Functional and Molecular Characterization of Nucleobase Transport by Recombinant Human and Rat Equilibrative Nucleoside Transporters 1 and 2 CHIMERIC CONSTRUCTS REVEAL A ROLE FOR THE ENT2 HELIX 5–6 REGION IN NUCLEOBASE TRANSLOCATION*

The human (h) and rat (r) equilibrative (Na (cid:1) -independ-ent) nucleoside transporters (ENTs) hENT1, rENT1, hENT2, and rENT2 belong to a family of integral membrane proteins with 11 transmembrane domains (TMs) and are distinguished functionally by differences in sensitivity to inhibition by nitrobenzylthioinosine and coronary vasoactive drugs. Structurally, the proteins have a large glycosylated loop between TMs 1 and 2 and a large cytoplasmic loop between TMs 6 and 7. In the present study, hENT1, rENT1, hENT2, and rENT2 were produced in Xenopus laevis oocytes and investigated for their ability to transport pyrimidine and purine nucleobases. hENT2 and rENT2 efficiently transported radiolabeled hypoxanthine, adenine, guanine, uracil, and thymine (ap-parent K m values 0.7–2.6 m M ), and hENT2, but not rENT2, also transported cytosine. These findings were

Plasma membrane transport processes for nucleosides and nucleobases play key roles in many aspects of mammalian physiology and pharmacology (1)(2)(3)(4)(5). In particular, uptake of exogenous nucleosides and nucleobases is the first step of nucleotide synthesis in tissues such as bone marrow and intestinal epithelium (and certain parasitic organisms) that lack de novo pathways for purine biosynthesis (5)(6)(7). The same transport processes also mediate cellular uptake of many synthetic nucleoside and nucleobase analogs used in cancer, viral, and parasite chemotherapy (3)(4)(5)8). Independent transport processes specific for nucleosides or nucleobases as well as shared mechanisms of nucleoside and nucleobase transport have been described (2,4).
In human and other mammalian cells and tissues, uptake of nucleosides is brought about by members of the concentrative (Na ϩ -dependent) nucleoside transporter (CNT) 1 and equilibrative (Na ϩ -independent) nucleoside transporter (ENT) families (3,5). CNTs have been described primarily in specialized epithelia, whereas ENTs occur in most, possibly all, cell types and tissues. Three CNT and two ENT isoforms have been identified. Human (h) and rat (r) CNT1 and CNT2 both transport uridine, but are otherwise selective for pyrimidine (hCNT1 and rCNT1) and purine (hCNT2 and rCNT2) nucleosides (9 -14). In contrast, hCNT3 and its mouse (m) ortholog mCNT3 transport both purine and pyrimidine nucleosides (15). Human and rat ENT1 and ENT2, which are also broadly selective for purine and pyrimidine nucleosides, are distinguished functionally by a difference in sensitivity to inhibition by nitrobenzylthioinosine (NBMPR), hENT2 and rENT2 being NBMPR-insensitive (16 -19). They also differ in sensitivity to inhibition by the coronary vasodilator drugs dipyridamole, dilazep, and draflazine (hENT1 Ͼ hENT2 Ͼ rENT1 ϭ rENT2). The relationships of these proteins to the transport processes defined by functional studies are: CNT1 (cit), 2 CNT2 (cif), CNT3 (cib), ENT1 (es), and ENT2 (ei). ENTs are widely distributed in other eukaryotes, but appear to be absent in prokaryotes, whereas CNTs are present in both (3,5).
Nucleobase transport has been studied most extensively in microorganisms. The processes present in mammalian cells are less well defined, although both equilibrative (Na ϩ -independ-ent) and concentrative (Na ϩ -dependent) nucleobase-specific transport activities have been described in a variety of different cell types and tissues (2,4). Equilibrative nucleobase transport has been found, for example, in human erythrocytes, human T-lymphoblastoid cells, pig renal epithelial cells, and S49 mouse-derived lymphoma cells (20 -22), whereas concentrative nucleobase transport occurs in kidney, intestine, placenta, and choroid plexus (23)(24)(25)(26)(27). Unlike nucleosides, little is known about the molecular basis of nucleobase transport in mammalian cells, and no cDNA encoding a functional mammalian nucleobase transporter has been cloned. Of three families of nucleobase transport proteins identified in bacteria, fungi, and plants (designated nucleobase-ascorbate transporters, PRT, and purine permease in Ref. 4), only one (nucleobase-ascorbate transporters) has known orthologs in mammals. Of these, human and rat SVCT1 and SVCT2 function as Na ϩ -dependent ascorbate transporters (28 -32), whereas mYspl1 is a mouse protein of unknown function (33).
Functional and molecular distinctions between nucleoside and nucleobase transport processes are not absolute. FUR4, for example, is a yeast PRT permease that transports both uracil and uridine (34), whereas FUI1, another deficient Saccharomyces cerevisiae PRT family member, transports uridine, but not nucleobases (35). PfENT1 is a purine nucleoside-selective ENT family member from Plasmodium falciparum that also transports nucleobases (36). Similarly, there is evidence from flux studies for a role for mammalian nucleoside transporters in nucleobase transport (2,4). In the human umbilical vein endothelial cell line EVC 304 (37), radiolabeled hypoxanthine transport was inhibited by adenosine, thymidine, and uridine, and radiolabeled adenosine transport was inhibited by hypoxanthine. A shared mechanism for thymidine and hypoxanthine transport has also been reported for human breast MCF7 and T-47D adenocarcinoma cells (38). The molecular entities responsible for these dual nucleoside/nucleobase transport activities were not determined, although inhibition by high (micromolar) concentrations of NBMPR (EVC 304) and dipyridamole (MCF7 and T-47D) and the lack of Na ϩ dependence suggested involvement of the ei rather than the es, cit, cif, or cib processes. Results of earlier studies of hypoxanthine transport in mouse S49 lymphoma cells (39) had suggested a minor role of the es process in hypoxanthine translocation.
Most cells express multiple nucleoside transport activities, making it technically difficult to evaluate the contributions of individual nucleoside transport processes to nucleobase translocation. This limitation can be overcome using recombinant DNA technology. Transport experiments in which recombinant human, rat, and mouse CNTs were produced individually in Xenopus oocytes and then evaluated for uridine/uracil uptake found that CNT1-3 did not transport uracil (12,14,15). Similarly, we show here that recombinant hCNT1, rCNT1, hCNT2, rCNT2, and hCNT3 do not transport hypoxanthine, the nucleobase permeant used in the experiments with cultured EVC 304, MCF7, T-47D, and S49 cells. To determine the role of ENT proteins in nucleobase translocation, we also examined nucleobase fluxes in Xenopus oocytes producing recombinant hENT1, rENT1, hENT2, or rENT2. hENT2 was also produced in a purine nucleobase transport-deficient strain of S. cerevisiae. Our results demonstrate that the human and rat ENT2 isoform has the dual capability of transporting both nucleosides and nucleobases, establishing ENT2 as the first identified mammalian nucleobase transporter protein. This newly demonstrated functional activity of human and rat ENT2 was exploited in chimeric studies of rENT1 and rENT2 to identify potential pore-lining regions of rENT2 responsible for nucleobase translocation.
Radioisotope Flux Studies-Transport in Xenopus oocytes was traced using the appropriate 14 C-or 3 H-labeled permeant (Moravek Biochemicals (Brea, CA) or Amersham Biosciences) at a concentration of 1 and 2 Ci/ml for 14 C-and 3 H-labeled compounds, respectively. Flux measurements were performed at room temperature (20°C) as described previously (9,12,16,17,40) on groups of 12 oocytes in 200 l of transport medium, which consisted of 100 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES, pH 7.5. Except where otherwise indicated, the concentration of radiolabeled permeant was 20 M. At the end of the incubation period, extracellular label was removed by six rapid washes in ice-cold transport medium, and individual oocytes were dissolved in 5% (w/v) SDS for quantitation of oocyte-associated radioactivity by liquid scintillation counting (LS 6000 IC, Beckman, Mississauga, Ontario, Canada). The flux values shown are means Ϯ S.E. of 10 -12 oocytes, and each experiment was performed at least twice on different batches of cells. Kinetic parameters were determined using programs of the ENZFITTER software package (Elsevier-Biosoft, Cambridge, UK). Calculations of apparent K i values assumed a competitive mechanism of inhibition.
The uptake of [ 14 C]hypoxanthine (Moravek Biochemicals) by proliferating yeast was measured as described previously (48) using the "oil-stop" method. Briefly, yeast were grown in CMM/glucose to an A 600 of 0.8 -1.5, washed three times in fresh medium, and resuspended to an A 600 of 2. Transport reactions were carried out at room temperature and initiated by the rapid addition of a small volume of radiolabeled permeant to a final concentration of 10 M (1 Ci/ml). Uptake was terminated at graded time intervals by centrifugation (12,000 ϫ g, 2 min) of 200-l portions of yeast through equal volumes of transport oil contained in 1.5-ml microcentrifuge tubes. Oil was removed by aspiration, and the yeast pellets solubilized overnight in 5% (v/v) Triton X-100 for liquid scintillation counting. Blank values, corresponding to trapped extracellular space, were determined by mixing yeast with radiolabeled hypoxanthine (10 M, 1 Ci/ml), followed by immediate centrifugation through oil. The flux values shown, corrected for trapped extracellular space, are presented as means Ϯ S.E. of triplicate determinations. Each experiment was performed at least twice on different batches of cells.

RESULTS AND DISCUSSION
Membrane transport studies in various mammalian cell and tissue preparations have produced evidence that equilibrative (Na ϩ -independent) cellular uptake of nucleosides involves two mediated processes (es and ei), both of which may also have a role in the transport of nucleobases (1,2,4,(37)(38)(39). Recently, the proteins responsible for these functional activities have been identified as members of the ENT protein family, and are designated in humans and rats as hENT1 and rENT1 (the es transporters) and hENT2 and rENT2 (the ei transporters) (16 -19).
Transport assays of recombinant human and rat ENT1 and ENT2 proteins produced in Xenopus oocytes (16 -18) and in mammalian transport-deficient cell lines (19,49) have verified that both isoforms function as broad specificity, purine and pyrimidine nucleoside transporters, ENT2 having generally lower K m values for nucleoside influx than ENT1 (49). Indirect evidence of a dual role of ENT2 in nucleoside and nucleobase transport has been provided by competition experiments that demonstrated inhibition of hENT2-mediated uridine transport by the nucleobase hypoxanthine, with no corresponding effect on hENT1 (19,49). In other recent studies, it has been established that a protozoan ENT family member from P. falciparum (PfENT1) functions as both a nucleobase and nucleoside transporter (36). Here, we have combined similar direct nucleobase transport studies in Xenopus oocytes with production in FCY2 purine-cytosine permease-deficient S. cerevisiae and chimeric approaches to investigate the functional and molecular characteristics of nucleobase transport by recombinant human and rat ENT1 and ENT2.
Hypoxanthine Uptake by Recombinant Human and Rat Nucleoside Transporters-Uridine is accepted as a permeant by all mammalian ENT and CNT proteins, and the apparent K m values for uridine inward transport are similar for human and rat ENT1 and ENT2 (3, 16 -18, 49). Hypoxanthine was selected as the test nucleobase permeant because of its previous use in nucleobase transport studies in cultured mammalian cells (37)(38)(39) and its ability to inhibit uridine transport mediated by recombinant hENT2 (19,49). Fig. 1 (A and B) presents a representative radioisotope flux experiment that compares the nucleoside (uridine) and nucleobase (hypoxanthine) transport capabilities of recombinant hENT1, hENT2, rENT1, and rENT2 produced in Xenopus oocytes. As shown in Fig. 1A, uridine (20 M) was taken up to similar extents by each of the four recombinant transporters during 2-min uptake intervals, which, we have established, measure initial rates of uridine transport (influx) (9,12,16,17,40). Values for the ENT-mediated components of transport, determined by subtracting the uptake in water-injected oocytes (control) from uptake in RNA transcript-injected oocytes, were 0.41 Ϯ 0.04, 0.52 Ϯ 0.05, 0.68 Ϯ 0.03, and 0.95 Ϯ 0.04 pmol/oocyte for hENT1, rENT1, hENT2, and rENT2, respectively. The corresponding data for hypoxanthine ( correspondingly large mediated components of nucleobase uptake by hENT2 and rENT2 (1.03 Ϯ 0.13 and 1.70 Ϯ 0.06 pmol/ oocyte, respectively), and no detectable transport for hENT1 or rENT1. Basal uptake of hypoxanthine in water-injected oocytes (0.20 Ϯ 0.02 pmol/oocyte) was higher than that of uridine (0.008 Ϯ 0.004 pmol/oocyte), a finding consistent with the previously established presence of low level endogenous nucleobase transport activity in Xenopus oocytes (27). In contrast, oocytes lack endogenous nucleoside transporters, and the basal uptake of uridine by water-injected oocytes, therefore, represents passive diffusion across the plasma membrane lipid bilayer (9,15,16,40).
For comparison, the relative abilities of the human concentrative (Na ϩ -dependent) transporters hCNT1, hCNT2, and hCNT3 and the rat orthologs rCNT1 and rCNT2 to transport uridine and hypoxanthine were also assessed in Fig. 2 (A and  B). These results, which showed no mediated hypoxanthine transport by any of the proteins, were consistent with the previously established inability of hCNT1-3 to transport the nucleobase uracil (12,14,15).
Time Courses of Hypoxanthine Transport-The results presented in Fig. 1B represent the first direct evidence that mammalian ENT2 proteins transport nucleobases in addition to nucleosides. In Fig. 3 (A and B) are shown representative experiments in which time courses (1-30 min) of uptake of hypoxanthine (20 M) were determined in hENT2-and rENT2producing oocytes compared with those in control water-injected cells. There were large differences in cellular uptake of hypoxanthine between oocytes producing hENT2 and rENT2 and those injected with water. Also shown as insets to Fig. 3 (A  and B) are expanded time courses of hypoxanthine uptake measured during the first 5 min of incubation. Mediated transport of hypoxanthine (uptake in RNA transcript-injected oocytes minus uptake in water-injected oocytes) was rapid for both transporters and was approximately linear with time for the first 2 min of incubation.
Corresponding time courses for hypoxanthine (20 M) uptake in hENT1-and rENT1-producing oocytes compared with water-injected oocytes are shown in Fig. 3C. Although there was no significant difference between RNA transcript-injected cells and those injected with water at early time intervals (see also Fig. 1B), more prolonged incubation revealed a minor component of mediated transport in both hENT1-and rENT1-producing oocytes. The small magnitudes of these hypoxanthine fluxes relative to those for ENT1-mediated uptake of uridine ( Fig. 1A) or ENT2-mediated uptake of both hypoxanthine and uridine (Fig. 1, A and B) demonstrated a large difference in nucleobase transport capability between the ENT2 and ENT1 isoforms. These results were consistent with the low efficiency of hypoxanthine transport observed for the NBMPR-sensitive (es) uptake process in cultured S49 cells (39). When similar experiments were conducted with the CNT proteins under the same prolonged uptake conditions (30-min incubation), no mediated transport of hypoxanthine was observed for any of the human or rat isoforms (data not shown). Subsequent in depth nucleobase transport studies focused on hENT2 and rENT2. Uptake intervals of 2 min were used in experiments (cross-inhibition and concentration dependence studies) where it was necessary to measure initial rates of nucleobase transport (influx). In other studies comparing uptake of a panel of different radiolabeled nucleobases, more prolonged (30-min) uptake intervals were used.
Selectivity of Nucleobase Transport by Recombinant hENT2 and rENT2-Complex patterns of nucleobase selectivity have been described for the equilibrative and concentrative nucleobase-specific transport processes of mammalian cells (reviewed in Refs. 2 and 4). Some accept both purine and pyrimidine nucleobases, whereas others are either purine or pyrimidine nucleobase-specific. In the cross-inhibition experiment shown in Fig. 4A, the initial rate of hypoxanthine transport (20 M, 2-min flux) by hENT2-producing oocytes was strongly inhibited by excess (5 mM) unlabeled purine (hypoxanthine, adenine) and pyrimidine nucleobases (thymine, uracil), as well as by inosine and uridine. The pyrimidine nucleobase cytosine also exhibited significant, but less marked inhibition, of hypoxanthine influx. Similar data were obtained for rENT2 (Fig. 4B), except that cytosine had no detectable effect on hypoxanthine transport. Therefore, both transporters appeared to be broadly selective for a range of purine and pyrimidine nucleobases. The finding that ENT2-mediated transport of hypoxanthine was inhibited by both uridine and inosine argues against the (unlikely) possibility that the observed fluxes were a consequence of h/rENT2-specific up-regulation of endogenous nucleobase-specific transport mechanisms. As described in subsequent sections, this possibility was also excluded by NBMPR and dipyridamole inhibition experiments, by kinetic determinations of apparent K i values for uridine inhibition of hypoxanthine influx and hypoxanthine inhibition of uridine influx, and by an independent demonstration of hypoxanthine transport by recombinant hENT2 produced in purine-cytosine permease (FCY2)-deficient yeast.
Because transport inhibition can occur in the absence of translocation of the inhibiting substance, the ability of hENT2 and rENT2 to transport nucleobases other than hypoxanthine was measured directly in oocyte experiments with a panel of radiolabeled purine (adenine, guanine, hypoxanthine) and pyrimidine nucleobases (cytosine, uracil). Fig. 5 shows uptake of these potential nucleobase permeants (20 M) compared with uridine in hENT2-and rENT2-producing oocytes and in control, water-injected cells. Similar to previous studies (9, 12, 16, 17, 40), a longer (30-min) uptake interval was used to detect permeants with both high and low ENT2-mediated transport activities. Consistent with the broad inhibition profile shown in Fig. 4A, hENT2 transported all of the purine and pyrimidine nucleobases tested, including guanine, which was not investigated in the cross-inhibition study because of its relatively low solubility (Fig. 5A). Cytosine was transported to a lesser extent than the other nucleobases or uridine (mediated flux 3.13 Ϯ 0.70 pmol/oocyte⅐30 min Ϫ1 compared with 9.2-31.0 pmol/oo-cyte⅐30 min Ϫ1 for the other permeants), which, together with its partial inhibition of hypoxanthine uptake observed in Fig.  4A, suggested that hENT2 has relatively low affinity for cytosine. hENT2 has also been shown to have a low apparent affinity for transport of cytidine and gemcitabine (an anticancer deoxycytidine analog) (49,50). The distinction between cytosine and other nucleobases was even more marked for rENT2, which, in the experiments of Fig 4B, was found be insensitive to cytosine inhibition. As shown in Fig 5B, rENT2 exhibited a very low mediated cytosine flux of Ͻ0.2 pmol/ oocyte⅐30 min Ϫ1 compared with 2.6 -12.7 pmol/oocyte⅐30 min Ϫ1 for the other nucleobases and uridine.
NBMPR and Dipyridamole Inhibition of Nucleobase Transport by Recombinant hENT2-Uridine transport by recombinant hENT2 produced in oocytes is inhibited by NBMPR and dipyridamole in the micromolar concentration range (18). Consistent with this pharmacologic profile, the inset to Fig. 4A shows that 1 M NBMPR and dipyridamole had no measurable effect on hENT2-mediated hypoxanthine influx, whereas both compounds at the higher concentration of 10 M caused partial inhibition of hypoxanthine transport activity.
Kinetics of Nucleobase Transport by Recombinant hENT2 and rENT2-Figs. 6 (A-D) and 7 (A-D) show representative concentration-dependence curves for uptake of adenine, hypoxanthine, thymine, and uracil measured as initial rates of transport (2-min flux) in hENT2-and rENT2-producing oocytes and in control water-injected cells. The hENT2-and rENT2-mediated components of influx were saturable and conformed to simple Michaelis-Menten kinetics, and the resulting kinetic constants are presented for each nucleobase in Table I. Apparent K m values varied between 0.74 and 2.6 mM for hENT2 (hypoxanthine, adenine Ͻ thymine Ͻ uracil) and between 1.0 and 2.5 mM for rENT2 (hypoxanthine, thymine Ͻ uracil Ͻ adenine). These K m values, which are similar to the K m and K i values estimated for nucleobase transport by system ei in human EVC 304 cells (37), are higher than the apparent K m values for transport of the corresponding nucleosides by recombinant hENT2 and rENT2 (Table I) (17)(18)(19)49), indicating a difference in the apparent affinities of ENT2 for nucleobases and nucleosides (nucleobases Ͻ nucleosides).
For uracil and uridine, the kinetic data in Table I gave calculated nucleobase:nucleoside K m ratios of 13 for hENT2 and 6.0 for rENT2. Similar K m ratios (10 for uracil/uridine, 14 for hypoxanthine/inosine, 7.9 for adenine/adenosine, and 2.4 for thymine/thymidine) were obtained for hENT2 using K m values for nucleoside transport by the recombinant protein in transfected pig PK15NTD cells (49). These lower apparent affinities of hENT2 and rENT2 for nucleobases were compensated by higher V max values. Calculated V max :K m ratios (a measure of "transport efficiency") for nucleobases were therefore either in the same range as uridine, or higher (Table I). As illustrated in Figs. 1 (A and B) and 5 (A and B), this allows hENT2 and rENT2 to transport nucleobases in the physiological (micromolar) concentration range at rates greater than or equal to uridine. Kinetic studies of adenine and adenosine transport by PfENT from P. falciparum suggest that this transporter also exhibits comparable efficiencies of nucleobase and nucleoside transport (36).
Xenopus oocytes possess endogenous Na ϩ -dependent and Na ϩ -independent processes for nucleobase transport that have K m values in the low micromolar range and exhibit low activities (27). These processes, therefore, did not contribute significantly to the concentration-dependence profiles shown in Figs. 6 and 7. In the case of hypoxanthine, for example, the published kinetic constants (27) suggest a contribution of Յ10% to the 5 mM uptake seen in water-injected oocytes in this study. Nucleobase uptake in water-injected oocytes exhibited an apparent linear concentration dependence, suggesting that entry oc-  Table I. curred primarily by passive diffusion across the plasma membrane lipid bilayer.
Because endogenous oocyte nucleobase transporters could possibly be up-regulated in response to ENT2 expression, we undertook a further series of kinetic experiments to verify formally that ENT2 was responsible for the dual nucleobase/nucleoside fluxes seen in hENT2-and rENT2-producing oocytes. In the quantitative test elaborated by Christensen and others (51), two interacting permeants (A and B) are transported by the same system (in this case ENT2) if the K i for inhibition of permeant A by inhibitor   Table I. B matches the K m of substrate B for its own transport and vice versa. Fig. 8 (A-D) therefore shows a series of reciprocal crossinhibition experiments in hENT2-and rENT2-producing oocytes measuring the abilities of graded concentrations of uridine or hypoxanthine (0.05-5.0 mM) to inhibit mediated influx (20 M, 2-min flux) of the other permeant. As anticipated from the results presented in Fig. 4 (A and B) and other studies (19,49), uridine and hypoxanthine effectively inhibited transport of the other permeant. Uridine and hypoxanthine apparent K i values derived from Fig. 8 IC 50 values (Table I) corresponded closely with their respective K m values for hENT2-and rENT2-mediated transport and were therefore consistent with a common ENT2 mechanism of cellular uptake. The inset to Fig. 8B presents a Dixon plot of a control experiment measuring uridine inhibition (0.5-2.0 mM) of hENT2-mediated hypoxanthine transport at two different substrate concentrations (0.2 and 0.7 mM). As required by the Christensen test (51), the interaction between hypoxanthine and uridine was competitive. Furthermore, the Dixon plot apparent K i value of 0.25 mM was consistent with the other hENT2 uridine K i and K m values summarized in Table I.
Recombinant hENT2 Transport in S. cerevisiae-As a further independent test for ENT2-mediated nucleobase transport, we used hENT2 cDNA contained in the yeast/E. coli shuttle vector pYPGE15 to produce recombinant hENT2 in yeast deficient in FCY2 purine-cytosine permease. Uptake of radiolabeled hypoxanthine measured in the presence and absence of excess (10 mM) uridine was then compared with corresponding hypoxanthine fluxes in control cells transfected with the empty pYPGE15 vector (Fig. 9, A and B). Consistent with FCY2 being the primary mechanism for hypoxanthine transport in normal yeast, uptake of hypoxanthine by control FCY2-deficient cells was slow, reaching values of Ͻ0.01 pmol/g protein in 60 min (Fig. 9B). Uptake of hypoxanthine in hENT2-producing cells was 9.4-fold higher (Fig. 9A), and similar in magnitude to fluxes of uridine seen in corresponding hENT2-mediated uridine transport experiments (45). Hypoxanthine transport was reduced back to basal levels in the presence of unlabeled uridine (Fig. 9A), in contrast to control cells where uridine had no effect on hypoxanthine uptake (Fig.  9B). Therefore, hENT2-producing yeast exhibited a uridine-  Table I. sensitive hypoxanthine flux not found in yeast transfected with the empty vector. These findings complement the Xenopus oocyte cross-inhibition data shown in Figs. 4 (A and B) and 8 (A and B) and prove, using an entirely different expression system, that ENT2 functions as a transporter for both nucleobases and nucleosides.
Chimeric Studies between rENT1 and rENT2-Structurally, ENT transporters have a common membrane architecture of 11 predicted TMs with a large extracellular glycosylated loop between TMs 1 and 2 and a large cytoplasmic loop between TMs 6 and 7. Sequence homology between family members is greatest within putative TM regions. In a previous study (52), the difference in vasodilator sensitivity between hENT1 (vasodilator-sensitive) and rENT1 (vasodilator-resistant) was used as the rationale for analysis of structural chimeras between the two proteins to identify domains involved in vasoactive drug binding. It was established that vasoactive drug inhibition involved two domains in the amino-terminal half of hENT1 (TMs 1-2 and TMs 3-6), with TMs 3-6 being the major site of interaction. Because functional studies suggest that vasodilators and NBMPR compete with nucleosides for binding to common or overlapping exofacial sites on the transporter (1-3, 53-56), it was hypothesized that these regions contribute to the permeant translocation channel of the transporter. Similarly, we have recently employed chimeras to map transporter interactions with NBMPR (42). To determine NBMPR-binding domains free from any structural elements required for vasodilator interactions, the rat transporters rENT1 and rENT2 were employed, both of which are resistant to inhibition by vasoactive drugs (17). These studies established that TMs 3-6 were also responsible for NBMPR binding, because microdomain swaps within this region indicated contributions from both TMs 3-4 and TMs 5-6 (42). In the present study, we have taken advantage of the difference in nucleobase/nucleoside selectivity between rENT1 and rENT2 to use the same series of rENT1/2 chimeras to map transporter interactions with nucleobases.
As shown in Fig. 10A, rENT1 and rENT2 contain 457 and 456 residues, respectively, and are 49% identical and 68% similar in amino acid sequence. The nomenclature and composition of the five rENT1/2 chimeras used in the study of nucleobase transportability are illustrated in Fig. 10B, with the three junction points represented by arrows A, B, and C in Fig. 10A. Graft sites A and C were in positions corresponding to those used previously to generate chimeras between hENT1 and rENT1 (52). The chimeric proteins were produced in Xenopus oocytes and evaluated for nucleoside (uridine) and nucleobase (hypoxanthine) transport activities (20 M). As in previous studies with these chimeras (42), we used a 30-min uptake interval to maximize fluxes mediated by low activity constructs.
Each of the five chimeras transported uridine, suggesting that the native conformation was generally retained in all of the constructs (42) (Fig. 11A). R(3-6) and R(3-4) displayed lower levels of uridine transport activity than the other chimeras, possibly as a result of reduced plasma membrane targeting of the recombinant chimeric proteins. Nevertheless, all constructs gave fluxes large enough to determine their nucleobase transport capability. Consistent with the data presented in Fig. 3C, wild-type rENT1 showed only modest hypoxanthine transport activity, and there was the expected large difference in hypoxanthine uptake between the two native transporters (rENT2 Ͼ Ͼ rENT1) (Fig.  11B). The first chimera of the series, R(1-6), was a 50:50 construct in which TMs 1-6 of rENT1 (representing the aminoterminal half of rENT1) were replaced by those of rENT2 (the splice site is shown as arrow C in Fig. 8A). In the representative experiment of Fig. 11B, the resulting chimera was hypoxanthine transport-positive, suggesting that the site(s) conferring nucleobase transport activity lay within the amino-terminal half of rENT2. To identify the critical nucleobase transport domains within this region, two additional chimeric transporters, R(1-2) and R (3)(4)(5)(6), with a splice site designed near the middle of TMs 1-6 (arrow A in Fig. 10A) were compared. Chimera R(1-2), composed of rENT2 from the amino terminus to the end of TM 2 and rENT1 for the rest of the protein, did not transport hypoxanthine, narrowing the region of interest to TMs 3-6 and the 12 residues of the preceding intracellular loop. For simplicity, this region is designated hereafter as TMs 3-6, and the abbreviation TM 1-2 denotes the domain from the amino terminus to the end of TM 2.
Conclusions-We have demonstrated that both the human and rat ENT2 proteins transported purine and pyrimidine nucleobases, including hypoxanthine, adenine, guanine, thymine, and uracil. hENT2, but not rENT2, also transported cytosine, indicating a species difference between the two ENT2 orthologs. In contrast to the human and rat ENT2 proteins, ENT1 proteins exhibited negligible hypoxanthine transport activity. Nucleobases were transported by hENT2 and rENT2 with lower apparent affinities than nucleosides, but with higher maximum velocities, such that the transport efficiencies of both proteins for nucleobases and nucleosides were very similar in the physiological concentration range. The ability of the ENT2 proteins, but not the ENT1 proteins, to transport nucleobases as well as nucleosides is a key functional difference between these ENT isoforms and provides a possible explanation why many cell types co-express both proteins. Although the ribose 3Ј-hydroxyl group has been shown to be important for es (ENT1)-mediated nucleoside transport (57), the demonstration that human and rat ENT2 both transport nucleobases suggests that recognition of the sugar moiety is less critical for ENT2. This conclusion is supported by the observation that antiviral nucleoside drugs (3Ј-azido-3Ј-deoxythymidine, 2Ј,3Јdideoxycytidine, and 2Ј,3Ј-dideoxyinosine), which lack the ribose 3Ј-hydroxyl group are more efficiently transported by human and rat ENT2 than by human and rat ENT1 (58). Another ENT family member (PfENT) from the malaria parasite P. falciparum also efficiently transports nucleobases (36), demonstrating that dual nucleobase and nucleoside transport capability is not unique to the human and rat ENT2 proteins.
Although functional studies have provided clear evidence for the existence of other, nucleobase-specific uptake mechanisms in different cell types and tissues (reviewed in Refs. 2 and 4), the present study establishes human and rat ENT2 as the first mammalian proteins with demonstrated nucleobase transport activity. The nucleobase-specific transport processes that have been defined kinetically in functional studies tend to have apparent K m values for nucleobase transport (2-200 M) that are lower than those reported here for human and rat ENT2 (0.7-2.6 mM). The contributions of these different mechanisms FIG. 10. Topographical model of rENT1 and rENT2 and schematic representation of rENT1/rENT2 chimeric constructs. A, potential membrane-spanning ␣-helices in the topographical model are numbered and putative glycosylation sites in rENT1 and rENT2 are indicated by solid and open stars, respectively. Residues identical in the two proteins are shown as darkened circles. Residues corresponding to insertions in the sequences of rENT1 or rENT2 are indicated by circles containing "ϩ" and "Ϫ" signs, respectively. Splice sites used for the construction of chimeras are represented by arrows A-C. B, a diagrammatic representation of the junction points of the chimeric species used in this study. These correspond to the start of the first cytoplasmic loop, the middle of the second cytoplasmic loop, and the start of the third cytoplasmic loop. Regions derived from rENT2 are shown as closed boxes, and those from rENT1 are shown as open boxes. The nomenclature used in this study is indicated; numbers in parentheses indicate putative TM segments of rENT2 that were transplanted into equivalent positions in rENT1.
FIG. 11. Uptake of hypoxanthine and uridine by recombinant rENT1, rENT2, and rENT1/2 chimeras produced in Xenopus oocytes. Uptake of hypoxanthine and uridine (20 M, 20°C, 30 min) in oocytes injected with water containing RNA transcripts for the different transporters was measured in transport medium containing 100 mM NaCl and compared with basal uptake in control, water-injected oocytes.
to overall cellular uptake of nucleobases will depend on their relative membrane abundance. In addition to transporting physiological nucleobases, it is possible that ENT2 may also mediate cellular uptake of nucleobase drugs (3)(4)(5)8).
Structurally, the major sequence differences between the ENT1 and ENT2 isoforms lie in the large extracellular loops between TMs 1 and 2 and in the putative cytoplasmic loops between TMs 6 and 7. It might be anticipated, therefore, that these loops would be involved in the different nucleobase transport capabilities displayed by the two transporters. However, the results of the present chimeric study have eliminated involvement of these regions and, as well, the carboxyl-terminal half of ENT2 in nucleobase transport. These suggest instead that the structural requirements for nucleobase transport reside within the same general region (TMs 3-6) previously implicated in vasodilator and NBMPR binding (42,52). Although the amino terminus up to the end of TM 2 (including the large extracellular loop) also contributes to vasodilator sensitivity (52), it plays no apparent role in either NBMPR inhibition (42) or nucleobase transport (this study). Microdomain swaps within TMs 3-6 of rENT1 and rENT2 further localized nucleobase transport activity to TMs 5-6 of rENT2 (between splice sites B and C in Fig. 10A). Earlier studies revealed a unique exofacial PCMBS-reactive Cys residue (Cys 140 ) in the outer half of rENT2 TM 4 (59). PCMBS binding to rENT2 Cys 140 was prevented by extracellular uridine, suggesting that this residue and the helix to which it belongs (TM 4) lie within or are closely adjacent to the permeant translocation channel of rENT2. Our demonstration that TMs 5-6 are involved in nucleobase transport suggests that residues in TMs 5-6 also contribute to the permeant translocation channel. Involvement of multiple residues in TMs 3-6 in nucleoside and nucleobase transport is supported by the demonstration that (i) TMs 3-4 and 5-6 contribute to NBMPR binding (42) and (ii) hypoxanthine, like uridine (59), protected Cys 140 in TM 4 of rENT2 from PCMBS inhibition. 3 TMs 5-6 of rENT1 and rENT2 are 50% identical and 60% similar in amino acid sequence. Future studies of this region will involve site-directed mutagenesis to define the specific amino acid residues responsible for nucleobase transport activity.