Epithelial water absorption: Osmosis or cotransport?

A paradigm lost? Not according to the studies of Duquette, Bissonnette, and Lapointe (1), which present a direct test of the recently advanced cotransport hypothesis of water absorption. The results of these studies argue, strongly, that the widely accepted notion that osmotic coupling is the single mechanism responsible for water transport across biological membranes, in general, and epithelia, in particular, is secure and that the recent challenges to this notion can be dismissed.

The Road to the Present Paradigm.

Approximately a century ago, E. Waymouth Ried (2) reported that a number of epithelia are capable of absorbing fluid in the absence of an external, osmotic, driving force when viable, but that this ability is lost following procedures that destroy viability. Years later, this pioneering finding was confirmed and extended by Curran and Solomon (3) who found that isosmotic reduction of the NaCl concentration of solutions perfusing rat small intestine by replacement with nonabsorbable solutes resulted in a proportional decrease in the rate of fluid absorption, and that fluid absorption remained isosmotic even when the NaCl concentration in the lumen was lower than that in plasma. Thus, it was NaCl that was being absorbed “uphill” and water followed “passively” in isosmotic proportion. Similar conclusions could be drawn from the results of studies carried out by Windhager et al. (4) on renal proximal tubule of the amphibian Necturus maculosa. Later, Diamond (5) demonstrated that rabbit gallbladder was capable of transporting a fluid isotonic with that in the lumen when the latter was varied over an 8-fold range.

These findings, and many similar findings by others, suggested that water absorption by epithelia is the result of intraepithelial osmotic coupling to the primary flow of solutes; i.e., solute transport generates an intraepithelial region of hypertonicity that serves as the osmotic driving force that energizes transepithelial water flow in the absence of, or even against, external differences in osmolarity. Over the years, several models have been proposed to account for isotonic water absorption and have recently been reviewed elsewhere (6–8). Suffice it to say in this brief commentary that they all share the essential features illustrated in Fig. 1. In essence, the entry of solutes into the cell across the apical membrane establishes an osmotic difference across that barrier that energizes the entry of water; the subsequent extrusion of these solutes across the basolateral membranes establishes a region(s) of hyperosmolarity in the lateral interspaces and/or subepithelial spaces that energizes the exit of water. The transmembrane movements of water take place through the lipid bilayer itself and/or through specialized pores termed aquaporins (9, 10). The increases in the rates of water absorption by small intestine and renal proximal tubule by sugars and amino acids, which enter the cell across the apical membrane coupled to the entry of Na and exit by means of facilitated diffusion mechanisms, could be readily explained by this model in which water absorption is independent of the nature of the transported solute and traverses a parallel pathway across the two limiting cell membranes.

Figure 1.

Illustration of the basic tenets of the osmotic model of solute-coupled fluid absorption. Solute enters and exits the cells through specific pathways; in this illustration Na-coupled entry mechanisms (e.g., SGLT1) and facilitated/equilibrating carrier exit mechanisms (e.g., GLUT2) plus the Na-K ATPase. Water follows through separate pathways driven by osmotic differences. These flows are shown taking place in two different cells for convenience. B.M., basement membrane; CAP., capillary.

Although an osmotic-coupling model such as that shown in Fig. 1 is attractive for its simplicity and the ease with which it can be generalized to all solute and water absorbing epithelia, the underlying, essential features have proven to be extremely difficult to verify experimentally. In particular, because the basolateral membranes appear to have very high hydraulic conductivities (i.e., osmotic permeabilities to water or L_p; ref. 11), the increase in osmolarity in the lateral and subepithelial spaces needed to drive high rates of water absorption is very small. For obvious reasons, measuring the osmolarity in the lateral intercellular spaces is no easy task. Xia et al. (12) have estimated that this increase in osmolarity is less than 15 mM for MDCK cells grown on permeable supports, but there are no reliable estimates for small intestine or renal proximal tubule. Nonetheless, despite these unresolved issues, the osmotic coupling model of water absorption has become, in Thomas Kuhn's words (13) the “normal science”—accepted by the scientific community, incorporated into textbooks, taught in the classroom, and essentially unchallenged until the early 1990s.

The Appearance of Anomalies.

The notion of cotransport of salt and water and secondary active transport of water was first put forth by Zeuthen (14, 15) as a result of his studies on Necturus choroid plexus. This notion was further elaborated on by Wright and his collaborators (16–18) in a series of elegant studies in which the Na-coupled glucose carrier, SLGT1, found in small intestine and renal proximal tubule was expressed in Xenopus laevis oocytes. These investigators found that simply expressing the protein in the oocyte markedly increased the osmotic permeability, L_p, of the membrane and that this increase could be inhibited by phlorizin; inasmuch as this agent is a potent, competitive inhibitor of the carrier this finding suggests that the pathway for water flow is through the pathway for sugar transport. But, perhaps more striking, these investigators also found that activating the carrier by the addition of sugar brought about an apparently immediate increase in oocyte volume that continued for many minutes and paralleled the rate of Na-coupled sugar uptake by the oocyte; inactivation of the carrier with phlorizin after a brief (15 sec) period of activity resulted in the abrupt, simultaneous inhibition of both Na-sugar cotransport and volume increase. The rate of increase in volume corresponded to the uptake of some 200–260 water molecules for each sugar molecule coupled to two Na ions or roughly 70–90 water molecules per solute molecule. In addition, this stoichiometry appeared to be invariant in the face of different Na concentrations, sugar concentrations, clamping potentials, bath osmolarities, etc. Furthermore, an increase in oocyte volume could take place against an adverse osmolarity of 10 mM (17), suggesting “a water pump” or secondary active transport of water driven by the electrochemical potential for Na⁺ across the oocyte membrane.

Finally, it appears that a number of Na-coupled cotransporters share the same properties when overexpressed in Xenopus oocytes. These include the renal Na-dicarboxylate cotransporter (NaDC-1; ref. 19), the rat Na-iodide cotransporter (rNIS), and the human Na-Cl-GABA cotransporter (hGAT1; ref. 20). Furthermore, the water pathway traversing all of these cotransporters is also permeable to urea, and their activation by the appropriate substrate promotes urea absorption as well as water absorption. The rate of urea absorption is proportional to the ratio of the molar concentration of urea to that of water (21).

As pointed out by Wright and his collaborators (16, 17) and Diamond (22) the presence of water pumps would dramatically alter our understanding of the mechanisms responsible for intestinal water absorption in response to nutrients under physiological and nonphysiological conditions (e.g., oral rehydration therapy), and a large fraction of renal proximal tubule water reabsorption under physiological conditions.

Toward Resolution of the Anomaly and Support of the Paradigm.

Two important findings have recently surfaced that provide strong reassurance in the now classic osmotic-coupling paradigm.

The first comes from the laboratories of Verkman and his collaborators (23) and serves to highlight the essential role of water channels, or aquaporins, in water reabsorption by renal proximal tubule. In 1976, Burg et al. (24) demonstrated that a large fraction of the fluid absorbed by perfused isolated rabbit proximal tubule depends on the presence of glucose and alanine in the luminal perfusate. Employing aquaporin-1-knockout mice, Verkman and his collaborators demonstrated that elimination of these water channels, when glucose and alanine are present in the luminal perfusate, resulted in an 80% decrease in the osmotic water permeability of perfused proximal tubule and a 50% decrease in net fluid absorption determined in both perfused tubules and in vivo micropuncture studies of this segment of the nephron. Subsequent studies (25) demonstrated that the luminal fluid of aquaporin-1-knockout mice becomes markedly hypotonic with respect to plasma and suggested that this increased osmotic gradient provides the driving force for water reabsorption through the lipid bilayer, thereby reconciling an 80% decrease in osmotic permeability with only a 50% decrease in net fluid reabsorption. These investigators concluded that there is no evidence of a pathway for transcellular water transport in renal proximal tubule other than water channels and the lipid bilayer.

Whereas the results reported by Verkman and his colleagues question the physiological significance and necessity for the cotransport model suggested by the Wright laboratory, the results reported by Duquette et al. (1) question the data suggesting the very existence of water cotransport. These investigators followed a similar approach to that used by the Wright laboratory. They confirmed that expressing SGLT1 in Xenopus oocytes resulted in a significant increase in oocyte water permeability that could be abolished by phlorizin. They then found that exposure of these oocytes to α-methylglucose (αMG) resulted in oocyte swelling that reached a steady-state by 5–8 min, at which time the stoichiometry was 815 molecules of water per sugar molecule transported, close to that expected for isosmotic uptake. When phlorizin was rapidly added after 15 min of αMG uptake, the current generated by the cotransporter abruptly vanished but swelling persisted (through the endogenous water leak pathway), indicating that αMG uptake accompanied by Na and water was not isotonic and that a considerable amount of solute had accumulated within the oocyte by that time. By using the information gained from studies of this type and a simple relation relating the total water flow to the flows through the endogenous leak pathway and the pathway resulting from SGLT1 expression alone and the flow due to cotransport, the investigators demonstrate that the total water uptake can be attributed to flow through the two leak pathways and that the contribution from cotransport is negligible. A similar conclusion could be drawn from the results of experiments in which αMG uptake was abruptly abolished after a brief (60 s) exposure by changing the membrane potential from −100 mV to 0 mV; the current generated by the cotransport process rapidly decreased by 94%, whereas the rate of volume increase was reduced by only 25%.

Finally, Duquette et al. (1) demonstrate that their data can be nicely fit by the predictions of a simple model for solute diffusion within the oocyte after entry across the membrane, leading to the conclusion that water uptake can be entirely attributed to osmotic flow driven by an αMG transport-dependent region of hypertonicity that is developed in an intracellular unstirred layer below the membrane. It has long been appreciated that a major problem complicating the study of osmotic water flow across membranes is the possibility of local osmotic phenomena arising from the presence of unstirred layers. The Wright group was certainly aware of this bug-a-boo and attempted to rule it out as a serious complication (16). However, they used an equation that is intended for a flat surface and, further, assumed that the intracellular diffusion coefficient for the sugar is equal to that in free solution. The approach used by Duquette et al. (1) seems more appropriate and suggests that the intracellular diffusion coefficient is only one-fourth that used by Loo et al. (16).

The conclusion that water flow is simply the result of passive flow driven by an osmotic gradient raises the possibility that the urea transport observed by Loo et al. (21) can be attributed to solvent-drag, but this suggestion is entirely conjectural.

Thus, Duquette et al. (1) present direct, compelling evidence that what was previously interpreted as cotransport is nothing more than osmotic flow through passive leak pathways into a hypertonic unstirred layer; such a process can also account for what appeared to be “secondary active transport” against an “apparently” adverse osmotic gradient. If these findings are confirmed the challenge posed to the “normal science” will have been thwarted and osmotic coupling will retain its position as the singular mechanism responsible for water absorption by epithelia.