Neutralization of a conserved amino acid residue in the human Na+/glucose transporter (hSGLT1) generates a glucose-gated H+ channel.

The role of conserved Asp204 in the human high affinity Na+/glucose cotransporter (hSGLT1) was investigated by site-directed mutagenesis combined with functional assays exploiting the Xenopus oocyte expression system. Substitution of H+ for Na+ reduces the apparent affinity of hSGLT1 for glucose from 0.3 to 6 mm. The apparent affinity for H+ (7 microm) is about three orders of magnitude higher than for Na+ (6 mm). Cation/glucose cotransport exhibits a coupling ratio of 2 Na+ (or 2 H+):1. Pre-steady-state kinetics indicate that similar Na+ - or H+ -induced conformational changes are the basis for coupled transport. Replacing Asp204 with Glu increases the apparent affinity for H+ by >20-fold with little impact on the apparent Na+ affinity. This implies that the length of the carboxylate side chain is critical for cation selectivity. Neutralization of Asp204 (Asp --> Asn or Cys) reveals glucose-evoked H(+) currents that were one order of magnitude greater than Na(+) currents. These phlorizin-sensitive H+ currents reverse and are enhanced by internal acidification of oocytes. Together with a H(+) to sugar stoichiometry as high as 145:1, these results favor a glucose-gated H+ channel activity of the mutant. Our observations support the idea that cotransporters and channels share common features.

The human high affinity Na ϩ /glucose cotransporter (hSGLT1) 1 is a member of a family of secondary transporter proteins encompassing more than 55 homologues from archaea, bacteria, yeast, insects, and mammals (1,2). This family uses electrochemical Na ϩ gradients to drive the coupled uphill transport of a variety of substrates (sugars, amino acids, vitamins, osmolytes, ions, myo-inositol, urea, and water). The expression of hSGLT1 in Xenopus laevis oocytes has resulted in a comprehensive study of both steady-state and pre-steady-state kinetics (3)(4)(5). A six-state ordered binding model has been proposed in which transport results from ligand-induced conformational changes (6,7). In this model Na ϩ binds before sugar, with a coupling ratio of 2 Na ϩ :1 glucose, and voltage influences both Na ϩ binding and the conformational states of the unloaded transporter. Functional analysis of SGLT chimeras and truncated proteins strongly suggests that the sugar pathway is located in the C-terminal domain of the protein (8,9). Site-directed thiol labeling of a residue in the proposed sugar pathway indicates that conformational changes are responsible for the coupling of Na ϩ and sugar transport (10). We suggest that these conformational alterations are induced by cation binding in the N-terminal domain of the protein.
Although the functional importance of the N terminus in cation binding/translocation was shown for another SGLT family member (the Na ϩ /proline transporter (PutP) of Escherichia coli) (11)(12)(13), there is little information on the role of the Nterminal domain in hSGLT1. We have initiated a study to explore the role of N-terminal residues in hSGLT1 in cotransport. In the present study, we have targeted a conserved residue, Asp 204 , located in a short cytoplasmic loop of hSGLT1 connecting transmembrane domains V and VI, which has been implicated in cation selectivity in PutP (12). Replacing Asp 204 in hSGLT1 with Asn, Cys, or Glu dramatically modulated the steady-state and pre-steady-state kinetics of the transporter. Remarkably, although a transporter with a negative amino acid (Asp or Glu) exhibited cation/glucose cotransport with a stoichiometry of 2 (Na ϩ or H ϩ ) to 1, neutralization of Asp 204 (by Asn or Cys) resulted in the activation of a glucose-activated H ϩ channel.

EXPERIMENTAL PROCEDURES
Molecular Biology-A plasmid containing human SGLT1 (hSGLT1) cDNA was used as template for site-directed mutagenesis. Replacement of Asp 204 with Asn, Cys, and Glu was performed using a two-step polymerase chain reaction protocol (14). For each pair of mutagenic oligonucleotides the sequence of the sense primer is presented with the altered nucleotide(s) underlined: D204C, 5Ј-GATTTACACGTGCACCT-TGC-3Ј; D204E, 5Ј-GATTTACACGGAAACCTTGC-3Ј; D204N, 5Ј-GATTTACACGAACACCTTGC-3Ј. Polymerase chain reaction products were digested with BglII and Eco47III, and the resulting 428-bp fragments were ligated into a similarly treated wild-type hSGLT1-containing plasmid. The fidelity of the inserted DNA fragments was confirmed by sequencing double-stranded DNA (Sequenase version 2.0, DNA sequencing kit, United States Biochemical, Cleveland, OH) after alkaline denaturation (15). Each mutagenized DNA template was linearized with XbaI, transcribed, and capped in vitro using the T3 RNA promoter (MEGAscript kit, Ambion, Austin, TX). X. laevis oocytes were injected with 50 ng of mRNA and were incubated in Barth medium containing gentamicin (5 mg/ml) at 18°C for 3-7 days (4).
Transport Assays and Electrophysiological Techniques-For transport and electrophysiological experiments, oocytes were bathed in an assay buffer composed of 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES-Tris, pH 7.5, and a combination of Na ϩ or choline ϩ chloride salts to give a final concentration of 100 mM. For H ϩ activation experiments the pH of 100 mM choline buffer was varied between 8.0 and 4.5 by titration with Tris or Mes. Uptake of methyl-␣-D-[U-14 C]glucopyranoside (293 Ci/mol, Amersham Pharmacia Biotech) and electrophysiological measurements using the two-microelectrode voltage clamp technique were performed as described (16,17). The stoichiometry of cation-coupled D-[U-14 C]glucose (316 Ci/mol, ICN Radiochemicals) uptake was determined under voltage clamp conditions (18).
Data Analysis-Sugar-evoked steady-state currents were fitted to Eq. 1, where I Glc and I max Glc represent D-glucose-induced current and maximal D-glucose-induced current, respectively, at saturating [cation], [Glc] is the concentration of D-glucose, and K 0.5 Glc is [Glc] at 0.5 I max Glc . Kinetic parameters for the phlorizin-inhibited cation leak or cation-activated glucose transport were determined by using Eq. 2, where I cation and I max cation is the cation-evoked (leak) current and the maximal cation (leak) current at saturating cation concentration [C], respectively, K 0.5 cation is [C] at 0.5 I max cation , and n represents the Hill coefficient. Charge-voltage (Q-V) relations for each membrane voltage (V m ) were obtained by integrating pre-steady-state current transitions (after subtracting the capacitive and the steady-state currents from the total currents) with time and were fitted to Eq. 3 (17,19), with Q hyp and Q dep for Q at hyperpolarizing and depolarizing limits, respectively. V 0.5 represents V m at which 50% of the total charge in the membrane electric field has moved, z is the apparent valence of the moveable charge, and F, R, and T have their usual meanings. All experiments were repeated at least three times with oocytes from different donor frogs. Data fits were performed using the non-linear regression algorithm in SigmaPlot (version 5.0, SPSS Inc., Chicago, IL). Unless otherwise noted, figures are based on data obtained from a typical experiment on a single oocyte, and errors represent S.E. of the fit.

RESULTS
In these studies all comparisons between Na ϩ and H ϩ kinetics were performed in the same oocyte expressing WT, D204C, D204E, or D204N cRNA. For comparison of WT kinetics and the kinetics of a transporter with a substitution at position 204, oocytes of the same batch were analyzed on the same day within 10 h.
Uptake Experiments-The most conservative substitution Asp 204 3 Glu resulted in a reduction of Na ϩ -dependent ␣MDG uptake by 87%, whereas the uptake rates of D204C and D204N were reduced by ϳ50% (Fig. 1). In the absence of Na ϩ ions, the rate of ␣MDG uptake in oocytes expressing D204E was about 1 pmol/h, a value comparable to oocytes with the native transporter (1.7 pmol ϫ h Ϫ1 ϫ oocyte Ϫ1 ). However, replacing Asp 204 with Cys or Asn increased the uptake rate by 7-and 16-fold. This increase may reflect the enhanced apparent affinity for sugar (see Table I).
Steady-State Currents- Fig. 2 shows representative sugarinduced steady-state current-voltage (I-V) relationships for WT, D204C, and D204E in the presence of Na c and H c at saturating [Glc]. For WT both curves were sigmoidal and saturated at hyperpolarizing voltages. A similar I-V relation was observed for D204E in Na c , but in H c the sugar-induced currents exhibited a supralinear increase with hyperpolarization. D204C and D204N (data not shown) exhibited essentially identical I-V relations with no saturation of the sugar-evoked currents at the most negative potential (Ϫ150 mV) in either Na c or in H c . The glucose-evoked currents in H c for these transporters were much greater than the Na ϩ currents (see Table I). In H c for the latter transporters, glucose activated outward currents at potentials more positive than ϳϩ20 mV (ϳϩ100 nA at ϩ50 mV). A comparison of the sugar-induced currents at V m ϭ Ϫ150 mV revealed similar values in the presence of H c for WT (Ϫ1220 nA), D204C (Ϫ1120 nA), and D204N (Ϫ870 nA), and a reduction of about 75% for D204E (Ϫ300 nA). In Na c , WT and D204E showed comparable currents generated by 100 mM Dglucose at Ϫ150 mV (Ϫ1480 and Ϫ930 nA). Under the same test conditions, D204C and D204N exhibited less than 10% of the current observed for WT.
We measured the sugar kinetics of each transporter by varying external [Glc] from 0.05 to 100 mM in the presence of Na c or H c . The apparent affinity constants for D-glucose (K 0.5 Glc ) at V m ϭ Ϫ110 mV are shown in Table I. As described earlier (20), hSGLT1 exhibited a K 0.5 Glc of 0.3 mM in Na c . When H c substituted for Na c the apparent affinity for Glc was reduced by 20-fold (K 0.5 Glc ϭ 6 mM). A similar reduction of the apparent Glc affinity in H c was also observed for each transporter with a replacement of Asp 204 . A closer examination of K 0.5 Glc revealed two classes of mutations. First, Glu in place of Asp 204 reduced the apparent affinity for Glc in either Na c or H c about 4-fold. The second class encompasses transporters with a neutral side chain at position 204 (Asp 204 3 Asn or Cys), which showed increased apparent affinities for Glc. Although the K 0.5 Glc of D204C in H c was only slightly decreased (3 mM), a ϳ5-fold reduction of K 0.5 Glc was observed in Na c (0.07 mM). D204N exhibited a more dramatic increase of the apparent affinity for Glc with a 10-and 15-fold reduction of K 0. 5 Glc in H c (0.6 mM) and in Na c (0.02 mM), respectively. Over a range from Ϫ150 to Ϫ30 mV, the K 0.5 Glc determined in Na c was essentially voltage-insensitive for all transporters. From Ϫ150 to Ϫ50 mV, WT exhibited an approximate 4-fold increased K 0. 5 Glc in H c , but no voltage dependence of the K 0. 5 Glc was observed for the other transporters. The I max Glc values for each transporter increased (more negative) with hyperpolarizing potentials and were comparable to the currents generated by 100 mM D-glucose (see Fig. 2). The cation had no impact on the magnitude of I max Glc for WT (Table I). WT and D204E exhibited a sigmoidal I max Glc -V relation in Na c and thus, the similar I max Glc values of these two transporters shown at Ϫ110 mV (Table I) were not significantly changed by more hyperpolarizing potentials. With the exception of WT, no saturation of the I max Glc -V relationship was observed for each transporter at hyperpolarizing potentials in H c .
To elucidate whether the substitution of Asp 204 in hSGLT1 changed the sugar specificity of the transporter, oocytes were held at Ϫ50 mV and currents generated by 10 mM sugar were monitored in the presence of Na c or H c . Under either test condition the selectivity pattern for each transporter was in the order Glc Ն ␣MDG Ն D-galactose Ͼ 3-O-methyl-D-glucose (data not shown). Consistent with the effect of the cation on the apparent Glc affinity for WT and D204E, each sugar induced ϳ10-fold lower currents in the presence of H c than in Na c . The hSGLT1 H ϩ Channel effect on the apparent Glc affinity of D204E was reflected by ϳ4-fold lower sugar-induced currents in Na c and H c than with WT. In H c , D204C and D204N exhibited sugar-induced currents similar to WT, but the currents generated by each sugar in Na c were ϳ50-fold lower than the sugar-evoked currents observed for WT.
We determined the apparent half-maximal cation concentration for Na ϩ -and H ϩ -activated glucose transport (K 0.5 cation ). Table I summarizes K 0.5 cation for all transporters at V m ϭ Ϫ110 mV. For WT the K 0.5 H ϩ (7 M, pH 5.2) was about three orders of magnitude smaller than the K 0.5 Na ϩ (6 mM). Again, the nature of the substitution of Asp 204 grouped transporters with a neutral amino acid side chain (Asp 204 3 Asn or Cys) in one group. These transporters exhibited a 10-fold reduced K 0.5 Na ϩ , but the K 0.5 H ϩ was not significantly altered. On the other hand, Glu in place of Asp 204 reduced K 0. 5 Na ϩ by about 3-fold (2 mM). However, this conservative substitution reduced K 0.5 H ϩ of this transporter by a factor of Ͼ20, shifting the pH of the apparent half-maximum concentration for H ϩ -activated glucose transport from pH 5.2 to ϳ6.5. In general, for each transporter K 0.5 Na ϩ was slightly voltage-dependent and increased about 4-fold from Ϫ150 to Ϫ50 mV, while K 0.5 H ϩ over the same voltage range increased by a factor of about 2.
The Hill coefficient (n) of WT for cation-activated glucose transport was larger than unity for both Na ϩ (n Na ϩ Ͼ 1.5) and H ϩ (n H ϩ Ͼ 1.2) and was not affected by voltage (Ϫ50 to Ϫ150 mV). n Na ϩ of D204E was Ͼ1.3 but for H ϩ /sugar cotransport n H ϩ was Ͻ1. Independent on the cation, D204C and D204N exhibited a n Na ϩ or n H ϩ of Ͻ 1.
Whereas at Ϫ110 mV I max Na ϩ for each transporter was essentially indistinguishable from I max Glc in Na ϩ , this similarity in H c was only observed for WT and D204E (note that the I max Na ϩ -V relation of D204C and D204N and the I max H ϩ -V relation of D204E did not saturate at hyperpolarizing potentials). On the other hand, at V m ϭ Ϫ110 mV, a transporter with a neutral amino acid side chain at position 204 exhibited a ϳ3-fold larger I max H ϩ (D204C, Ϫ1630 nA; D204N, Ϫ2450 nA) than I max Glc in H c (D204C, Ϫ675 nA; D204N, Ϫ769 nA) and I max H ϩ increased supralinearly with more negative potentials.
To determine the net Na ϩ and H ϩ leaks through each transporter, [Na ϩ ] or [H ϩ ] was varied between 0.1 and 100 mM Na ϩ or between 0.03 and 32 M H ϩ and phlorizin-sensitive currents and choline currents at pH 8.0 were subtracted from the total currents. The kinetics of the cation leak were calculated at V m ϭ Ϫ110 mV by fitting the data to Eq. 2. In general, for each transporter K 0.5 Na ϩ for glucose cotransport and leak were essentially identical, but the K 0.5 H ϩ value for the leak was ϳ10-fold smaller than K 0.5 H ϩ . WT exhibited a leak I max Na ϩ of Ϫ97 Ϯ 4 nA and replacement of Asp 204 with Asn (54 Ϯ 5 nA), Cys (18 Ϯ 2 nA), or Glu (47 Ϯ 3 nA) reduced the leak I max Na ϩ ϳ2to 5-fold. The leak I max H ϩ for WT (430 Ϯ 4 nA) was 2.5 to 3 times larger than for D204C (Ϫ172 Ϯ 16 nA) and D204N (Ϫ142 Ϯ 7 nA), but there was apparently no H ϩ leak through D204E. Although the Hill coefficient (n) for the Na ϩ or the H ϩ leak pathways was Ͼ1.2 only for WT, the value of n for each modified transporter was Յ1.
Pre-Steady-State Charge Movement- Fig. 3A shows representative current records of oocytes injected with WT or D204E cRNA in the presence of Na c or H c after stepping the membrane potential from the holding potential (Ϫ50 mV) to the test potential (V m ϭ Ϫ150 to ϩ50 mV in 20-mV decrements). After the initial fast membrane capacitive transients (Ͻ 1 ms) each transporter exhibited currents that relaxed to a steady-state with a single time constant (see Fig. 4). These relaxations were abolished after addition of saturating [sugar] and/or [phlorizin] (data not shown). The removal of a negative amino acid side chain at position 204 caused a dramatic reduction of the current transients (compare Q max in Table I).
The charge movement was obtained by integration of the transporter-mediated relaxation. At each V m , a plot of the charge (Q) versus the membrane voltage (V m ) yielded a sigmoidal charge-voltage (Q-V) curve (Fig. 3B). Data were fitted to Eq. 3 to obtain the Boltzmann parameters Q max (maximal charge), V 0.5 (V m at which 50% of the charge has moved in the membrane electric field), and z (the apparent valence of the charge).  hSGLT1 H ϩ Channel Q max of WT and D204E were comparable in Na c and H c (ϳ23 nanocoulombs (nC)), and Q max of D204C and D204N were reduced about 5-fold (Table I). The substitution of Na c by H c shifted V 0.5 of WT from Ϫ36 Ϯ 1 mV to Ϫ74 Ϯ 4 mV. Glu in place of Asp 204 exhibited a V 0.5 in Na c of Ϫ50 Ϯ 2 mV. However, the V 0.5 of this transporter in H c was shifted to more positive potentials (V 0.5 ϳϩ40 Ϯ 5 mV). In Fig. 3B the effect of varying [cation] on the charge-voltage (Q-V) relationship is shown for WT and D204E. At [Na ϩ ] or [H ϩ ] below 25 mM or 3.2 M, respectively, the Q-V curve for WT didn't become saturated at negative potentials. D204E exhibited a sigmoidal Q-V curve over the entire [Na ϩ ] range tested. At [H ϩ ] higher than 3.2 M, the Q-V relationship of the latter transporter did not become saturated at ϩ50 mV. Relaxation Time Constants-Substraction of capacitive and steady-state currents from the total currents (see Fig. 3A) revealed a monoexponential time constant () of the pre-steadystate relaxation currents. was voltage-dependent in the ON response but not dependent on voltage in the OFF response (Fig. 4). In Na c the shapes of the ON -V curves for the transporters tested were bell-shaped and described a Gaussian fit. The maximum ( max ) for WT and D204E was ϳ20 ms and ϳ27 ms at ϳϪ100 mV. In contrast, in H c the -V curve of WT didn't reach max over the applied V m (ϩ50 mV to Ϫ150 mV) but the bell-shaped -V relation of D204E was dramatically shifted to the right ( max of ϳ32 ms at ϳϪ22 mV). As reported above, D204C and D204N exhibited very small pre-steady-state currents that were at the border of resolution and precluded us from determining reliable max values over the entire voltage range.
Stoichiometry-The stoichiometry of ion/sugar cotransport was determined by simultaneously measuring the unidirectional flux of D-[ 14 C]glucose into oocytes expressing WT or D204N and monitoring the sugar-evoked inward current. Integration of the inward current with time revealed the net positive charges that entered the oocyte. A plot of the net charge against glucose uptake by oocytes expressing WT revealed a linear relation in both Na c and H c with a slope of ϳ2 inward charges per glucose molecule transported (2.1 Ϯ 0.1 in Na c ; 2 Ϯ 0.3 in H c ) (Fig. 5). In the presence of Na c the slope of the inward charge/glucose relation of D204N was 2.1 Ϯ 0.4. Replacing Na c with H c changed the slope for D204N to 39 Ϯ 3. By reducing the final D-[ 14 C]glucose concentration in H c from 5 to 0.1 mM, the slope for D204N was 144 Ϯ 7 (Fig. 5, inset), but no significant effect on the H ϩ /glucose stoichiometry was observed for WT (data not shown). No Na c -or H c -dependent glucose uptake or sugar-evoked current was observed in control (H 2 O-injected) oocytes.
Outward Currents-We next studied the effect of intracellular acidification on glucose-induced H ϩ currents. Acidification of the intracellular compartment was produced by rapid replacement of external choline chloride with potassium acetate (21). Fig. 6 shows representative I-V curves on the effect of internal acidification of oocytes expressing D204N or WT in the presence of 31.6 M H ϩ (pH 4.5). In the absence of acetate, the glucose-induced currents for D204N increased with hyperpo-larization and reversed at ϩ23 mV (Fig. 6A). Internal acidification increased the magnitude of outward currents with depolarization (ϩ800 nA at ϩ50 mV) but had only minor effects on currents at potentials more negative than Ϫ70 mV. No outward currents were observed after internal acidification of oocytes expressing WT under either condition (Fig. 6B), despite hSGLT1 H ϩ Channel the fact that the amount of WT in the plasma membrane was about four times that of D204N (Q max for WT ϭ 14 nC, for D204N ϭ 4 nC). Addition of 0.2 mM phlorizin reversibly blocked the inward and outward glucose-induced currents for D204N completely (Fig. 6C).

DISCUSSION
The present study confirms and extends our previous reports on the kinetics of SGLT1 (4,5): First, the steady-state and pre-steady-state kinetics of the transporter in Na ϩ are in close agreement with previous observations on hSGLT1 (10,17,20,(22)(23)(24). Second, we confirm and extend our findings on rabbit SGLT1 (25,26) that H ϩ can drive sugar cotransport through hSGLT1. The kinetics of H ϩ and Na ϩ sugar cotransport are quite similar, even though the apparent affinity of hSGLT1 for H ϩ is ϳ1000-fold greater than for Na ϩ , and the apparent affinity for sugar is about 10-fold lower in H c than in Na c (see Table I). In fact, the maximum rates of cotransport, the cationto-sugar stoichiometry (2:1), the Hill coefficients (Ͼ1), turnover numbers (ϳ50 s Ϫ1 ), and voltage dependence of transport are virtually identical for Na ϩ and H ϩ cotransport when measured in the same oocyte. Furthermore, the pre-steady-state kinetic parameters Q max , V 0.5 , and max are comparable in Na c and H c . The only quantitative differences are that the turnover number for the leak pathway (uniport) (27) is about 5-fold higher in H c (22 s Ϫ1 ) than in Na c (4.4 s Ϫ1 ) and that K 0. 5 H ϩ is about one order of magnitude smaller for the leak than for the cotransport. The latter observation is consistent with the results for hSGLT1 reported by Chen et al. (24). In contrast to their work (they reported a Hill coefficient for Na ϩ uniport ϭ 1 and for H ϩ uniport Ͼ 1), the present study clearly emphasizes that the Hill coefficients for both Na ϩ and H ϩ uniport are Ͼ1. Together with the concentration dependence of V 0.5 (95 mV per 10-fold change in [Na ϩ ] or [H ϩ ]), this finding supports our hypothesis that two cations, Na ϩ or H ϩ , bind to SGLT1 before sugar. This is in direct contradiction to the Na ϩ (or H ϩ )-sugar-Na ϩ binding sequence model proposed by Chen et al. (24) based on their experiments. However, Chen's model is inconsistent with their reported results, because it does not take into account that the Hill coefficient for the H ϩ leak current was Ͼ1 and that the currents for the Na ϩ leak were no greater than 15 nA (com- hSGLT1 H ϩ Channel pared with 100 nA in the present study). Furthermore, their conclusion that Na ϩ is the last substrate bound to the transporter cannot account for the steady-state kinetics, because the maximum rate of Na ϩ /glucose cotransport is independent of [Na ϩ ] (4,5).
To explore the role of the N terminus of hSGLT1 in cation/ sugar cotransport, we have analyzed the effect of replacing a conserved amino acid residue, Asp 204 . This residue is aligned with Asp 187 in the Na ϩ /proline cotransporter (PutP), which is functionally important for cation selectivity and Na ϩ -dependent proline binding and transport (12). Replacing Asp 204 revealed two functional classes of mutations, depending on the presence or absence of a negative-charged amino acid side chain at position 204. This stands in contrast to PutP, where a polar rather than a charged residue at position 187 is essential for function.
The first class of mutations consists only of a transporter with a Glu in place of Asp 204 . Independent of the coupling cation, the apparent affinity for glucose is increased by ϳ4-fold, indicating a cation species-independent effect on the glucose translocation through D204E. The apparent affinity of the transporter for Na ϩ is increased by 3-fold, whereas the I-V relation and I max Glc in Na c ϩ , I max Na ϩ , and the leak I max Na ϩ are all comparable to WT. These minor kinetic differences are reflected by similar turnover numbers by these transporters. On the other hand, D204E exhibits an increased apparent H ϩ affinity. This implies that lengthening the side chain at position 204 by only one methylene group (about 1.5 Å) dramatically influences, directly or indirectly, the geometry of the cation site(s). A similar observation was reported for the glutamate transporter (GLT-1 or EAAT-2) (28). The increased apparent H ϩ affinity of D204E is mirrored by the alkaline shift in the Q-V relation (Fig. 3B), suggesting that, even at very low [H ϩ ], binding of H ϩ induces conformational changes that are the basis for coupled transport (10). This fact is most likely the reason for the apparent independence of the Q-V relation of D204E on [Na ϩ ], because H ϩ is the preferred cation species. At [H ϩ ] larger than 3 M no saturation at depolarizing potentials is detectable, precluding reliable calculations in this [H ϩ ] range. However, between 0.1 and 3 M H ϩ ⌬ V 0. 5 10 of D204E is similar to ⌬ V 0. 5 10 for WT from 1 M to 32 M H ϩ . The ϳ20-fold increase in apparent affinity is also reflected by a ϳ110-mV shift of V 0.5 toward positive potentials and a max of D204E at Ϫ25 mV (no max is detectable for WT in H c from Ϫ150 mV to ϩ50 mV). According to computer simulations of the 6-state kinetic model for SGLT1 (5), this right shift of max and the generally higher values are due to higher binding and lower dissociation rate constants for the cation. This prediction is consistent with the observation that no H ϩ leak through D204E is observed and may explain why the glucose-evoked H ϩ currents (Fig. 2) were about ϳ5-fold smaller for D204E than for WT. Because the I-V relation of D204E in H ϩ does not saturate at hyperpolarizing potentials, the turnover number for H ϩ / glucose cotransport for D204E at Ϫ110 mV must be greater than 9 s Ϫ1 .
The second class of mutations encompasses transporters with a neutral side chain at position 204 (Asp 3 Asn or Cys). By using thiol labeling, the predicted cytoplasmic location of Asp 204 (1,29) is confirmed, because Cys 204 is only accessible from the cell inside (data not shown). Furthermore, introduction of a positive charge (by 2-aminoethyl methanethiosulfonate) doesn't have a significant effect on the kinetics of the transporter. Based on this observation, we may exclude a possible role of Asp 204 in salt-bridge formation with a positive amino acid as proposed for Asp residues in the lactose permease (LacY) (30 -32). However, as indicated by the 5-fold reduction of Q max for D204C and D204N, removal of the negative charge at position 204 reduces the number of transporters in the plasma membrane. These data imply a trafficking defect of the protein to the plasma membrane: a common problem observed with the expression of a membrane protein in a eukaryotic expression system (33,34). This trafficking problem precludes us from a detailed analysis of the pre-steady-state kinetics of D204C or D204N.
Analysis of steady-state kinetics of D204C and D204N shows a Ͼ10-fold reduction of K 0.5 Glc in Na c and H c . The finding that this single mutation also produces a 10-fold reduction in K 0.5 Na ϩ raises the idea that cation and sugar sites are in close proximity or may overlap. A similar hypothesis is proposed for the melibiose permease of E. coli, where an N-terminal domain of the transporter plays a fundamental role in connecting cationand sugar-binding sites in terms of coupling (35,36). Although the apparent affinity parameters of the (co)substrates for D204C and D204N were altered and the steady-state activation of the currents revealed a Hill coefficient Ͻ 1, neither the coupling ratio of Na ϩ to glucose transport (Fig. 5) nor the turnover number for Na ϩ /glucose cotransport (I max Na ϩ /Q max , see Table I) was significantly altered.
Unexpectedly, all experimental evidence indicates that neutralization of Asp 204 (D204C and D204N) in hSGLT1 results in the activation of a glucose-activated H ϩ channel. This conclusion is based on the observations that: 1) The glucose-evoked H ϩ currents are an order of magnitude greater than the Na ϩ currents (Fig. 2). Because the H ϩ currents do not become saturated within the applied voltage range, this would imply that the turnover number (I max H ϩ /Q max ) increases by more than one order of magnitude over that in Na ϩ (Table I). 2) The H ϩ currents reverse at positive voltages, unlike either the Na ϩ currents for D204C or D204N or the H ϩ and Na ϩ currents for WT or D204E (Fig. 2). These outward H ϩ currents are enhanced by internal acidification of the oocyte (Fig. 6A) and are blocked by the addition of phlorizin (Fig. 6C). 3) The H ϩ currents greatly exceed those expected for the strict coupling of H ϩ /sugar cotransport. The H ϩ /sugar stoichiometry increases from the expected value of 2 to as high as 144 (Fig. 5). The Na ϩ /glucose stoichiometry for both WT and D204N was 2:1. 4) There is no change in the H ϩ leak currents for these mutants. The turnover numbers for the H ϩ leak currents for D204C and D204N (Ϫ172 nA/5 nC ϭ 34.5 s Ϫ1 ; Ϫ142 nA/4.4 nC ϭ 32 s Ϫ1 ) are comparable to that for the wild-type protein (22 s Ϫ1 ).
These results imply that minor structural changes in hSGLT1, i.e. replacing a carboxyl side chain with an amine or sulfhydryl group, are sufficient to open a glucose-activated H ϩ channel. This channel activity is intimately associated with sugar transport, because the competitive blocker phlorizin is unable to activate the H ϩ channel but is only able to block the effect of glucose. Clearly, based on our results we cannot distinguish whether neutralization of this conserved acidic residue generates a new "artificial" H ϩ pore or affects interactions with the coupling cation and/or other parts of the protein, thereby modifying the "original" H ϩ pathway. Because Asp 204 is located in a cytoplasmic loop, it seems unlikely that this residue is part of the membrane-spanning H ϩ pathway. However, with the lack of high resolution structural data of this transporter, this question will remain an enigma.
Our observations lend support to the suggestion that cotransporters and channels share features in common. For example, the glutamate cotransporters also behave as ligand-induced chloride channels (37,38, see also Ref. 39 for review). However, this report shows that hSGLT1 appears to be unique in that: 1) ion channel activity is only observed in the mutagenized protein; 2) the ligand (sugar) opens a channel for the driving cation (H ϩ ) for cotrans-hSGLT1 H ϩ Channel port, and 3) SGLT1 functions as a monomer (40). The glutamate cotransporter EAAT3 occurs as a homopentamer in oocyte plasma membranes, and it has been suggested that the chloride channel is formed by the oligomer (41).