Passive water and ion transport by cotransporters

METHODS

Xenopus laevis oocytes were isolated, injected with 50 ng of rabbit SGLT1 (Hediger et al. 1987) or human GAT1 (Nelson et al. 1990) cRNA, and maintained in Barth's medium at 18°C for 5-14 days (Parent et al. 1992a; Loo et al. 1996). The frogs were chilled by immersion in 0°C ice water for 30 min and killed by an overdose of Nembutal (60 mg for 60 min) in accordance with the protocol approved by the UCLA Chancellor's Animal Protection Committee. Substrate (glucose, GABA) transport was monitored using the two-electrode voltage clamp (Loo et al. 1993).

The osmotic water permeability of oocytes was estimated from the rate of change of cell volume when an osmotic gradient (-20 mosmol l⁻¹) was imposed by removal of mannitol from the superfusing solution. The oocyte cross-sectional area was measured every 0.5 s from bright-field images, and the images were digitized as previously described (Loo et al. 1996; Zeuthen et al. 1997; Meinild et al. 1998a). The relative volume of the oocytes (V/V_o) is related to the cross-sectional area A by: V/V_o = (A/A_o)^3/2, where V is the volume, V_o is the initial volume and A_o is the initial cross-sectional area. Osmotic water permeability (L_p) is proportional to the rate of change of cell volume: V_o(d/dt V/V_o) =L_p×S × V_w×Δπ, where S is the oocyte surface area (0.4 cm², Loo et al. 1996), V_w is the partial molar volume of water (18 cm³ mol⁻¹) and Δπ is the osmotic gradient. Oocytes were impaled with microelectrodes during the course of all experiments. Impalement of the oocytes with microelectrodes had no effect on L_p (Loo et al. 1996; Meinild et al. 1998a,b). The experimental chamber had a functional volume of 15 μl, and all experiments were performed under continuous superfusion (at a rate of 17 μl s⁻¹) of the bath solution, which was usually a NaCl buffer (mM: 90 NaCl, 20 mannitol, 2 KCl, 1 MgCl₂, 1 CaCl₂ and 10 Hepes (pH 7.4); osmolarity was 212 mosmol l⁻¹). The osmolarity of all solutions was measured with a vapour pressure osmometer (Wescor, Logan, UT, USA). All experiments were repeated on at least three oocytes from different donor frogs. With the exception of experiments to determine the activation energies for the Na⁺ leak and osmotic water transport, all experiments were performed at 20-21°C.

RESULTS

The Na⁺ leak current of SGLT1 is illustrated in Fig. 1A. Shown are continuous current records for an oocyte expressing SGLT1 and a control non-injected oocyte. The holding potential was -50 mV. In the SGLT1 oocyte (Fig. 1A, left panel), external phlorizin (100 μM) blocked a component of the inward holding current in the presence of Na⁺. After phlorizin blockade, the holding current was the same as that of control oocytes. The Na⁺ leak, which was the current blocked by a saturating concentration of phlorizin, was 8 % (29 nA) of the maximum sugar-coupled Na⁺ current (I_max; 356 nA) measured in the presence of a saturating concentration of sugar (5 mM α-methyl-D-glucopyranoside, α-MDG). Phlorizin had no effect on the holding current when external Na⁺ was replaced by choline. The Na⁺ leak was not observed in control oocytes (Fig. 1A, right panel). The Na⁺ leak saturated with increasing external Na⁺ concentration and negative membrane voltages. At maximal transport rate (V_m, -150 mV), the apparent affinity constant (K_0.5) for Na⁺ was 2.5 mM and the Hill coefficient was 2. For example, in one experiment, K_0.5 for Na⁺ was 3.8 ± 0.4 mM and the Hill coefficient was 2.3 ± 0.5. The inhibitory constant (K_i) for phlorizin inhibition of the Na⁺ leak current was 3-5 μM (data not shown). In addition to Na⁺, Li⁺ and H⁺ also carried the leak current, and the K_i for phlorizin inhibition of the Li⁺ leak current was 100 μM (data not shown).

The passive (osmotic) water permeability of SGLT1 is illustrated in Fig. 1B. Upon application of the osmotic gradient (-20 mosmol l⁻¹), the oocyte volume increased linearly with time. The rate of swelling was reduced to that of control oocytes by the addition of 100 μM phlorizin to the external solution (this study; Loo et al. 1996).

The phlorizin-inhibited passive water permeability was related to the level of expression of SGLT1. Figure 2 shows the direct correlation between passive volume flow (J_v) and the density of SGLT1 in the membrane. J_v was the total water flux under an osmotic gradient of -20 mosmol l⁻¹, and the number of transporters was obtained by dividing the maximal rate of sugar transport, I_max, in the presence of Na⁺ (the current induced by 5 mM α-MDG) by the turnover rate of rabbit SGLT1 (25 s⁻¹, Panayotova-Heiermann et al. 1994, 1995). The data were pooled from experiments performed in bath solution containing Na⁺, Li⁺ and choline. Volume flows measured in saturating concentrations of phlorizin (100 μM) with Na⁺ were similar to those of control oocytes (this study; Loo et al. 1996). The collinearity of the data indicate that the passive water permeability of SGLT1 is independent of the external cation. The slope of the regression line was 1.4 (± 0.4) × 10⁻¹⁴ cm s⁻¹ per transporter.

The blockade of the SGLT1 L_p by phlorizin (100 μM) in the presence of Na⁺, Li⁺ and choline is shown in Fig. 3. The L_p of the oocyte (the background L_p) has been subtracted. Phlorizin (100 μM) blocked 100 % of the SGLT1 L_p in Na⁺ buffer, but only 41 and 21 %, respectively, in Li⁺ and choline. The K_i for phlorizin blockade of the L_p was similar to the K_i for the leak currents: 3-5 μM in Na⁺ and ∼100 μM in Li⁺ (data not shown). The weak inhibition of the SGLT1 L_p by phlorizin in choline is consistent with the lack of effect of 100 μM phlorizin on current in choline (Fig. 1A).

The Na⁺ leak and passive water permeability were also sensitive to phloretin. The apparent affinities were similar: 100 μM external phloretin blocked the SGLT1 Na⁺ leak current by 47 ± 6 % (n = 5) and the SGLT1 L_p by 36 ± 6 % (n = 6). In control oocytes, 100 μM phloretin increased the holding current slightly, by 9 ± 4 nA (n = 3) at -50 mV, but had no effect on the L_p. In three control oocytes, the L_p was 2.2 (± 0.4) × 10⁻⁴ cm s⁻¹ in the absence, and 2.4 (± 0.2) × 10⁻⁴ cm s⁻¹ in the presence, of 100 μM phloretin.

The activation energies (E_a) of the passive water permeability and the Na⁺ leak were compared. The E_a values were obtained by measuring the phlorizin-sensitive Na⁺ leak current, I_Na, and passive water permeability, L_p, as the bath temperature was varied between 10 and 30°C. E_a values were determined from the slope of the Arrhenius plot of I_Na and L_p (Fig. 4). E_a for the SGLT1 L_p was 5 ± 1 kcal mol⁻¹ (n = 4); for the SGLT1 Na⁺ leak it was 19 ± 1 kcal mol⁻¹ (n = 3).

We performed experiments on the human Na⁺-Cl⁻-GABA cotransporter GAT1 to examine the differences between the water and Na⁺ leak pathways. SKF89976A is a specific inhibitor of GABA transport by GAT1 (see Mager et al. 1993). Figure 5A shows that there was a water permeability associated with GAT1 which was blocked by SKF89976A (25 μM). After blockade by SKF89976A, L_p was the same as in control oocytes (Fig. 5B). GAT1 did not exhibit a Na⁺ leak: in four oocytes expressing GAT1 (the GABA-induced current at -50 mV was 325 ± 86 nA), 200 μM SKF89976A had no effect on the holding current. The background current carried by Na⁺, measured as the difference in current when external Na⁺ was replaced by choline when membrane potential was held at -50 mV, was slightly higher in oocytes expressing GAT1 (20 ± 4 nA, n = 4) than controls (12 ± 1 nA, n = 3). Phloretin (100 μM, n = 3) had no effect on the passive water permeability of oocytes expressing GAT1 (data not shown).

DISCUSSION

The main conclusion of this study is that the Na⁺ leak and passive water transport through cotransporters exhibit very different characteristics. (1) Saturability. The Na⁺ leak saturates with increasing external Na⁺ concentration with an apparent affinity constant, K_0.5, for Na⁺ of 2.5 mM and a Hill coefficient of 2 at -150 mV (present study; Panayotova-Heiermann et al. 1998). In contrast, passive volume flow is linear with the imposed osmotic gradient (Loo et al. 1996; Meinild et al. 1998a). (2) Activation energy. The Na⁺ leak has a high activation energy, E_a, of 19 kcal mol⁻¹, while E_a was 5 kcal mol⁻¹ for passive volume flow. (3) Ionic specificity. The ion leak pathway is supported by Na⁺, H⁺ and Li⁺ but not choline (this study; Umbach et al. 1990; Panayotova-Heiermann et al. 1998). In contrast, the SGLT1 L_p is identical in the presence of Na⁺, Li⁺ and choline (Fig. 2). (4) Functionality. Data from flux studies of radiolabelled glucose and Na⁺ under voltage clamp indicate that the Na⁺ leak is not present under saturating glucose conditions (Mackenzie et al. 1998). In contrast, SGLT1 passive water transport is present under saturating conditions (Loo et al. 1996; Meinild et al. 1998a). The observation that GAT1 exhibits a passive water permeability, but not a Na⁺ leak (Fig. 5) is also consistent with different mechanisms for Na⁺ leak and passive water transport. Passive water transport has also been reported for the cystic fibrosis transmembrane conductance regulator (CFTR) and, similar to SGLT1, the passive water pathway appears to be separate from the ion permeation pathway (Schreiber et al. 1997).

We have previously proposed a 6-state ordered kinetic model for Na⁺-glucose cotransport (Parent et al. 1992b; Loo et al. 1998). The model describes Na⁺-coupled sugar transport as a series of ligand-induced conformational changes (Fig. 6). On the external membrane surface, two sodium ions bind to the empty transporter (conformation 1, C1), forming the complex C2 before the binding of the sugar molecule. Sugar is transported via a conformational change (C3 ⇌ C4) of the fully loaded transporter (C3) between the external and internal faces of the membrane. After the dissociation of sugar and Na⁺ from the transporter at the internal surface, the empty transporter (C6) is returned to the external surface via another conformational change (C6 ⇌ C1). In the presence of Na⁺ and sugar, phlorizin locks the transporter protein in one conformation (the phlorizin-bound form of C3). The model also incorporates an uncoupled Na⁺ flux through SGLT1 (uniporter or Na⁺ leak mode, C2 ⇌ C5) in the absence of sugar. The data obtained in the present study provide further independent evidence for this kinetic model, namely that: (1) the Na⁺ leak (C2 ⇌ C5) saturates with external Na⁺ concentration (see also Panayotova-Heiermann et al. 1998); (2) the Na⁺ leak (C2 ⇌ C5) and Na⁺-glucose cotransport (C3 ⇌ C4) share high activation energies, 19 kcal mol⁻¹ for Na⁺ leak and 26 kcal mol⁻¹ for Na⁺-glucose cotransport (Loo et al. 1996); (3) the apparent affinity constant, K_0.5, for Na⁺ (2.5 mM) and the Na⁺ Hill coefficient (n_H = 2) are the same for Na⁺ leak as for Na⁺-glucose cotransport; and (4) the Na⁺ leak and Na⁺-glucose cotransport are both blocked by phlorizin with the same sensitivity (K_i, ∼5 μM). The model also provides a framework for accounting for the effects of phlorizin on the leak current and passive water transport in the presence of Na⁺ and Li⁺, i.e. the similar K_i values for phlorizin blockade of the leak currents and L_p in Na⁺ and Li⁺ (Fig. 3).

The L_p is the same in the active (sugar-transporting) and passive (Na⁺ uniporter) modes (Loo et al. 1996; Meinild et al. 1998a), as well as when Na⁺ is replaced by choline (present study). Computer simulations of the kinetic model (see legend to Fig. 6) showed that at -50 mV: (1) in the presence of choline, 60 % of the cotransporters were in C1 and 40 % were in C6; (2) in Na⁺ but in the absence of sugar, 80 % were in C2, 10 % were in C1 and 10 % were in C3; and (3) in the presence of Na⁺ and sugar, 80 % were in C5 and 20 % were in C6 (see also Parent et al. 1992b). Thus C1, C2, C5 and C6 may each contribute to passive water permeability. Phlorizin, which has been shown to block the conformational changes of SGLT1, locks the protein in the phlorizin-bound form of C3, and in this conformation the transporter is able neither to undergo conformational changes (Loo et al. 1993, 1998), nor to transport water passively. Thus passive water transport by SGLT1 depends on the conformational state of the protein.

The inhibitors of L_p in SGLT1 and GAT1 block the conformational changes in both transporters (Mager et al. 1993; Loo et al. 1998). The common action of these compounds in locking the proteins in a conformation that does not allow water transport suggests that SGLT1 and GAT1 share a common mechanism for passive water transport.

The low activation energy (5 kcal mol⁻¹) of passive volume flow is comparable to that of aquaporins (3-4 kcal mol⁻¹, Preston et al. 1992; 6-7 kcal mol⁻¹, Meinild et al. 1998b), suggesting that passive water flow through SGLT1 is similar to water flow through an aqueous pore. It is noteworthy that the unitary L_p for SGLT1 is about 100 times lower than that of aquaporin1 (Zampighi et al. 1995; Loo et al. 1996). Aquaporin1 is strictly selective for water (Meinild et al. 1998b), implying that the pore is very narrow. The low SGLT1 L_p may be due to: (1) a longer effective pore length; (2) a lower mobility of H₂O in the pore (for instance, more binding sites and hydrogen bonding); or (3) a lower open probability for the SGLT1 water channel.

A number of transport proteins appear to exhibit the ability to transport water passively. These include the facilitative glucose transporter GLUT1 (Fischbarg et al. 1990), the CFTR (Hasegawa et al. 1992; Schreiber et al. 1997), the Na⁺-glucose cotransporter (Zampighi et al. 1995; Loo et al. 1996; Loike et al. 1996; Zeuthen et al. 1997; Meinild et al. 1998a) and the Na⁺-Cl⁻-GABA cotransporter (this study). In preliminary experiments, we have also observed passive water transport in the Na⁺-dicarboxylate transporter NaDC1 (Pajor, 1995), and the plant H⁺-amino acid transporter AAP5 (Fischer et al. 1995). Loike et al. (1996) have found that the passive water permeability of rabbit SGLT1 depends on external Na⁺, contrary to our present finding of independence from the cation. The reason for this discrepancy is not clear but it may be that we measured initial rates (within 30 s) while the experiments of Loike et al. (1996) were considerably longer (up to 30 min). Meinild et al. (1998b) have recently found that osmotically induced volume changes in Xenopus oocytes exhibit significant time-dependent decreases. This would lead to large discrepancies between the L_p measured during initial rates (within 10 s) and that for longer periods.

What is the physiological relevance of passive water transport by SGLT1? The human small intestine absorbs 10 l of water a day, but the mechanism of water transport is not well understood. Fluid transport occurs in the absence of or even against bulk osmotic gradients, and is secondary to active solute transport. Water movement has been postulated to be driven by local osmotic gradients within the tissue (Diamond & Bossert, 1967). Since osmotic gradients have not been found within epithelial tissues and water channels have not yet been reported for the small intestine (overall, water permeabilities are low in the intestine), an alternative hypothesis has been proposed where water is transported across the intestine by cation-driven transporters, such as the Na⁺-glucose transporter in the apical and the K⁺-Cl⁻ transporter in the basolateral membrane (Zeuthen, 1991; Loo et al. 1996; Zeuthen et al. 1997; Meinild et al. 1998a). Based on the stoichiometry of 200-250 water molecules transported with each glucose molecule, we estimated that approximately 4-5 l of water are transported across the brush border membrane of the intestine by SGLT1 by secondary active transport (Loo et al. 1996; Zeuthen et al. 1997; Meinild et al. 1998a). The accompaniment of 200-250 H₂O molecules with two sodium ions and each glucose molecule is equivalent to a coupling ratio of 65-85 H₂O molecules per solute molecule. Since isotonicity requires 275 H₂O molecules per solute molecule, this results in a fluid which is hypertonic (Diamond, 1996). The finding that the passive water pathway is operational under sugar-transporting conditions (Meinild et al. 1998a) indicates that the hypertonic fluid transported across the apical membrane by SGLT1 per se would create an osmotic gradient, which could then provide a driving force for water movement through the passive water pathway of SGLT1. Thus according to our hypothesis, isotonic water transport across the brush border membrane of the intestine is the sum of secondary active water transport (coupled to Na⁺ and sugar transport) and a passive component through the SGLT1 water channel.