Portal de Periódicos da CAPES

Residue 457 Controls Sugar Binding and Transport in the Na+/Glucose Cotransporter

2001; Elsevier BV; Volume: 276; Issue: 52 Linguagem: Inglês

10.1074/jbc.m108286200

1083-351X

Ana Dı́ez-Sampedro, Ernest M. Wright, Bruce A. Hirayama,

Ion Transport and Channel Regulation

The Na+/glucose cotransporter (SGLT1) is highly selective for its natural substrates,d-glucose and d-galactose. We have investigated the structural basis of this sugar selectivity on the human isoform of SGLT1, single site mutants of hSGLT1, and the pig SGLT3 isoform, expressed in Xenopus oocytes using electrophysiological methods and the effects of cysteine-specific reagents. Kinetics of transport of glucose analogues, each modified at one position of the pyranose ring, were determined for each transporter. Correlation of kinetics with amino acid sequences indicates that residue Gln-457 sequentially interacts with O1 of the pyranose in the binding site, and with O5 in the translocation pathway. Furthermore, correlation of the selectivity characteristics of the SGLT isoforms (SGLT1 transports both glucose and galactose, but SGLT2 and SGLT3 transport only glucose) with amino acid sequence differences, suggests that residue 460 (threonine in SGLT1, and serine in SGLT2 and SGLT3) are involved in hydrogen bonding to O4 of the pyranose. In addition, the results show that substrate specificity of binding is not correlated to substrate specificity of transport, suggesting there are at least two steps in the sugar translocation process. The Na+/glucose cotransporter (SGLT1) is highly selective for its natural substrates,d-glucose and d-galactose. We have investigated the structural basis of this sugar selectivity on the human isoform of SGLT1, single site mutants of hSGLT1, and the pig SGLT3 isoform, expressed in Xenopus oocytes using electrophysiological methods and the effects of cysteine-specific reagents. Kinetics of transport of glucose analogues, each modified at one position of the pyranose ring, were determined for each transporter. Correlation of kinetics with amino acid sequences indicates that residue Gln-457 sequentially interacts with O1 of the pyranose in the binding site, and with O5 in the translocation pathway. Furthermore, correlation of the selectivity characteristics of the SGLT isoforms (SGLT1 transports both glucose and galactose, but SGLT2 and SGLT3 transport only glucose) with amino acid sequence differences, suggests that residue 460 (threonine in SGLT1, and serine in SGLT2 and SGLT3) are involved in hydrogen bonding to O4 of the pyranose. In addition, the results show that substrate specificity of binding is not correlated to substrate specificity of transport, suggesting there are at least two steps in the sugar translocation process. Na+/glucose cotransporter 1-deoxy-d-glucopyranoside 2-deoxy-d-glucopyranoside 2-fluoro-2-deoxy-d-glucopyranoside 3-fluoro-3-deoxy-d-glucopyranoside 3-O-methyl-d-glucopyranoside 4-fluoro-4-deoxy-d-glucopyranoside 5-thio-d-glucopyranoside 6-deoxy-d-glucopyranoside 6-fluoro-6-deoxy-d-glucopyranoside α-methyl-d-glucopyranoside β-methyl-d-glucopyranoside glucose methyl methanethiosulfonate methanethiosulfonate 2-aminoethyl methanethiosulfonate 2-hydroxyethyl methanethiosulfonate Na+/glucose cotransporter, isoform 1, human Na+/glucose cotransporter, isoform 3, pig wild-type hSGLT1 3-(cyclohexylamino)-1-propanesulfonic acid facilitated glucose transporter The Na+/glucose cotransporters (SGLTs)1 transform the energy of the Na+ electrochemical gradient into mechanical work to drive sugar across the membrane, potentially against its concentration gradient (e.g. Refs. 1Parent L. Supplission S. Loo D.D.F. Wright E.M. J. Membr. Biol. 1992; 125: 63-79PubMed Google Scholar, 2Loo D.D.F. Hazama A. Supplission S. Turk E. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5767-5771Crossref PubMed Scopus (206) Google Scholar, 3Loo D.D.F. Hirayama B.A. Gallardo E.M. Lam J.T. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 7789-7794Crossref PubMed Scopus (142) Google Scholar). Previously we have used the interaction of large aromatic glycosides to probe the vestibule to the sugar-binding site and the translocation pathway. These studies determined essential aglycone-protein interactions of human SGLT1 (hSGLT1) and pig SGLT3 (pSGLT3) and provided a minimum size for the vestibule and sugar transport "pore" (e.g. Refs. 4Lostao M.P. Hirayama B.A. Wright E.M. J. Membr. Biol. 1994; 142: 161-170Crossref PubMed Scopus (92) Google Scholar, 5Dı́ez-Sampedro A. Lostao M.P. Wright E.M. Hirayama B.A. J. Membr. Biol. 2000; 176: 111-117Crossref PubMed Scopus (62) Google Scholar, 6Hirayama B.A. Dı́ez-Sampedro A. Wright E.M. Br. J. Pharmacol. 2001; 134: 484-495Crossref PubMed Scopus (52) Google Scholar). In this study we focus our attention on the importance of sugar-protein hydrogen bonding in substrate specificity of transport. Determination of interactions between the sugar and the protein is an important step in elucidating the structural factors that determine specificity in the sugar-binding site and will allow future studies on characterization of the translocation pathway. Our strategy was to: 1) determine the importance of the interactions at each position of the sugar for transport through hSGLT1; 2) test the proposed interactions in cotransporters mutated in a single residue identified through sequence analysis; and 3) compare the interactions of these sugars in an SGLT isoform with known differences in sugar recognition (pSGLT3). In this work we have taken advantage of advances in gene technology and application of biophysical methods to cotransport to extend classic studies of the selectivity of sugar cotransport to the molecular level. We studied sugar specificity in four clones: hSGLT1, two hSGLT1 mutants, Q457C and Q457E, and pSGLT3. Interactions of sugars with the SGLTs were studied by analyzing the transport of sugars that differ in only one position of the ring. The importance of each position in the glucose ring was evaluated by determining the apparent affinity (K0.5) for each compound and comparing it with the K0.5 for glucose. The results indicate that residue 457 is important in both sugar recognition and translocation through hydrogen bond interactions with O1 and O5 of the pyranose ring. A plasmid containing human SGLT1 (hSGLT1 (7Hediger M.A. Turk E. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 5748-5752Crossref PubMed Scopus (223) Google Scholar)) cDNA was used as a template for site-directed mutagenesis. Glutamine at the 457 position was replaced by glutamic acid (Q457E) using a two-step polymerase chain reaction protocol with the oligonucleotide primers: Q457E-sense, 5′-TTCGATTACATCGAATCGATTACCAGT-3′; and Q457E-antisense, 5′-ACTGGTAATCGATTCGATGTAATC-3′. The underlined letters represent the mutation at amino acid position 457 to a glutamic acid and the ClaI site that was silently introduced as an aid in screening. PCR products were purified and combined in a final PCR reaction using the normal primers flanking the mutation site to produce an insert with AvrII andEco47III restriction sites. The resulting 500-bp fragment was ligated into a similarly treated wild-type-containing plasmid. One transformed colony was screened by sequencing to verify the presence of only the desired mutation. The mutagenized DNA template was linearized with XbaI, and mRNA was transcribed and capped in vitro using the T3 RNA promoter (MEGAscript kit, Ambion, Austin, TX). Mature Xenopus laevis oocytes were injected with 50 ng of mRNA coding for hSGLT1 (7Hediger M.A. Turk E. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 5748-5752Crossref PubMed Scopus (223) Google Scholar), hSGLT1 Q457C (3Loo D.D.F. Hirayama B.A. Gallardo E.M. Lam J.T. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 7789-7794Crossref PubMed Scopus (142) Google Scholar), hSGLT1 Q457E, or pSGLT3 (pig renal Na+/glucose cotransporter, pSAAT1 (8Kong C.T. Yet S.F. Lever J.E. J. Biol. Chem. 1993; 268: 1509-1512Abstract Full Text PDF PubMed Google Scholar, 9Mackenzie B. Panayotova-Heiermann M. Loo D.D.F. Lever J.E. Wright E.M. J. Biol. Chem. 1994; 269: 22488-22491Abstract Full Text PDF PubMed Google Scholar)). Oocytes were maintained in Barth's medium supplemented with gentamicin (5 mg/ml)/penicillin (100 units/ml)/streptomycin (100 μg/ml) at 18 °C from 3–6 days until used. Experiments were performed at 22 °C using the two-electrode voltage-clamp method in a rapid perfusion chamber (4Lostao M.P. Hirayama B.A. Wright E.M. J. Membr. Biol. 1994; 142: 161-170Crossref PubMed Scopus (92) Google Scholar). For the experiments, the oocytes were bathed in Na+ buffer composed of (in millimolar) 100 NaCl, 2 KCl, 1 MgCl2, 1 CaCl2, 10 HEPES/Tris, pH 7.4, and in the Na+-free buffer, choline-Cl replaced NaCl. The membrane potential was held at −50 mV and stepped for 100 ms from +50 mV to −150 mV, in 20 mV decrements. The sugar-dependent current was the difference between the current recorded in the sugar and the previous record in Na+ buffer alone. All sugar solutions were freshly prepared. The experiments were controlled and data were acquired using pClamp software (Axon Instruments, Foster City, CA). The steady-state sugar-dependent currents (I) were obtained at each voltage and sugar concentration as the difference between the current measured at steady state in the presence and absence of the sugar. The apparent affinity constant (K0.5) and the maximal current for saturating sugar (Imax) were obtained by fitting the steady-state currents at each membrane potential to the equation,I = Imax × [S]/(K0.5 + [S]), using the non-linear fitting method in SigmaPlot (SPSS, Chicago, IL) where [S] is the sugar concentration. Working concentrations of MTSHE (1 mm) and MeMTS (0.5 mm) in Na+buffer were prepared from 200 mm stock solutions (in anhydrous dimethyl sulfoxide, kept frozen at −20 °C). MTSEA was dissolved immediately before use. The transport level at saturating sugar concentrations was measured before labeling (100 mmαMDG). Labeling with MTSEA and MTSHE was performed with the oocyte membrane clamped at −50 mV, and then the reagent was added to the bath solution for 1 min. MeMTS was applied for 15 min. Iodoacetamide was dissolved at a concentration of 2 mm in Na+buffer at pH 9 (buffered with CAPS). The oocyte, clamped at −50 mV, was bathed with this solution for 20 min. After exposure to the test reagent, the remaining unlabeled transporters were inactivated by exposure to MTSEA for 1 min (3Loo D.D.F. Hirayama B.A. Gallardo E.M. Lam J.T. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 7789-7794Crossref PubMed Scopus (142) Google Scholar). d-Glucose, α-methyl-d-glucose, β-methyl-d-glucose, 2-deoxy-d-glucose, 3-O-methyl-d-glucose, 3-fluoro-3-deoxy-d-glucose, d-galactose, 6-deoxy-d-glucose, 6-fluoro-6-deoxy-d-glucose, and iodoacetamide were purchased from Sigma Chemical Co. (St. Louis, MO); 5-thio-d-glucose was from Aldrich (Milwaukee, WI); and 1-deoxy-d-glucose (1,5-anhydro-d-glucitol), 4-fluoro-4-deoxy-d-glucose, methyl methanethiosulfonate (MeMTS), 2-hydroxyethyl methanethiosulfonate (MTSHE), and 2-aminoethyl methanethiosulfonate hydrobromide (MTSEA) from Toronto Research Biochemicals (Downsview, Ontario, Canada). All other reagents were purchased from Sigma or Research Organics (Cleveland, OH), unless otherwise specified. Sugar transport was measured by using an inherent property of Na+/glucose cotransporters: Transport of the substrate requires cotransport of Na+, which generates an electrical current (e.g. Refs. 10Mackenzie B. Loo D.D.F. Wright E.M. J. Membr. Biol. 1998; 162: 101-106Crossref PubMed Scopus (106) Google Scholar, 11Dı́ez-Sampedro A. Eskandari S. Wright E.M. Hirayama B.A. Am. J. Physiol. Renal Physiol. 2001; 49: F278-F282Crossref Google Scholar). Fig.1 shows an example of an experiment using an oocyte expressing Q457C hSGLT1 with its membrane potential clamped at −50 mV. Initially the oocyte was bathed in Na+ buffer and, when 100 mm αMDG (a saturating concentration) was added, an inward current of −230 nA was generated, representing the maximal substrate-dependent current of the oocyte. The bath solution was changed to choline chloride, to wash the substrates from the chamber, and the current was eliminated. After re-equilibration of the oocyte with Na+ buffer the current returned to the baseline and when 20 mm galactose was added a small current (−15 nA) was generated. After washing and re-equilibration in Na+ buffer to return to the baseline, 20 mm4F4DOglc was added to the bath and a −150 nA current was recorded, suggesting that Q457C has a higher affinity for 4F4DOglc than galactose. From these currents, using different sugar concentrations, the apparent sugar affinities (K0.5) and their maximal transport (Imax) were obtained.K0.5 for 4F4DOglc and galactose were calculated to be 0.33 and 52 mm, respectively (TableI), and the maximal transport for both sugars were around 75% of the αMDG Imax.Table ISummary of apparent affinities for each sugar in hSGLT1, the two mutants: Q457C, Q457E, and pSGLT3SugarhSGLT1Q457CQ457EpSGLT3K0.5RatioglcK0.5RatioglcK0.5RatioglcK0.5RatioglcmmmmmmmmGlucose0.5 ± 0.02111 ± 1.214.0 ± 0.4110 ± 0.41αMDG0.7 ± 0.041.44.6 ± 0.60.42.8 ± 0.60.75 ± 0.40.5βMDG0.5 ± 0.15132 ± 32.914 ± 2.03.57 ± 2.00.71DOglc10 ± 120∞∞55 ± 1.71426 ± 102.62DOglc>100>200∞∞∞∞∞∞3OMglc6 ± 112134 ± 161268 ± 1917∞∞Galactose0.6 ± 0.021.252 ± 74.721 ± 55.2>200>204F4DOglc0.07 ± 0.010.140.33 ± 0.050.030.3 ± 0.050.0756 ± 45.65Thioglc4 ± 186 ± 1.50.5Ki = 40 ± 2410Ki = 17 ± 101.76DOglc3 ± 0.564 ± 0.60.366 ± 0.51.54.0 ± 0.40.4All values are for measurements at −150 mV. TheK0.5 and Ki values were obtained according to the "Experimental Procedures." Steady-state currents for each sugar, obtained at each sugar concentration, were calculated by fitting to equation, I = Imax × [S]/(K0.5 + [S]). Ratioglc is the proportional affinity for a sugar compared to the K0.5 value for glucose ( K 0.5sugar/K 0.5glc). All kinetic values were confirmed in at least two different oocytes. Errors are reported as standard error of the mean for triplicate or quadruplicate determinations. For duplicate determinations the error reported is the propagated error of the two fits. In hSGLT1 theImax values of all the sugars, as the percentage of the Imax for αMDG, were 83 ± 5% (S.E.,n = 16). In pSGLT3 galactose Imaxcould not be determined due to its high K0.5 (>200 mm); 2DOglc, 30Mglc, and 5thioglc were not transported, and the remaining sugars were transported at 73 ± 13% (n = 11) of the αMDG Imax. In Q457C, 1DOglc and 2DOglc did not bind, but the remaining sugars were transported as well as αMDG (90 ± 4%, n = 19). Also in Q457E the sugars were transported at 88 ± 5% of the αMDG Imax (n = 16), excluding 2DOglc (non-binding) and 5thioglc, which was not transported. Maximal αMDG currents for Q457C and Q457E were around 200 nA. Open table in a new tab All values are for measurements at −150 mV. TheK0.5 and Ki values were obtained according to the "Experimental Procedures." Steady-state currents for each sugar, obtained at each sugar concentration, were calculated by fitting to equation, I = Imax × [S]/(K0.5 + [S]). Ratioglc is the proportional affinity for a sugar compared to the K0.5 value for glucose ( K 0.5sugar/K 0.5glc). All kinetic values were confirmed in at least two different oocytes. Errors are reported as standard error of the mean for triplicate or quadruplicate determinations. For duplicate determinations the error reported is the propagated error of the two fits. In hSGLT1 theImax values of all the sugars, as the percentage of the Imax for αMDG, were 83 ± 5% (S.E.,n = 16). In pSGLT3 galactose Imaxcould not be determined due to its high K0.5 (>200 mm); 2DOglc, 30Mglc, and 5thioglc were not transported, and the remaining sugars were transported at 73 ± 13% (n = 11) of the αMDG Imax. In Q457C, 1DOglc and 2DOglc did not bind, but the remaining sugars were transported as well as αMDG (90 ± 4%, n = 19). Also in Q457E the sugars were transported at 88 ± 5% of the αMDG Imax (n = 16), excluding 2DOglc (non-binding) and 5thioglc, which was not transported. Maximal αMDG currents for Q457C and Q457E were around 200 nA. Similar experiments were performed with each of the four clones with different sugars. As an example, in Fig.2 we show the kinetics obtained for αMDG and 3F3DOglc in hSGLT1 from the same oocyte. TheK0.5 values for the sugars were very different (0.5 ± 0.01 and 9 ± 1.0 mm), but the estimated maximal currents (Imax) were similar: 863 ± 5 and 777 ± 25 nA. We note that the functional expression of hSGLT1 mutants was lower than the wild-type transporter, as reflected by the lower Imax. The sugar specificity of each clone was determined by analyzing apparent kinetic constants of each sugar, and the results are summarized in Table I. All kinetics were determined on at least two different oocytes, and the "reference" Imaxwas estimated for each oocyte using a saturating concentration of αMDG. In general, the Imax for all transported sugars were similar to that for glucose and αMDG: theImax values of the sugars are within 75% of theImax of αMDG except as noted (Table I). Ten hexoses with differences at each position of the ring were tested in hSGLT1-expressing oocytes.Imax values, measured in the same oocyte, were similar for all the transported sugars, but the affinities were very different (Table I). The d-glucoseK0.5 for this transporter was 0.5 mm, but 1DOglc had a higher K0.5 (10 mm), indicating the importance of this –OH. TheK0.5 for α-methyl-d-glucopyranoside (αMDG) which has a methyl group in the first position in the equatorial orientation, was 0.7 mm; thus, the presence of this group had no effect on the affinity. We also tested βMDG, to see if the orientation of this methyl group modified the kinetics, but the K0.5was similar to αMDG (0.5 mm). The most dramatic change in the affinity occurred when the second position of the sugar lost its equatorial –OH group, theK0.5 for 2DOglc was 112 mm. The –OH group here is critical for recognition, because replacement of the –OH with –F (2F2DOglc) or –NH2 (glucosamine) resulted in extremely poor substrates, both having K0.5values of ≫100 mm (not shown). Despite its poor affinity, the Imax for 2DOglc was estimated to be similar to the Imax for glucose. 3OMglc had a K0.5 of 5.5 mm, and hSGLT1 will accept an –F at position 3 (3F3DOglc) with aK0.5 of ∼9 mm (Fig. 2). The only increase in the apparent affinity, compared with glucose, was found when the equatorial –OH in the fourth position was replaced with –F (4F4DOglc) where the K0.5 = 0.07 mm. Changing the orientation in the fourth –OH to the axial position (galactose) had no effect: the K0.5 was identical to d-glucose (0.55 mm). Changing the pyranose O to S in the ring of the sugar, 5-thio-d-glucose, increased theK0.5 to 4.4 mm. The apparent affinity for 6DOglc was 2.7 mm, similar to that for 6F6DOglc (1.3 mm, not shown). The pentose,d-xylose, had a K0.5 of about 100 mm and myo-inositol, a cyclic polyhydroxyalcohol, had a K0.5 of ∼200 mm (not shown). The same sugars were tested in oocytes expressing pSGLT3, and differences in affinity and in selectivity were found (Table I). d-Glucose had a K0.5 of 9.7 mm, similar to βMDG (7.1 mm), but higher than αMDG (4.8 mm). 1DOglc had aK0.5 of 26 mm. Neither 2DOglc nor 3OMglc induced currents, even at concentrations up to 200 mm, indicating that they were not transported. Furthermore, neither 200 mm 2DOglc nor 3OMglc inhibited the current generated by 4 mm αMDG, suggesting that they are also not inhibitors. Galactose had a very high K0.5(>200 mm), and replacing the –OH in the fourth position of glucose for –F (4F4DOglc) improved the K0.5, but it was still higher than the K0.5 for glucose (55 versus 10 mm). 100 mm5thioglc did not induce any current, but it did inhibit 4 mm αMDG-induced current with a Ki of 17 mm. The K0.5 for 6DOglc was 4 mm. The K0.5 for glucose was 11 mm (∼20 times higher than in hSGLT1), and for αMDG it was 4.6 mm (6 times). Surprisingly, the βMDG affinity in Q457C was 32 mm, about 70 times higher than in the WT. Neither 200 mm 1DOglc nor 2DOglc generated any current or inhibited 4 mm αMDG-induced current, i.e. they were non-interacting sugars. For 3OMglc and galactose,K0.5 values were significantly higher for Q457C than in hSGLT1: >100 versus 5.5 mm and 52versus 0.55 mm, respectively. TheK0.5 for galactose was five times higher than that for Glc (52 versus 11 mm). However, the 4F4DOglc K0.5 for this mutant was 0.33 mm, 30-fold lower that for glucose. The sugars 5thioglc and 6DOglc had K0.5 values for Q457C of 5.5 and 4.5 mm, respectively. We have previously shown that, after alkylating Cys-457 with MTSEA, Q457C is still able to bind sugars, but it loses its ability to translocate the sugar across the membrane (3Loo D.D.F. Hirayama B.A. Gallardo E.M. Lam J.T. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 7789-7794Crossref PubMed Scopus (142) Google Scholar). Fig. 3A shows the current induced by glucose in a Q457C oocyte clamped at −50 mV. The currents were −30 nA at 10 mm and −160 nA at 150 mm concentration. After washing and re-equilibrating the oocyte in Na+ buffer, 1 mm MTSEA was added for 1 min. After treatment, transport was eliminated. Treatment with MTSHE produced the same result, as do a variety of alkylating reagents (e.g. N-ethylmaleimide, [2-(trimethylammonium)ethyl] methanethiosulfonate and tetramethylrhodamine maleimide (3Loo D.D.F. Hirayama B.A. Gallardo E.M. Lam J.T. Wright E.M. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 7789-7794Crossref PubMed Scopus (142) Google Scholar)). The reason for inhibition of the translocation part of the transport cycle is unclear; however, although many of the reagents had hydrogen bond-donating or -accepting groups, none contained both functionalities. Our hypothesis was that by replacing the side chain of cysteine with the functional groups of the native glutamine, using iodoacetamide, we could restore WT function. The glucose K0.5 in Q457C was 20 times higher than in WT protein (0.5 versus 11 mm). In Fig. 3B the glucose-induced currents are shown in a different oocyte expressing Q457C. Treatment of this oocyte first with 2 mm iodoacetamide for 20 min and then 1 mm MTSEA for 1 min did not abolish the glucose current (the MTSEA treatment inactivated all the Q457C transporters that had not reacted with iodoacetamide). The second part of the panel shows that the glucose-induced current after the treatment was 70% compared with the initial current indicating that iodoacetamide reacted with 70% of the transporters. A kinetic analysis (at −50 mV) on a single oocyte, before and after labeling with iodoacetamide, showed that the K0.5 of glucose before treatment was 17 ± 2 mm, and after treatment the K0.5 decreased to 8 ± 0.5 mm, with no significant change inImax (not shown). Control experiments with WT protein gave identical currents before and after treatment with 2 mm iodoacetamide and MTSEA. Furthermore, iodoacetamide altered sugar selectivity of Q457C at the 5 position. 5Thioglc was a transported substrate, but after derivatization by iodoacetamide, 5thioglc became an inhibitor. Similarly, derivatizing Q457C with MeMTS also did not abolish transport. In contrast to the effect of iodoacetamide, MeMTS appears to manifest its effect as a decrease in turnover without affecting theK0.5. For example, in two oocytes from the same batch, before MeMTS treatment, the K0.5 for αMDG (at −50 mV) was 13 ± 1 mm, and after derivatizing, the K0.5 was 13 ± 2 mm, but the Imax decreased by 75% after treatment (not shown). The d-glucoseK0.5 (4 mm) was similar to theK0.5 for αMDG (2.8 mm) showing that addition of a methyl group in the alpha position did not change the affinity, but in the beta position K0.5increased to 14 mm. 1DOglc had a very poor affinity (K0.5 = 55 mm). This mutant did not transport or bind 200 mm 2DOglc. Other sugars with poor affinities were galactose (K0.5 = 21 mm) and 3OMglc (K0.5 = 68 mm). On the other hand, 4F4DOglc had the lowestK0.5 found in this mutant (0.3 mm). The only sugar that had similar affinity to the WT was 6DOglc (6versus 3 mm in WT). 5-Thioglucose was not transported by Q457E, but it was an inhibitor of the 4 mmαMDG induced-current (Ki = 40 mm). In this study we have examined the hydrogen bonding interactions involved in substrate specificity in hSGLT1 expressed in X. laevis oocytes. SGLT1 strongly discriminates between its natural substrates, d-glucose and d-galactose, and other hexoses. Our experiments show that by varying the amino acid side chain at position 457 (by site-directed and chemical mutagenesis) we can modify sugar affinity and selectivity. Finally, the differences in sugar affinity and selectivity between hSGLT1 and pSGLT3 emphasize the importance of specific amino acid sequences in determining sugar transport. As shown for sugar binding proteins (e.g. Refs. 12Quiocho F.A. Pure Appl. Chem. 1989; 61: 1293-1306Crossref Scopus (343) Google Scholar, 13Rao V.S.R. Qarba P.K. Balaji P.V. Chandrasekaran R. Conformations of Carbohydrates. Harwood Academic Publishers, Amsterdam1998: 303-353Google Scholar), the affinity and selectivity of sugar transporters should be largely controlled by the position, number, and type of hydrogen bonds that the protein can make with the sugar. In sugar-binding proteins, the –OH groups and the ring O are involved in extensive hydrogen bonding, and these bonds are of three types: cooperative, where an –OH group simultaneously acts as a hydrogen bond donor and acceptor; bidendate, where adjacent –OH groups hydrogen bond with different atoms of the same polar side chain; and network, where the cooperative and bidendate hydrogen bonds lead to the formation of a network of bonds between the sugar and the protein. Sugar-protein hydrogen bonds are almost equally distributed between the neutral-neutral and neutral-charged types, and the residues making these bonds are, more often than not, those containing planar polar side chains with two functional groups (Asn, Asp, Glu, Gln, Arg, and His). Amino groups, and to a lesser extent –OH groups, are the hydrogen bond donors, and acidic residues (Asp and Glu) are often important when there is little discrimination between anomers or epimers, e.g.d-glucose andd-galactose. In addition to hydrogen bonds, hydrophobic bonding may contribute to sugar-protein interactions, i.e.aromatic residues (Phe, Trp, and Tyr) promote van der Waals interaction with hydrophobic patches on one or both sides of the pyranose ring. Finally, the strength of hydrogen bonding between the sugar and the protein depends on the hydration of the active site; the lower the degree of hydration, the lower the dielectric constant, and the greater the strength of the hydrogen bond. Glucose analogues (each with one modification to the pyranose ring) were used as probes of the structure of the binding site of the human isoform of the Na+/glucose cotransporter (hSGLT1) to identify important interactions and residues that determine transport specificity. The relative importance of these interactions was mapped by comparing the kinetics of the series of glucose analogues on hSGLT1, single-site mutants Q457C and Q457E, and pig SGLT3 (pSGLT3). Each modification of the sugar changed the capacity of the sugar to make hydrogen bonds with the protein, so, if the part of the sugar that has been modified is important in substrate recognition and/or translocation, we would expect to see changes in transport kinetics. The hydroxyls were replaced with either: –F, which can accept hydrogen bonds; –SH, a donor but very rarely an acceptor; –NH2, which only donates; –S–, which can only accept, or; –CH3, which does not donate or accept hydrogen bonds (14Jeffrey G.A. An Introduction to Hydrogen Bonding. Oxford, New York, NY1997: 15Google Scholar). The majority of the changes in potential hydrogen bonding interactions at every position in the sugar decreased affinity with respect to glucose. This indicates that the protein-sugar interactions for the transporter are specific and involve each position of the pyranose ring. These results, combined with the effect of alkylation of Q457C hSGLT1, indicate that residue 457 strongly interacts with O1 of the pyranose in the sugar recognition process. The analysis also implicates O5, in the ring, as interacting with the 457 residue in the translocation pathway. An interpretation of the results (Table I) is as follows. The transport protein must donate a hydrogen bond to O1 of the sugar for optimal affinity. 1DOglc cannot accept a hydrogen bond here, and its K0.5 was more than 20 times higher than glucose. If a methyl group was added to an equatorial or axial oxygen at the first position of the glucose (αMDG or βMDG) the affinity was similar to d-glucose, which indicates that the sugar does not need to donate a hydrogen bond from this position, suggesting the presence of a protein–OH or –NH2 group. This also indicates that, although the protein must donate to the sugar, the hydrogen bond-donating group is either positioned to interact with both anomers or is not highly constrained. The α or β methyl group may also make hydrophobic interactions that compensate for the lost of hydrogen bond donation at position 1. The additional mass of the methyl group is easily accommodated, because our previous studies have shown that large structures can be added to the pyranose ring, especially in the β-configuration (4Lostao M.P. Hirayama B.A. Wright E.M. J. Membr. Biol. 1994; 142: 161-170Crossref PubMed Scopus (92) Google Scholar, 5Dı́ez-Sampedro A. Lostao M.P. Wright E.M. Hirayama B.A. J. Membr. Biol. 2000; 176: 111-117Crossref PubMed Scopus (62) Google Scholar). The second position in the sugar has a very specific hydrogen bonding requirement that is not compatible with anything other than –OH in the equatorial position. TheK0.5 changed from 0.5 mm for glucose to >100 mm for 2DOglc. This sugar has lost the ability to donate or accept hydrogen bonds at position 2. Both 2F2DOglc (hydrogen bond acceptor only) and glucosami