Substrate Selectivity of YgfU, a Uric Acid Transporter from Escherichia coli*

Background: YgfU is homologous to nucleobase transporters of the ubiquitous family NCS2. Results: YgfU transports uric acid. A single-amino acid replacement (T100A) converts YgfU to a dual-selectivity transporter for uric acid and xanthine. Conclusion: Thr-100 is essential for the uric acid selectivity. Significance: Dissecting selectivity determinants in family NCS2 is crucial for understanding evolution of purine uptake. The ubiquitous nucleobase-ascorbate transporter (NAT/NCS2) family includes more than 2,000 members, but only 15 have been characterized experimentally. Escherichia coli has 10 members, of which the uracil permease UraA and the xanthine permeases XanQ and XanP are functionally known. Of the remaining members, YgfU is closely related in sequence and genomic locus with XanQ. We analyzed YgfU and showed that it is a proton-gradient dependent, low-affinity (Km 0.5 mm), and high-capacity transporter for uric acid. It also shows a low capacity for transport of xanthine at 37 °C but not at 25 °C. Based on the set of positions delineated as important from our previous Cys-scanning analysis of permease XanQ, we subjected YgfU to rationally designed site-directed mutagenesis. The results show that the conserved His-37 (TM1), Glu-270 (TM8), Asp-298 (TM9), and Gln-318 and Asn-319 (TM10) are functionally irreplaceable, and Thr-100 (TM3) is essential for the uric acid selectivity because its replacement with Ala allows efficient uptake of xanthine. The key role of these residues is corroborated by the conservation pattern and homology modeling on the recently described x-ray structure of permease UraA. In addition, site-specific replacements at TM8 (S271A, M274D, V282S) impair expression in the membrane, and V320N (TM10) inactivates the permease, whereas R327G (TM10) or S426N (TM14) reduces the affinity for uric acid (4-fold increased Km). Our study shows that comprehensive analysis of structure-function relationships in a newly characterized transporter can be accomplished with relatively few site-directed replacements, based on the knowledge available from Cys-scanning mutagenesis of a prototypic homolog.

Conservation of sequence and overall structure in many evolutionarily broad families of transporters often correlates with high functional variations between homologs. In most cases, however, it is common that few of the members have been studied for structure-function relationships, whereas high-resolution evidence from crystallography is scarce. To explore the spectrum of substrate selectivity determinants in such families, it is important to introduce effective mutagenesis designs for the comprehensive ab initio study of new homologs, based on existing evidence from known members. Capitalization on data from Cys-scanning or other systematic analyses of a well studied homolog, if available, offers a powerful approach to this end.
The above considerations apply promptly to the nucleobase-ascorbate transporter (NAT) 2 or nucleobase-cation symporter-2 (NCS2) family, an interesting example of a conserved but functionally diverse group of transporters present in all major taxa of organisms. NAT transporters model on the template of the uracil permease UraA, the first and only x-ray structure to be described recently for a member of this family, which represents a novel fold (1). With respect to function, only 15 of more than 2,000 predicted members have been characterized in detail; these are specific for the cellular uptake of uracil, xanthine, or uric acid (microbial, plant, and nonprimate mammalian genomes) or vitamin C (mammalian genomes) (1)(2)(3). Two of them, the xanthine permease XanQ of Escherichia coli (4 -10) and the uric acid/xanthine permease UapA of Aspergillus nidulans (11)(12)(13)(14)(15), have been studied extensively with Cysscanning mutagenesis and reverse and forward genetics, respectively. These studies have shown striking similarities between key NAT determinants of the two transporters, reinforcing the idea that few residues at conserved motifs of the family may be invariably critical for function or underlie specificity differences (for a summary of current knowledge, see supplemental Fig. S1). Based on homology modeling, most of these residues are found at the vicinity or at the periphery of the binding site in transmembrane segments TM1, TM3, TM8, and TM10 (1,10,15).
In the current work, we enrich the data set of functionally known NAT members by studying previously uncharacterized homologs from the genome of E. coli K-12, and we analyze * This work was supported by the research project EVOTRANS co-funded by function, substrate selectivity, and the role of key residues in one of them (YgfU), using mutagenesis designs that are based on the well studied homolog XanQ. Strikingly, the genome of E. coli K-12 contains 10 predicted members, of which the uracil permease UraA (16) and the xanthine permeases XanQ and XanP (4) are functionally known. Of the remaining members, three cluster together with UraA, XanQ, and XanP in COG2233 (YgfU, RutG, YbbY), and four cluster separately in COG2252 (YgfQ, YjcD, YicO, PurP). Characteristic NAT sequence motifs are retained only by the three COG2233 members (see Fig. 1A). Of them, RutG is related in sequence with the uracil permease UraA and is considered to be part of an unconventional pyrimidine utilization operon (17)(18)(19), whereas YgfU is found along with XanQ in a cluster of putative purine catabolic genes (20) and is related in sequence with xanthine or uric acid permeases from Gram-positive bacteria (21) (supplemental Table S1). The COG2252 members are more related in sequence with the fungal and plant AzgA-like adenine-guanine-hypoxanthine transporters (22,23), and one of them (PurP) has been proposed to be a high-affinity adenine transporter, based on genetic (24) and systems biology evidence (25).
In the context of this work, we cloned and overexpressed YgfU and showed that it is a proton-gradient dependent, lowaffinity (K m 0.5 mM), and high-capacity transporter for uric acid that also transports xanthine, but with disproportionately low capacity. Subsequently, we subjected YgfU to site-directed mutagenesis, based on data available for the homologous xanthine permease XanQ, and found that residues irreplaceable for the mechanism occur at five highly conserved positions, whereas a single-amino acid replacement (T100A) converts the uric acid-selective YgfU to a dual-selectivity transporter for both uric acid and xanthine. These results are supported with mirror-image replacements made in XanQ and homology modeling on the recently described structure of UraA.
DNA Manipulations-Construction of expression plasmids and biotin acceptor domain (BAD)-tagged versions of NAT homologs was essentially as described previously for XanQ and XanP (4). Briefly, the coding sequences of NAT genes were amplified by PCR on the template of genomic DNA prepared from E. coli T184 and transferred to plasmid vector pT7-5 by restriction fragment replacement; BAD-tagged versions were prepared using two-stage (overlap extension) PCR (28) on the templates of pT7-5/ygfU (or other NAT) and pT7-5/xanQ-BAD. For construction of mutants, two-stage PCR was performed on the template of YgfU-BAD or XanQ-BAD, as indicated. The entire coding sequence of all engineered constructs was verified by double-strand DNA sequencing in an automated DNA sequencer (MWG Biotech) (supplemental Table  S2).
Growth of Bacteria-E. coli T184 harboring given plasmids was grown aerobically at 37°C in Luria-Bertani medium containing streptomycin (0.01 mg/ml) and ampicillin (0.1 mg/ml). Fully grown cultures were diluted 10-fold, allowed to grow to mid-logarithmic phase, induced with isopropyl-1-thio-␤-D-galactopyranoside (0.5 mM) for an additional 2 h at 37°C, harvested, and washed with appropriate buffers.  (4). For kinetic uptake measurements, initial rates were assayed in T184 cells, at 5-20 s, and data were fitted to the Michaelis-Menten equation using Prism4, to determine K m and V max values. For ligand competition experiments with XanQ mutants, uptake of [ 3 H]xanthine (1 M) was assayed in the absence or presence of unlabeled analogues (1 mM) (4 -6).
In Silico Analysis-Comparative sequence analysis of NAT/ NCS2 homologs was based on BLAST-p search and ClustalW alignment; the most recent genome annotations were used for retrieving sequence data. Initial analysis of transmembrane topology of YgfU was performed using the program TMHMM (29). Homology threading was performed using the known x-ray structure of UraA as a template (Protein Data Bank (PDB) code 3QE7) on the web-based SWISSPROT modeling server (30); results were displayed and analyzed with PyMOL version 1.4 (Schrödinger, LLC).

RESULTS
YgfU Is a Uric Acid Transporter-The NAT genes ygfU, rutG, ybbY, ygfQ, yjcD, yicO, and purP were mobilized from the E. coli K-12 genome, transferred to transcriptional control of the lacZ promoter/operator in plasmid vector pT7-5, and induced for overexpression in E. coli T184 at conditions of negligible endogenous activity of oxidized purine or uracil uptake (4). Our data show that with one exception, all NAT constructs can be expressed in the E. coli membrane at high levels, comparable with the ones of XanQ used as a control (Fig. 1B), whereas transport activity at 25°C is observed only with YgfU, which transports [ 14 C]uric acid (1 mM) to high levels, and RutG, which has a marginally detectable activity for [ 3 H]xanthine (1 M) (Fig. 1C). Kinetic analysis reveals that RutG transports xanthine with 50-fold lower affinity (K m 0.15 mM) in comparison with XanQ or XanP (4), and YgfU transports uric acid with low affinity (K m 0.5 mM) and high capacity (V max 715 nmol min Ϫ1 mg Ϫ1 ) ( Table 1). The steady-state level of uric acid accumulation achieved with YgfU after 2-15 min of exposure to uric acid (1 mM) (Fig. 1C) is 381 Ϯ 4 nmol mg Ϫ1 ; taking into account an internal volume for E. coli of about 5.8 l mg Ϫ1 (31), this level corresponds to a concentration gradient of 128-fold (in/out). In addition, this uptake activity is abolished in the presence of the protonophore carbonyl cyanide m-chlorophenyl hydrazone, implying that transport is dependent on proton symport (32). With respect to xanthine uptake, YgfU displays a marginal xanthine uptake activity at 25°C, which precludes kinetic analysis, but it can transport xanthine at 37°C, albeit with very low affinity (K m 0.3 mM; 100-fold higher than the K m values of XanP or XanQ) and capacity (V max 14 nmol min Ϫ1 mg Ϫ1 ) ( Table 1).
Delineation of Mutagenesis Targets for Structure-Function Analysis of YgfU-The Cys-scanning and site-directed analysis of the homologous XanQ (5-10) offers a basis to search for targets of efficient mutagenesis designs in YgfU. Important residues of XanQ have been delineated at positions where a side chain is irreplaceable with respect to function (Glu-272, Asp-304, Gln-324, Asn-325) or expression in the membrane (Gly-83, Pro-318), replaceable with a limited number of other side chains (His-31, Asn-93, Phe-94, Asp-276, Ala-279, Thr-280, Gly-305, Gly-351, Pro-354, Ile-432), or highly sensitive to inactivation of a substituted Cys mutant by N-ethylmaleimide (Leu-86, Val-261, Gly-275, Ser-284, Ala-323, Asn-326, Gly-327, Val-328, Ile-329, Thr-332, Gly-333, Ser-336, Asn-430). Most of these 29 residues are highly conserved in the NAT/NCS2 family (Fig. 2), and YgfU differs from XanQ at 9 of the 29 positions (supplemental Fig. S1). To examine whether these residues contribute to the different substrate preference of YgfU, we engineered mutants replacing each one of the 9 residues with the corresponding amino acid found in XanQ. Mutagenesis was also performed at conserved side chains corresponding to the 4 functionally irreplaceable residues of XanQ and His-31, which is essential for substrate affinity (7). Thus, our initial mutagen-FIGURE 1. Sequence, immunoblot, and nucleobase uptake analysis of NAT homologs from E. coli. The full-length sequences of NAT members from E. coli K-12 were aligned using ClustalW, and part of this alignment referring to conserved sequence motifs of the family is shown in A. The members include the known nucleobase transporters XanQ (P67444), XanP (P0AGM9), and UraA (P0AGM7) and the unknown YgfU (Q46821), RutG (P75892), YbbY (P77328), YgfQ (Q46817), YjcD (P0AF52), YicO (P31440), and PurP (P31466). The motifs shown are characteristic for functionally known members of the family and have been shown to contain functionally important residues in XanQ (6 -10). Sequence identity scores are shown on the right. The coding sequences from the unknown homologs were mobilized from the genome, induced for overexpression in E. coli T184 under control of the lacZ promoter/operator in pT7-5/NAT-BAD, and assayed for expression levels in the membrane and active transport of xanthine, uracil, and uric acid. B, samples containing 100 g of membrane protein were subjected to SDS-PAGE (12%) and immunoblotting using HRP-conjugated avidin. C, cells expressing each homolog were normalized to 0.7 mg of protein/ml of cell suspension and assayed for active transport of [ 14 C]uric acid (1 mM, 25°C) in the presence or absence of carbonyl cyanide m-chlorophenyl hydrazone (CCCP) (10 M). Transport assays were also performed for [ 3 H]xanthine (1 M, 25°C), using XanQ as a positive control (4), and for [ 3 H]uracil (1 M, 25°C). Uracil uptake could not be detected with any of the unknown homologs; xanthine uptake was observed only with RutG, albeit to a marginal extent (not shown). esis targets included His-37, Thr-100, Thr-259, Glu-270, Met-274, Leu-278, Val-282, Asp-298, Ser-317, Gln-318, Asn-319, Val-320, Arg-327, and Ser-426 (residues conserved in XanQ are underlined) (Fig. 3).
Site-directed Mutagenesis of YgfU-Based on the above, we constructed and analyzed single-replacement mutants (a) E270D, E270N, E270Q, D298E, D298N, Q318N, Q318E, N319Q, N319D (conservative replacements of residues that are invariant and functionally irreplaceable in XanQ); (b) H37N, H37K, H37L, T100N, T100S, T100A (replacements of side chains that correspond to affinity-or specificity-related residues in XanQ); and (c) T259V, M274D, L278T, V282S, S317A, V320N, R327G, S426N (replacements of nonconserved residues of the important set with the corresponding amino acid found in XanQ). In addition, prompted by our initial data (see below), we engineered replacements at positions of TM8 that correspond to moderately N-ethylmaleimide-sensitive mutants (S256Q and S271A). The complete set of YgfU residues sub-jected to mutagenesis and their counterparts in XanQ is given in supplemental Fig. S1. After verification of the sequence, each YgfU mutant was transferred and expressed in E. coli T184 and assayed for its expression levels in the membrane and its ability to catalyze active transport of uric acid and xanthine (Fig. 4). Of the 25 mutants, four are undetectable in the membrane (T100N, S271A, M274D, V282S). Of the remaining 21 mutants, which display high protein levels comparable with wild type, nine show negligible uptake activities for uric acid or xanthine (E270D, E270N, E270Q, D298E, N319Q, N319D, H37K, H37L, V320N). Of the 12 active mutants, four show marginal activity for uric acid uptake (5-12% of wild-type) (Q318N, Q318E,  H37N, D298N), two accumulate uric acid at low (T100S, 20%) or intermediate levels (S317A, 40%), and six are highly active (60 -120%) (T100A, S256Q, T259V, L278T, R327G, S426N). With respect to xanthine uptake, significant activities above the wild-type threshold are only attained by mutants T100A, T100S, and S317A (Fig. 4C). Mutant T100A, in particular, which is highly active with both uric acid and xanthine, displays very significant xanthine uptake activity (about 50% of wildtype XanQ used as a control in these assays) even at 25°C, where wild-type YgfU is inactive (Fig. 5). Kinetic transport analysis shows that three mutants (T100S, R327G, S426N) have significantly lower affinity for uric acid than wild type (4-fold higher K m values), whereas T100A does not differ significantly from wild type in affinity but displays a 10-fold higher V max for uptake of xanthine at 37°C (Table 1). When comparing uric acid with xanthine uptake activity for wild-type and T100A, it is clear that this replacement converts the uric acid-selective YgfU to a dual-selectivity transporter for both uric acid and xanthine (Fig. 5D).
Mirror-image Replacements in XanQ-To evaluate further the role of nonconserved YgfU residues, we performed reverse site-directed mutagenic analysis in XanQ. Each one of the 11 important residues of XanQ that differ in YgfU was replaced individually with the corresponding amino acid found in YgfU. Thus, we constructed XanQ mutants Q258S, V261T, A273S, T280L, S284V, A323S, N326V, and N430S and assayed them for expression, uptake activity, and substrate selectivity in parallel with mutants N93T (7), N93A (10), D276M (9), and G333R (8) that had been engineered by us in previous studies. As shown in Fig. 6, all these mutants are expressed in the E. coli membrane to high levels. Of them, two show negligible uptake activities for xanthine or uric acid (Q258S, D276M), two display very low xanthine uptake activities (10 -20% of wild type) (N93T, N430S), seven display intermediate xanthine uptake activities (40 -60% of wild type) (N93A, V261T, T280L, S284V, A323S, N326V, G333R), and one is highly active for xanthine uptake (100 -120%) (A273S). On the other hand, only N93A shows a low but detectable uptake activity for uric acid (Fig. 6C). Kinetic transport analysis shows that five mutants have significantly lower affinity for xanthine than wild type (3-fold higher K m values for mutants N93T, A323S, N326V, and N430S and 6-fold higher K m value for A273S) (Table 1). Finally, the specificity profile of each mutant was investigated by assaying the extent of inhibition of [ 3 H]xanthine uptake in the presence or absence of a series of purine analogues ( Table 2). It is very interesting that despite their inefficiency for uric acid uptake, most 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. ND, assays were performed, but kinetic values were not determined due to very low uptake rates. All mutants and the wild-type versions used contained a carboxyl-terminal BAD.

DISCUSSION
The results in this study yield important new knowledge in two respects. First, it is shown for the first time that the E. coli K-12 genome encodes a uric acid permease activity (YgfU). Second, key determinants of the substrate binding site function and specificity of this permease are delineated, and Thr-100 in TM3 is demonstrated to be essential for the selectivity for uric acid.
The finding that the ygfU gene product in E. coli is a highcapacity transporter for uric acid raises questions on the putative physiological significance of such an uptake activity because E. coli is not currently known to contain enzymes for uric acid catabolism or to have a complete catabolic pathway for purines. Previous studies have shown that E. coli K-12 is unable to use purines other than adenine as sole nitrogen sources and, even with adenine, nitrogen assimilation is incomplete (20), but it can grow on allantoin as a sole nitrogen source, albeit only at anaerobic conditions (33). It also contains xanthine dehydrogenase activity (20), which can divert hypoxanthine and xanthine from the purine salvage pathway and produce uric acid. The missing link, however, is in the catabolic oxidation of uric acid to allantoin.
As shown relatively recently (34), the oxidation of uric acid to allantoin is expected to involve three enzymatic steps, the first one catalyzed by urate oxidase and yielding 5-hydroxy-isourate (HIU), the second one catalyzed by HIU hydrolase, which converts HIU to 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline (OHCU), and the third one catalyzed by OHCU decarboxylase, which converts OHCU to dextrorotatory S-(ϩ)-allantoin, the substrate for (S)-allantoin amidohydrolase (AllB) (35). In E. coli K-12, no enzymatic activity or any candidate E. coli gene has been reported for the step of oxidation of uric acid to HIU, whereas HIU hydrolase, and suggestively, OHCU decarboxylase activity has been assigned experimentally to the product of gene yedX (EcoGene database). 3 Interestingly, E. coli HIU hydrolase and its Bacillus subtilis homolog PucM (36) show clear sequence and structural similarity with the vertebral thyroid hormone distributor protein transthyretin, which probably originated from a duplication of an ancestral HIU hydrolase gene (37). In contrast to most bacterial homologs, the HIU hydrolase genes from E. coli and other enterobacteria appear not to be associated with purine metabolism operons and have been suggested to encode proteins with periplasmic localization, which might utilize HIU produced by a spontaneous, nonenzymatic reaction of uric acid acting as an antioxidant in the periplasm (38).
On the other hand, a complete gene cluster for purine utilization (hpx cluster) has been recently identified in Klebsiella pneumoniae (39) and Klebsiella oxytoca (40) containing separate genes for all three enzymatic steps of uric acid oxidation, namely hpxO (urate hydroxylase), hpxT (HIU hydrolase), and hpxQ (OHCU decarboxylase). Apart from HpxT (HIU hydrolase), which is homologous to E. coli HIU hydrolase and B. subtilis PucM, HpxO was found to function as a novel FAD-dependent urate oxidase (41), distinct from the classical urate oxidase (uricase) known for B. subtilis (36) and other species with complete purine utilization pathways (42). Parts of the hpx gene cluster of Klebsiella including the HpxO homolog are also encountered in other ␥-proteobacterial genera such as Xanthomonas and Acinetobacter (40) but not in E. coli. Finally, the hpx cluster contains a single putative permease gene (hpxP), which was suggested to encode a general purine permease (40). HpxP belongs to the NAT/NCS2 family and is closely homologous with the E. coli xanthine permease XanQ (45% sequence identity) but shares limited sequence homology with YgfU (25% identity).
Overall, it appears that there is more to learn on the purine utilization pathways in E. coli K-12, and our identification of YgfU as a uric acid transporter is an important contribution to this end. Although YgfU is found along with XanQ in a cluster of putative purine catabolic genes (20), this cluster does not appear to contain homologs for catabolic utilization of uric acid. In addition, YgfU has a limited array of orthologs in ␥-Proteobacteria (for example, only six homologs with Ͼ50% identity in other enterobacteria), but it shares up to 50% sequence identity with many homologs from Gram-positive bacteria including the known xanthine permease PbuX and uric acid permeases PucJ and PucK from B. subtilis (supplemental Table S1). Apart from uric acid, YgfU can utilize as a substrate xanthine, but with disproportionately low capacity, which is only apparent at 37°C (Fig. 5 and Table 1). In addition, the affinity of YgfU for xanthine is very low relative to the affinities of the xanthine permeases XanP and XanQ (4). Based on the above, the major substrate transported by YgfU appears to be uric acid, and we would like to suggest UacT (uric acid transporter) as an alternative, function-based designation for YgfU.
The comprehensive mutagenesis study of YgfU (UacT) presented herein reveals the key role of His-37 (TM1), Glu-270 (TM8), Asp-298 (TM9), and Gln-318 and Asn-319 (TM10), which are functionally irreplaceable, and of Thr-100 (TM3), which is essential for the uric acid selectivity. Structural modeling of YgfU based on the known x-ray structure of the uraciltransporting homolog UraA (1) shows that five of the six important residues fall at transmembrane helices, which form a binding shelter around substrate, and at least three of them (Glu-270, Gln-318, Thr-100) are in the substrate binding pocket (Fig. 7). In other members of the NAT/NCS2 family, these residues are either absolutely conserved (His-37, Glu-270, Asn-319) or conform to a distinct conservation pattern with limited side chain changes (Fig. 2). Most importantly, as well, critical roles have been described for the corresponding residues of other known NAT homologs (1,(7)(8)(9)(10)(11)(12)(13)43), as addressed in more detail below.
The five irreplaceable YgfU residues are conserved in the xanthine-transporting homolog XanQ and the dual-specificity uric acid/xanthine transporter UapA, and four out of five are also irreplaceable for function in XanQ (8,9) and in UapA (12,13). The irreplaceable role of the counterparts of Gln-318 and Asn-319, which belong to the conserved NAT signature motif (11), has been linked with defining the function of the binding  1 mM, 37°C). A, samples containing ϳ100 g of membrane protein were subjected to SDS-PAGE (12%) and immunoblotting using HRP-conjugated avidin. Prestained molecular mass standards (in kDa; Bio-Rad, low range) are shown on the left. B, initial rates of uric acid uptake (closed histogram bars) were measured at 5-15 s, and steady-state levels of uric acid accumulation (open histogram bars) were measured at 2-10 min. Control values obtained from T184 harboring pT7-5 alone were subtracted from the sample measurements in all cases. Results are expressed as a percentage of the activity of wild-type control with standard deviations from three independent determinations. C, comparison of initial rates of uric acid and xanthine uptake for the active YgfU mutants. Results are expressed as a percentage of the rate of YgfU and XanQ, respectively. The percentage ratio of the xanthine uptake value (red) to the uric acid uptake value (black) is given for each mutant in a third histogram bar (blue). site in both XanQ (5,8) and UapA (12,13), and site-directed alkylation studies in XanQ have implied that the corresponding residues interact directly with purine substrate (8). In the known x-ray structure of the pyrimidine-specific homolog UraA (1), the position of Gln-318 is occupied by a Glu (Glu-290), which forms two direct hydrogen bonds with substrate (uracil); the counterpart of Asn-319 (Asn-291) stabilizes the structure through an indirect hydrogen-bonding effect involving the counterpart of His-37 (His-24) in TM1. The role of another irreplaceable YgfU residue (Glu-270) might also be linked directly with binding because its counterpart in UraA (Glu-241) also forms two direct hydrogen bonds with substrate in the crystal structure (1). The fourth invariably irreplaceable residue (Asp-298 in YgfU) corresponds to a position of UraA, which is at the same depth in the membrane but distal from the binding site in the crystal structure and might have a distinct conformational role in the purine translocation mechanism, as suggested for permease XanQ (7,9). In any event, the fact that four conserved residues associated either directly or indirectly with binding are functionally irreplaceable in XanQ, YgfU (UacT), and UapA highlights the functional conservation of the purine binding site in NAT/NCS2 homologs of different selectivity.
His-37 (TM1) is absolutely conserved in the NAT/NCS2 family. Although replaceable in transporters other than YgfU, this histidine appears to be irreplaceable for function in YgfU (Fig. 4). From the other homologs, which include the bacterial XanQ (7), the fungal UapA (15), and the mammalian hSVCT1 (43) and hSVCT2 (44), a conserved hydrogen-bonding role is apparent for this residue, which is crucial for either the overall structural folding or the constitution of a proper affinity and specificity binding site or for both. Namely, the presence of an uncharged hydrogen-bonding residue at this position is essential for high-affinity binding and transport in XanQ (7) and in the hSVCTs (43,44), whereas in UapA, it has been suggested as crucial for proper folding and targeting to the plasma membrane (15). The recently elucidated crystal structure of the different-specificity homolog UraA (1) as well as homology models built on this structure (10,15) show that the conserved His of TM1 hydrogen-bonds to the conserved Asn of the NAT signature motif at the binding site periphery, substantiating further the previous findings. Based on the mutagenesis data of Fig. 4, it appears that this hydrogen-bonding interaction between His-37 and Asn-319 (Fig. 7) is more stringent in YgfU than in other homologs, and it may play an irreplaceable role with respect to the binding site integrity in this transporter.
Replacement of Thr-100 (TM3) with Ala changes the uric acid-selective YgfU to a dual-selectivity variant transporting both uric acid and xanthine with comparable efficiencies (Fig.  5), and replacement of the same residue with Ser also allows improved xanthine uptake relative to the uric acid uptake activity (Fig. 4). Thus, the side chain of Thr-100 is linked with the transporter ability to differentiate between uric acid (8-oxyxanthine) and xanthine. A similar role has been proposed for the corresponding amino acid in permease XanQ (Asn-93) because replacement of Asn-93 with Ala or Ser converts the  Table 1). The abbreviations are: UA, uric acid; and X, xanthine.
xanthine-selective XanQ to dual-selectivity variants (7,10); in the case of XanQ, however, the dual-selectivity variants transport the non-wild-type substrate (uric acid) with very low capacity (10). The selectivity-related role of Thr-100/Asn-93 appears to be unique because no other single-replacement mutant of either XanQ (6 -10) or YgfU (this study) has been shown to convert the native transporter to a dual-selectivity one. Interestingly, a similar albeit less prominent effect was observed with mutants replacing the corresponding residue (Ser-154) in the fungal homolog UapA, where introduction of an Ala leads to higher affinity for xanthine relative to uric acid, thus shifting the dual-selectivity profile to the xanthine-selective direction (15).
Although the side chains of Thr-100 or Asn-93 per se are poorly conserved in other homologs (10), the polar character of Thr-100/Asn-93 is conserved invariably in the known nucleobase-transporting NAT members (Asn, Thr, or Ser), whereas the ascorbate-transporting members have an Ala (Fig. 2). Furthermore, all the dual-selectivity uric-acid/xanthine transporters (Xut1, UapA, UapC, AfUapA, Lpe1) have a Ser at this position (Fig. 2). To understand the structural relevance of this difference, we have built models of the xanthine-selective XanQ, the uric acid-selective YgfU, and the dual-selectivity fungal homologs on the template of the known structure of UraA ( Fig. 7 and data not shown). The models indicate that the relevant position of TM3 is vicinal to the substrate binding site formed between residues of the middle parts of TM3, TM8, and TM10 (1). Strikingly, however, the Thr or Ser replacing Asn-93 in YgfU and all dual-selectivity NATs is distal from the conserved, substrate-relevant glutamate of TM8 (minimal distance between oxygen atoms, 6.0 Å), whereas Asn-93 in XanQ is sig-nificantly closer (distance between oxygen atoms, 4.5 Å). This difference is most prominent in YgfU (Fig. 7) and UapA (10), which conserve nearly all the other functionally important residues of TM1, TM3, TM8, or TM10, except Asn-93.
In UapA (10,15), occupation of the Asn-93 position by Ser orients this residue away from the substrate-relevant Glu of TM8, leading to relaxation of a constraint for the recognition of analogues at position 8 and allowing binding and transport of uric acid (8-oxy-xanthine), which shifts the NAT selectivity to a less stringent, dual-substrate profile. Accordingly, the XanQ mutants replacing Asn-93 with Ser or Ala yield efficient recognition of 8-methylxanthine and low but significant uptake of uric acid, mimicking in part the fungal, dual-selectivity NATs (10).
In YgfU, Thr-100 is also oriented away from the carboxyl group of Glu-270 (TM8), leaving more space in the substrate binding pocket but, at the same time, the pK a of Glu-270 may be distorted significantly relative to the corresponding carboxylic acid in XanQ due to its vicinity with hydrophobic groups from Thr-100 and Met-274 (Fig. 7B). These changes on the substrate binding Glu-272/Glu-270 might account for the selectivity difference between YgfU (uric acid) and XanQ (xanthine). Interestingly, however, replacement of Asn-93 with Thr in XanQ cannot imitate the YgfU profile but leads to low affinity for all xanthine analogues, possibly due to interference of the methyl group of Thr-93 in the vicinity of the essential Glu-272, which is not counterbalanced by other changes (10). An even more dramatic effect is observed in the current study with the mirrorimage YgfU replacement T100N, which leads to complete lack of expression in the membrane (Fig. 4). Based on these observations, it is evident that further combinatorial replacements are needed to promptly convert XanQ to the YgfU selectivity profile or vice versa, and targets for such replacements can be selected from residues of the important set analyzed here (Fig.  2), taking into account their orientation relative to the binding site in the modeled structures and functional properties of the relevant single-replacement mutants. 4 Apart from the results discussed above, our initial mutagenic analysis of YgfU permease revealed that site-specific replacements at TM8 (S271A, M274D, V282S) impair expression in the membrane, V320N (TM10) inactivates the permease, S317A (TM10) allows substantial uptake of xanthine, and R327G (TM10) or S426N (TM14) reduces the affinity for uric acid (4-fold increased K m ). The significance of these additional findings will be evaluated further with ongoing mutagenesis 4 E. Karena, K. Papakostas, and S. Frillingos, manuscript in preparation. FIGURE 7. Arrangement of important residues in modeled structure of YgfU. The sequences of YgfU and XanQ were threaded on the known x-ray structure of UraA (1) using the SWISSPROT modeling server. The structural models were displayed with PyMOL version 1.4. Shown are four TMs that contain binding residues and form a shelter around substrate in UraA, namely TM1 (gray), TM3 (wheat), TM8 (teal), and TM10 (salmon). A, arrangement of the four TMs in the YgfU model. B, the same model (YgfU) excluding the ␤-strand regions of TM3 and TM10 and indicating the side chains of residues analyzed with mutagenesis in this study. C, arrangement of the corresponding residues and helices in XanQ. The side chains of important residues of YgfU (B) and XanQ (C) are shown as thick sticks (irreplaceable and/or substrate binding-relevant) or spheres (sensitive to alkylation or bulky replacement in XanQ; crucial for the protein expression or affinity in YgfU). The side chain of Thr-100 (or Asn-93), which is associated with the uric acid selectivity, is shown with red label. studies in this and other NAT homologs. 5 In any event, it is important to note that several mutants of YgfU at positions of TM8, TM10, TM3, and TM1 flanking the binding site in the modeled structure lead to grossly impaired activity (H37N, V320N) or lack of expression in the membrane (T100N, S271A, M274D, V282S), whereas their counterpart or mirror-image replacement mutants in XanQ are active and expressed normally (Fig. 6). This observation implies that the binding site of YgfU is more prone to destabilization induced by peripheral side chain changes, whereas the reciprocal changes in XanQ are permissive for structure and function.
Finally, from a methodological point of view, it should be emphasized that all the results presented here for YgfU permease were based on site-directed mutagenesis designs deduced from Cys-scanning analysis data of a well studied homolog (XanQ). In this respect, we have shown that comprehensive analysis of structure-function relationships in a newly characterized transporter can be accomplished with relatively few site-directed replacements, based on the knowledge available from Cys-scanning mutagenesis of a prototypic homolog.