Demonstration of Equilibrative Nucleoside Transporters (hENT1 and hENT2) in Nuclear Envelopes of Cultured Human Choriocarcinoma (BeWo) Cells by Functional Reconstitution in Proteoliposomes*

The equilibrative nucleoside transporters (ENTs) are a newly recognized family of membrane proteins of which hENT1 is the nitrobenzylmercaptopurine ribonucleoside (NBMPR)-sensitive (es) and hENT2 the NBMPR-insensitive (ei) transporter of human cells. BeWo cells exhibit large numbers (>107/cell) of NBMPR-binding sites and highes and ei nucleoside transport activities relative to other cell types. In this work, we have demonstrated that proliferating BeWo cells possess (i) mRNA encoding hENT1 and hENT2 and (ii) hENT1-specific immunoepitopes. We examined NBMPR binding and its inhibition of uridine transport in various BeWo membrane fractions and proteoliposomes derived therefrom to determine if NBMPR binding to intracellular membranes represented interaction with functionales transporters. Unfractionated membranes and fractions enriched 5-fold in plasma membranes relative to postnuclear supernatants exhibited high NBMPR binding activity. Intact nuclei and nuclear envelopes also exhibited abundant quantities of NBMPR-binding sites with affinities similar to those of enriched plasma membranes (K d = 0.4–0.9 nm). When proteoliposomes were made from octyl glucoside-solubilized membranes, high affinity NBMPR-binding sites were not only observed in crude membrane preparations and plasma membrane-enriched fractions but also in nuclear envelope fractions. Proteoliposomes prepared from either unfractionated membranes or nuclear envelopes exhibited both hENT1-mediated (82–85%) and hENT2-mediated (15–18%) transport of [3H]uridine. These results provided evidence for the presence of functional es and ei transporters in nuclear membranes and endoplasmic reticulum, suggesting that hENT1 and hENT2 may function in the translocation of nucleosides between the cytosol and the luminal compartments of one or both of these membrane types.

Multiple nucleoside transport (NT) 1 -mediated processes exist in mammalian cells and are divided into two groups, depending on whether they are equilibrative or concentrative in nature (for recent reviews, see Refs. [1][2][3]. The equilibrative transporters accept both purine and pyrimidine nucleosides as permeants and are found in most, possibly all, cell types, whereas the concentrative transporters have relatively narrow selectivities for nucleoside permeants and are limited to specialized cell types. The transporters are low abundance proteins, and the equilibrative transporters of human and pig erythrocytes are the only ones that have been purified to homogeneity (4,5). The equilibrative transporters have been subdivided on the basis of sensitivity to nitrobenzylthioinosine (NBMPR); one subtype (es) is inhibited by Յ1 nM NBMPR, as a result of high affinity (K d Ͻ 5 nM) binding of NBMPR at or near the permeant binding site (6 -8), whereas the other subtype (ei) is unaffected by low concentrations (Ͻ1 M) of NBMPR. The concentrative nucleoside transporters comprise several functional subtypes (1) that differ in their substrate selectivities and tissue distributions and, with some exceptions (9), are unaffected by high concentrations (Ͼ10 M) of NBMPR.
Cultured human choricarcinoma (BeWo) cells have been studied mainly for their ability to undergo differentiation from proliferating cytotrophoblast-like cells to larger syncytiotrophoblast-like cells when exposed to methotrexate in a process that resembles normal in utero development (17,18). Previous studies with proliferating BeWo cells (19) revealed unusual NBMPR binding characteristics in comparison with other cell types. BeWo cells have extraordinarily large numbers of high affinity sites (Ͼ2 ϫ 10 7 /cell) of two apparent classes (K d ϭ 0.6 and 14.5 nM), whereas most cultured cell types have far fewer high affinity sites of a single apparent class (e.g. HeLa cells possess 4.1 ϫ 10 5 /cell with K d ϭ 0.7 nM). NBMPR-binding proteins have been shown in reconstitution studies to mediate es NT activity in human erythrocytes (20,21), cultured leukemia (CEM) cells (22), and Ehrlich ascites tumor cells (23). The existence of two classes of binding sites, together with the lack of proportionality between NBMPR-binding sites and maximal transport activities of BeWo and HeLa cells, led to the suggestion (19) that the second class of NBMPR-binding sites in BeWo cells may be es transporters associated with intracellular membranes.
In the present study, we established that BeWo cells, which possess es-and ei-mediated nucleoside transport activities, express hENT1 and hENT2 mRNAs. We quantified binding of NBMPR to various BeWo membrane fractions by analysis of isolated and detergent-solubilized membranes and of reconstituted proteoliposomes. A single class of high affinity NBMPRbinding sites was observed in all membrane preparations, including intact nuclei and nuclear envelopes. There was no evidence of the lower affinity NBMPR-binding sites previously observed in intact BeWo cells in any of the membrane preparations, suggesting loss of these sites during cell disruption. Functional reconstitution of both NBMPR-sensitive (es) and NBMPR-insensitive (ei) uridine transport activities was demonstrated in proteoliposomes derived from both crude membrane and nuclear envelope-enriched fractions. Our results suggest that hENT1 and hENT2 may play a role in the translocation of nucleosides between the cytosol and the lumen of nuclear envelopes and/or endoplasmic reticulum and provide the first evidence for functional ENTs in intracellular membranes of mammalian cells. Cell Culture-The origin and characteristics of BeWo cells have been described (24,25). Cultures were initiated from mycoplasma-free frozen stocks and grown at 37°C in a humidified atmosphere of 5% CO 2 in air as described previously (25). Cells were harvested at weekly intervals from proliferating cultures by trypsin treatment and maintained in plastic culture flasks in Roswell Park Memorial Institute (RPMI) 1640 basal medium supplemented with 5% fetal bovine serum and 5% NUserum type 1V. Cell numbers were determined with an electronic particle counter after dissociation with trypsin/EDTA and resuspension in 0.15 M NaCl solution. For preparation of membranes, cells (5 ϫ 10 6 / flask) were grown in 15 Corning disposable tissue culture flasks (15-cm 2 surface area), and actively proliferating cells were harvested 10 -14 days later, yielding from 8 ϫ 10 8 to 1 ϫ 10 9 cells.

Materials-[G-
RNA Blotting-Poly(A) ϩ RNA was isolated from 1.0 ϫ 10 8 cells with a Fast Track 2.0 kit (Invitrogen, Carlsbad, CA) and used on RNA blots and in reverse transcription polymerase chain reaction (RT-PCR) amplification. To prepare the RNA blots, BeWo mRNA was subjected to electrophoresis on 1% denaturing agarose-formaldehyde gels and transferred to Hybond-N ϩ nylon filters (Amersham). The resulting RNA blots were probed with 32 P-labeled cDNA specific to hENT1 (10) or hENT2 (16) using ExpressHyb (CLONTECH, Palo Alto, CA) according to the manufacturer's recommendations. The blots were washed to high stringency and exposed to x-ray film for autoradiography. To perform the RT-PCR reactions, BeWo cDNA was made using the SuperScript Preamplification System (Life Technologies, Inc.) and subjected to PCR amplification using Taq polymerase (Life Technologies) and primers specific to hENT1 (5Ј-CCTGTGTCCTGTTTCTTGAC and 5Ј-TTGTCA-CACAATTGCCCGGAACAG) or hENT2 (5Ј-GCCTGCGGTTCCTGT-TCGTGC and 5Ј-GCAGCATGCTCAGAGCAGCGCCTTGAAG). The resulting PCR products were resolved on 0.8% agarose gels and either (i) stained with ethidium bromide for visualization using a 1-kilobase pair ladder (Life Technologies) as a size standard or (ii) purified and sequenced using the PCR primers, a dRhodamine cycle sequencing kit (Perkin-Elmer), a model 310 Genetic Analyzer (Perkin-Elmer), and MacVector sequence analysis software (Oxford Molecular, Oxford, UK).
Isolation of BeWo Membranes-Membrane fractions were prepared from trypsin-treated cultures of BeWo cells by sucrose gradient centrifugation (26,27). Cells were resuspended in phosphate-buffered saline (PBS) composed of 20 mM phosphate and 0.15 M NaCl at pH 7.4 and centrifuged (10 min, 1000 ϫ g). Packed cells were resuspended in 1 mM ZnCl 2 and fragmented with a Polytron homogenizer (Brinkmann Instruments). Nuclei and unbroken cells were removed by gentle centrifugation (2 min, 900 ϫ g), after which the supernatants were centrifuged (30 min, 11,000 ϫ g) and the resulting membrane pellets were resuspended in 9.25% (w/w) sucrose. The membrane fractions were then separated using a 15-45% sucrose gradient and stored at Ϫ80°C in 15% (v/v) dimethyl sulfoxide. Sucrose solutions were in 0.5 mM ZnCl 2 , 5 mM K 2 HPO 4 , pH 7.4. The fractions collected at 25% sucrose had 4 -5 times the specific activity of crude nuclei-free membranes when assayed for alkaline phosphatase (EC 3.1.3.1) and 5Ј-nucleotidase (EC 3.1.3.5) and thus were enriched in plasma membranes (Sigma Diagnostics 5Ј-nucleotidase assay kit). Nearly 40% of the total protein applied to the sucrose gradient was recovered in this sucrose fraction. Nuclear membranes were prepared from the nuclei collected during the initial stage of membrane preparation according to Kay et al. (28), whereby nuclei were treated with a low level of DNase I in the presence of Mg 2ϩ at slightly alkaline pH, after which membranes were collected by centrifugation (30 min, 11,000 ϫ g) and stored at Ϫ80°C in 15% (v/v) dimethyl sulfoxide.
Solubilization and Reconstitution of BeWo Membranes-Membranes were thawed at room temperature, washed extensively to remove dimethyl sulfoxide, and resuspended in ice-cold reconstitution buffer (pH 7.4) composed of 100 mM KCl, 10 mM Tris, 0.1 mM MgCl 2 , and 0.1 mM CaCl 2 . Membranes were then mixed with a solution of OCTG (1% final concentration) and asolectin (0.15% final concentration) prepared in reconstitution buffer and incubated on ice for 1 h with frequent mixing. The insoluble material was removed by centrifugation (60 min, 100,000 ϫ g, 4°C), and the supernatant was retained on ice until use. Because solubilization efficiency of detergents is often more dependent on the ratio of detergent to protein than on the actual detergent concentration used (29,30), the membrane protein concentrations were monitored using a modified Bradford assay (23) and adjusted to approximately 200 g of protein/ml. The reconstitution technique was based on a method used previously (23). The solubilized membranes (2 ml) were supplemented with a sonicated preparation of lipids (0.3 ml) consisting of phosphatidylcholine (bovine brain), cholesterol, phosphatidylethanolamine (egg) and phosphatidylserine (bovine brain) in molar ratios of 33:33:26:8, respectively, plus a trace (10 5 dpm/ml) of [ 14 C]cholesteryl oleate. The detergent was removed by gel filtration (Sephadex G-50 medium, 1.5 ϫ 38-cm column) at a flow rate of 1 ml/min; this procedure is known to result in the formation of unilamellar vesicles (31). The void volume fractions containing proteoliposomes were pooled, frozen in ethanol and dry ice, and stored at Ϫ80°C for up to 1 month. For functional assays, proteoliposomes were thawed at room temperature, centrifuged (40,000 ϫ g, 20 min), resuspended in approximately 1 ml of reconstitution buffer, sonicated for 15 s in a cylindrical tank sonicator (Laboratory Supplies Company Inc., Hicksville, NY), and then stored on ice until use. Brief sonication after the freeze/thaw cycle is known to enhance the activity of the reconstituted erythrocyte nucleoside transporter (21).
Equilibrium Binding of [ 3 H]NBMPR-High affinity binding of [ 3 H]NBMPR, which has been used extensively in the characterization of equilibrative nucleoside transport proteins (1,8,32,33), was performed as follows. Membrane preparations, proteoliposomes (ϳ10 g of protein/ml final concentration), and intact nuclei and nuclear membranes (at ϳ40 g of protein/ml) were incubated for 45 min at room temperature with a range of concentrations (0.24 -24 nM) of [ 3 H]NBMPR in either the absence (total binding) or presence (nonspecific binding) of 10 M NBMPR (1-ml final assay volume). Proteins were precipitated by incubation (15 min) with a mixture of ␥-globulins (1.65 mg/ml final concentration) and polyethylene glycol (M r 8000, 10% (w/v) final concentration) and then filtered under vacuum through Whatman GF/B filters. The filters were washed twice with reconstitution buffer containing 8% (w/v) polyethylene glycol at 4°C and analyzed for radioactive content by standard liquid scintillation counting techniques as described previously (23).

Uptake of [ 3 H]Uridine by Proteoliposomes-
The uptake assays were initiated by the rapid addition of 100 l of proteoliposome suspension (ϳ5 g of protein, with or without inhibitors) to 25 l of reconstitution buffer with [ 3 H]uridine (20 M final concentration) and mixed by repeated pipetting. After specific incubation times, 100 l of the reaction mixtures were layered on ice-cold Sephadex G-50 fine minicolumns (see below) and centrifuged (45 s, 700 ϫ g). The effluents were collected in preweighed tubes from which portions were removed for the determination of protein content and 3 H and 14 C contents (dual label counting). The [ 3 H]uridine content of each sample was normalized for intraexperimental variation in eluate volume and phospholipid content based on the 14 C-labeled lipid concentration of the minicolumn effluent. Approximately 50% of the volume added to the minicolumns was recovered in the effluents. The minicolumns used were prepared by filling 1-ml syringes (fitted with a polyethylene filter) with Sephadex G-50 equilibrated in reconstitution buffer containing 10 M dilazep and nitrobenzylthioguanosine (NBTGR). The gel-filled minicolumns were centrifuged (45 s, 700 ϫ g) and placed on ice for at least 30 min prior to use. Estimates of zero time uptake values were obtained by measuring the uptake of [ 3 H]uridine at ϳ2 s in the presence of ice-cold solutions containing 10 mM adenosine, 10 M dipyridamole, and 10 M NBTGR to inhibit all transporter-mediated influx. The zero-time uptake values were subtracted from all other data points, and the accumulation of [ 3 H]uridine was expressed as picomol taken up/mg of proteoliposomal protein. Mediated influx of [ 3 H]uridine was defined as the total uptake minus uptake observed in the presence of 1 mM adenosine, 10 M dipyridamole, and 10 M NBTGR (nonmediated). For inhibition assays, proteoliposomes were incubated with test compounds for 10 min before initiating transport reactions. Protein was measured using bovine serum albumin as the standard with appropriate corrections for detergent interference (34,35).
Flow Cytometry-Nuclei (obtained by cell disruption as described above) were collected by centrifugation (5 min, 500 ϫ g), washed twice in PBS, and resuspended at a density of 10 6 nuclei/ml. The nuclei were then incubated for 1 h at 4°C with either (i) fluorescein isothiocyanate (FITC)-conjugated monoclonal antibodies directed against nuclear pore complex proteins (M414-FITC from Babco (Richmond, CA)) at 0.01 mg/ml or (ii) with rabbit polyclonal antibodies (1 g/ml) raised against a synthetic peptide corresponding to amino acid residues 55-64 (EL-SKDAQASA) of the predicted hENT1 sequence (designated hENT1-(55-64)) and immediately thereafter with goat anti-rabbit FITC-conjugated antibodies. The nuclei were then washed twice with PBS and resuspended in 500 l of 1% paraformaldehyde-PBS solution prior to being analyzed on a FACSort flow cytometer (Becton Dickinson Immunocytochemistry Systems, San Jose, CA). Rabbit anti-mouse FITCconjugated antibodies were used as the negative control for the M414-FITC, and goat anti-rabbit FITC-conjugated antibodies were used as a negative control for the hENT1-(55-64) antibodies. The synthetic peptide (hENT1-(55-64)) was prepared and conjugated to keyhole limpet hemocyanin (Alberta Peptide Institute, Edmonton, Canada) from an analysis of the hENT1 sequence with an algorithm (Surface Plot) that predicts exposed, potentially immunogenic residues. The antibodies that recognized hENT1-(55-64) were purified by first passing rabbit sera through a Protein G (Amersham Pharmacia Biotech Inc.) column to collect the IgG fraction, followed by dot blot affinity purification using the synthetic peptide. The specificity of the hENT1-(55-64) antibodies was assessed by comparing fluorescence intensity of hENT1-containing cells in the presence and absence of excess hENT1-(55-64).

RESULTS
BeWo Cells Express hENT1 and hENT2 mRNA-BeWo cells have previously been shown (19) to exhibit es and ei functional transport activity. Since hENT1 and hENT2 were initially identified by molecular cloning of a placental cDNA library (10,16) and the BeWo cell line is of placental origin (18), BeWo RNA was examined for evidence of expression of hENT1 and hENT2 mRNA. RNA blot analysis at high stringency demonstrated a single band of 2.4 or 2.6 kilobase pairs with either the hENT1 or hENT2 probe, respectively (data not shown). The identity of the two bands was subsequently confirmed by RT-PCR amplification (Fig. 1). Primers specific to hENT1, which were designed to amplify a 385-base pair DNA fragment, generated identically sized bands from a hENT1-containing plasmid (control) and from cDNA prepared by reverse transcription of BeWo poly(A) ϩ RNA. Likewise, hENT2-specific primers, which were designed to amplify a 245-base pair DNA fragment, generated PCR products of the expected size from a hENT2containing plasmid (control) and from BeWo cDNA. Control reactions, in which either (i) a reverse transcriptase-negative preparation was used or (ii) template cDNA was omitted from the PCR mixtures, were negative. Direct sequencing of the PCR products shown in Fig. 1 confirmed that they were derived from hENT1 and hENT2 mRNA (data not shown). These results suggested that the es and ei nucleoside transport activities of BeWo cells were mediated by the hENT1 and hENT2 gene products.
NBMPR Binding to BeWo Membrane Fractions-Because [ 3 H]NBMPR binds specifically and with high affinity to functional nucleoside transporters, equilibrium binding analysis has been used extensively as a surrogate marker for the presence of the es transporter protein (1,8,32,33). BeWo membrane fractions collected at different sucrose concentrations were first assayed for NBMPR binding activity at a single concentration (5 nM) under conditions previously shown in studies with intact cells (19) to be sufficient to label over 40% of the high affinity sites. Crude membranes (the "postnuclear" supernatant) prepared from ϳ1 ϫ 10 9 cells bound 6.5 pmol of NBMPR/mg of protein when incubated with 5 nM NBMPR for 45 min at room temperature. The BeWo membrane fractions collected at different sucrose concentrations after density gradient centrifugation had different specific binding activities (data not shown), with the maximum activity observed in the fraction isolated from the 25% sucrose layer (26 pmol/mg of protein). Thus, there was an approximate 4-fold increase in binding activity relative to that observed in the crude membranes, and the protein recovery in this fraction amounted to roughly 40% of the total protein present in the various mem- brane fractions. Subsequent experiments with "plasma membranes" utilized the membranes collected from the 25% sucrose layer, since this fraction was enriched 4 -5-fold in plasma membranes relative to crude membranes when assayed for the activities of the plasma membrane marker enzymes alkaline phosphatase and 5Ј-nucleotidase (Table I).
Crude membranes and plasma membranes were subjected to detailed analyses of NBMPR binding to quantify the number of binding sites and their relative affinities for NBMPR (see Fig.  2 for results of a typical experiment with each preparation and Table II for a summary). Since the Scatchard plots were linear (Fig. 2, insets), a one-site binding model was used to estimate K d and B max values. Plasma membranes (Fig. 2B, Table II) exhibited K d and B max values of 0.9 Ϯ 0.2 nM and 75.0 Ϯ 5 pmol/mg of protein, respectively (mean Ϯ S.E., n ϭ 4). The K d value was essentially the same as that determined previously (0.6 nM) for the high affinity binding sites in intact BeWo cells (19), and there was no indication of a second set of lower affinity binding sites. There was a nearly 4-fold increase in specific binding activity relative to that of the crude membrane preparations ( Fig. 2A; Table II), indicating, since high affinity binding of NBMPR represents a specific interaction with the es transporter, that density gradient centrifugation in 25% sucrose yielded a plasma membrane fraction that was enriched in es transporter protein.
NBMPR Binding to Detergent-solubilized Plasma Membranes and Proteoliposomes-Achieving efficient solubilization while maintaining functional integrity is a prerequisite for reconstitution studies, since the solubilized proteins will be used in making proteoliposomes. Solubilization of [ 3 H]NBMPR binding activity from the 25% sucrose fraction was achieved using 1% OCTG, and the concentration of membrane protein was maintained at ϳ200 g/ml. Under these conditions, nearly 70% of total protein was solubilized. The resulting mixed micelles exhibited a single class of high affinity binding sites (K d , 3.5 nM) and retained the same number of NBMPR-binding sites/mg of protein as the nonsolubilized plasma membrane preparations (Table II). The slight decrease in affinity for [ 3 H]NBMPR after solubilization with OCTG could have been due to the presence of the detergent; similar reductions in binding affinity have been observed with micellar preparations of solubilized murine es transporters (23).
Proteoliposomes prepared from solubilized plasma membranes also exhibited only one class of NBMPR-binding sites (Table II). The proteoliposomes bound 35 pmol of [ 3 H]NBMPR/ mg of protein with an affinity (K d , 1.1 nM) that was identical to that observed for binding of NBMPR to unsolubilized plasma membranes. The efficiency of incorporation of NBMPR-binding sites from the plasma membranes into proteoliposomes was approximately 45% based on a comparison of the observed B max values for NBMPR binding.
NBMPR Binding to Nuclei-In an earlier study conducted with intact BeWo cells (19), the slow time courses of association of NBMPR and the presence of two classes of high affinity binding sites were interpreted as evidence for intracellular NBMPR-binding sites. In the present study, only one class of sites (K d ϳ 1 nM) was observed with the various membrane fractions. It was possible that the second class of sites (K d ϭ 14.5 nM) seen previously in intact cells was associated with internal components (e.g. nuclei) that were discarded in the preparation of membranes. Thus, intact nuclei and nuclear envelopes were isolated from BeWo cells for analysis of sitespecific binding of [ 3 H]NBMPR.
The nuclei and nuclear membranes were first tested for the presence of plasma membranes and organelles such as mitochondria and lysosomes, since these organelles have been reported previously (36, 37) to exhibit either NBMPR binding or  nucleoside transport activities. The nuclei and the nuclear envelope preparations contained relatively few contaminating plasma membranes (2-3%), as indicated by the low activities of marker enzymes relative to those observed in crude and plasma membrane fractions ( Table I). The nuclear preparations were not contaminated with mitochondria and lysosomes, based on the absence of (Ͻ1%) detectable succinate-INT reductase and ␤-D-galactosidase activity, respectively. For comparison, the crude membranes had an approximate mitochondrial content of 9 Ϯ 1% and lysosomal content of 7 Ϯ 1%.
The nuclear preparations were analyzed by flow cytometry using murine monoclonal antibodies (M414-FITC) directed against a related family of nuclear pore complex proteins (38) and polyclonal antibodies directed against a synthetic peptide derived from the predicted extracellular loop between transmembrane domains 1 and 2 of hENT1 (Fig. 3). The preparations consisted almost completely of nuclei, since 99% of the population reacted with the M414-FITC antibodies, with a substantial increase in fluorescence intensity relative to that of nuclei stained with FITC-conjugated antibodies against murine IgG (the negative control). When polyclonal antibodies directed against hENT1 were used, 92% of the nuclei reacted positively, suggesting that most nuclei possessed immunoepitope(s) recognized by the hENT1-(55-64) antibodies (Fig.  3B). These results, although consistent with the presence of the hENT1 protein in nuclei, could also reflect the presence of cross-reactivity with other proteins.
Equilibrium analysis of site-specific binding of [ 3 H]NBMPR to nuclei and nuclear membranes was undertaken in the experiments of Fig. 4 (see also Table II). The nuclei bound 7.0 pmol of [ 3 H]NBMPR/mg of protein with an affinity (K d ϭ 0.7 nM) that was virtually identical to that observed with plasma membranes enriched in NBMPR binding activity. The NBMPR binding capacity of the nuclear envelopes was nearly one-fifth of the binding capacity of plasma membranes in the 25% sucrose fractions. The observed NBMPR binding activities associated with the nuclei and the nuclear envelopes were not due to plasma membrane contamination, since the alkaline phosphatase and the 5Ј-nucleotidase activities were only 2-3% of that observed with the plasma membranes recovered from the 25% sucrose fractions (Table I).
[ 3 (1), and the passive permeability of liposomal membranes to uridine is lower than that for adenosine (39). The representative experiment shown in Fig. 6 examined the initial rates of uptake of 20 M [ 3 H]uridine by proteoliposomes prepared from the plasma membrane-enriched fractions. The value (mean Ϯ S.E.) from six separate experiments for the uninhibited initial rate was 55.0 Ϯ 5.0 pmol/mg of protein/s, and preincubation of proteoliposome with a mixture of nucleoside transport inhibitors (adenosine, NBTGR, dilazep) decreased this rate to 10.0 Ϯ 2.5 pmol/mg of protein/s, or 27% of total uptake. The mediated component of uridine uptake, which was obtained by subtracting the inhibited value from the total value, was 45.0 Ϯ 5.0 pmol/mg of protein/s. The presence of NBMPR binding activity in nuclear membrane fractions and of the hENT1 immunoepitope on intact nuclei suggested that nuclear membranes and/or the associ-  Table I. Binding assays were conducted at room temperature using 45-min incubations with a range of concentrations (0.24 -24 nM) of [ 3 H]NBMPR. Specific binding was calculated as the difference between total binding and binding observed in the presence of 10 M NBMPR and 10 M dilazep. B max and K d values (mean Ϯ S.E., n ϭ 4) were determined by mass law analysis (Scatchard plot) of data obtained from binding assays. ated endoplasmic reticulum may contain functional es transporters. Reconstituted proteoliposomes were prepared from solubilized nuclear membrane fractions and tested for their ability to transport [ 3 H]uridine in the presence and absence of a mixture of nucleoside transport inhibitors (Fig. 7A). The values (mean Ϯ S.E.) from three independent experiments for the uninhibited and inhibited initial rates were 11.0 Ϯ 2.0 and 3.5 Ϯ 1.0 pmol/mg of protein/s. The mediated component of uridine uptake, which was obtained by subtracting the inhibited value from the total value, was 7.5 Ϯ 1.0 pmol/mg of protein/s. To determine if the mediated component was composed of both es and ei activities, NBMPR was tested for its ability to inhibit the transport-mediated influx of 20 M [ 3 H]uridine in proteoliposomes prepared from the nuclear envelopes (Fig. 7B). Cellular accumulation of [ 3 H]uridine was inhibited by about 85% with 60 nM NBMPR, and the remaining component (about 15%) was resistant to 5 M NBMPR. These results indicated the operation of both es and ei NT processes in proteoliposomes prepared from nuclear envelopes.

DISCUSSION
Cultured BeWo cells exhibit high levels of es and ei transport activities and unusually large numbers (Ͼ10 7 /cell) of NBMPRbinding sites (19). Because of these features, BeWo cells have been a useful model system for analysis of relationships between NBMPR binding and es-mediated transport capability. In this work, we have established that BeWo cells possess both hENT1 and hENT2 mRNA, and we conclude that the transporter proteins responsible for mediating the high levels of es and ei activity previously reported (19) in BeWo cells are hENT1 (es) and hENT2 (ei).
NBMPR is a tightly binding and specific inhibitor of equilibrative transport of nucleosides in mammalian cells and, in its radioactive form, has been used to identify transporter proteins in several cell lines (23,40,41). NBMPR-binding proteins purified from human erythrocytes exhibit both uridine transport and NBMPR-binding activities when reconstituted in unilamellar phospholipid vesicles (20), leading to the conclusion that NBMPR-binding proteins represent es transporters. NBMPRbinding site abundance has frequently been taken as a measure of functional es transporters, since the total number of NBMPR-binding sites was shown to be proportional to V max values for uridine influx in erythrocytes of several mammalian species (42). However, earlier studies with BeWo cells (19), which possess an elevated number of NBMPR-binding sites and high thymidine transport activity, revealed (i) two classes of high affinity NBMPR-binding sites (K d ϭ 0.6 and 14.5 nM) and (ii) a lack of proportionality between NBMPR binding site densities and thymidine transport activity when compared with other cell types. These observations, which suggested that BeWo cells may possess NBMPR-binding proteins that are not  functional es transporters, were the basis of the current investigation.
We first examined binding of 5 nM [ 3 H]NBMPR to crude BeWo membranes and membrane fractions collected at different sucrose concentrations after density gradient centrifugation. The sucrose density fractions had different NBMPR binding activities, with the maximum activity observed with fractions obtained from the 25% sucrose layer. Equilibrium binding analysis with the latter fraction indicated a nearly 4-fold increase in specific binding activity (B max ) over that observed with crude membranes. Only one class of NBMPRbinding sites were observed in the crude membrane-containing and plasma membrane-containing 25% sucrose fractions. NBMPR is lipophilic and diffuses into cells, where it could interact with intracellular binding sites other than those associated with the various membrane preparations. For this reason, the nuclei, which had been removed and discarded at the initial stage of membrane fractionation, were retained and tested for NBMPR-binding sites. Intact nuclei and nuclear envelopes bound NBMPR with high affinity, although only one class of sites was detected. The second class of sites was evidently lost during disruption of cells. We speculate that the second class of binding sites previously observed in intact BeWo cells (19) represented functional es transporters associated with intracellular membranes for which the binding constants were incorrectly estimated by assuming that the intracellular concentration of free NBMPR was equal to that of the extracellular compartment. The second class of NBMPR-bind-ing sites observed in BeWo cells in the earlier study (19) was most likely an artifact for the aforementioned reasons.
The es and ei transporter proteins have been defined by their functional activities in plasma membranes (1)(2)(3). However, our results indicated that nuclear membranes and/or the endoplasmic reticulum contain functional es and ei transporter proteins. Other transporter proteins known primarily for their plasma membrane functions have also been observed in nuclear and/or endoplasmic reticulum membranes (43)(44), and the cystic fibrosis transmembrane conductance regulator, a plasma membrane-localized chloride channel, has been shown to exhibit functional activity in the endoplasmic reticulum (45).
BeWo nuclei and nuclear envelopes used in the present study were not contaminated to any significant level by plasma membranes, mitochondria, or lysosomes as revealed by marker enzyme assays. Mitochondria, which appear to have the capacity for uptake of nucleosides (3), have low numbers of NBMPRbinding sites (B max Ͻ 2 pmol/mg of protein) (36), whereas nuclear envelopes used in the present study exhibited a B max value of 13 pmol/mg of protein. Lysosomes have high levels of nucleoside transport activity but with low sensitivity (25 M) to inhibition by 3 NBMPR or dipyridamole (37). The proteoliposomes reconstituted from BeWo nuclear membranes were inhibited by 3 nM NBMPR (Fig. 7B).
While the ability of NT proteins to bind [ 3 H]NBMPR with high affinity is a well established indicator of functionality, the ultimate test is the ability to mediate transmembrane fluxes of [ 3 H]nucleosides. Thus, OCTG-solubilized plasma membrane preparations were incorporated into liposomal membranes, and the resulting proteoliposomes were shown to accumulate [ 3 H]uridine by mediated processes. [ 3 H]Uridine uptake by the reconstituted proteoliposomes was blocked by nucleoside transport inhibitors, with a rank order of potency of NBMPR Ͼ dilazep Ͼ dipyridamole (data not shown). The high level of NBMPR binding activity associated with nuclear envelopes corresponded to approximately 18% of the total membrane pool of NBMPR-binding sites and evidently represented functional es transporters, since NBMPR-sensitive uridine transport was observed in proteoliposomes prepared from detergent-solubilized nuclear envelopes. The observed initial rate of uridine influx in the latter preparations was 75 pmol/mg of protein/10 s, whereas the rate obtained with the proteoliposomes from solubilized plasma membranes was 450 pmol/mg of protein/10 s. These values were comparable when normalized for the NT protein content in these membranes as revealed by B max values for NBMPR-binding capacities (13 and 75 pmol/mg of protein, respectively).
BeWo cells possess both es and ei transporters, as defined by their sensitivities to NBMPR and the presence of both hENT1 and hENT2 mRNA. Approximately 15% of the mediated influx of [ 3 H]thymidine in intact BeWo cells was attributed to the ei process (19). Reconstitution of protein solubilized from the crude, unfractionated BeWo membranes resulted in both esand ei-mediated [ 3 H]uridine uptake by proteoliposomes, with the ei-mediated process representing about 18% of the total activity. Proteoliposomes prepared from nuclear envelopes accumulated [ 3 H]uridine by both es and ei nucleoside transporters, with the ei-mediated process amounting to about 15% of the total activity, suggesting the presence of both hENT1 and hENT2 proteins in the nuclear-envelope fractions.
In conclusion, we have demonstrated that BeWo cells express hENT1 and hENT2 mRNA, suggesting that es and ei transport in BeWo cells is mediated by the hENT1 and hENT2 proteins. We achieved the solubilization of nearly 100% of the teins retained the NBMPR binding characteristics observed in studies with isolated membranes. Reconstituted proteoliposomes, which retained approximately 45% of NBMPR binding activity, were capable of accumulating [ 3 H]uridine by processes that were inhibited by NBMPR and dipyridamole with characteristics that indicated the presence of both es and ei-mediated transport processes. hENT1 immunoreactivity, NBMPR binding activity, and both es-and ei-mediated transport activities were observed in the nuclear envelope preparations, providing the first indication that the ENT transporters are associated with intracellular membranes in a functional state.
Lysosomes and mitochondria exhibit nucleoside transport activity (3,36,37), and our finding of es and ei activities in nuclear envelope preparations, which are comprised of nuclear membranes and endoplasmic reticulum, provides further evidence for the occurrence of nucleoside transporters in intracellular membranes of mammalian cells. The finding that both hENT1 and hENT2 are functionally active when isolated from nuclear envelope preparations raises two possibilities. Endoplasmic reticulum contains many enzymes of nucleotide metabolism, and it is possible that hENT1 and hENT2, which are bidirectional exchangers, may play a role in translocation of nucleosides between the cytosolic and luminal compartments. It is also possible that the nuclear envelope-associated hENT1 and hENT2 proteins serve as a reserve pool for translocation to the plasma membrane when additional transport capability is required. Studies are in progress to determine turnover rates of hENT1 and hENT2 in cells under different growth conditions and to determine more precisely the subcellular location of the hENT1 and hENT2 proteins.