The aquaporin family of molecular water channels.

The aquaporins (AQPs) are a family of intrinsic membrane proteins that function as water-selective channels in the plasma membranes of the cells of many watertransporting tissues (1, 2). The first AQP to be identified, CHIP28 (later called AQP-CHIP), was purified by Agre and his colleagues from human erythrocytes in the late 1980s and its cDNA sequence was reported in 1991 (reviewed in ref. 2). In 1992, the functional identification of AQP-CHIP as a water channel protein was established by complementary RNA expression studies in Xenopus oocytes (3) and by functional reconstitution of water transport activity in liposomes after the incorporation of the purified protein (4). These findings sparked a veritable explosion of work with an impact in several longstanding areas of investigation, such as the biophysics of water permeation across cell membranes, the physiology of fluid transport in the kidney and other organs, and the pathophysiological basis of inherited and acquired disorders of water balance. Significant progress has also been made in the analysis of the structural basis of selective water permeation through the AQPCHIP channel. The identification of AQP-CHIP as a molecular water channel also has spawned the identification of three related AQP water channels (Table 1), which are expressed in mammalian tissues and one of which (AQP3) is reported in this issue by Ishibashi and colleagues (7). Functional Role of Water Channels. In general, the plasma membranes of all animal cells (even those without water channels) are at least moderately permeable to water and can maintain osmotic equilibrium in static external environments (8). Studies in artificial lipid membranes have demonstrated osmotic water permeability (Po) values in excess of 50 pm/s in pure phospholipid membranes (9). Addition ofcholesterol, an important component of natural plasma membranes, lowers Pf. For example, in artificial lipid membranes with a 1:1 ratio of cholesterol/phospholipid, the Pf falls to the range 10-20 ma/s, a low level but nonetheless theoretically sufficient to maintain osmotic equilibrium in most cells without the aid of water channels. However, the water permeabilities of many cell types have been found to exceed values that can be explained by lipid-phase water permeation alone. For example, the initial portion of the descending limb of Henle's loop in the chinchilla kidney has an epithelial Pf value of 1500-2500 pm/s due to the fact that the plasma membranes of this epithelium contain large amounts of AQPCHIP (10). In contrast, the distal portion of the descending limb in the chinchilla has a much lower Pf value (68 pm/s), corresponding with an absence of detectable AQP-CHIP (10). The extremely high water permeabilities that result from water channel expression in a variety of tissues subserve a number of physiological roles. (i) Water channels can raise plasma membrane water permeability to the levels required for efficient coupling between NaCl transport and water transport in epithelia that carry out isosmotic fluid transport (e.g., AQP-CHIP in renal proximal tubule and choroid plexus). (ii) Water channels allow rapid osmotically driven water transport in epithelial and endothelial cells of the renal medulla, which is critical for the countercurrent processes that concentrate the urine. (iii) Water channels provide a target for regulation of water transport [e.g., AQPCD, the target for vasopressin-mediated regulation of water reabsorption in the renal collecting duct (CD)]. Biophysical Basis of the Water-Selective Channel Concept. The concept ofa water pore or channel was originated by Koefoed-Johnsen and Ussing in 1953 (11) to explain the observation that osmotic water permeability is higher than diffusional water permeability in frog skins. The increase in Pf brought about by antidiuretic hormone exposure in this epithelium was explained as a widening ofpore diameter, an event that was assumed to explain antidiuretic hormone-induced solute fluxes. Subsequently, numerous other biophysical investigations solidified the water pore concept but raised doubts about the ability of small solutes to penetrate water pores. An important step forward was made by Macey and Farmer (12), who demonstrated that water (molecular radius, 1.5 A) and the small solute urea (molecular radius, 2.0 A) penetrate the plasma membrane of erythrocytes by independent pathways. They observed that the organomercurial sulfliydryl-reactive reagent p-chloromercuribenzenesulfonate (pCMBS) markedly decreased erythrocyte Pf (from w200 plm/s to 20 ,pm/s), consistent with the elimination of an aqueous channel. pCMBS inhibited urea and glycerol permeability in addition to water permeability. However, another agent, phloretin [,8-(p-hydroxyphenyl)phloropropriophenone], strongly inhibited ureaand glycerol transport without affecting water permeability. From these and numerous other biophysical studies in erythrocytes and other cell types, the view emerged that water and small nonelectrolytes can traverse plasma membranes via specialized transporters (presumably proteinaceous in nature) whose selectivity is based not on molecular size alone but also on other physical properties of the transported substrates. Thus, based on knowledge gained from biophysical studies, it was predictable that independent selective transporters for water and urea would be identified. The prediction has come to fruition in the past 2 years with the identification offour mammalian water channels (all expressed in the renal medulla) and a renal medullary urea transporter (Table 1). Tissue Distribution and Functional Roles of AQPs. Table 1 summarizes several key characteristics of the known AQPs and compares them to those of the vasopressin-regulated urea transporter UT-2. AQP-CHIP is the major water channel of the erythrocyte plasma membrane and is heavily expressed at sites of constitutively rapid water transport in the kidney (proximal tubule and the descendinglimb ofHenle's loop) (13, 14). In addition, AQP-CHIP is heavily expressed in epithelia and endothelia thought to be involved in fluid transport at diverse sites throughout the body (1517). AQP-CD (5) is the vasopressinregulated water channel of the renal CD and thus has a critical function in the regulation ofwater excretion. It does not appear to be expressed outside the kidney. Mercurial-insensitive water channel (MIWC) is expressed in the renal medulla (perhaps in the medullary vasculature) and in the lung (6). Its physiological role is as yet undefined. AQP3, reported in this issue (7), is expressed in the CD of