Alteration of substrate affinities and specificities of the Chlorella Hexose/H+ symporters by mutations and construction of chimeras.

The cDNAs HUP1 and HUP2 of Chlorella kessleri code for monosaccharide/H+ symporters that can be functionally expressed in Schizosaccharomyces pombe. By random mutagenesis three HUP1 mutants with an increased Km value for D-glucose were isolated. The 40-fold increase in Km of the first mutant is due to the amino acid exchange N436I in putative transmembrane helix XI. Two substitutions were found in a second (G97C/I303N) and third mutant (G120D/F292L), which show a 270-fold and 50-fold increase in Km for D-glucose, respectively. An investigation of the individual mutations revealed that the substitutions I303N and F292L (both in helix VII) cause the Km shifts seen in the corresponding double mutants. These mutations together with those previously found support the hypothesis that helices V, VII, and XI participate in the transmembrane sugar pathway. Whereas for most mutants obtained so far the Km change for D-glucose is paralleled by a corresponding change for other hexoses tested, the exchange D44E exclusively alters the Km for D-glucose. Moreover the pH profile of this mutant is shifted by more than 2 pH units to alkaline values, indicating that the activity of the transporter may require deprotonation of the corresponding carboxyl group. Chimeric transporters were constructed to study the 100-fold lower affinity for D-galactose of the HUP1 symporter as compared with that of the HUP2 protein. A crucial determinant for the differential D-galactose recognition was shown to be associated with the first external loop. The effect could be pinpointed to a single amino acid change: replacement of Asn-45 of HUP1 with isoleucine, the corresponding amino acid of HUP2, yields a transporter with a 20 times higher affinity for D-galactose. The reverse substitution (I47N) decreases the affinity of HUP2 for D-galactose 20-fold.

The green alga Chlorella kessleri possesses an inducible transport system, capable of accumulative uptake of a variety of monosaccharides using an electrochemical proton gradient as driving force (1)(2)(3)(4). Three cDNAs coding for highly homologous Chlorella monosaccharide/H ϩ symporters were cloned by differential screening (5,6) and named HUP1Ϫ3 (hexose uptake protein). Their identities have been confirmed by heterologous expression in Schizosaccharomyces pombe (6,7). Fur-thermore, the HUP1 transporter retains its uptake activity after solubilization from the membrane of transgenic fission yeast, purification to homogeneity, and reconstitution into proteoliposomes (8,9).
The HUP symporters belong to a large family of substrate transporters, called the "major facilitator superfamily" (10). Members of this major facilitator superfamily are thought to consist of 12 ␣-helical transmembrane segments connected by internal and external loops. Support for this topological model comes from alkaline phosphatase fusion protein analysis of the Escherichia coli lactose permease lacY (11) and N-glycosylation scanning mutagenesis studies on the human glucose facilitator GLUT1 (12). However, hard structural data on the nature of the binding sites and translocation pathways of substrates and cosubstrates have not been obtained. Since no three-dimensional structure of a transporter is in sight, one has to be content with indirect evidence, deduced for example from mutagenesis studies.
Structure-function analysis of the HUP1 transporter (13,14) was carried out in a sugar uptake deficient S. pombe strain (15). Several mutants with an increased K m value for D-glucose uptake were found by site-directed mutagenesis (13) and by polymerase chain reaction random mutagenesis with subsequent selection for decreased sensitivity toward the toxic sugar 2-deoxyglucose (14). The affected amino acids cluster in the middle of the transmembrane helices V (Gln-179), VII (Gln-298 and Gln-299), and XI (Val-433 and Asn-436), with the exception of Asp-44 putatively located at the beginning of the first external loop (Fig. 1). The fact that predominantly acidic amino acids and their amides were identified correlates well with the finding that binding sites of periplasmic sugar-binding proteins are built up by such residues (16,17).
The symporters HUP1 and HUP2 differ significantly in their substrate specificity (6,18). Especially, the affinity for D-galactose is more than 100 times higher for the HUP2 protein. The amino acids of the HUP1 protein probably involved in substrate recognition (see above) are also present in the HUP2 transporter. The different substrate specificities of the two transporters must, therefore, be determined by differing residues at still unidentified positions. Recently, a study using chimeric proteins revealed that the exchange of a 30-amino acid span at the beginning of the first extracellular loop of HUP1 for that of HUP2 increases the affinity for D-galactose by about 15-fold (18).
The present work tries to find answers to the following questions. 1) Do additional residues exist in the HUP1 symporter, which give rise to an increased K m value for D-glucose uptake upon replacement? 2) Do all these HUP1 mutants also exhibit decreased affinities for other sugars, or do some of them show substrate specific effects? 3) Is it possible to narrow down the segment of the first external loop of HUP2 participating in D-galactose recognition?

EXPERIMENTAL PROCEDURES
Strains and Growth Conditions-All cloning steps were carried out in E. coli DH5␣ with the plasmid vector pUC18. E. coli TG1 served as host for the phagemid pUC118 and the helper virus M13KO7 in sitedirected mutagenesis. The Leu Ϫ and sugar uptake deficient strain S. pombe YGS-B25 (15) used for heterologous expression of the various transporter cDNAs was grown in 2% gluconate, 2% yeast extract. Transformed S. pombe cells were cultivated in minimal medium containing 2% gluconate and 0.67% yeast nitrogen base without amino acids.
Random Mutagenesis by Polymerase Chain Reaction-The fulllength cDNA of HUP1 was amplified by polymerase chain reaction under suboptimal conditions as described previously (14) in order to achieve one error per cDNA fragment on average. The pool of randomly mutagenized cDNAs was ligated into pEVP11 or pART3 and introduced in S. pombe cells leading to the RMY and RGY transformants, respectively.
Recovery and Sequencing of the Mutated HUP1 cDNAs-Plasmid reisolation from RGY52 was performed by the phenol/chloroform/isoamyl alcohol procedure (21) with the following addition. The aqueous phase containing the recovered plasmids was purified and precipitated by successive treatment with phenol, diethyl ether, and ethanol prior to transformation of E. coli DH5␣. However, this procedure for plasmid isolation from yeast cells as well as several others failed in the cases of RMY126 and RMY254. Therefore one big colony of each of these transformants was picked and directly applied to a standard polymerase chain reaction. Plasmids were released from the cells due to preincubation at 94°C for 10 min. The mutated HUP1 cDNAs were amplified afterward using flanking primers that bind in the promotor and the polylinker region of pEVP11. Then they were subcloned via SacI/ BamHI into pUC18 and their nucleotide changes were determined by sequence analysis using the T7 Sequencing TM kit (Pharmacia Biotech) and synthetic oligonucleotides.
Separation of the Mutations of the Double Mutants RMY126 and RGY52-The transformants RMY126 and RGY52 both exhibited two point mutations in the HUP1 gene (see "Results"). These mutations could be separated using a unique KpnI restriction site lying in between (Fig. 1). The SacI/KpnI fragment and the KpnI/BamHI fragment coding for the N-and C-terminal part were ligated to the respective missing sequences from the wild-type clone. This resulted in HUP1 coding regions carrying either the one or the other mutation. Those originating from RMY126 were resubcloned into pEVP11, those originating from RGY52 into pART3. S. pombe YGS-B25 was transformed as described above.
Site-directed Mutagenesis-Preparation of the single-stranded HUP1 and HUP2 template DNA was performed as described previously (13,18). Site-specific mutagenesis was carried out with the Sculptor TM in vitro mutagenesis system (Amersham) according to the instructions of the manufacturer. The sequences of the synthetic oligonucleotides Construction of the Chimeric HUP1/2/1 cDNAs-We previously reported (18) the construction of a chimeric HUP1/2/1 transporter (C6), which consists mainly of HUP1 sequence. Only the front part of the first extracellular loop is derived from the HUP2 symporter. The HUP1 fragment coding for this loop section was excised by EcoRV/BsgI digestion (see Fig. 1) and replaced with the equivalent HUP2 fragment. The EcoRV restriction site had first to be introduced into the cDNAs of the wild-type transporters without changing the amino acid sequences. Approximately in the middle of the exchanged loop segment exists a unique Asp700I restriction site ( Fig. 1) at homologous positions in HUP1 and HUP2. Substitution of the Asp700I/BsgI fragment of HUP1 for that of HUP2 resulted in the chimeric cDNA C7. C8 was generated in the same way by replacement of the EcoRV/Asp700I fragment. All chimeric HUP1/2/1 cDNAs possess exactly the same 5Ј-untranslated sequence as the wild-type HUP1 and were cloned via SacI/BamHI into the expression vector pEVP11.
Transport Assays-Five to 20 ml of S. pombe cells (OD 578 ϭ 1.0) were harvested, washed once in 5 ml of 100 mM potassium phosphate buffer, pH 6.0, and resuspended in the same buffer to a final volume of 1 ml. Cells were optimally energized by adding ethanol to a final concentration of 120 mM. After 2 min of shaking at 30°C the test was started by adding radioactive sugar. Samples were withdrawn at given intervals, filtered through nitrocellulose filters (0.8 m pore size), and washed once with distilled water. Incorporation of radioactivity was determined by scintillation counting. In order to obtain the K m and V max values, initial uptake rates were measured at different substrate concentrations and plotted according to Lineweaver-Burk. Since nontransformed S. pombe cells (YGS-B25) do not show measurable hexose uptake activity, transport rates of mutated HUP gene products as low as 0.1% of that of wild-type HUPs expressed in S. pombe could reliably be measured. When the effect of external pH was tested, cells were washed and resuspended in McIlvaine buffer (adjusted to a given pH in the range from pH 3 to 7 by mixing 100 mM citric acid with 200 mM Na 2 HPO 4 ) or 50 mM Tris/HCl buffer (adjusted to a given pH in the range from pH 7 to 9). All radioactive sugars were D-[U-14 C]compounds purchased from Amersham.
Isolation of Total Membranes, SDS-Polyacrylamide Gel Electrophoresis, and Immunoblotting-S. pombe cells of a 30-ml culture (OD 578 Ϸ 1) were pelleted by centrifugation. Their membranes were isolated as described (13). The protein content was assayed by the method of Bradford (22). SDS-polyacrylamide gel electrophoresis was carried out according to Laemmli (23); proteins were transferred electrophoretically to nitrocellulose and incubated overnight with polyclonal anti-HUP1-A antibody (13). The blot was immunodetected with the ECL kit of Amersham. Expression of mutant cDNAs was compared with that of wild-type HUP1 cloned in the same vector (pEVP11 or pART3).

Random Generation of HUP1 Mutants with an Increased K m
Value for D-Glucose-The K m value is an expression independent measure of the affinity of the HUP1 transporter for its sugar sustrate(s). Therefore K m mutants should lead to the identification of amino acid residues most probably involved in substrate binding. Recently we reported an unbiased functional screening for such K m mutants (14). It is based on a 1000-fold increase in 2-deoxyglucose sensitivity upon transformation of a sugar uptake deficient S. pombe strain with the HUP1 cDNA. A pool of randomly mutated HUP1 cDNAs was generated by polymerase chain reaction and used for transformation. Transformants with intermediate 2-deoxy-D-glucose sensitivity were selected and tested for decreased affinity for D-glucose. Four K m mutants had been obtained in this way (14). In the meantime further use of this strategy has been made and three additional mutants have been isolated (Table I).
The HUP1 cDNA isolated from RMY254 exhibits three nucleotide changes, but only one of them affects the primary structure of the HUP1 protein. The substitution of the asparagine residue at position 436 in helix XI for isoleucine decreases the affinity for D-glucose 40-fold. The cDNAs of the other K m mutants carry only two point mutations, which both result in amino acid changes. Thus, the HUP1 symporter of RMY126 contains cysteine and asparagine instead of glycine 97 (helix II) and isoleucine 303 (helix VII), respectively. The K m value for D-glucose is dramatically increased by a factor of about 270. The point mutations in the HUP1 cDNA of RGY52 alter the transport protein in the following way. Glycine 120 (helix III) is replaced by an aspartic acid and phenylalanine 292 (helix VII) is changed to leucine. As a consequence the K m value for D-glucose rises approximately 50-fold as compared with that of the wild-type.
These K m mutants also show a dramatic decrease in V max value, which correlates well, however, with the poor level of expression (Table I). A reduced amount of transport protein is not unusual for clones originating from the random mutagenesis approach, since a lower uptake rate also contributes to higher 2-deoxy-D-glucose resistance (14). It is not really under-stood, however, why these mutations lead to such a low expression of transport protein. The possibility that mutations affect protein secondary structure and/or protein stability, leading to indirect effects on K m values, unfortunately cannot be ruled out. Although, in principle this is true for all mutations described here or elsewhere, particular caution is required for those reducing the expression level.
Separation of the Mutations in the HUP1 Double Mutants-In order to elucidate whether the K m change found in RMY126 and RGY52 is brought about by both substitutions acting additively or whether it is simply caused by one of them, the mutations were separated from each other as described under "Experimental Procedures." Table II lists the K m values of the single mutants. Obviously, the effect of substituting Gly-97 for Cys on the D-glucose affinity of HUP1 is negligible. The nearly identical K m values of mutant and wild-type transporter manifest that glycine 97 is not important for the interaction with the substrate. On the other hand mutant I303N exhibits a K m value very similar to that of the double mutant RMY126. Therefore, it is suggested that the isoleucine residue in helix VII is involved in D-glucose binding.
In the case of RGY52 the results are more complex. Both substitutions in question generate effects on D-glucose affinity, but they are not additive. The replacement F292L causes a 50-fold increase of the K m value equal to that of the double mutant. In the G120D mutant the K m value is raised somewhat less but still significantly (about 30-fold). It is not clear why the effects are not additive to some extent. It seems that the structural aberration caused by F292L includes the one induced by G120D. Since G120D is a much more drastic change when compared with F292L, the results are taken as indication that the phenylalanine residue in helix VII is important for correct substrate binding. Finally, it should be emphasized that for all single mutants the V max value correlates very well with the amount of transport protein detected by immunoblotting (Table  II).
Substrate Specificity of Several Mutated HUP1 Transporters-The Chlorella HUP1 symporter enables transformed S. pombe cells to take up a great number of monosaccharides. The affinities for the particular sugar substrates differ widely, however. Thus, the K m values for the uptake of D-glucose, D-mannose, D-fructose, D-xylose, and D-galactose turned out to be in the range of 1.5 ϫ 10 Ϫ5 , 1.5 ϫ 10 Ϫ4 , 3 ϫ 10 Ϫ4 , 1.5 ϫ 10 Ϫ3 , and 3 ϫ 10 Ϫ3 M, respectively (Table III). In previous publications (13,14), as well as in this paper, several HUP1 mutants with an increased K m value for D-glucose were described. It was of interest to find out whether these mutants show less efficient binding also of other substrates, i.e. whether sugar specific effects or more general ones are produced by the mutations. For this purpose the mutated HUP1 transporters listed in Table III were chosen for a detailed analysis of their substrate specificities and compared with that of the wild-type protein.
Three mutations, Q179E, I303N, and N436Y, equally diminish the affinities for all sugars tested (D-glucose, D-mannose, and D-fructose). In the case of Q298N and N436Q the K m values for D-glucose, D-xylose, and D-galactose transport are significantly increased while those for D-mannose and D-fructose uptake are only mildly affected. The conservative exchange D44E, however, influences the substrate specificity in an extraordinary way. As previously pointed out, this substitution increases the K m for D-glucose uptake by more than 10 times (13). Surprisingly, none of the K m values for other sugar substrates is altered in the D44E mutant significantly, i.e. by more than a factor of 2. These results suggest a special role of aspartate 44 in the glucose transport process. Therefore, this amino acid position has been investigated more thoroughly.
Characterization of Position Asp-44 -In the transporter model of Fig. 1 aspartate 44 is situated at the very beginning of the first extracellular loop of HUP1. This is based on the simple energetic reason that charged amino acids are expected to avoid the lipophilic environment of the membrane. In the lactose permease of E. coli, however, two aspartate residues are thought to exist in transmembrane helix VII. They both are neutralized by forming salt bridges with lysine residues in helix X and XI, respectively (24). It is a characteristic feature of these charge pairs that simultaneous replacement of both partners by neutral amino acids does not impair the translocation process, whereas the exchange of only one partner, leaving the other one unpaired, inactivates the permease completely. Since the mutation D44N leads to a total inactivation of the HUP1 transporter (13), it might be suspected that the aspartic acid residue likewise is involved in a salt bridge. The primary structure of the Chlorella transporter provides several basic amino acids that might act as the positively charged counterpart. Some of them were replaced individually or in combination by   (Table IV). The ability to take up D-glucose is maintained in the mutants R144L, R204L, and K59Q/K60M, albeit with dramatically reduced overall activities. These reductions, however, primarily reflect the low expression levels.
There is no indication that the catalytic activity per se is impaired. Moreover D-glucose transport tests of all three mutants yield K m values nearly identical to that of wild-type HUP1. These results make it unlikely that one of the basic amino acids at position 59, 60, 144, or 204 in the transport protein interacts with aspartate 44 via a salt bridge. Previous studies not only demonstrated the absolute necessity of a carboxyl group at position 44 but also the importance of its precise location for the transporter to be active. Thus, increasing the side chain length by changing aspartic into glutamic acid (D44E) decreases the V max of D-glucose uptake by 90% and raises the K m by 15-fold as compared with the wildtype, although similar amounts of transport protein are present in the cells (13). In order to elucidate whether the potential for protonation/deprotonation of Glu-44 as compared with that of Asp-44 is affected, the influence of extracellular pH on the translocation reaction catalyzed by the two symporters has  3.5 ϫ 10 Ϫ4 5.5 ϫ 10 Ϫ4 4 ϫ 10 Ϫ4 Ն5 ϫ 10 Ϫ2 Ն5 ϫ 10 Ϫ2 a This mutant is actually a double mutant, but the second exchange, F497S, is in all probability unimportant, since it is located in the C-terminal tail, that can be deleted without affecting substrate binding (14). b N.D., not determined. been examined. The D-glucose uptake activity of HUP1 is optimal at about pH 4.5 and declines steeply toward lower and higher values (Fig. 2). The D44E mutant shows a completely different pH dependence. As ambient pH is increased from 3.0 to 9.0, D-glucose uptake is accelerated gradually, reaching an optimum at about pH 7.0.
Determination of Substrate Specificities of HUP1 and HUP2-Despite the high homology (74%), the Chlorella symporters HUP1 and HUP2 differ significantly in substrate specificity (6,18). Whereas both carriers efficiently transport Dglucose with comparable affinities, there exists a large difference concerning D-galactose. This hexose is only poorly accepted by HUP1 and favored by HUP2. The K m values for D-galactose uptake differ from one another by 2 orders of magnitude (3 ϫ 10 Ϫ3 M versus 2.5 ϫ 10 Ϫ5 M). In a study using chimeric proteins it was recently shown that exchanging only the front part of the first extracellular loop of HUP1 for that of HUP2 (chimera C6), gives rise to a 15-fold higher affinity for D-galactose as compared with that of the wild-type HUP1 (18). A sequence alignment of the interchanged segments demonstrates that HUP1 and HUP2 differ in 16 out of 29 positions in this region (Fig. 3).
To localize the crucial amino acid(s) more precisely, the segment was bisected using an internal Asp700I site and two additional HUP1/2/1 chimeras, C7 and C8, were generated by exchanging the posterior and anterior halves, respectively (Fig.  3). As to the D-glucose uptake, C7 and C8 show K m values identical to that of C6 and very similar to that of HUP1 (Table  V), indicating that the homologous replacements do not interfere with D-glucose binding. On the other hand, remarkable differences are clearly visible regarding the transport of Dgalactose. The K m values of C7 and C8 coincide well with those of HUP1 and C6, respectively. Therefore, it is reasonable to conclude that the improved D-galactose binding of chimera C6 is mediated solely by one or at most a few residues in the anterior part of the exchanged loop fragment. Within this region only seven amino acids differ in the two transporters (Fig.  3).
If the critical determinant were a single amino acid, it should be identifiable by individual replacement of these seven residues in HUP1 with those of HUP2. Table V summarizes the data obtained with the first two HUP1 mutants, V52T and N45I. The exchange at position 52 does not influence the D-galactose affinity. In contrast, the K m value for D-galactose is positively affected in mutant N45I (Fig. 4) and corresponds to that of chimeras C6 and C8, clearly indicating that the enhanced D-galactose recognition of these chimeras can be explained by the substitution of a single amino acid, i.e. of asparagine 45 for isoleucine. It is suggested that this position is a major determinant for D-galactose binding in HUP2, a view strongly supported by the fact that the affinity of the HUP2 symporter to D-galactose is reduced 20-fold when the reverse mutation, I47N, is established (Table V). The three mutations, V52T and N45I in HUP1 and I47N in HUP2, affect the K m value for D-glucose only to a minor extent if at all (Table V).
Finally it should be mentioned that the V max values of the chimeras and the HUP1 and HUP2 mutants correlate without exception well with the levels of expressed transport protein (data not shown). Although a specific antibody for HUP2 is lacking, the amounts of wild-type HUP2 and I47N transporter could be estimated, because the polyclonal anti-HUP1-B antiserum (13) shows weak cross-reactivity. DISCUSSION The HUP proteins of Chlorella catalyze the uptake of several monosaccharides in co-transport with protons. How transport function is accomplished by the Chlorella symporters or, for that matter by transporters of any other organism, is an open and intriguing question. Regarding the nature of the sugarbinding site and the translocation pathway, it is assumed that residues directly interacting with the substrate should not be replaceable without causing a significant change in the affinity of the transporter for the substrate. Several HUP1 mutants with an increased K m value for D-glucose were found previously (13,14). Apart from Asp-44, which will be discussed later, the following amino acids were affected: Gln-179 (in helix V), Gln-298 and Gln-299 (both in helix VII), Val-433 and Asn-436 (both in helix XI).
Studies on the human glucose facilitator GLUT1 support the notion that at least some of these residues are intimately involved in substrate binding, e.g. glutamine 161, which is homologous to Gln-179 in HUP1 (25), and glutamine 282, which corresponds to Gln-298 of HUP1 (26). There is also good evidence, that helix XI of GLUT1 interacts directly with Dglucose (27).
Here, three additional randomly generated HUP1 mutants with decreased affinity to D-glucose have been reported (Table  I). The N436I mutation confirms previous findings concerning this special asparagine (14). The fact that the same amino acid has been found repeatedly in random mutagenesis studies indicates that the screen for K m mutants approaches saturation. The HUP1 symporter of mutant RMY126 revealed two amino acid exchanges, G97C (in helix II) and I303N (in helix VII), but FIG. 2. pH dependence of D-glucose uptake by S. pombe YGS-B25 expressing the Chlorella wild-type HUP1 or the D44E symporter. The initial rates of D-[U-14 C]glucose transport were measured for the wild-type (f) and the D44E mutant (q) as described under "Experimental Procedures." The uptake rate at pH 6.0 corresponds to 131 and 2.8 mol/h ϫ g fresh weight for the wild-type HUP1 and the D44E mutant, respectively. FIG. 3. Construction of chimeras C6 -C8. The chimeric transporter C6 consists mainly of HUP1 sequence. Only the front part of the first external loop, represented by the hatched box, is replaced by the corresponding HUP2 sequence. In the alignment shown above only those amino acids of HUP2 are listed that differ from the HUP1 sequence. The exchanged loop segment can be bisected by Asp700I restriction as indicated by the dotted line. Replacing the posterior and anterior section independently leads to chimeras C7 and C8, respectively.
only the latter proved to be responsible for the detected K m effect (Table II). A parallel situation exists in mutant RGY52, where only F292L (in helix VII) induces a K m shift identical to that in the double mutant. However, the efficiency of D-glucose binding is also influenced by the second exchange, G120D in helix III, but the insertion of a new negative charge in a non-polar transmembrane domain may more likely perturb the secondary structure of the protein.
Taken together, the new K m mutants strengthen the importance of residues within helices VII and XI, which have previously been suggested to play a role in D-glucose binding. Remarkably, residues Phe-292, Gln-298, Gln-299, and Ile-303 cluster on one face of putative helix VII when viewed on a helical wheel plot (Fig. 5). One might speculate therefore that Phe-292, Gln-299, and Ile-303, which are located in closest vertical proximity, are probably guiding the sugar substrate along the translocation path. Stacking of hydrophobic patches of the glucopyranose ring with aromatic residues is clearly visible in the binding pocket of the D-glucose-binding protein of E. coli (17). Considering phenylalanine 292, the same could hold for D-glucose binding of HUP1. By x-ray crystallography it was shown that sugar binding in periplasmic binding proteins is mediated mainly by charged amino acids and their amides via hydrogen bonds (16,17). It is striking that most of the HUP1 residues identified as good candidates for sugar interaction are amides (Gln-179, Gln-298, Gln-299, and Asn-436). In addition, there is a charged residue, Asp-44, in HUP1 that cannot be changed without affecting D-glucose binding. The topological model of HUP1 (Fig. 1) puts Asp-44 outside the hydrophobic membrane, but a location within helix I is imaginable as well. Supposing it were so, then helices I, V, VII, and XI would participate in sugar binding. Although it might be a coincidence that the first and fifth transmembrane spanning domains of the N-and C-terminal half of HUP1 would then have been identified, it is nevertheless remarkable, since the 12-helix arrangement of the major facilitator superfamily transporters most likely has arisen by a gene duplication event of an ancestral gene encoding a protein with six transmembrane spans (28). Helices with functional importance may very well be corresponding ones in each half of the protein.
When the proposed helix packing of HUP1 is compared with that of the intensively studied lactose permease of E. coli a number of parallels like the postulated neighborhood of helices V, VII, and XI can be seen (29). The importance of helices I and VII of lactose permease is, furthermore, underpinned by sugar specificity mutants (Refs. 30 and references therein). Only for helix X, potentially playing a major role in lactose transport (30,31), corresponding evidence for HUP transporters is missing.
Oppositely charged residues in transmembrane segments sometimes neutralize each other by forming a salt bridge (24). Provided that Asp-44 is located in helix I, it might also be paired with a basic amino acid. Candidates that could possibly act as positive counterions for Asp-44 were selected and replaced by neutral residues (Table IV). Evidence that Asp-44 is linked to a basic amino acid via a salt bridge has not been obtained, however. The shift in the pH optimum from pH 4.5 of the wild-type HUP1 protein to pH 7.0 of the D44E transporter indicates that the carboxyl group of the mutant is located in a drastically changed environment (increased hydrophobicity), leading to such a large increase in its pK a . The shift also suggests that a negative charge is required at this position for transport activity; in the case of the mutant one would have to assume that the proton dissociates only under the more alkaline condition. Of course, this is in accordance with the obser-  vation that transport in the mutant D44N is abolished (13).
Ten monosaccharide/H ϩ symporters with proven transport activity have been cloned from plants so far: HUP1Ϫ3 from C. kessleri (5,6), STP1Ϫ4 from Arabidopsis thaliana (32) 1 , MST1 from Nicotiana tabacum (33), HEX3 from Ricinus communis (34), and MTST1 from Medicago truncatula (35). A sequence comparison of these transporters reveals that HUP1 residues that most likely contribute to sugar binding (Asp-44, Gln-179, Gln-298, Gln-299, Ile-303, Val-433, Asn-436) are absolutely conserved, the only exception being Phe-292. On the one hand, this fact clearly emphasizes the great importance of these residues. On the other hand, additional positions must be postulated to explain the differing substrate specificities of the highly homologous transporters. The HUP2 symporter, for example, transports D-galactose more efficiently and with a 100 times higher affinity than the HUP1 symporter does. Various chimeras between these two transporters were previously constructed in order to identify crucial site(s) for the differential D-galactose recognition (18). A chimera consisting of the Nterminal half of HUP2 and the C-terminal half of HUP1 shows a K m value for D-galactose uptake indistiguishable from that of the high affinity transporter HUP2, indicating that the critical positions lie within the N-terminal half (18). In the chimera C6 only the first 29 amino acids of putative loop 1 were removed from HUP1 and replaced by the corresponding residues of HUP2 (Fig. 3). This chimera still has a 15-fold increased affinity for D-galactose, implying that there are at least two separate determinants in the N-terminal half of HUP2, one inside the interchanged loop segment and one outside.
By use of additional chimeras and site-directed mutants it has been proven that one single exchange within loop 1, N45I, causes the improved D-galactose affinity seen in C6 (Table V). A HUP2 mutant carrying the reverse substitution, I47N, reveals a 20-fold increased K m value for D-galactose uptake and therefore strongly supports the notion that the exchanged position is a crucial determinant.
Recently the construction of various chimeras between two closely related facilitators of S. cerevisiae have been reported, i.e. the Gal2 protein, which transports D-galactose and D-glucose, and the Hxt2 protein, which is specific for D-glucose (36,37). In contrast to the findings for HUP2, a 35-amino acid segment around helix X was identified as the D-galactose recognition domain of Gal2. Taking into account that the corresponding affinities of the Chlorella and S. cerevisiae transporters differ by orders of magnitude, this could mean that the mechanism of D-galactose binding varies among different members of the major facilitator superfamily. However, one should be aware of the fact that only those regions within the HUP2 and the Gal2 protein can be detected by the chimera method, which differ in sequence between the homologous partners. In other words, some positions that are essential but not sufficient for good D-galactose binding may be conserved in HUP1 and HUP2 as well as in Gal2 and Hxt2, respectively. If such conserved regions in the two Chlorella transporters in question differ from those conserved in the two S. cerevisiae transporters, the method of constructing chimeras may very well uncover different parts of a related binding pocket. Therefore, residues within the two identified regions of the proteins may cooperate in D-galactose binding and the pocket may in principal be similarly constructed in the HUP2 and the Gal2 protein.
Finally, it should be mentioned that an interesting model for ligand specificity of closely related opioid receptors was recently proposed (38). In this case extracellular loops are thought to act as selective barriers that control which ligands can enter the rather unspecific transmembrane binding pocket common to all receptor subtypes. If applied to HUP1, asparagine 45 (and also Asp-44) might actually be located in the first external loop, where it may restrict the passage of D-galactose, not, however, that of D-glucose.