Contribution to Substrate Recognition of Two Aromatic Amino Acid Residues in Putative Transmembrane Segment 10 of the Yeast Sugar Transporters Gal2 and Hxt2*

The comprehensive study of chimeras between the Gal2 galactose transporter and the Hxt2 glucose transporter ofSaccharomyces cerevisiae has shown that Tyr446is essential and Trp455 is important for galactose recognition by Gal2. Consistent with this finding, replacement of the corresponding Phe431 and Tyr440 residues of Hxt2 with Tyr and Trp, respectively, allowed Hxt2 to transport galactose, suggesting that the two amino acid residues in putative transmembrane segment 10 play a definite role in galactose recognition (Kasahara, M., Shimoda, E., and Maeda, M. (1997) J. Biol. Chem. 272, 16721–16724). Replacement of Trp455 of Gal2 with any of the other 19 amino acids was shown to reduce galactose transport activity to between 0 and <20% of that of wild-type Gal2. The role of Phe431 in Hxt2 was similarly studied. Other than Phe, only Tyr at position 431 was able to support glucose transport activity, at the reduced level of <20%. In contrast, replacement of Tyr440 of Hxt2 with other amino acids revealed that most replacements, with the exception of Pro and charged amino acids, supported glucose transport activity. The importance of residue 431 in sugar recognition was more pronounced in a modified Hxt2 in which Tyr440 was replaced with Trp. Glucose transport was supported only by the aromatic amino acids Phe, Tyr, and Trp at position 431, and galactose transport was supported only by Tyr. These results suggest that an aromatic amino acid located in the middle of transmembrane segment 10 (Tyr446 in Gal2 and Phe431 in Hxt2) plays a critical role in substrate recognition in the yeast sugar transporter family to which Gal2 and Hxt2 belong.

The yeast Saccharomyces cerevisiae possesses nearly 20 homologous genes that potentially encode proteins belonging to the sugar transporter family (1)(2)(3)(4). This family includes both Hxt2, a major high affinity glucose transporter and Gal2, a major high affinity galactose transporter that also transports glucose with almost the same affinity (2,5). We have previously taken a three-step comprehensive approach to identify the amino acid residues responsible for substrate recognition in Gal2 and Hxt2 (5)(6)(7). In the first step (5), three types of systematic chimeras between Gal2 and Hxt2 were constructed with the use of the Escherichia coli homologous recombination system. The site responsible for differential recognition of galactose and glucose was localized to a COOH-terminal region of 101 amino acids. In the second step (6), the 101-amino acid region was subdivided into transmembrane (TM) 1 segment 10, TM11, TM12, and the proximal half of the COOH-terminal hydrophilic tail by introducing five restriction enzyme sites into the corresponding segment of each gene, without changing the encoded amino acids. By analyzing the 16 clones encoding all possible combinations of subdomains, we identified TM10 as the segment responsible for differential recognition of galactose and glucose. In the third step (7), a mixture of ϳ25,000 distinct plasmids that encoded all possible combinations of the 12 amino acids in TM10 that differ between Gal2 and Hxt2 was produced. All the 19 galactose transport-positive clones selected on galactose-limited agar plates encoded transporters containing the Tyr 446 residue of Gal2 (see Fig. 8). Fourteen of the 19 clones also encoded Trp 455 of Gal2 and the other five encoded Cys 455 , a residue not found in either Gal2 or Hxt2. To confirm that Tyr 446 is important for substrate recognition, we replaced this residue of Gal2 with each of the other 19 amino acids. None of the resulting transporters exhibited galactose transport activity. Replacement of Phe 431 and Tyr 440 of Hxt2 with the corresponding Tyr and Trp of Gal2 allowed the modified Hxt2 to transport galactose. These results demonstrated that Tyr 446 is essential and Trp 455 is important for the discrimination of galactose and glucose by Gal2.
The present study was designed to answer the following questions: (i) Is the apparently essential role of Tyr 446 in substrate recognition by Gal2 attributable to a contribution to the affinity of the transporter for galactose? (ii) Which amino acid residues can substitute for Trp 455 of Gal2? (iii) Is Phe 431 of Hxt2 essential for glucose recognition? (iv) Is Tyr 440 of Hxt2 important for glucose recognition? We obtained the results indicating that Tyr 446 is exclusively required for galactose recognition and two aromatic amino acids in TM10 play important roles in substrate recognition both in Gal2 and Hxt2.

EXPERIMETNAL PROCEDURES
Production of GAL2 and HXT2 Cassette Vectors-The cassette vector GAL2-pTV3e was constructed as described previously (5). Briefly, GAL2 (8) was introduced into the YEp-based vector pTV3 (YEp TRP1 bla) (9). After the disruption of an EcoRI site situated at the boundary of GAL2 and the vector by blunting, a new EcoRI site was introduced immediately downstream of the initiation codon by PCR (10), which resulted in a change in the encoded amino acid sequence from Met-Ala-Val-Glu to Met-Ala-Glu-Phe. A new ClaI site was also introduced immediately downstream of the termination codon of GAL2. Since HXT2 contains a single EcoRI site at the position corresponding to that of the newly created site in GAL2, it was necessary to introduce only a new ClaI site at the position corresponding to that in GAL2. These two sites were used to replace the open reading frame of GAL2 in GAL2-pTV3e * This work was supported by grants from the Ministry of Education, Science, Sports and Culture of Japan and from Teikyo University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. with that of HXT2, yielding HXT2-pTV3e. Both GAL2-pTV3e and HXT2-pTV3e were further modified to create five restriction enzyme sites (SacI, MluI, SpeI, StuI, and NcoI) in the distal half of each gene (6), yielding GAL2D-pTV3e and HXT2D-pTV3e, respectively (the StuI site in GAL2 and the NcoI site in HXT2 are preexisting sites).
Replacement of Specific Residues in Gal2 or Hxt2-Residues Tyr 446 and Trp 455 of Gal2 and the corresponding Phe 431 and Tyr 440 of Hxt2 were targeted for by each of the other 19 amino acids. The nucleotide sequence of Gal2 coding for Trp 455 was randomly modified by PCR with a degenerate primer. A forward primer 5Ј-CACGG TAAAA GCCAG CCGAG CTCTA AAGGT GCCGG TA encompassing the SacI site and a degenerate reverse primer 5Ј-TTCGA CTTGA CGCGT AGTGG GAATG ATTCT GCTGT GATGA CNNNG GCA AC TGGGG CCCAG GT that creates a new ApaI site at a position corresponding to that in HXT2 were used for long PCR (Ex Taq, Takara). Twenty-five cycles of 95°C for 30 s, 50°C for 2 min and 72°C for 1 min were performed with a thermocycler (model 2400, Perkin-Elmer). The PCR products were digested with SacI and MluI and introduced into the corresponding position of GAL2D-pTV3e. A total of 40 clones was sequenced, and 14 encoding the desired amino acids were selected (Table I). The remaining five clones were generated with the use of a specific primer for each amino acid. This series of clones was designated Gal2(Y-X). Replacement of Tyr 446 with each of the other 19 amino acids was performed in a similar manner, with this series of clones designated Gal2(X-W). For the Hxt2 transporter, four series of clones, designated Hxt2(X-Y), Hxt2(F-X), Hxt2(X-W), and Hxt2(Y-X) according to the same nomenclature were produced.
Other Assays-The nucleotide sequences of the substituted fragments and the surrounding regions in each clone were verified by sequencing both strands with a DNA sequencer (model 373A, Perkin-Elmer). Yeast cells were cultured as described (5-7). Transport of galactose or glucose in yeast strain LBY416(MAT␣ hxt2::LEU2 snf3::HIS3 gal2 lys2 ade2 trp1 his3 leu2 ura3) harboring the various Gal2 or Hxt2 plasmids was measured at 30°C for 5 s (7). Transport activities with 0.1 mM galactose or glucose were expressed as picomoles/ 10 7 cells/5 s. For comparison, the background obtained with control cells harboring the empty vector was subtracted from transport activity obtained with each clone, and the activity was normalized by expression as a percentage of that obtained with cells expressing Gal2 (galactose transport) or Hxt2 (glucose transport). Although in some instances, values of Ͻ10% were significant by Student's t test (p Ͻ 0.05), we considered only values of Ͼ10% as significant, since cell numbers determined with Burker-Turk cell counters were somewhat variable, and some transporters showed activities that were less than the control values, sometimes as low as Ϫ5 to Ϫ10%. Immunoblot analysis of cell homogenates was performed as described previously (5). Briefly, polyclonal rabbit antibody to Gal2 or Hxt2 was produced by using the COOH-terminal 13 or 14 oligopeptide, respectively, that was coupled to keyhole limpet hemocyanin. Yeast cells grown to an early log phase were washed with H 2 O and disrupted with glass beads. The homoge-nate was subjected to SDS-gel electrophoresis and blotted onto a polyvinylidine difluoride membrane (Immobilon P SQ , Millipore), followed by incubation with 125 I-protein A (IM144, Amersham Pharmacia Biotech) overnight. Autoradiography was performed with imaging plates (BAS2000, Fuji Film). Under the present conditions, a linear relation of amounts of protein and radioactivity was observed. These antibodies reacted with Gal2 or Hxt2, but not with other homologous proteins revealed by the yeast genome sequence (5,6). To evaluate the possible degradation of modified Hxt2 transporters in vivo and in vitro, we used the protease-deficient strain BJ3505 (MATa pep4::HIS3 prb1-⌬1.6R his3 lys2 trp1 ura3 gal2 can1) in place of LBY416 or added a mixture of protease inhibitors consisting of 5 mM EDTA, pepstatin A (2 g/ml), leupeptin (2 g/ml), aprotinin (100 units/ml), and 2 mM phenylmethylsulfonyl fluoride (final concentrations) to the homogenization solution. Neither approach appeared to affect the amounts of these transporters.

Effects of Substrate Concentration on Galactose and Glucose
Transport by Gal2(X-W) Transporters-Our previous studies with systematic series of chimeras of Gal2 and Hxt2 revealed that Tyr 446 in Gal2 is essential for the differential recognition of galactose and glucose (5-7). To characterize further the role of Tyr 446 in substrate recognition, we measured transport of galactose or glucose by Gal2(X-W) transporters at a substrate concentration (10 mM) 100 times that used previously (Fig. 1). Even at this high substrate concentration, substantial galactose transport was mediated only by Gal2(Y-W); Gal2(F-W) showed 12% of the galactose transport activity of Gal2, whereas the other Gal2(X-W) transporters showed no significant activity (Fig. 1A). Glucose transport at the high substrate concentration was mediated by many of the Gal2(X-W) transporters (Fig. 1B). Gal2(Y-W) and Gal2(F-W) were most active, with Gly, Ala, Val, Leu, Met, Cys, Ser, Thr, and Asn substitutions also conferring transport activity at the high glucose concentration. Comparison of glucose transport activities between substrate concentrations of 0.1 and 10 mM indicates that the transporters with glucose transport activities of ϳ20% at 10 mM glucose possess K m values of ϳ100 mM. The differences in transport activity were not due to differences in the extent of expression of Gal2(X-W) transporters, as revealed by immunoblot analysis of cell homogenates ( Fig. 2A).
Galactose and Glucose Transport by Gal2(Y-X) Transporters-We have previously shown that Trp 455 of Gal2 is important for the differential recognition of galactose and glucose (7). The replacement of Trp 455 with any of the other 19 amino acids The nucleotide sequences of GAL2 and HXT2 were modified to encode Gal2(X-W), Gal2(Y-X), Hxt2(X-Y), Hxt2(F-X), Hxt2(X-W), and Hxt2(Y-X) transporters as described under "Experimental Procedures." The original codons are UAU(Y 446 ) in GAL2 and UUC(F 431 ) and UAC(Y 440 ) in HXT2.
markedly reduced galactose transport to Ͻ20% of that of Gal2, although significant activity was exhibited by transporters containing Tyr, Cys, Thr, Ile, or Met at this position (Fig. 3). In glucose transport assays, Gal2(Y-W), or wild-type Gal2, showed only 28% of the activity of Hxt2; this may be mainly due to Tyr 446 since replacement of this residue with Phe increased the glucose transport activity to 60% of that of Hxt2 (Fig. 1B). Gal2(Y-X) transporters containing Tyr, Phe, or Thr showed almost the same glucose transport activity as did the original Gal2. Thus, for galactose transport, Trp was the most favorable amino acid at position 455, with other amino acids conferring little activity; whereas for glucose transport, replacement of Trp with many other amino acids supported activity, although not to the same extent as did Trp. The expression levels were similar for each of the Gal2(Y-X) transporters (Fig. 2B). Importance of Phe 431 and Tyr 440 of Hxt2 for Glucose Transport-Our previous study (7) and the results described above indicate that Tyr 446 in Gal2 is essential for the recognition of galactose. Since Gal2 and Hxt2 share an amino acid sequence identity of ϳ70% (5), it is reasonable to hypothesize that Phe 431 of Hxt2, corresponding to Tyr 446 of Gal2, may be important for recognition of glucose. Transport of glucose or galactose in cells expressing Hxt2(X-Y) transporters was measured at a substrate concentration of 0.1 mM (Fig. 4A). The only transporter other than Hxt2(F-Y), or wild-type Hxt2, that exhibited significant glucose transport activity was Hxt2(Y-Y), which showed an activity of 17% of that of Hxt2. In contrast, none of the by cells expressing each mutant transporters was measured. After subtraction of the control value (3.9 Ϯ 0.2 pmol/10 7 cells/5 s (n ϭ 36) at 0.1 mM glucose; 153 Ϯ 8 pmol/10 7 cells/5 s (n ϭ 9) at 10 mM glucose), the transport activity in experimental cells was expressed relative to that of cells expressing Hxt2 (67.4 Ϯ 2.9 pmol/10 7 cells/5 s (n ϭ 28) at 0.1 mM glucose; 973 Ϯ 22 pmol/10 7 cells/5 s (n ϭ 9) at 10 mM glucose). Data shown are mean Ϯ S.E. (n Ն 3). Values of Ͼ10% were considered significant. Although a part of data on galactose and glucose transport at 0.1 mM substrate concentration has been reported in the previous study (7), additional data are included and described in this study.

FIG. 2. Expression of Gal2(X-W) and Gal2(Y-X) transporters as detected by immunoblot analysis. LBY416 cells harboring plasmids
containing GAL2, each of 20 Gal2(X-W) genes (A), or each of 20 Gal2(Y-X) genes (B) were cultured to early log phase and homogenized. A portion of each homogenate (10 g of protein) was subjected to immunoblot analysis with antibodies to the COOH terminus of Gal2 and 125 I-labeled protein A (5). The 53,000 (53k) recombinant transporters were detected by autoradiography with imaging plates (BAS2000, Fuji Film). Hxt2(X-Y) transporters supported galactose transport. Some of the Hxt2(X-Y) transporters were expressed at a low level, as revealed by immunoblot analysis (Fig. 5A). Transporters containing Pro or charged amino acids at position 431 tended to show a reduced level of expression, suggesting that a proper conformation of the protein is necessary for it to maintain its integrity, although the possibility of protein degradation during preparation can not be excluded at present (see "Experimental Procedures" and "Discussion").
To study the role of Tyr 440 of Hxt2, we measured glucose transport in cells expressing Hxt2(F-X) transporters (Fig. 4B). In contrast to Hxt2(X-Y) transporters, many Hxt2(F-X) transporters showed high glucose transport activities. However, no significant glucose transport activity was apparent with transporters containing Pro or the charged amino acids Asp, Glu, His, Lys, or Arg at position 440. None of the Hxt2(F-X) transporters exhibited galactose transport activity. Since the expression of transporters containing these charged amino acids was markedly reduced (Fig. 5B), the results obtained with the corresponding cells were inconclusive.
Galactose Transport by Doubly Modified Hxt2(X-W) and Hxt2(Y-X) Transporters-In the previous study (7), to confirm that Tyr 446 and Trp 455 of Gal2 are important for galactose recognition, we replaced the corresponding amino acids of Hxt2, Phe 431 , and Tyr 440 , with Tyr and Trp, respectively. The resulting Hxt2(Y-W) transporter mediated galactose transport in addition to glucose transport. In the present study, we constructed the Hxt2(X-W) series of transporters to determine whether amino acids other than Tyr at position of 431 of Hxt2 can support galactose transport. Hxt2(Y-W) was the only transporter in the Hxt2(X-W) series that exhibited galactose transport activity (Fig. 6A). This apparently essential role for Tyr 431 in galactose transport by doubly modified Hxt2 suggests that, in wild-type Hxt2, Phe 431 together with Tyr 440 acts to admit glucose and reject galactose. Hxt2(Y-W) showed a K m for galactose of 43 mM, which is about eight times that of Gal2 and almost identical to that of Hxt2-8, which possesses a Gal2derived TM10 (6). For glucose transport, the importance of aromatic amino acids at position 431 was apparent, with only Hxt2(X-W) transporters containing Phe, Tyr, or Trp exhibiting glucose transport activity.
Hxt2(Y-X) transporters behaved differently (Fig. 6B). For galactose transport, Trp, Met, Cys, Thr, and Leu were effective, but the aromatic amino acids Phe and Tyr were not. Thus, the combination of Tyr 431 and Trp 440 present in Gal2 was also most effective in conferring galactose transport activity on the background of Hxt2. Glucose transport in cells expressing Hxt2(Y-X) transporters was at most ϳ25% of that in cells expressing Hxt2, a value similar to that obtained with cells expressing Gal2. Glucose transport at this reduced level was observed with Hxt2(Y-X) transporters containing a wide range of amino acids, including Trp, Phe, Tyr, Cys, Thr, Met, Leu, and Ile. Taken together, our data indicate that the combination of Phe 431 and Tyr 440 is most accepting of glucose and rejects galactose efficiently. The extent of expression of Hxt2(Y-I) and Hxt2(Y-D) was low, indicating that no definitive conclusions can be drawn for these mutants (Fig. 7B).

DISCUSSION
In this study, we have demonstrated the importance of two aromatic amino acid residues in putative TM10 for substrate recognition by Gal2 and Hxt2. Hereafter, the aromatic amino acid situated in the middle of TM10, Tyr 446 in Gal2 and Phe 431 in Hxt2, is referred to as the middle aromatic site (Fig. 8), and

FIG. 5. Expression of Hxt2(X-Y) and Hxt2(F-X) transporters as detected by immunoblot analysis.
Cells harboring plasmids containing HXT2, each of 20 Hxt2(X-Y) genes (A) or each of 20 Hxt2(F-X) genes (B) were cultured to early log phase and homogenized. A portion of 10 g was subjected to immunoblot analysis with antibodies to the COOH terminus of Hxt2 (5). The bands immediately below the 47,000 (47k) recombinant proteins appear to be degradation products; their abundance increased by repeated freeze-thaw treatment of homogenates. that situated at the cytoplasmic end of TM10, Trp 455 in Gal2 and Tyr 440 in Hxt2, is referred to as the cytoplasmic aromatic site.
We previously showed (7) that Tyr 446 of Gal2 is essential for the differential recognition of galactose and glucose on the basis that none of 19 other amino acids was able to replace it for galactose transport. However, it remained possible that the apparent inability of other amino acids to support galactose transport may have been overcome by higher concentrations of substrate. In the present study, to determine whether the role of the middle aromatic site is merely to contribute to the affinity for galactose, we compared galactose transport at substrate concentrations of 0.1 and 10 mM. Our observation that, even at the high galactose concentration, substantial galactose transport was mediated only by Gal2(Y-W), with Gal2(F-W) showing a small amount of activity, is consistent with the idea that Tyr at the middle aromatic site is not merely a contributor to the affinity for galactose, but rather is exclusively required for the selection of galactose. In contrast, for glucose transport, Tyr at the middle aromatic site of Gal2 was not essential. The observation that other amino acid residues were able to replace it, especially at the high substrate concentration, indicates that the role of this Tyr residue in glucose transport is to increase the affinity for glucose. Glucose transport in cells expressing Hxt2(X-Y) transporters was supported by Phe and, to a much lesser extent, by Tyr. A similar tendency was apparent with Hxt2(X-W) transporters, in which the requirement for an aromatic amino acid showed an order of preference of Phe Ͼ Ͼ Tyr Ͼ Trp, as well as with Gal2(X-W) transporters, which showed the same rank order of preference at the low glucose concentration. These observations suggest that the glucose recognition process is similar in both transporters. The apparently different roles of the middle aromatic site in galactose and glucose transport might be indicative of multiple functions of the amino acid at this site (see below).
With regard to the cytoplasmic aromatic site, galactose transport by Gal2(Y-X) transporters was preferentially supported by Trp, with several other amino acids, including Tyr Ͼ Cys Ϸ Met Ϸ Thr Ϸ Ile, also conferring activity. Glucose transport by Hxt2(F-X) transporters was supported by most amino acids, excluding Pro and charged residues. Similar patterns of glucose transport activities were observed with Hxt2(Y-X) and Gal2(Y-X) transporters, although Tyr at the middle aromatic site appeared to reduce the activities of these series. These results indicated that the interaction of the cytoplasmic site with glucose may be similar in Gal2 and Hxt2. It should be mentioned that the variable level of expression of Hxt2(F-X) and Hxt2(Y-X) transporters revealed by immunoblot analysis may not necessarily reflect the amounts of the transporters in intact cells, since the expression level of Gal2(Y-X) transporters was not variable and yet they showed a pattern of glucose transport activity similar to those of the Hxt2(F-X) and Hxt2(Y-X) transporters.
A total of 18 closely related sugar transporters has been identified in S. cerevisiae, the cluster I of sugar permease homologs (4) that is equivalent to the hexose transporter family (3), excluding Snf3 and Rgt2, which appear to be glucose sensor (11). Of these 18 homologs, only Gal2 contains Tyr at the middle aromatic site and Trp at the cytoplasmic site. All the other 17 homologs, including Hxt2, contain Phe at the middle site. At the cytoplasmic site, Tyr is present in most homologs and Phe in Hxt4, Hxt12, and Hxt14. Together with our demonstration that Phe was the second most active residue in Hxt2(F-X) transporters in terms of glucose transport, these observations are consistent with the idea that Gal2 is the only galactose transporter and that the other 17 transporters are glucose transporters. If this is the case, low affinity galactose transport noted in previous studies (2) may be attributable to one or more of the glucose transporters carrying galactose with low affinity. This notion is also consistent with the selection of galactose transport-positive clones from gal2 mutants (7) and the selection of glucose transport-positive clones from strains carrying gal2 and multiple hxt mutations in the presence of antimycin (12).
Several functions for the two aromatic sites in galactose transport by Gal2 have been proposed (7), including roles in steric hindrance, hydrogen bonding, a stacking effect with the sugar, and structural changes in other amino acids that interact directly with the sugar. In addition, the possibility that the two amino acids function independently has also been suggested. Recent structural studies on porins (13,14) may be worth mentioning in this respect. The three-dimensional structures of maltoporin and the sucrose-specific porin SrcY analyzed by x-ray crystallography indicate that maltose and sucrose passes through the corresponding porins via a relay of aromatic amino acids. Since most galactose and glucose molecules are in the form of ␤and ␣-pyranose in aqueous solution (15), differential recognition of galactose and glucose requires differentiation between ␣and ␤-anomers and C4 epimers.
Crystallographic studies of binding proteins in the periplasm of bacteria (16) and of lectins (17), in addition to the porin studies, have shown that the binding of sugars to these proteins is mediated by hydrogen bonding to various amino acids and H 2 O and stacking of the sugars with aromatic amino acids. Thus, further insight into the molecular mechanism of substrate recognition by Gal2 and Hxt2 should be provided by identification of the amino acid residues that presumably form hydrogen bonds with galactose and glucose located at the middle site. In addition, it will be important to determine whether other aromatic amino acids in these transporters contribute to substrate binding.  (1). Of the 35 amino acids in TM10, those common to Gal2 and Hxt2 are shown as shaded circles, and those that differ are indicated by the one-letter amino acid code. The aromatic amino acids located at the middle of TM10, Tyr 446 in Gal2, and Phe 431 in Hxt2 as well as those located at the cytoplasmic end, Trp 455 in Gal2 and Tyr 440 in Hxt2, are also indicated.