Transport of volatile solutes through AQP1

doi:10.1113/jphysiol.2002.023218

Transport of volatile solutes through AQP1

Gordon J. Cooper, Yuehan Zhou, Patrice Bouyer, Irina I. Grichtchenko and Walter F. Boron

Department of Cellular and Molecular Physiology, Yale University School of Medicine, New Haven, CT 06520, USA

ABSTRACT

Top
Abstract
Introduction
Methods
Results
Discussion
References

For almost a century it was generally assumed that the lipid phases of all biological membranes are freely permeable to gases. However, recent observations challenge this dogma. The apical membranes of epithelial cells exposed to hostile environments, such as gastric glands, have no demonstrable permeability to the gases CO₂ and NH₃. Additionally, the water channel protein aquaporin 1 (AQP1), expressed at high levels in erythrocytes, can increase membrane CO₂ permeability when expressed in Xenopus oocytes. Similarly, nodulin-26, which is closely related to AQP1, can act as a conduit for NH₃. A key question is whether aquaporins, which are abundant in virtually every tissue that transports O₂ and CO₂ at high levels, ever play a physiologically significant role in the transport of small volatile molecules. Preliminary data are consistent with the hypothesis that AQP1 enhances the reabsorption of HCO₃^- by the renal proximal tubule by increasing the CO₂ permeability of the apical membrane. Other preliminary data on Xenopus oocytes heterologously expressing the electrogenic Na⁺-HCO₃^- cotransporter (NBC), AQP1 and carbonic anhydrases are consistent with the hypothesis that the macroscopic cotransport of Na⁺ plus two HCO₃^- occurs as NBC transports Na⁺ plus CO₃^2- and AQP1 transports CO₂ and H₂O. Although data - obtained on AQP1 reconstituted into liposomes or on materials from AQP1 knockout mice - appear inconsistent with the model that AQP1 mediates substantial CO₂ transport in certain preparations, the existence of unstirred layers or perfusion-limited conditions may have masked the contribution of AQP1 to CO₂ permeability.

(Received 24 April 2002; accepted after revision 28 May 2002)
Corresponding author W. F. Boron: Department of Cellular and Molecular Physiology, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06520-8026, USA. Email: walter.boron{at}yale.edu

INTRODUCTION

Top
Abstract
Introduction
Methods
Results
Discussion
References

Channels permeable to small, neutral molecules

Overton's rule

The pioneering work by Overton at the end of the 19th century, provided many insights into the processes by which small neutral solutes cross cell membranes. The major finding of this work was that the permeability coefficient of a solute is linearly related to its oil-water partition coefficient, a principle later adopted as Overton's rule. However there are deviations from this relationship, with small solutes having higher permeabilities than predicted (Walter & Gutknecht, 1986). This increase in permeability is inversely related to the molecular volume of the solute. Further refinements led to the proposal of the 'solubility-diffusion' model (Finkelstein, 1986), which states that in order for a solute to cross a membrane it must enter the interfacial hydrocarbon region before diffusing through the membrane to reappear on the other side. Thus, this solubility-diffusion model predicts that not only will the oil-water partition coefficient having a bearing on permeability, but also the ease with which a solute moves through the membrane hydrocarbon. The application of these principles to the movement of dissolved gases across cell membranes has led to the traditional view that gases are freely permeant across cell membranes. However several recent observations have challenge this dogma. Not only has it become clear that some membranes can have extremely low permeability to gases, but the movement of some gases across membranes is not governed by the 'solubility-diffusion' model. Additionally, it appears that specific channels can augment the movement of some gases across some membranes.

Evidence for water pores

The first small, neutral molecule recognized as violating Overton's rule was water. In 1935, Krogh and coworkers found that upon exposing amphibian skin to a transepithelial osmotic gradient, the H₂O flux is 3- to 5-fold larger than predicted if H₂O were simply diffusing through lipid (Hevesy et al. 1935). The authors concluded that the membranes have two pathways, a non-watery pathway accounting for diffusional H₂O flux and a watery pathway consisting of pores that permit the osmotic flow of H₂O. In 1953, Koefoed-Johnsen & Ussing showed that a posterior-pituitary hormone extract increases the osmotic H₂O permeability (i.e. P_H2O) of amphibian skin by 100 % to 200 %, without substantially increasing the diffusional flux. They suggested that the increased P_H2O reflected an increase in the diameter of the H₂O-filled pore. H₂O-filled pores have been proposed for many cell types, including amoebae and oocytes from frog and zebra fish (Prescott & Zeuthen, 1953). However, trout eggs have a P_H2O of zero, implying a membrane devoid of pores. In a study on human erythrocytes in 1957, Paganelli & Solomon found a P_H2O at least 30-fold greater than for any other cell type tested at the time. The ratio of osmotic to diffusional H₂O permeabilities indicated the presence of H₂O pores that constituted 0.1-1 % of the membrane surface, and had a diameter of 3.5Å.

Identifying the water pore - the aquaporins (AQPs)

Whilst purifying a 32-kDa 'Rh' polypeptide from human red blood cells (RBCs), Preston & Agre (1991) noted that a 28-kDa protein co-precipitated at several stages. Based on the N-terminal sequence of this 28-kDa protein, they designed degenerate primers that yielded a promising PCR product. Using this PCR product to screen a human bone-marrow cDNA library, they isolated a full-length clone that encodes 269 amino acids. Their clone, originally named CHIP28 (for channel-forming integral protein), is homologous to MIP-26, which encodes the 26-kDa major intrinsic protein (263 amino acids) from the ocular lens, and the bacterial glycerol facilitator (GlpF; Sweet et al. 1990). The function of MIP-26, which had been cloned in 1984 (Gorin et al. 1984), is still unknown, although disruption of this protein in the lens fibre membrane is associated with cataract formation (Shiels & Bassnett, 1996).

The major advance in understanding the function of the MIP gene family was the discovery that expressing CHIP28 in oocytes markedly increases P_H2O (Preston et al. 1992), and CHIP28 has now been designated aquaporin 1 (AQP1). Aquaporin 1 is expressed at high levels in the proximal tubule of the kidney and erythrocytes. The high water permeability of the erythrocytes is greatly reduced by mercurial agents. The AQP1-induced increase in oocyte permeability is also sensitive to Hg²⁺, leading to the conclusion that AQP1 accounts for the high water permeability of these tissues. By selectively mutating the cysteine residues in AQP1 to serine residues, Preston et al. (1993) were able to identify one residue, Cysteine 189, which is responsible for the mercurial sensitivity of AQP1. Oocytes expressing the C189S mutant have the same high water permeability as oocytes expressing the wild-type channel, but this increase in H₂O permeability is no longer reduced by Hg²⁺ (Preston et al. 1993).

Well over 100 members of the MIP family have now been identified. In terms of function these proteins fall into two classes - the aquaporins, which are primarily permeable to H₂O, and the aquaglyceroporins, which transport other small solutes as well as H₂O. The properties and distribution of the MIP family have been reviewed in depth (Schäffner, 1998; Heymann et al. 1998; Engel et al. 2000; Nielsen et al. 2002). A defining characteristic of members of the AQP family is that the N- and C-terminal halves of the molecules are homologous, presumably the result of an ancient gene duplication event. Moreover, each half has an NPA (Asn-Pro-Ala) motif. In 1994 Jung et al. put forward the hourglass model (Fig. 1), which suggests that AQP1 consists of six transmembrane spans and that the two NPA motifs in the middle of loops dip into the membrane and contribute to the formation of the channel pore. Moreover, the Hg²⁺-sensitive C189 would lie near the entrance to the channel pore. This model has aged well, and the atomic structure of AQP1, solved by high-resolution electron microscopy (Murata et al. 2000), confirms these predictions (Nielsen & Agre, 1995). The structural data also demonstrate the presence of the water pore with a diameter of about 3Å at the narrowest point.

View larger version
[in this window]
[in a new window]

Figure 1. Model of aquaporin 1
A, schematic representation of the structure of AQP1. The molecule has intracellular amino and carboxy termini, six membrane-spanning segments, and five loops. Loops B and E fold back into the membrane, contain the tandem repeat of the amino-acid sequence NPA. B, the hourglass model predicts that loops B and E overlap in the membrane to allow the formation of a channel pore. (Data from Nielsen et al. 2002.)

Membranes with a low permeability to small neutral molecules

Among the smallest of neutral molecules of biological importance are CO₂ and NH₃. Both are volatile and both cause changes in intracellular pH (pH_i) when crossing the plasma membrane. (The changes in pH associated with transport of these molecules are reviewed by Roos & Boron (1981).) These properties make them ideally suited for probing the permeability of cell membranes.

Thick ascending limb

The thick ascending limb of Henle's loop (TAL) reabsorbs NaCl in preference to H₂O, diluting the contents of the tubule lumen. The barrier is the apical membrane, which is impermeable to H₂O and H⁺ (Hebert & Andreoli, 1984). Kikeri et al. (1989), perfusing single TAL segments, found that exposing the basolateral membrane to NH₃/NH₄⁺ elicits the typical pH_i changes (i.e. a pH_i increase due to NH₃ influx, followed by a pH_i decrease due to NH₄⁺ influx), indicating that both NH₃ and NH₄⁺ can cross the membrane. However, exposing the apical membrane to NH₃/NH₄⁺ failed to produce even a transient pH_i increase: pH_i simply fell as the robust apical uptake of the acidic NH₄⁺ dominated pH_i. Whether the apical NH₃ influx was in fact very low could not be ascertained because it was impossible to fully block the NH₄⁺ transport. Nevertheless, this work provided the first hint that a cell membrane might have a low permeability to a volatile molecule (i.e. a low P_NH3).

Gastric gland

The chief and parietal cells of gastric glands experience a hostile environment: pepsin at a pH as low as 0.7. What prevents the glands - which lack a protective mucus coating - from digesting themselves? The answer appears to be an extraordinary apical membrane that has no demonstrable permeability to H⁺, nor CO₂ or HCO₃^-, nor to NH₃ or NH₄⁺. Waisbren et al. (1994a,b) microperfused single hand-dissected gastric glands as if they were kidney tubules (Burg et al. 1966). Exposing the basolateral solution ('bath') to pH 6.4 produced a large pH_i decrease, whereas exposing the lumen to pH 1.4 caused none (Fig. 2A). Similarly, exposing the bath to 20 mM NH₃/NH₄⁺ at pH 7.4 increased pH_i, whereas exposing the lumen to 20 mM NH₃/NH₄⁺ at pH 8.0 caused no pH_i change, even though the luminal [NH₃] was ~4-fold greater than bath [NH₃] (Fig. 2B). Finally, exposing the bath to 1 % CO₂/5 mM HCO₃^- caused the expected pH_i decrease, indicating that the basolateral membrane is CO₂ permeable (Fig. 2C). However, exposing the lumen to even 100 % CO₂/22 mM HCO₃^- produced no change in pH_i. Thus, chief and parietal cells have apical membranes that have an undetectably low permeability to H⁺, NH₃ and CO₂.

View larger version
[in this window]
[in a new window]

Figure 2. Unique permeability properties of the apical membranes of the gastric gland
A, low H⁺ permeability. The experiments in all three panels were performed at 37 °C on isolated perfused gastric glands from the rabbit, and the records are from a single parietal cell. Intracellular pH (pH_i) was monitored using the fluorescent pH-sensitive dye BCECF in conjunction with a digital imaging system. Reducing the pH of the luminal fluid from 7.4 to 1.4 in a CO₂/HCO₃^--free system produced no detectable pH_i change, whereas reducing bath (i.e. basolateral) pH by only 1 pH unit caused a large and rapid fall in pH_i. B, low NH₃/NH₄⁺ permeability. Exposing the lumen to a Hepes-buffered solution containing 20 mM NH₄⁺/NH₃ at pH 8.0 ([NH₃]_o 1.46 mM) produced no detectable pH_i increase, whereas exposing the bath to 20 mM NH₄⁺/NH₃ at pH 7.4 ([NH₃]_o 0.4 mM) produced a rapid pH_i increase followed by the usual series of changes. C, low CO₂/HCO₃^- permeability. Exposing the lumen to a solution equilibrated with 100 % CO₂ and containing 22 mM HCO₃^- (pH 6.1, [CO₂]_o 24 mM) produced no discernable pH_i decrease, whereas exposing the bath to 1 % CO₂/4.4 mM HCO₃^- (pH 7.4, [CO₂]_o 0.24 mM) or 5 % CO₂/22 mM HCO₃^- (pH 7.4, [CO₂]_o 1.2 mM) caused the expected pH_i decreases. The gland was bathed in 200 µM DIDS to eliminate the pH_i recovery from the intracellular acid load that would otherwise have occurred. (In A, data from Waisbren et al. 1994a. In B, data from Waisbren et al. 1994b.)

Colonic crypt

Singh et al. also found that the apical membrane of colonic-crypt cells - which are also exposed to an inhospitable environment - has no detectable permeability to NH₃ (Singh et al. 1995), CO₂ or butyric acid (S. K. Singh & W. F. Boron, unpublished observations).

Xenopus oocytes

Another cell with unusual permeability properties is the Xenopus oocyte. Its inhospitable environment is pond water, which would osmotically lyse ordinary vertebrate cells. The oocyte's NH₄⁺ permeability swamps its NH₃ permeability (Burckhardt & Frömter, 1992; Cougnon et al. 1996). Blocking this NH₄⁺ permeability reveals that P_NH3 permeability is indeed very low (S. K. Singh, A. Cimini, H. Binder & W. F. Boron, unpublished observations), as is the P_H (Gunshin et al. 1997) and, of course, P_H2O. The pattern that emerges is that certain membranes have a low permeability to H₂O, H⁺, NH₃ and CO₂. The oocyte is unusual in that it is reasonably permeable to CO₂, suggesting that it may have a special (e.g. protein-mediated) pathway for CO₂ transport.

Factors influencing permeability to small neutral molecules

So far we have seen that there are membranes with extremely low permeabilities to CO₂ or NH₃. The molecular basis of these unusual permeability properties is unclear. Electron micrographs of gastric-gland apical membranes reveal no mucus or other barrier (Ito, 1987; Waisbren et al. 1994a). What makes these membranes special?

The solubility-diffusion model of membrane permeability states that, in the first step of permeation, the solute must enter the interfacial hydrocarbon region of the membrane. The polar heads of most phospholipids are likely to provide little barrier to H₂O, NH₃ and CO₂ movement across bilayers; the hydrocarbon tail probably is the barrier. Therefore, we expect the permeability of a solute to be linked to how easily the solute dissolves into the membrane's hydrocarbon, and then to the rate at which it diffuses through the hydrocarbon. In turn, the diffusion of solutes through this hydrocarbon-tail region of the membrane is governed by (i) saturation and length of the lipid hydrocarbon tail, and (ii) stabilizing molecules. We can combine these latter two variables in a parameter referred to as membrane fluidity. Thus, if a solute crosses a membrane according to the solubility-diffusion model, then both solubility in the hydrocarbon and membrane fluidity ought to be major determinants of permeability.

The permeability of membranes to H₂O and NH₃ obeys the solubility-diffusion model

The effect of membrane fluidity and asymmetry on the permeability of membranes to H₂O, NH₃ and CO₂ has been studied in depth using membrane vesicles. In the case of H₂O and NH₃, changes in the saturation and length of the hydrocarbon region have the predicted effects on permeability, with a decrease in membrane fluidity decreasing the permeability to H₂O and NH₃. In the same way addition of cholesterol, which stabilizes bilayers by increasing the packing density of the hydrocarbon tails and reducing tail mobility, also reduces membrane permeability to H₂O and NH₃ (Lande et al. 1995).

The permeability of membranes to CO₂ does not obey the solubility-diffusion model

If the movement of CO₂ across membranes obeyed the solubility-diffusion model, one might expect that a major predictor of the CO₂ permeability of different membranes would be the partition coefficient for CO₂ between water and the membrane lipid. For many small non-electrolytes, the partition coefficient varies substantially among different organic solvents. However, in the case of CO₂ the oil-water partition coefficient does not vary greatly amongst various organic solvents (Simon & Gutknecht, 1980). Thus, one might predict that differences in lipid solubility would not underlie differences in the P_CO2 of various membranes.

In two recent studies, the group of Zeidel has examined the effect on P_CO2 of varying membrane fluidity. In the first study, the authors generated liposomes with three different lipid compositions and a wide range of fluidities as determined by fluorescence anisotropy (Lande et al. 1995). Although the P_H2O of these vesicles varied by a factor of 130, there was no difference in CO₂ permeability (Prasad et al. 1998). In the second study the fluidity of vesicles was manipulated either by changing the temperature or introducing praseodymium (Hill et al. 1999). As the temperature of the membrane increases, fluidity increases sharply as the membrane passes the gel to liquid-crystalline phase transition. Praseodymium binds to the phosphate group heads of the phospholipids and reduces their mobility, leading to a decrease in fluidity. Neither of these treatments affected the CO₂ permeability of the vesicles. Taken together these two studies demonstrate that the fluidity of the liposome membrane and hence diffusion of CO₂ through the hydrocarbon portion of the membrane does not limit the permeability to CO₂.

If the solubility-diffusion model does not describe CO₂ diffusion across a liposome membrane, what enables gas-tight membranes, such as the apical membrane of the gastric gland, to exclude CO₂? One possibility is that these membranes incorporate unusual stabilizing molecules that alter the way in which CO₂ diffuses through the hydrocarbon phase of the membrane. For example, the glycolipid cerebroside contributes to the low H₂O permeability of the urinary bladder epithelium (Hicks et al. 1974). Another possibility is that specific intrinsic proteins are introduced into the membrane to help reduce the permeability to CO₂. With regard to membrane water permeability a similar role of membrane stabilization has been suggested for the uroplakin family of proteins (Hicks et al. 1974).

Contribution of proteins to gas permeability

From the above discussion, it is clear that specialized cells can construct membranes that are virtually impermeable to the gases NH₃ and CO₂. However, a question that arises is whether cells can use precise controllable mechanisms to increase gas permeability. Recently, several studies have indicated that the movement of gases through membrane proteins can contribute to membrane gas permeability.

AQP1 is a CO₂ channel

Experiments on oocytes with an intact vitelline membrane

Because of the unusual permeability properties of Xenopus oocytes, we thought that oocytes might be a useful expression system for asking whether proteins influence the apparent membrane permeability to NH₃ and CO₂. Initially working with oocytes having intact vitelline membranes, we found that expressing AQP1 produced a slight (~18 %) but statistically insignificant increase in the rate at which introducing extracellular CO₂ caused pH_i to fall (Nakhoul et al. 1998). Reasoning that CO₂ build-up at the intracellular surface could slow CO₂-induced acidification, we injected a mixture of carbonic anhydrase (CA) I and II protein into Xenopus oocytes and repeated the experiments. Indeed, CA increased the rate of pH_i decrease ~4-fold. Moreover, in oocytes expressing AQP1, the acidification was ~40 % faster than in water-injected oocytes (Fig. 3A). In both groups, ethoxzolamide (a permeant CA inhibitor) reduced the acidification rate to values only slightly above those in oocytes not injected with CA.

View larger version
[in this window]
[in a new window]

Figure 3. Expression of aquaporin 1 increase the permeability of Xenopus oocytes to CO₂
A, experiments on oocytes with vitelline membrane intact. Switching the extracellular buffer from Hepes to 1.5 % CO₂/10 mM HCO₃^- (constant pH_o = 7.5) causes a sustained fall of intracellular pH, the initial rate of which is indicated by the filled bars. The initial rate of pH_i recovery (i.e. alkalinization) caused by removing the CO₂/HCO₃^- solution is summarized by the open bars. Injection of carbonic anhydrase (CA) caused a significant increase in the initial CO₂-induced rate in both water-injected oocytes and those expressing AQP1. Moreover, in CA-injected oocytes, expression of AQP1 significantly increased the CO₂-induced acidification rate. B, experiments on oocytes with vitelline membrane removed, and in absence of injected CA. Each symbol represents data from a separate experiment (i.e. oocyte). There was a linear relationship between the initial rate of CO₂-induced acidification and lysis time (an index of the AQP1 expression level), with acidification rate increasing as lysis time decreased. (In A, data from Nakhoul et al. 1998. In B, data from Cooper & Boron, 1998a.)

The results of this study suggested that AQP1 could have increased the CO₂-induced acidification rate by acting as a CO₂ conduit. However, expression of AQP1 could have speeded the acidification via alternative mechanisms: (i) altering the lipid composition of the cell membrane (e.g. increasing the proportion of short-chain fatty acids) and thereby increased P_CO2 through the membrane's lipid phase; (ii) causing overexpression of a native 'gas' channel; or (iii) increased the effectiveness of CA.

Experiments on de-vitellinized oocytes

We found that, with the vitelline membrane dissected away from the outer surface of oocytes, AQP1 (in the absence of CA) measurably increased the rate of CO₂-induced acidification (Cooper & Boron, 1998a). Moreover, the acidification rate increased in parallel with APQ1 expression, as judged by the lysis time of oocytes placed in deionized water (Fig. 3B).

If CO₂ passes through AQP1, then mercurial agents that block AQP1's water permeability (Preston et al. 1992) might inhibit the CO₂ permeability as well. As shown in Fig. 4A the mercurial p-chloromercuribenzenesulfonate (pCMBS) causes only a slight slowing of the slow CO₂-induced acidification of a water-injected oocyte. The pCMBS record was made in the same oocyte as the control record, but after a 15-min pre-incubation in 1 mM pCMBS. In an oocyte expressing AQP1 (Fig. 4B), the acidification during the control CO₂ pulse (no inhibitor) was substantially faster than the control pulse in Fig. 4A. Moreover, in this AQP1-expressing oocyte, pCMBS substantially slowed the paired CO₂-induced acidification (Fig. 4B).

View larger version
[in this window]
[in a new window]

Figure 4. Effect of pCMBS on the CO₂-induced acidification
A, H₂O-injected control oocyte. Two records represent paired data from an experiment on a single oocyte exposed to 1.5 % CO₂/10 mM HCO₃^- before and after a 15-min exposure to 1 mM pCMBS. B, AQP1-expressing oocyte. C, C189S-expressing oocyte. D, change in CO₂-induced acidification rate produced by pCMBS. Each bar represents the mean of paired differences, before and after treating the oocyte with pCMBS. (Reproduced from Cooper & Boron, 1998a with permission of the American Physiological Society.)

Because it is possible that the pCMBS exerted its effects by acting on a molecule whose activity was upregulated by AQP1 expression, we examined an AQP1 mutant whose water permeability is mercurial insensitive - the C189S mutant, in which the Cys at position 189 is replaced by Ser (Preston et al. 1993). When expressed in oocytes, AQP1-C189S mediates a normal CO₂-induced acidification. However, treatment with pCMBS only slightly reduces the CO₂-induced acidification rate - as is the case in water-injected oocytes (Fig. 4C). Thus, pCMBS slows the acidification only slightly in water-injected and C189S-expressing oocytes, but substantially in oocytes expressing wild-type AQP1 (Fig. 4D).

Experiments on AQP1 reconstituted into membrane vesicles

The group headed by Zeidel has also reported an AQP1-dependent increase in CO₂ permeability. In initial experiments they measured the CO₂ permeability of proteoliposomes which had been constructed artificially with known lipid compositions. Although the H₂O permeability of these proteoliposomes varied greatly, in proportion to the vesicle fluidity, there was no difference in the CO₂ permeability of these vesicles (Prasad et al. 1998). However in membrane vesicles constructed from E. coli membrane lipid, the CO₂ permeability was significantly lower than for the 'artificial' proteoliposomes. Reconstituting AQP1 into these E. coli vesicles, increased the CO₂-induced acidification rate ~4-fold (Prasad et al. 1998) (Fig. 5A), an effect that was abolished by HgCl₂ (Fig. 5B).

View larger version
[in this window]
[in a new window]

Figure 5. Aquaporin 1 increases the CO₂ permeability of liposomes made from E. coli membrane lipids
A, effect of incorporating AQP1 into liposomes. In vesicles prepared from E. coli membranes, adding AQP1 produced a significant increase in the rate at which intravesicular pH decreased upon exposure to a CO₂-containing solution. B, effect of blocking AQP1 with HgCl₂. The AQP1-dependent increase in CO₂ permeability was reduced by addition of HgCl₂, and reversed by $beta$ -mercaptoethanol. (Reproduced from Prasad et al. 1998 with permission of the Journal of Biological Chemistry.)

NH₃ movement through the peribacteroid membrane may be mediated by nodulin 26

In legumes, biological nitrogen fixation from N₂ to NH₃/NH₄⁺ occurs in specialized root nodules that play host to the bacteria rhizobia. The rhizobia are enclosed in a symbiosome compartment bounded by the peribacteroid membrane. The symbiotic arrangement found in the root nodules of these plants is of particular interest with respect to gas transport. Low O₂ levels are necessary to induce expression of the nitrogenase enzyme in rhizobia, which is responsible for nitrogen fixation. Indeed, high O₂ levels inhibit the enzyme. For efficient nitrogen fixation the peribacteroid membrane needs to limit O₂ movement whilst maintaining a high permeability to N₂ and NH₃. Recent work has shown that the uptake of NH₃ into vesicles formed from the peribacteroid membrane is sensitive to mercurials. Addition of HgCl₂ reduced the vesicle NH₃ permeability by 42 %. These results suggest that a fraction of the peribacteroid membrane NH₃ transport is channel mediated (Niemietz & Tyerman, 2000). The protein nodulin-26, a member of the MIP family, is expressed at high levels in the peribacteroid membrane, and it has been shown that this protein is permeable to water and several other small solutes (Rivers et al. 1997). The channel-mediated fraction of peribacteroid membrane NH₃ transport was attributed to nodulin-26, although the high activation energy of the NH₃ transport suggests that there are distinct pathways through the channel for H₂O and NH₃ (Niemietz & Tyerman, 2000). In addition to the hypothesized channel-mediated pathway for NH₃ transport, the peribacteroid membrane also appears to be able to mediate NH₄⁺ transport via a NH₄⁺-permeable channel.

Cooper et al. (1999) have examined the effect of expressing nodulin-26 on the NH₃/NH₄⁺ permeability of Xenopus oocytes. They found that the presence of nodulin-26 increased the permeability of the oocytes to both NH₃ and NH₄⁺. The mercurial agent pCMBS abolished the component of NH₃ permeability that required expression of nodulin-26. These data support the hypothesis of Niemietz & Tyerman (2000) - that nodulin-26 accounts for the protein-mediated flux of NH₃ across the peribacteroid membrane.

The relative contribution of nodulin-26 to the transport of NH₃/NH₄⁺ out of the symbiosome is linked to the pH of the symbiosome space. One would expect that increased acidity of the symbiosome space - which would lower the [NH₃]/[NH₄⁺] ratio within the symbiosome space - would favour NH₃ transport into, rather than out of, the symbiosome (opposite to what happens during N₂ fixation), whereas increased alkalinity would favour the efflux of NH₃ from the symbiosome space into the cytoplasm. Thus, one might expect that regulation of symbiosome pH might influence the net movement of NH₃ across the symbiosome membrane. Indeed, pre-incubation of peribacteroid membrane vesicles with ATP (which supports the H⁺ pump and thus would be expected to lower intravesicular pH) reduces the uptake of NH₃.

The CO₂ permeability of red blood cells is inhibited by DIDS

The red blood cell membrane has an extremely high permeability to O₂ and CO₂, a property that goes hand in hand with the major role of these cells - the carriage of respiratory gases between the pulmonary and systemic capillaries. The stilbene derivative 4,4'-diisothiocyanatostilbene-2,2'-disulfonic acid (DIDS), a compound commonly used as an inhibitor of bicarbonate transporters, inhibits the apparent CO₂ permeability of red blood cells by ~90 % (Forster et al. 1998). However, the action of DIDS on CO₂ permeability apparently did not involve inhibition of either the Cl^--HCO₃^- exchanger (Band 3) or intracellular carbonic anhydrase. The authors concluded that the transport of CO₂ into the red blood cell involves a membrane protein.

Might the CO₂ permeability of AQP1 be physiologically relevant?

Whether or not the permeability of any of the AQPs to CO₂ or other gases proves to be important to organisms, studying the permeability of the AQPs to these gases may provide insight into the workings of the AQPs, just as probing ion channels with unphysiological ions has proved important for understanding ion-channel function. Nevertheless, a key, unresolved question remains: do any of the AQPs mediate the flux of quantitatively significant quantities of a small, volatile molecule?

Evidence 'for'

Circumstantial evidence

There is no doubt that many MIP family members are highly H₂O permeable. Moreover, it is clear that the H₂O transport through some AQPs is sometimes physiologically important (e.g. AQP1 in the renal proximal tubule, AQP2 and AQP3 in the renal collecting duct). Doubtlessly, other examples will emerge. On the other hand, AQPs are often present in high abundance in cells for which one might think that a high H₂O permeability is not important (e.g. RBCs). In most of these cases, the presence of high AQP levels probably would do no harm. However, why would cells invest the resources necessary to produce huge amounts of AQPs if the increased H₂O permeability conferred no obvious advantage? Might these AQP pores have a function other than raising P_H2O?

In the case of RBCs, it has been pointed out that AQP1 permits rapid volume changes as RBCs enter and leave the hypertonic renal medulla. But what is the advantage to changing volume rapidly? Would it be better not to change volume at all? Moreover, in patients with sickle cell anaemia, rapid RBC shrinkage would tend to concentrate the haemoglobin and thus might be deleterious. Why, then, should RBCs maintain 200 000 copies of AQP1? Why should the lung risk the rapid development of pulmonary oedema (e.g. under pathological conditions of high hydrostatic and/or low oncotic pressure in the capillaries) by expressing high levels of AQP1 in pulmonary capillaries and high levels of AQP5 in alveolar type-I pneumocytes (the thin cells through which gas exchange occurs)? Why should the brain risk the rapid development of cerebral oedema by expressing high levels of AQP4 in the astrocytes that surround the capillaries (Nielsen et al. 1997)? Is it possible that, at least in some of the restricted cases such as those just mentioned, one of the roles of the AQP may be to enhance gas transport? Indeed, membranes with high gas-transport rates always seem to have high levels of an AQP.

DIDS reduces P_CO2 in human RBCs

As mentioned above, Forster and coworkers found that treating RBCs with DIDS causes P_CO2 to fall by ~90 % (Forster et al. 1998). It has long been known that DIDS does not affect P_H2O in RBCs (Benga et al. 1983; Macey, 1984). In preliminary work on oocytes expressing AQP1, we found that DIDS reduces the AQP1-dependent P_CO2 to zero, while leaving P_H2O intact (Boron & Cooper, 1998). Taken together, the data of Forster and colleagues and our data are consistent with the hypothesis that a sizeable fraction of CO₂ crosses the RBC membrane via AQP1.

Bicarbonate transport by the proximal tubule

The proximal tubule (PT) of the kidney reabsorbs approximately 80 % of the HCO₃^- filtered at the glomerulus. The proximal-tubule cell secretes protons that, in the presence of CA IV, which is coupled to the outer surface of the apical membrane, rapidly titrate luminal HCO₃^- to H₂O and CO₂. The CO₂ and H₂O enter the PT cells and then, under the influence of cytoplasmic CA II, re-combine to form HCO₃^- and H⁺. The HCO₃^- leaves the cell via a basolateral electrogenic Na⁺-HCO₃^- cotransporter and the protons recycle across the apical membrane. Thus, the reabsorption of HCO₃^- from the lumen is, in fact, the uptake of CO₂ across the apical membrane. The following thought experiment illustrates that this apical uptake of CO₂ is indeed prodigious.

Let us make three reasonable assumptions: (i) the glomerular filtrate contains 24 mM HCO₃^-, (ii) the PT reabsorbs 80 % of this HCO₃^-, and (iii) the PT also reabsorbs two-thirds of the filtered fluid. Moreover, let us assume that the PT reabsorbs the two-thirds of the filtered fluid before it reabsorbs any HCO₃^-. Thus, luminal [HCO₃^-] would triple, rising to 72 mM. Reabsorbing 80 % of this HCO₃^- requires that the PT form 0.8 times 72 mM 58 mM of dissolved CO₂ - a concentration that corresponds to a CO₂ partial pressure of 2.5 atmospheres! If we make the not unreasonable assumption that the contact time of the tubule fluid with the PT is 5 s, then the PT must be able to take up the equivalent of 0.5 atmospheres of dissolved CO₂ per second. It is interesting to note that the PT expresses AQP1 at high levels on both the apical and basolateral membranes. This AQP1 clearly plays an important role in transepithelial H₂O transport. We were also keen to address the question of whether the presence of AQP1 in the proximal tubule also contributes to HCO₃^- reabsorption.

The apical membrane. Preliminary work indicates that either 100 µM DIDS (Boron & Cooper, 1998) or 100 µM Zn²⁺ (Cooper & Boron, 1998b) blocks the CO₂ permeability of AQP1 expressed in oocytes, without substantially affecting P_H2O. If AQP1 contributes to the transport of CO₂ across the apical membrane of the PT cell then both Zn²⁺ and DIDS should inhibit HCO₃^- reabsorption. In preliminary studies, we worked with isolated perfused S2 segments of the proximal tubule, perfusing the lumen with 5 % CO₂/22 mM HCO₃^- and ³H-methoxyinulin (a volume marker). We computed the rate of HCO₃^- reabsorption (J_HCO3) from the [HCO₃^-] and [³H-methoxyinulin] in the collected fluid. Other experiments had shown that we could increase J_HCO3 by ~50 % by exposing the basolateral surface of the tubule ('bath') to a pH 7.4 out-of-equilibrium (OOE) CO₂/HCO₃^- containing 5 % CO₂ but virtually no HCO₃^- (Zhao et al. 1995), rather that to a similar equilibrated solution that also contained 22 mM HCO₃^-; in other words, bath HCO₃^- inhibits HCO₃^- reabsorption (Zhou et al. 1999). Under these conditions, adding either 100 µM Zn²⁺ or 100 µM DIDS to the lumen reduced J_HCO3 by ~30 % (Zhou et al. 2000). Other experiments had shown that, in the absence of AQP1 inhibitors, switching the bath from a 5 % CO₂/nominally HCO₃^--free OOE solution to a 20 % CO₂/nominally HCO₃^--free solution causes a substantial increase in J_HCO3 (Y. Zhou & W. F. Boron, unpublished observations); in other words, bath CO₂ directly stimulates HCO₃^- reabsorption. Testing the AQP1 inhibitors when the bath contained 20 % CO₂ but virtually no HCO₃^-, we found that each of the inhibitors reduced J_HCO3 by ~50 %.

In another series of experiments, we examined the effects of luminal Zn²⁺ when the bath contained physiological levels of HCO₃^- (22 mM) at pH 7.4. Luminal Zn²⁺ reduced by J_HCO3 by ~40 % when 5 % CO₂ was in the bath (the standard equilibrated solution). When the bath contained 20 % CO₂ and the basal J_HCO3 was higher, Zn²⁺ reduced J_HCO3 by more than 60 %. To increase J_HCO3 in a more physiological way, we introduced 10^-11 M angiotensin II into the lumen. Under these conditions, when the bath contained 5 % CO₂/22 mM HCO₃^-, luminal Zn²⁺ reduced the elevated J_HCO3 by ~50 %. When the bath contained 20 % CO₂/22 mM HCO₃^-, Zn²⁺ reduced the even-more elevated J_HCO3 by nearly 65 %. Thus, it appears that the greater the basal J_HCO3, the greater the reduction in J_HCO3 by inhibitors of AQP1 gas permeability.

The basolateral membrane. The classical model for the reabsorption of HCO₃^- by the proximal tubule has electrogenic NBC transporting Na⁺ and HCO₃^- across the basolateral membrane in a ratio of 1:3, with two net negative charges moving out of the cell. In principle, NBC could transport the same number of charges and perform the same acid-base work if, instead of carrying 3 HCO₃^-, it carried 1 CO₃^2- and 1 HCO₃^-. Indeed, preliminary work by Grichtchenko & Boron (2002a) indicates that NBC does indeed transport CO₃^2-. They studied voltage-clamped oocytes co-expressing NBC and CA IV, and monitored external-surface pH before and after blocking CA IV with acetazolamide (ACZ). However, in terms of atoms transported, an NBC carrying CO₃^2- + HCO₃^- comes up short by 1 carbon, 2 hydrogens and 3 oxygens. We suggest that renal NBC may work according to the following manner, which borrows some elements from the work of Eberhard Frömter (Muller-Berger et al. 2001) and of Irina Grichtchenko and Mitch Chesler (Grichtchenko & Chesler, 1994). Three HCO₃^- ions approach the membrane from the cytoplasmic side, but NBC directly transports only one of them. A second HCO₃^- dissociates to form CO₃^2-, which NBC transports along with the aforementioned HCO₃^- and a Na⁺. Immediately adjacent to the membrane, the H⁺ created during the formation of this CO₃^2- causes such a large fractional increase in [H⁺] that the third HCO₃^- that approaches the membrane undergoes the reaction HCO₃^- + H⁺ right CO₂ + H₂O. In order to complete the overall transport process, the newly formed CO₂ plus H₂O must cross the membrane in parallel with the Na⁺, HCO₃^- and CO₃^2-. Once outside the cell, all of the above reactions would reverse, so that the CO₃^2-, CO₂ and H₂O would regenerate two HCO₃^- ions. For this mechanism to work efficiently, carbonic anhydrases would be required on both the intra- and extracellular faces of the basolateral membrane, and the permeability of the basolateral membrane to CO₂ and H₂O would have to be high. Our preliminary work suggests that cytoplasmic CA II binds to the cytoplasmic C terminus of NBC (J. D. Rojas, I. Choi, B. A. Davis, L. V. Virkki & W. F. Boron, unpublished observations), consistent with a similar observation made on AE1 by Reithmeier (Vince & Reithmeier, 1998, 2000). CA IV is present on the outer surface of the basolateral membrane of PT cells. Finally, it is tempting to speculate that the AQP1 that is abundantly present on the PT basolateral membrane serves as a conduit for at least some of the CO₂ and H₂O that would have to exit the cell in parallel with Na⁺, HCO₃^- and CO₃^2-.

To test the validity of this above model, Grichtchenko & Boron (2002b) co-expressed NBC, CA II, CA IV and AQP1 in voltage-clamped oocytes, and monitored the pH just above the oocyte membrane. Shifting the holding potential from -50 to +50 mV causes a marked acceleration in the transport of Na⁺, HCO₃^- and CO₃^2- into the oocyte. As extracellular HCO₃^- buffers the H⁺ created in the formation of the CO₃^2-, we would expect a small increase in [CO₂] on the outer surface of the oocyte - the price paid for creating the CO₂ gradient necessary to drive CO₂ influx. If AQP1 contributes to the influx of CO₂, then blocking AQP1's CO₂ permeability with Zn²⁺ ought to increase the surface [CO₂] still further, slow the reaction H⁺ + HCO₃^- right CO₂ + H₂O, and thus lead to a buildup of H⁺ on the cell surface. This is exactly what Grichtchenko and Boron found in their preliminary work. Moreover, this Zn²⁺-dependent acidification of the oocyte surface was almost completely eliminated by deleting the AQP1 (i.e. the oocyte expressed NBC, CA II and CA IV, but not AQP1). These experiments provide a proof of concept that AQP1, acting in parallel with NBC and in concert with carbonic anhydrases, could contribute to the overall process of 'HCO₃^- transport' across the basolateral membrane of the PT.

Evidence 'against'

Colton-null humans

In humans, AQP1 is the basis for the Colton blood-group antigen. One might argue that the lack of an overt phenotype in Colton-null individuals means that AQP1 must not be mediating substantial gas transport. However, by the same logic, we would be forced to conclude that AQP1 does not mediate substantial water transport. Detecting a deficiency might require stressing the system. In the case of pulmonary gas exchange, the stress might be exercise at altitude - and the experiments would have to be designed to detect impaired gas exchange in the face of compensatory hyperventilation. In the case of the acid-base transport by the kidney, the stress might be respiratory acidosis. As is often the case in knockouts, the body may compensate for the deficient gene, so that no overt phenotype is apparent, even under stress. Thus, conclusions drawn from Colton-null humans as well as AQP1-knockout mice must be taken with a grain of salt.

AQP1-deficient mice

Yang et al. (2000) have used AQP1-knockout mice to examine whether AQP1 mediates a quantitatively important flux of CO₂ in the lungs and erythrocytes (stopped-flow studies).

Lungs. When they artificially ventilated mice with air containing 5 % CO₂, Yang et al. (2000) found an arterial P_CO2 of 77 Torr. On switching the inspired gas to 0 % CO₂, they found that arterial P_CO2 fell to 39 Torr over the same 10 min period (with a half-time of ~2 min), regardless of whether they worked with +/+ (i.e. wild-type) or - /- mice (i.e. AQP1 knockout). The authors concluded that AQP1 plays no role in the alveolar exchange of CO₂. This analysis is incorrect. As blood flows down a pulmonary capillary of a normal mammal, the blood probably travels only about one-third of the way along the capillary before CO₂ in the blood comes into diffusion equilibrium with the CO₂ in alveolar air. Blood from many capillaries mixes in the left ventricle, and a few seconds later, mixed blood appears in a systemic artery where one measures P_CO2. Thus, at any instant in time, arterial P_CO2 reflects a weighted average of alveolar P_CO2 values a few seconds earlier. If, after a sudden shift in the composition of the inspired gas, the arterial P_CO2 changes with a half-time of 2 min, then one can conclude that the average alveolar P_CO2 also changed with a half-time of 2 min. In conclusion, the arterial P_CO2 changes monitored by Yang et al. (2000) reflect the washout of the alveoli by a new inspired-gas mixture, and make no statement about the diffusion of CO₂ across the alveolar blood-gas barrier.

In a more recent effort, Fang et al. (2002) filled the alveoli of mouse lungs with a fluid containing the fluorescent pH indicator BCECF-dextran as well as carbonic anhydrase. Simultaneously, they perfused the pulmonary artery alternately with either a CO₂/HCO₃^--free saline or a saline buffered with 5 % CO₂/25 mM HCO₃^-. The authors found that switching to the CO₂-containing pulmonary-artery perfusate caused a rapid acidification of the alveolar space - reflecting the influx of CO₂ from the pulmonary capillaries - and that the speed of the acidification was indistinguishable among lungs from wild-type mice, AQP1 -/- mice, AQP5 - /- mice, and the double-knockout AQP1/5 -/- mice. In parallel experiments, the authors increased the osmolality of the perfusate and used the resulting increases in BCECF fluorescence as an index of osmotic permeability. The deletion of AQP1 or AQP5 caused a 10-fold reduction in the osmotic water permeability, and the double deletion caused a further 3-fold reduction. These are indeed elegant experiments and, assuming no differences in the surface area for gas exchange, would have convincingly demonstrated that AQP1 and AQP5 do not contribute significantly to CO₂ permeability in the lung - except for one concern. If the pulmonary arteries had been perfused with blood (preferably with the AQP status of the RBCs matching that of the perfused lungs) then carbonic anhydrase in the RBCs could have catalysed the reaction HCO₃^- + H⁺ right H₂O + CO₂ and thereby have replenished the CO₂ lost by diffusion into the alveolar air spaces. In lungs perfused with saline, capillary [CO₂] may have fallen rapidly to levels in the alveolar space, so that the diffusion of CO₂ would have become 'perfusion limited' rather than 'diffusion limited.'

Erythrocytes. Using a stopped-flow approach together with an intracellular fluorescent pH-sensitive dye, Yang et al. (2000) measured the rate at which the pH_i of RBCs fell upon exposure to CO₂. The computed P_CO2 was indistinguishable in +/+ and -/- mice, 0.011-0.012 cm s^-1. Although a 5 min pretreatment with HgCl₂ greatly reduced the osmotic water permeability in RBCs from +/+ animals, a < 2 min pretreatment with HgCl₂ failed to reduce P_CO2. However, in RBCs from +/+ mice, blocking CA with ACZ slowed the CO₂-induced pH_i decline by ~90 %. More recently, Fang et al. (2002) obtained similar results using membrane vesicles derived from proximal-tubule apical membranes. Both sets of raise technical and theoretical concerns.

First, the measured P_CO2 in RBCs from wild-type mice was 0.012 cm s^-1, which is more than an order of magnitude less than Roughton's value of 0.15 cm s^-1 (Roughton, 1959), nearly 50-fold less than Forster's value (Forster, 1968), and about two orders of magnitude less than Gros's value (Forster et al. 1998). The most likely explanation for the very low apparent P_CO2 values in the experiments of Yang et al. (2000) is the presence of substantial unstirred layers around RBCs; indeed, unstirred layers are a major limitation of the stopped-flow technique. A 10- to 100-fold underestimation of the overall CO₂ transport rate in the study of Yang et al. would make it impossible to detect even very large AQP1-dependent increases in CO₂ permeability.

Second, pH_i measurements are an indirect way of assessing the influx of CO₂ across the cell membrane. CO₂ must move from the bulk extracellular fluid, through an extracellular unstirred layer, across the cell membrane, and through an intracellular unstirred layer to reach the bulk intracellular fluid. There, the reaction CO₂ + H₂O right HCO₃^- + H⁺ generates the protons that can bind to a pH-sensitive dye and thereby alter a fluorescence signal (for example) that one detects. One should not confuse the time constant of the pH_i change with CO₂ permeability. Imagine that we suddenly increase [CO₂] at the outer surface of a cell membrane from zero to 1 mM, thereby producing a large initial influx of CO₂. In the presence of a high intracellular carbonic anhydrase activity, the incoming CO₂ leads to the rapid generation of H⁺, and thus a rapid fall in pH_i. If we repeat the experiment in the absence of CA, the same initial influx of CO₂ produces only a very slow fall in pH_i. Although CA greatly speeds the pH_i decrease, it has no effect on the initial burst of CO₂ influx. Only some time later does CA indirectly speed CO₂ entry: by consuming incoming CO₂, CA keeps [CO₂] lower for a longer period of time, and thus helps to sustain the CO₂ gradient that drives the passive influx.

The following analogy illustrates the limitations of a CA-dependent pH_i measurement for estimating the initial CO₂ influx (or P_CO2) in a system with an unstirred layer large enough to decrease the apparent P_CO2 by 10- to 100-fold. Imagine that we are given the task of judging the winner of a 100 m dash. One contestant is a world-class athlete who will finish the race in less than 10 seconds. The other is a retired physiologist who will require at least a minute. Rather than photographing the contestants as they cross the finish line, we instruct each to use a book of matches (the CA) to light a candle (the fluorescence) at the finish line. The first to light a candle wins. We fire the starting gun and then go off to have a cup of tea. Returning 15 minutes later (to simulate the unstirred layer), we find both candles lit and declare the contest a tie! In a second race, we replace the matches with pieces of flint and steel (the uncatalysed reaction) to light the candles, fire the starting gun, and return 15 minutes later to find two frustrated contestants but no lit candles. Our conclusions: (i) the runners are equally slow, and (ii) a book of matches is the most important element in running a race.

In conclusion, it is critical to distinguish what is rate limiting in the assay (e.g. the rate constant of the reaction CO₂ + H₂O right HCO₃^- + H⁺) from what is rate limiting in the biological process under study (e.g. the movement of CO₂ through the membrane). If the assay lacks sufficient time resolution, no conclusions can be drawn concerning the biological process. Thus, in the experiments of Yang et al. (2000) no conclusions can be reached concerning the role of AQP1 in CO₂ transport.

AQP1 reconstituted into membrane vesicles

Using a stopped-flow approach and artificially constructed liposomes into which they reconstituted AQP1, Yang et al. (2000) attempted to reproduce the aforementioned findings of Prasad et al. (1998). These latter authors had tested vesicles with three defined lipid compositions, finding in all cases that the baseline CO₂ permeability was too high to detect easily any increase induced by AQP1. However, when Prasad et al. made liposomes from E. coli phospholipids, baseline P_CO2 fell by a factor of ~3, and incorporating AQP1 increased P_CO2 4-fold. Yang et al. (2000) used only the lipids with higher baseline P_CO2 and - consistent with the observations of Prasad et al. (1998) - did not detect an increase in apparent P_CO2 when they incorporated AQP1. However, Yang et al. did not test the liposomes that, according to the data of Prasad et al. might have been expected to yield a positive result. Aside from the issue of the lipids chosen, the liposome study of Yang et al. has the same technical limitation outlined above for the RBC study: the unstirred layers were so large that it would have been impossible to detect any AQP1-dependent increase in P_CO2.

Two other issues remain in comparing the liposome work of Prasad et al. (1998) and Yang et al. (2000). First, how is it that Prasad et al. were able to use a stopped-flow approach to detect a 4-fold P_CO2 increase in liposomes containing AQP1? One possibility is that (i) the use of E. coli phospholipids lowered the baseline rate of CO₂ entry sufficiently and (ii) the addition of AQP1 raised the true P_CO2 sufficiently that - even with the unstirred layers inherent with the stopped-flow technique - Prasad et al. were able to detect a difference in the acidification rate of the liposomes. In other words, the worse the unstirred layer, the less the baseline CO₂ permeability would have to be and/or the greater the CO₂ permeability of AQP1 would have to be in order to detect CO₂ moving through AQP1. A corollary is that technical limitations lead to underestimates of AQP1' s true CO₂ permeability.

The second remaining issue concerns the action of HgCl₂. Prasad et al. (1998) working with AQP1-containing liposomes made from E. coli phospholipids, found that treating with HgCl₂ caused the CO₂-induced acidification rate to fall to a value indistinguishable from that observed with AQP1-free liposomes. Moreover, treatment with $beta$ -mercaptoethanol reversed the effects of HgCl₂ on AQP1-containing liposomes. Yang et al. working with the leakier liposomes of defined lipid composition, found that HgCl₂ slowed the CO₂-induced acidification in both AQP1-containing and control liposomes. These latter authors suggested that, in the experiments of Prasad et al. HgCl₂ exhibited its effect not by reacting with C189 on AQP1, but by leaking into the liposome and inhibiting CA. Again, it is unfortunate that the two groups did not perform precisely the same set of experiments. Prasad et al. did not examine the effect of HgCl₂ in their tight control (i.e. AQP1 free) liposomes made from E. coli phospholipids. Thus, we cannot rule out a nonspecific effect of HgCl₂ on CA, the pH-sensitive dye, or on the basal CO₂ permeability of the membrane lipid. On the other hand, if HgCl₂ were producing non-specific effects in the experiments of Prasad et al. it is curious that adding HgCl₂ to AQP1-containing vesicles reduced the CO₂-induced acidification rate to almost exactly the same value as in control vesicles lacking AQP1. Thus we cannot rule out the possibility that HgCl₂ had non-specific effects on the leakier liposomes used by Yang et al.

Conclusions

Data on model systems indicates that CO₂ can move through AQP1, and that NH₃ can move through nodulin-26. A key question at this juncture is whether the movement of small volatile molecules through MIPs or other membrane proteins is ever physiologically important. It is very unlikely that membrane proteins make a substantial contribution to gas permeability in all membranes. When a cell membrane consists of lipids with high intrinsic gas permeability, and when the demands on gas transport are low, the cell can likely meet those demands by simple diffusion of the gas through the membrane lipid. On the other hand, membrane proteins might play an important role where cell membranes have a relatively low intrinsic gas permeability and/or where rates of gas transport are very high. Membranes that are likely to have a relatively low intrinsic gas permeability are those exposed to a hostile environment, and might thus include erythrocytes (which are subject to shear stresses and have a limited capacity for self repair), eukaryotic cells exposed to pond water (which are subject to osmotic bursting), the apical membranes of many epithelia (which may be exposed to toxic chemicals as H⁺, NH₃, enzymes and bile salts), the ciliary body as well as the choroid plexus and blood-brain barrier (where blood would be hostile to neurons). High rates of gas transport occur across erythrocytes, pulmonary epithelia, proximal-tubule apical membranes, and the choroid plexus/blood-brain barrier.

Determining whether membrane proteins play a significant role in these preparations will be extremely challenging technically. Dynamic measurements of gas transport that rely on transient changes in some measured parameter (e.g. pH or P_CO2), and approaches that are foiled by unstirred layers may be especially prone to yielding false-negative results. More promising are techniques that rely on steady-state measurements, such as isotope exchange (Dodgson et al. 1990), rates of HCO₃^- reabsorption by the proximal tubule (Zhou et al. 2000), and surface-pH changes produced by CO₂ transport (Grichtchenko & Boron, 2002b).

REFERENCES

Top
Abstract
Introduction
Methods
Results
Discussion
References

BENGA, G., POP, V. I., POPESCU, O., IONESCU, M. & MIHELE, V. (1983). Water exchange through erythrocyte membranes: nuclear magnetic resonance studies on the effects of inhibitors and of chemical modifications of human membranes. Journal of Membrane Biology 76, 129-137 [Medline]

BORON, W. F. & COOPER, G. J. (1998). Effect of DIDS on the CO₂ permeability of the water channel AQP1. FASEB Journal 12, A374

BURCKHARDT, B. C. & FRÖMTER, E. (1992). Pathways of NH₃/NH₄⁺ permeation across Xenopus laevis oocyte cell membrane. Pflügers Archiv 420, 83-86 [Medline]

BURG, M., GRANTHAM, J., ABRAMOW, M. & ORLOFF, J. (1966). Preparation and study of fragments of single rabbit nephrons. American Journal of Physiology 210, 1293-1298 [Medline]

COOPER, G. J. & BORON, W. F. (1998a). Effect of PCMBS on CO₂ permeability of Xenopus oocytes expressing aquaporin 1 or its C189S mutant. American Journal of Physiology 275, C1481-1486 [Medline]

COOPER, G. J. & BORON, W. F. (1998b). Effect of zinc on the CO₂ permeability of the water channel AQP1. Journal of the American Society of Nephrology 9, 17A

COOPER, G. J., VIRKKI, L. V. & BORON, W. F. (1999). Effect of expressing Nodulin 26 on the membrane permeability of Xenopus oocytes to NH₃/NH₄⁺ and acetic acid. FASEB Journal 13, LB26., COUGNON

DODGSON, S. J., GROS, G., KRAWIEC, J. A., LIN, L., BITTERMAN, N. & FORSTER, R. E. (1990). Comparison of ¹⁸O exchange and pH stop-flow assays for carbonic anhydrase. Journal of Applied Physiology 68, 2443-2450 [Medline]

ENGEL, A., FUJIYOSHI, Y. & AGRE, P. (2000). The importance of aquaporin water channel protein structures. Embo Journal 19, 800-806 [Abstract/Full Text]

FANG, X., YANG, B., MATTHAY, M. A. & VERKMAN, A. S. (2002). Evidence against aquaporin-1-dependent CO₂ permeability in lung and kidney. Journal of Physiology 542, 63-69 [Abstract/Full Text]

FINKELSTEIN, A. (1986). Water Movement Through Lipid Bilayers, Pores, and Plasma Membranes. Theory and Reality. John Wiley & Sons, New York

FORSTER, R. E. (1968). The rate of CO₂ equilibration between red cells and plasma. In CO₂: Chemical, Biochemical and Physiological Aspects, ed. FORSTER, R. E., EDSALL, J. T., OTIS, A. B. & ROUGHTON, F. J. W., pp. 275-284. NASA SP-188, Washington

FORSTER, R. E., GROS, G., LIN, L., ONO, Y. & WUNDER, M. (1998). The effect of 4,4'-diisothiocyanato-stilbene-2,2'-disulfonate on CO₂ permeability of the red blood cell membrane. Proceedings of the National Academy of Sciences of the USA 95, 15815-15820 [Abstract/Full Text]

GORIN, M. B., YANCEY, S. B., CLINE, J., REVEL, J. P. & HORWITZ, J. (1984). The major intrinsic protein (MIP). of the bovine lens fiber membrane: characterization and structure based on cDNA cloning. Cell 39, 49-59 [Medline]

GRICHTCHENKO, I. I. & BORON, W. F. (2002a). Evidence for CO⁼₃ transport by NBCe1, based on surface-pH measurements in voltage-clamped Xenopus oocytes co-expressing NBCe1 and CAIV: evidence for CO⁼₃ transport. FASEB Journal 16, A795

GRICHTCHENKO, I. I. & BORON, W. F. (2002b). Evidence for CO₂ permeability of AQP1 based on surface-pH measurements in Xenopus oocytes co-expressing NBCe1, CAII, CAIV and AQP1. FASEB Journal 16, A808

GRICHTCHENKO, I. I. & CHESLER, M. (1994). Depolarization-induced acid secretion in gliotic hippocampal slices. Neuroscience 62, 1057-1070 [Medline]

GUNSHIN, H., MACKENZIE, B., BERGER, U. V., GUNSHIN, Y., ROMERO, M. F., BORON, W. F., NUSSBERGER, S., GOLLAN, J. L. & HEDIGER, M. A. (1997). Cloning and characterization of a mammalian proton-coupled metal-ion transporter. Nature 388, 482-488 [Medline]

HEBERT, S. C. & ANDREOLI, T. E. (1984). Control of NaCl transport in thick ascending limb. American Journal of Physiology 246, F745-756 [Medline]

HEVESY, G., HOFER, E. & KROGH, A. (1935). The permeability of the skin of frogs to water as determined by D₂O and H₂O. Skandinavisches Archiv fuer Physiologie 72, 199-214

HEYMANN, J. B., AGRE, P. & ENGEL, A. (1998). Progress on the structure and function of aquaporin 1. Journal of Structural Biology 121, 191-206 [Medline]

HICKS, R. M., KETTERER, B. & WARREN, R. C. (1974). The ultrastructure and chemistry of the luminal plasma membrane of the mammalian urinary bladder: a structure with low permeability to water and ions. Philosophical Transactions of the Royal Society B 268, 23-38

HILL, W. G., RIVERS, R. L. & ZEIDEL, M. L. (1999). Role of leaflet asymmetry in the permeability of model biological membranes to protons, solutes, and gases. Journal of General Physiology 114, 405-414 [Abstract/Full Text]

ITO, S. (1987). Functional gastric morphology. In Physiology of the Gastrointestinal Tract, ed. JOHNSON, L., pp. 817-851. Raven Press, New York

JUNG, J. S., PRESTON, G. M., SMITH, B. L., GUGGINO, W. B. & AGRE, P. (1994). Molecular structure of the water channel through aquaporin CHIP. The hourglass model. Journal of Biological Chemistry 269, 14648-14654 [Abstract]

KIKERI, D., SUN, A., ZEIDEL, M. L. & HEBERT, S. C. (1989). Cell membranes impermeable to NH₃. Nature 339, 478-480 [Medline]

KOEFOED-JOHNSEN, V. & USSING, H. H. (1953). The contributions of diffusion and flow to the passage of D₂O through living membranes. Acta Physiologica Scandinavica 28, 60-76

LANDE, M. B., DONOVAN, J. M. & ZEIDEL, M. L. (1995). The relationship between membrane fluidity and permeabilities to water, solutes, ammonia and protons. Journal of General Physiology 106, 67-84 [Abstract]

MACEY, R. I. (1984). Transport of water and urea in red blood cells. American Journal of Physiology 246, C195-C203 [Medline]

MULLER-BERGER, S., DUCOUDRET, O., DIAKOV, A. & FRÖMTER, E. (2001). The renal Na-HCO₃-cotransporter expressed in Xenopus laevis oocytes: change in stoichiometry in response to elevation of cytosolic Ca²⁺ concentration. Pflügers Archiv 442, 718-728 [Medline]

MURATA, K., MITSUOKA, K., HIRAI, T., WALZ, T., AGRE, P., HEYMANN, J. B., ENGEL, A. & FUJIYOSHI, Y. (2000). Structural determinants of water permeation through aquaporin-1. Nature 407, 599-605 [Medline]

NAKHOUL, N. L., DAVIS, B. A., ROMERO, M. F. & BORON, W. F. (1998). Effect of expressing the water channel aquaporin-1 on the CO₂ permeability of Xenopus oocytes. American Journal of Physiology 274, C543-548 [Medline]

NIELSEN, S. & AGRE, P. (1995). The aquaporin family of water channels in kidney. Kidney International 48, 1057-1068 [Medline]

NIELSEN, S., FROKIAER, J., MARPLES, D., KWON, T. H., AGRE, P. & KNEPPER, M. A. (2002). Aquaporins in the kidney: from molecules to medicine. Physiological Reviews 82, 205-244 [Abstract/Full Text]

NIELSEN, S., NAGELHUS, E. A., AMIRY-MOGHADDAM, M., BOURQUE, C., AGRE, P. & OTTERSEN, O. P. (1997). Specialized membrane domains for water transport in glial cells: high-resolution immunogold cytochemistry of aquaporin-4 in rat brain. Journal of Neuroscience 17, 171-180 [Abstract/Full Text]

NIEMIETZ, C. M. & TYERMAN, S. D. (2000). Channel-mediated permeation of ammonia gas through the peribacteroid membrane of soybean nodules. FEBS Letters 465, 110-114 [Medline]

PAGANELLI, C. V. & SOLOMON, A. K. (1957). The rate of exchange of tritiated water across the human red cell membrane. Journal of General Physiology 41, 259-277

PRASAD, G. V., COURY, L. A., FIN, F. & ZEIDEL, M. L. (1998). Reconstituted aquaporin 1 water channels transport CO₂ across membranes. Journal of Biological Chemistry 273, 33123-33126 [Abstract/Full Text]

PRESCOTT, D. M. & ZEUTHEN, E. (1953). Comparison of water diffusion and water filtration across cell surfaces. Acta Physiologica Scandinavica 28, 77-94

PRESTON, G. M. & AGRE, P. (1991). Isolation of the cDNA for erythrocyte integral membrane protein of 28 kilodaltons: member of an ancient channel family. Proceedings of the National Academy of Sciences of the USA 88, 11110-11114 [Abstract]

PRESTON, G. M., CARROLL, T. P., GUGGINO, W. B. & AGRE, P. (1992). Appearance of water channels in Xenopus oocytes expressing red cell CHIP28 protein. Science 256, 385-387

PRESTON, G. M., JUNG, J. S., GUGGINO, W. B. & AGRE, P. (1993). The mercury-sensitive residue at cysteine 189 in the CHIP28 water channel. Journal of Biological Chemistry 268, 17-20 [Abstract]

RIVERS, R. L., DEAN, R. M., CHANDY, G., HALL, J. E., ROBERTS, D. M. & ZEIDEL, M. L. (1997). Functional analysis of nodulin 26, an aquaporin in soybean root nodule symbiosomes. Journal of Biological Chemistry 272, 16256-16261 [Abstract/Full Text]

ROOS, A. & BORON, W. F. (1981). Intracellular pH. Physiological Reviews 61, 296-434 [Medline]

ROUGHTON, F. J. W. (1959). Diffusion and simultaneous chemical reaction velocity in haemoglobin solutions and red cell suspensions. Progress in Biophysical Chemistry 9, 56-104

SCHÄFFNER, A. R. (1998). Aquaporin function, structure, and expression: are there more surprises to surface in water relations? Planta 204, 131-139 [Medline]

SIMON, S. A. & GUTKNECHT, J. (1980). Solubility of carbon dioxide in lipid bilayer membranes and organic solvents. Biochimica et Biophysica Acta 596, 352-358 [Medline]

SINGH, S. K., BINDER, H. J., GEIBEL, J. P. & BORON, W. F. (1995). An apical permeability barrier to NH₃/NH₄⁺ in isolated, perfused colonic crypts. Proceedings of the National Academy of Sciences of the USA 92, 11573-11577 [Abstract]

SWEET, G., GANDOR, C., VOEGELE, R., WITTEKINDT, N., BEUERLE, J., TRUNIGER, V., LIN, E. C. & BOOS, W. (1990). Glycerol facilitator of Escherichia coli: cloning of glpF and identification of the glpF product. Journal of Bacteriology 172, 424-430 [Medline]

VINCE, J. W. & REITHMEIER, R. A. (1998). Carbonic anhydrase II binds to the carboxyl terminus of human band 3, the erythrocyte Cl^-/HCO₃^- exchanger. Journal of Biological Chemistry 273, 28430-28437 [Abstract/Full Text]

VINCE, J. W. & REITHMEIER, R. A. (2000). Identification of the carbonic anhydrase II binding site in the Cl^-/HCO₃^- anion exchanger AE1. Biochemistry 39, 5527-5533 [Medline]

WAISBREN, S. J., GEIBEL, J. P., BORON, W. F. & MODLIN, I. M. (1994a). Luminal perfusion of isolated gastric glands. American Journal of Physiology 266, C1013-1027 [Medline]

WAISBREN, S. J., GEIBEL, J. P., MODLIN, I. M. & BORON, W. F. (1994b). Unusual permeability properties of gastric gland cells. Nature 368, 332-335 [Medline]

WALTER, A. & GUTKNECHT, J. (1986). Permeability of small nonelectrolytes through lipid bilayer membranes. Journal of Membrane Biology 90, 207-217 [Medline]

YANG, B., FUKUDA, N., VAN HOEK, A., MATTHAY, M. A., MA, T. & VERKMAN, A. S. (2000). Carbon dioxide permeability of aquaporin-1 measured in erythrocytes and lung of aquaporin-1 null mice and in reconstituted liposomes. Journal of Biological Chemistry 275, 2686-2692 [Abstract/Full Text]

ZHAO, J., HOGAN, E. M., BEVENSEE, M. O. & BORON, W. F. (1995). Out-of-equilibrium CO₂/HCO₃^- solutions and their use in characterizing a new K/HCO₃ cotransporter. Nature 374, 636-639 [Medline]

ZHOU, Y., BOUYER, P., VIRKKI, L., DAVIS, B. A. & BORON, W. F. (2000). Effect of Zn or DIDS on HCO₃^- reabsorption by S2 proximal tubule. FASEB Journal 14, A355

ZHOU, Y., ZHAO, J., BOUYER, P. & BORON, W. F. (1999). Effect of independently varying basolateral [CO₂] and [HCO₃^-] on HCO₃^- reabsorption by rabbit S2 proximal tubule. FASEB Journal 13, A65

Acknowledgements

This work was supported by a grant from the Office of Naval Research, an NIH Program Project Grant (PO1 DK17433), and an NIH research grant (RO1 DK30344). G.J.C. was supported by a HFSPO traveling fellowship. I.I.G. was supported by the National Kidney Foundation and NIH training grant T32 NS07455. P.B. was supported by the National Kidney Foundation.

Author's present address

G. J. Cooper: Department of Biomedical Science, Alfred Denny Building, University of Sheffield, Western Bank, Sheffield S10 2TN, UK.

Presented at The Journal of Physiology Synthesium on Water Transport Controversies, Christchurch, New Zealand, 30 August, 2001.

This article has been cited by other articles:

A. Missner, P. Kugler, S. M. Saparov, K. Sommer, J. C. Mathai, M. L. Zeidel, and P. Pohl
Carbon Dioxide Transport through Membranes
J. Biol. Chem., September 12, 2008; 283(37): 25340 - 25347.
[Abstract] [Full Text] [PDF]

N. Uehlein and R. Kaldenhoff
Aquaporins and Plant Leaf Movements
Ann. Bot., January 1, 2008; 101(1): 1 - 4.
[Abstract] [Full Text] [PDF]

Y. Wang and E. Tajkhorshid
Molecular Mechanisms of Conduction and Selectivity in Aquaporin Water Channels
J. Nutr., June 1, 2007; 137(6): 1509S - 1515S.
[Abstract] [Full Text] [PDF]

W. F. Boron
Acid-Base Transport by the Renal Proximal Tubule
J. Am. Soc. Nephrol., September 1, 2006; 17(9): 2368 - 2382.
[Abstract] [Full Text] [PDF]

<

BENGA, G., POP, V. I., POPESCU, O., IONESCU, M. & MIHELE, V. (1983). Water exchange through erythrocyte membranes: nuclear magnetic resonance studies on the effects of inhibitors and of chemical modifications of human membranes. Journal of Membrane Biology 76, 129-137	[Medline]
BORON, W. F. & COOPER, G. J. (1998). Effect of DIDS on the CO₂ permeability of the water channel AQP1. FASEB Journal 12, A374
BURCKHARDT, B. C. & FRÖMTER, E. (1992). Pathways of NH₃/NH₄⁺ permeation across Xenopus laevis oocyte cell membrane. Pflügers Archiv 420, 83-86	[Medline]
BURG, M., GRANTHAM, J., ABRAMOW, M. & ORLOFF, J. (1966). Preparation and study of fragments of single rabbit nephrons. American Journal of Physiology 210, 1293-1298	[Medline]
COOPER, G. J. & BORON, W. F. (1998a). Effect of PCMBS on CO₂ permeability of Xenopus oocytes expressing aquaporin 1 or its C189S mutant. American Journal of Physiology 275, C1481-1486	[Medline]
COOPER, G. J. & BORON, W. F. (1998b). Effect of zinc on the CO₂ permeability of the water channel AQP1. Journal of the American Society of Nephrology 9, 17A
COOPER, G. J., VIRKKI, L. V. & BORON, W. F. (1999). Effect of expressing Nodulin 26 on the membrane permeability of Xenopus oocytes to NH₃/NH₄⁺ and acetic acid. FASEB Journal 13, LB26., COUGNON
DODGSON, S. J., GROS, G., KRAWIEC, J. A., LIN, L., BITTERMAN, N. & FORSTER, R. E. (1990). Comparison of ¹⁸O exchange and pH stop-flow assays for carbonic anhydrase. Journal of Applied Physiology 68, 2443-2450	[Medline]
ENGEL, A., FUJIYOSHI, Y. & AGRE, P. (2000). The importance of aquaporin water channel protein structures. Embo Journal 19, 800-806	[Abstract/Full Text]
FANG, X., YANG, B., MATTHAY, M. A. & VERKMAN, A. S. (2002). Evidence against aquaporin-1-dependent CO₂ permeability in lung and kidney. Journal of Physiology 542, 63-69	[Abstract/Full Text]
FINKELSTEIN, A. (1986). Water Movement Through Lipid Bilayers, Pores, and Plasma Membranes. Theory and Reality. John Wiley & Sons, New York
FORSTER, R. E. (1968). The rate of CO₂ equilibration between red cells and plasma. In CO₂: Chemical, Biochemical and Physiological Aspects, ed. FORSTER, R. E., EDSALL, J. T., OTIS, A. B. & ROUGHTON, F. J. W., pp. 275-284. NASA SP-188, Washington
FORSTER, R. E., GROS, G., LIN, L., ONO, Y. & WUNDER, M. (1998). The effect of 4,4'-diisothiocyanato-stilbene-2,2'-disulfonate on CO₂ permeability of the red blood cell membrane. Proceedings of the National Academy of Sciences of the USA 95, 15815-15820	[Abstract/Full Text]
GORIN, M. B., YANCEY, S. B., CLINE, J., REVEL, J. P. & HORWITZ, J. (1984). The major intrinsic protein (MIP). of the bovine lens fiber membrane: characterization and structure based on cDNA cloning. Cell 39, 49-59	[Medline]
GRICHTCHENKO, I. I. & BORON, W. F. (2002a). Evidence for CO⁼₃ transport by NBCe1, based on surface-pH measurements in voltage-clamped Xenopus oocytes co-expressing NBCe1 and CAIV: evidence for CO⁼₃ transport. FASEB Journal 16, A795
GRICHTCHENKO, I. I. & BORON, W. F. (2002b). Evidence for CO₂ permeability of AQP1 based on surface-pH measurements in Xenopus oocytes co-expressing NBCe1, CAII, CAIV and AQP1. FASEB Journal 16, A808
GRICHTCHENKO, I. I. & CHESLER, M. (1994). Depolarization-induced acid secretion in gliotic hippocampal slices. Neuroscience 62, 1057-1070	[Medline]
GUNSHIN, H., MACKENZIE, B., BERGER, U. V., GUNSHIN, Y., ROMERO, M. F., BORON, W. F., NUSSBERGER, S., GOLLAN, J. L. & HEDIGER, M. A. (1997). Cloning and characterization of a mammalian proton-coupled metal-ion transporter. Nature 388, 482-488	[Medline]
HEBERT, S. C. & ANDREOLI, T. E. (1984). Control of NaCl transport in thick ascending limb. American Journal of Physiology 246, F745-756	[Medline]
HEVESY, G., HOFER, E. & KROGH, A. (1935). The permeability of the skin of frogs to water as determined by D₂O and H₂O. Skandinavisches Archiv fuer Physiologie 72, 199-214
HEYMANN, J. B., AGRE, P. & ENGEL, A. (1998). Progress on the structure and function of aquaporin 1. Journal of Structural Biology 121, 191-206	[Medline]
HICKS, R. M., KETTERER, B. & WARREN, R. C. (1974). The ultrastructure and chemistry of the luminal plasma membrane of the mammalian urinary bladder: a structure with low permeability to water and ions. Philosophical Transactions of the Royal Society B 268, 23-38
HILL, W. G., RIVERS, R. L. & ZEIDEL, M. L. (1999). Role of leaflet asymmetry in the permeability of model biological membranes to protons, solutes, and gases. Journal of General Physiology 114, 405-414	[Abstract/Full Text]
ITO, S. (1987). Functional gastric morphology. In Physiology of the Gastrointestinal Tract, ed. JOHNSON, L., pp. 817-851. Raven Press, New York
JUNG, J. S., PRESTON, G. M., SMITH, B. L., GUGGINO, W. B. & AGRE, P. (1994). Molecular structure of the water channel through aquaporin CHIP. The hourglass model. Journal of Biological Chemistry 269, 14648-14654	[Abstract]
KIKERI, D., SUN, A., ZEIDEL, M. L. & HEBERT, S. C. (1989). Cell membranes impermeable to NH₃. Nature 339, 478-480	[Medline]
KOEFOED-JOHNSEN, V. & USSING, H. H. (1953). The contributions of diffusion and flow to the passage of D₂O through living membranes. Acta Physiologica Scandinavica 28, 60-76
LANDE, M. B., DONOVAN, J. M. & ZEIDEL, M. L. (1995). The relationship between membrane fluidity and permeabilities to water, solutes, ammonia and protons. Journal of General Physiology 106, 67-84	[Abstract]
MACEY, R. I. (1984). Transport of water and urea in red blood cells. American Journal of Physiology 246, C195-C203	[Medline]
MULLER-BERGER, S., DUCOUDRET, O., DIAKOV, A. & FRÖMTER, E. (2001). The renal Na-HCO₃-cotransporter expressed in Xenopus laevis oocytes: change in stoichiometry in response to elevation of cytosolic Ca²⁺ concentration. Pflügers Archiv 442, 718-728	[Medline]
MURATA, K., MITSUOKA, K., HIRAI, T., WALZ, T., AGRE, P., HEYMANN, J. B., ENGEL, A. & FUJIYOSHI, Y. (2000). Structural determinants of water permeation through aquaporin-1. Nature 407, 599-605	[Medline]
NAKHOUL, N. L., DAVIS, B. A., ROMERO, M. F. & BORON, W. F. (1998). Effect of expressing the water channel aquaporin-1 on the CO₂ permeability of Xenopus oocytes. American Journal of Physiology 274, C543-548	[Medline]
NIELSEN, S. & AGRE, P. (1995). The aquaporin family of water channels in kidney. Kidney International 48, 1057-1068	[Medline]
NIELSEN, S., FROKIAER, J., MARPLES, D., KWON, T. H., AGRE, P. & KNEPPER, M. A. (2002). Aquaporins in the kidney: from molecules to medicine. Physiological Reviews 82, 205-244	[Abstract/Full Text]
NIELSEN, S., NAGELHUS, E. A., AMIRY-MOGHADDAM, M., BOURQUE, C., AGRE, P. & OTTERSEN, O. P. (1997). Specialized membrane domains for water transport in glial cells: high-resolution immunogold cytochemistry of aquaporin-4 in rat brain. Journal of Neuroscience 17, 171-180	[Abstract/Full Text]
NIEMIETZ, C. M. & TYERMAN, S. D. (2000). Channel-mediated permeation of ammonia gas through the peribacteroid membrane of soybean nodules. FEBS Letters 465, 110-114	[Medline]
PAGANELLI, C. V. & SOLOMON, A. K. (1957). The rate of exchange of tritiated water across the human red cell membrane. Journal of General Physiology 41, 259-277
PRASAD, G. V., COURY, L. A., FIN, F. & ZEIDEL, M. L. (1998). Reconstituted aquaporin 1 water channels transport CO₂ across membranes. Journal of Biological Chemistry 273, 33123-33126	[Abstract/Full Text]
PRESCOTT, D. M. & ZEUTHEN, E. (1953). Comparison of water diffusion and water filtration across cell surfaces. Acta Physiologica Scandinavica 28, 77-94
PRESTON, G. M. & AGRE, P. (1991). Isolation of the cDNA for erythrocyte integral membrane protein of 28 kilodaltons: member of an ancient channel family. Proceedings of the National Academy of Sciences of the USA 88, 11110-11114	[Abstract]
PRESTON, G. M., CARROLL, T. P., GUGGINO, W. B. & AGRE, P. (1992). Appearance of water channels in Xenopus oocytes expressing red cell CHIP28 protein. Science 256, 385-387
PRESTON, G. M., JUNG, J. S., GUGGINO, W. B. & AGRE, P. (1993). The mercury-sensitive residue at cysteine 189 in the CHIP28 water channel. Journal of Biological Chemistry 268, 17-20	[Abstract]
RIVERS, R. L., DEAN, R. M., CHANDY, G., HALL, J. E., ROBERTS, D. M. & ZEIDEL, M. L. (1997). Functional analysis of nodulin 26, an aquaporin in soybean root nodule symbiosomes. Journal of Biological Chemistry 272, 16256-16261	[Abstract/Full Text]
ROOS, A. & BORON, W. F. (1981). Intracellular pH. Physiological Reviews 61, 296-434	[Medline]
ROUGHTON, F. J. W. (1959). Diffusion and simultaneous chemical reaction velocity in haemoglobin solutions and red cell suspensions. Progress in Biophysical Chemistry 9, 56-104
SCHÄFFNER, A. R. (1998). Aquaporin function, structure, and expression: are there more surprises to surface in water relations? Planta 204, 131-139	[Medline]
SIMON, S. A. & GUTKNECHT, J. (1980). Solubility of carbon dioxide in lipid bilayer membranes and organic solvents. Biochimica et Biophysica Acta 596, 352-358	[Medline]
SINGH, S. K., BINDER, H. J., GEIBEL, J. P. & BORON, W. F. (1995). An apical permeability barrier to NH₃/NH₄⁺ in isolated, perfused colonic crypts. Proceedings of the National Academy of Sciences of the USA 92, 11573-11577	[Abstract]
SWEET, G., GANDOR, C., VOEGELE, R., WITTEKINDT, N., BEUERLE, J., TRUNIGER, V., LIN, E. C. & BOOS, W. (1990). Glycerol facilitator of Escherichia coli: cloning of glpF and identification of the glpF product. Journal of Bacteriology 172, 424-430	[Medline]
VINCE, J. W. & REITHMEIER, R. A. (1998). Carbonic anhydrase II binds to the carboxyl terminus of human band 3, the erythrocyte Cl^-/HCO₃^- exchanger. Journal of Biological Chemistry 273, 28430-28437	[Abstract/Full Text]
VINCE, J. W. & REITHMEIER, R. A. (2000). Identification of the carbonic anhydrase II binding site in the Cl^-/HCO₃^- anion exchanger AE1. Biochemistry 39, 5527-5533	[Medline]
WAISBREN, S. J., GEIBEL, J. P., BORON, W. F. & MODLIN, I. M. (1994a). Luminal perfusion of isolated gastric glands. American Journal of Physiology 266, C1013-1027	[Medline]
WAISBREN, S. J., GEIBEL, J. P., MODLIN, I. M. & BORON, W. F. (1994b). Unusual permeability properties of gastric gland cells. Nature 368, 332-335	[Medline]
WALTER, A. & GUTKNECHT, J. (1986). Permeability of small nonelectrolytes through lipid bilayer membranes. Journal of Membrane Biology 90, 207-217	[Medline]
YANG, B., FUKUDA, N., VAN HOEK, A., MATTHAY, M. A., MA, T. & VERKMAN, A. S. (2000). Carbon dioxide permeability of aquaporin-1 measured in erythrocytes and lung of aquaporin-1 null mice and in reconstituted liposomes. Journal of Biological Chemistry 275, 2686-2692	[Abstract/Full Text]
ZHAO, J., HOGAN, E. M., BEVENSEE, M. O. & BORON, W. F. (1995). Out-of-equilibrium CO₂/HCO₃^- solutions and their use in characterizing a new K/HCO₃ cotransporter. Nature 374, 636-639	[Medline]
ZHOU, Y., BOUYER, P., VIRKKI, L., DAVIS, B. A. & BORON, W. F. (2000). Effect of Zn or DIDS on HCO₃^- reabsorption by S2 proximal tubule. FASEB Journal 14, A355
ZHOU, Y., ZHAO, J., BOUYER, P. & BORON, W. F. (1999). Effect of independently varying basolateral [CO₂] and [HCO₃^-] on HCO₃^- reabsorption by rabbit S2 proximal tubule. FASEB Journal 13, A65

				N. Uehlein and R. Kaldenhoff Aquaporins and Plant Leaf Movements Ann. Bot., January 1, 2008; 101(1): 1 - 4. [Abstract] [Full Text] [PDF]

				Y. Wang and E. Tajkhorshid Molecular Mechanisms of Conduction and Selectivity in Aquaporin Water Channels J. Nutr., June 1, 2007; 137(6): 1509S - 1515S. [Abstract] [Full Text] [PDF]

				W. F. Boron Acid-Base Transport by the Renal Proximal Tubule J. Am. Soc. Nephrol., September 1, 2006; 17(9): 2368 - 2382. [Abstract] [Full Text] [PDF]