Transport of water against its concentration gradient: fact or fiction?

Molecular models of water cotransport. Resent structural information show that cotransporters of the symport type (red) work according to the alternating‐access concept.5–7 (a) Hydration of the access cavity (blue) allows the substrate (yellow) to enter from the outside solution and to bind in the cavity. Subsequently, water and substrate are occluded (b) and further conformational changes cause the cotransporter to open an aqueous cavity toward the inside solution (c), which allows substrate to be expelled. However, how about water transport? There will be a net transport of water, if the protein adopts one of the subsequent conformations (d1, d2, or d3) before returning to the initial, outward open conformation (a). In the occlusion model (d1), the protein attains a closed conformation that squeezes out the water before returning to (a).8 On this model, there is a net transport of water equal to the occluded water volume. In the hyperosmolar‐cavity model (d2), the substrate generates an intramolecular osmotic driving force while occupying the cavity.9 Water is transported by osmosis via an aqueous pathway that is open through the protein. In the Brownian piston model (d3), the channel is too narrow for the water molecules to circumvent the substrate.10 During exit, therefore, the substrate will push water molecules out, while the void behind the substrate molecule is taken up by water molecules that enter from the outside solution via the aqueous channel.

The unstirred layer model. In this model, the substrate flow generated by the membrane protein (red arrow) gives rise to an osmotic gradient across the membrane. As the substrate molecules diffuse toward and are captured by the protein from the outside solution, the number of substrate molecules is depleted in the solution surrounding the orifice of the protein. Consequently, the osmolarity decreases gradually toward the orifice of the protein with a maximum decrease at the membrane. Likewise, when the substrate molecules leave the membrane protein and diffuse into the inside solution, usually the cytosol, there will be an increase in substrate concentration in the solution near the protein and the osmolarity will be increased. A difference in substrate concentration and hence osmolarity (Δosmolarity) will therefore be set up across the membrane as an indirect result of the substrate transport. The osmotic gradient will pull water in the same direction as the substrate transport (blue arrow). Clearly, low diffusion coefficients in the solutions will give rise to higher osmotic gradients, i.e., a higher Δosmolarity across the membrane. High passive water permeability of the membrane will also make unstirred layer effects more efficient.

How to distinguish between cotransport of water and unstirred layer effects. (a) Cotransport of water and unstirred layer effects differ in the speed at which they respond to substrate application. If substrate is added at t = 0, the cotransport model predicts an immediate influx of water and cell swelling (vertical broken line); the molecular coupling between substrate and water is indicated by a black triangle. In contrast, the unstirred layer model predicts an initial period of no swelling, because it depends on the intracellular osmolarity to build up. In the Xenopus oocyte, it usually takes about 20 seconds before the intracellular osmolarity increases sufficiently to generate measurable influxes of water (vertical broken line). (b) This panel shows water fluxes induced in the Na⁺/K⁺/2Cl⁻ cotransporter (NKCC1) when activated by increasing (isosmotically) the K⁺ concentrations of the outside solution by 15 mM (at t = 0). The NKCC1 isoform is known to transport water while the isoform NKCC2 has no inherent capacity for water transport.19,20 It is seen that the water flux induced by NKCC1 can be explained by cotransport of water, while that of NKCC2 is best explained by the unstirred layer model. In the two examples, the rates of cation transport by the two isoforms were the same, data from proteins expressed in Xenopus oocyte.19 (c) Cotransport of water is also observed for the Na⁺‐coupled glucose transporter (SGLT1). Addition of sugar at t = 0, initiates an abrupt influx of water. This is compared with the comparatively minor water fluxes induced by sugar transport in the uniporter GLUT2 and by Na⁺ transport in the ionophore gramicidin. In this example, the combined transport of sugar and Na⁺ transport by GLUT2 and the gramicidin equals that mediated by SGLT1. Despite this, the influx of water maintained by SGLT1 was larger than the sum of the water fluxes maintained by GLUT2 and the gramicidin‐treated oocyte. The water transport observed for SGLT1 is best described by the cotransport model. The water fluxes induced by GLUT2 and the gramicidin are best described by unstirred layer models, data from Xenopus oocyte.21

Uphill water transport by the K⁺/Cl⁻ cotransporter (KCC) in epithelial cells. The volume trace was obtained by ion‐selective microelectrodes in amphibian choroid plexus.8 The gray boxes illustrate the effects of adding 100 mOsm to the extracellular solution, either as 100 mM mannitol, 50 mM NaCl, or 50 mM KCl. In case mannitol or NaCl were added the cell shrank instantly. In contrast, if the osmolarity was increased by adding KCl the cell swelled instantly despite the extracellular osmolarity being 100 mOsm higher than the inside solution. This is a clear example of uphill cotransport of water; importantly, the intracellular concentrations of K⁺ and Cl⁻ increased only by a few millimolar during KCl addition (not shown). If the cells had been treated with 1 mM furosemide (an inhibitor of KCC), addition of 50 mM of KCl caused a shrinkage of the cells comparable with that obtained with mannitol or NaCl as osmolyte.

Cellular water homeostasis as a balance between water pumps and water leaks. In this example, the molecular water pump Na⁺/glucose cotransporter (SGLT1) is energized by the Na⁺ gradient and the K⁺/Cl⁻ cotransporter (KCC) water pump by the K⁺ gradient, the cationic gradients being maintained by the Na⁺/K⁺ATPase. The single blue arrow represents the passive osmotic water leak and could result from aquaporins, other water permeable proteins, and the lipid bilayer. If the transporters were placed asymmetrically, transcellular water transport would follow as observed for epithelia, in the present example from left to right. In the kidney proximal tubule the water leak would mainly result from aquaporins (AQP). The small intestine, however, has very few aquaporins, and the leak would result from the lipid and other membrane proteins.