INTRODUCTION

Cotransporters are found in bacteria, plants and animals, and the driving cations are either Na⁺ (e.g. Na⁺/glucose; Ref. 1), K⁺ (e.g. insect K⁺/amino acid; Ref. 2) or H⁺ (e.g. Lac-permease; Ref. 3). Do all of these cotransporters share a common mechanism? All of these integral membrane proteins use the energy of the transmembrane electrochemical ion gradient to drive the accumulation of a substrate against its concentration gradient into the cell. This is commonly described as an ordered process, in which the binding of the first substrate (e.g. Na⁺) increases the affinity of the transporter for its co-substrate (e.g. sugar): the essential activator model. The transporter then undergoes another conformational change(s), which results in release of the substrates into the cell. Cotransporters are characterized as being highly specific for one cation above all others, but in earlier studies (e.g. Refs. 4 and 5), it was reported that Li⁺, for example, could partially substitute for Na (6, 7), and the bacterial melibiose cotransporter MelB can use gradients of H⁺, Li⁺, or Na⁺ (8). In this study we investigated the effects of cation substitution on function of the Na⁺/glucose cotransporter (SGLT1) in the steady state using electrophysiological methods and compared effects of substitution of Na⁺, H⁺, and Li⁺ on SGLT1 with those described for MelB (8). Fitting the results to a six-state kinetic model suggests that the largest part of the cation substitution effect can be attributed to cation binding on the internal and external faces of SGLT1. The similarity of the effects of cation substitution on sugar transport by SGLT1 and the melibiose cotransporter (8) suggests that the cotransporters share a common mechanism.

MATERIALS AND METHODS

Mature oocytes from Xenopus laevis were injected with cRNA encoding the rabbit isoform of SGLT1 (9). Ionic currents, which are stoichiometrically coupled to Na⁺ and sugar fluxes (10), were measured using the two-electrode voltage-clamp and a pulse protocol as described previously (100-ms pulses over the range +50 to 150 mV from a holding potential of 50 mV). Sugar-induced currents were the difference between the records taken in cation + sugar and the preceding record taken in cation alone (11). Oocytes were held in a perfusion chamber and bathed in (in mM) 100 NaCl (or LiCl or choline chloride), 2 KCl, 1 MgCl₂, 1 CaCl₂, 10 HEPES-Tris, pH 7.5. The pH of the choline buffer was varied between pH 5.5 and pH 7.5 by using mixtures of MES,¹ Tris, and HEPES buffers.

Kinetic constants were determined by fitting the data to the Hill equation: I = (I_max)([S]ⁿ)/([S]ⁿ + K_0.5ⁿ) using the non-linear fitting algorithm in Sigmaplot (Jandel, Foster City, CA). Here [S] is the substrate concentration, I is current, I_max = maximal current, the apparent affinity, K_0.5, is the concentration of S which gives 0.5 I_max, and n is the Hill coefficient. For sugar activation, n = 1. Cation activation was from 0.5-100 mM for Na⁺ and Li⁺. For H⁺ activation the pH range was from 5.5 (3.2 µM) to 7.5 (0.032 µM). Figures are from individual oocytes, unless noted in the legend; however, all experiments were repeated at least three times on different oocytes taken from different animals with similar results. Statistics for the kinetics are either the standard error for the fit or standard error of the mean. Error bars are not shown if they are smaller than the symbol.

All chemicals were purchased from Sigma, Aldrich, or Research Organics (Cleveland, OH) and were of the highest grade available.

The effects of cation substitution on transport were modeled by fitting the experimental data to the six-state kinetic model of SGLT1 (12-14). As this is a simplified model of SGLT1 function, a conservative approach was used. Rate constants were initially adjusted to match a representative data set for Na⁺. After this, only the rate constants affecting affinity for cation (k₁₂^o, k₂₁^o, k₅₆^o, and k₆₅^o) were altered to match the data from H⁺ or Li⁺. For simulations of cation activation the MG concentrations were: Na⁺, 10 mM; Li⁺, 100 mM; H⁺, 50 mM. For simulations of sugar activation, [Na⁺] and [Li⁺] were 100 mM, and [H⁺] was 3.2 µM (pH 5.5).

RESULTS

Fig. 1 shows the sugar-induced current recorded from an oocyte expressing SGLT1 to show the consistency of the transport characteristics for Na⁺, Li⁺, and H⁺; sugar is transported with the cation, which causes an inward cationic current, and this transport is sensitive to the classic inhibitor of SGLT1, phlorizin.

Activation of 100 mM sugar transport by [Li⁺] (0.5-100 mM) is shown in Fig. 2. Representative sugar-dependent currents for 5, 25, and 100 mM Li⁺ (Fig. 2A) are [Li⁺]- and voltage-dependent, and show saturation with 25 and 100 mM Li⁺ at 150 mV. The data at each voltage were fit to the Hill equation to determine the kinetic parameters (Fig. 2B). Values for the Hill coefficient (n) ranged from 1.3 ± 0.1 to 1.8 ± 0.1 at V_m = 50 mV (n = 3).

Fig. 3 compares current-voltage curves for sugar-dependent currents for the three cations. The sugar-dependent Na⁺ current showed the characteristic curve (11); there was no transport at +50 mV, and a substantial current at 0 mV (about 40% of the maximum), which approached saturation by 150 mV. Both the Li⁺ and H⁺ traces were shifted toward hyperpolarizing potentials, relative to the Na⁺ curve. Like Na⁺, neither supported transport at +50 mV. In choline, with 3.2 µM H⁺ (pH 5.5) in the bath, MG induced a small current at 0 mV (~10% of the Na⁺ current), but a gradient of Li⁺ was unable to support transport in the absence of an electrical driving force. As the membrane potential increased to 50 mV, the transport rate for both H⁺ and Li⁺ increased, but were only 30% and 10% of the Na⁺-driven current, and, while the Li⁺ supported transport saturated by 150 mV, the H⁺ current curve did not show saturation at 150 mV.

We measured cation and sugar kinetics to determine the origin of these functional differences. Fig. 4A shows the effect of the cation on the apparent affinity for MG (K_0.5^MG). The points are data from one oocyte (used in panels A-D). The K_0.5^MG (V_m = 150 mV) in Na⁺ was 0.15 ± 0.03 mM (n = 3) and essentially voltage-insensitive. The K_0.5^MG in H⁺ was 3.8 ± 1.0 (n = 3) mM and increased about 2-fold (to 6.8 ± 0.4 mM) as the membrane depolarized to 50 mV. When the driving ion was Li⁺, the K_0.5^MG was similar to that in H⁺ (1.7 ± 0.6 mM, n = 4), but the apparent affinity decreased by almost 15-fold (to 28 ± 8 mM) as the membrane depolarized from 150 to 50 mV.

The I_max/V curves for MG are plotted in Fig. 4B. The influence of membrane potential on I_max in each cation is similar to Fig. 2 (I_max^MG curves for H⁺ and Li⁺ are shifted toward hyperpolarizing values, relative to Na⁺, and the H⁺-driven transport does not saturate by 150 mV). I_max^MG is determined by the driving cation; at all voltages I_max^MG in Li⁺ is lower than I_max^MG in Na⁺, and I_max^MG in H⁺ is greater. In this oocyte, where the I_max^MG in Li⁺ was 70-90% of I_max^MG in Na⁺; I_max^MG in H⁺ was 170-200% of I_max^MG Na⁺.

Fig. 4C shows that the effect of the cation on the K_0.5^Glu for glucose was similar to MG: When Na⁺ was the driving ion the K_0.5^Glu was 0.1 ± 0.01 mM and voltage-independent; when the cation was H⁺ the K_0.5^Glu increased by an order of magnitude (1.7 ± 0.3 mM) and was about 2-fold higher (4.1 ± 0.6 mM) as the membrane potential decreased to 50 mV; and when the cation was Li⁺ the K_0.5^Glu increased about 15-fold as the voltage depolarized from 150 to 50 mV (0.9 ± 0.2 to 13.2 ± 1.2 mM).

This pattern changed for galactose (Fig. 4D). The K_0.5^Gal at 150 mV was 2 orders of magnitude higher in H⁺ compared to the K_0.5^Gal in Na⁺ (16 ± 1.6 mM versus 0.2 ± 0.02 mM, n = 3). The pattern for Na⁺ and Li⁺ was similar to that measured for the other two sugars. The highest affinity was measured when SGLT1 used Na⁺ and was insensitive to voltage. When SGLT1 used Li⁺, the K_0.5^Gal at 150 mV was an order of magnitude higher (4.9 ± 2.1 mM, n = 3) and highly voltage sensitive (at 70 mV, 57 ± 16 mM, n = 3). The 2-fold sensitivity of K_0.5^Gal in H⁺ to depolarization (45 ± 18 mM at 50 mV) was similar to that measured when SGLT1 used H⁺ to transport MG and glucose.

The effect of the individual cation on the I_max/V relationship was quantitatively identical for all three sugars. Fig. 4B shows the curves for MG, but the curves for the I_max/V relationship for the other sugars, measured in the same oocyte, were identical.

Na⁺-driven sugar transport has been described by an ordered six-state kinetic model, shown in Fig. 5 (12-14). Transport is envisaged as a series of conformational changes induced by ligand binding. The states are: the empty transporter, [C]; the transporter bound to Na⁺, [CNa₂]; and the sugar-cation complex, [SCNa₂]. The empty transporter binds 2 Na⁺ before the sugar. The sugar-cation complex then undergoes a conformational change, which results in transport of the cosubstrates into the cell. The substrates are released on the inside, and the substrate binding sites again become accessible at the external surface. The steps that are sensitive to the membrane voltage are the conformational change of the empty transporter between the internal and external membrane surfaces ([C]  [C]") and the cation binding/dissociation steps ([C]  [CNa₂]). The internal cation binding step in the simulation is insensitive to voltage (11).

By simply changing the rate constants k₁₂^o, k₂₁^o, k₅₆^o, and k₆₅^o, as shown in Table I, we could reproduce both the apparent cation and sugar affinity, the influence of membrane potential on affinity, and the magnitude of the change in I_max (I_max^Li = 75% I_max^Na; I_max^H = 200% I_max^Na, data not shown), although the present model does not reproduce the shape of the Li⁺ and H⁺ I_max/V curve. The activation of sugar transport by Na⁺, H⁺, and Li⁺ is shown in Fig. 6A as symbols, and the prediction using the simulation are the smooth curves. We used the same values to predict the effect each cation had on affinity for MG. The model predicts that when SGLT1 uses Li⁺ the K_0.5^MG will be the most sensitive to V_m, that H⁺ will be intermediate to Li⁺ and Na⁺, and suggests that K_0.5^MG approaches a constant value at extremely hyperpolarizing values, independent of the identity of the cation.

Table I. Rate constants for simulation of the six-state kinetic model of Fig. 4 The values for k12o and k21o used for simulation of MG affinity (K0.5MG) in the three cations (Fig. 2A and 5) are listed. k23 was adjusted to fit the Na+ data used in the figure. The other rate constants were unchanged from previously published values (13): k16o, 100 s1; k61o, 35 s1; k25, 0.01 s1; k23, 150,000 mol1 s1; k32, 20 s1; k45, 800 s1; k34, 50 s1; k43, 50 s1. k52 and k54 are determined by the other rate constants to satisfy requirements for microscopic reversibility (11). The constants  = 0.3, " = 0, and  = 0.7 describe the fraction of the membrane electrical field sensed by the ion binding to the external and internal sites, and that sensed by the empty carrier. Cation k12o k21o k56o k65o (mol2 s1) × 104 (s1) × 103 s1 (mol2 s1) × 102 Na 2 0.3 16 0.5 Li 3 150 10 0.5 H 20,000 1.5 40 1000

The H⁺-driven transport did not saturate under the conditions used for these experiments. This is at least partially due to the K_0.5^H of ~pH 5.2 and our maximum experimental pH of 5.5, so we could not saturate the cation binding site by 150 mV. We could therefore expect non-saturating I/V curves in H⁺, consistent with the data from Na⁺ (11) and Li⁺ (i.e. Fig. 2A). (The H⁺/sugar I/V curves approach saturation under other experimental conditions, unpublished observations.) The model predicts that the cation binding steps, k₁₂^o and k₆₅^o, are greatly increased for H⁺ (10⁴ and 2 × 10³-fold compared to Na⁺) and that k₂₁^o and k₅₆^o are similar to those for Na⁺. For Li⁺ K₂₁ was increased 500-fold over Na⁺. More than one set of parameters was found to fit the H⁺ data. The parameter set in Table I was selected based on our assumption that the H⁺ affinity would increase symmetrically.

DISCUSSION

These experiments show that under hyperpolarizing membrane potentials SGLT1 can use the electrochemical gradient of Na⁺, Li⁺ and H⁺ to drive sugar transport using a common mechanism. The effects of cation binding are to: 1) induce a conformational change in SGLT1, which enhances the cotransporter's affinity for sugar; and 2) alter a rate-limiting step in the transport cycle. Using these criteria, the favored cation was Na⁺ at physiological values of the membrane potential (V_m = 50 mV) where Na⁺ bestows the highest sugar affinity (K_0.5^MG = 0.15 mM) and transport capacity. Use of either Li⁺ or H⁺ imposed a kinetic penalty on apparent sugar affinity (by at least an order of magnitude) and altered the I_max (~80% of I_max^Na in Li⁺, ~190% of I_max^Na in H⁺). The maximal rate in Li⁺ was lower than Na⁺ for all sugars, and was highly regulated by the voltage. The fact that H⁺ was capable of supporting a significantly higher maximal rate of transport may play a physiological role in the very proximal parts of the gut where the pH of the chyme is acidic and the lumenal sugar concentration will be high.

The original experiments that defined the cationic requirements for intestinal Na⁺/glucose cotransport (4, 5, 7) did not detect cotransport supported by other cations. From the present experiments the reason becomes clear; transport kinetics, both affinity and maximal rate, are highly modulated by the membrane potential. Since in tissue or vesicle experiments the membrane potentials are nominally between 60 and 0 mV, it is expected that in these classic experiments transport energized by H⁺ should be small, and that by Li⁺ should be barely detectable.

The cation functions as an essential activator (Fig. 5), so when the cation binds it increases the affinity of SGLT1 for sugar. The highest sugar affinity will be measured when the cation binding sites are saturated, and the lowest sugar affinity when the cation sites are empty. Therefore, anything that affects cation affinity will also have an effect on the apparent affinity for sugar. When SGLT1 used Na⁺ it had the highest affinity for sugar, in the order glucose (0.1 mM at 150 mV) < MG (0.15 mM) < galactose (0.25 mM), and sugar affinity was essentially unaffected by the membrane potential (9, 11). This is consistent with Na⁺ serving as an essential activator, as the Na⁺ concentration was at least 5 times higher than the lowest Na⁺ affinity measured. In this case the Na⁺ binding site is essentially saturated at all values of V_m, and so the K_0.5^sugar should be constant (Fig. 4, A, C, and D). All three cations appear to activate SGLT1 using the same mechanism; all increase the apparent K_0.5^sugar with increasing [cation], and have a Hill coefficient > 1, suggesting that the stoichiometry (cation:sugar) remains constant at 2:1 (9, 11, 14).

When Li⁺ was the activator the apparent sugar affinity was an order of magnitude lower than in Na⁺, but the same pattern of affinity was maintained; highest affinity for glucose (0.9 mM) < MG (1.7 mM) < galactose (4.9 mM). The K_0.5^Li was very sensitive to the V_m (Fig. 6A) so when V_m depolarized the cation binding site became progressively less saturated. Since cation binding is required to increase sugar affinity, and Li⁺ affinity is much more sensitive to V_m than Na⁺ or H⁺, we would expect that the apparent sugar affinity in Li⁺ would greatly decrease as V_m depolarized (Fig. 4, A, C, and D).

The apparent affinity for sugar in H⁺, however, showed a different pattern. The profiles of MG and glucose affinity followed the expected order: glucose (1.7 mM) < MG (3.8 mM), but for galactose sugar affinity was decreased by a factor of about 10 (16 mM). As expected from the K_0.5^H/V relationship (Fig. 6A), the voltage sensitivity of the apparent affinity for all three sugars remained slight. This effect suggests that the binding of each cation induces a unique conformational change in the sugar binding pocket. In the case of galactose, since the only difference between it and glucose is the orientation of the hydroxyl at C-4, we expect that H⁺ binding produces a conformational change that results in misalignment of a residue, which is important in recognition of the C-4 hydroxyl.

Since fluorescence studies suggest that the cation binding site is remote (~35 Å) from the sugar binding site (15) the question arises "Can binding of these monovalent cations produce conformational changes in distant parts of the protein?" Hohenester et al. (16) have shown that the monovalent cation-dependent enzyme dialkylglycine decarboxylase undergoes specific conformational rearrangements, both at the reaction center (11 Å from the cation; Ref. 17) and even small ternary alterations in the arrangements of the dimers. These changes appear to be solely due to the mechanism by which the protein compensates for the differences in cationic radius and coordination number between the activators, the large cations K⁺ and Rb⁺, and the smaller inhibitors, Li⁺ and Na⁺. This mechanism centers on how the rigidity of the cation binding site is compensated by addition of a single water molecule, which both accommodates a reduction in coordination number (from 6 to 4 or 5) and replaces ion-ligand bonding. This leads to "an altered local protein structure around metal binding site 1 that, in turn, leads to changes in both the dialkylglycine decarboxylase active site structure and the quaternary structure of the dialkylglycine decarboxylase tetramer" (16). Such a mechanism may account for the cation selectivity as well as the effect of the cation on sugar recognition by SGLT1.

We have simulated the effect of cation substitution on the apparent affinity for both cation and sugar (Fig. 6) using the six-state model (12, 13). The simulations suggest that the greatest influence of cation substitution is at the cation binding steps, as changing only the rate constants for these steps, k₁₂^o, k₂₁^o, k₅₆^o, and k₆₅^o, can simulate not only the observed K_0.5 values, but how the membrane potential influences affinity for both cation and sugar. The model predicts that SGLT1 will have the highest affinity for sugar when it uses Na⁺, followed by H⁺ and then Li⁺. Using the same rate constants, it also correctly predicts the apparent affinity and voltage dependence of the cations: highest affinity for H⁺ and lowest for Li⁺.

The fact that the maximal transport rate, while set by the cation (highest in H⁺; lowest in Li⁺), is independent of the transported sugar (quantitatively identical for all three sugars in a given cation regardless of K_0.5^sugar), in turn suggests that the mechanism of translocation is separate from the process of sugar binding. In this scheme the molecular mechanisms underlying the kinetic model might be described as follows. 1) Binding of the cation causes a conformational change in the sugar binding site, which results in a increase in affinity for sugar. 2) When sugar binds it is transported to the inside, at a rate independent of the identity of the cation. 3) Maximal transport rate is determined by recycling of the empty transporter to the "outside-facing" conformation, and this is controlled by cation binding on the inside. Note that our experiments measure the overall kinetics, and all 14 rate constants influence the kinetics, so the translocation event (k₃₄  k₄₃) can remain constant even though the I_max increases.

The effect of cation substitution on SGLT1 function is similar to that described for the bacterial melibiose cotransporter (MelB) (for example, Refs. 8 and 18), which can also use Na⁺, H⁺, and Li⁺ to drive sugar transport. Table II is a comparison of the kinetic parameters for these transporters (MelB data taken from Ref. 8). In both cotransporters the order of decreasing cation affinity is H⁺ > Na⁺ ~ Li⁺, and both transporters preferred H⁺ by several orders of magnitude over Na⁺ or Li⁺. The same order follows for maximal velocity of transport; H⁺ supported a higher transport rate than Na⁺ or Li⁺, which were similar. And, like SGLT1, the driving cation determined the preferred substrate in MelB, presumably by differences in conformation of the sugar-binding site (18, 19).

Table II. Comparison of the effects of cation substitution on the melibiose cotransporter and SGLT1 The affinity data for SGLT1 is the average ± S.E. of 3-4 trials. The data for maximal uptake is from an experiment in which kinetics for all three cations was done on one oocyte; standard errors reported are for the fit. Note that the melibiose affinities are Kd and SGLT1 is apparent affinity, K0.5. The reference for the melibiose data is Pourcher et al. (8). Errors were not reported. Cation Affinity for cation Maximum uptake rate Affinity for sugar MelB (Kd) SGLT1 (K0.5) MelB (Vmax) SGLT1 (Imax) MelB (Kd) SGLT1 (K0.5) mM mM nmol/mg·min Na mM mM H+ 0.0005 0.007  ± 0.005 42 1860  ± 200 0.02 3.8  ± 1.6 Na+ 0.3 4  ± 4 2 920  ± 40 0.7 0.15  ± 0.05 Li+ 0.5 11.6  ± 3.9 1 666  ± 11 0.7 11  ± 1.0

On the other hand, there is a difference in how the membrane potential affected MelB transport kinetics in Na⁺ and H⁺ (20). If MelB used Na⁺ for melibiose transport, depolarization caused a decrease in V_max but no change in K_t^sugar. If the driving cation was H⁺, however, the same decrease in membrane potential resulted in a decreased affinity for sugar, and no change in V_max. In SGLT1 depolarization caused an increase in K_0.5^sugar in all three cations, and there was a voltage-sensitive range of I_max for Na⁺ and Li⁺, as well as a voltage-independent range as the membrane hyperpolarized; in H⁺, transport was voltage-sensitive over the entire range. These similarities suggest that the basic mechanism of activation and transport has been conserved in evolution from bacteria to mammals. We anticipate that further investigations of the effects of cation substitution, using lanthanides (21, 22), for example, will provide further insights into the characteristics of the cation binding site and mechanism of activation of cotransporters, and the sources of divergence of eucaryotic and procaryotic transporters.