Equilibrative Nucleoside Transporters of Arabidopsis thaliana cDNA CLONING, EXPRESSION PATTERN, AND ANALYSIS OF TRANSPORT ACTIVITIES*

Equilibrative nucleoside transporters (ENTs) occur in diverse organisms. In the model plant Arabidopsis thaliana , eight potential ENTs (AtENTs) have been predicted by genome sequencing. We here report the cloning of the cDNAs for AtENTs 2, 3, 4, 6, 7, and 8. Conceptual trans-lation of the cDNAs of AtENTs 2, 3, 4, 6, 7, and 8 yielded polypeptides possessing strong similarities to ENTs characterized previously. Eleven putative transmembrane domains were identified in each of the six AtENTs. In suspension cells, the transcription of AtENTs 1, 3, 4, 6, and 8 was increased by two treatments (nitrogen deprivation, application of 5-fluorouracil and methotrexate) that inhibited the de novo pathway of nucleotide synthesis, indicating that multiple members of the Arabidopsis ENT family may function in the salvage pathway of nucleotide synthesis. Except for AtENT1, the transcription of the remaining six AtENTs showed varying degrees of organ specificity. However, all seven AtENTs were expressed in the leaf and flower. In plant, insect, and yeast cells, ectopically expressed AtENT3 was targeted to the plasma membrane. AtENT3 expressed in yeast cells transported adenosine and uridine with high affinity. Furthermore, the activities of AtENT3 appear not to require a transmembrane proton gradient because protonophores did not abolish adenosine or uridine transport. In into epidermal cells using a bombardment Goring The and photographed using confocal microscopy. In some experiments, onion epidermal strips were left for 15 min on filter papers soaked with 0.8 M mannitol before being examined. This osmotic shock treatment caused a partial separation of the plasma membrane from the cell wall, which gave a higher resolution image of the association of the AtENT3:GFP fusion protein with plasma membrane.

port activities are important to the salvage pathway of nucleotide synthesis (1)(2)(3)(4)(5)(6). According to the mechanism of transport, nucleoside transporters have been divided into two families. The equilibrative nucleoside transporters (ENTs) 1 generally transport nucleosides down their concentration gradients (5)(6)(7); against this generality is the finding that several ENTs recently characterized from kinetoplastid protozoans and the higher plant Arabidopsis thaliana are probable proton symporters (8 -13). The concentrative nucleoside transporters (CNTs) catalyze the transport of nucleosides against their concentration gradients and are either sodium or proton symporters (5,6,14). Topologically, ENTs studied thus far have all been predicted to possess 11 transmembrane helices (TMs) (5,15,16). In the studies on a human ENT (hENT1), evidence has been obtained indicating that the amino-and carboxyl-terminal regions are located on the cytoplasmic and extracellular faces of the membrane, respectively (16). The regions linking the TMs are all hydrophilic but differ considerably in their size. The large loop between TMs 1 and 2 is composed of ϳ41 residues and is glycosylated and extracellular (5,7,16). The second large loop (between TMs 6 and 7) contains ϳ66 amino acid residues and is cytoplasmic (5,16). From amino acid sequence comparisons, it has been suggested that ENTs from other organisms (including insects, nematodes, protozoa, yeasts, and higher plants) may adopt a topology similar to that of hENT1 (16).
The biochemical properties of several ENTs have been investigated extensively. In general, ENTs are broadly selective in substrate specificity (5,6). However, their transport processes are differentially inhibited by nucleoside analogues (such as nitrobenzylmercaptopurine ribonucleoside (NBMPR)) or vasodilator drugs (e.g. dipyridamole, dilazep, and draflazine). For example, the es type ENTs are sensitive to inhibition by NBMPR, whereas the ei type ENTs are not (1,2,17). Whereas the es type transporter from humans is potently inhibited by vasodilator drugs, the corresponding transporters from rodents and the ei type transporters from both humans and rodents are much less sensitive to such drugs (6, 18 -22). Through molecular studies on mammalian ENTs, a region encompassing TMs 1-6 has been deduced to be involved in substrate/inhibitor binding (23,24). More recently, it has been found that the replacement of a glycine residue (residue 154) in TM 4 of hENT1 by serine results in reduced NBMPR binding and that residue 33 in TM 1 of hENTs 1 and 2 is an important determinant of vasodilator drug sensitivity (25,26). The mutation of a single glycine residue (residue 179) in TM 5 alters both nucleoside transport activity and sensitivity to NBMPR in hENT1 (27).
In addition to playing a key role in the salvage pathway of nucleotide synthesis, mammalian ENTs have been found to affect many physiological processes including neurotransmission and cardiovascular activity through the regulation of adenosine concentrations (1)(2)(3)(4)(5)(6). Clinically, ENTs are targets of vasodilator drugs as well as routes for uptake of nucleoside drugs used in treating cancers and viral infections (3). Mammalian ENTs are found in a variety of tissues and cells, and evidence for the presence of more than one type of ENTs in the same tissue has been obtained (28,29).
Compared with mammals and parasites, progress in understanding the expression, structure, and function of higher plant ENTs has been slow. Although nucleoside transport processes were first recorded in plants more than 20 years ago and have since been discovered in several cell types (30 -32), the corresponding nucleoside transporters have not been identified. By taking advantage of the genetic information from the genome sequencing project, researchers have recently started molecular and biochemical studies of the ENTs encoded by the model plant A. thaliana (13,33). A total of eight genes encoding potential ENTs have been predicted in Arabidopsis (5). Their putative products (designated as AtENT1 to -8) form an independent branch in the phylogenetic tree constructed using ENTs from mammals, insects, nematodes, protozoa, yeasts, and plants (5). The gene encoding AtENT1 is expressed constitutively and in multiple tissues (33). Functional expression of AtENT1 in yeast cells has also been achieved (13). It was shown that, in intact yeast cells, AtENT1 catalyzed a protonlinked, high affinity adenosine transport and that the transport mediated by AtENT1 was resistant to inhibition by the nucleoside analog NBMPR and by the vasodilator drugs dilazep and dipyridamole (13). In addition, it was found that uridine, which is commonly transported by various types of mammalian ENTs, might not be significantly transported by AtENT1 (13). These results suggest that at least some (if not all) of the AtENTs predicted by genome sequencing are likely to be functional proteins. In this context, it would be important to study all members of the Arabidopsis ENT family in order to gain a more comprehensive understanding of the function of higher plant ENTs.
In protozoan parasites and some mammalian cell types, the importance of ENTs in the salvage pathway of nucleotide synthesis is indicated by the lack of the de novo pathway of nucleotide synthesis (4, 34 -36). In higher plants, evidence for the participation of ENTs in the salvage pathway has mainly come from observations that exogenously supplied nucleosides (such as [ 3 H]thymidine) and analogs (for example, bromodeoxyuridine) can be taken up and incorporated into DNA synthesis (37)(38)(39)(40). So far there have been no investigations on whether the transcription of ENT genes or the activities of ENT proteins would change in response to inhibition of the de novo pathway of nucleotide synthesis in plant cells. Experiments of this kind are essential for not only providing molecular evidence on the function of ENTs in the salvage pathway of nucleotide synthesis but also yielding insights into the relationship between the two pathways of nucleotide synthesis in plant cells. Based on the above discussions, we decided to conduct a more comprehensive study of the ENTs of A. thaliana predicted by genome sequencing. Below, we report the cloning of the cDNAs for AtENTs 2, 3, 4, 6, 7, and 8, comparative analysis of the amino acid sequences of AtENTs with those from other organisms, regulation of AtENT transcription by alterations in the de novo pathway of nucleotide synthesis, transcription patterns of seven AtENTs in various Arabidopsis organs, and functional expression of AtENT3 in yeast cells. The potential role of ENTs in A. thaliana is discussed in light of our results and those published previously.

Cloning of AtENT cDNAs and Comparative Analysis of Amino Acid
Sequences of AtENTs-The Col-1 ecotype of A. thaliana was used throughout this study. To amplify cDNAs for the complete coding regions of AtENTs 2, 3, 4, 6, 7, and 8 using RT-PCR, oligonucleotide primers (Table I) were synthesized according to genomic sequence information generated by the Arabidopsis genome initiative (AGI). Total RNA samples were prepared from suspension cells, leaves, stems, roots, flowers, or immature siliques of A. thaliana as described previously (41). RT-PCR experiments using appropriate primers and enzymes were conducted following published protocols (41). The expected cDNA fragments were cloned into the pGEM®-T Easy plasmid vector (Promega). The inserts in the resultant positive clones were sequenced commercially (TaKaRa) from both strands. The obtained cDNA sequences were compared with those of predicted AtENT coding regions using various programs at the NCBI Web site (www.ncbi.nlm.nih.gov).
For calculating amino acid sequence identities using the software DNAstar (DNAstar Inc), the amino acid sequences of AtENTs 2, 3, 4, 6, 7, and 8 were deduced from cloned cDNAs and compared with those of AtENT1 and ENTs from other sources for which nucleoside transport activities have been demonstrated (Table II) (6). To investigate phylogenetic relationships of AtENTs to ENTs from other sources, the amino acid sequences were aligned using the ClustalW program at the EBI Web site (www.ebi.ac.uk/clustalw/index.html). The aligned sequences were converted into MEGA format, which was subsequently employed for constructing phylogenetic trees at the MEGA 2 Web site (www. megasoftware.net) (42). The putative transmembrane domains of AtENTs were predicted using the HMMTOP program (available on the World Wide Web at www.enzim.hu/hmmtop/server/hmmtop.cgi) (43) with manual adjustment aided by the comparison of the amino acid sequences of AtENTs with those of hENTs 1 and 2 (for which there is currently more structural information available).
Transcriptional Responses of AtENTs to Nitrogen Deprivation and Drug Treatment in Arabidopsis Suspension Cells Using Semiquantitative RT-PCR-For investigating transcriptional responses of AtENTs to nitrogen deprivation or drug treatment, a suspension cell culture of A. thaliana was initiated and maintained as previously described (41,44). Suspension cells grown in nitrogen-sufficient liquid medium (containing 25 mM KNO 3 and 1 mM (NH 4 ) 2 SO 4 ) were collected, washed, and divided into two batches. Both batches of cells were resuspended in nitrogen-deficient medium (devoid of any NO 3 Ϫ and NH 4 ϩ ). For the first batch of cells, a sample was immediately taken as the day 0 control. Further samples were taken 1, 3, and 5 days later. Control samples were prepared at identical time points from cells cultured in nitrogensufficient medium. The second batch of cells, which had been in nitrogen-deficient medium for 5 days, was collected, washed, and dispersed into nitrogen sufficient medium. Cell samples were then taken at 1, 3, and 5 days after nitrogen resumption. The strategies for preparing suspension cells for drug treatment and sample collection were essentially similar to those described above except that nitrogen-sufficient medium was used. The two drugs employed were 5-fluorouracil (2 M, diluted from a 250 mM solution prepared in 1 M NH 4 OH) and methotrexate (1 M, diluted from a 50 mM solution prepared in 1 M NH 4 OH). Control samples were prepared from suspension cells cultured in the liquid medium lacking the two drugs. Total RNA was extracted from the harvested cell samples and was converted to cDNA by reverse transcription (41). The cDNA contents in all reverse transcription mixtures were normalized by amplifying the transcripts of tubulin using the primers TuF and TuR (Table I) (41). Evaluation of the transcript levels of the seven AtENTs in the different cell samples by PCR using the normalized cDNA mixtures and gene-specific primers (Table I) was then carried out as described previously (41).
Evaluation of Organ Specificities of AtENTs Transcription in A. thaliana-Root, stem, leaf, flower, and immature silique samples were prepared from 7-week-old A. thaliana plants grown under normal physiological conditions. Total RNA was extracted from the five samples and was transcribed into cDNA (41). The cDNA contents in the reverse transcription mixtures were normalized as described above. The transcript levels of the seven AtENTs in the five organs were evaluated by PCR using the normalized cDNA samples and genespecific primers (Table I).
Localization of AtENT3 Expressed in Insect, Yeast, or Plant Cells-For all DNA cloning experiments in this study, the methods described by Sambrook et al. (45) were generally followed. The correctness of each cloning step was verified by restriction enzyme digestion and sequencing through the cloning junction(s). To facilitate the construction of an AtENT3:GFP fusion cistron for insect or yeast cell expression, the vector pFB-GFP was created by ligating a BglII/NotI fragment, which contains the EGFP open reading frame and several restriction sites (SalI, ApaI, BamHI, and AgeI) from the pEGFP-1N plasmid (Clontech), to BamHI/NotI-digested pFastBac vector (Invitrogen). The unique SalI site upstream of the EGFP open reading frame in pFB-GFP was then employed for accepting the AtENT3 open reading frame that was amplified by PCR using primers 3F and 3R (Table I) and the high fidelity polymerase ExTaq (TaKaRa). The resultant construct pFB-AtENT3: EGFP and the control construct (pFB-GFP) were expressed in insect cells using the Bac-to-Bac TM baculovirus expression system (Invitrogen) following the instructions provided by the supplier. The expressing cells were examined using a confocal microscope (FV500; Olympus) at 48 h post-transfection. Optical sections were prepared for selected cells at a thickness of 0.35 m. For yeast cell expression, the construct pY-AtENT3:EGFP was created by ligating an EcoRI/XbaI fragment of AtENT3:EGFP (derived from pFB-AtENT3:EGFP) to the pYES2 vector (Invitrogen) linearized with the same two enzymes. A control construct, pYEGFP, was prepared by removing the AtENT3 coding sequence from pYAtENT3:EGFP using SalI digestion and self-ligation of the resultant plasmid backbone. Induction of the expression of pYAtENT3:EGFP and pYEGFP in yeast cells was carried out following the instructions by Invitrogen. Expressing cells were examined using confocal microscopy. For transient expression in plant cells, the construct p35S-AtENT3: GFP was created by ligating the SalI fragment of AtENT3 (see above) into the SalI site upstream of the coding sequence of GFP in the vector p35S-GFP (provided by Dr. Jinsong Zhang, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences). Plasmid DNA of p35S-AtENT3:GFP or p35S-GFP was introduced into onion epidermal cells using a particle bombardment method as described by Silva and Goring (46). The expressing cells were examined and photographed using confocal microscopy. In some experiments, the bombarded onion epidermal strips were left for 15 min on filter papers soaked with 0.8 M mannitol before being examined. This osmotic shock treatment caused a partial separation of the plasma membrane from the cell wall, which gave a higher resolution image of the association of the AtENT3:GFP fusion protein with the plasma membrane.
Adenosine Transport of AtENT3 Expressed in the BMA64 -1A ade2 Strain of Saccharomyces cerevisiae-Under normal conditions, S. cerevisiae cells do not take up adenosine because of the lack of an endogenous transport system for the nucleoside (11). An ectopically expressed nucleoside transporter may enable S. cerevisiae to take up adenosine if it is functional (11,13). Previous investigators have shown that the W303 ade2 strain of S. cerevisiae (W303, Mat ␣, leu2-3, leu2-112, trp1-1, ura3-1, his3-11, his3-15, ade2-1, can1-100), which is defective in adenine biosynthesis and forms red colonies on medium lacking adenine, can be employed for functional expression of AtENT1 (13). However, when we attempted to use this strain for functional expression of AtENT3, many white colonies formed on medium containing adenosine. To avoid potential difficulties in using the W303 ade2 strain, we chose to perform functional expression of AtENTs in an alternative ade2 strain (BMA64 -1A, Mat ␣, ura3-52, trp1⌬2, leu2-3 -112, his3-11, ade2-1, can1-100, purchased from Euroscarf). To verify the suitability of the BMA64 -1A ade2 strain, two AtENT1 constructs, pYAtENT1 and pYGFP:AtENT1, were prepared and expressed. To prepare pYAtENT1, a BamHI/EcoRI fragment of the AtENT1 open reading frame (33) was cloned into the pYES2 vector that had previously been cut with the same two enzymes. pYGFP:AtENT1 was constructed by inserting a BamHI fragment of GFP (33) into the BamHI site of pYAtENT1. Two constructs, pYAtENT3 and pYAtENT3:EGFP, were used for functional expression of AtENT3 in the BMA64 -1A ade2 strain. The preparation of pYAtENT3:EGFP was described above. pYAtENT3 was constructed by cloning the complete coding region of AtENT3, which was amplified by PCR using primers 3F and 3RS (Table I) and ExTaq, into the XhoI site of pYES2. The four expression constructs (pYAtENT1, pYGFP: AtENT1, pYAtENT3, and pYAtENT3:EGFP) and the control plasmid pYES2 were each transformed into the cells of the BMA64 -1A ade2 strain. For plate assays, the cells were grown on three types of medium: minimal medium containing full complement of nutrients (as described by Invitrogen), minimal medium lacking uracil, and minimal medium lacking both uracil and adenine but containing 150 M adenosine (Sigma).
The yeast cells expressing pYAtENT1 or pYAtENT3 were used for uptake experiments. Cells were cultured to an A600 of 0.7-1.5 in liquid medium containing galactose (as described by Invitrogen), collected by centrifugation, and washed twice with 25 mM phosphate buffer (pH 6.0). The washed cells were resuspended to an A 600 of 3 in 25 mM phosphate buffer and were kept on ice. For uptake assays, 100 l of cells were gently mixed with an equal volume of transport medium (25 mM phosphate buffer, pH 6.0) containing the desired concentrations of [ 3 H]adenosine (Amersham Biosciences). The mixture was layered on top of 200 l of oil in an Eppendorf tube and was kept at 25°C for the required time (47). Uptake was terminated by a 2-min centrifugation at 12,000 ϫ g. The supernatant (containing unincorporated [ 3 H]adenosine) was removed by aspiration. The pelleted cells were washed twice with 25 mM phosphate buffer and were solubilized with 5% Triton X-100 for measuring the level of [ 3 H]adenosine uptake in a scintillation counter (MicroBeta Trilux; PerkinElmer Life Sciences). [ 3 H]Adenosine uptake rate was expressed as pmol/mg of yeast protein. The quantification of yeast protein was carried out using a Bio-Rad protein assay kit (Bio-Rad). Kinetic parameters (K m , V max ) were calculated by nonlinear regression using the SigmaPlot 2000 software (SPSS Inc., Chicago, IL). For investigating the sensitivity of AtENT3-mediated transport to inhibitors, vasodilator drug (dilazep or dipyridamole; Sigma), nucleoside analog (NBMPR; Sigma), or protonophore (CCCP or DNP; Sigma) of desired concentrations were added to the yeast cell suspension (in transport medium) 30 min before the addition of [ 3 H]adenosine and the initiation of the uptake assay (26).
Uridine Transport of AtENT3 Expressed in the fui1 Strain of S. cerevisiae-In S. cerevisiae, Fui1p is a cell surface uridine transporter (47). The mutation of its coding gene, fui1, could eliminate uridine influx into the yeast cells (47). Based on this property, a fui1 knock-out strain, fui1:TRP1, has been used successfully for studying uridine transport activities of hENT 1 and 2 (26,48). To investigate potential uridine transport activities of AtENT3, we expressed AtENT3 cDNA in a similar fui1 knock out strain (BY4741, Mat ␣, his3⌬1, leu2⌬0, met15⌬0, ura3⌬0, YBL042C:KanMX4, purchased from Euroscarf). The coding sequence of AtENT3 was amplified by PCR using primers 3F2 and 3R2 (Table I). The amplified fragment was digested with the restriction enzymes EcoRI and NotI, followed by cloning into the p181AINE vector (49) that had been treated with the same two enzymes. The DNA of the resultant plasmid p181AtENT3 was used to transform fui1 cells. The uptake of [ 3 H]uridine (Amersham Biosciences) by AtENT3 expressing fui1 cells in the absence or presence of inhibitors (dilazep, dipyridamole, NBMPR, CCCP, or DNP) was examined as described above.

RESULTS
cDNA Cloning and Comparative Analysis of the Deduced Amino Acid Sequences of Arabidopsis ENTs-Using sequence information generated from the AGI project, PCR primers (Table I) were synthesized for amplifying the cDNAs of the complete coding regions of seven potential ENTs (AtENTs 2-8) by RT-PCR. To maximize the chance of identifying the cDNAs, total RNA samples were prepared from suspension cells, leaves, stems, roots, flowers, or immature siliques. During the amplification experiments, it was common to find the cDNA for AtENTs 2, 3, 4, 6, 7, or 8. However, a cDNA for AtENT5 was not obtained in repeated trials. Further attempts involving the use of alternative PCR primers, Taq polymerases, and/or cycling conditions did not lead to the identification of the cDNA specific for AtENT5 either (data not shown). Sequencing the obtained cDNAs revealed that, for AtENTs 2, 3, and 4, the coding region sequences derived from cDNAs agreed with those predicted by AGI. In AtENTs 6, 7, and 8, the coding region sequences deduced from cDNAs differed from those predicted by AGI. A comparison of the cDNA sequences, the predicted coding region sequences, and the genomic sequences of AtENTs 6 -8 showed that the disagreement was caused by imprecise prediction of the intron and exon boundaries during the annotation process of the AGI project (data not shown).
The amino acid sequences of AtENTs 2, 3, 4, 6, 7, and 8 were deduced from their respective cDNAs and compared with those of AtENT1 and ENTs from other sources (Table II). The identities among the amino acid sequences of the seven AtENTs ranged from 27.0% (between AtENT1 and AtENT4) to 91.1% (between AtENT3 and AtENT6) (Table II). AtENT1 and AtENT8 were more similar to each other (47.6% identical in amino acid sequences) than either to AtENTs 2, 3, 4, 6, and 7 (Table II). AtENTs 2, 3, 4, 6, and 7 were more closely interrelated (at least 50% identical in pairwise comparisons of amino acid sequences) than any of them to AtENTs 1 and 8 (Table II). Amino sequence identities among AtENTs and ENTs from other sources (mammals, Caenorhabditis elegans, S. cerevisiae, parasites) varied from 11.0% (between AtENT2 and Leishmania donovani NT2) to 26.5% (between AtENT8 and rat ENT1) (Table II). In phylogenetic analysis, the seven AtENTs formed an independent cluster, which contained two subgroups, one composed of AtENTs 1 and 8 and the other of AtENTs 2, 3, 4, 6, and 7 ( Fig. 1). Because of the existence of significant levels of identities, the amino acid sequences of the seven AtENTs could be aligned with those of hENT1 and hENT2 (Fig. 2). Aided by the alignment and the computer software HMMTOP, 11 putative transmembrane domains were predicted for AtENTs 2, 3, 4, 6, 7, and 8, respectively (Fig. 2). In the amino acid position corresponding to the residue 33 of hENTs 1 and 2, an Ile was 5Ј-TATGGATCCCTGATGAGCCAGAGCCAACC-3Ј a A, for cDNA amplification; B, for transcript detection by semiquantitative RT-PCR; C, for constructing pYAtENT3 (in conjunction with primer 3F); D, for constructing p181AtENT3.
b The primer combination (5F, 5R) was one of the seven primer sets that were designed and used in attempting to amplify the cDNA, or detecting the transcripts, of At4g05140. c Because of a high level of nucleotide sequence identity between the coding regions of At4g05110 and At4g05120, primers 6F and 3R could be used for amplifying the cDNA of At4g05110. found for AtENTs 1 and 8 and a Leu for AtENTs 2, 3, 4, 6, and 7 (Fig. 2, indicated by an arrow). The glycine residue, which is located at the 179-position of hENT1 and important for the interaction of hENT1 with inhibitors and substrates, was conserved at the corresponding positions of hENT2 and all seven AtENTs (Fig. 2, indicated by inverted filled triangle). However, the glycine residue at the 154-position of hENT1 required for NBMPR binding was not conserved in either hENT2 or the seven AtENTs (Fig. 2, indicated by inverted open triangle).
Expression Patterns of AtENTs-The existence of multiple ENTs in A. thaliana raises the question of whether they are all involved in the function of the salvage pathway of nucleotide synthesis or if their expression possesses organ specificity. To obtain evidence on the function of AtENTs in the salvage pathway of nucleotide synthesis, we attempted to inhibit the de novo pathway of nucleotide synthesis in Arabidopsis suspension cells by nitrogen deprivation or application of fluorouracil and methotrexate, followed by investigating changes in the transcript levels of the different AtENTs using semiquantitative RT-PCR. Both treatments led to clear increases in the transcript levels of AtENTs 1, 3, 4, 6, and 8 (Fig. 3, A and B). Resumption of nitrogen supply or withdrawal of fluorouracil and methotrexate from the growth medium decreased the transcript levels of AtENTs 1, 3, 4, 6, and 8 to those in the control cell samples (data not shown). During these experiments, we were unable to detect the transcripts of AtENTs 2 and 7 in either control or treated suspension cell samples. Nevertheless, the results shown in Fig. 3, A and B, demonstrated that the transcription of multiple AtENTs was elevated in response to the inhibition of the de novo pathway of nucleotide synthesis in Arabidopsis suspension cells.
To address the second question, the transcript levels of the seven AtENTs in five Arabidopsis organs were compared in semiquantitative RT-PCR experiments. The transcripts of AtENT1 accumulated abundantly in all five organs (Fig. 3C). In contrast, the transcription of the remaining six AtENTs showed varying degrees of organ specificities (Fig. 3C). The transcripts of AtENT3 were undetectable in stem, whereas those of AtENT4 were not found in either root or silique (Fig.  3C). The transcripts of AtENT7 were not detected in root, stem, or silique although it was highly expressed in both leaf and flower (Fig. 3C). The seven AtENTs were all expressed in leaf and flower (Fig. 3C). In contrast, fewer AtENTs were expressed to high levels in root (AtENTs 1 and 3), stem (AtENTs 1, 4, and 8), or silique (AtENTs 1, 3, 6, and 8) (Fig. 3C).
Transport Activities of AtENT3-Before studying the transport activities of AtENT3, an AtENT3:EGFP fusion cistron was expressed in both insect and yeast cells. Confocal microscopy of GFP fluorescence showed that the AtENT3:EGFP fusion protein was localized specifically to the plasma membrane in either yeast or insect cells (data not shown). An AtENT3:GFP fusion cistron was also constructed and expressed in plant cells. In the control cell expressing the GFP cistron (Fig. 4A), the GFP fluorescence was distributed throughout the entire cell. In contrast, in the cell expressing the AtENT3:GFP fusion cistron (Fig. 4B), the GFP fluorescence was localized to the periphery of the cell. A clear association of the AtENT3:GFP fusion protein with the plasma membrane (Fig. 4C, indicated by arrows) but not the cell wall (Fig. 4C, indicated by arrowheads) was found when the expressing cells were subject to osmotic shock. The same treatment did not change the distribution of GFP throughout the cytoplasm in the control experiment (Fig. 4D).
Following the strategies and methods described under "Experimental Procedures," five expression constructs (pY-AtENT1, pYAtENT1:GFP, pYAtENT3, and pYAtENT3:GFP) as well as the control vector pYES2 were introduced into the cells of the BMA64 -1A ade2 strain of S. cerevisiae. After inducing the expression of the cloned sequences by galactose, the ade2 cells harboring AtENT1 or AtENT3 coding sequences could grow in the presence of 150 M adenosine (Fig. 5 and data not shown), indicating that adenine dependence of the ade2 cells was rescued by adenosine transport activities of the proteins expressed from the cloned sequences in the four expres-  Table II.
Comparative Analysis of Arabidopsis ENTs sion constructs. Using the ade2 cells expressing pYAtENT1, the main characteristics of adenosine transport mediated by AtENT1 were investigated. The K m and V max of adenosine transport by AtENT1 were 3.6 M and 134.7 pmol/mg protein/ min, respectively. Adenosine transport by AtENT1, although resistant to dilazep and NBMPR, was abolished by CCCP. These results were comparable with those obtained by Möhlmann et al. (13) on adenosine transport by AtENT1 using the original W303 ade2 strain of S. cerevisiae.
Following the above experiments on AtENT1, the characteristics of nucleoside transport mediated by AtENT3 were investigated in more detail. In initial uptake experiments, we found that BMA64 -1A ade2 cell cultures expressing AtENT3 or AtENT3:GFP can both transport [ 3 H]adenosine, but the uptake rate by AtENT3:GFP-expressing cells was generally 10% lower than that by AtENT3-expressing cells. Therefore, yeast cells expressing AtENT3 rather than its GFP fusion protein were used for subsequent experiments. In initial time course experiments with [ 3 H]adenosine or [ 3 H]uridine, the uptake of both radiolabels was approximately linear for the first 5 min. Consequently, uptake periods of 2 min were employed in subsequent experiments in order to estimate initial rates of substrate transport. The K m and V max of adenosine transport by AtENT3 expressed in the BMA64 -1A ade2 strain were found to be 2.9 M and 269.9 pmol/mg protein/min, respectively (  (46). The subcellular distribution of the ectopically expressed proteins was revealed by examining the location of GFP fluorescence using confocal microscopy. In C and D, the bombarded epidermal strips were subject to osmotic shock (by 0.8 M mannitol) for 15 min before being examined for GFP fluorescence. The arrows indicate the association of AtENT3:GFP with the plasma membrane (C). The filled arrowheads mark the cell wall (C). Bars, 100 m. 6A). The corresponding values of uridine transport by AtENT3 expressed in the fui1 strain were 3.2 M and 232.5 pmol/mg protein/min, respectively (Fig. 6B). CCCP reduced [ 3 H]adenosine transport of AtENT3 by about 12% (Fig. 6C, Table III). DNP inhibited [ 3 H]adenosine transport of AtENT3 by ϳ38% (Table III). A similar effect of CCCP and DNP on [ 3 H]uridine transport of AtENT3 was also found (Fig. 6D, Table III). Vasodilator drugs partly inhibited [ 3 H]adenosine and [ 3 H]uridine transport by AtENT3 (Table III). However, AtENT3 transport of the two nucleosides was strongly resistant to NBMPR (Table  III). Employing [ 3 H]adenosine, the substrate specificity of AtENT3 was further investigated. In competition experiments, AtENT3 transport of [ 3 H]adenosine was most significantly and consistently reduced by both purine and pyrimidine nucleosides and 2Ј-deoxynucleosides (Table IV). In addition, certain nucleobases (cytosine and uracil) and nucleotides (e.g. ADP and ATP) also inhibited [ 3 H]adenosine uptake albeit at lower efficiencies (Table IV). DISCUSSION As a model species, A. thaliana has frequently been used in studying fundamental questions in higher plants. ENTs exist in a wide variety of organisms, including higher plants (5,6,33,50,51). However, their roles in plant biology remain to be investigated in a systematic manner. The complete sequencing of the genome of A. thaliana makes possible to investigate and compare the functions of ENTs in a higher plant species from a whole genome level. But prior to the results reported in this paper, only a limited amount of information had been obtained on one (AtENT1) of the eight predicted ENTs in A. thaliana (13,33). In this study, the cDNAs for AtENTs 2, 3, 4, 6, 7, and 8 were identified and sequenced. Despite repeated efforts, we

FIG. 6. Characters of [ 3 H]adenosine and [ 3 H]uridine transport by recombinant AtENT3. [ 3 H]
Adenosine transport was carried out using the BMA64 -1A ade2 cells expressing AtENT3, whereas [ 3 H]uridine transport was conducted using the AtENT3 expressing fui1 cells. A and B, kinetics of adenosine or uridine transport into intact yeast cells. Transport was allowed to proceed for 2 min in the indicated concentrations of substrates. Uptake rates and S.E. values were calculated using the data (after subtracting the uptake rates of control cells) from three independent assays. The K m and V max values were derived from nonlinear fit to the uptake plot. Under such conditions, the K m and V max of adenosine transport by AtENT3 were 2.9 M and 269.94 pmol/mg of protein/min, respectively, whereas the corresponding values of uridine transport by AtENT3 were 3.2 M and 232.5 pmol/mg of protein/min, respectively. The inserted Eadie-Hofstee plots are for illustration purposes. C and D, effect of the protonophore CCCP on adenosine or uridine transport by recombinant AtENT3 in time course experiments. Yeast cells expressing AtENT3 were suspended in a transport medium containing 1.8 M [ 3 H]adenosine or [ 3 H]uridine (filled circles) or in the transport medium containing an identical concentration of the radioactive nucleoside plus 5 M CCCP (open circles). Uptake rates and S.E. values were calculated using the data from three independent assays. As a control, [ 3 H]adenosine or [ 3 H]uridine uptake (in the absence of CCCP) by yeast cells harboring the empty vector (pYES2 for adenosine uptake, p181AINE for uridine uptake) was also determined (filled triangles). failed to identify a cDNA for AtENT5. The failure may be caused by one or more of the following possibilities. The gene encoding AtENT5 may not be transcribed, its transcript level may be too low to be detected by RT-PCR, or its expression may be highly regulated both spatially and temporally.
Amino acid sequence comparisons indicated that the seven AtENTs are more closely related to each other than any of them are to ENTs from other sources. It is therefore understandable that the seven AtENTs formed an independent cluster in phylogenetic investigation. The same analysis also showed that the seven AtENTs could be divided into two subgroups (one containing AtENTs 1 and 8, the other containing AtENTs 2, 3, 4, 6, and 7), indicating that there may be genetic and/or functional differentiation among ENTs from a single plant species. AtENTs and their counterparts from mammals, C. elegans, S. cerevisiae, and parasites all have 11 putative TMs, suggest-ing that the conservation of 11 transmembrane TMs may be essential for all ENTs to transport their substrates and/or interact with cellular membrane systems. It is interesting to note that the glycine residues located at the 179-position of hENT1 was conserved at the corresponding positions of hENT2 and all seven AtENTs. Further experiments will be carried out to assess the importance of this residue in maintaining the structure and transport function of AtENTs. In all seven AtENTs, the position corresponding to the glycine residue at the 154-position of hENT1 was occupied by Asp. In previous mutagenesis experiments, it has been found that a conservative substitution of this Gly residue with Ser resulted in a mutated hENT1 protein with reduced sensitivity to NBMPR (25). Because the size and charge properties of Asp are very different from those of Gly or Ser, it will be interesting to test whether the presence of Asp at this position is associated with the strong resistance of AtENT3 (or AtENT1) to NBMPR. Finally, in all seven AtENTs, the position corresponding to residue 33 of hENT1 was occupied by one of two aliphatic amino acids (Ile in AtENTs 1 and 8, Leu in AtENTs 2, 3, 4, 6, and 7). Because it has been shown that the mutation of residue 33 from Met to Ile in hENT1 caused a significant reduction in dilazep and dipyridamole sensitivity (26), it will be important to examine whether the presence of aliphatic amino acids in the positions corresponding to the residue 33 of hENT1 is associated with the resistance of AtENT1 to dilazep and dipyridamole observed previously (13) or with the relative insensitivity of AtENT3 to dilazep demonstrated in the current study (see below).
In Arabidopsis suspension cells, the transcription of AtENTs 1, 3, 4, 6, and 8 was clearly increased by nitrogen deprivation or the application of fluorouracil and methotrexate. Nitrogen deprivation can have many consequences on the metabolism and growth of plant suspension cells. One such consequence may be the inhibition of the de novo pathway of nucleotide synthesis, because the execution of the pathway needs low molecular weight nitrogenous precursors, the synthesis of which relies on the absorption of nitrogen nutrient from the culture medium. It is possible that nitrogen deprivation reduced the efficiency of the de novo pathway of nucleotide synthesis and lowered the nucleotide pool of the suspension cells. This led to an increase in the activity of the salvage pathway of nucleotide synthesis, which required enhanced transcription (expression) of multiple AtENTs. Thus, the results of our nitrogen deprivation experiments provided useful evidence on the function of multiple AtENTs in the salvage pathway of nucleotide synthesis. Another important piece of evidence on the function of AtENTs in the salvage pathway of nucleotide synthesis came from the experiments using fluorouracil and methotrexate. Fluorouracil and methotrexate inhibit thymidylate synthetase and dihydrofolate reductase, respectively (52). Because both enzymes are required for dTMP synthesis through the de novo pathway, their inhibition may result in depletion of the dTMP pool in the suspension cells. To compensate for the reduction of cellular dTMP pool, the synthesis of dTMP via the salvage pathway may be augmented, which would require increased transcription (expression) of multiple AtENTs. Because the transcription of AtENTs 2 and 7 was undetectable in the suspension cells, their potential role in the salvage pathway of nucleotide synthesis could not be investigated in our nitrogen deprivation or drug treatment experiments using suspension cells. Because AtENTs 2 and 7 were transcribed in leaf and flower in planta, their potential role in nucleoside transport (salvage) will be investigated in future using intact plants.
Under normal growth conditions, AtENT1 was expressed constitutively and abundantly. In contrast, the transcription of the remaining six AtENTs showed varying degrees of organ  specificities. All AtENTs were expressed in leaf and flower, suggesting that nucleoside salvage (and the de novo pathway of nucleotide synthesis) may be an integral part of the physiology of the two organs. Past studies have shown that in higher plants nucleotides can be synthesized via both de novo and salvage pathways in leaf tissues and that there are nucleotides (such as ATP) and derivatives (e.g. ADP, AMP, adenosine, etc.) in the intercellular space of leaf cells (53)(54)(55). Furthermore, when leaves undergo senescence, nucleic acids in leaf cells are degraded by various nucleases and phosphatases to yield nucleosides, nucleobases, and phosphate for reuse in the growth of newer organs (56). It is therefore probable that a higher level of nucleoside transport activities is maintained in the leaf cells for transport and salvage of nucleosides. Flower development in higher plants involves the formation of various floral meristems through mitosis and the development of gametic cells via meiosis. Both processes require the replication of genomic DNA, for which a sufficient supply of nucleotide precursors is essential. It is possible that, in flowers, the two pathways of nucleotide synthesis are both operated in order to maintain an adequate supply of nucleotide precursors. If the de novo pathway is inhibited due to nutrient deficiency (by lack of soil nitrate, etc.) or inhibition of photosynthesis (by environmental stresses such as drought, etc.), the salvage pathway would still ensure some supply of the nucleotides required for the formation of floral meristematic cells and gametic cells. The expression of all seven AtENTs in the flower may thus be a reflection of the importance of the salvage pathway in the development of flower and associated gametic cells. Evidence for an essential role of the salvage pathway of nucleotide synthesis in flower comes also from the finding that mutation of an enzyme required for the salvage pathway results in defective pollen cells and male sterility in A. thaliana (57). Compared with the leaf and flower, fewer AtENTs were expressed in the root, stem, and silique. AtENTs 1, 3, 6, and 8 were strongly transcribed in the silique. This indicates that nucleoside transport activities (and the salvage pathway of nucleotide synthesis) may also be important in the development of the maternal (pod and seed coat) and embryonic (cotyledon and embryonic axis) tissues in silique. AtENTs 1, 4, and 8 were highly transcribed in the stem. They may be involved in loading and unloading of nucleosides and analogs (such as nucleoside cytokinins; see below) in the vascular system for long distance transport. Only two AtENTs (1 and 3) were transcribed to high levels in root. This indicates that nucleoside transport (salvage) in the root may not be as extensive as that in the leaf, flower, silique, and stem. Nucleosides may be produced from nucleic acid degradation in decaying plant or microbial cells in the soil. The two AtENTs expressed in the root may have a role in the salvage of the nucleosides present in the rhizosphere. The root-expressed AtENTs 1 and 3 may also be involved in the transport of the phytohormone cytokinins. Cytokinins, which are adenine analogs, are synthesized mainly in the root apical meristem (58,59). They are transported from root to shoot through the xylem in the form of nucleoside cytokinins (58 -60). Because there is currently little information on how nucleoside cytokinins enter into and exit from the xylem, it would be important to examine whether AtENTs expressed in root (or stem) may aid the transport of cytokinins through the vascular system of higher plants. Taken together, our results suggest that nucleoside transport (salvage) may be important in all parts of an Arabidopsis plant. However, the extent of nucleoside transport (salvage) may differ among different organs, which may mainly be controlled by differential expression of AtENTs 2, 3, 4, 6, 7, and 8. Judging from patterns of organ specificities and transcript levels, AtENTs 1 and 3 may be considered as the major ENTs whose activities are more widespread in A. thaliana. The activities of AtENTs 4, 6, and 8 are less prevalent, and those of AtENTs 2 and 7 are more restricted to specific organs.
Previous investigation has shown that AtENT1 expressed in the W303 ade2 strain of S. cerevisiae was a high affinity adenosine transporter (13). In the present study, we confirmed previous results on major characteristics of nucleoside transport mediated by AtENT1 using an alternative ade2 strain of S. cerevisiae and constructs prepared using an alternative yeast expression vector. For our investigation on the nucleoside transport activities of AtENT3, we first demonstrated that AtENT3 expressed from the cloned cDNA was targeted to the plasma membrane of plant, yeast, or insect cells. Subsequently, we found that AtENT3 expressed in yeast cells transported adenosine (K m , 2.9 M) and uridine (K m , 3.2 M) with high affinity. Furthermore, AtENT3 is also likely to be a transport with broad substrate specificity on the basis of the data from the competition experiments (Table IV). We have recently obtained evidence on the transport of nucleotides (AMP, ADP, and ATP) by AtENT3 expressed in the BMA64 -1A ade2 strain of S. cerevisiae. 2 The transport processes mediated by AtENT3 and AtENT1 share both similarities and differences. The similarities include resistance to nanomolar concentration of NBMPR and transport of both nucleosides and 2Ј-deoxynucleosides. The major differences were in the following three aspects. First, protonophores (CCCP and DNP) inhibited adenosine transport of AtENT1 by almost 100% (13). In contrast, the inhibition of AtENT3-mediated adenosine or uridine transport by CCCP and DNP was only around 10 and 35%, respectively (Table III). Second, adenosine transport of AtENT3 was partly inhibited by nanomolar concentrations of dilazep and dipyridamole, whereas the two chemicals had no effect on adenosine transport by AtENT1 (13). Third, by expressing AtENT3 cDNA in the fui1 strain of yeast, we obtained direct evidence on uridine transport by AtENT3 (Table III), which is a common permeant of mammalian ENTs (61). However, AtENT1 was previously deduced not to transport uridine (13). Based on the above comparisons, we suggest that AtENT3 is an ei type ENT.
While investigating the effect of vasodilator drugs on nucleoside transport of AtENT3, it was interesting to note that a large increase in the concentration of dilazep or dipyridamole was not followed by a more dramatic reduction in AtENT3's activity (Table III). If the two drugs were acting as competitive inhibitors, the results obtained with of 20 nM would suggest a K i of ϳ30 nM. With this calculation, the application of 20 M dilazep or dipyridamole would eliminate AtENT3 transport activity. The present data, therefore, indicate that the two vasodilator drugs may act in a noncompetitive manner in inhibiting nucleoside transport activity of AtENT3.
In conclusion, we have presented in this paper a more extensive study of the ENTs of A. thaliana predicted by genome sequencing. The cDNAs for seven of the eight predicted At-ENTs were cloned; their putative proteins could be subdivided into two subgroups. The majority of the AtENTs may function in the salvage pathway of nucleotide synthesis. Under normal growth conditions, there were at least two AtENTs highly transcribed in each of five Arabidopsis organs, suggesting that ENT activities were important and regulated in all parts of an Arabidopsis plant. In contrast to AtENT1 that was highly transcribed in all five Arabidopsis organs and may function as a concentrative, proton-linked transporter, AtENT3, transcribed in most Arabidopsis organs, represented a typical ei type ENT. The results and resources generated in this study have laid the foundation for more detailed comparative studies on mechanism of nucleoside transport, regulation, and biological role of AtENTs in the growth and development of A. thaliana in the future.