Cloning and functional characterization of two bacterial members of the NAT/NCS2 family in Escherichia coli

Abstract

The coding potential of the genome of E. coli K-12 includes YgfO and YicE, two members of the evolutionarily conserved NAT/NCS2 transporter family that are highly homologous to each other (45% residue identity) and closely related to UapA of Aspergillus nidulans, a most extensively studied microbial member of this family. YgfO and yicE were cloned from the genome, over-expressed extrachromosomally and assayed for uptake of [³H]xanthine and other nucleobases, in E. coli K-12, under conditions of negligible activity of the corresponding endogenous systems. Alternative, essentially equivalent functional versions of YgfO and YicE were engineered by C-terminal tagging with an epitope from the E. coli lactose permease and a biotin-acceptor domain from Klebsiella pneumoniae. Both YgfO and YicE were shown to be present in the plasma membrane of E. coli and function as specific, high-affinity transporters for xanthine (K_m 4.2–4.6 µM for YgfO, or 2.9–3.8 µM for YicE), in a proton motive force-dependent manner; they display no detectable transport of uracil, hypoxanthine, or uric acid at external concentrations of up to 0.1 mM. Both YgfO and YicE are inefficient in recognizing uric acid or xanthine analogues modified at position 8 of the purine ring (8-methylxanthine, 8-azaxanthine, oxypurinol, allopurinol), which distinguishes them from their fungal homologues UapA and Xut1.

• Purine transport
• Escherichia coli
• xanthine
• uric acid
• specificity

Introduction

Transporters are an important class of integral membrane proteins represented by more than 10% of the known gene products in most genomes and highly relevant to microbial adaptations and human physiology and disease. Active transporters, that transduce energy released mostly from ATP or ion electrochemical gradients into a solute concentration gradient, are crucial for uptake of nutrients or release of toxic substances from the cell, at limiting concentrations. A few such transport proteins have been studied extensively to yield high-resolution models of the active transport mechanism, based on crystallography and a cohort of fine site-directed techniques (see Iwata & Kaback [2004], Abramson et al. [2003], [2004]). On the other hand, however, a large number of putative transporter genes in the current sequence databases (Saier [2000], Ren et al. [2004]) remain functionally uncharacterized. As a consequence, many transporter families exist for which numerous putative members are encountered but few are characterized or studied at the molecular level.

The ubiquitous Nucleobase-Ascorbate Transporter (NAT) (de Koning & Diallinas, [2000]) or Nucleobase-Cation Symporter-2 (NCS2) family (TC 2.A.40; http://saier-144-164.ucsd.edu/tcdb/) illustrates clearly such an example. The family is represented in essentially every species, with exception of some degenerate yeast, protozoan and bacterial genomes; although it consists of over ninety sequenced members, few of them are functionally characterized to date (see Figure 1). Based on this information, the family accommodates three subsets of transporters, one specific for oxidized purines (xanthine and/or uric acid), one specific for the pyrimidine uracil and one, confined in mammals, specific for L-ascorbate (see also Tsukaguchi et al. [1999], Liang et al. [2001], de Koning & Diallinas [2000], Goudela et al. [2005]). Apart from the mammalian SVCT1 and SVCT2 that function as ascorbate:Na⁺ symporters (Tsukaguchi et al. [1999]), the only well-studied, nucleobase-transporting member of NAT/NCS2 family is UapA, a high-affinity uric acid and xanthine:H⁺ symporter from the ascomycote Aspergillus nidulans (Diallinas & Scazzocchio [1989], Gorfinkiel et al. [1993], Diallinas et al. [1995], [1998], Meintanis et al. [2000], Amillis et al. [2001], [2004]). Chimeric purine transporter analysis between UapA and its lower-specificity, A. nidulans paralogue, UapC (Diallinas et al. [1998]) coupled with site-directed mutagenesis and second-site suppressor analysis of selected UapA mutants (Meintanis et al. [2000], Amillis et al. [2001]) have defined a conserved NAT/NCS2-motif region that includes residues affecting substrate recognition and selectivity (Figure 1). However, fundamental aspects of the structure-function relationships in UapA, such as the binding site determinants or the exact topology, especially at the C-terminal part of the molecule that includes the NAT/NCS2 motif (Diallinas et al. [1998]), remain unsolved.

Figure 1. ClustalW alignment of the coding sequences of YgfO (AAC75920; 466 codons), YicE (AAC76678; 463 codons), YgfU (Q46821; 482 codons), YgfQ (AAT48153; 455 codons), YcdG (AAC74091; 464 codons), YbbY (AAC73615; 435 codons), YjcD (AAC77034; 449 codons), YicO (AAC76687; 470 codons) and YieG (AAC76737; 445 codons) with functionally known proteins of NAT/NCS2 family (http://saier-144–164.ucsd.edu/tcdb/; for Lpe1, see Argyrou et al. [2001]; for Xut1, see Goudela et al. [2005]) showing the NAT/NCS2 signature-motif region. Principal substrates for functionally characterized homologues (including YgfO and YicE from this work) are given in parentheses. Positions of conserved [Q/E/P], N, G, T, and [R/K/G] residues are highlighted (see text for details).

Functionally known bacterial members of the NAT/NCS2 family include uracil transporters from both Gram-positive (PyrP) (Martinussen et al. [2001], Turner et al. [1994]) and Gram-negative bacteria (UraA) (Andersen et al. [1995]), as well as xanthine (PbuX) (Christiansen et al. [1997]) or uric acid transporters (PucJ, PucK) (Schultz et al. [2001]) from Gram-positive species of the Bacillus/Clostridium group. No oxidized-purine transporter members have been characterized functionally in Gram-negative bacteria. In contrast to B. subtilis (Schultz et al. [2001]) and other bacteria that can grow on all natural purine bases as sole nitrogen sources, aerobically grown Escherichia coli cannot use purines other than adenine (or adenosine) as a sole nitrogen source; it can, however, use purines like hypoxanthine, guanine (guanosine) or xanthine (xanthosine) to potentiate aerobic growth when aspartate serves as the basic nitrogen source (Xi et al. [2000]). Although the observations imply that those exogenous purines might be tunneled thoroughly to nucleotide synthesis via salvage pathways (Zalkin & Nygaard [1996]), it is likely that the purines are also partially catabolized (Xi et al. [2000]) but the corresponding transporter(s) and enzymes are not sufficient to support exclusive growth, due to poor uptake efficiency or weak promoter activity or inadequate regulation of the relevant genes (Xi et al. [2000], Reitzer & Schneider [2001]). In this respect, it is interesting that the genome of E. coli contains multiple NAT/NCS2 paralogues that could serve as specific xanthine transporters, based on bioinformatic evidence (http://www.membranetransport.org).

In this work, we have cloned and over-expressed in functional form two of the putative E. coli homologues of the NAT/NCS2 family, namely YgfO and YicE. Both proteins were shown to function as specific, proton motive force (PMF)-dependent high-affinity transporters for xanthine, that cannot use uric acid, hypoxanthine, uracil, or any other naturally occuring nucleobase as a substrate. An 106-amino acid epitope tag was also engineered onto the carboxyl terminus of either YgfO or YicE and shown not to affect their activity or substrate specificity profile. The importance of these two newly characterized bacterial transporters with respect to developing model systems for the structure-functional analysis of microbial NAT/NCS2 is discussed.

Materials and methods

Materials

[8-³H]xanthine (18 Ci/mmol and 27.6 Ci/mmol), [2,8-³H]hypoxanthine (20 Ci/mmol), [5,6-³H]uracil (59 Ci/mmol), and [8-¹⁴C]uric acid (50 mCi/mmol) were purchased from Moravek Biochemicals (Brea, CA). Non-radioactive nucleobases were from Sigma (St. Louis, MO) or provided by H. de Koning and G. Diallinas (see Goudela et al. [2005]). Oligodeoxynucleotides were synthesized by BioSpring GmbH (Frankfurt, Germany). Restriction endonucleases were from MBI Fermentas (St. Leon-Rot, Germany). High fidelity Taq Polymerase (Expand High Fidelity PCR system) was from Roche Molecular Biochemicals (Manheim, Germany). Site-directed rabbit polyclonal antiserum against the C-terminal dodecapeptide of E. coli LacY was generously donated by H. R. Kaback (UCLA) and prepared by BabCo (Richmond, CA). Avidin–HRP and protein A–HRP conjugates were from Amersham Pharmacia Biotech (London, UK). All other materials were reagent grade and obtained from commercial sources.

Bacterial strains and plasmids

E. coli strains DH5α (Gibco-BRL) or BW25113 [lacI^q, rrnB_T14, ΔlacZ_WJ16, hsdR514, ΔaraβAD_A?33, ΔrhaBAD_LD78] (Datsenko & Wanner [2000]) or MC1061 [lacX74, araD139, (ara, leu)7697, galU, galK, hsr, hsm, strA] (Casadaban & Cohen [1980]) were used for growth assays on minimal media. E. coli TOP10F′ (Invitrogen) was used for initial propagation of recombinant plasmids. E. coli DH5α or T184 [lacI⁺O⁺Z^-Y^-(A), prsL, met^-, thr^-, recA, hsdM, hsdR/F?, lacI^qO⁺Z^D118] (Teather et al. [1980]) harbouring pT7–5 (Sahin-Tóth et al. [1995]), pT7–5/melY-BAD (Frillingos & Kaback [2001]), pT7–5/ygfO or pT7–5/yicE with or without given C-terminal tags (this study; see below) was used for IPTG-inducible expression of transporters from the lacZ promoter/operator (p/o). E. coli was transformed according to Inoue et al. ([1990]).

Cloning, construction of expression plasmids and engineering of C-terminal tags

For construction of expression plasmids, the coding sequence of ygfO (AAC75920; with a change in position for the initiator Met codon, yielding an ORF of 466 codons) or yicE (AAC76678; 463 codons) was amplified by PCR on the template of genomic DNA prepared from E. coli T184, using sense and antisense primers with non-annealing overhangs at their 5′ end containing a BamHI and a HindIII restriction site, respectively; the PCR products were used for BamHI-HindIII restriction fragment replacement in plasmid vector pT7–5/lacY(cassette) (EMBL-X56095) to construct plasmids pT7–5/ygfO and pT7–5/yicE.

The epitope-tagged versions pT7–5/ygfO-BAD and pT7–5/yicE-BAD were constructed using two-stage (overlap-extension) PCR (Ho et al. [1989]) on the templates of plasmids pT7–5/ygfO or pT7–5/yicE (see above) and pT7–5/melY-BAD (Frillingos & Kaback [2001]), with internal sense and antisense primers carrying the terminal 12-bp sequence from the coding region of YgfO or YicE at their 5′ and 3′ halves, respectively, and the initial 14-bp sequence from the biotin-acceptor domain (BAD) of Klebsiella pneumoniae oxaloacetate decarboxylase (Consler et al. [1993]), at their 3′ and 5′ halves, respectively. Second-stage PCR products were digested with BamHI and HindIII and ligated to similarly treated plasmid vectors pT7–5/ygfO or pT7–5/yicE; the resulting constructs contain a 94-amino acid BAD domain followed in-frame by the C-terminal 12peptide of E. coli LacY and a stop codon.

The entire coding sequence of the above constructs was verified by double-strand DNA sequencing (MWG-Biotech, Ebersberg, Germany).

Growth of bacteria

For growth assays, the indicated substrains of E. coli K-12 or DH5α harbouring pT7–5, pT7–5/ygfO or pT7–5/yicE were grown aerobically at 37°C on minimal glucose medium M9 supplemented with NH₄Cl (17 mM) or L-aspartate (8 mM) or other nitrogen sources (as indicated), containing ampicillin (0.1 mg/mL), when needed.

For assays of active transport and immunoblotting, E. coli T184 harbouring pT7–5, pT7–5/ygfO or epitope-tagged ygfO, or pT7–5/yicE or epitope-tagged yicE was grown aerobically at 37°C in Luria-Bertani (LB) medium containing streptomycin (0.01 mg/mL) and ampicillin (0.1 mg/mL). Alternatively, growth of the transformed T184 was performed in minimal glucose medium M9 supplemented with NH₄Cl (17 mM) as the sole nitrogen source and the required amino acids L-threonine (0.17 mM) and L-methionine (0.14 mM) (Teather et al. [1980]) together with the appropriate antibiotics. Fully grown cell cultures were diluted 10-fold, allowed to grow to mid-logarithmic phase and induced with isopropyl 1-thio-β,D-galactopyranoside (IPTG; 0.5 mM) for an additional 2 h at 37°C. Cells were harvested, washed at the appropriate buffers, and used for transport assays or preparation of membranes.

Transport assays and kinetic analysis

E. coli T184 (or DH5α, where indicated) were washed twice in KP_i (0.1 M), pH 7.5, normalized to an OD₄₂₀ of 10 (35 µg of total protein per 50 µL), and assayed for active transport of [³H]xanthine (0.1–100 µM), [³H]hypoxanthine (1–100 µM), [³H]uracil (1–100 µM) or [¹⁴C]uric acid (1–100 µM), by rapid filtration at 25°C, pH 7.5, as described (see Frillingos et al. [1994]). For kinetic uptake measurements, initial rates were assayed at 5–10 sec in the concentration range of 0.1–100 µM [³H]xanthine; V_max and K_m values were determined from non-linear regression fitting to the Michaelis-Menten equation (Prism4; http://www.graphpad.com/). At least three determinations from three independent assays were performed in all cases.

For gross competition experiments, xanthine uptake was assayed in the simultaneous presence of [³H]xanthine (1 µM) and unlabeled analogues (1 mM) that had been pre-equilibrated in the cell mixture for 5 min.

Immunoblot analysis

Membrane fractions of E. coli T184 harbouring given plasmids were prepared by osmotic shock, ethylenodiaminotetraacetic acid (EDTA)/lysozyme treatment, and sonication (Konings et al. [1971], Frillingos et al. [1994]) and subjected to sodium dodecyl sulphate (NaDodSO₄) polyacrylamide (12%) gel electrophoresis. Proteins were electroblotted to poly(vinylidene difluoride) membranes (Immobilon-PVDF; Pall Corporation). YgfO-BAD or YicE-BAD were probed with horseradish peroxidase (HRP)-conjugated avidin or with a site-directed polyclonal antibody against the C terminus of E. coli LacY (Carrasco et al. [1984]) followed by protein A-HRP. The signal was developed with enhanced chemiluminescence (ECL; Amersham).

In silico analysis

DNA and deduced amino-acid sequences were retrieved from the most recent annotation of the genome sequence of E. coli K-12 (U00096; update of 24 June 2004), including a corrected DNA sequence for YgfQ (AAT48153) and a change in position of the deduced initiator codon for YgfU (Q46821); the position of initiator codon for YgfO has also been changed (in accordance with the EcoGene suggestion) to yield a predicted gene product of 466 instead of 485 residues (AAC75920). Comparative analysis of transporters was based on TransportDB (http://www.membranetransport.org) (Ren et al. [2004]), TC-DB (http://saier-144-164.ucsd.edu/tcdb/) (Saier [2000]), and COGs (http://www.ncbi.nlm.nih.gov/COG/) (Tatusov et al. [1997]).

Results

Homologues of the NAT/NCS2 family in E. coli

Based on in silico analysis, ten members of the NAT/NCS2 family are present in the genome of E. coli K-12 (Blattner et al. [1997], U00096; update of 24 June 2004) (http://www.membranetransport.org). Of those, only UraA has been experimentally characterized, as a specific uracil permease (Andersen et al. [1995]); of the remaining members, five (Yice, YgfO, YgfU, YcdG, YbbY) are clustered together with UraA in COG2233 (as putative xanthine/uracil permeases) and four (YjcD, YgfQ, YicO, YieG) are clustered separately, in COG2252 (http://www.ncbi.nlm.nih.gov/COG/). A sequence comparison with known transporters of the NAT/NCS2 family shows that YgfO, YicE, YgfU, and YcdG are the only ones retaining the characteristic sequence motif [Q/E/P]-N-X-G-X-X-X-X-T-[R/K/G] (Diallinas et al. [1998], Amillis et al. [2001]) (Figure 1). YcdG is mostly related to the E. coli uracil transporter UraA (35% identity of residues; 80% in motif), while YgfO and YicE are mostly related to the xanthine/uric acid transporters UapA and UapC of Aspergillus nidulans (∼30% identity of residues; 70% in motif), and YgfU is related to the xanthine transporter PbuX and the uric acid transporter PucJ of Bacillus subtilis (∼40% identity of residues; 70% in motif) (Figure 1).

To date, no oxidized-purine transporting system has been identified and characterized in E. coli K-12. In general consistence with previous studies (Xi et al. [2000]), we have found that oxidized purines (xanthine, hypoxanthine, uric acid) cannot support aerobic growth of E. coli K-12 (DH5α, BW25113, Top10F′, MC1061, or T184) when used as sole nitrogen sources in M9 minimal media. However, xanthine or hypoxanthine (but not uric acid) substantially potentiated the growth of E. coli DH5α when added (at concentrations of 50 µM or higher) in cultures utilizing L-aspartate (8 mM) or NH₄Cl (17 mM) as the sole nitrogen sources, implying that xanthine or hypoxanthine are utilizable (through purine salvage or catabolism) but are unable to provide the full nitrogen requirement for growth. On the other hand, no toxic effects were observed with allopurinol or oxypurinol or uric acid when supplemented at concentrations of up to 1 mM in the minimal growth media of E. coli DH5α, an inhibitory effect on growth being observed only with 5 mM of allopurinol or 5 mM of uric acid (data not shown).

Early studies have indicated that E. coli harbours high-affinity uptake systems for xanthine and hypoxanthine (Roy-Burman & Visser [1975]) and that these systems are driven by the proton motive force (Burton [1983]). Partially responsible for these activities might be homologues of the NAT/NCS2 family in E. coli, as also noted by other authors (Xi et al. [2000]). As presented below (Figures 2–6), we have found that extrachromosomal over-expression of the cloned ygfO or yicE genes in E. coli DH5α or T184 provide efficient, high-affinity systems for the uptake of xanthine, although evidence for exclusive growth on xanthine (or hypoxanthine) or a more efficient growth of the transformed DH5α cells at limiting xanthine supplement concentrations could not be obtained (data not shown).

Figure 2. Xanthine uptake activities of YgfO and YicE. E. coli T184 harbouring pT7–5/ygfO (upper panel) or pT7–5/yicE (lower panel) was grown aerobically at 37°C in complete medium to mid-logarithmic phase, induced with IPTG (0.5 mM) for 2 h, and assayed for transport of [³H]xanthine (external concentration 1 µM), as described in Methods. Assays were performed in the absence (closed triangles) or presence of CCCP, 5 µM (open triangles) or 30 µM (closed rectangles); in all cases, control assays performed with T184 harbouring pT7–5 alone yielded values indistinguishable from the ones of YgfO or YicE treated with CCCP, 30 µM (closed rectangles).

Figure 3. Substrate transport specificity of YgfO and YicE. E. coli T184 harbouring pT7–5/ygfO (upper panels) or pT7–5/yicE (lower panels) was grown aerobically at 37°C in complete medium to mid-logarithmic phase, induced with IPTG (0.5 mM) for 2 h, and assayed for transport of [³H]xanthine (rhombuses; panels on the left, X; panels on the right), [¹⁴C]uric acid (triangles; panels on the left, UA; panels on the right), [³H]hypoxanthine (HX; panels on the right) or [³H]uracil (URA; panels on the right), at the indicated external concentrations, as described in Methods. Initial uptake rates were measured at 5 and 10 sec, and maximal uptake values were taken from measurements over a 1 to 20-min incubation period. Control values obtained from T184 harboring vector pT7–5 alone (maximal uptake averaging 0.01, 0.02, 0.01, or 0.01 nmol/mg and 0.2, 1.2, 0.3, or 0.5 nmol/mg, for 1 µM and 100 µM of xanthine, hypoxanthine, uracil, or uric acid, respectively) were subtracted from the sample measurements in all cases. Data with standard deviation bars represent the means of three independent determinations.

Figure 4. Ligand recognition profiles of YgfO and YicE. E. coli T184 harbouring pT7–5/ygfO (upper panel) or pT7–5/yicE (lower panel) was grown, induced and assayed for transport of [³H]xanthine (1 µM), as described in the legend to Figure 2. Prior to initiation of the assay, cells were equilibrated with the indicated non-radiolabelled nucleobases (1 mM) for 5 min. Uptake rates were measured at 5 and 10 sec and control values obtained with T184 harboring pT7–5 alone were subtracted from the sample measurements in all cases. Results are presented as the percentage of transport activity retained relative to the activity of untreated positive controls, pre-equilibrated with solvent vehicle alone. Values represent the means of three determinations with standard deviations shown. Nucleobases and analogues used were: X, xanthine; U, uric acid; H, hypoxanthine; Ur, uracil; A, adenine; T, thymine; C, cytosine; G, guanine; Th, theophylline (1,3-dimethylxanthine); Cf, caffeine (1,3,7-trimethylxanthine); Ap, allopurinol; Op, oxypurinol; 1, 1-methylxanthine; 3, 3-mthylxanthine; 2, 2-thioxanthine; 6, 6-thioxanthine; 7, 7-methylxanthine; 8, 8-methylxanthine; 9, 9-methylxanthine; 8z, 8-azaxanthine; 9d, 9-deazaxanthine; 7d, 7-deazaxanthine; F, 5-fluoro-uracil; Al, allantoin; Im, imidazol.

Figure 5. Transport analysis and expression levels of YgfO-BAD and YicE-BAD. E. coli T184 harbouring pT7–5/ygfO, pT7–5/ygfO-BAD or pT7–5 alone (panels A and B) or pT7–5/yicE, pT7–5/yicE-BAD or pT7–5 alone (panel C) was grown, induced and assayed for transport of [³H]xanthine (1 µM), in the absence (A, C) or presence (B) of the indicated non-radiolabelled nucleobases (1 mM), as described in the legend to Figures 2 and 4. In parallel experiments, membrane fractions of induced T184 harbouring pT7–5/melY-BAD (as a control; Frillingos & Kaback, [2001]) (lane 1) or pT7–5/yicE-BAD (lane 2) or pT7–5/ygfO-BAD (lanes 3 and 4) were prepared and subjected to immunoblot analysis (100 µg of protein per lane) (panel D), as described in Methods. Proteins on the blot were probed both with anti-LacY-epitope antibody and with HRP-conjugated avidin, as indicated. Migration positions of prestained molecular weight standards (Bio-Rad, low range) are shown on the left (panel D). Nucleobases and analogues used in panel B for analysis of YgfO-BAD (dark-coloured bars) and YgfO (light-coloured bars) were: X, xanthine; U, uric acid; H, hypoxanthine; Ur, uracil; A, adenine; T, thymine; C, cytosine; G, guanine; Ap, allopurinol; Op, oxypurinol; 1, 1-methylxanthine; 3, 3-mthylxanthine; 2, 2-thioxanthine; 6, 6-thiolxanthine; 7, 7-methylxanthine; 8, 8-methylxanthine; 9, 9-methylxanthine; 8z, 8-azaxanthine.

Figure 6. IPTG-inducible expression and activity of YgfO-BAD. E. coli T184 harbouring pT7–5/ygfO-BAD or vector pT7–5 alone was grown aerobically at 37°C in complete medium, to mid-logarithmic phase, and induced with IPTG (0.5 mM) for the indicated periods of time. Part of the harvested cells was used to prepare membrane fractions and subjected to immunoblot analysis (100 µg protein per lane) using the anti-LacY-epitope antibody (upper panel), and part of them was resuspended in reaction buffer (35 µg of protein per 50 µL) to assay transport of [³H]xanthine (1 µM) (lower panel), as indicated. Prestained molecular weight standards (Bio-Rad, low range) are shown on the right of the upper panel.

Cloning and functional over-expression of YgfO and YicE

To examine whether the products of ygfO or yicE can function as specific purine transporters and provide efficient uptake systems for the utilization of xanthine and/or hypoxanthine in E. coli K-12, we have mobilized the corresponding coding sequences from the E. coli genome and transferred them to transcriptional control of the lacZ p/o in plasmid vector pT7–5. Expression of YgfO or YicE from these plasmids in transformed T184 or DH5α cells results in E. coli capable of rapid, highly efficient, concentrative uptake of [³H]xanthine while control T184 or DH5α transformed with vector pT7–5 alone display insignificant activity (Figure 2).

Active transport of xanthine (1 µM, 25°C) by T184 harbouring YgfO or YicE is linear for 5–15 sec with an initial uptake rate of about 1.5 or 0.15 nmol/mg protein/min, respectively, and approximates a steady state at 45–60 sec. The steady-state level of accumulation achieved by YgfO after an 1–10 min exposure to 1 µM xanthine is 0.80±0.18 nmol/mg protein (n=10); taken into account that the internal volume of E. coli is about 5.8 µL/mg protein (Felle et al. [1980]), this level corresponds to a concentration gradient of 310-fold (in/out). Similarly, YicE achieves a xanthine accumulation level of 0.08±0.01 nmol/mg protein (n=8), corresponding to a concentration gradient of 15-fold (in/out). Xanthine uptake activity of either YgfO or YicE is abolished in the presence of the protonophore carbonyl cyanide-m-chlorophenyl hydrazone (CCCP) (Figure 2), implying that transport is dependent on proton symport (Kaback et al. [1974]); this is further substantiated by the finding that YgfO and YicE are inactivated upon blockade of the respiratory chain by N-ethylmaleimide and that exogenously added electron donors such as reduced phenazine methosulphate (PMS) (Konings et al. [1971]) are able to restore transport activity (data not shown). Kinetic analysis (Table I) reveals that both YgfO and YicE are high-efficiency, high-affinity transporters for xanthine (K_m 4.2 and 2.9 µM, respectively).

Table I.  K_m and V_max values of YgfO and YicE for xanthine uptake¹.

Permease	Km(µM)	Vmax (nmol min^−1 mg ^−1 protein)	Vmax/Km (µL min^−1 mg ^−1)
YgfO	4.2±0.7	6.36±0.52	1514
YgfO-BAD	4.6±0.3	6.34±0.56	1378
YicE	2.9±0.5	0.59±0.14	203
YicE-BAD	3.8±0.6	0.85±0.06	224

¹E. coli T184 expressing the corresponding constructs were assayed for initial rates of xanthine uptake at 5 and 10 sec, in the concentration range of 0.1–100 µM, as described in the Methods section; negative control values obtained from T184 harboring vector pT7–5 alone (see Figure 2) were subtracted from the sample measurements in all cases. Kinetic parameters were determined from non-linear regression fitting to the Michaelis-Menten equation using Prism4; values represent the means of three independent determinations with standard deviations shown.

Specificity and substrate recognition analysis of YgfO and YicE

Both YgfO and YicE were assayed and found incapable for uptake of [¹⁴C]uric acid, [³H]hypoxanthine, or [³H]uracil, at concentrations of 1–100 µM (Figure 3). In addition, non-radiolabelled uric acid, hypoxanthine, uracil, adenine, thymine, or cytosine was unable to compete effectively against [³H]xanthine uptake in inhibition assays, even at a 1000-fold excess, while guanine competed by approximately 40% (Figure 4). In similar assays, a number of xanthine analogues substituted at specified positions with sulphur or methyl groups (2- or 6-thioxanthine, 1-, 3-, or 9-methylxanthine) were shown to be very efficient in inhibiting xanthine uptake by either YgfO or YicE (Figure 4). Importantly, however, xanthine analogues modified at positions 7 or 8 either inhibited marginally (oxypurinol) or were fully unable to inhibit xanthine uptake (7- or 8-methylxanthine, 7-deazaxanthine, allopurinol), even at a 1000-fold excess (Figure 4), implying that these positions are either crucial per se for substrate binding or hinder severely, when substituted, the formation of other crucial binding interactions. In any event, the inability to bind 8-modified analogues gives additional credence to the finding that uric acid (8-oxy-xanthine) is not recognised as a substrate by either YgfO or YicE and clearly distinguishes these two bacterial transporters from their fungal homologues UapA and Xut1 (Goudela et al. [2005]).

Gross competition analysis with 25 different xanthine analogues (Figure 4) showed that differences in the specificity profile between YgfO and YicE are minor. Dose-response analysis of inhibition rates indicated that YicE is distinguished from YgfO by a two-fold decrease in IC₅₀ for 6-thioxanthine and two-fold increase in IC₅₀ for 3-methylxanthine, with no major differences observed for the other analogues (data not shown). A more detailed study of ligand specificity is presented for YgfO in the accompanying manuscript (Goudela et al. [2005]).

Epitope-tagged versions of YgfO and YicE

In order to identify the ygfO and yicE gene products and facilitate further analyses and purification studies, we constructed alternative versions of YgfO and YicE by engineering additional sequence tags onto their carboxyl termini (see Methods). As shown in Figures 5 and 6, [³H]xanthine (1 µM) uptake activity of YgfO or YicE is only marginally affected by the BAD tag containing a biotin-acceptor domain from K. pneumoniae and the C-terminal dodecapeptide of E. coli LacY. Kinetic analysis shows that both YgfO-BAD and YicE-BAD display a similar affinity (K_m 4.6 vs 4.2 µM, and 3.8 vs 2.9 µM, respectively) and a similar V_max relative to their untagged counterparts, while the ratio V_max/K_m remains essentially unaffected (Table I). Finally, gross competition analysis with a series of 18 xanthine analogues showed clearly that the C-terminally engineered BAD domain does not affect the ligand specificity profile of YgfO (Figure 5B).

Levels of YgfO-BAD or YicE-BAD expression were assessed in immunoblot analyses performed in parallel with the xanthine uptake assays. It was found that YgfO-BAD is expressed in the cytoplasmic membrane fraction at high levels, while YicE-BAD is expressed poorly (Figure 5D); this difference in expression, revealed by both the anti-LacY-epitope antibody and avidin-HRP, may sufficiently account for the observed 10-fold difference in V_max between YgfO and YicE constructs (as expressed in nmol per min per mg of total cell protein; Table I). The immunoreactive signal of YgfO-BAD was also followed upon induction of T184 harbouring pT7–5/YgfO-BAD for increasing periods of exposure to IPTG and found to increase rapidly (from undetectable at time 0 to maximal at ≤60 min of induction), with a concomitant increase in xanthine uptake activity (Figure 6).

Discussion

We have shown that YicE and YgfO, two of the known sequence homologues of the NAT/NCS2 family in Escherichia coli, can work as specific, high-affinity transporters for xanthine (K_m 4.2 µM for YgfO, 2.9 µM for YicE), in a PMF-dependent manner. This is consistent with early biochemical indications on low µM-affinity, PMF-driven systems for uptake of this purine in E. coli (Burton [1983], Roy-Burman & Visser [1975]) and with the existing bioinformatic evidence (www.membranetransport.org; www.ncbi.nlm.nih.gov/COG/). The two identified proteins share similar characteristics with respect to substrate selectivity and ligand-recognition profile, including their inability to transport uric acid, uracil or hypoxanthine, inability to recognize other natural nucleobases or xanthine analogues modified at positions 7 or 8 of the imidazol ring and same pattern of inhibition preferences, as determined from assaying a series of 25 related ligands. The apparent, 10-fold higher uptake activity of YgfO relative to YicE observed in all experiments (V_max 6.36 nmol.mg⁻¹.min⁻¹ for YgfO, 0.59 nmol.mg⁻¹.min⁻¹ for YicE) is likely due to a concomitant difference in expression levels between the two transporter constructs, as revealed by Western blot analysis of the corresponding, epitope-tagged versions.

Interestingly, ygfO belongs to a cluster of genes at min 65 of the E. coli genome that constitute several putative purine catabolic operons (Xi et al. [2000]); in particular, ygfO is downstream of xdhD (a putative xanthine oxidase; Xi et al. [2000]) and of a predicted allantoinase gene, and upstream of ygfP (a predicted allantoate amidohydrolase) and ygfQ (a weakly associated NAT/NCS2 paralogue; Figure 1) in a putative operon controlled by two potential nitrogen-inducible, σ⁵⁴-dependent promoters (Reitzer & Schneider [2001]). In this respect, using the λ recombinase method of Datsenko and Wanner ([2000]), we have disrupted chromosomal ygfO in E. coli BW25113 but found no detectable impairment of the xanthine-associated growth phenotype (P. Karatza and S. Frillingos, unpublished work). In addition, evidence for a more efficient utilization of xanthine and improved growth of E. coli K-12 upon over-expression of YgfO (or YicE) could not be obtained from our data. It should be noted, however, that a clear xanthine-related phenotype is not consistently present in E. coli and there is considerable strain-to-strain variation possibly due to variations in regulation and endogenous levels of purine transporters (Xi et al. [2000]; and our unpublished data). Furthermore, the genome of E. coli contains multiple NAT/NCS2 paralogues predicted (Figure 1) or identified (this study) as putative xanthine transporters and elucidation of different regulation patterns or specificity of the products of these genes might require a much more intense and systematic approach, that is beyond the scope of this manuscript.

In comparison with UapA from A. nidulans, the most well-studied microbial homologue of NAT/NCS2 family, or with the other known xanthine transporters of the family in Eukarya (UapC, Lpe1, Xut1) (Diallinas et al. [1995], [1998], Argyrou et al. [2001], Amillis et al. [2001], Goudela et al. [2005]), both YgfO and YicE differ markedly in their inefficiency to utilize uric acid (8-oxy-xanthine) as a substrate. Furthermore, they differentiate from UapA and Xut1 (Goudela et al. [2005]) by their inability to recognize analogues of xanthine with bulky substitutions or N at position 8 of the imidazol ring, while no significant differences are evident in the recognition pattern of xanthine analogues modified at other positions. It appears, therefore, that modification at position 8 is a key determinant for different substrate binding orientations and interactions and for evolutionary divergence between eukaryal and bacterial purine transporters of the NAT/NCS2 family. The significance of these findings is studied further and analysed in more detail in the accompanying manuscript (Goudela et al. [2005]).

Xanthine uptake activity of either YgfO or YicE is not affected by epitope-tagging that extends their carboxy-terminal regions by an additional 106 amino acid residues (BAD version). In particular, the biotin-acceptor domain (BAD) tag was found not to affect K_m or V_max for xanthine transport by either YgfO or YicE or alter the overall substrate/ligand recognition profile of YgfO to any extent whatsoever. Similar observations have been made with two alternative C-terminal tags, including the C-terminal dodecapeptide of LacY followed by His₁₀ (22 residues; Smirnova & Kaback [2003]) and the green fluorescent protein sGFP (239 residues; Tavoularis et al. [2001]) (P. Karatza and S. Frillingos, unpublished work). The findings imply that the carboxy-terminal tail of these transporters does not interfere with substrate binding or play a major role in the transport mechanism. On the other hand, the engineered tags might prove valuable tools for future expression/purification, topological analysis or structure/function-relation studies, as evidenced from their use in other polytopic E. coli inner membrane proteins (Frillingos et al. [1998], Drew et al. [2002], Smirnova & Kaback [2003]).

In conclusion, this is the first report on purine-specific members of the NAT/NCS2 family in E. coli K-12. Since the work on UapA of A. nidulans and related fungal proteins (Diallinas & Scazzocchio [1989], Diallinas et al. [1995], [1998], Meintanis et al. [2000], Amillis et al. [2001], [2004], Goudela et al. [2005]) has indicated important evolutionary implications for this class of xanthine/uric acid transporters but little is known to date on the details of their mechanism, it is challenging to us to exploit the identified E. coli homologues as model systems to gain insight on the structure-function relationships, topology, binding-site interactions, structural dynamics and other mechanistic aspects of these transporters.

This paper was first published online on prEview on 28 April 2005.

We thank Dr George Diallinas for introducing us to the NAT family and for many helpful discussions and comments. We also thank H. Ronald Kaback for his constant encouragement, support and advice, in the course of this work. This research was funded by the program ‘Heraklitos’ of the Operational Program for Education and Initial Vocational Training of the Hellenic Ministry of Education under the 3rd Community Support Framework and the European Social Fund.