Effect of PCMBS on CO2 permeability of Xenopus
oocytes expressing aquaporin 1 or its C189S mutant
GORDON J. COOPER AND WALTER F. BORON
Department of Cellular and Molecular Physiology,
Yale University School of Medicine, New Haven, Connecticut 06520
p-chloromercuriphenylsulfonic acid; intracellular pH; carbon
dioxide
of this century, a dogma of biology has
been that small molecules such as H2O, CO2, O2, and
NH3 cross cell membranes by passing between membrane lipids. However, beginning in 1935 (11) and
continuing in 1953 (14, 18), evidence began to point to
the existence of ‘‘pores’’ that can mediate fluxes of H2O,
at least in certain cell membranes. The cloning of the
H2O channel aquaporin (AQP) 1 (19) and the demonstration that expressing AQP1 in Xenopus oocytes markedly increases the H2O permeability of these cells (20)
confirmed that H2O does pass through channels. The
cloning of AQP2 from the renal collecting duct (7), as
well as its association with the membrane vesicles
inserted into the cell apical membrane under the
FROM THE TURN
The costs of publication of this article were defrayed in part by the
payment of page charges. The article must therefore be hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
direction of vasopressin (6), showed how H2O transport
could be regulated. The only other example of a membrane protein that can mediate substantial transport of
a small, neutral molecule is the UT family of urea
transporters (28).
As far as gas transport is concerned, the first suggestion that gases might not cross all membranes simply
by dissolving in the membrane lipid was the observation (13) that NH3/NH1
4 , when applied to the lumen of
the renal thick ascending limb, acidifies the cells (due
to NH1
4 uptake) rather than alkalinizes them (due to
NH3 uptake). Later work demonstrated that the apical
membranes of gastric-gland cells have an immeasurably low permeability to both NH3 and NH1
4 as well as
to both CO2 and HCO2
(27)
and
that
apical
membranes
3
of colonic-crypt cells similarly have no measurable
2
permeability to NH3 or NH1
4 (24) or to CO2 or HCO3
(25).
The above studies demonstrate that cell membranes
exist that are impermeable to H2O or gases and that
either expressing AQPs or inserting preexisting AQPs
into the membrane can increase the permeability to
H2O. However, still unresolved is the question of
whether channels might increase the gas permeability
of a membrane in the same way that the AQPs increase
H2O permeability. A recent study on Xenopus oocytes
injected with carbonic anhydrase (CA) demonstrated
that expressing AQP1 increases by ,40% the rate at
which exposing the cell to CO2 causes intracellular pH
(pHi ) to fall (15). Although it is possible that AQP1
serves as a conduit for CO2, this previous study could
not rule out the possibility that overexpressing AQP1
increases the apparent permeability to CO2 by 1)
stimulating injected CA, 2) enhancing the CO2 solubility of the membrane’s lipid, or 3) increasing the expression of a native ‘‘gas channel.’’
The aim of the present study was to determine
whether CO2 can actually pass through AQP1. We
eliminated CA as a variable by performing the experiments in the absence of injected CA; oocytes have no
detectable, native CA activity (16). We found that the
CO2 permeability of oocytes increases with the degree
of AQP1 expression and that the AQP1-dependent CO2
permeability is blocked by the mercury derivative
p-chloromercuriphenylsulfonic acid (PCMBS). Mercurials are known to reduce the H2O permeability of AQP1
(20) and do so by interacting with the cysteine at
position 189 (21). We found that the C189S mutant of
AQP1 induces the same CO2 permeability as the wildtype AQP1 but that the C189S-dependent increase in
CO2 permeability is insensitive to PCMBS.
0363-6143/98 $5.00 Copyright r 1998 the American Physiological Society
C1481
Downloaded from http://ajpcell.physiology.org/ by 10.220.33.6 on June 16, 2017
Cooper, Gordon J., and Walter F. Boron. Effect of
PCMBS on CO2 permeability of Xenopus oocytes expressing
aquaporin 1 or its C189S mutant. Am. J. Physiol. 275 (Cell
Physiol. 44): C1481–C1486, 1998.—A recent study on Xenopus oocytes [N. L. Nakhoul, M. F. Romero, B. A. Davis, and W.
F. Boron. Am. J. Physiol. 274 (Cell Physiol. 43): C543–548,
1998] injected with carbonic anhydrase showed that expressing aquaporin 1 (AQP1) increases by ,40% the rate at which
exposing the cell to CO2 causes intracellular pH to fall. This
observation is consistent with several interpretations. Overexpressing AQP1 might increase apparent CO2 permeability
by 1) allowing CO2 to pass through AQP1, 2) stimulating
injected carbonic anhydrase, 3) enhancing the CO2 solubility
of the membrane’s lipid, or 4) increasing the expression of a
native ‘‘gas channel.’’ The purpose of the present study was to
distinguish among these possibilities. We found that expressing the H2O channel AQP1 in Xenopus oocytes increases the
CO2 permeability of oocytes in an expression-dependent
fashion, whereas expressing the K1 channel ROMK1 has no
effect. The mercury derivative p-chloromercuriphenylsulfonic
acid (PCMBS), which inhibits the H2O movement through
AQP1, also blocks the AQP1-dependent increase in CO2
permeability. The mercury-insensitive C189S mutant of AQP1
increases the CO2 permeability of the oocyte to the same
extent as does the wild-type channel. However, the C189Sdependent increase in CO2 permeability is unaffected by
treatment with PCMBS. These data rule out options 2–4
listed above. Thus our results suggest that CO2 passes
through the pore of AQP1 and are the first data to demonstrate that a gas can enter a cell by a means other than
diffusing through the membrane lipid.
C1482
AQP1 AS A CO2 CHANNEL
METHODS
RESULTS
Determination of oocyte viability. We assessed the
viability of the oocytes from the measurement of Vm,
rejecting oocytes if Vm was more positive than 30 mV.
We found that neither removing the vitelline membrane nor expressing AQP1 had a statistically significant effect on Vm. The mean Vm of H2O-injected oocytes
was 52.2 6 3.3 mV (n 5 14) with the vitelline membrane intact and 48.8 6 2.1 (n 5 36) with the vitelline
membrane removed. AQP1-expressing oocytes had a
Vm of 49.0 6 4.9 mV (n 5 10) with the vitelline
membrane intact and 53.2 6 2.2 mV (n 5 34) with the
vitelline membrane removed.
Effect of AQP1 expression on rate of CO2-induced
acidification. In oocytes in which vitelline membranes
had been removed, the rate of CO2-induced acidification
increased in an expression-dependent manner (Fig. 1,
A-C). Figure 1A shows experiments from three individual oocytes with different expression levels, as judged
by lysis time in deionized H2O. As the level of expression increases, the rate of acidification also increases.
Data from 34 experiments showed a good correlation
between lysis time and rate of acidification (Fig. 1B).
We grouped these data into three bins on the basis of
cell lysis time: AQP1/Hi for lysis times ,60 s, AQP1/
Mid for lysis times of 61–120 s, and AQP1/Low for lysis
times of 121–180 s. Compared with the rate of acidification in H2O-injected controls (Fig. 1C), the rate was
significantly higher in the AQP1/Hi group and AQP1/
Mid group. The rate of acidification for the AQP1/Low
Downloaded from http://ajpcell.physiology.org/ by 10.220.33.6 on June 16, 2017
Solutions. The control amphibian Ringer solution ND-96
contained (in mM) 96 NaCl, 2 KCl, 1 MgCl2, 1.8 CaCl2, and 5
HEPES (pH adjusted to 7.50 using NaOH or HCl). In
HCO2
3 -buffered solutions, 10 mM NaCl was replaced with
NaHCO3, and the solution was equilibrated with a gas
mixture of 1.5% CO2-98.5% O2 to gave a solution pH of 7.50.
In the 0 Ca21 ND-96 Ringer solution used in oocyte isolation,
CaCl2 was omitted and replaced by NaCl. The OR3 culture
medium contained 6.85 g/l of Leibovitz L-15 cell culture
medium (GIBCO-BRL, Gaithersburg, MD), 10,000 U/ml penicillin G sodium, 10,000 mg/ml streptomycin sulfate (GIBCOBRL), and 5 mM HEPES (pH adjusted to 7.50 with NaOH or
HCl). The osmolarity of all solutions was adjusted to between
195 and 200 mosmol/l with NaCl or H2O. PCMBS (Sigma, St.
Louis, MO) was powdered into solutions to give a final
concentration of 1 mM on the day of experiments.
cRNA preparation. The cDNA encoding AQP1 and C189S
was a generous gift of Dr. P. Agre. In both cases, the cDNA had
been inserted in the plasmid pXbG-ev1, flanked by the 38 and
58 untranslated regions of the Xenopus b-globin gene (20).
The plasmid was linearized by cutting with XBA1, and cRNA
was made using a T3 polymerase kit (Ambion, Austin, TX).
The cRNA encoding ROMK1 was a gift from Drs. Carmel
McNicholas and Gerhard Giebisch.
Oocyte preparation. Stage V–VI oocytes were isolated from
Xenopus laevis. Toads were anesthetized by immersion in
0.2% tricaine (buffered to pH 7.50 with 5 mM HEPES). After
the anesthetized animals were placed on ice, a 1-cm incision
was made in the abdominal wall, and a lobe of an ovary was
removed. The abdominal muscle was sutured using 5–0 cat
gut, and the skin was sutured using 6–0 silk. The removed
ovarian tissue was cut into sections of ,5 3 5 3 5 mm and
agitated in 0 Ca21 ND-96 for 1 h. The oocytes were incubated
two times for 20 min in 0 Ca21 ND-96 containing 2 mg/ml type
I collagenase (Sigma), separated by a 15-min wash in 0 Ca21
ND-96. At the end of the second collagenase treatment, the
oocytes were washed in 0 Ca21 ND-96 for 30 min. This was
followed by a further 30-min wash in the standard Ca21containing ND-96. The oocytes were then transferred to OR3
media and sorted by size and stage. The sorted oocytes were
kept in OR3 at 18°C. On the day after isolation, oocytes were
injected with 50 nl of sterile H2O (GIBCO-BRL) or 50 nl of a
cRNA solution containing either 1 ng of AQP1 cRNA, 1 ng of
C189S cRNA, or 12.5 ng of ROMK1 cRNA. Oocytes were used
in physiological experiments 3–10 days after injection. Just
before each experiment, the vitelline membrane of a selected
oocyte was removed by manual dissection. All experiments
were performed at 22°C.
Quantification of AQP1 expression. The expression level of
AQP1 or its C189S mutant was determined by measuring the
time taken for the oocyte to ‘‘lyse’’ when we switched the
bathing solution to deionized H2O. Oocytes with intact vitelline membranes swell and explode dramatically (20). Devitellinized oocytes also swell but, instead of exploding, gently
emit a plume of debris (‘‘lysis’’). Because cell pressure probably rises less during cell swelling in devitellinized oocytes, it
is likely that the relationship between lysis time and H2O
permeability is more linear than in oocytes with an intact
vitelline membrane. We did not note a substantial change in
lysis time during the 3–10 days after injection of H2O or
cRNA.
Electrophysiological measurements. Oocytes were transferred to a perfusion chamber that had a long, thin channel
and a volume of ,200 µl. Solutions constantly flowed down
the length of the channel, delivered by syringe pumps (Harvard Apparatus, South Natick, MA) at a rate of 4 ml/min. The
switching of solutions was controlled by pneumatically oper-
ated valves (Clippard Instrument Laboratory, Cincinnati,
OH). Two thin strands of nylon were stretched across the top
of the channel and formed an ‘‘X’’ when viewed from the top of
the chamber. The solution flowing down the channel pushed
the oocyte against the notch formed by the intersection of the
nylon strands and thus held the oocyte in place.
pHi and membrane potential (Vm ) were measured using
microelectrodes, as described in detail previously (22). Briefly,
electrodes were pulled from borosilicate capillary glass
(Warner Instruments, West Haven, CT) using a horizontal
puller (Sutter Instrument, Novato, CA). The Vm electrodes
were filled with 3 M KCl and had resistances of 2–5 MV. The
pH microelectrodes were silanized at 200°C for 5 min, using
bis-di-(methylamino)dimethylsilane (Fluka Chemical, Ronkonkoma, NY). The tips of the electrodes were filled with
hydrogen ionophore I cocktail B (Fluka) and then backfilled
with a buffer containing 0.04 M KH2PO4, 0.023 M NaOH, and
0.150 M NaCl (pH 7.0). The electrodes were calibrated at pH 6
and 8 and had slopes of 55–60 mV/pH unit. An additional,
single-point adjustment was made in the perfusion chamber
by calibrating the electrode against the pH 7.50 ND-96
solution, just before impalement.
Both the Vm and pH microelectrodes were connected to
high-impedance electrometers. The voltage due to pH was
obtained from the subtraction of the pH and Vm signals (see
Ref. 23). Vm was the difference between the Vm electrode and
an external calomel reference electrode. The signals were
digitized and recorded by an 80486-based personal computer.
Statistics. The initial rate of intracellular acidification was
calculated by linear regression. Significance is assumed at
the 5% level. Analysis was performed using paired t-tests or
ANOVA, as appropriate, in SigmaStat for Windows. If ANOVA
indicated a difference, comparison between groups was performed using the Student-Newman-Keuls method.
AQP1 AS A CO2 CHANNEL
C1483
group was not different from the H2O-injected controls.
As observed in a previous study (15), expression of
AQP1 had no effect on the rate of CO2-induced acidification in oocytes with intact vitelline membranes.
Fig. 1. Effect of expressing aquaporin 1 (AQP1) on CO2 permeability
of Xenopus oocytes. A: intracellular acidification caused by exposing
devitellinized oocytes, previously injected with AQP1 cRNA, to 1.5%
CO2/10 mM HCO2
3 . pHi, intracellular pH. Extracellular pH was
always 7.5. Trace i, oocyte with a low AQP1 expression level (lysis
time in deionized H2O 165 s) and low CO2-induced acidification rate
(9.6 3 1024 pH · s21 ); trace ii, oocyte with a moderate expression level
(lysis time 67 s) and moderate acidification rate (25.1 3 1024 pH · s21 );
trace iii, oocyte with a high expression level (lysis time 35 s) and high
acidification rate (35.8 3 1024 pH · s21 ). B: relationship between cell
lysis time (t; which is inversely related to AQP1 expression) and
acidification rate for 34 devitellinized oocytes injected with AQP1
cRNA. Open circles represent individual oocytes, and closed circles
represent means for oocytes with lysis times of 0–60, 61–120, and
121–180 s. SE bars are shown if SE exceeded the size of the circle.
Line plotted through the data is the result of linear regression [y 5
(20.144x 1 35.8) 3 1024, r2 5 0.82]. C: CO2-induced acidification rate
in groups of oocytes. For each group, no. of observations is shown in
parentheses, and SE is indicated. Oocytes with vitelline membranes
removed are indicated by cross-hatched bars, and those with vitelline
membranes intact are indicated by open bars. H2O, H2O-injected
controls; AQP1/Hi, lysis time of 0–60 s; AQP1/Mid, lysis time of
61–120 s; AQP1/Low, lysis time of 121–180 s; ROMK1, injected with
cRNA encoding the ROMK1 K1 channel; AQP1, injected with AQP1
cRNA. * Acidification rate for group 2 was significantly greater than
that for all other groups by ANOVA (degrees of freedom 5 8,105, F 5
38.3). † Acidification rate for group 3 was significantly different from
all other groups (degrees of freedom 5 8,105, F 5 38.3). There are no
significant differences among the acidification rates in groups 1 and 4–7.
Effect of expressing ROMK1 on rate of CO2-induced
acidification. In oocytes expressing the K1 channel
ROMK1 (12), the rate of CO2-induced acidification was
not significantly different from that in controls. Expression of ROMK1 caused a shift in the resting Vm from
48.8 6 2.1 mV (n 5 36) to 96.8 6 3.0 mV (n 5 6, P ,
0.0001), which is close to the predicted equilibrium
potential for K1 under the conditions used.
Effect of PCMBS on the rate of CO2-induced acidification. If CO2 passes through AQP1, then one might
expect the CO2-induced acidification to be inhibited by
mercurial compounds, which are known to block the
H2O permeability of AQP1 (20). We examined this
possibility by twice exposing an oocyte to 1.5% CO2,
first in the absence and then in the presence of PCMBS,
an organic mercury derivative. In H2O-injected oocytes
(Fig. 2A), a 15-min incubation in 1 mM PCMBS produced a small but statistically significant decrease in
the rate of the CO2-induced acidification1 compared
with the matched control (P , 0.0001, paired t-test). In
oocytes expressing AQP1 (Fig. 2B), a 15-min incubation
in PCMBS also significantly reduced the rate of the
CO2-induced acidification (P 5 0.0009). However, the
magnitude of the PCMBS effect was 3.5-fold greater in
1 In all three groups of oocytes treated with PCMBS (i.e., H O2
injected oocytes, oocytes expressing AQP1, and oocytes expressing
the C189S mutant), a 15-min exposure to PCMBS had no significant
effect on the pHi prevailing just before application of CO2. However a
15-min exposure to PCMBS produced a significant depolarization in
all three groups. The magnitude of this depolarization was the same
for each group: 12.9 6 4.4 mV (n 5 6) for H2O-injected oocytes, 17.2 6
1.7 mV (n 5 6) for AQP1-expressing oocytes, and 15.1 6 3.5 mV (n 5
6) for C189S-expressing oocytes (degrees of freedom 5 2,17, F 5 0.45,
P 5 0.63).
Downloaded from http://ajpcell.physiology.org/ by 10.220.33.6 on June 16, 2017
Fig. 2. Effect of p-chloromercuriphenylsulfonic acid (PCMBS) on
CO2-induced acidification. A: H2O-injected control oocyte. Two records represent a paired experiment on a single oocyte exposed to
1.5% CO2/10 mM HCO2
3 before and after a 15-min exposure to 1 mM
PCMBS. B: AQP1-expressing oocyte.
C1484
AQP1 AS A CO2 CHANNEL
AQP1-expressing oocytes than in H2O-injected controls. As expected, the treatment with PCMBS also
reduced the H2O permeability of AQP1-expressing oocytes.2
Effect of expressing C189S on the rate of CO2-induced
acidification. The C189S mutant of AQP1 has a normal
H2O permeability but is immune to inhibition by mercurial derivatives (21). The triangles in Fig. 3A represent
CO2-induced acidification rates in oocytes with different levels of C189S expression. These data fall on the
same line (taken from Fig. 1B) as that derived for
oocytes expressing wild-type AQP1. Moreover, comparing oocytes that lysed in the middle time range (60–
120 s), we find no significant difference in the CO2induced acidification rates between oocytes expressing
wild-type AQP1 (21.7 6 1.1 3 1024 pH · s21, n 5 15) and
2 From the rates of CO -induced acidification in the absence of
2
PCMBS, as well as the regression line describing the relationship
between lysis time and acidification rate (Fig. 1B), we were able to
predict the lysis time in deionized H2O, if the oocyte had not
subsequently been treated with PCMBS. In AQP1-expressing oocytes, this computed control lysis time was more than threefold
greater than the actual lysis time after treatment with PCMBS.
However, in oocytes expressing the C189S mutant, the computed
control and actual PCMBS lysis times were identical.
those expressing the C189S mutant (23.6 6 0.4 3 1024
pH · s21, n 5 7). Thus oocytes expressing the C189S
mutant display the same CO2 permeability properties
as those expressing the wild-type AQP1 channel.
Effect of PCMBS on the rate of CO2-induced acidification in C189S-expressing oocytes. Exposing an oocyte
expressing C189S to 1 mM PCMBS for 15 min produced
a small reduction in the rate of CO2-induced acidification compared with the preceding CO2 exposure in the
same cell (Fig. 3B). PCMBS also did not affect lysis
time.2 In six such paired experiments, PCMBS produced a decrease in the CO2-induced acidification rate
(Fig. 3, C and D) that was statistically significant and
virtually identical to that produced by PCMBS in
H2O-injected control oocytes (Fig. 3, C and D). Indeed,
there was no difference between the effects of PCMBS
in the control and C189S groups as judged by one-way
ANOVA. However, the effect of PCMBS on the rate of
the CO2-induced acidification was significantly greater
in AQP1 oocytes (Fig. 3, C and D) than in either the
control or C189S groups (degrees of freedom 5 2,17;
F 5 20.4; P , 0.005 in both cases).
Determination of CO2 permeability. Of the CO2 molecules entering the cell, the fraction that forms H2CO3
and then dissociates into H1 and HCO2
3 is K/(H 1 K),
Downloaded from http://ajpcell.physiology.org/ by 10.220.33.6 on June 16, 2017
Fig. 3. Effect of expressing the AQP1 mutant C189S on the CO2 permeability of devitellinized Xenopus oocytes.
A: relationship between lysis time and acidification rate for 11 devitellinized oocytes injected with C189S cRNA.
Open triangles represent individual oocytes. Line plotted through the data is the regression line from Fig. 1B
(oocytes expressing wild-type AQP1). B: effect of PCMBS on CO2-induced acidification in C189S-expressing oocytes.
Protocol was the same as for Fig. 2, A and B. C: effect of PCMBS on absolute initial rates of CO2-induced acidification
in H2O-injected oocytes (H2O), oocytes injected with AQP1 cRNA (AQP1), and oocytes injected with cRNA encoding
the AQP1 mutant C189S (C189S). Open bars represent data obtained in the absence of PCMBS, and hatched bars
represent data obtained after a 15-min incubation in PCMBS. No. of observations is shown in parentheses, and SE
is indicated. * Rate of acidification is significantly slower than for the control group (absence of PCMBS in the same
oocytes) as judged by a paired t-test. 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. Data were obtained on
3 groups of oocytes displayed in C. For each group, no. of observations is shown in parentheses, and SE is indicated.
* Change in rate was significantly larger compared with the 2 other groups as judged by one-way ANOVA.
AQP1 AS A CO2 CHANNEL
DISCUSSION
The oocyte is a convenient system for studying
membrane CO2 permeability. Our approach was to
measure the rate of intracellular acidification produced
by exposing an oocyte to a solution, at constant pH,
containing 1.5% CO2. As the CO2 enters the cell, it
slowly combines with H2O to form H2CO3. The H2CO3
then rapidly dissociates to yield HCO2
3 and the proton
that we detect. CA catalyzes CO2 1 OH2 k HCO2
3,
which has the effect of bypassing the slow step in the
above reaction sequence. Because the oocyte is devoid
of both native CA and HCO2
3 transporters (22), the
initial rate at which pHi falls in the presence of a CO2/
HCO2
3 solution is proportional to rate at which CO2
enters the cell.
In agreement with a previous study by Nakhoul et al.
(15), we observed that expression of AQP1, by itself,
had no effect on the rate of CO2-induced acidification.
Those authors were able to unmask an AQP1-dependent increase in CO2 permeability by injecting CA
protein into oocytes (15). However, this result left open
the possibility that AQP1 acted by stimulating the
injected CA. In the present study, we did not inject the
oocytes with CA, and thus the oocytes had negligible
CA activity. We found, instead, that we could unmask
the effect of AQP1 by stripping away the vitelline
membrane. Thus our data rule out the possibility that,
in the experiments of Nakhoul et al. (15), AQP1 acted
by stimulating injected CA. Our data also suggest that
there are two rate-limiting steps in the intracellular
acidification produced by exposing an oocyte to CO2 as
follows: 1) the flux of CO2 across the vitelline membrane and 2) the conversion of intracellular H2CO3 to
1
HCO2
3 and H .
One of our goals was to determine whether the effect
of AQP1 was specifically related to AQP1 or merely a
consequence of inserting large numbers of any channel
into the cell membrane. We found that expressing the
K1 channel ROMK1 produced no alteration in CO2
permeability, indicating that there is at least some
specificity for the class of channel expressed in the cell
membrane.
To address further the issue of whether CO2 passes
through AQP1, we examined the effect of the mercury
derivative PCMBS, which does not permeate cell membranes (26), on the CO2-induced acidification. PCMBS
produced a small reduction in the rate of CO2-induced
acidification in control oocytes, an effect that might be
explained in the following two ways: 1) PCMBS interacts with the membrane lipids, decreasing the permeability of this pathway to CO2, or 2) the Xenopus oocyte
contains a native gas channel that is partially sensitive
to mercurials. PCMBS also decreased the CO2-induced
acidification rate in AQP1-expressing oocytes, but to a
much greater extent (3.5-fold) than in control oocytes.
The above results are consistent with the hypothesis
that CO2 can pass through AQP1 but do not rule out the
possibility that AQP1 is somehow increasing CO2 permeability by either altering the composition of the
membrane lipids or inducing a native CO2 pathway
that is sensitive to PCMBS. To rule out these two
options, we expressed the mercury-insensitive mutant
of AQP1, C189S. This mutant increased the H2O and
CO2 permeability to the same extent as the wild-type
channel (Fig. 3A). However, neither the C189S-dependent H2O permeability nor the C189S-dependent CO2
permeability was sensitive to PCMBS (Fig. 3, B-D).
Thus, unless the C189S mutation not only knocks out
the ability of mercury to inhibit the H2O permeability of
AQP1 but also knocks out AQP1’s ability to alter
membrane lipid composition and/or recruit native gas
channels, we can conclude that CO2 does indeed pass
through AQP1. These data, complemented by those in
the earlier paper by Nakhoul (15), provide the first
evidence for a gas crossing a membrane by a route other
than the membrane lipid.
The conclusion that CO2 can pass through the AQP1
channel raises the following two questions. 1) Compared with H2O, do significant amounts of CO2 pass
through the channel? 2) Are movements of CO2 through
AQP1 physiologically significant? To answer the first
question, we have estimated the AQP1-dependent CO2
permeability of the oocyte and found it to be 18.4 3 1024
cm · s21 (see Determination of CO2 permeability). The
AQP1-dependent osmotic H2O permeability of oocytes
is 129 3 1024 cm · s21 (20). Thus the CO2-to-H2O permeability ratio is ,0.14. This figure is orders of magnitude
larger than for any other solute suggested to pass
through an AQP-type channel. For example, in AQP1
the H2O-to-solute permeability ratio is ,0.0008 for glycerol
and ,0.0001 for urea (1). Thus AQP1 is far more permeable to CO2 than to any other molecule aside from H2O.
The second question must still be answered. We
would expect AQP1-dependent CO2 permeability to be
most important in cells with a low intrinsic CO2 permeability, those with high levels of AQP1 expression,
and/or cells that mediate very high CO2 fluxes. Indeed,
AQP1 is present at high levels in tissues or cells with
high gas-transport rates: pulmonary capillary endothelium (17), renal proximal tubule (4), choroid plexus,
Downloaded from http://ajpcell.physiology.org/ by 10.220.33.6 on June 16, 2017
where H is intracellular H1 concentration and K is the
apparent dissociation constant (i.e., 7.24 3 1027 ) for
1
the equilibrium CO2 1 H2O k HCO2
3 1 H (8). Thus the
CO2 influx (JCO2) is the product of (H 1 K)/K and the
rate of intracellular H1 formation (JH; see Ref. 2). JH is
the product of the initial rate of intracellular acidification on exposure to CO2 (dpHi/dt), the intrinsic intracellular buffering power, and the volume-to-surface ratio
estimated by Preston et al. (20) to be 0.02 cm. From the
steady-state CO2-induced pHi decrease evoked by applying 1.5% CO2 (23), we obtained a mean intrinsic
buffering power of 19.9 6 1.0 mM (n 5 25) in the
present experiments. There is no evidence that resting
oocytes express any significant acid-base transporters,
suggesting that our estimates of intrinsic buffering
power are reliable. The CO2 permeability is JCO2 divided by the initial transmembrane CO2 concentration
gradient. We assumed that the extracellular solution
was equilibrated with 1.5% CO2, that the intracellular
solution was equilibrated with 0.03% CO2, and that the
CO2 solubility coefficient was 0.0364 mM/mmHg (9).
C1485
C1486
AQP1 AS A CO2 CHANNEL
placenta (10), and erythrocytes (19). It is intriguing to
speculate that AQP1 homologues with no known function, such as major intrinsic protein, which is the major
membrane protein in the lens of the eye (3), or Nodulin26, which is similarly abundant in the membranes of
legume root nodules (5), may be gas channels.
Received 8 June 1998; accepted in final form 31 August 1998.
REFERENCES
1. Abrami, L., F. Tacnet, and P. Ripoche. Evidence for a glycerol
pathway through aquaporin 1 (CHIP28) channels. Pflügers Arch.
430: 447–458, 1995.
2. Boron, W. F., and P. De Weer. Intracellular pH transients in
squid giant axons caused by CO2, NH3, and metabolic inhibitors.
J. Gen. Physiol. 67: 91–112, 1976.
3. Broekhuyse, R. M., E. D. Kuhlmann, and H. J. Winkens.
Lens membranes VII. MIP is an immunologically specific component of lens fibers membranes and is identical with the 26K band
protein. Exp. Eye Res. 29: 303–313, 1979.
3a. Cooper, G. J., and W. F. Boron. The CO2 permeability of the
AQP1 water channel, expressed in Xenopus oocytes (Abstract). J.
Am. Soc. Nephrol. 8: 16A, 1997.
4. Denker, B. M., B. L. Smith, F. P. Kuhajda, and P. Agre.
Identification, purification, and partial characterization of a
novel Mr 28000 integral membrane protein from erythrocyes and
renal tubules. J. Biol. Chem. 263: 15634–15642, 1988.
5. Fortin, M. G., N. A. Morrison, and D. P. S. Verma. Nodulin-26
a peribacterial membrane nodulin is expressed independently of
the development of the peribacterial compartment. Nucleic Acids
Res. 15: 815–824, 1987.
6. Fushimi, K., S. Sasaki, T. Yamamoto, M. Hayashi, T. Furukawa, S. Uchida, M. Kuwahara, K. Ishibashi, M. Kawasaki, I. Kihara, and F. Marumo. Functional characterization and cell immunolocalization of AQP-CD water channel in
kidney collecting duct. Am. J. Physiol. 267 (Renal Fluid Electrolyte Physiol. 36): F573–F582, 1994.
7. Fushimi, K., S. Uchida, Y. Hara, Y. Hirata, F. Marumo, and
S. Sasaki. Cloning and expression of apical membrane water
channel of rat kidney collecting tubule. Nature 361: 549–552,
1993.
8. Harned, H. S., and F. T. Bonner. The first ionization of
carbonic acid in aqueous solutions of sodium chloride. J. Am.
Chem. Soc. 67: 1026–1031, 1945.
9. Harned, H. S., and R. Davis. The ionization constant of
carbonic acid in water and solubility of carbon dioxide in water
and aqueous salt solutions from 0 to 50°. J. Am. Chem. Soc. 65:
2030–2037, 1943.
Downloaded from http://ajpcell.physiology.org/ by 10.220.33.6 on June 16, 2017
We thank Dr. Paul De Weer for suggesting to us that a channel
might be permeable to CO2. We also are extremely grateful to Dr. P.
Agre for the gift of the cDNA encoding AQP1 and the C189S mutant.
We also thank Drs. B. A. Davis and M. F. Romero for help in
generating the cRNA for AQP1 and C189S, Drs. G. Giebisch and C. M.
McNicholas for providing the cRNA for ROMK1, and E. M. Hogan for
technical assistance.
This work was supported by National Institute of Neurological and
Communicative Disorders and Stroke Grant NS-18400 (to W. F.
Boron) and by the Human Frontiers Scientific Program Organization
(to G. J. Cooper).
Portions of this work have been published in preliminary form
(3a).
Address for reprint requests: G. J. Cooper, Dept. of Cellular and
Molecular Physiology, Yale University School of Medicine, 333 Cedar
St., New Haven, CT 06520.
10. Hasegawa, H., S. C. Lian, W. E. Finkbeiner, and A. S.
Verkman. Extrarenal tissue distribution of CHIP28 water channels by in situ hybridization and antibody staining. Am. J.
Physiol. 266 (Cell Physiol. 35): C893–C903, 1994.
11. Hevesy, G., E. Hofer, and A. Krogh. The permeability of the
skin of frogs to water as determined by D2O and H2O. Scand.
Arch. Physiol. 72: 199–214, 1935.
12. Ho, K., C. G. Nichols, W. J. Lederer, J. Lytton, P. M. Vassilev,
M. V. Kanazirska, and S. C. Hebert. Cloning and expression of
an inwardly rectifying ATP-regulated potassium channel. Nature
362: 31–38, 1993.
13. Kikeri, D., A. Sun, M. L. Zeidel, and S. C. Hebert. Cell
membranes impermeable to NH3. Nature 339: 478–480, 1989.
14. Koefoed-Johnsen, V., and H. H. Ussing. The contributions of
diffusion and flow to the passage of D2O through living membranes. Acta Physiol. Scand. 28: 60–76, 1953.
15. Nakhoul, N. L., M. F. Romero, B. A. Davis, and W. F. Boron.
Effect of expressing the water channel aquaporin-1 on the CO2
permeability of Xenopus oocytes. Am. J. Physiol. 274 (Cell
Physiol. 43): C543–C548, 1998.
16. Nakhoul, N. L., M. F. Romero, A. Waheed, B. A. Davis, G.
Hageman, W. S. Sly, and W. F. Boron. Processing and functional expression of carbonic anhydrase isoforms in Xenopus
laevis oocytes (Abstract). FASEB J. 10: A88, 1996.
17. Nielsen, S., B. L. Smith, E. L. Christensen, and P. Agre.
Distribution of the aquaporin CHIP in secretory and resorptive
epithelia and capillary endothelia. Proc. Natl. Acad. Sci. USA 90:
7275–7279, 1993.
18. Prescott, D. M., and E. Zeuthen. Comparison of water diffusion and water filtration across cell surfaces. Acta Physiol.
Scand. 28: 77–94, 1953.
19. Preston, G. M., and P. Agre. Isolation of the cDNA for
erythrocyte integral membrane protein of 28 kilodaltons: member of an ancient channel family. Proc. Natl. Acad. Sci. USA 88:
11110–11114, 1991.
20. Preston, G. M., T. P. Carroll, W. B. Guggino, and P. Agre.
Appearance of water channels in Xenopus oocytes expressing red
cell CHIP28 protein. Science 256: 385–387, 1992.
21. Preston, G. M., J. S. Jung, W. B. Guggino, and P. Agre. The
mercury-sensitive residue at cysteine 189 in the CHIP28 water
channel. J. Biol. Chem. 268: 17–20, 1993.
22. Romero, M. F., M. A. Hediger, E. L. Boulpaep, and W. F.
Boron. Expression cloning and characterization of a renal
electrogenic Na1/HCO2
3 cotransporter. Nature 387: 409–413,
1997.
23. Roos, A., and W. F. Boron. Intracellular pH. Physiol. Rev. 61:
296–434, 1981.
24. Singh, S. K., H. J. Binder, J. P. Giebel, and W. F. Boron. An
apical permeability barrier to NH3/NH1
4 in isoloated, perfused
colonic crypts. Proc. Natl. Acad. Sci. USA 92: 11573–11577, 1995.
25. Singh, S. K., A. C. Cimini, and W. F. Boron. Permeability
properties of apical vs. basolateral barriers of colonic-crypt
epithelial cells (Abstract). Physiologist 39: 144, 1996.
26. Vanstevenick, J., R. I. Weed, and A. Rothstein. Localization
of erythrocyte membrane sulfhydryl groups essential for glucose
transport. J. Gen. Physiol. 48: 617–632, 1965.
27. Waisbren, S. J., J. Geibel, I. M. Modlin, and W. F. Boron.
Unusual permeability properties of gastric gland cells. Nature
368: 332–335, 1994.
28. You, G., C. P. Smith, Y. Kanai, W.-S. Lee, M. Stelzner, and
M. A. Hediger. Cloning and characterization of the vasopressinregulated urea transporter. Nature 365: 844–847, 1993.