn-338 was not an important determinant for hENT1 interactions with these inhibitors (Fig. 4B). In a follow-up to the study that identified TM 8 LdNT2 residues as important for transporter function and targeting, a secondary mutation at Asn-175 was identified in the intracellular loop between TMs 4 and 5 that rescued the transport-deficient D389N LdNT2 mutant, although this effect was more likely because of compensatory effects on the tertiary structure of the protein than to a direct interaction with TM 8 residues (49). These results suggest that the intracellular half of TM 8 is conformationally linked to but not necessarily in close proximity to TMs 4 and 5, which have been demonstrated in several studies to be important for, or implicated to directly participate in, permeant and inhibitor binding (3, 26, 27, 50 -55).
Our previous studies identified residue 33 in TM 1 as important for dilazep and dipyridamole interactions and showed that it may be close to Leu-442 in TM 11 forming part of the dipyridamole-binding site (21)(22)(23). This study suggests that TM 8 is also important for dipyridamole interactions and may also be in FIGURE 8. Helical wheel projections of TM 8. The helical wheel projection, viewed from the extracellular side of the membrane, was generated using the HelixWheel program on the EXPASY molecular biology server and subsequently transposed onto a high resolution helical wheel template. Based on multiple sequence alignments of 43 ENT sequences, residues indicated by darkened circles were hydrophilic (Ser, Thr, Asp, Glu, Asn, Gln, Lys, or Arg) in Ն40% of the sequences analyzed and residues indicated by dark"spokes" were identical in Ն40% of the sequences analyzed. The residues under investigation in this study are marked (*). close proximity to TMs 1 and 11, although its specific structural role in the dipyridamole-binding site may be indirect.
Our results, taken together with those of previously published studies, suggest that conserved residues in close proximity to each other on the same face of TM 8 play critical roles in influencing inhibitor interactions, rates of conformation change during transport, and/or protein folding. Future work will be aimed at addressing the potentially interesting role of Asn-338 in hENT1 synthesis and trafficking, as suggested by the results of this study.