Comparative substrate recognition by bacterial and fungal purine transporters of the NAT/NCS2 family

Abstract

We compared the interactions of purines and purine analogues with representative fungal and bacterial members of the widespread Nucleobase-Ascorbate Transporter (NAT) family. These are: UapA, a well-studied xanthine-uric acid transporter of A. nidulans, Xut1, a novel transporter from C. albicans, described for the first time in this work, and YgfO, a recently characterized xanthine transporter from E. coli. Using transport inhibition experiments with 64 different purines and purine-related analogues, we describe a kinetic approach to build models on how NAT proteins interact with their substrates. UapA, Xut1 and YgfO appear to bind several substrates via interactions with both the pyrimidine and imidazol rings. Fungal homologues interact with the pyrimidine ring of xanthine and xanthine analogues via H-bonds, principally with N1-H and =O6, and to a lower extent with =O2. The E. coli homologue interacts principally with N3-H and =O2, and less strongly with N1-H and =O6. The basic interaction with the imidazol ring appears to be via a H-bond with N9. Interestingly, while all three homologues recognize xanthines with similar high affinities, interaction with uric acid or/and oxypurinol is transporter-specific. UapA recognizes uric acid with high affinity, principally via three H-bonds with =O2, =O6 and =O8. Xut1 has a 13-fold reduced affinity for uric acid, based on a different set of interactions involving =O8, and probably H atoms from positions N1, N3, N7 or N9. YgfO does not recognize uric acid at all. Both Xut1 and UapA recognize oxypurinol, but use different interactions reflected in a nearly 26-fold difference in their affinities for this drug, while YgfO interacts with this analogue very inefficiently.

• Aspergillus nidulans
• Escherichia coli
• specificity
• purine analogues

Introduction

Purines and pyrimidines are widely used as antimicrobial, antiviral or anticancer agents. However, their use suffers from a lack of selectivity, leading to severe side effects. The selectivity and efficacy of purine antimetabolites is achieved at two levels: the plasma membrane transporters that mediate cellular uptake, and the enzymes of the purine metabolic pathways that convert pro-drugs to cytotoxic metabolites, usually nucleotide analogues.

To exploit the purine transporters of pathogenic microorganisms (bacteria, fungi, protozoa) for the delivery of new drugs, several conditions must be met to confer selectivity and efficacy; these include: (1) high affinity of the microbial transporter, combined with low affinity for the mammalian transporters, (2) low abundance of competing substrates for the microbial transporters in its natural environment, and, ideally, (3) concentrative rather than equilibrative uptake. Microbial purine transporters appear to satisfy at least the last two requirements. Physiological purine concentrations in the blood or other tissues are low (up to 1 µM; Plagemann et al. [1988]), except from uric acid, which can be up to 200 µM. In addition, all known microbial nucleobase transporters are PMF (proton motive force)-driven and can transport purines against concentration gradients. Recently, several studies have also addressed the first criterion for nucleobase and nucleoside transporters of pathogenic protozoa (de Koning & Jarvis [1999], Wallace et al. [2002]). More specifically, substrate recognition interactions have been shown to accurately predict the uptake of chemotherapeutic agents by T. b. brucei. Such models were developed for the T. b. brucei P1 and P2 adenosine transporters and showed that the ‘structural determinants’ for substrate recognition by P2, but not by P1, were present in trypanocides as different as diamidines, melaminophenyl arsenicals, and isometamidium. P2 binds and, probably, transports these drugs with sub-micromolar affinity. Similar studies comparing substrate recognition determinants for the T. b. brucei H2 purine transporter and the human facilitative diffusion adenine-guanine transporter hFNT1, showed that selective uptake of nucleobase analogues by trypanosomes is possible, and thus provided a structural rationale for the selection of such compounds (Wallace et al. [2002]).

Herein, we describe for the first time a similar biochemical approach to study structure-function relationships in model representative bacterial and fungal homologues of the ubiquitous Nucleobase-Ascorbate Transporter family, known as NAT (de Koning & Diallinas [2000]) or NCS2 (Nucleobase-Cation Symporter; http://saier-144-164.ucsd.edu/tcdb/). The NAT/NCS2 family consists of over ninety currently sequenced proteins derived from Gram-negative and Gram-positive bacteria, archaea, fungi, diatoms, plants and animals. Most functionally characterized members are specific for either oxidized purines, such as xanthine and uric acid, or the pyrimidine uracil. Several microbial NAT proteins have been shown to be substrate:H⁺ symporters. However, two closely related mammalian members of the family, SVCT1 and SVCT2, with distinct tissue distribution profile, co-transport L-ascorbate and Na⁺ with a high degree of specificity and high affinity for the vitamin (Tsukaguchi et al. [1999]; Liang et al. [2001]). Clustering of NAT family members on the phylogenetic tree is complex with bacterial proteins and eukaryotic proteins each falling into three distinct clusters. The plant and animal proteins cluster loosely together, but the fungal proteins branch from one of the three bacterial clusters. Proteins of the NAT family are 414–650 amino acid residues in length and probably possess 12–14 transmembrane α-helical spanners (TMSs). The NAT family might be distantly related to the NCS1 family, which is restricted to prokaryotes, fungi and plants, and includes transporters for purines, cytosine, allantoin or thiamine (http://www.membranetransport.org) and to the AzgA-like family, with homologues in bacteria, archaea, fungi and plants (Cecchetto et al. [2004]). The only studied member of this family is the AzgA adenine-hypoxanthine-guanine transporter of A. nidulans (Amillis et al. [2004], Cecchetto et al. [2004]).

Based on purine uptake inhibition data, a structure-activity relationship for the binding of various analogues was formulated to identify and compare the specific substrate recognition determinants for three NAT transporters, namely UapA (A. nidulans), Xut1 (C. albicans) and YgfO (E. coli). UapA is a well-studied xanthine-uric acid transporter. Apart from several articles describing its characterization and regulation of expression, in both mycelia and germinating conidiospores (Darlington & Scazzocchio [1967], Diallinas & Scazzocchio [1989], Gorfinkiel et al. [1993]; Diallinas et al. [1995]; Valdez-Taubas et al. [2000], [2004], Amillis et al. [2004]), specific molecular determinants for the substrate specificity of UapA have also been partially characterized (Diallinas et al. [1998], Meintanis et al. [2000], Amillis et al. [2001]). Xut1 is a high-affinity, high-capacity, xanthine-uric acid transporter, described for the first time herein. Finally, YgfO is a recently characterized, high-affinity, high-capacity, xanthine transporter, described in the accompanying paper (Karatza & Frillingos [2005]). The substrate-recognition interactions proposed here for UapA, Xut1 and YgfO are compared with respect to each other and discussed in relevance to designing highly selective purine-based antifungals and antibacterials.

Materials and methods

Strains and media

Standard media for A. nidulans and C. albicans (Cove [1966], Scazzocchio et al. [1982]) and E. coli (Sambrook et al. [1989]) were used. The Escherichia coli K-12 strains used were DH5α and T184 (see also Karatza & Frillingos [2005]). E. coli harbouring plasmids pT7-5 or pT7-5/YgfO(wt) was grown aerobically at 37°C on either Luria-Bertani broth or minimal medium M9 with ammonium as the sole nitrogen source, containing ampicillin (0.1 mg/mL). The basic A. nidulans strain used in uptake studies, AZAC::pAN510, has been previously described (Meintanis et al. [2000]; Amillis et al. [2001]). It is a triple mutated azgA4, uapA24, uapC201/401 strain, carrying total loss-of-function mutations in the genomic loci encoding the three major purine transporters, and an arginine auxotrophy (argB2), which has been used for the selection of a transformant carrying a single copy of plasmid (pAN510) with the wild-type uapA gene. The C. albicans strain used, DSM3454, was purchased from Deutsche Sammlung von Mikroorganismen und Zellkulteren. DSM3454, also known as CBS5736, was isolated from human vagina in South Africa and is unrelated to the genome sequence strain SC5314. (Maria C. Costanzo, Senior Scientific Curator, Candida Genome Database, and Dr Peter Hoffmann,DSMZ-Deutsche Sammlung von Mikroorganismen und Zellkulturen, personal communication).

Cloning of xut1

xut1 was initially identified in silico (Contig: Chr4_Ctg10162) using the uapA sequence as a ‘probe’ against the C. albicans genome (http://genome-www.stanford.edu/fungi/Candida/). The following oligonucleotides, specific for the 5′ and 3′ regions flanking the open reading frame (orf), were designed: 5′-CGGGATCCAGCTATTCAAAAGACTTTAAAAACG-3′, 5′-GCTCTAGATTACTTCAAATCAGGGAATC-3′. A PCR reaction was performed, using 20 ng DNA as a template, as follows: 5 min at 94°C, followed by 35 cycles of 45 sec 94°C, 1 min 51°C, 2 min 72°C, and finally 10 min 72°C. The product of this reaction, which had the expected 1776 bp length and restriction map (not shown), was cloned in pGEM-easy vector™ (Promega) and sequenced. The sequence, named xut1, was deposited in GenBank and assigned accession number AY928185. This sequence was 99.3% identical to the GenBank sequence with accession EAK96574 of the genome strain SC5314. The 0.7% DNA differences corresponds to 12 base-pair substitutions, of which five lead to amino acid differences in the corresponding translation products, while the remaining seven concern third codon bases and do not affect amino acid sequences.

Northern blot analysis and standard nucleic acid manipulations

Northern blot analysis was performed using the glyoxal method (Sambrook et al. [1989]). [³²P]-dCTP labelled DNA molecules used as argB, uapA- or xut1-specific probes were prepared using a random hexanucleotide-primer kit following the supplier's instructions (Promega). RNA extraction from C. albicans was performed using the RNeasy Plant Mini Kit (Qiagen) as described by the supplier. C. albicans genomic DNA was purified from spheroplasts, which were isolated using lyticase (Sigma), using a standard SDS-acetic acid method (Lockington et al. [1985]). Total genomic DNA isolation from A. nidulans strains has been described (Lockington et al. [1985]). All other techniques involving DNA manipulations were according to Sambrook et al. ([1989]).

Radiolabelled purine uptake measurements

A. nidulans

[³H]-xanthine or [¹⁴C]-uric acid uptake was assayed in conidiospores at 37°C (Diallinas et al. [1995], Meintanis et al. [2000], Amillis et al. [2001], Cecchetto et al. [2004]). All experiments were carried out in triplicate. Initial velocities (measured at 1–2 min of incubation with the radioactive substrate) were corrected by subtracting background uptake values, measured in the triple azgA⁻uapA⁻uapC⁻ mutant.

C. albicans

Analogous measurements in C. albicans were done with cells harvested from cultures in minimal media supplemented with biotin and urea (5 mM) as a N source, at 37°C, and pH 6.8, unless otherwise indicated (pH 8 and pH 4, in Figure 2). For all kinetic experiments, cells were collected at the exponential phase of growth (OD₆₀₀ 0.6), where the predominant form of this fungus is yeast-like. For developmental studies, uptake was assayed at OD₆₀₀ 0.2, 0.4, 0.6, 0.8 and >3. At OD₆₀₀ 0.2–0.6 C. albicans is mostly yeast-like, at 0.6–0.8 it forms yeast-like and pseudohyphal cells, and in stationary phase (OD₆₀₀>3) it forms only true mycelium. For physiological studies, purine induction was achieved by replacing urea with 0.1 mg/ml of uric acid or adenine, N repression by 5 mM ammonium chloride, and N starvation by a shift to media lacking any N source, for 2 h. The H⁺-uncoupler carbonylcyanide chlorophenylhydrazone (CCCP) and the H⁺-ATPase inhibitor N,N′-dicyclohexylcarbodiimide (DCCD) were added at final concentrations of 30 µM and 100 µM respectively, for 10 min before initiating the uptake assay. The method of uptake measurements is essentially identical to that described for A. nidulans. All experiments were carried out in triplicate.

Figure 1. Physiological and biochemical identification of Xut1, a C. albicans xanthine-uric acid transporter. (a) Growth of C. albicans on purines as sole N sources, (b) Time course of uptake of 0.5 µM [³H]-xanthine or [³H]-uric acid, (c) Inhibition of [³H]-xanthine (0.5 µM) or [³H]-uric acid (0.5 µM) uptake by increasing concentrations of unlabelled xanthine or uric acid, respectively, (d)% inhibition of [³H]-xanthine (0.5 µM) uptake by increasing concentrations of unlabelled uric acid, and% inhibition of [³H]-uric acid (0.5 µM) uptake by increasing concentrations of unlabelled xanthine, (e) Energetic dependence of xanthine uptake. For details see Materials and methods. Results represent mean values from three independent determinations with standard deviations shown.

Figure 2. Regulation of expression of Xut1. (a) Regulation of xut1 mRNA steady-state levels in response to the N source, purine availability, pH or T. The nitrogen sources for each lane shown were: Ur, urea; NH₄, ammonium chloride; Ua, uric acid; Ad, adenine; -Nst, no nitrogen source; pH:8, pH:4 and 30°C were also in urea. Approximately equal RNA loading, except from the lane corresponding to uric acid (Ua) which was 50% of the rest, was controlled by estimating the amount of rRNA as previously described (Sambrook et al. [1989]; not shown), (b) Xut1 transport activity in response to the N source, purine availability, or pH, (c) Xut1 transport activity at different developmental stages. For details see Materials and methods.

E. coli

[³H]-xanthine uptake in E. coli K-12 was assayed by rapid filtration, at 25°C, pH 7.5, as described (Karatza & Frillingos [2005]). Briefly, fully grown E. coli T184 harbouring pT7-5 or pT7-5/YgfO(wt) was diluted 10-fold, allowed to grow to mid-logarithmic phase and induced with IPTG (0.5 mM) for YgfO expression. Cells were harvested, 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 used for active transport assay. Levels of YgfO expression were assessed in parallel immunoblot analyses using an antibody against the C-terminal dodecapeptide of E. coli LacY (Frillingos et al. [1994]) that had been engineered genetically at the C-terminus of YgfO (Karatza & Frillingos [2005]).

Gross competition assays were performed in the simultaneous presence of [³H]-xanthine (0.5 µM or 1 µM, as indicated) and 1 mM of non-radiolabelled putative competitors that had been pre-equilibrated in the assay mixture for 5 min. All transport assays were carried out in at least three independent experiments. Radiolabelled [³H]-xanthine (18 or 27.6 Ci/mmol) and [¹⁴C]-uric acid (13 Ci/mmol) were from Moravek Biochemicals, Brea, CA.

Kinetic analysis

Data were fitted to the appropriate equations with the use of the Prism 3 software package (http://www.graphpad.com/) to obtain K_m, K_i and IC₅₀ values. IC₅₀ values for compounds inhibiting the uptake of radiolabelled permeants were determined from full dose-response curves with a minimum of eight points spread over the relevant range. In all cases, the Hill coefficients were close to −1, consistent with competitive inhibition. Moreover, in previous studies, it has been demonstrated that a number of the compounds tested in the present study, inhibited UapA-mediated xanthine transport competitively (Diallinas et al. [1998], Meintanis et al. [2000], Amillis et al. [2001]). Thus, all the available evidence suggests that a simple model of competition with the binding site of the transporter is applicable and that the criteria for use of the Cheng and Prusoff ([1973]) equation to determine K_i value have been met. Hence, K_i values were calculated from the Cheng and Prusoff ([1973]) equation K_i=IC₅₀/[1+(L/K_m)], in which L is the permeant concentration. It should be noted that the K_i value is an affinity constant implying binding to the transporter but does not indicate whether the ligand is being transported across the membrane. Free Gibbs energy ΔG⁰ was calculated from ΔG⁰= − RTln(K_i), where R is the ideal gas constant and T the absolute temperature (°K).

Purines and analogues used

Purines: purine, adenine, guanine, hypoxanthine, xanthine, uric acid. Purine analogues: 1-methylxanthine, 3-methylxanthine, 7-methylxanthine, 9-methylxanthine, 8-methylxanthine, theophylline (1,3-dimethylxanthine), caffeine (1,3,7-trimethylxanthine), 7-deazaxanthine, 9-deazaxanthine, 8-azaxanthine, 2-thioxanthine, 2-thiouric acid, 6-thioxanthine, 6-thiouric acid, 8-thiouric acid, 2,8-dithiouric acid, isoguanine (2-hydroxy-6-aminopurine), 6-thioguanine, 8-azaguanine, 2-aminopurine, 2-oxypurine, allopurinol, oxypurinol. Pyrimidines: uracil, cytosine, thymine. Pyrimidine analogues: 5-fluorouracil, 5-fluorocytosine, 5-aminouracil. Purine nucleosides : adenosine, guanosine, inosine. Pyrimidine nucleosides: uridine, cytidine, thymidine. Ureides: allantoin. Most compounds were purchased from Sigma (St. Louis, MO). Several of these analogues were gifts from Dr H de Koning, Prof. C Scazzocchio, or Dr M Perez-Perez.

Several novel analogues (1-butylxanthine, 1-allylxanthine, 1-propylxanthine, 1-propargylxanthine, 1-benzylxanthine, 1-cyclopentylxanthine, 1-methyl-3-propylxanthine, 1-methyl-2-methylthiopurin-6-one, 1-propyl-2methoxypurin-6-one, 1-propyl-9-deazaxanthine, 3,7-diallylxanthine, 3,7-dimethyl-1-propargylxanthine, 3,7-dimethyl-6-methylthioxanthine, 6-thio-3,7-dimethylxanthine, 7-propargylxanthine, 7-ethylxanthine, 7-benzylxanthine, 7-(2-hydroxyethyl)-theophylline, 8-chlorotheophylline, 9-deaza-1,3-dimethylxanthine) were synthesized and provided by Dr Christa Muller, and tested only as substrates for UapA. None of these acted as a substrate and they were not tested further (results not shown).

Results and discussion

Identification and functional characterization of a C. albicans NAT homologue

Evidence for a specific xanthine-uric acid uptake system in C. albicans

We have identified by blast searches a single C. albicans UapA homologue, which we named Xut1 (http://genome-www.stanford.edu/fungi/Candida/). Xut1 is 55% identical in amino acid sequence with UapA and 30% identical with the bacterial homologue YgfO, and shows a similar putative topology with both UapA and YgfO (data not shown). These observations suggest that Xut1 might be a xanthine and/or uric acid transporter, as well. However, no data existed on the ability of C. albicans to take up these purines. We have examined this possibility in several ways. First, we examined whether C. albicans can grow, similarly to several fungi, on purines, and specifically on uric acid or xanthine, as sole nitrogen sources. In silico analysis showed that although C. albicans does not possess a xanthine dehydrogenase (XDH), an enzyme necessary for the utilization of purines as a N source (Glatigny & Scazzocchio [1995]), it does have a gene encoding an homologue of XanA, that is an enzyme of the alternative pathway of xanthine oxidation, characterised in A. nidulans (Darlington & Scazzocchio [1968], A. Cultrone, M. Rochet, C. Scazzocchio & R. Fernández, unpublished results). This suggested that C. albicans, if capable of taking up uric acid or/and xanthine, might also be able to use them as N sources. Our results showed that these purines can indeed be used as N sources in C. albicans (see Figure 1a). Comparing growth rates, uric acid was the best N source and xanthine was better than hypoxanthine. This suggested either different uptake systems or rate limiting catabolism of hypoxanthine to xanthine and uric acid.

By direct uptake measurements of radiolabelled purines (see Materials and methods), we have confirmed that C. albicans takes up hypoxanthine, xanthine and uric acid very efficiently (see below). A high-affinity, high-capacity, hypoxanthine uptake system has also been detected previously (Rao et al. [1983]). Kinetic analysis of this system (Rao et al. [1983], S. Goudela & G. Diallinas, unpublished) shows functional similarity to the Fcy2p hypoxanthine-adenine-guanine-cytosine transporter of S. cerevisiae (Weber et al. [1990]). This system does not recognize at all xanthine, oxypurinol or uric acid (S. Goudela & G. Diallinas, unpublished).

A C. albicans high-affinity, high-capacity, xanthine-uric acid/H⁺ symporter (Xut1)

Results in Figure 1b summarize the kinetic analysis of a novel xanthine-uric acid transport system of C. albicans, and show that: (a) uptake rates for both purines are linear for at least 2 min, (b) for both purines, the dependence on substrate concentration is characteristic of the kinetics of a single transporter, (c) the K_m for xanthine and uric acid is 4 and 50 µM respectively, (d) the V_max for both xanthine and uric acid uptake is high (1.2–2.5 pmoles/min 10⁷ cells), (e) xanthine and uric acid are transported by the same transporter, as this is judged by cross-competition experiments, and (f) xanthine and uric acid uptake are dependent on the H⁺ electrochemical gradient. These results strongly suggested that C. albicans has a single transporter with kinetic, mechanistic and specificity characteristics very similar to the UapA transporter of A. nidulans. Considering the similarity of the two transport systems, and the existence of a single C. albicans open reading frame with similarity to the uapA gene, we called this novel gene Xut1 and the corresponding C. albicans xanthine-uric acid transporter Xut1 (Xanthine Uric acid Transporter 1).

Interestingly, despite the structural and functional similarity of UapA and Xut1, the two transporters showed noticeable differences, as well. While UapA has a similar affinity for both uric acid and xanthine (7 and 8 µM respectively; Diallinas et al. [1998] and herein), Xut1 has a nearly 13-fold lower affinity for uric acid compared to xanthine. This preliminary observation highlighted, not only the need for studying in detail several NAT homologues before drawing general conclusions concerning the specificity and kinetics of NAT members, but also stresses the usefulness of such comparisons for the identification of specific amino acid residues critical for the function and specificity of NAT proteins.

Regulation of expression of Xut1 in response to N and purine availability

We have studied the regulation of expression of Xut1, at the level of transcript accumulation and transport activity. Results shown in Figure 2a demonstrate that xut1 is transcriptionally regulated in response to the presence or absence of purines from the medium, and in response to the N source supplied. xut1 mRNA levels are relatively low in urea, 3-fold inducible by purines (uric acid or adenine), and totally repressible by ammonia. In addition, N starvation for a period of 2 h, also led to a 2-fold increase in xut1 transcript levels. Finally, xut1 mRNA levels were not affected by pH or temperature. This profile of regulation is similar to that of uapA transcription regulation in A. nidulans (Gorfinkiel et al. [1993], Amillis et al. [2004]) and further supports that the two transporters carry out similar metabolic functions.

Figure 2b shows that transcriptional regulation is partially in line with transport activities. Transcriptional induction by uric acid, which in A. nidulans has been shown to be the physiological inducer of genes involved in purine uptake and catabolism, or repression by ammonia, are directly reflected to analogous increase and decrease in xanthine uptake activity. On the other hand, adenine or N starvation had no significant effect on carrier-mediated xanthine uptake activities. In fact, under these conditions, C. albicans showed increased diffusion of xanthine (not shown). It seems that the presence of adenine or N starvation have a pleiotropic affect on C. albicans cell membranes, and that this effect might ‘mask’ or partially suppress the transcriptional activation of xut1.

Expression of Xut1 in response to C. albicans development

We have also investigated the expression of Xut1 in response to the phase of growth and morphological status of C. albicans.Figure 2c shows that both xanthine and uric acid uptake, attributed to the Xut1 transport activity, peak at the mid exponential phase when the predominant form of C. albicans is yeast-like, and drop dramatically at later stages when pseudohypha, and especially true mycelium, develop.

All further uptake assays performed in this work were performed in the presence of urea as a non-repressing N source. Purine-induction conditions were not used, in order to avoid several steps of washing the cells before adding the radiolabelled purines and because it was not needed to obtain significant transport activities.

NAT homologues from A. nidulans, C. albicans and E. coli have distinct kinetics and specificities

Natural purines, pyrimidines and a variety of purine analogues were assayed for their ability to inhibit [³H]-xanthine transport by UapA in A. nidulans, Xut1 in C. albicans and YgfO in E. coli. By calculating the free energies (ΔG⁰) involved in the interaction between a single transporter and a variety of related putative substrates, structure-activity relationships were modeled and compared. Transport of 0.5–1.0 µM [³H]-xanthine was systematically assessed in the presence of a range of concentrations of 64 potential substrates (purines, pyrimidines, nucleoside, ureides and nucleobase-related analogues; for a complete list of compounds used see Materials and methods). For studies in A. nidulans, the strain used, known as ACZF::pAN510, carries loss-of-function mutations in all genes, other than uapA, encoding purine transporters (azgA and uapC), and a genetically non-identified mutation leading to lack of uracil transport (Amillis et al. [2001]), and thus allows the direct kinetic analysis of UapA-mediated purine uptake (for the detailed strain genotype see Materials and methods). Transport measurements in C. albicans were carried out in the wild-type strain DSM3454 (Deutsche Sammlung von Mikroorganismen und Zellkulteren). Despite the lack of a xut1 knock-out strain, Xut1 should be the only xanthine-uric acid transporter in this fungus. This is supported by both the in silico genomic analysis and the monophasic kinetic behavior (Hill coefficient close to 1.0; results not shown) of xanthine or uric acid uptake in the wild-type C. albicans strain (see below). Finally, transport rates in E. coli were assessed in a strain expressing YgfO from a plasmid-borne, IPTG-inducible promoter, under conditions of negligible activity of the endogenous genes involved in purine uptake and catabolism (Karatza & Frillingos [2005]).

Initially all compounds were tested for their ability to compete with [³H]-xanthine uptake, when present at 1 mM, that is an approximately 1000-fold excess. For UapA-mediated and Xut1-mediated transport in A. nidulans and C. albicans, respectively, excess non-labeled xanthine, uric acid, 1-, 3-, 8- and 9-methylxanthine, 2-thioxanthine, 2-thiouric acid, 6-thioxanthine, 8-thiouric acid and 8-azaxanthine, led to [³H]-xanthine uptake rates close or lower than 50% of the control with no inhibitor (results not shown). However, several analogues inhibited differentially Xut1 and UapA. Excess 6-thiouric acid and 2,8-dithiouric acid inhibited significantly only Xut1. Oxypurinol fully inhibited Xut1-mediated xanthine uptake but had a less prominent effect on UapA-mediated xanthine transport (14% of the control rate; results not shown). Allopurinol inhibited moderately (40% of the control rate) Xut1-mediated transport but led to an increase in the UapA-mediated xanthine uptake rate (up to 170% of the control rate). In the presence of any of the other compounds tested, [³H]-xanthine uptake rates were 80–110% of the wild-type for both fungi (results not shown).

YgfO-mediated [³H]-xanthine transport) was efficiently inhibited (>50%) by excess xanthine, 1-, 3- and 9-methylxanthine, 2-thioxanthine and 6-thioxanthine. Unlike UapA and Xut1, YgfO-mediated [³H]-xanthine transport was not inhibited by uric acid, uric acid analogues, 8-methylxanthine, or 8-azaxanthine In addition, YgfO was not inhibited by allopurinol but weakly inhibited by oxypurinol (results not shown). What arises from this observation is that YgfO, in contrast to the fungal homologues, cannot recognize substrates with bulky groups, or a N at position 8 of the purine ring. The significance of this observation is discussed later. Finally, YgfO was not inhibited by uracil, which was ‘weakly’ recognized by UapA and Xut1 (results not shown). As already mentioned, allopurinol, led to a ‘paradoxical’, but consistent, 40–70% increase in [³H]-xanthine uptake rates in A. nidulans. Previous genetic evidence has unequivocally confirmed that allopurinol is an excellent UapA substrate at concentrations as low as 3 µM. More specifically, while wild-type strains (uapA⁺) are sensitive to allopurinol, as judged on growth on hypoxanthine as sole N source (Darlington & Scazzocchio [1967], Scazzocchio et al. [1973], Lewis et al. [1978]), uapA⁻ loss of-function mutants are totally resistant to allopurinol (Diallinas & Scazzocchio [1989]). Furthermore, physiological evidence from simple growth tests also supports that xanthine and its analogue 2-thioxanthine, when in excess, compete with UapA-mediated allopurinol uptake (M. Koukaki & G. Diallinas, unpublished). In other words, xanthine, uric acid and their analogues seem to share the same transporter, and possibly to bind to the same binding site, with allopurinol. However, the kinetic analysis performed in this work reveals that the mechanism of uptake of this drug does not obey simple kinetics. The non-availability of radiolabelled allopurinol of high specific activity in commerce, and the very low capacity of UapA for allopurinol uptake, did not allow us to further investigate this apparent kinetic paradox.

Fungal and E. coli transporters bind differently to the pyrimidine ring of purines

The K_i values for 23 selected compounds were determined from dose-response curves, and ΔG⁰ values for the interaction between UapA, Xut1 or YgfO and purine-related substrates were calculated as described under Materials and methods. Table I lists K_i values and binding energies for the three transporters. Based on the inhibition data, a structure-activity relationship for the binding of different molecules to each transporter was formulated to identify the specific interactions with the purine ring. The rationale of the models is described below. It should be noticed that purine structural formulas shown in this work, and used to develop speculative models for transporter-substrate interactions, correspond to preferred tautomers at neutral pH (Hernandez et al. [1996a], Hernandez et al. [1996b], Stoychev et al. [2002], Kulikowska et al. [2004]). In the case of oxypurinol, two tautomers (N8-H and N9-H) had to be considered for the rationale of the models proposed herein (Hernandez et al. [1996b]).

Table I.  K_i (µM) values and ΔG⁰ binding energies (kJ/mol) of UapA, Xut1 and YgfO. n.d., not determined, see text. Significant differences are highlighted in bold. All nucleobase analogues, among those listed in Materials and methods, not present in the table, were shown not to compete UapA-, Xut1- or YgfO-mediated xanthine uptake at concentrations as high as 2 mM.

	UapA		Xut1		YgfO
Substrates	Ki	ΔG^0	Ki	ΔG^0	Ki	ΔG^0
Uric acid	7	30.6	50	25.5	>1500	<16.1
Xanthine	8	30.3	4	32.0	1.5	33.2
Hypoxanthine	>1500	<16.8	>1500	<16.8	>1500	<16.1
Guanine	>1500	<16.8	>1500	<16.8	>1500	<16.1
Uracil	1200	17.3	824	18.3	>1500	<16.1
1-Methylxantine	280	21.0	219	21.7	36	25.3
2-Thioxanthine	63	24.9	30	26.8	91	23.0
2-Thiouric acid	315	20.8	97	23.8	>1500	n.d.
3-Methylxanthine	28	27.0	22	27.7	72	23.6
6-Thioxanthine	346	20.5	228	21.6	41	25.0
6-Thiouric acid	>1500	<16.8	84	24.2	>1500	n.d.
7-Methylxanthine	>1500	<16.8	>1500	<16.8	>1500	<16.1
7-Methyluric acid	>1500	<16.8	1410	17.0	n.d	n.d
7-Deazaxanthine	>1500	<16.8	>1500	<16.8	>1500	<16.1
8-Azaguanine	>1500	<16.8	>1500	<16.8	n.d.	n.d.
8-Methylxanthine	100	23.7	80	24.3	>1500	<16.1
8-Azaxanthine	247	21.4	55	25.0	>1500	<16.1
9-Methylxanthine	200	22.0	160	22.5	53	24.4
9-Deazaxantine	>1500	<16.8	>1500	<16.8	>1500	<16.1
8-Thiouric acid	213	21.8	344	20.5	57	24.2
2,8-Dithiouric acid	1000	17.8	253	21.4	n.d.	n.d.
Allopurinol	n.d.	n.d.	802	18.4	>1500	<16.1
Oxypurinol	103	23.7	4	32.0	827	17.6

The strict specificity preference of the three NAT homologues for purines over pyrimidines shows that constituents of the imidazol part of the purine ring are essential for binding interactions (Table I). Very weak binding of uracil (K_is 0.8–1.2 mM) could be detected only with the fungal transporters. In those cases, comparing ΔG⁰ values of uracil (17.3–18.3 kJ/mol) to those of xanthine (30.3–32 kJ/mol), we assumed that the pyrimidine part of xanthine contributes approximately 17–18 kJ/mol and the imidazol part 13–14 kJ/mol to substrate binding. Four possible H-bond interactions can be made with the pyrimidine part of purines. These could involve keto groups at positions 2 and 6 (=O2 and =O6) and H atoms at positions 1 and 3 (N1-H and N3-H). The individual contributions of the different positions of the pyrimidine ring were found to be different in the fungal and E. coli homologues. In UapA and Xut1, position N3-H should not be involved in significant interactions with purines, as 3-methylxanthine has a K_i value (28 and 22 µM in UapA and Xut1, respectively) close to that of xanthine (8 and 4 µM respectively), corresponding to binding energy loss of only 3.5–4.3 kJ/mol (Table I). A slightly more important ΔG⁰ loss of 5.2–5.4 kJ/mol was estimated when the =O2 of xanthine was substituted by =S2 in 2-thioxanthine. The loss in ΔG⁰ between xanthine and 6-thioxanthine or 1-methylxanthine showed that the most important contribution, approximately 10 kJ/mol, in binding of the pyrimidine ring of xanthine comes from H-bonds with either N1-H or =O6 or both. It is evident that the sum of individual interactions with positions 1, 2, 3 and 6 (approximately 30 kJ/mol) exceeds that estimated for the total contribution of the pyrimidine part (17–18 kJ/mol). Given however that the analogues used have bulky substitutions (–CH₃ or =S), this overestimation is attributable to extra energy loss due to indirect steric effects on neighbouring H-bonds. The similarity in energy loss with analogues substituted at either positions 1 or 6 suggests that the same 10 kJ/mol H-bond was ‘lost’ either directly, due to lack of the appropriate H acceptor or donor, or indirectly, due to steric hindrance, at positions N1-H or/and =O6. The more significant loss in energy binding of 2-thioxanthine compared to that of 3-methylxanthine might also suggest that the H-bond lost involves directly =O2, while the loss with 3-methylxanthine is due to an indirect steric effect. Thus, considering the sum of the estimated interactions with positions 1 and/or 6, on the one part, and 2, on the other, the overall contribution of the pyrimidine ring can be calculated, in either transporter case, to about 14.5–15.8 kJ/mol, a value that approximates the observed ΔG⁰ for uracil (17.3–18.3 kJ/mol).

Interestingly, while UapA, Xut1 and YgfO (8, 4 and 1.5 µM respectively) have similar affinities for xanthine, contributions of the different positions of the pyrimidine ring to YgfO binding, were different from those of UapA and Xut1. Comparing ΔG⁰ values of xanthine, 1-methylxanthine, 2-thioxanthine, 3-methylxanthine and 6-thioxanthine, it was evident that positions 2 and 3 have a major role in YgfO binding (9.8–10.2 kJ/mol each), while positions 1 and 6 contribute less, 7.9–8.2 kJ/mol each. However, the sum of energetic contributions for xanthine binding from these individual calculations comes to 36.1 kJ/mol, which is in contrast to the negligible binding of uracil, even at mM concentrations (ΔG⁰<16.8 kJ/mol). This apparent discrepancy between ΔG⁰_obs and S(ΔG⁰) is suggestive of significant steric effects, probably in the vicinity of positions =O2 and N3-H, not present in fungal transporters.

Taken together, the above results indicate a clear difference in the architecture and/or plasticity of the binding site between fungal and E. coli NAT homologues. This observation should be promptly considered in the design of novel NAT substrates with antifungal or antibacterial properties. For example, while UapA and Xut1 recognize 3-methylxanthine very efficiently, the bacterial homologue YgfO has a relatively low affinity for it (47-fold lower than the affinity of xanthine). In contrast, YgfO recognizes 1-methylxanthine with relatively higher affinity compared to the two fungal homologues (36 µM compared to 219–280 µM). Thus, while position 3 should be considered as the most appropriate for modification and the development of novel xanthine analogues with antifungal activities, position 1 might be more appropriate for the development of purine drugs against E. coli and related bacteria.

Interactions with the imidazol part of purines

The proposed interactions with the pyrimidine part of purines can already justify why adenine, isoguanine or cytosine, which all lack both an =O6 and a N1-H, are not NAT substrates. However, the very weak (UapA, Xut1) or insignificant (YgfO) interaction with uracil, the inefficient recognition of many xanthine analogues substituted at positions 7, 8 or 9 and the extensive differences in the observed affinity for oxypurinol among the three transporters (Table I), all argue for a critical involvement of the imidazol part in substrate recognition/binding. As already rationalized above, this should be at least 13 kJ/mol. NAT homologues, in principle, can H-bond with positions 7, 8 or 9 of the imidazol ring, depending on the purine involved. These interactions could involve an 8-keto group (=O8 in uric acid and related analogues), non-protonated N atoms as H acceptors (N9 in xanthine, 7-methylxanthine and oxypurinol tautomer N8-H, N7 in 9-methylxanthine, or N8 in allopurinol and oxypurinol tautomer N9-H), protonated N atoms as H donors (N7-H in xanthine and uric acid or related analogues, or N9-H in 8-azaxanthine, uric acid or related analogues, allopurinol and oxypurinol tautomer N9-H).

Binding of xanthine

Binding to the imidazol ring of xanthine could involve interactions with N7-H and/or N9. The inability of all transporters tested to recognize 7-deazaxanthine or 9-deazaxanthine, two analogues devoid of any non-protonated N (Table I), suggested that interaction with the imidazol ring requires a major H-acceptor, for xanthine at neutral pH being the non-protonated N9. Further evidence for the importance of N9 came from the observation that UapA, Xut1 and YgfO bind 9-methylxanthine with reduced affinity (25-, 40- and 35-fold, respectively). However, the fact that all transporters still bind 9-methylxanthine with moderate affinities (K_is 53–200 µM, ΔG⁰ 22–24.4 kJ/mol), suggested that a non-protonated N7 present in this analogue might also contribute to binding. Comparing ΔG⁰ values of 9-methylxanthine (22–24.4 kJ/mol) and uracil (16–18 kJ/mol), N7 contribution might in that case be 4–8 kJ/mol. On the other hand, the nil binding of 7-methylxanthine in all three transporters, notwithstanding the presence of the critical non-protonated N9, is possibly due to severe steric hindrance. Binding of 8-methylxanthine and 8-azaxanthine distinguished fungal from E. coli homologues. While Xut1 and UapA bind these analogues with moderate affinities (K_is at the order of 55–344 µM), YgfO does not bind 8-methylxanthine or 8-azaxanthine, even at mM concentrations. This result was in line with the inability of YgfO for recognizing uric acid, uric acid analogues or oxypurinol (see below), which all differ from xanthine at position 8 of the imidazol ring.

Binding of uric acid

Recognition of uric acid reflects the most important physiological difference among the three transporters tested. As discussed earlier, interaction of uric acid with UapA is of high affinity (K_m/i 7 µM), with Xut1 is moderate (K_m/i 50 µM), while YgfO does not recognise uric acid, even when present at mM concentrations. This observation suggested that interactions with uric acid in the three transporters are different, a conclusion further supported by the different affinities found for 2-thiouric acid, 6-thiouric acid, 8-thiouric acid, or 2,8-dithiouric acid. A number of our data suggest that not only UapA and Xut1 bind uric acid using interactions different from those used for xanthine binding, but also UapA and Xut1 per se use different contacts to bind uric acid. Uric acid lacks the non-protonated N9 that appears to be critical for xanthine binding, as outlined above. Positions =O2, =O6 and =O8 seem to contribute all the necessary energy for binding in UapA, since individual substitutions of the oxo-groups with thio-groups lead to significant energy losses, summing up to ≥34 kJ/mol, a value that approximates roughly the observed ΔG⁰ for uric acid (30.5 kJ/mol). On the other hand, in Xut1, the sum of energy losses from the same substitutions is very low, approximately 8 kJ/mol, compared to the observed ΔG⁰ value for uric acid (25.5 kJ/mol). This implies that UapA interacts principally via the three oxo-groups, while Xut1 may interact mostly via the H atoms present at positions 1, 3, 7 or/and 9. The unavailability of appropriate analogues did not allow us to investigate the role of N1-H, N3-H, N7-H or N9-H on uric acid binding by Xut1.

Despite the structural identity of the pyrimidine rings of xanthine and uric acid, the different energetic contributions of =O2 (5.2 versus 9.8 kJ/mol) and =O6 (9.3 versus 14 kJ/mol) and the possible involvement of a different H acceptor of the imidazol ring (N9 versus =O8), suggest that these two substrates might be positioned differently in the binding site of UapA. Differential positioning of substrates, and in particular of purines, in the binding pockets of enzymes is not uncommon (Hernandez et al. [1996b]), and could not have been predicted without the extensive analysis presented here, since overall affinities for these two substrates are similar (Table I).

The fact that YgfO did not recognize at all uric acid, 2-thiouric acid, 6-thiouric acid or analogues with substitutions at position 8 (8-methylxanthine, 8-azaxanthine), while having very similar overall binding affinity for xanthine, strongly suggests that the part of the bacterial binding site interacting with the imidazol ring is significantly different from the fungal binding sites. Consistently, YgfO recognizes oxypurinol with very low affinity, >500-fold lower than that for xanthine (Table I), in sharp contrast to what is seen with the fungal homologues (see next).

Binding of oxypurinol

Xut1 binds oxypurinol with very high affinity (K_i 4 µM), indistinguishable from that for xanthine. The only differences between xanthine and oxypurinol are in the relative N positions in the imidazol ring. When considering the N8-H oxypurinol tautomer, which has been proposed to be the predominant bio-active form (Hernandez et al. [1996b]), the imidazol ring of both xanthine and oxypurinol might well interact with Xut1 via N9, explaining the similarity in K_i values. UapA binds oxypurinol with reduced affinity (103 µM), equivalent to a loss of approximately 6.3 kJ/mol in binding energy compared to xanthine. This indicates that the presence of a proton donor at position 8 (N8-H) has a negative effect on binding of oxypurinol by UapA, but not by Xut1. Alternatively, oxypurinol in the binding site of UapA might predominantly exist as tautomer N9-H, ‘weakening’ the proposed interaction via a non-protonated N9. In any case, the relative kinetic estimations for oxypurinol and xanthine binding strongly support the involvement of position N9 in H-bonds with both Xut1 and UapA. On the other hand, the 26-fold difference in the affinities of Xut1 and UapA for oxypurinol also supports that the binding sites in the two transporters, and especially the parts interacting with positions N8 or/and N9, are not identical.

Allopurinol binding further supported important differences between Xut1 and UapA. Xut1 binds allopurinol with nearly 180-fold reduced affinity (K_i 802 µM) compared to oxypurinol. The 13.5 kJ/mol ‘lost’ may be due to differences in positions =O2 and N3 (see above), and to different levels of tautomerism between positions N8 and N9, in allopurinol and oxypurinol. The 18.5 kJ/mol estimated for allopurinol binding by Xut1 may derive from interactions with =O6/N1-H and N8 or N9. Finally, despite the genetic evidence that allopurinol is a UapA substrate, kinetic assays could not evidence any binding of this analogue to UapA. Despite the lack of a current explanation, this kinetic paradox further emphasizes the existence of important differences between the binding sites of Xut1 and UapA.

The distinct ligand recognition profiles presented here for three paradigmatic NAT homologues from filamentous fungi (UapA), yeast (Xut1) and Gram-negative bacteria (YgfO) emphasize similar aspects as well as important differences in the pattern of NAT-substrate interactions employed. The substrate-binding sites of the two fungal NATs examined appear to be similar with respect to interactions with xanthine, but differ with respect to interactions with uric acid and oxypurinol. The bacterial homologue YgfO shares similarities with the fungal transporters with respect to the overall contribution of the pyrimidine part in xanthine binding, but fine differences are revealed on a more detailed kinetic examination of individual positions. For example, while the contribution of the different positions in the fungal transporters follows the order N1 > N3 and =O6 > = O2, in E. coli the order is reversed, N3 > N1 and =O2 > = O6.

The important differences in the proposed interactions of Xut1 and UapA with uric acid (7-fold) and oxypurinol (26-fold), as opposed to their overall similarity of interactions with xanthine, strongly suggests that modification at position 8 is a key determinant for different substrate binding orientations. This in turn implies important differences in the part of the binding sites of Xut1 and UapA interacting with this position. This is even more prominent for the bacterial homologue YgfO, which does not bind uric acid and binds oxypurinol with very low affinity (500-fold lower than xanthine). Thus, it seems that NAT homologues while sharing significant similarities in interactions with the pyrimidine ring, they have diverged profoundly in their interactions with the imidazol ring of purines. Interestingly, in both UapA and YgfO, several mutations that differentially affect recognition of xanthine, uric acid or/and oxypurinol have been recently characterized (M. Koukaki, A. Vlanti, S. Goudela, A. Pantazopoulou, H. Gioule, S. Tournaviti & G. Diallinas, under revision; P. Karatza & S. Frillingos, in preparation).

The present study has resulted in formulating preliminary structure-activity relationship for the binding of NAT transporters. The models proposed in Figure 3 should be considered only as speculative, as our approach cannot be used for estimating precisely the contribution of steric affects, the contribution of thio-groups in H-bonds, or the percentage of tautomerism in the binding site of transporters. Nevertheless, for each transporter studied herein, a number of purine analogues emerge as promising templates for the development of novel, specific antifungals or antibacterials. For example, UapA-mediated transport of purine drugs could be based on 3-substituted xanthines, Xut1-mediated transport on 3-substituted xanthines or oxypurinol, and YgfO-mediated transport on 1- or 6-substituted xanthines.

Figure 3. Speculative models for the possible interactions of UapA, Xut1 and Ygfo with their substrates discussed in the text. ΔG⁰_obs is the ΔG⁰ calculated directly from the K_i of the relevant substrate. ∑(ΔG⁰) is the sum of the differences in ΔG⁰ values obtained from binary comparisons of the relevant substrate and selected analogues. For oxypurinol only ΔG⁰_obs values are shown. In that case, the contribution of different positions is based on xanthine binding (for the pyrimidine ring) and the difference between the ΔG⁰_obs and the contribution of the pyrimidine part (for the imidazol ring). Estimated apparent ΔG⁰ values are in kJ/mol. R stands for an amino acid in the transporter interacting, most probably via H-bonds, with a specific purine position. Shaded areas depict positions of interactions with R. Preferred tautomers under conditions used in this work are shown (Hernandez et al. [1996a], [1996b]; Stoychev et al. [2002], Kulikowska et al. [2004]).

Last but not least, we have characterized a novel transport system in C. albicans, and presented evidence that it corresponds to the putative protein sequence called Xut1. This system, similarly to what has been shown previously for UapA in A. nidulans (Gorfinkiel et al. [1993], Amillis et al. [2004]), is regulated in response to both physiological and developmental signals. This information should be promptly considered in putative future attempts to use transporters of fungi as specific gateways for efficient targeted pharmacological interventions.

The results presented here aim to provide starting models to further study structure-function relationships in the NAT family, a most important purine transporter family of microorganisms. In this vein, more specific experimental evidence for the NAT-substrate side chain interactions is expected to be provided by our ongoing analysis of several specificity and affinity mutations of UapA (M. Koukaki, A. Vlanti, S. Goudela, A. Pantazopoulou, H. Gioule, S. Tournaviti & G. Diallinas, manuscript submitted and unpublished results) and Cys-scanning mutagenesis of YgfO (P. Karatza & S. Frillingos, in preparation).

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

We thank, first of all, Dr H. de Koning, for introducing us in the idea of identifying transporter-substrate interactions using a kinetic approach, for providing several purine analogues, for his time spent in long critical discussions concerning this work, for his friendship. We thank Dr M. Perez-Perez for 7-deazaxanthine. We thank Areti Pantazopoulou, Sotiris Amillis, George Zacharioudakis and Anna Vlanti for ideas, moral support and technical help. We also thank Dr Christa Muller for providing us with several purine analogues and the Biochemistry Department of the University of Athens for providing the scintillation counter. Finally we thank Prof. C. Scazzocchio for 2-thioxanthine and thiouric acid analogues, and for critically reading the manuscript. S.G. was supported by the Archimedes 2002 prize of the EU, by the ‘Plato’ Franco-Hellenic Collaboration Grant (2002), and by the National Grant PYTHAGORAS 2003. M.K. was supported by the National Grants PENED 1999 and PYTHAGORAS 2003, and The University of Athens (ELKE). Work in the laboratory of G.D. was supported by the National Grants PENED 1999, PYTHAGORAS 2003, The University of Athens (ELKE) and the ‘Plato’ Franco-Hellenic Collaboration Grant (2002). Work in the laboratory of S. F. 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 (2002).