Molecular Identification and Characterization of Novel Human and Mouse Concentrative Na 1 -Nucleoside Cotransporter Proteins (hCNT3 and mCNT3) Broadly Selective for Purine and Pyrimidine Nucleosides (System cib )*

The human concentrative (Na 1 -linked) plasma mem brane transport proteins hCNT1 and hCNT2 are selective for pyrimidine nucleosides (system cit ) and purine nucleosides (system cif ), respectively. Both have homologs in other mammalian species and belong to a gene family (CNT) that also includes hfCNT, a newly identified broad specificity pyrimidine and purine Na 1 -nucle-oside symporter (system cib ) from the ancient marine vertebrate, the Pacific hagfish ( Eptatretus stouti ). We now report the cDNA cloning and characterization of cib homologs of hfCNT from human mammary gland, differentiated human myeloid HL-60 cells, and mouse liver. The 691- and 703-residue human and mouse proteins, designated hCNT3 and mCNT3, respectively, HL-60 Radioisiotope Flux Studies— Nucleoside uptake by differenti- ated HL-60 cells was measured as described previously (25) by exposing replicate cultures at room temperature to 3 H-labeled permeant (10 m M , 1 m Ci/ml) in sodium or sodium-free transport medium (130 m M NaCl or 130 m M NMDG/HCl and 3 m M K 2 HPO 4 , 2 m M CaCl 2 , 1 m M MgCl 2 , 20 m M Tris/HCl, and 5 m M glucose, pH 7.4). Radiochemicals (Moravek Biochemicals) were 98–99% pure as assessed by high performance liquid chromatography using water-methanol gradients on a C18 reverse phase column, and transport for timed intervals of 1–6 min was terminated by immersion of the culture dish in an excess volume of ice-cold transport solution. Assays to detect concentrative transport were performed in the presence of 100 m M dilazep (a gift from Hoffman La Roche & Co., Basel, Switzerland) to block equilibrative transport of the test nucleoside. Transport by nonadherent parental HL-60 cells was performed as described previously (26) using the inhibitor oil stop method. Values are presented as the means of triplicate measurements 6 S.D. Tissue and Cell Distribution of hCNT3 mRNA— A human multiple tissue expression (MTE TM ) RNA array (CLONTECH) and dot blots of mRNA (0.5 m g) from parent and differentiated HL-60 cells on Bright-Star-Plus nylon transfer membrane (Ambion) were incubated with a cDNA probe corresponding to hCNT3 amino acid residues 359–549 labeled with 32 P using the T7 QuickPrime kit (Amersham Pharmacia Biotech). Hybridization at high stringency (68 °C) was performed using ExpressHyb hybridization solution (CLONTECH) and 100 m g/ml of sheared herring sperm DNA. Wash conditions were as described in the CLONTECH ExpressHyb user manual. Signals on exposed blots were converted to a high resolution tiff image (Hewlett Packard ScanJet 4C) and quantified using

gene mapped to chromosome 9q22.2 and included an upstream phorbol myristate acetate response element.
Most nucleosides, including those with antineoplastic and/or antiviral activities (1,2), are hydrophilic, and specialized plasma membrane nucleoside transporter (NT) 1 proteins are required for uptake into or release from cells (3,4). NT-mediated transport is therefore a critical determinant of metabolism and, for nucleoside drugs, their pharmacologic actions (5). NTs also regulate adenosine concentrations in the vicinity of cell surface receptors and have profound effects on neurotransmission, vascular tone, and other processes (6,7).
Seven nucleoside transport processes 2 that differ in their cation dependence, permeant selectivities and inhibitor sensitivities have been observed in human and other mammalian cells and tissues. The major concentrative systems (cit, cif, and cib) are inwardly directed Na ϩ -dependent processes and have been primarily described in specialized epithelia such as intestine, kidney, liver, and choroid plexus, in other regions of the brain, and in splenocytes, macrophages, and leukemic cells (3,4). Concentrative NT transcripts have also been found in heart, skeletal muscle, placenta, and pancreas. The equilibrative (bidirectional) transport processes (es and ei) have generally lower substrate affinities and occur in most, possibly all, cell types (3,4). Epithelia (e.g. intestine and kidney) and some nonpolarized cells (e.g. leukemic cells) coexpress both concentrative and equilibrative NTs, whereas other nonpolarized cells (e.g. erythrocytes) exhibit only equilibrative NTs (3,4). Systems cit and cif are generally pyrimidine nucleoside selective and purine nucleoside selective, respectively, whereas systems cib, es, and ei transport both pyrimidine and purine nucleosides. System ei also transports nucleobases.
Molecular cloning studies have isolated cDNAs encoding the human and rat proteins responsible for four of these NT processes (cit, cif, es, and ei) (8 -17). These proteins and their homologs in other mammalian species comprise two previously unrecognized families of integral membrane proteins (CNT and ENT) with quite different predicted architectural designs (3,4). The relationships of these NT proteins to the processes defined by functional studies are: CNT1 (cit), CNT2 (cif), ENT1 (es), and ENT2 (ei). Although the NT protein(s) responsible for mammalian cib have remained elusive, we have recently identified a CNT protein with cib-type transport activity from the ancient marine vertebrate, the Pacific hagfish (Eptatretus stouti) (18). 3 The CNT family also includes the Escherichia coli proton/nucleoside symporter NupC (19). Human and rat CNT1 (650 and 648 residues, 71 kDa), designated hCNT1 and rCNT1, respectively, are 83% identical in amino acid sequence (8,11) and contain 13 putative TMs with an exofacial glycosylated tail at the carboxyl terminus (18). 4 hCNT2 (658 residues) (12,13) is 83% identical to rCNT2 (659 residues) (9, 10) and 72% identical to hCNT1 (11). The hagfish transporter hfCNT (683 residues) (18) is 50 -52% identical to h/rCNT1/2 and has a similar predicted membrane topology. NupC (19), in contrast, is a smaller protein with 27% identity to mammalian CNTs, with the major difference being the absence of the equivalents of TM 1-3 and the amino-and carboxyl-terminal regions of the other proteins.
In structure/function studies, the characteristics of hCNT1/2 chimeras and sequence comparisons between h/rCNTs and hf-CNT have identified two sets of adjacent residues in TMs 7 and 8 of hCNT1 that, when converted to the corresponding residues in hCNT2, changed the specificity of the transporter from cit to cif (18). Mutation of the two residues in TM 7 alone produced a protein with intermediate, cib-like activity. In this cit/cib conversion, mutation of hCNT1 Ser 319 to Gly was sufficient to enable transport of purine nucleosides, whereas mutation of the adjacent residue Gln 320 to Met (which had no effect on its own) augmented this transport. TMs 7 and 8 have also been identified as potential determinants of substrate selectivity in rCNT1/2 (21), and mutation of rCNT1 Ser 318 (the rat counterpart of hCNT1 Ser 319 ) resulted in a cib-type phenotype similar to that seen with the hCNT1 Ser 319 mutation (22).
Although an earlier study had identified a member of the SGLT glucose transporter family, SNST1, as a candidate cibtype transporter (23), its nucleoside-transport activity is very low, and we hypothesized that the missing mammalian concentrative NT was more likely to be a CNT transporter. Following a search for additional mammalian CNT isoforms, we now report the cDNA cloning of new human and mouse members of the CNT transporter family. The encoded proteins, designated hCNT3 and mCNT3, respectively, exhibit strong cib-type functional activity when expressed in Xenopus oocytes and have primary structures that place them together with hfCNT in a CNT subfamily separate from h/rCNT1/2.

EXPERIMENTAL PROCEDURES
Molecular Cloning of hCNT3-BLAST searches of CNT sequences in the GenBank TM data base identified overlapping human ESTs from mammary gland (AI905993) and colon adenocarcinoma (AW083022) different from established members of the CNT transporter family. Together, they formed a composite cDNA fragment 807 bp in length with an open reading frame of 245 residues followed by 69 bp of 3Ј-untranslated sequence. The cDNA was 62% identical in nucleotide sequence to corresponding regions of the hCNT1 (U62968) and hCNT2 (AF036109) cDNAs and 68% identical to the hfCNT (AF132298) cDNA. The encoded amino acid sequence was 79% identical to the carboxyl terminus of hfCNT and 58 and 62% identical, respectively, to hCNT1 and hCNT2.
These indications of a novel human CNT distinct from hCNT1 and hCNT2 were tested by RT-PCR in a panel of total RNA samples from human mammary gland, small intestine, kidney (CLONTECH, Palo Alto, CA), and liver (13). Because the close sequence similarity between the EST composite sequence and hfCNT suggested that the new CNT might correspond to system cib, we also performed RT-PCR on differentiated human myeloid HL-60 cells, a source of functional cib-type transport activity (see below). First strand cDNA was synthesized using the Superscript Preamplification system (Life Technologies, Inc.) and oligo(dT) as primer. The PCR reaction (30 l) contained 50 ng of template first-strand cDNA, 2.5 units of Taq-DeepVent DNA polymerase (100:1) and 10 pmol each of the 5Ј-and 3Ј-oligonucleotide primers 5Ј-GAAACATGTTTGACTACCCACAG-3Ј and 5Ј-GTGGAGTTGAAGG-CATTCTCTAAAACGT-3Ј. Amplification for one cycle at 94°C for 55 s, 54°C for 55 s, and 72°C for 70 s, two cycles at 94°C for 55 s, 55°C for 55 s, and 72°C for 70 s, and 30 cycles at 94°C for 55 s, 58°C for 55 s, and 72°C for 70 s (Robocycler 40 Temperature Cycler, Stratagene, La Jolla, CA) generated visible PCR products of the predicted size (480 bp) from four of the samples (differentiated HL-60 cells, mammary gland, small intestine, and liver).
We extended the partial EST cDNA sequence by 5Ј-rapid amplification of cDNA ends amplification of mRNA from differentiated HL-60 cells using the FirstChoice RLM-RACE kit (Ambion, Austin, TX). Poly(A) ϩ -selected RNA was treated with calf intestinal phosphatase to degrade 5Ј-truncated transcripts, followed by tobacco acid pyrophosphatase to remove cap from the remaining full-length mRNAs. A synthetic RNA adaptor from the kit was then ligated to the full-length 5Ј-monophosphate transcript population using T4 RNA ligase, followed by first strand cDNA synthesis with oligo(dT) as primer. For the initial PCR, the 5Ј-primer was the outer adaptor primer provided by the kit and the gene-specific 3Ј-primer was 5Ј-GATATATATTGCTGCACAC-CGTTTACAA-3Ј. Amplification by Taq-DeepVent DNA polymerase (100:1) was for 40 cycles at 94°C for 55 s, 65°C for 55 s, and 72°C for 3 min and 1 cycle at 72°C for 10 min, the reaction mixture being heated to 94°C for 1 min before addition of the Taq-DeepVent DNA polymerase mixture. The PCR reaction mixture was resolved on a 1% agarose gel, and faint bands between 1.5 and 2.0 kb in size were isolated and purified (QIAEX II Gel Extraction kit; Qiagen Inc.). This product was then reamplified by nested PCR (35 cycles at 94°C for 55 s, 65°C for 55 s, and 72°C for 3 min and 1 cycle at 72°C for 10 min) using an inner 5Ј-primer from the kit and the gene-specific 3Ј-primer 5Ј-TTAGCT-CAAAACTCAGCTGTGGGTAGTC-3Ј. A defined band of ϳ1.7 kb was isolated, cloned into pGEM-T (Promega, Madison, WI), and sequenced by Taq DyeDeoxyterminator cycle sequencing using an automated model 373A DNA Sequencer (Applied Biosystems, Foster City, CA). The inset overlapped the 807-bp EST sequence by 114 bp and generated an additional 1633 bp of upstream sequence. The new composite 2440-bp sequence was 66% identical to the hfCNT cDNA and contained an open reading frame of 691 amino acids. cDNAs containing the complete coding sequence were then obtained by RT-PCR from differentiated HL-60 cells and mammary gland, as described previously, using 5Ј-and 3Ј-primers flanking the open reading frame (5Ј-CTAAATGAAGAGCG-CTTGGGACCT-3Ј and 5Ј-AGCATCTGTACTTCAGAGTTCCACTGG-3Ј). The resulting ϳ2.2-kb products were ligated into pGEM-T and sequenced in both directions to give identical 691-residue open reading frames flanked by 92 bp of 5Ј-untranslated nucleotide sequence and 41 bp of 3Ј-untranslated sequence.
As expected from their identical nucleotide and predicted amino acid sequences, there was no difference in hCNT3 transport function between cDNA clones isolated from HL-60 cells or mammary gland. Radioisotope transport studies reported in this paper were performed with the HL-60 clone in pGEM-T.
Molecular Cloning of mCNT3-BLAST searches of mouse ESTs in the GenBank TM data base identified 630-and 635-bp sequences from two mammary gland IMAGE clones with 73 and 83% sequence identity to parts of the hCNT3 cDNA sequence. IMAGE clone 1514965 aligned with the 5Ј-coding region, whereas 1515408 ended 54 bp short of the predicted stop codon. Both clones were obtained from the IMAGE Consortium through the American Type Culture Collection (Manassas, VA). PCR showed that they were incomplete, and sequencing of 1515408 gave an additional 85 bp of sequence to complete the 3Ј-end of the open reading frame. A cDNA with the complete coding sequence was then obtained by RT-PCR from mouse liver RNA (Jackson Laboratories, Bar Harbor, ME) with 5Ј-primer 5Ј-AGGATGTCCAGGGCAGACCCG-GGAAAGA-3Ј and 3Ј-primer 5Ј-AGATCACAATTTATTAGGGATCCAA-TTG-3Ј. First strand cDNA was synthesized using the Thermoscript RT-PCR System (Life Technologies, Inc.), and amplification by Taq-DeepVent DNA polymerase (100:1) was for 2 cycles at 94°C for 2 min, 64°C for 1 min, and 72°C for 2.5 min, 2 cycles at 94°C for 1 min, 62°C for 1 min, and 72°C for 2.5 min, 30 cycles at 94°C for 1 min, 60°C for 1 min, and 72°C for 2.5 min, and one final extension cycle for 10 min at 72°C. The resulting ϳ2.0-kb product was ligated into pGEM-T and subcloned into the enhanced Xenopus expression vector pGEM-HE (24). Each was sequenced in both directions, giving identical 703-amino acid residue open reading frames flanked by short 3-bp regions of 5Ј-untranslated or 3Ј-nucleotide sequence. By providing additional 5Ј-and 3Ј-untranslated sequences from a Xenopus ␤-globin gene, the pGEM-HE construct gave greater functional activity and was used in subsequent transport characterization of the mouse protein. pGEM-HE was also used for electrophysiological studies of hCNT3.
Expression of Recombinant hCNT3 and mCNT3 in Xenopus Oocytes-hCNT3 and mCNT3 plasmid DNAs were linearized with NotI (pGEM-T) or NheI (pGEM-HE) and transcribed with T7 polymerase mMESSAGE mMACHINE TM (Ambion). Stage VI oocytes of Xenopus laevis (8) were microinjected with 20 nl of water or 20 nl of water containing capped RNA transcripts (20 ng) and incubated in modified Barth's medium (changed daily) at 18°C for 72 h prior to the assay of transport activity.
hCNT3 and mCNT3 Radioisotope Flux Studies-Transport was traced using the appropriate 14 C/ 3 H-labeled nucleoside, nucleoside drug, or nucleobase (Moravek Biochemicals, Brea, CA or Amersham Pharmacia Biotech) at a concentration of 1 and 2 Ci/ml for 14 C-labeled and 3 H-labeled compounds, respectively. [ 3 H]Gemcitabine (2Ј,3Ј-difluorodeoxycytidine) was a gift from Eli Lilly Inc. (Indianapolis, IN). Radiochemicals were 98 -99% pure (see HL-60 transport studies). Flux measurements were performed at room temperature (20°C) as described previously (8, 11) on groups of 12 oocytes in 200 l of transport medium containing 100 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES, pH 7.5. Except where otherwise indicated, the nucleoside concentration was 20 M. At the end of the incubation period, extracellular label was removed by six rapid washes in ice-cold transport medium, and individual oocytes were dissolved in 5% (w/v) SDS for quantitation of oocyte-associated radioactivity by liquid scintillation counting (LS 6000 IC; Beckman). Initial rates of transport (influx) were determined using an incubation period of 5 min (8). Choline replaced sodium in Na ϩ dependence experiments, and the transport medium for adenosine uptake contained 1 M deoxycoformycin to inhibit adenosine deaminase activity. The flux values shown are the means Ϯ S.E. of 10 -12 oocytes, and each experiment was performed at least twice on different batches of cells. Kinetic (K m and V max ) and Na ϩ activation parameters (K 50 and Hill coefficient) Ϯ S.E. were determined using ENZFITTER (Elsevier-Biosoft, Cambridge, UK) and SigmaPlot (SPSS Inc., Chicago, IL) software, respectively.
Measurement of hCNT3-induced Sodium Currents-Oocytes were voltage clamped using the two-electrode voltage clamp. Membrane currents were measured at room temperature by use of a GeneClamp 500B oocyte clamp (Axon Instruments, Foster City, CA). The microelectrodes were filled with 3 M KCl and had resistances that ranged from 1-2.5 M⍀. The GeneClamp 500B was interfaced to a computer via a Digidata 1200 A/D convertor and controlled by Axoscope software (Axon Instruments, Foster City, CA). Current signals were filtered at 20 Hz (four-pole Bessel filter) and sampled at intervals of 20 ms. For data presentation, the signals were further filtered at 0.5 Hz by use of pCLAMP software (Axon Instruments). Cells were not used if resting membrane potentials were unstable or less than Ϫ30 mV. For measurements of hCNT3generated currents, oocyte membrane potentials were clamped at Ϫ50 mV. Oocytes were perfused with the same medium used for radioisotope flux studies, and transport assays were initiated by changing the substrate-free solution to one containing nucleoside (200 M). In experiments examining Na ϩ dependence, sodium in the medium was replaced by choline.
HL-60 Cell Culture and Differentiation-The human promyelocytic cell line, HL-60, obtained from the American Type Culture Collection, was propagated as suspension cultures in RPMI 1640 medium, supplemented with 10% fetal calf serum using reagents purchased from Life Technologies, Inc. Stock cultures were maintained in 5% CO 2 without antibiotics at 37°C, subcultured every 3-4 days and demonstrated to be mycoplasma-free. Cell numbers were determined using a Coulter Counter model Z2 (Coulter Electronics Inc., Luton, UK).
To induce differentiation, HL-60 cells (3 ϫ 10 6 ) growing in logarithmic phase were placed in 10-cm Falcon Primaria tissue culture plates (Becton Dickinson) in the presence of phorbol 12-myristate 13-acetate (200 ng/ml) (Sigma) freshly dissolved in acetone. After 48 h, the plates were washed once with transport buffer (see below) to remove nonad-herent cells and then incubated for 15 min in the presence or absence of 100 M dilazep. Transport assays were performed on the remaining adherent cells. Total RNA and mRNA were prepared from exponentially growing parent and adherent HL-60 cells using the RNeasy Mini Protocol (Qiagen) and Fast Track 2.0 Isolation kit (Invitrogen, Carlsbad, CA), respectively.
HL-60 Radioisiotope Flux Studies-Nucleoside uptake by differentiated HL-60 cells was measured as described previously (25) by exposing replicate cultures at room temperature to 3 H-labeled permeant (10 M, 1 Ci/ml) in sodium or sodium-free transport medium (130 mM NaCl or 130 mM NMDG/HCl and 3 mM K 2 HPO 4 , 2 mM CaCl 2 , 1 mM MgCl 2 , 20 mM Tris/HCl, and 5 mM glucose, pH 7.4). Radiochemicals (Moravek Biochemicals) were 98 -99% pure as assessed by high performance liquid chromatography using water-methanol gradients on a C18 reverse phase column, and transport for timed intervals of 1-6 min was terminated by immersion of the culture dish in an excess volume of ice-cold transport solution. Assays to detect concentrative transport were performed in the presence of 100 M dilazep (a gift from Hoffman La Roche & Co., Basel, Switzerland) to block equilibrative transport of the test nucleoside. Transport by nonadherent parental HL-60 cells was performed as described previously (26) using the inhibitor oil stop method. Values are presented as the means of triplicate measurements Ϯ S.D.
Tissue and Cell Distribution of hCNT3 mRNA-A human multiple tissue expression (MTE TM ) RNA array (CLONTECH) and dot blots of mRNA (0.5 g) from parent and differentiated HL-60 cells on Bright-Star-Plus nylon transfer membrane (Ambion) were incubated with a cDNA probe corresponding to hCNT3 amino acid residues 359 -549 labeled with 32 P using the T7 QuickPrime kit (Amersham Pharmacia Biotech). Hybridization at high stringency (68°C) was performed using ExpressHyb hybridization solution (CLONTECH) and 100 g/ml of sheared herring sperm DNA. Wash conditions were as described in the CLONTECH ExpressHyb user manual. Signals on exposed blots were converted to a high resolution tiff image (Hewlett Packard ScanJet 4C) and quantified using the public domain NIH Image program, version 1.60. For Northern analysis, 5-g samples of mRNA from human pancreas, bone marrow, trachea, intestine, liver, brain, heart, and kidney (CLONTECH) were separated on a 0.8% formaldehyde-agarose gel, blotted on to BrightStar-Plus nylon transfer membrane, and hybridized with the same hCNT3 probe (residues 359 -549) under identical high stringency conditions.
Possible cross-hybridization between CNT family members was tested on dot blots of dilutions (0.5 g-5 ng RNA) of hCNT1, hCNT2, and hCNT3 in vitro transcripts. Three identical series of blots were incubated either with hCNT3 probe or with equivalent probes for hCNT1 or hCNT2. The hCNT3 probe, which was 63 and 58% identical in nucleotide sequence to the corresponding regions of hCNT1 and hCNT2, respectively, showed no cross-hybridization with hCNT1 or hCNT2 transcripts. Similarly, there was no cross-reactivity between the hCNT1 and hCNT2 probes and hCNT3 RNA. Some cross-hybridization was seen between the hCNT1 and hCNT2 probes (73% nucleotide sequence identity) and their respective transcripts at RNA loadings Ն50 ng. Under the conditions of high stringency used in our experiments, the hCNT3 probe was therefore specific.
Quantitative Real Time RT-PCR-In TaqMan TM quantitative RT-PCR (Applied Biosystems), an oligonucleotide probe, labeled with a fluorescent tag at the 5Ј-end and a quenching molecule at the 3Ј-end, is located between two PCR primers. The 5Ј-nucleotidase activity of Taq polymerase cleaves the fluorescent dye from the probe during each PCR cycle. The fluorescent signal generated is monitored in real time and is proportional to the amount of starting template in the sample.
RNA from parent or differentiated HL-60 cells was reverse transcribed using the TaqMan TM Gold RT-PCR kit (Applied Biosystems) and subjected to real time PCR using an Applied Biosystems PRISM 7700 Sequence Detection System and TaqMan TM Universal PCR Master Mix kit. Amplification conditions were a single cycle at 50°C for 2 min and 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 1 min at 60°C using hCNT3 probe and primers designed using Primer Express software (Applied Biosystems). The hCNT3 probe 5Ј-6-carboxyfluorescein-CGGACTCACATCCATGGCTCCTTC-6-carboxy-tetramethylrhodamine-3Ј was purchased from Applied Biosystems, whereas the 5Јand 3Ј-primers were 5Ј-GGGTCCCTAGGAATCGTGATC-3Ј and 5Ј-CGAGGCGATATCACGCTTTC-3Ј, respectively. GADPH and 18 S ribosomal RNA probes and primers, used as internal controls, were purchased as a TaqMan TM RNA Control Reagent kit. Relative quantification of hCNT3 message was determined as described previously (27).
Chromosomal Fluorescence in Situ Hybridization-Analysis of normal human lymphocyte metaphase chromosomes was performed by methods described previously (28) using a PCR probe corresponding to hCNT3 amino acid residues 86 -685. Chromosomal localization of the gene was also determined by screening an RPCI-11 human male BAC library (29).

RESULTS AND DISCUSSION
Membrane transport studies in various human and other mammalian cell and tissue preparations have produced evidence that concentrative (Na ϩ -linked) cellular uptake of nucleosides and nucleoside drugs is mediated by at least three distinct mechanisms (3,4). Systems cit and cif are found primarily in specialized epithelia such as intestine, liver, and kidney and have characteristic, overlapping substrate specificities for pyrimidine and purine nucleosides, respectively. As well, a broadly selective transport activity for both pyrimidine and purine nucleosides (system cib) has been described (3,4). Expression cloning and other recombinant DNA strategies have recently established that systems cit and cif are mediated by isoforms of the CNT transporter family, designated in humans as hCNT1 and hCNT2, respectively (8 -13). In the present study, we report the molecular identification and characterization of the human and mouse cib transporters and demonstrate that they represent a new mammalian isoform of the CNT transporter family (hCNT3 and mCNT3, respectively).
Molecular Identification of hCNT3 and mCNT3-The possibility that cib might be a CNT transporter (18) led to identification of ESTs encoding partially overlapping regions at the carboxyl terminus of a new, previously unrecognized human CNT distinct from hCNT1 or hCNT2. The full-length cDNA obtained by 5Ј-rapid amplification of cDNA ends/RT-PCR amplification of phorbol 12-myristate 13-acetate-differentiated human myeloid HL-60 cells and by RT-PCR of human mammary gland encoded a 691-residue protein (77 kDa), designated here as hCNT3. 5 Differentiated HL-60 cells represent a functionally defined source of cib transport activity (see below), whereas human mammary gland was the origin of one of the carboxyl terminus ESTs. The hCNT3 sequence enabled us, in turn, to identify ESTs from mouse mammary gland encoding the amino-and carboxyl terminus ends of a mouse homolog. The corresponding full-length mouse cDNA, obtained by RT-PCR from liver (also a source of hCNT3 transcript), encoded a 703-residue protein designated here as mCNT3. 5 hCNT3 and mCNT3 Amino Acid Sequences-hCNT3 was 57% identical in amino acid sequence to hfCNT and 48 and 47% identical to hCNT1 and hCNT2, respectively (Fig. 1A). Protein structure algorithms predicted a topology for hCNT3 similar to that of hfCNT and hCNT1/2 (20), with relatively large extramembraneous amino and carboxyl termini (carboxyl terminus external) linked by 13 TMs and short hydrophilic sequences (Յ22 residues), with the exception of larger extracellular loops between TMs 5-6, 9 -10, and 11-12 (Fig.  1B). Residues within TMs 4 -13 were particularly highly conserved between hCNT3 and hfCNT (67% sequence identity), whereas TMs 1-3 and the amino and carboxyl termini were much more divergent. The conserved TM 4 -13 domains of hCNT3 and hfCNT corresponded closely to the predicted membrane architecture of the shorter E. coli CNT proton/nucleoside cotransporter NupC (19), suggesting that these regions represent the functionally important core structure of the proteins. We engineered an amino-terminal truncated form of rCNT1 and established that the TM 1-3 region is not required for transport activity. 6 mCNT3 contained additional amino acids at the amino terminus (Fig. 1A) and was 78% identical in sequence to hCNT3, 57% identical to hfCNT, and 48% identical to mCNT2 (AF079853), the other known mouse CNT.
Since we first identified rCNT1 from rat jejunum by expression selection in Xenopus oocytes in 1994, more than 40 members of the CNT protein family have been identified from mammals, lower vertebrates, insects, nematodes, pathogenic yeast, and bacteria. As shown in Fig. 1C, phylogenetic analysis identified discrete clusters of proteins, including two for bacteria and one for vertebrate transporters. hCNT3 and mCNT3 were placed together with hfCNT in a different vertebrate CNT subfamily from the human and other mammalian CNT1 and CNT2 proteins. Characteristically conserved motifs of the CNT transporter family present in hCNT3 and mCNT3 included GX 21 GX 2 FXFG between TMs 5 and 6, (G/A)XKX 3  In common with other CNTs, hCNT3 and mCNT3 also contained multiple consensus sites for N-linked glycosylation, grouped at the carboxyl terminus (hCNT3 Asn 636 and Asn 664 , mCNT3 Asn 648 and Asn 676 ). The extracellular location of this region has been confirmed by mutagenesis of rCNT1, which is glycosylated at Asn 605 and Asn 643 (20).
Previously (18), we have identified two adjacent pairs of residues (Ser 319 /Gln 320 and Ser 353 /Leu 354 ) in the TM 7-9 region of hCNT1 that, when mutated together to the corresponding residues in hCNT2 (Gly 313 /Met 314 and Thr 347 /Val 348 ), converted hCNT1 (cit) into a transporter with cif-type functional characteristics. An intermediate broad specificity cib-like transport activity was produced by mutation of the two TM 7 residues alone; mutation of Ser 319 to Gly allowed for transport of purine nucleosides, and this was augmented by mutation of Gln 320 to Met. Mutation of Ser 353 in TM 8 to Thr converted the cib-like transport of the TM 7 double mutant into one with cif-like characteristics but with relatively low adenosine transport activity. Mutation of Leu 354 to Val increased the adenosine transport capability of the TM 7/8 triple mutant, producing a full cif transport phenotype. On its own, mutation of Ser 353 converted hCNT1 into a transporter with novel uridine-selective transport properties. The sequences of hCNT3 and mCNT3 at these positions were intermediate between hCNT1 and hCNT2, one member of each pair of residues being identical to the corresponding residue in hCNT1 and the other to that in hCNT2. These sequences in hCNT3 and mCNT3 were identical to hfCNT (Gly 340 /Gln 341 and Ser 374 /Val 375 in the case of hCNT3).
Functional Expression and Substrate Specificity of Recombinant hCNT3 and mCNT3-hCNT1 and hCNT2 display cit-and cif-type Na ϩ -dependent nucleoside transport activities (11,13). Therefore, although both hCNT1 and hCNT2 transport uridine and certain uridine analogs, they are otherwise selective for pyrimidine (hCNT1) and purine (hCNT2) nucleosides (except for modest transport of adenosine by hCNT1). hfCNT, in contrast, exhibits cib-type Na ϩ -dependent nucleoside transport activity and is broadly selective for both pyrimidine and purine nucleosides. Fig. 2A shows a representative transport experiment in Xenopus oocytes measuring uptake of uridine and a panel of other radiolabeled pyrimidine and purine nucleosides (cytidine, thymidine, adenosine, guanosine, and inosine) and nucleobases (uracil and hypoxanthine) in cells injected with water alone (control) or with water containing hCNT3 transcripts. Uptake of uridine (20 M, 30-min flux) by hCNT3-expressing oocytes was Na ϩ -dependent (60.7 Ϯ 4.5 and 6.1 Ϯ 0.7 pmol/oocyte in 5 GenBank TM /EBI Data Bank accession numbers AF305210 and AF305211. 6  Na ϩ and choline medium, respectively) and concentrative (60.7 pmol/oocyte corresponds to an in-to-out concentration ratio of ϳ3:1, calculated assuming an oocyte water content of 1 l). In Na ϩ medium, uridine uptake in control water-injected oocytes was only 0.5 Ϯ 0.1 pmol/oocyte, giving a mediated flux (uptake by RNA-injected oocytes minus uptake in water-injected oocytes) of 60.2 pmol/oocyte and a mediated-to-basal flux ratio of 120:1. Consistent with cib-type functional activity, each of the other pyrimidine and purine nucleosides tested (cytidine, thymidine, adenosine, guanosine, and inosine) gave similar mediated fluxes. mCNT3 (Fig. 2B) exhibited a similar pattern of Na ϩ -dependent cib-type functional activity, and neither protein transported uracil or hypoxanthine. Fig. 3 compares the differences in substrate specificity between hCNT3, mCNT3, hfCNT, hCNT1, and hCNT2 by measuring the mediated uptake of three diagnostic nucleoside permeants (uridine, thymidine, and inosine). All five proteins transported uridine. However, hCNT1 (cit) exhibited pyrimidine nucleoside selective characteristics (marked thymidine uptake, low inosine transport), whereas hCNT2 (cif) was purine nucleoside selective (low thymidine uptake, marked in-osine transport). hCNT3, mCNT3, and hfCNT exhibited similar cib-type profiles, with marked transport of both thymidine and inosine. Subsequent in depth transport experiments focussed on the human transporter hCNT3.
Kinetic Properties and Inhibitor Sensitivity of Recombinant hCNT3- Fig. 4 shows representative concentration dependence curves for uridine, cytidine, thymidine, adenosine, guanosine, and inosine, measured as initial rates of transport (5-min flux) in hCNT3-expressing oocytes and in control waterinjected cells. Kinetic constants for the hCNT3-mediated component of influx are presented in Table I. K m values varied between 15 and 53 M (cytidine, adenosine Ͻ uridine, thymidine Ͻ guanosine, inosine) and were within the range expected for native cib-type transporters (30 -32) and for hfCNT in oocytes (17-54 M). 3 They were also similar to K m values obtained previously for permeants of recombinant mammalian CNT1/2 transporters. For example, the hCNT3 K m for uridine was 22 M compared with 37-45 M for hCNT1, rCNT1, and hCNT2 (8,11,13). hCNT3 K m values for thymidine and inosine were 21 and 53 M, respectively, compared with 13 M for thymidine transport by rCNT1 (33)  transport by rCNT2 (21). hCNT3 V max values were in the range 24 and 51 pmol/oocyte.5 min Ϫ1 (uridine, thymidine Ͻ cytidine, adenosine Ͻ guanosine, inosine), giving V max :K m ratios of 0.9 -2.1 (Table I). These data support the cib-type specificity profile of hCNT3 shown in Fig. 2A and demonstrate that hCNT3 transports different pyrimidine and purine nucleosides with very similar efficiencies. For all of the nucleosides tested (Fig.  4), influx in water-injected oocytes was linear with concentration, consistent with nonmediated simple diffusion through the lipid bilayer.
In addition to the three major mammalian concentrative nucleoside transport systems cit, cif, and cib, there are two minor Na ϩ -dependent nucleoside transport processes, csg and cs, which have been described only in leukemic cells (34,35). Although their permeant preferences have not been well defined, the csg process (34) accepts guanosine, and the cs process (35) accepts adenosine analogs as permeants. In contrast to cit, cif, and cib, both are inhibited by nanomolar concentrations of NBMPR (34,35). hCNT3 was unaffected by NBMPR or other equilibrative nucleoside transport inhibitors, dipyridamole and dilazep, at concentrations up to 10 M (100 M for dilazep, which is more soluble), eliminating hCNT3 as a possible contributor to csg or cs transport activity (Fig. 5).
hCNT3 Na ϩ :Nucleoside Cotransport-Na ϩ /nucleoside coupling ratios of 1:1 have been described for various cit and cif transport activities in different mammalian cells and tissues (reviewed in Ref. 3). In contrast, a coupling ratio of 2:1 has been reported for system cib in choroid plexus and microglia (30,31). In Fig. 6 (A and C), we show for both hCNT3 and mCNT3 that the relationship between uridine influx (10 M) and Na ϩ concentration was sigmoidal. Fitting the data to the Hill equation, v ϭ V max ⅐[Na ϩ ] n /(K 50 n ϩ [Na ϩ ] n ), gave Hill coefficients (n) of 2.2 Ϯ 0.2 (hCNT3) and 2.3 Ϯ 0.1 (mCNT3), indicating a Na ϩ / nucleoside coupling ratio of at least 2:1. Similar values of n (2.0 Ϯ 0.2 and 2.0 Ϯ 0.1, respectively) were determined from the slopes of Hill plots of the data (Fig. 6, B and D), and in five independent experiments, Hill plot transformations gave a mean hCNT3 Hill coefficient of 2.1 Ϯ 0.3. Because rCNT1 exhibited a Hill coefficient of 1 in similar experiments (10), our data establish, for the first time, that the stoichiometry of Na ϩ /nucleoside coupling is different in different CNT family members. In this respect, the CNTs resemble the SGLT glucose transporter family, where examples of proteins with 2:1 and 1:1 Na ϩ /sugar coupling ratios (SGLT1 and SGLT2, respectively) have been described (36 -39). Similarly, the PepT2 and PepT1 proton-linked peptide transporters have 2:1 and 1:1 H ϩ /peptide coupling ratios, respectively (40). There was an interesting difference in K 50 values for Na ϩ activation between hCNT3 and mCNT3 (16 Ϯ 1 and 7 Ϯ 1 mM, respectively), although both transporters were fully saturated with Na ϩ at cation concentrations approaching the physiological concentration range (Fig. 6, A and C).
In addition to radioisotope flux studies, we also used the two-electrode voltage clamp technique to investigate the Na ϩ dependence of hCNT3-mediated nucleoside transport. As shown in Fig. 7, external application of uridine, thymidine, and inosine (200 M) to oocytes expressing recombinant hCNT3 induced inward currents for all three nucleosides that returned to baseline upon removal of permeant. No currents were seen in water-injected oocytes or when Na ϩ in the extracellular medium was replaced by choline, demonstrating that hCNT3 functions as a broad specificity electrogenic Na ϩ /nucleoside symporter. In parallel with the radioisotope transport data shown in Fig. 6 (A and C), there was a sigmoid relationship between uridine-evoked current and Na ϩ concentration (data not shown). In preliminary experiments to determine directly the Na ϩ /nucleoside coupling ratio by simultaneous measurement of Na ϩ currents and [ 14 C]uridine influx under voltage clamp conditions, as described previously for the SDCT1 rat kidney dicarboxylate transporter (41), we have confirmed that hCNT3 has a Na ϩ /nucleoside coupling ratio Ն2:1, whereas for hCNT1 the ratio is 1:1. 6 A two-Na ϩ /one-nucleoside symporter FIG. 1-continued will have a greater ability to transport permeant against its concentration gradient than a one-Na ϩ /one-nucleoside symporter, and they may have evolved to transport nucleosides under different conditions. Experiments are in progress with hfCNT, mCNT3, and other CNTs to determine whether the 2:1 stoichiometry is limited to members of the CNT3/hfCNT subfamily ( Fig. 2A) or has a more widespread distribution.
hCNT3-mediated Transport of Anticancer and Antiviral Nucleoside Drugs-The difference in substrate specificity between CNT1 and CNT2 for physiological pyrimidine and purine nucleosides is reflected in their complementary roles for transport of pyrimidine and purine antiviral and anticancer nucleoside drugs. For example, we have used Xenopus oocyte expression to establish that mammalian CNT1/2 proteins transport antiviral dideoxynucleosides: h/rCNT1 transports the AIDS drugs 3Ј-azido-3Ј-deoxythymidine (AZT) and 2Ј,3Ј-dideoxycytidine but not 2Ј,3Ј-dideoxyinosine, whereas hCNT2 transports only 2Ј,3Ј-dideoxyinosine (8,11,13,42). Gemcitabine, a cytidine analog used in therapy of solid tumors, is a good hCNT1 permeant but is not transported by hCNT2 either in oocytes (43) or in transfected HeLa cells (44). As shown in Fig. 8,   hCNT3, a cib-type NT, efficiently transported both pyrimidine (5-fluorouridine, 5-fluoro-2Ј-deoxyuridine, zebularine, and gemcitabine) and purine (cladribine and fludarabine) anticancer nucleoside drugs. Lower, but still significant, uptake was observed for pyrimidine (AZT and 2Ј,3Ј-dideoxycytidine) and purine (2Ј,3Ј-dideoxyinosine) antiviral nucleoside drugs, the magnitudes of the fluxes being similar to those found previously for hCNT1 (AZT and 2Ј,3Ј-dideoxycytidine) and hCNT2 (2Ј,3Ј-dideoxyinosine) (11,13). Only ganciclovir, an antiviral drug with an acyclic ribose moiety, was not transported. Therefore, by virtue of its ability to transport both pyrimidine and purine nucleosides, hCNT3 is capable of transporting a broader range of therapeutic nucleosides than either hCNT1 or hCNT2. Consistent with the present results, thymidine transport by the microglial cib transporter was inhibited by AZT (31).
Tissue and Cell Distribution-The cib process has been described functionally in rabbit choroid plexus (30), rat MSL-9 microglia cells (31), Xenopus oocytes injected with rat jejunal mRNA (32), human leukemic (45) and colorectal carcinoma CaCo cells (46) and, after induction of differentiation, in human HL-60 cells (see below). The human and mouse ESTs that led to the identification of h/mCNT3 were from human/mouse mammary gland and human colon adenocarcinoma, whereas the full-length transporter cDNAs were isolated from differentiated HL-60 cells and mammary gland (hCNT3) and liver (mCNT3). Fig. 9 shows a multiple tissue expression RNA array for 76 human tissues and cells probed with hCNT3 cDNA (a second commercial RNA array from the same supplier gave essentially identical results). As described under "Experimental Procedures," this analysis was performed under conditions of high stringency where there was no cross-hybridization between hCNT3 and hCNT1 or hCNT2. The distribution pattern of hCNT3 transcripts, although selective, was surprisingly widespread. Highest levels were found in a number of normal tissues, including mammary gland, pancreas, bone marrow, and trachea, with substantial levels in various regions of the intestine (but very much less in kidney) and more modest levels in liver, lung, placenta, prostrate, testis, and other tissues, including some regions of the brain and heart. hCNT3 transcripts were generally present in fetal tissues but were low in various cultured cell lines, including K562, HeLa, and undifferentiated HL-60 (see also below). In contrast, h/rCNT1 and h/rCNT2 transcripts are found primarily in specialized epithe-

FIG. 2. Uptake of 14 C/ 3 H-labeled nucleosides and nucleobases by recombinant hCNT3 and mCNT3 expressed in Xenopus oocytes.
Uptake of nucleosides and nucleobases (20 M, 20°C, 30 min) in oocytes injected with RNA transcripts or water alone was measured in transport medium containing 100 mM NaCl or 100 mM choline chloride (ChCl).
lia, including small intestine, kidney, and liver (8 -13). Other reported sources of h/rCNT1 and h/rCNT2 transcripts include brain, spleen, heart, pancreas, and skeletal muscle (9,12,47). A systematic analysis of CNT1/2 transcript distribution similar to that shown in Fig. 9 for hCNT3 would be helpful to more fully characterize the different expression patterns of the three transporters.
In parallel with the multiple tissue expression RNA array, we also investigated the distribution of hCNT3 transcript in selected tissues by Northern blotting. This less sensitive technique detected hCNT3 transcripts in pancreas, bone marrow, and trachea but not in intestine, liver, brain, or heart (Fig. 10). Kidney, as expected from Fig. 9, was also negative. In pancreas, bone marrow, and trachea, three bands were apparent: a major 5.3-kb transcript and secondary bands at 6.5 and 4.8 kb. Although two of the bands were similar in size to the major transcripts of hCNT1 (3.4 kb) and hCNT2 (4.5 -kb) (13), the blot was probed at high stringency under conditions where there was no cross-reactivity with hCNT1/2 (see "Experimental Procedures"). It is likely, therefore, that they represent alternate hCNT3 gene transcripts rather than cross-hybridization with other CNT family members. The absence of bands in kidney, which contains transcripts for both hCNT1 and hCNT2 (13), is further evidence of the specificity of the hCNT3 probe. The same tissues (plus mammary gland) were also analyzed by TaqMan TM quantitative RT-PCR using hCNT3-specific primers as described below for HL-60 cells. Relative levels of hCNT3 transcript by this method were pancreas Ͼ bone marrow, trachea, intestine Ͼ mammary gland Ͼ Ͼ liver, brain, heart Ͼ kidney (data not shown).
HL-60 Cells (Functional Studies)-The human promyelocytic leukemia cell line HL-60 can be induced to differentiate into adherent monocyte/macrophage-like cells treatment with phorbol esters (48,49). Upon differentiation, the cells exhibit a decrease in equilibrative, Na ϩ -independent nucleoside transport that is accompanied by an increase in concentrative, Na ϩdependent transport of both pyrimidine and purine nucleosides (50). To rigorously identify the concentrative transport process-(es) contributing to this uptake, we first determined the uridine transport profile of parent and phorbol 12-myristate 13-acetate-treated HL-60 cells under conditions previously shown to be optimal for induction of the concentrative transport activity (50). Equilibrative transport was measured by replacing Na ϩ in the transport medium with NMDG, and concentrative transport was determined in the presence of Na ϩ but with the addition of dilazep (100 M) to inhibit equilibrative transport activity. We have shown that this concentration of dilazep has no effect on hCNT3 transport activity (Fig. 5). Although previ-  were measured in transport medium containing 100 mM NaCl. Kinetic parameters calculated from the mediated component of transport (uptake in RNA-injected oocytes minus uptake in water-injected oocytes) are presented in Table I. ous studies have reported a small amount of Na ϩ -dependent transport activity in untreated HL-60 cells (50,51), our assays did not detect any concentrative transport in the parent cell line, which exhibited only equilibrative uptake of 10 M uridine (Fig. 11A). However, there was a notable increase in Na ϩ -dependent uridine transport in differentiated, adherent HL-60 cells (Fig. 11B). Uptake of thymidine and formycin B (a nonmetabolized analog of inosine) was then used to define which of the concentrative transport processes (cit, cif, cib) was active in differentiated HL-60 cells (Fig. 11, C and D). Both nucleosides were taken up by the concentrative process(es) of differentiated HL-60 cells. Transport of thymidine was totally inhibited by unlabeled thymidine, inosine, and uridine, whereas formycin B uptake was reciprocally inhibited by thymidine. Thus, cib (rather than cit ϩ cif) was the dominant concentrative transport activity in differentiated HL-60 cells. Consistent with this result, concentrative uridine transport was inhibited by uri-dine, thymidine, and inosine (data not shown). Dot blot analysis (Fig. 9A) and nonquantitative RT-PCR (Fig. 12A) established that the appearance of cib functional activity correlated with substantially increased levels of hCNT3 transcripts in differentiated versus parent HL-60 cells, the latter exhibiting only small amounts of hCNT3 mRNA. In RT-PCR experiments, parent and differentiated HL-60 cells were negative for hCNT2 mRNA and expressed only very small amounts of hCNT1 mRNA, most likely as a consequence of bleed-through transcription (data not shown). These results provided further evidence that the concentrative nucleoside transport activity seen in differentiated HL-60 cells was mediated by cib and not by (cit ϩ cif).
HL-60 Cells (TaqMan TM Quantitative Real Time RT-PCR)-The relative levels of hCNT3 transcripts in HL-60 parent and differentiated cells were determined by quantitative real time RT-PCR (Fig. 12B). Glyceraldehyde-3-phosphate dehydrogen- FIG. 5. Recombinant hCNT3 is not inhibited by NBMPR, dipyridamole, or dilazep. Initial rates of transportermediated uridine uptake (20 M, 20°C, 5 min) were measured in transport medium containing 100 mM NaCl in the absence or presence of 1-10 M NBMPR and dipyridamole or 1-100 M dilazep. Oocytes were incubated with inhibitor for 30 min before the addition of permeant. Mediated transport was calculated as uptake in RNA-injected oocytes minus uptake in water-injected oocytes (uptake of uridine by water-injected oocytes was unaffected by NBMPR, dipyridamole, or dilazep).
FIG. 6. Sodium dependence of influx of uridine mediated by recombinant hCNT3 and mCNT3. Initial rates of transporter-mediated uptake of uridine (10 M, 20°C, 5 min) by hCNT3 (A) and mCNT3 (C) were measured in transport medium containing 0 -100 mM NaCl using choline chloride to maintain isosmolality. Mediated transport was calculated as uptake in RNA-injected oocytes minus uptake in water-injected oocytes (uptake of uridine by water-injected oocytes was not Na ϩ -dependent). B and D are Hill plots of the hCNT3 and mCNT3 data, respectively. K 50 values and Hill coefficients (n) are given in the text. Broken lines in B and D correspond to n values of 1. ase and hCNT3 were demonstrated to amplify with equal efficiency, and glyceraldehyde-3-phosphate dehydrogenase was therefore used as the internal control to normalize levels of expression of hCNT3 mRNA between samples. To compare samples, a threshold line was set at the phase of the PCR reaction during which the fluorescent signal accumulated exponentially. As shown in Fig. 12B, there was a substantial difference between the HL-60 parent and differentiated samples in the PCR cycle numbers at the threshold line, and three independent experiments gave a mean (Ϯ S.E.) ratio of 4.08 Ϯ 0.09, indicating (because PCR amplification is an exponential process) that there was 16.9 Ϯ 1.05 more hCNT3 mRNA in differentiated versus parent HL-60 cells. Similar results were obtained when the data were normalized to 18 S gene expression (data not shown). The human tumor cell lines K562 (erythroleukemia) and HeLa (cervical carcinoma) were also tested in this assay (data not shown) and gave signals that were close to background levels (see also Fig. 9A). These results, taken with those of the transport experiments, indicated that the small amount of hCNT3 transcription in the HL-60 parent cells did not result in enough protein to be functionally detected, whereas the differentiated cells that expressed 16-fold more hCNT3 mRNA had a readily measurable cib transport process. Because the analyses were performed on exponentially growing parent and differentiated cells, the difference in transcript levels between parent and differentiated HL-60 cells could not be attributed to cell proliferation.
Chromosomal Localization of the hCNT3 Gene-Although the genes encoding hCNT1 and hCNT2 have both been mapped to chromosome 15 (15q25-26 (11) and 15q15 (13), respectively), fluorescence in situ hybridization analysis localized the hCNT3 gene to 9q22.2. The same chromosomal band location was determined by screening a human BAC library. Searches of the Sanger Center human genomic sequence data base, and the Unfinished High Throughput Genomic Sequence GenBank TM data base identified two chromosome 9 clones (GenBank TM accession numbers AL356134 and AL353787) containing multiple hCNT3 genomic fragments that, when aligned, revealed 27.3-kb of composite hCNT3 gene sequence containing 74% of the hCNT3 coding sequence. The coding sequence that was obtained was an exact match with corresponding regions of the hCNT3 cDNA sequence. 5 Analysis of hCNT3 5Ј-genomic sequence in the potential upstream promoter region of the gene revealed the presence of a eukaryotic phorbol myristate acetate (ester) response element (52) with the sequence 5Ј-TGAGTCA-3Ј that may potentially contribute to the transcriptional regulation of hCNT3 seen in HL-60 cells. Studies are in progress to compare the organization of the hCNT3 gene with that for hCNT1 (32 kb), which has been determined to contain 18 exons separated by 17 introns (GenBank TM accession numbers 187967-187978).
Conclusions-The CNT protein family in humans is represented by three members, hCNT1, hCNT2, and the presently described hCNT3. Searches of the Unfinished High Throughput Genomic Sequence GenBank TM data base have so far revealed no other closely related members of this family in humans. 6 hCNT3 is a transcriptionally regulated electrogenic transport protein that, unlike hCNT1 and hCNT2, has a broad permeant selectivity for pyrimidine and purine nucleosides and FIG. 7. Sodium currents induced by exposure of recombinant hCNT3 to nucleoside permeants. Upper left panel, inward currents caused by perfusing an hCNT3-expressing oocyte at room temperature with 200 M inosine, uridine, or thymidine in Na ϩ -containing transport medium (NaCl). Upper right panel, the same oocyte perfused with 200 M inosine, uridine, or thymidine in transport medium with Na ϩ replaced by choline (ChCl). No inward currents were generated. Lower panels, the same experiments described for upper panels above but with a control water-injected oocyte. No inward currents were generated. nucleoside drugs. Hill-type analysis of the relationship between uridine influx and Na ϩ concentration indicated a Na ϩ : uridine coupling ratio of at least 2:1, compared with 1:1 for hCNT1/2. These characteristics and the induction of hCNT3 mRNA in HL-60 cells following phorbol ester treatment identified hCNT3 as the physiological human cib transporter. A mouse homolog of hCNT3 (mCNT3) was also cloned, suggesting that CNT3 is widely distributed in mammals.
A candidate cib-type transporter SNST1 that is related to the Na ϩ -dependent glucose transporter SGLT1 was identified in 1992 in rabbit kidney (23). There is no sequence similarity between SNST1 and either the CNT or ENT protein families. Although recombinant SNST1, when expressed in oocytes, stimulated low levels of Na ϩ -dependent uptake of uridine that was inhibited by pyrimidine and purine nucleosides (i.e. cibtype pattern), the function of this protein remains unclear because the rate of uridine transport in oocytes was only 2-fold above endogenous (background) levels, whereas a Ͼ100-fold stimulation was observed with h/mCNT3. It is likely that the true physiological substrate of SNST1 is a low molecular weight metabolite for which there is overlapping permeant recognition with nucleosides.
hCNT3 and mCNT3 are more closely related to the hagfish transporter hfCNT than to mammalian CNT1/2 and thus form a separate CNT subfamily. Hagfish diverged from the main line of vertebrate evolution about 550 million years ago and represent the most ancient class of extant vertebrates. The high degree of amino acid sequence similarity between h/mCNT3 and hfCNT in the TM 4 -13 region (67% sequence identity) may indicate functional constraints on the primary structure of this domain and suggests that cib-type concentrative NTs fulfill important physiological functions. The tissue distribution of hCNT3 transcripts was more widespread than anticipated from previous studies of cib functional activity and is different from that of either CNT1 or CNT2. Although transcripts for mammalian CNT1 and CNT2 have been described in jejunum, kidney, liver, and brain (CNT1) and jejunum, kidney, FIG. 9. Tissue distribution of hCNT3 mRNA. A and B, a commercial human multiple tissue expression RNA array probed with a 32 P-labeled cDNA corresponding to hCNT3 amino acid residues 359 -549. The inset in A is a dot blot of mRNA (0.5 g) from suspension parent and adherent differentiated HL-60 cells probed with the same cDNA. The numbered samples are: 1, leukemia (undifferentiated HL-60); 2, HeLa S3; 3, leukemia (K-562); 4, leukemia (MOLT-4); 5, Burkitt's lymphoma (Raji); 6, Burkitt's lymphoma (Daudi); 7, colorectal adenocarcinoma (SW480); 8, lung carcinoma (A549); 9, whole brain; 10, cerebral cortex; 11 10. High stringency Northern analysis of mRNA from human tissues probed with 32 P-labeled hCNT3 cDNA. Samples of human tissue mRNA (5 g) were separated on a 0.8% formaldehydeagarose gel and blotted on to BrightStar-Plus nylon transfer membrane. Hybridization with a radiolabeled cDNA probe for the coding sequence of hCNT3 amino acid residues 359 -549 was performed under high stringency conditions where there was no cross-reactivity with hCNT1 or hCNT2. Arrows indicate the positions of three bands in pancreas, bone marrow with sizes of 3.5, 4.2, and 6.5 kb. liver, spleen, heart, skeletal muscle, and pancreas (CNT2), the highest levels of hCNT3 transcripts were found in pancreas, bone marrow, trachea, mammary gland, and duodenum. Clinically, hCNT3 may be expected to contribute to the concentrative cellular uptake of both anticancer and antiviral nucleoside drugs. Future studies of the transcriptional regulation of the hCNT3 gene will enhance our understanding of its physiological function(s) and therapeutic potential.