Does aquaporin-1 pass gas? An opposing view

The possibility that aquaporin-1 (AQP1) facilitates CO₂ transport across biological membranes is an intriguing one with potentially numerous consequences in mammalian physiology. However, such an unorthodox conjecture must pass tests of physical reasonableness and be validated by incisive yet simple-to-understand experimental methods. The conjecture that AQP1 passes CO₂ is unorthodox because biomembranes are in general highly gas permeable, so much so that unless extraordinary experimental approaches are used CO₂ permeability is limited not by the intrinsic membrane permeability to CO₂ but by ‘unstirred layers’ outside of the membrane (Geers & Gros, 2000).

Can CO₂ pass through AQP1, and if so, can the increase in intrinsic membrane CO₂ permeability conferred by AQP1 significantly augment CO₂ transport in real tissues? It is not possible on theoretical grounds to exclude the possibility that some CO₂ can pass through AQP1 or other proteins. Recent molecular dynamics simulations of water movement through AQP1 suggest, however, that water molecules twist and turn as they wind through the narrow, single-file pore in AQP1 monomers. Whether the larger CO₂ molecule can do so is unknown. Of greater theoretical concern is that membranes where AQP1 is expressed – erythrocytes, lung endothelia and kidney proximal tubule – are highly CO₂ permeable. CO₂ transport is probably rate limited in these tissues by unstirred layers, and so is unlikely to increase even if AQP1 were infinitely permeable to CO₂.

What about the experimental evidence to support or refute the idea that AQP1 is a CO₂ channel? As detailed in the accompanying review in this issue of The Journal of Physiology by Boron and colleagues (Cooper et al. 2002), the principal evidence is that CO₂ permeability in Xenopus oocytes expressing large amounts of AQP1 is ∼40 % greater than that of control oocytes (Nakhoul et al. 1998). These are technically complex experiments in which oocytes are first microinjected with cRNA. At a later time they are microinjected with carbonic anhydrase to catalyse CO₂/ HCO₃⁻/pH equilibration. The oocytes are then impaled for a third time with a pH-sensing microelectrode to measure cytoplasmic acidification in response to addition of CO₂ to the bathing solution. In a different protocol designed to test whether CO₂ permeability through AQP1 could enhance NBC-mediated Na⁺/HCO₃⁻ cotransport by locally shunting CO₂, zinc-dependent oocyte acidification was measured after microinjection with a mixture of cRNAs encoding NBC, two carbonic anhydrase isoforms and AQP1. The results of this complex study were interpreted in terms of AQP1-dependent CO₂ shunting, however the possibility of NBC-adjacent CO₂ transport seems unlikely because CO₂ diffuses rapidly in cytoplasm and there is no evidence for physical coupling of NBC and AQP1, which are probably expressed in very different amounts.

As detailed in Yang et al. (2000) and in a new study reported in this issue of The Journal of Physiology (Fang et al. 2002), we investigated whether CO₂ transport by AQP1 could be detected under physiological and super-physiological conditions in erythrocytes, lung, kidney and reconstituted proteoliposomes. Organ studies were done comparing normal mice and transgenic mice lacking AQP1. Carbon dioxide permeability in erythrocytes and kidney proximal tubule membranes was very fast (equilibration time a few milliseconds) and was not reduced in AQP1 knockout mice. CO₂ permeability in AQP1-reconstituted proteoliposomes was not different from that in control liposomes, even though water permeability was >100-fold different. In each of these systems intrinsic membrane CO₂ permeabilities should be even higher due to unstirred layer effects. CO₂ blow-off by the lung was not AQP1-dependent under physiological conditions in living mice, and airspace-capillary CO₂ permeability was not affected by AQP1 deletion under super-physiological conditions in a rapidly perfused lung preparation.

From these considerations and experimental results it appears unlikely that CO₂ transport by AQP1, if it occurs, is important in mammalian physiology. However, the possibility that small gases such as CO₂, NH₃ and NO are transported by proteins should not be ignored. The relatively non-selective aquaglyceroporins, or other abundant proteins such as Rh factors, should be evaluated for their gas transporting ability.