UPS1 and UPS2 from Arabidopsis Mediate High Affinity Transport of Uracil and 5-Fluorouracil* □ S

Salvage pathways play an important role in providing nucleobases to cells, which are unable to synthesize suf-ficient amounts for their needs. Cellular uptake systems for pyrimidines have been described, but in higher eukaryotes, transporters for thymine and uracil have not been identified. Two plant transporters, AtUPS1 and PvUPS1 , were recently identified as transporters for allantoin in Arabidopsis and French bean, respectively. However, , in contrast to tropical legumes, uses mainly amino acids for long distance transport. cloned into pCR4TopoBlunt and sequenced. The UPS1 antisense fragment was excised with XbaI/XhoI and ligated into the pCR4TopoBluntAAP6intron construct. The resulting construct was subsequently cleaved with HindIII/SpeI and the UPS1 sense fragment, excised by HindIII/SpeI, was introduced. The complete UPS1 sense- AAP6- intron- UPS1- antisense construct was subcloned into pCB302.3 (27) using SpeI/XbaI. Promoter-GUS Constructs— The AtUPS1 ( AtUPS2 ) promoter region comprising the 647-bp (721 bp) genomic DNA extending from the 3 (cid:3) - untranslated region of the preceding gene to ATG was PCR-amplified from genomic DNA (BAC T4M8) using primers 5 (cid:3) -ATAACTAGTCTTA- CAAGAACAGCAAGCTTT-3 (cid:3) and 5 (cid:3) -ATAGGATCCCTTTCTATCTCTC-TTTTATAT-3 (cid:3) (5 (cid:3) -ATATCTAGAATGATCTTATCATAGTTGTAT-3 (cid:3) and 5 (cid:3) -ATAGGATCCACAAGCTATGGCACCTCCTTT-3 (cid:3) ), cloned into pCR4TopoBlunt, and sequenced. The AtUPS1 ( AtUPS2 ) promoter was excised by SpeI/BamHI (XbaI/BamHI) and introduced into pCB308

Purine and pyrimidine nucleotides are building blocks for DNA and RNA synthesis and essential precursors for enzyme cofactors (NAD, FAD, 5-adenosylmethionin, thiamin), signal molecules (e.g. cyclic nucleotides or the phytohormone cytokinins), and plant secondary metabolites (e.g. caffeine). They can be synthesized de novo or obtained by recycling of nucleosides and nucleobases (salvage pathways) (1,2). The enzymes involved in salvage pathways frequently act with cellular import systems for their substrates in a concerted manner. Thus, most unicellular organisms including prokaryotes, fungi, and cellular parasites possess specialized transporters for the capture of nucleosides and nucleobases from their environment, which are used for nucleotide synthesis with lower net energy cost compared with de novo synthesis. In some extreme cases obligate parasites have lost the capacity to produce nucleotides via de novo synthesis, and import of nucleobases or nucleosides represents the only route to nucleotide formation. Uptake systems for nucleosides and nucleobases are also present in multicellular organisms. Moreover, the expression of uptake and release systems in distinct animal cell types seems to play a crucial role in the recovery and the distribution of nucleobases and nucleosides passing tissue barriers, e.g. in the kidney, intestine, and placenta (3). These transport systems, therefore, play not only an important role in organism physiology but also in the pharmaco-dynamics of anti-tumor compounds, such as 5-FU 1 (4).
In plant cells adenylates comprise the largest nucleotide pool followed in size by uridylates. UDP-sugars serve as activated intermediates in the synthesis of sucrose and cell wall polymers (5). Of the free pyrimidine bases, plants apparently can only salvage uracil, whereas they are unable to salvage cytosine due to the lack of cytosine deaminase (5). Cells often rely on salvage pathways when the activities of enzymes for de novo synthesis are low or absent (e.g. during early stages of seed germination) or when demand for nucleobases is high (e.g. when cells divide rapidly). Although salvaging of nucleobases and nucleosides predominates at the inception of germination, its importance declines when germination proceeds and de novo synthesis takes over (6). Salvaging is often accompanied by transport of nucleobases and nucleosides from storage tissues, as was demonstrated for germinating castor bean (7). A high affinity transporter system for uracil with a K m of ϳ40 M was described (7).
Despite the apparent physiological importance of high affinity transport of free pyrimidines, in particular of uracil, no transporters for these substrates have been identified in plants or other higher eukaryotes until now. A number of membrane protein families mediating transport of nucleobases or nucleosides into the cell have been described to date; they comprise (i) the nucleobase ascorbate transporters (NAT; Archaea, eubac-teria, fungi, plants, and animals) (8,9), (ii) the purine-related transporter family (PRT; Archaea, bacteria, fungi, and plants but not in animals) (10), (iii) the purine-related permeases (PUP; only in plants) (11,12), (iv) the ureide permeases (UPS; only in plants) (13,14), (v) the concentrative nucleoside transporters (fungi and animals but not in plants) (15), and (vi) the equilibrative nucleoside transporters (ENT; fungi and animals and plants) (16,17). Only two of these families include members with demonstrated capacity to transport uracil and other free pyrimidines; NAT, e.g. UraA in Escherichia coli (18), and PRT, e.g. FUR4 in Saccharomyces cerevisiae (19). However, the investigated NAT homologues in animals mediate Na ϩ -coupled ascorbate transport, and maize Lpe1 (leaf permease), the only NAT protein characterized so far in plants, transports xanthine and uric acid (9). PRT homologues are not present in higher eukaryotes with the exception of an expressed Arabidopsis protein with unknown function (At5g03555).
Recently, members of the UPS family mediating allantoin transport were identified in Arabidopsis and French bean (13,14). Allantoin is an important transport form of organic nitrogen in legumes, whereas Arabidopsis is considered to preferentially transport amino acids. Therefore, the substrate specificities of AtUPS1 and AtUPS2, related members of the UPS family from Arabidopsis, were analyzed in detail. Using heterologous expression in yeast and in Xenopus oocytes, it was demonstrated that pyrimidines are excellent substrates for both AtUPS1 and AtUPS2. The affinity for uracil was severalfold higher than for allantoin. In plants, expression of AtUPS1 and AtUPS2 was high during periods of increased demand of nucleotides, i.e. early seedling development, indicating a function in salvage pathways, e.g. by the utilization of pyrimidines from seed storage tissue. In planta, AtUPS function as major transporters for uracil analogs and potentially uracil itself, as demonstrated by the increased tolerance to 5-FU of plants in which AtUPS1 was inhibited by PTGS or in plants that carry T-DNA insertions in AtUPS2.

EXPERIMENTAL PROCEDURES
UPS1 and UPS2 Expression Constructs-AtUPS2 was amplified by PCR from a cDNA library from 5-day-old seedlings from Arabidopsis thaliana (ecotype Col-0) with the primers 5Ј-ATAGGATCCATCCATT-TAGAGCCCGAGAAT-3Ј and 5Ј-ATATCTAGATTACTTTCTATGTCCA-GAAGA-3Ј and cloned into pCR4TopoBlunt (Invitrogen). Sequencing revealed mutations of A to T (at position 276 of the open reading frame) and A 678 to G not resulting in an altered protein sequence. In addition, T 308 was mutated to C, causing a valine to alanine exchange. This mutation was reverted by site-directed mutagenesis. The AtUPS1-coding sequence was PCR-amplified from pFL61UPS1 (13) and cloned into pCR4TopoBlunt. The AtUPS1-coding sequences were BamHI/XbaI excised from pCR4TopoBluntUPS1 and cloned into the oocyte expression vector pOO2 (20) cut by BamHI/XbaI. The UPS1 oocyte expression construct was described previously (13). The UPS2 coding sequence was BamHI/XhoI-excised from pOO2UPS2 and subcloned into the yeast expression vector pDR199. pDR199UPS1 was derived by amplifying UPS1 from the cDNA library with primers 5Ј-CCCAGCCTTGAAAAT-AATTAAAATTTATAATTTTAAAAGATAAAAGAGAGATAGAAAGAT-G-3Ј and 5Ј-AAGCTGGATCGCTCGAGTCGACTGCAGGCCGCCC-GGGCCGTCATTTTCTATGTCCCGAGGAAG-3Ј. The PCR product together with the EcoRI/BamHI-linearized vector pDR199 was introduced into a yeast dal4 dal5 knock out strain (13) by in vivo cloning (21). After selection on uracil-free medium, plasmids were isolated, and the cDNA insertion was verified by sequencing. Two T to C mutations (positions 997 and 1152 of the open reading frame), which do not alter the protein sequence, were found in the coding sequence. pMD200 was derived from pDR195 (22) by disruption of the URA3 gene by ApaI/NcoI digestion and introduction of the LEU2 gene. LEU2 was PCR-amplified with primers 5Ј-ATAGGGCCCGTGGGAA-TACTCAGGTATCG-3Ј and 5Ј-ATACCATGGCTGGACGTAAACT-CCTCTTC-3Ј from pACT2 (BD Biosciences). The UPS1 (UPS2)coding sequences were excised from pCR4TopoBluntUPS1 (pCR4TopoBluntUPS2) and cloned into pMD200; UPS1 was excised EcoRI, blunted, and introduced into the blunted NotI site of pMD200; UPS2 was excised by XbaI/NotI, XbaI was blunted, and the construct was introduced into pMD200 (BamHI-blunted/NotI).
Expression in Yeast-The yeast dal4 dal5 mutant strain was transformed with pDR199 or pDR199UPS1 (pDR199UPS2) (23). Growth tests were performed on medium (synthetic defined medium without ammonium sulfate) with 5 mM allantoin. The yeast strain CEN.PK 2-1C (Euroscarf) was used to generate a fur4 knockout mutant strain by using a loxP-KanMX-loxP disruption cassette (24). The fur4 mutant strain was selected on synthetic defined medium (with 38 mM ammonium sulfate, 200 mg/liter G418, 0.1 mM L-histidine, 0.1 mM L-tryptophan, 0.2 mM uridine), confirmed by PCR, and subsequently transformed with pMD200 or pMD200UPS1 (pMD200UPS2). Growth tests were performed on medium with 0.2 mM uracil instead of uridine. Transport measurements with 14 C-labeled allantoin were performed as previously described (13).
Expression in Xenopus laevis Oocytes-Oocytes were surgically removed from female X. laevis frogs. Follicular cell layers and connective tissues were removed by digesting with 2 mg/ml collagenase A (Roche Applied Science) for 1.5-2 h at 22°C in a Ca 2ϩ -free solution composed of 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 5 mM HEPES, pH 7.4. Oocytes were subsequently washed 5-8 times with ND96 solution (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES, pH 7.4). Healthy oocytes at developmental stages V-VI were selected and maintained at 16°C in ND96 solution supplied with 50 mg/liter gentamycin and 2.5 mM sodium pyruvate (ND96-G). Capped cRNAs were transcribed in vitro using the mMessage mMachine mRNA SP6 kit (Ambion) from linearized plasmids pOO2UPS1 and pOO2UPS2. RNA yields were estimated by agarose gel electrophoresis, and the concentrations were finally adjusted to ϳ1 g/l with nuclease-free water. Oocytes were microinjected with 50 nl of cRNA (20 -50 ng/oocyte) and incubated at 16°C in ND96-G solution to allow the expression of AtUPS genes.
Electrophysiology-Two-electrode voltage clamp was performed starting 2 days after injection. The standard perfusion solution (wash solution) contained 100 mM choline chloride, 2 mM CaCl 2 , 2 mM MgCl 2 in 5 mM Mes-Tris, pH 5.5. Substances were dissolved in this solution to final concentrations of 200 M, and the pH values were adjusted to 5.5 with Mes or Tris. For voltage-current analysis, currents were recorded while clamping the oocytes at Ϫ100 mV or by applying an I/V protocol. The I/V protocol consisted of 200-ms pulses of voltages ranging from Ϫ140 to 0 mV in steps of 20 mV. Substrate evoked currents were constant during these short pulses. Net currents induced by substrates were obtained by subtracting the background currents (measured with "wash solution" alone) before and after the addition of substrates. To avoid endogenous currents, which were slowly activated at more negative voltages, currents elicited at more negative voltages than Ϫ120 mV were not analyzed (in some cases only voltages in the range of 0 to Ϫ100 mV were analyzed). No differences between non-injected and waterinjected oocytes were observed; thus, in most cases non-injected oocytes were used as controls.
T-DNA Insertion Lines-A. thaliana ecotype Col-0 was used throughout this study. Seeds of the T-DNA insertion lines Salk_ 143013 (designated ups2-1) and Salk_044551 (designated ups2-2) were obtained from the Arabidopsis Biological Resource Center (Columbus, Ohio) (25). Homozygous plants were isolated from T3 plants and confirmed in the T4 generation. Genomic DNA was prepared as described (www.dartmouth.edu/ϳtjack/TAILDNAprep.html). Homozygous lines were screened by two sets of PCR. The first PCR was performed using gene-specific primers 5Ј-GTAGATAAGTTTTCCGTG-ACG-3Ј and 5Ј-GAACCTAGCAAAGGTAGTGTC-3Ј (Salk_044551) or 5Ј-CATTGCAAACATAACAGACG-3Ј and 5Ј-GTATCTTTACATTCTCTGG-AC-3Ј (Salk_143013). The second PCR was carried out with the T-DNA left border primer 5Ј-CTTTGACGTTGGAGTCCAC-3Ј and the primer 5Ј-GAACCTAGCAAAGGTAGTGTC-3Ј (Salk_044551) or 5Ј-GTATCTT-TACATTCTCTGGAC-3Ј (Salk_143013). Plants were considered homozygous if no PCR product was obtained by the first PCR reaction, and fragments of expected size confirming the T-DNA insertion were amplified by the second PCR reaction. Localization of T-DNA insertions was analyzed by sequencing. Homozygous lines were subsequently taken for further investigations.
Plant Transformation-pCB302. 3 and pCB308 constructs were introduced into Agrobacterium strain GV3101 for plant transformation. Arabidopsis plants grown in the greenhouse were transformed with Agrobacterium using the floral dip method (28) and transgenic plants (T1 to T3 generation) were recovered by BASTA TM (phosphinothricin). For analysis of GUS activity three lines (T3) of the translational fusions of AtUPS1 or AtUPS2 promoter regions with the uidA gene were used for further investigation. Five randomly chosen T3 or T4 lines containing the UPS1-PTGS construct, three of which were homozygous and two heterozygous (based on BASTA resistance segregation), were analyzed for mRNA levels and phenotypic alterations.
Plant Growth-Seeds were vernalized at 4°C for 2 days before imbibition. Plants were grown in a Percival Scientific growth chamber under a 16 h of light/8 h of dark photoperiod at 22°C and a light intensity of 100 -150 microeinsteins m Ϫ2 s Ϫ1 . Plates with MS medium (Invitrogen) supplemented with 1% (w/v) sucrose and 0.7% (w/v) phytagar (Invitrogen) were used for all experiments except for studies with N-(phosphonacetyl)-L-aspartate (1 mM) and exogenous pyrimidines (1 mM). In the latter studies seedlings were grown in liquid MS medium containing 1% (w/v) sucrose on a rotary shaker at 100 rpm. For growth assays on 5-FU, the medium was supplemented with 200 M quantities of this compound.
Promoter GUS assays-Plants were harvested from days 1 to 11 after imbibition and analyzed for ␤-glucuronidase activity (29). Staining was performed for 15 h at 37°C. For root sections, 6-day-old plants were prefixed in 50 mM phosphate buffer, pH 7.2, containing 1.4% formaldehyde. Plants were washed in 50 mM phosphate buffer, pH 7.2, followed by GUS staining for 8 h at 37°C and subsequent fixation in 50 mM phosphate buffer, pH 7.2, containing 2% paraformaldehyde and 1% glutaraldehyde. Dehydration was performed in an ethanol series followed by embedding in hydroxyethyl methacrylate (Technovit 7100, Heraeus Kulzer). 6-m-thick sections were prepared with a LKB Ultratome and counterstained in purple by periodic acid-Schiff staining.

RESULTS
AtUPS1 and -2 Mediate Allantoin Uptake-Out of the five members of the UPS family in Arabidopsis, AtUPS1 and AtUPS3 seem to have arisen from a recent gene duplication as well as AtUPS 2 and AtUPS4. To be able to compare the properties of these two groups, the AtUPS2 cDNA (At2g03530) was isolated from a seedling cDNA library. AtUPS2 showed 74% sequence identity to AtUPS1 on the amino acid level. The yeast mutant dal4 dal5, deficient in the uptake of allantoin and allantoic acid, previously used to isolate AtUPS1 (13), was used to demonstrate that AtUPS2 was able to suppress the mutant phenotype in a growth assay using 5 mM allantoin as the sole source of nitrogen (Fig. 1A). Colonies of cells expressing AtUPS2 or AtUPS1 were several times bigger as those formed by control cells transformed with the empty vector. AtUPS1 colonies grew slightly faster than AtUPS2 colonies. In addition, uptake of [ 14 C]allantoin, due to AtUPS2 expression, was directly shown. The uptake rate was ϳ75% of the rate observed for AtUPS1-expressing yeast, whereas the uptake rates of dal4 dal5 yeast mutant cells transformed with the empty vector was ϳ3% of the rate observed upon AtUPS1 or AtUPS2 expression (Supplemental Fig. 1). Thus, both proteins mediate allantoin transport.
AtUPS1 and AtUPS2 Also Transport Uracil-Because there is no evidence that Arabidopsis uses allantoin as a long distance transport form of nitrogen, it was conceivable that in Arabidopsis UPS has a different physiological substrate. Previous studies had shown that AtUPS1 recognizes oxo derivatives of N-heterocycles (13). Because the structure of uracil fits the criteria established previously, and since uracil transport in plants has been described but transporters have not been identified, we tested whether AtUPS1 or AtUPS2 would be able to mediate uracil uptake. The yeast mutant fur4, deficient in uracil uptake, grew on 0.2 mM uracil as the sole source of pyrimidines when expressing AtUPS1 or AtUPS2, demonstrating that both transporters mediate uracil uptake (Fig. 1B).
Functional Expression of AtUPS1 and AtUPS2 in Xenopus Oocytes-To test the substrate specificities of AtUPS1 and AtUPS2 more directly, both transporters were functionally expressed in X. laevis oocytes. Expression of AtUPS1 induced inward currents in oocytes bathed in buffer with allantoin in a flow system. When clamped at Ϫ100 mV, significant inward currents were also recorded from AtUPS2-expressing oocytes upon the addition of allantoin ( Fig. 2A), whereas no significant currents were detected in non-injected control oocytes (Fig. 2, B and C). To define the substrate specificity in detail, a large number of compounds were tested. Criteria for the selection of potential substrates were (i) molecules belonging to the group of nucleobases and derivatives, (ii) compounds transported by other nucleobase transporters, and (iii) compounds with similar structural features. Currents were induced by cyclic purine degradation products and pyrimidines. In case of allantoin, hydantoin-, cytosine-, thymine-or uracil-induced currents remained stable or slowly decreased until washout of the substrates. However, when flushed with xanthine, the evoked current decreased within 10 s to a lower steady state level ( Fig.  2A). A similar effect was found for uric acid (not shown). Despite lower current levels, AtUPS1 behaved in a manner similar to AtUPS2. Clamping oocytes to Ϫ100 mV for longer periods resulted in small shifts of the base line ( Fig. 2A). To minimize errors conferred by this process, further investigations were done using brief substrate flushing periods and the I/V protocol described under "Experimental Procedures." The allantoin-induced currents recorded from AtUPS1-or AtUPS2expressing oocytes increased in response to further hyperpolarization of the membrane potential (Fig. 2C) and displayed clear voltage dependence. Steady state currents induced by different substances were plotted as a percentage of those caused by allantoin set to 100% (Fig. 3, A and B). The cyclic purine degradation products uric acid and xanthine induced currents in AtUPS1-and AtUPS2-expressing oocytes (ϳ60 and 50% for AtUPS1, 75 and 40% for AtUPS2, respectively). All pyrimidines and their tested derivatives (cytosine, uracil, thymine, dihydrouracil, and 5-FU) induced currents of different magnitude (Fig. 3A). Slight differences were found between AtUPS1 and AtUPS2 in response to thymine and cytosine (ϳ92 and 105% for thymine, 50 and 75% for cytosine, respectively), whereas uracil, dihydrouracil, and 5-FU were transported with comparable rates (ϳ 91, 110, and 140%, respectively). Transporters for some of the tested compounds have already been identified in plants, e.g. for nucleosides (AtENT) (31), for the secondary metabolite caffeine (At-PUP), and for the purine bases adenine, guanine, and hypoxanthine (AtPUP) (11,12). None of these substrates induced currents in AtUPS1 of AtUPS2-expressing oocytes (Fig.  3, A and B). Moreover, ascorbic acid failed to produce currents ( Fig. 3A). Allantoic acid and urea, which are linear degradation products of purines, were also not transported, confirming that the ureido residue is not a structural determinant for substrates. Hydantoin, a synthetic compound lacking the ureido substitution at position 5 of allantoin, induced currents in AtUPS1-and AtUPS2-expressing oocytes (ϳ108 and 127%, respectively) (Fig. 3A). No inward currents could be recorded for barbituric acid and orotic acid even though they seemed to resemble the structure of pyrimidines, except in having additional substitutions in the meta position to the oxo residues (Fig. 3B). A mechanism of co-transport of one proton with the neutral substrate had been suggested for AtUPS1 (13). The percentage of neutral and charged molecules varies between substrates at pH 5.5 used in the assays. In the case of allantoin, hydantoin, and uracil more than 99% of all molecules are uncharged, whereas in the case of cytosine 89% of all molecules are negatively charged (ϳ80 for uric acid and 20% for xanthine). At least 99.9% of orotic and barbituric acids carry one negative net charge. To elucidate whether these almost exclusively negatively charged compounds are transported without being electrically detected due to proton co-transport, competition experiments were performed. Allantoin and other compounds were mixed at equal concentrations (200 M), resulting in almost the same current levels as induced by allantoin alone. These results indicate that these compounds are not transported or bound by AtUPS1 or AtUPS2 (Fig. 3B). In summary, all transported substrates were heterocyclic compounds with similar electron density distribution established by the N-hetero-atoms and the substituents. AtUPS1 and AtUPS2 transport a similar range of substrates covering cyclic purine degradation products and pyrimidines.
AtUPS1 and AtUPS2 Have a High Affinity to Uracil-Because the affinities of transporters for their substrates, in conjunction with the local metabolite concentration in the plant, can be informative about the physiological function of the transporters, the kinetic parameters for allantoin and uracil transport were determined for AtUPS1 and AtUPS2. Currents mediated by allantoin or uracil depended on the concentrations of the compounds and displayed saturation kinetics (Fig. 4). For AtUPS1, an apparent K m for allantoin of 75 Ϯ 2.7 M was determined. This value is significantly higher as compared with that obtained by measuring [ 14 C]allantoin uptake in yeast (52 Ϯ 2 M) (13). The discrepancy might be due to the use of different expression systems (yeast versus Xenopus oocytes) and/or different measurement techniques (detection of radiotracers versus electrophysiological detection of co-transported protons). AtUPS2 had a 3-fold lower K m of 26 Ϯ 5.4 M. Both transporters showed higher affinities to uracil with K m values in the range of 5.9 Ϯ 0.7 M (AtUPS1) and 6.2 Ϯ 1.6 M (AtUPS2; Fig. 4). Thus, AtUPS1 and AtUPS2 have ϳ11and 5-fold higher affinity to uracil than to allantoin.
AtUPS1 and AtUPS2 Are Expressed during Seedling Development-AtUPS1 expression was found to be induced under conditions of nitrogen starvation (13). Nitrogen starvation causes a multitude of effects including nucleobase deficiency (31). To investigate whether AtUPS transcript levels might be increased under conditions of increased demand for nucleobases, mRNA levels of AtUPS1 and AtUPS2 were studied in seedlings during the first 11 days of development. AtUPS1 levels were high at day 1, declining continuously to steady levels after day 5 (Fig. 5). In contrast, expression of AtUPS2 was hardly detectable during the first 2 days of seedling development, whereupon expression levels increased until day 9. To study tissue-specific expression, promoter regions of AtUPS1 and AtUPS2 were isolated and fused translationally to the uidA gene. GUS expression, driven by the AtUPS1 promoter, was observed in 2-day-old seedlings (Fig. 6A). GUS staining was detected in the whole seedling with exception of the radicle. At day 3 GUS activity was present in the hypocotyl and in cotyledons but limited to hypocotyl and hydathodes (staining of hydathodes was not always observed) from day 4, further decreasing with age.
GUS activity driven by the AtUPS2 promoter was in agreement with the reverse transcription-PCR data. Starting at day 4, blue staining was observed in cotyledons, increasing in intensity and distribution during subsequent days (Fig. 6B). At later stages (11 days) GUS activity was high in the cotyledons and primary leaves. In addition, staining was observed in the stele of roots beginning at about day 5 (Figs. 6B and 7). To localize AtUPS2 expression in the root stele at the cellular level, transverse sections of plant roots were analyzed. Highest GUS activity was found in the pericycle, and weaker activity was detectable in all cells of the stele, except in mature xylem tracheary elements (Fig. 7).
To investigate if AtUPS expression is directly regulated by the availability of pyrimidines, the effects of uracil or uridine supplementation and of N-(phosphonacetyl)-L-aspartate, a potent transition-state analog inhibitor of aspartate transcarbamoylase causing pyrimidine starvation (32), were studied in Arabidopsis seedlings. Pyrimidine supplementation or starvation caused no more than slight variations of AtUPS expression levels ( Supplemental Fig. 2). These data do not provide evidence for a principal regulation of AtUPS expression by pyrimidine availability and suggest that other regulatory mechanisms may predominate in the context of plant development.
AtUPS1 PTGS Lines and AtUPS2 Knock-out Mutants Are Less Sensitive to 5-Fluorouracil-To test whether AtUPS1 and AtUPS2 serve as uracil transporters in planta, AtUPS2 T-DNA insertion lines ups2-1 and ups2-2 and several AtUPS1 PTGS lines were analyzed (Supplemental Fig. 3). The insertion in ups2-1 is localized in the first exon 16-bp downstream of the ATG, causing a deletion of 25 bp of the coding sequence. Line ups2-2 carries a T-DNA insertion in the third intron of AtUPS2. In either line no AtUPS2 mRNA was detectable by reverse transcription-PCR, indicating that both T-DNA insertions prevent the generation of stable full-length transcripts (Fig. 8). AtUPS1 mRNA levels were severely reduced in five independent PTGS lines, as shown by reverse transcription-PCR (Fig. 8). Due to the high sequence similarity of AtUPS1 and AtUPS2 in the region used for PTGS, mRNA levels of AtUPS2 were also reduced in 2 of the 5 lines. No phenotypic alterations were found for any of the lines when plants were grown in soil or on agar. However, when germinated on the toxic uracil analog 5-FU (200 M), wild type seedlings showed severe developmental retardation (33), whereasAtUPS1 PTGS plants as well as ups2-1 and ups2-2 mutants grew significantly better (Fig.  9). The reduced sensitivity to 5-FU suggests that the UPS transporters serve as the major uptake systems for 5-FU.

A Novel Way for Cellular Uptake of Pyrimidines in Higher
Eukaryotes-The present study indicates that AtUPS1 and AtUPS2 are proton-coupled transporters for uracil and other free pyrimidines. This conclusion is based on (i) the complementation of the uracil uptake-deficient yeast mutant fur4, (ii) the ability of uracil, thymine, dihydrouracil, 5-FU, and to a minor extent, cytosine to generate inward currents in Xenopus oocytes expressing AtUPS1 or AtUPS2, and (iii) the reduced sensitivity of Arabidopsis lines with reduced AtUPS1 or AtUPS2 transcript levels to 5-FU. Thus, AtUPS1 and AtUPS2 are the first identified proteins able to transport uracil and thymine in higher eukaryotes. Functional investigation of this transporter family might, therefore, be helpful in a broader context for understanding the role of transport in nucleobase metabolism and salvage in multicellular organisms.
The overall sequence similarity indicates that UPS constitutes a distinct class of plant membrane proteins (aramemnon-.botanik.uni-koeln.de). Although no phylogenetic relationships could be determined using sequence comparison tools between the bacterial UraA (NAT), the yeast FUR4 (PRT) and the plant UPS, they have in common the ability of proton-coupled high affinity transport of uracil into the cells. The lack of any evidence for a common ancestry, therefore, suggests that these transporters are the result of convergent evolution. However, it remains to be investigated whether certain amino acid residues, present at distinct topological positions, are crucial for substrate selectivity in the known uracil transporters. Studies on structure-function relationships of transporters are scarce due to the difficulty of crystallizing membrane proteins. Thus, mutational analysis has been used to explore possible structures determining substrate selectivity of nucleobase trans- porters (35,36). For example, substitution of Lys-272 located in a putative transmembrane domain of the S. cerevisiae uracil transporter FUR4 resulted in strong impairment of both binding and translocation of uracil (37).
Because allantoin, the substrate originally used to identify the UPS transporters, may not be of major relevance for long distance transport in Arabidopsis, one may hypothesize that pyrimidines are the primary substrates for UPS in Arabidopsis, whereas in tropical legumes, UPS has evolved during evolution as allantoin transporters for long distance transport of organic nitrogen (14). The finding that uracil, which is the principal free pyrimidine salvaged in plants, is transported with high affinity in the low micromolar range supports this notion. Furthermore, the expression pattern of AtUPS1 during the early events of germination is consistent with a role in uracil transport. It is, however, also conceivable that UPS also serves other functions, such as uptake of N-heterocycles from soil or in the translocation of still unknown N-heterocycles within the plant, as is suggested by the induction of AtUPS1 expression under N-starvation (13).
Given that germination and seedling development is not affected in the mutants with decreased expression of AtUPS1 and AtUPS2, other transporters for uracil or uridine may be able to compensate the reduced UPS transporter activity. The Arabidopsis genome encodes five members of the UPS family. Despite intensive screens, we have been unable to amplify transcripts for AtUPS3 and AtUPS4, although Expressed Service Tags have been identified for AtUPS4. AtUPS5 is transcribed and processed in two different mRNA splicing variants, but their functions remain unknown. In addition, a putative membrane protein (At5g03555) shares 30% sequence identity with the uracil transporter FUR4 of Schizosaccharomyces pombe on the protein level. To date efforts to demonstrate pyrimidine transport for this protein have been unsuccessful in our group. 2 However, it is important to note that the protein encoded by At5g03555 shares also similarity with other PRTs that are transporters for adenine and cytosine (FCY2) or the vitamin thiamine (THI10). Eight proteins of the ENT family are encoded in the Arabidopsis genome. Although substrate specificities and expression of several AtENTs have been recently reported (31,38), their functions, e.g. during germination, remain unclear.
AtUPS1 and AtUPS2 Serve Distinct Functions-AtUPS1 and AtUPS2 share a high degree of homology at the amino acid level (74% identity) and have similar predicted transmembrane structures (aramemnon.botanik.uni-koeln.de). Phylogenetically related proteins may either transport distinct substrates, have different affinities for a given substrate, or they may be expressed at different times or in different locations on cellular or subcellular level. For instance, the S. cerevisiae DAL4 is 70% identical to FUR4, but it serves different functions as a transporter for allantoin or uracil, respectively (34). In contrast, AtUPS1 and AtUPS2 seem to transport the same range of substrates. A notable difference is the 3-fold higher relative affinity of AtUPS2 for allantoin. All prediction programs of subcellular targeting in the data base Aramemnon consistently predict AtUPS1 to localize to the secretory pathway, whereas three programs suggest that AtUPS2 is in the secretory pathway; one predicts plastidic and one predicts mitochondrial localization (aramemnon.botanik.uni-koeln.de). The uptake efficiency observed in yeast and oocytes expressing AtUPS indicates that both proteins can be functionally targeted to the plasma membrane of these cells. Because mistargeting may occur in heterologous expression systems, subcellular localizations have to be experimentally validated in future studies. The differential expression of AtUPS1 and AtUPS2 mRNA indicates that, despite the similarity in biochemical function, the genes have distinct functions in the plant.
AtUPS1 is strongly expressed in seedlings during the early stages of germination and seedling development. Previous observations in germinating castor bean suggested that RNA from endosperm is degraded and released to the apoplasm in the form of nucleosides and free nucleobases (7). Castor bean cotyledons were shown to be able to take up nucleosides and nucleobases. The high expression of AtUPS1 at day 1 after imbibition may, thus, be taken as an indication that represents a transporter involved in salvaging of free pyrimidines originating from storage tissue mobilization during germination of Arabidopsis plants. The previously observed induction of AtUPS1 expression under N-starvation may be explained by a decreased availability of organic nitrogen (e.g. for nucleotide biosynthesis), necessitating an up-regulation of salvage pathways (13,31). It has been recently reported that members of the Arabidopsis ENT family were strongly induced by 5-FU-mediated thymidylate starvation (31). We also observed up-regulation of AtENT1 in response to N-(phosphonacetyl)-L-aspartate-mediated general pyrimidine starvation and down-regulation in response to exogenous uracil, 2 suggesting that transporter induction might be a general mechanism in response to changes in nucleotide availability. In contrast, only minor changes in AtUPS1 and AtUPS2 transcript levels were observed in response to changes in pyrimidine availability, suggesting that other factors may be more important in regulating expression of these transporters.
AtUPS2 presents a spatially and temporally distinct expression pattern as AtUPS1, indicating a different function. In tropical legumes high amounts of allantoin are produced in root nodules, transferred to the xylem vessels, and exported to the shoot via the transpiration stream. It has been postulated that PvUPS1 serves in the transfer of allantoin through the endodermis and for the loading into the xylem (14). Arabidopsis plants can use allantoin as the sole nitrogen source, which suggest that an allantoin transporter is localized in the roots. AtUPS2, but not AtUPS1, localizes in cells of the root stele. Thus, AtUPS2 may serve for utilization of a range of substrates present in the rhizosphere including pyrimidines but also allantoin, xanthine, and uric acid. Alternatively, AtUPS2 may be involved in utilization of pyrimidines synthesized in leaves and transported to the roots via the phloem.
Inhibition of either AtUPS1 or AtUPS2 expression led to similar seedling phenotypes when grown on plates containing 5-FU. However, AtUPS1 and AtUPS2 do not serve redundant functions. Thus, it is likely that both proteins are necessary for effective 5-FU transport from the external medium to the location where the toxic effect occurs, presumably in a sequential manner. According to the functions hypothesized above, AtUPS2 may allow 5-FU allocation in the plant, whereas AtUPS1 may transport it in cotyledon cells with photosynthetic capacity, where the toxic effects of 5-FU are maximal due to inhibition of the plastidic enzyme thymidylate synthetase (26).
In summary, the evidence presented in this work strongly suggests that UPS is involved in pyrimidine transport in Arabidopsis seedlings. Further studies using lines carrying mutations in AtUPS1, AtUPS2, and AtUPS5 and lines deficient in uridine transport will be required to determine the relative contribution of UPS to nucleobase salvaging and to identify potential other functions of this family of N-heterocycle proton cotransporters.