Biophysical characteristics of the pig kidney Na+/glucose cotransporter SGLT2 reveal a common mechanism for SGLT1 and SGLT2.

The Na+-dependent, low affinity glucose transporter SGLT2 cloned from pig kidney is 76% identical (at the amino acid level) to its high affinity homologue SGLT1. Using two-microelectrode voltage clamp, we have characterized the presteady-state and steady-state kinetics of SGLT2 expressed in Xenopus oocytes. The kinetic properties of the steady-state sugar-evoked currents as a function of external Na+ and α-methyl-D-glucopyranoside (αMG) concentrations were consistent with an ordered, simultaneous transport model in which Na+ binds first. Na+ binding was voltage-dependent and saturated with hyperpolarizing voltages. Phlorizin was a potent inhibitor of the sugar-evoked currents (KiPz ≈ 10 μM) and blocked an inward Na+ current in the absence of sugar. SGLT2 exhibited Na+-dependent presteady-state currents with time constants 3-7 ms. Charge movements were described by Boltzmann relations with apparent valence ≈ 1 and maximal charge transfer ≈ 11 nC, and were reduced by the addition of sugar or phlorizin. The differences between SGLT1 and SGLT2 were that (i) the apparent affinity constant (K0.5) for αMG (≈3 mM) was an order of magnitude higher for SGLT2; (ii) SGLT2 excluded galactose, suggesting discrete sugar binding; (iii) K0.5 for Na+ was lower in SGLT2; and (iv) the Hill coefficient for Na+ was 1 for SGLT2 but 2 for SGLT1. Simulations of the six-state kinetic model previously proposed for SGLT1 indicated that many of the kinetic properties observed in SGLT2 are expected by simply reducing the Na+/glucose coupling from 2 to 1.

In the proximal tubule of the kidney, reabsorption of filtered glucose at the apical membrane is mediated by one or more Na ϩ /glucose cotransporters. The low affinity glucose transporter (SGLT2) 1 (1), cloned from a pig kidney cell line (2), bears 76% identity at the amino acid level to its high affinity homologue SGLT1 (3), yet exhibits functional characteristics (1) that clearly distinguish it from SGLT1.
In the present study we examined the mechanisms of Na ϩ / glucose cotransport via the pig SGLT2 transporter expressed in Xenopus oocytes. We determined the dependence of the steadystate kinetics on Na ϩ and ␣MG concentrations, and examined presteady-state charge movements associated with SGLT2. We have also extended our analysis of the substrate selectivity of SGLT2, its cation activation, and the effects of phlorizin upon both the uncoupled and the sugar-coupled Na ϩ pathways through the transporter.

EXPERIMENTAL PROCEDURES
Stage V-VI oocytes from Xenopus laevis were injected with pSGLT2 cRNA as described previously (1). Electrophysiological experiments were performed using a two-microelectrode voltage clamp (1,4,5). Oocytes were superfused at 20 -22°C with experimental medium in which the Na ϩ concentration varied between 0 and 100 mM by equimolar replacement with choline or Li ϩ . Test solutions (additionally containing sugars and/or phlorizin) were always washed out by superfusing the oocyte with substrate-free, choline chloride medium (pH 7.5). Sugar-evoked currents (I) were fitted to Equation 1, for which I max is the current maximum, S is the substrate (sugar or cation) concentration, and K 0.5 S is the substrate concentration for half-maximal current. The Hill coefficient (n H ), an empirical description of cooperativity, for S refers to the least number of molecules of S transported in each cycle and is, in general, the same as the coupling coefficient (n).
I ϭ I max ⅐ S nH ͑K 0.5 S ͒ nH ϩ S nH (Eq. 1) Charge movements were obtained as the integral of the presteady-state currents and fitted to the Boltzmann relation (Equation 2) for which Q max ϭ Q dep Ϫ Q hyp (Q dep and Q hyp represent the charge Q at depolarizing and hyperpolarizing limits), z is the apparent valence, and V 0.5 is the potential for 50% Q max ; F, R, and T have their usual thermodynamic meaning.
Model simulations for SGLT2 (see Fig. 7) were performed as described for SGLT1 (6). The steady-state concentrations of each transporter state (C x ) were estimated using the King-Altman procedure (in Parent et al. (6), see Fig. A1 and Equations A14 -A19). The steady-state currents and the kinetic parameters K 0.5 and I max were derived as described (Equations A20 -A43) in Parent et al. (6), but with the empty transporter valence (z C ) changed to Ϫ1, Na ϩ -coupling (n) ϭ 1, and internal concentrations [Na ϩ ] i ϭ 10 mM and [␣MG] i ϭ 0. The steadystate current is given by Equation 3, for which the rate k xy describes the reaction step C x 3 C y (see Fig. 7).
The presteady-state current (I T ) is given by Equation 4.
and was obtained by numerical integration of Equations A44 -A49 in Parent et al. (6). The phenomenological constant ␣Ј describes the fraction of the membrane electric field between bulk solution and the Na ϩ -binding site at the external face, ␣Љ is its internal equivalent, and ␦ is the fraction of the membrane field sensed by the empty binding site on the carrier during translocation; microscopic reversibility requires that ␣Ј ϩ ␣Љ ϩ ␦ ϭ 1 (7). I T was integrated over time to yield the simulated Q/V m data. The slow ( 1 ) and fast ( 2 ) time constants of transient current decay are given by where R k ϭ k 16 ϩ k 61 ϩ k 12 ϩ k 21 (Eq. 7) and W ϭ ͙͑R k 2 Ϫ 4͓k 21 k 16 ϩ k 21 k 61 ϩ k 12 k 61 ͔͒ (Eq. 8)

RESULTS
Currents Associated with pSGLT2-Following step-changes in membrane potential (V m ) from the holding potential (V h ) of Ϫ50 mV, we observed pSGLT2 transporter-mediated currents in oocytes (Fig. 1A) in the presence of Na ϩ . After the onset of the pulse (left panel), the transient currents comprised two components, (i) a capacitive current (also observed in controlinjected oocytes), which decayed with time constant Ϸ 0.7 ms, and (ii) a more slowly decaying current (time constant, 3-6 ms). At the end of the 100-ms voltage pulse, the "off" currents again included similar transients with opposite polarity to the "on" currents, before returning to the steady-state current at Ϫ50 mV. Following the addition of sugar (right panel) the transporter-mediated transients were attenuated (especially obvious from the off response) and there was a large increase in inward current. The glucose-dependent steady-state currents (at 20 mM) were roughly linear as a function of voltage between ϩ50 and Ϫ150 mV (Fig. 1B) Selected data were plotted as a function of V m . At 100 mM Na ϩ , K 0.5 ␣MG was 3 mM and independent of voltage ( Fig. 3A), whereas at 5 mM Na ϩ , K 0.5 ␣MG increased with depolarization. At Ϫ150 mV, K 0.5 Na (0.8 mM) was essentially independent of [␣MG] o (Fig. 3B); K 0.5 Na was relatively unchanged with depolarization in the presence of high ␣MG, but increased with depolarization at low ␣MG. The I max ␣MG /V m relationship was independent of [Na ϩ ] o (Fig. 3C). The relationship of I max Na to V m was similar at each [␣MG] o (Fig. 3D) however the magnitude of I max Na fell markedly FIG. 2. Steady-state kinetics of pSGLT2-mediated Na ؉ /␣MG cotransport at ؊70 mV. The kinetic parameters K 0.5 and I max were each derived for ␣MG and Na ϩ as a function of [Na ϩ ] o and [␣MG] o , respectively, by fitting data to Equation 1 (error bars represent the error in the estimates). All data were derived from a single oocyte superfused with test solutions containing 0.2, 0.5, 1, 2, 5, 10, 20, or 50 mM ␣MG at each of 1, 5, 10, 20, and 100 mM Na ϩ (in random order). The results presented were all at a test potential (V m ) of Ϫ70 mV; however, the kinetics were qualitatively no different at any V m between Ϫ150 and Ϫ30 mV. The same oocyte was used for measurement of presteady-state charge movements (Fig. 4B).
FIG. 3. Voltage dependence of the steady-state parameters for pSGLT2-mediated Na ؉ /␣MG cotransport. Selected kinetic data, all derived from the same experiment as in Fig. 2, were plotted as a function of membrane potential (V m ): A and C, ␣MG kinetics at 5 mM (f) and 100 mM (q) Na ϩ ; B and D, Na ϩ kinetics at 1 mM (ࡗ), 5 mM (f), and 50 mM ( ) ␣MG. Error bars represent the error in the estimates of kinetic parameters calculated using Equation 1. Dotted lines represent the corresponding predictions from the model simulation, using the rate constants given in Fig. 7. as [␣MG] o was reduced. The Na ϩ :␣MG coupling stoichiometry of 1:1 was inferred from the Hill coefficients (n H ) for Na ϩ and for ␣MG. At Ϫ70 mV, n H for Na ϩ activation of the 20 mM ␣MG-evoked currents was 1.3 Ϯ 0.2, and that for ␣MG activation (100 mM Na ϩ ) was 1.2 Ϯ 0.2; n H for Na ϩ did not vary with either V m or [␣MG] o (not shown).
Presteady-state Charge Movements-We investigated the relaxation kinetics of the SGLT2 presteady-state currents (Figs. 1 and 4A). At 10 mM Na ϩ , the on transients decayed with time constants ( on ) of 4 -6 ms, with a maximal value ( max on ) at ϩ8 mV. Increasing [Na ϩ ] o (within the range 1-20 mM) had the effects of (i) increasing max on and (ii) shifting to more positive potentials the V m at which on was maximal (i.e. V max on ): V max on shifted by Ϸ ϩ50 mV per 10-fold increase in [Na ϩ ] o . For off currents, off (4 -7 ms) was voltage-independent.
The transporter-mediated transients were integrated with time to estimate charge transfer (Q). The Q/V m relationship ( Fig. 4B) at 10 mM Na ϩ was described by a single Boltzmann relation (Equation 2) with Q max of Ϸ11 nC and V 0.5 of ϷϪ34 mV. The apparent valence (z) of the movable charge on the transporter was Ϸ1. Increasing [Na ϩ ] o from 1 to 20 mM shifted V 0.5 to less negative potentials, with ⌬V 0.5 Ϸ ϩ96 mV per 10-fold increase in [Na ϩ ] o (inset). At 100 mM Na ϩ , the Q/V m relationship did not saturate at positive V m ; however, from plots of V 0.5 against [Na ϩ ] o we estimated V 0.5 to be Ϸϩ60 mV.
[Na ϩ ] o had no effect on Q max or z.
Li ϩ and H ϩ were effective substitutes for Na ϩ in driving ␣MG transport via pSGLT2 (not shown). At Ϫ150 mV, the current evoked by 50 mM ␣MG in Li ϩ was identical to that in Na ϩ , and that with H ϩ (i.e. choline chloride at pH 5.5) was 36% of that in Na ϩ . The I/V m relationships in Li ϩ and H ϩ were shifted in the hyperpolarizing direction compared with Na ϩ .
Inhibition by Phlorizin-Phlorizin blocked the uncoupled Na ϩ current observed in the absence of sugar (Fig. 6A); this current was up to 8% of the maximal ␣MG-evoked current, with a reversal potential at Ϸϩ10 mV. Phlorizin inhibited the ␣MG-evoked currents: the inhibition constant (K i Pz ) was determined by Dixon analysis of the currents evoked by 3 and 20 mM ␣MG in the presence of 0.3-100 M phlorizin (Pz). At Ϫ50 mV (Fig. 6B) K i Pz was 9 M, and between Ϫ150 mV and Ϫ10 mV K i Pz was 8 -14 M. The I max ␣MG value (Ϫ260 nA) inferred from the Dixon plot (Fig. 6B) was consistent with the I max ␣MG measured (Ϫ230 nA) in the same oocyte. An identical K i Pz (8 Ϯ 3 M) was obtained from plots (not shown) of the phlorizin concentrations that inhibited the 1, 2, 5, and 20 mM ␣MG-evoked currents by 50% (by extrapolating to zero ␣MG).

DISCUSSION
A molecular basis for the kinetic heterogeneity of Na ϩ /glucose cotransport activity in renal brush-border membrane vesicles (8,9) was established with the cloning of two distinct glucose cotransporters from kidney, SGLT1 (high affinity for glucose) and SGLT2 (low affinity). The amino acid sequence identity among the SGLT1 isoforms from rat, rabbit, and man exceeds 87% (10), and pig SGLT2 is 76% identical to pig SGLT1 (1). The residues which differ for SGLT2 but are conserved throughout the SGLT1 subfamily must account for the different characteristics of SGLT2. Several of the SGLT1 isoforms Presteady-state currents were recorded in pSGLT2 cRNA-injected oocytes in Na ϩ medium in the absence of sugar following step changes in V m from the holding potential (V h ) of Ϫ50 mV (as shown in the top panel). A, compensated membrane current records (with capacitive transient and the steady-state current substracted; filtered at 100 Hz for display) from 3 ms after the voltage step, in 10 mM Na ϩ . B, charge/ voltage relationship at 10 mM Na ϩ . The presteady-state currents were integrated with time and averaged for on and off currents; data are the mean Ϯ S.E. from six determinations in a single oocyte (the same oocyte as in Figs. 2 and 3). The data were fitted to the Boltzmann relation (solid line, Equation 2), from which we determined V 0.5 (the V m for 50% charge transfer) to be Ϫ34 mV. The fitted Boltzmann relation was then compared with the model prediction (dotted line, V 0.5 ϭ Ϫ28 mV) using the rate constants determined for pig SGLT2 (Fig. 7). The predicted values of Q were normalized by setting have been extensively characterized in terms of presteadystate and steady-state kinetics and cation/substrate selectivity, and a kinetic model has been proposed (4 -6, 10 -15). 2 In this study of the functional mechanism for SGLT2, we found that, although SGLT1 and SGLT2 share many similar properties, there were a number of striking differences, including sugar specificity and Na ϩ /glucose coupling.
Substrate Selectivity-The reactivity of SGLT2 with ␣MG (K 0.5 ␣MG Ϸ 3 mM), glucose (K 0.5 Glc Ϸ 6 mM) and galactose (K 0.5 Gal Ͼ Ͼ 20 mM), and its exclusion of 3-O-methyl-D-glucopyranoside, L-glucose and mannose, is entirely consistent with the substrate selectivity of the low affinity glucose transport system previously described by Turner and Moran (8,9) for brush-border membrane vesicles from rabbit outer cortex. SGLT2 reacted with its favored substrates (␣MG, glucose) with much lower apparent affinity than SGLT1: K 0.5 ␣MG was 6 -15-fold higher for SGLT2 than for the SGLT1 isoforms. That SGLT2 failed to react with either mannose or 2-deoxy-D-glucose indicated a requirement for the axial -OH at carbon-2 (C-2) in glucose, as was the case for SGLT1 (11). However, SGLT2 also required axial -OH at C-3 (as indicated by the exclusion of allose and 3-O-methyl-D-glucopyranoside) and at C-4 (as indicated by the weak interaction with galactose). Substrate recognition in SGLT2 may also involve hydrophilic interactions around C-6 since xylose and the L-enantiomer of glucose were both excluded.
Phlorizin, the classical inhibitor of Na ϩ -coupled glucose transport, is a phenyl-␤-glucoside, which prompted us to investigate the interactions of a range of phenylglucosides with pig SGLT2. Our group previously determined for rabbit SGLT1 (14) that conjugation of the phenyl ring to glucose in the ␤ configuration decreased the apparent affinity but that this was partially restored by the addition of -OH at the C-4 in the phenyl ring (i.e. arbutin). Arbutin evoked a greater current in SGLT2 than did glucose (Fig. 5), suggesting that the phenyl ring can nest in a vestibule possessing both a hydrophobic and a polar region near C-4 of the phenyl ring. SGLT2, in contrast to SGLT1 (14), also tolerated one of the phenyl-␣-glucosides, suggesting differences in the molecular architecture of the hydrophobic vestibule.
Phlorizin potently inhibited the ␣MG-evoked currents mediated by SGLT2. The phlorizin inhibition constant (K i Pz ) for SGLT2 was Ϸ10 M, higher than that for each of the SGLT1 isoforms (0.01-1.4 M) (10,15), consistent with the 10-fold lower affinity of SGLT2 for sugar.
These studies indicate that the substrate binding site in SGLT2 is distinct from that of SGLT1 with respect to (i) the residues interacting with the sugar at C-3 and C-4, and (ii) the dimensions of the binding site vestibule. The characteristics of a pig SGLT2-pig SGLT1 chimera indicate that sugar binding is mediated by the carboxyl-terminal half of the protein (16). In particular, the hydrophilic loops between putative membrane domains M10/11, M12/13, and M13/14 possess substituted residues in pSGLT2 which are otherwise conserved throughout the SGLT family and may underlie the observed differences in sugar binding.
Mechanisms of Na ϩ /Glucose Cotransport-Despite differences between SGLT1 and SGLT2 in Na ϩ coupling and voltage dependence, their steady-state properties were very similar. As for SGLT1 (13), Li ϩ and H ϩ could each substitute for Na ϩ in driving sugar transport mediated by SGLT2, most effectively at hyperpolarized V m . Na ϩ binding appears to be V m -dependent and saturable with hyperpolarization. K 0.5 Na was more steeply voltage-dependent in SGLT1 than in SGLT2: K 0.5 Na for SGLT1 was Ϸ 3-fold higher than SGLT2 at Ϫ150 mV, but 5-15-fold higher at Ϫ50 mV. However, in general the changes in K 0.5 and I max for SGLT2 arising from varying V m or cosubstrate concentration were very similar to SGLT1.
The data for SGLT2 are consistent with an ordered, simultaneous transport mechanism (Fig. 7) in which Na ϩ binds first. The K 0.5 increased sharply with diminishing cosubstrate concentration, indicating that both substrates were translocated simultaneously, rather than consecutively in which case the K 0.5 values are expected to decrease (17). The order of binding, Na ϩ first, sugar second, was revealed by the I max data: I max Na was dependent upon [␣MG] o , but I max ␣MG was not affected by varying [Na ϩ ] o ; i.e. saturating ␣MG could always drive the transporter to its maximal rate regardless of [Na ϩ ] o , whereas [␣MG] o imposed a limitation upon the maximal rate of the transporter even at saturating Na ϩ . The magnitude of the sugar-uncoupled Na ϩ influx being Ͻ8% of the maximal sugar-evoked current suggests that the preferred configuration for translocation across the membrane is the fully loaded "carrier-Na ϩ -sugar" complex ( Fig. 7, states 3 3 4).
Thus SGLT2 and SGLT1 share a common mechanism in which the cation (Na ϩ , Li ϩ , and H ϩ ) binds before glucose, followed by a simultaneous translocation step, but with distinct Na ϩ : glucose coupling stoichiometry, 1:1 for SGLT2 and 2:1 for SGLT1. Subtle differences in the shapes of the I/V m curves for SGLT2 and SGLT1 imply that the rate-limiting steps may differ between the two transporters.
Transient Charge Movements-The characteristics of the presteady-state charge movements for pig SGLT2 (Fig. 4) were similar to those obtained for SGLT1 and the other Na ϩ -and H ϩ -driven transporters (see Table I  3-7 ms, within the same range determined for the other cotransporters; in considering species differences, the SGLT2 time constants may not differ from SGLT1 (cf. 4 -25 ms for the rat, rabbit, and human isoforms). Whereas V 0.5 at saturating activator concentration for most other cotransporters ranged from Ϫ50 mV to 0 mV (including the SGLT1 isoforms), we estimated that V 0.5 for SGLT2 at saturating Na ϩ (100 mM) was Ϸϩ60 mV. This indicated that at resting membrane potential a greater proportion of the SGLT2 ligand binding sites are oriented extracellularly compared with the other cotransporters. For SGLT2, the shift in V 0.5 brought about by a 10-fold increase in activator (Na ϩ ) concentration was Ϸϩ100 mV, similar to that for other cotransporters. For both SGLT2 and SGLT1 (4), 2 the apparent valence (z) was Ϸ1 and did not change with [Na ϩ ] o .
Using the relation Q max ϭ C T ⅐z⅐e (for which e is the elementary charge) the density (C T ) of functional SGLT2 carriers in the oocyte membrane was in the order of 10 11 /oocyte. The validity of estimating transporter density from charge movement data for transporters and ion channels expressed in oocytes was confirmed using freeze-fracture electron microscopy (19). The turnover rate of the fully loaded transporter, given by I max /Q max , was Ϸ60s Ϫ1 and was similar to that determined for the other cotransporters. Thus we conclude that the presteadystate kinetic properties of SGLT2 are very similar to SGLT1.
Model Predictions Resulting from Changing Na ϩ /Glucose Coupling from 2:1 to 1:1-Our conclusion that the kinetic behavior of SGLT2 is very similar to SGLT1 suggests a common mechanism, but with different stoichiometry. What changes in kinetic parameters should we expect as a direct result of reducing the Na ϩ /glucose coupling stoichiometry from 2:1 (as for SGLT1) to 1:1 (as for SGLT2)? We simulated the six-state kinetic model of rabbit SGLT1 (6,12) and determined the specific consequences of changing the Na ϩ -coupling coefficient (n) from 2 to 1 (Fig. 8) without altering any other parameters. Allowing for some quantitative discrepancies that may arise from comparing SGLT homologues from different species, the n ϭ 1 model predicts a number of changes which are consistent with experimental observations for pig SGLT2: (i) the observed reduction in K 0.5 Na and the reduced voltage dependence on K 0.5 Na (Fig. 8A). The rabbit model predicted K 0.5 Na to decrease by an order of magnitude upon decreasing n from 2 to 1, whereas the observed change for pig SGLT2 was 3-5-fold; however, this difference was reconciled by a reduction in k 23 o (the rate constant for external sugar binding) by an order of magnitude (see below for justification, and Fig. 3B); (ii) no change in the apparent affinity for ␣MG at negative V m , and a less marked FIG. 8. Model prediction of the effects of changing the 2 Na ؉ : 1 ␣MG coupling stoichiometry of rabbit SGLT1 to 1:1. Presteadystate and steady-state currents were simulated (at 20°C) using the kinetic model for SGLT1 (6), with Na ϩ -coupling (n) of 2 and with the rate constants previously determined for rabbit SGLT1 (12) (k 54 o should correctly be 2.24 ϫ 10 7 M Ϫ1 ⅐s Ϫ1 ). These were compared with predictions using the n ϭ 1 model with identical rate constants. A and B, apparent affinity constants for ␣MG and Na ϩ (at 100 mM Na ϩ , and 2 mM ␣MG, respectively). C, predictions of the currents evoked by 2 mM ␣MG in 100 mM Na ϩ , normalized according to I o , the current at saturating hyperpolarization (Ϫ200 mV). Six-state kinetic model in which the empty carrier is negatively charged (the apparent valence of the movable charge is Ϫ1) and the Na ϩ -coupling coefficient is 1. The reaction scheme is described by 14 rates (where the rate k xy represents the reaction step x 3 y). Each rate is defined by its potential-independent rate constant (k xy o ), V m , and ligand (Na ϩ ) concentration, as well as the coefficients ␣Ј, ␣Љ, and ␦, which describe the fraction of the electric field sensed by the Na ϩ binding to its external site (␣Ј) or internal site (␣Љ) and by the empty ion binding site on the carrier during membrane translocation (␦), where ␣Ј ϩ ␣Љ ϩ ␦ ϭ 1 (7) voltage dependence (Fig. 8B). This predicted voltage dependence was observed for rabbit SGLT1 when [Na ϩ ] o was low: when each were compared at 10 mM Na ϩ , K 0.5 ␣MG for pig SGLT2 displayed significantly less voltage-dependence than that for rabbit SGLT1 (Fig. 6A of Ref. 6). At all V m , K 0.5 ␣MG was higher for SGLT2 than that predicted by the n ϭ 1 model, but since the molecular architecture of the SGLT2 sugar binding site is discrete from that of SGLT1 (see above, and Fig. 5), we should not expect such a change to arise simply from the reduction in n; (iii) shifts in the I/V m and Q/V m relationships toward more positive V m (Fig. 8, C and D). These predictions were consistent with presteady-state data, although the steady-state evoked currents in SGLT2 varied over the V m range as for SGLT1 rather than shifting to depolarized V m ; (iv) a 50% reduction in Q max (Fig. 8D), which may in part account for the lower sugardependent currents typically obtained for SGLT2 compared with SGLT1 assuming similar levels of expression; and (v) a reduction in z from 1.4 to 0.9 (Fig. 8D), close to the measured value for SGLT2. Thus, collectively, the kinetic properties of SGLT2 are largely accounted for by the change in Na ϩ -coupling from 2 to 1.
Model Simulation of Presteady-state Kinetics and Steadystate Na ϩ /␣MG Cotransport Kinetics for SGLT2-Presteadystate and steady-state currents for pig SGLT2 were simulated according to the six-state model (Fig. 7) assuming 1 Na ϩ : 1 ␣MG coupling stoichiometry as suggested by Hill analysis; the model is otherwise identical to that proposed for SGLT1 (6,12). We found that the model could account qualitatively for the steady-state (Fig. 3) and presteady-state (Fig. 4) phenomena without changing the rate constants for rabbit SGLT1 (12), except for two conservative modifications: (i) k 23 o was reduced by an order of magnitude to reflect the 10-fold increase in K 0.5 ␣MG for SGLT2; (ii) k 56 o , a principal rate-controlling step in rabbit SGLT1 (6,12,20) at negative V m , was increased from 16 s Ϫ1 to 48 s Ϫ1 to account for the Ͼ 3-fold increase in turnover rate for SGLT2 compared to rabbit SGLT1.
The model closely predicted the Q/V m relationship ( Fig. 4C) with V 0.5 ϭ Ϫ28 mV at 10 mM Na ϩ (cf. Ϫ34 mV determined experimentally). The kinetic data in the V m range Ϫ150 to Ϫ50 mV indicated that the Na ϩ binding and charge transfer steps in the cycle were at saturation, since K 0.5 Na was V m -independent within this range (at saturating ␣MG) and V 0.5 was extremely positive. However, that the measured (but not predicted) I/V m curves still displayed V m dependence within this V m range implies that the model requires an additional voltage-dependent step in the transport cycle, e.g. Na ϩ dissociation at the internal face (states 5 3 6).
Conclusion-We have presented kinetic data demonstrating that Na ϩ /glucose cotransport mediated by SGLT2 can be described by an ordered, simultaneous transport model in which Na ϩ binds first. In the absence of sugar, SGLT2 exhibited Na ϩ -dependent transient charge movements which we attribute to reorientation of the charged carrier within the membrane. Taken together, these observations suggest that SGLT1 and SGLT2 share a common mechanism despite their functional differences. Among these apparent differences are the change in apparent Na ϩ : ␣MG coupling from 2:1 for SGLT1 to 1:1 for SGLT2, and the reduction in K 0.5 Na observed for SGLT2. However, our model predicts that the reduced K 0.5 Na is a direct consequence of the change in Na ϩ coupling altering the rate k 12 (according to Equation A1). Since this change need not involve a change in the rate constant k 12 o (which is a reflection of ligand binding affinity per se), it is likely that the Na ϩ binding site in SGLT2 is identical to one of the binding sites in SGLT1. The general reduction in apparent affinities for sugar and the effective exclusion of galactose suggests that the molecular architecture of the sugar binding site in SGLT2 is quite distinct from that of SGLT1. In ongoing structure-function studies of the Na ϩ /glucose cotransporters, the challenge is now to identify the molecular basis for these functional differences between SGLT1 and SGLT2.