Cotransport of Water in the Choroid Plexus Epithelium: From Amphibians to Mammals

Abstract

Transport by the choroid plexus epithelium is central for cerebrospinal fluid formation. The final steps take place across the brush border membranes, and the availability of these membranes allows detailed studies of relevant transport mechanisms. Here we review data from amphibians and relate them to mammals. Osmotic water permeabilities have been determined in amphibians with high precision by means of microelectrodes. No significant unstirred layers were observed: this accords with vesicle studies of both amphibian and mammalian membranes. Simple osmotic mechanisms do not explain fluid transport in epithelia adequately; the water permeabilities are not large enough. Furthermore, osmotic models cannot describe fluid transport against large osmotic gradients of up to half plasma osmolarity. Cotransporter-mediated water transport has been demonstrated in luminal membranes of the choroid plexus epithelium: in amphibia a K⁺/Cl⁻ cotransporter and in mammals a Na⁺/K⁺/2Cl⁻ cotransporter. These cotransporters are water permeable and the water flux can be energized by the cotransport of ions. We suggest that a hyperosmolar compartment within the cotransport-protein is central to this coupling. The water fluxes can proceed against significant osmotic gradients. Models of the choroid plexus epithelium employing cotransporter-mediated water transport give a quantitative description of the properties of the living tissue.

Appendix: Molecular Mechanism of Water Cotransport

It is generally accepted that cotransport proteins of the symport type play a key role for the coupling between ion and water fluxes, although the exact molecular mechanism is not understood. Water follows the other substrates closely in the transport protein: the molar ratio between substrate and water transport is a constant irrespective of whether the transport is driven by concentration, electrical, or osmotic gradients. Cotransport of water has been described for a number of different cotransporters by a variety of techniques, for reviews see Zeuthen (2010), Zeuthen and MacAulay (2012b). The present review deals mainly with the K⁺/Cl⁻ cotransporter KCC, but the phenomenon has been established in several other cotransporters, such as the Na⁺/K⁺/2Cl⁻ (isoform NKCC1), the SGLT1, and the H⁺/lactate transporter (Zeuthen et al. 1996). Interestingly, the isoform NKCC2 did not cotransport water (Hamann et al. 2005; Zeuthen and MacAulay 2012a). Many of the cotransporters have been studied heterologeously expressed in Xenopus oocytes which enables the effects of site directed mutations to be investigated (Zeuthen et al. 2016).

To understand water cotransport, the structure of each conformational states of the same protein must be known. Furthermore, the transitions between these states must be described kinetically. At present, a complete set of data for one protein is not available and a putative model can only be set up piecing together information from different cotransporters. In broad terms, transport is maintained by the transition between two conformational states, an outwards facing and an inward facing, the so called alternating access model (Mitchell 1957). In the first state the substrate gains access to the binding site from the outer solution via an aqueous cavity (Fig. 4.7a). This is followed by the occlusion of the substrates and some water molecules (b). Finally, in the inward open state (c) the substrate and the occluded water gain access to the inner solution via an aqueous cavity. After the substrate has left the protein, the empty transporter may attain a closed conformation from which it returns to the inward-open conformation (not shown). It is generally held that the coupling mechanism is closely associated with the exit step, four models have been suggested, d1–d4.

Fig. 4.7

Molecular mechanisms of water cotransport. During transport the cotransport protein changes conformation. In the outward facing confirmation (a) the binding site(s) is available from the outer solution via an aqueous cavity. This is followed by an occluded conformation (b) in which the substrate and some water molecules are internalized. In the inward facing conformation (c), the substrate and water gain access to the internal solution. It is generally held that the mechanism of water cotransport is associated with the exit step and various mechanisms have been suggested: (d1) Substrate concentration increases significantly in the external solution abutting the protein (unstirred layers) and water is transported by conventional osmosis through the protein. (d2) The presence of substrate in the exit funnel gives rise to large intramolecular hyperosmolarities, water transport results from osmosis. (d3) When the substrate has left the protein, the exit cavity collapses and water in the cavity is squeezed out. (d4) Water in the protein is pushed through by the substrate. Figure adapted from Zeuthen and MacAulay (2012b)

The first two models (d1 and d2) are essentially osmotic and require an aqueous pathway in the protein. An aqueous pathway in the sodium coupled glucose cotransporter SGLT1 has been described both functionally (Loo et al. 1996, 1999) and structurally (Faham et al. 2008). This aqueous pathway, as well as that in members of other relevant cotransporters from the SSS superfamily (solute-sodium superfamily), has been analyzed in molecular dynamics simulations (Li et al. 2013), which demonstrates opening and closing of the pathway. Furthermore, it has been shown experimentally that water and substrates share a pathway inside the SGLT1 (Zeuthen et al. 2016). In the so-called unstirred layer model (d1) it is assumed that during transport the substrate concentration increases significantly in the solution at the exit side, and that water transport is driven by the accompanying osmotic gradient. The major problem with this model is however, that the diffusion rates in the external solutions are far too high to sustain significant gradients; the substrates diffuse away from the transporter before any osmotic gradient is build up. Even inside cells the diffusion for sugars and smaller substrates are high, about half of those in free solution (Zeuthen and MacAulay 2002). This problem is circumvented in the hyperosmolar-cavity model (d2). Here the hyperosmolarity builds up inside the exit cavity of the protein. As long as the substrate is held in here in a thermodynamically free state, there will be an osmotic driving force. Water will finally exit the protein driven by an intramolecular hydrostatic pressure. Given reasonable parameters for the sugar cotransporter SGLT1 and the monoport GLUT, this model predicts quantitatively several experimental results (Naftalin 2008). The water transport will depend on how long the substrate is present inside the cavity before entering the outer solution. In this context, it is interesting that the state in which the substrate leaves the cotransporter has been shown to be the rate limiting step, for a review on the SGLT see Wright et al. (2011). This might explain why hyperosmolarities in the exit solutions accelerates the rate of water cotransport (Zeuthen 1994). If a large inert osmolyte, say mannitol, was permanently present in the exit solution, the water remaining in the exit cavity after the substrate has left would be under osmotic stress. This would speed up its closure and hence the return of the protein to its outward open conformation ready for a new transport cycle. In fact, the experimental evidence for water cotransport could be explained by a combination of osmo-sensitive kinetics and a hyperosmolar-cavity model.

The last two models do not employ osmosis as such. In the Occlusion model (d3) it is assumed that a given amount of water is occluded together with the substrate inside the protein. This water is transferred with the substrate to the exit cavity and squeezed out as the cavity collapses. However, in the crystal structures determined so far, the occluded state are found to contain far less than the 90 water molecules required by this model. It should be noted however, that the transport experiments were performed on proteins from amphibians or mammals, while the structures were determined in bacterial proteins. It remains to be seen how much water can be held in cotransporters from vertebrates. In the Brownian piston model (d4) the water is pushed through the protein (Zeuthen 1996). The idea has been tested by molecular dynamics simulations lasting 200 ns (Choe et al. 2010). It could be argued that this is too short to give a realistic picture of the transport process which may last around 10 ms.

This analysis shows that it is possible to set up a working model for cotransport of water based on well-established experimental and structural findings. At present the hyperosmolar-cavity model (d2) holds most promise. It describes more experimental findings and conflicts with fewer facts than the other models. A number of questions for future investigation present themselves: For example, is there a pathway for water in the protein throughout the whole transport cycle? Can thermodynamics describe the substrate-created hyperosmolarity in the exit funnel; how high is the osmolarity and how many water molecules are transported osmotically through the protein in this period? How much water is occluded in mammalian proteins? Will a difference in substrate affinity between the entry and the exit side (Eskandari et al. 2005) lead to a preferred direction of water transport?

Transporters from the cation chloride cotransporters (CCC) differ from those of the SSS family. For both families a water flux is induced when the transporter is exposed to a transmembrane osmotic gradient. But for the CCC-cotransporters the water fluxes are associated with fluxes of ions and have high activation energies (Hamann et al. 2005). This contrasts with the SSS proteins which have a channel-like aqueous pathway with low activation energy. Functionally at least, this pathway works in parallel with the Na⁺/glucose/water cotransport system. It would appear that the CCC transporters are more tightly coupled to water than the SSS transporters.