Content uploaded by Gerolf Gros
Author content
All content in this area was uploaded by Gerolf Gros on Jan 26, 2015
Content may be subject to copyright.
The FASEB Journal •Research Communication
CO
2
permeability of cell membranes is regulated by
membrane cholesterol and protein gas channels
Fabian Itel,*
,1
Samer Al-Samir,
‡,1
Fredrik Öberg,
§
Mohamed Chami,
†
Manish Kumar,
储
Claudiu T. Supuran,
¶
Peter M. T. Deen,
#
Wolfgang Meier,* Kristina Hedfalk,
§
Gerolf Gros,
‡,2
and Volker Endeward
‡
*Department of Chemistry and
†
Center for Cellular Imaging and Nanoanalytics (C-CINA),
Biozentrum, Universität Basel, Basel, Switzerland;
‡
Department of Vegetative Physiology, Zentrum
Physiologie, Medizinische Hochschule Hannover, Hannover, Germany;
§
Department of Chemistry
and Molecular Biology, University of Gothenburg, Göteborg, Sweden;
储
Department of Chemical
Engineering, Pennsylvania State University, State College, Pennsylvania, USA;
¶
Dipartimento di
Chimica, Laboratorio di Chimica Bioinorganica, Universita` di Firenze, Florence, Italy; and
#
Department of Physiology, Radboud University Nijmegen Medical Centre, Nijmegen,
The Netherlands
ABSTRACT Recent observations that some mem-
brane proteins act as gas channels seem surprising in
view of the classical concept that membranes generally
are highly permeable to gases. Here, we study the gas
permeability of membranes for the case of CO
2
, using
a previously established mass spectrometric technique.
We first show that biological membranes lacking pro-
tein gas channels but containing normal amounts of
cholesterol (30–50 mol% of total lipid), e.g., MDCK
and tsA201 cells, in fact possess an unexpectedly low
CO
2
permeability (P
CO2
)of⬃0.01 cm/s, which is 2
orders of magnitude lower than the P
CO2
of pure
planar phospholipid bilayers (⬃1 cm/s). Phospholipid
vesicles enriched with similar amounts of cholesterol
also exhibit P
CO2
⬇0.01 cm/s, identifying cholesterol
as the major determinant of membrane P
CO2
. This is
confirmed by the demonstration that MDCK cells de-
pleted of or enriched with membrane cholesterol show
dramatic increases or decreases in P
CO2
, respectively.
We demonstrate, furthermore, that reconstitution of
human AQP-1 into cholesterol-containing vesicles, as
well as expression of human AQP-1 in MDCK cells,
leads to drastic increases in P
CO2
, indicating that gas
channels are of high functional significance for gas
transfer across membranes of low intrinsic gas perme-
ability.—Itel, F., Al-Samir, S., Öberg, F., Chami, M.,
Kumar, M., Supuran, C. T., Deen, P. M. T., Meier,
W., Hedfalk, K., Gros, G., Endeward, V. CO
2
perme-
ability of cell membranes is regulated by membrane
cholesterol and protein gas channels. FASEB J. 26,
5182–5191 (2012). www.fasebj.org
Key Words: aquaporin 1 䡠aquaporin Z 䡠phospholipid vesicles
䡠MDCK cells
It has been a long-standing paradigm in biology
that cell membranes are highly permeable to all gases
(1). This seemed to be supported by the observation
that artificial phospholipid bilayers show extremely
high gas permeability (2–5). In the case of the respira-
tory gases CO
2
and O
2
, the major resistances to the
release of CO
2
from the sites of cellular production to
the blood, and vice versa for O
2
uptake, are well
established for most tissues. These resistances are con-
stituted by 1) the speed of the chemical reactions in red
blood cells and the chemical binding capacity of the
blood for the two gases, and 2) the CO
2
or O
2
diffu-
sivities and the diffusion distances to be overcome
during the process of CO
2
release from mitochondria
into capillary red blood cells, or of O
2
uptake in the
reverse direction (6). Processes 1 and 2 are so fast that
neither in the tissue nor in the normal lung is a
difference expected between tissue or alveolar CO
2
or O
2
partial pressure (pCO
2
/pO
2
) and endcapillary
pCO
2
/pO
2
. Never, until recently, have indications
been observed that cell membranes constitute a signif-
icant barrier to CO
2
transfer, and neither has such a
barrier been reported for O
2
transfer. Just three excep-
tions from this apparent general principle have been
reported for CO
2
. After an early observation of the
apical membrane in the medullary thick ascending
limb of Henle being virtually impermeable to NH
3
(7),
it was reported that the apical membrane of gastric
gland epithelium is poorly permeable to CO
2
(8) and
that this holds also for the apical membrane of the
colonic gland (9) and colonic surface epithelium (10).
1
These authors contributed equally to this work.
2
Correspondence: Zentrum Physiologie, Vegetative Physi-
ologie 4220, Medizinische Hochschule Hannover, 30625
Hannover, Germany. E-mail: gros.gerolf@mh-hannover.de
doi: 10.1096/fj.12-209916
Abbreviations: -OG, n-octyl--d-glycopyranoside; AQP-1,
aquaporin-1; AqpZ, aquaporin Z; D
CO2
,CO
2
diffusion coeffi-
cient; DIDS, 4,4=-diisothiocyano-2,2=-stilbenedisulfonic acid;
EMEM, Eagle’s minimal essential medium; LPR, lipid to protein
ratio; MCD, methyl--cyclodextrin; MDCK, Madin-Darby ca-
nine kidney; PC, l-␣-phosphatidylcholine; pCO
2
,CO
2
partial
pressure; P
CO2
,CO
2
permeability; pO
2
,O
2
partial pressure; PS,
l-␣-phosphatidylserine; Rh, Rhesus; tsA201, transformed human
embryonic kidney 293
5182 0892-6638/12/0026-5182 © FASEB
These observations suggested that there are at least a
few exceptions to the rule that biological membranes
exhibit CO
2
permeabilities as high as those of artificial
phospholipid bilayers.
Another departure from the general view that cell
membranes, through their lipid phases, possess ex-
tremely high gas permeability comes from recent re-
ports indicating that the membrane proteins human
aquaporin-1 (AQP-1), a few other AQP isoforms, and
plant aquaporin (11–15) can increase CO
2
transfer
across membranes. Endeward et al. (16) were the first to
show that in addition to aquaporin, the Rhesus (Rh)
protein can constitute a channel for CO
2
. These studies
were confined to only a small number of experimental
models, essentially human red blood cells and the
Xenopus oocyte expression system. It has remained an
open question, however, how gas channels can be of
functional significance, when most cell membranes
have intrinsic CO
2
permeability (P
CO2
) values similarly
high as those of phospholipid bilayers. This apparent
discrepancy has led and continues to lead to an exten-
sive discussion fundamentally questioning the physio-
logical significance of gas channels (4, 5, 17–22). The
present work, therefore, addresses the hitherto unan-
swered question of what the CO
2
permeability of a
variety of gas channel-free mammalian cells actually is.
We show here that biological membranes containing
normal amounts of cholesterol (30–50 mol% of total
lipid) in fact possess a P
CO2
of ⬃0.01 cm/s, 2 orders of
magnitude lower than that of pure phospholipid bilay-
ers (⬃1 cm/s). Using two experimental models, phos-
pholipid vesicles and Madin-Darby canine kidney (MDCK)
cells, we demonstrate that the low P
CO2
value of biolog-
ical membranes is essentially due to their cholesterol
content. With the same two models, we go on to show
that the gas channel AQP-1, when inserted into a
cholesterol-containing membrane of low background
P
CO2
, does cause a physiologically highly significant
increase in membrane CO
2
permeability. We conclude
that the permeability of biological membranes for CO
2
,
and possibly for other gases, as well, is regulated by two
parameters: cholesterol, which imparts mechanical sta-
bility to the membrane but causes low gas permeability,
and protein gas channels, which can markedly raise the
gas permeability in membranes possessing a high con-
tent of cholesterol.
MATERIALS AND METHODS
P
CO2
from mass spectrometric measurement of
18
O
exchange
With this method, the CO
2
permeability of well-stirred sus-
pensions of isolated cells or of layers of intact, stripped colon
mucosa epithelium is determined (10). The principle is to
expose the cells to a solution containing 25 mM bicarbonate
(1%
18
O-labeled) in 125 mM NaCl. The approach to isotopic
equilibrium takes much longer than establishment of the
chemical equilibrium (which is complete after ⬃1 min) and
requires ⬃2 h when carbonic anhydrase-containing cells at a
cytocrit of ⬃0.05–0.5% are present. During this latter pro-
cess, most of the labeled CO
2
and HCO
3⫺
of the extracellular
space moves to the intracellular space, where both molecules
feed into the carbonic anhydrase-catalyzed CO
2
/HCO
3⫺
hy-
dration-dehydration reaction that runs back and forth con-
tinuously. In each dehydration step, there is a chance of 1/3
that the
18
O label leaves the CO
2
-HCO
3⫺
pool (⬃25 mM)
and enters into the much larger water pool (55 M), from
where it can virtually not come back into the CO
2
-HCO
3⫺
pool. This causes a relatively slow decay of the extracellular
C
18
O
16
O concentration, whose time course we measure by
mass spectrometry via a special inlet system (23). The kinetics
of the decay in C
18
O
16
O (such as shown in Fig. 1A; note the
logarithmic scale of the yaxis) is determined by the cell
permeabilities for CO
2
and HCO
3⫺
and by the intracellular
carbonic anhydrase activity. The latter quantity is determined
separately from lysed cells, and the two former parameters are
calculated from the recorded time courses of C
18
O
16
O decay,
as described previously (10). Phase 1 of the mass spectromet-
ric recordings in Fig. 1Arepresents the kinetics of the decay
of C
18
O
16
O in the absence of cells and/or carbonic anhy-
drase, and allows us to calculate the velocity constant of the
uncatalyzed CO
2
hydration under the conditions of the
experiment. Phase 2 is initiated by the addition of carbonic
anhydrase-containing vesicles (with 30% cholesterol in the
lower curve and 70% cholesterol in the upper curve), or, in
other cases, of carbonic anhydrase-containing cells. The time
course of this second phase is primarily, but not exclusively,
determined by the rapid entry of C
18
O
16
O from the bulk
solution into the vesicles and by the intravesicular carbonic
anhydrase activity. This phase is followed by phase 3, which is
linear in the semilogarithmic representation of Fig. 1Aand
reflects the influences of the membrane permeation by CO
2
and HCO
3⫺
, as well as of carbonic anhydrase activity. The
diffusion and reaction processes underlying phases 2 and 3
have been formulated in 6 differential equations, which are
solved as described to generate a “theoretical” curve of
[C
18
O
16
O] vs. time (10). The solution is then subjected to a
fitting procedure, which yields the optimal values of P
CO2
and
P
HCO3⫺
from a well-defined minimum of the sum of squares
of deviations between experimental and calculated values of
[C
18
O
16
O]. The theory and the mathematical procedure to
extract the cell membrane permeabilities for CO
2
and
HCO
3⫺
from the original mass spectrometric records, and
further experimental details of the method have been de-
scribed extensively (10, 14, 16). It may be noted that the
experiments with intact cells were performed in the presence
ofa5⫻10
⫺5
M concentration of the extracellular carbonic
anhydrase inhibitor 2,4,6-trimethyl-1-[(4-sulfamoylphenyl)ethyl]-
pyridinium perchlorate (C14H17ClN2O6S; ref. 24); this was
done to ensure that any carbonic anhydrase possibly set free
by minor lysis of the cells during stirring did not catalyze CO
2
hydration in the bulk solution of the mass spectrometer
measuring chamber.
We have previously reported the extent to which measure-
ment of the P
CO2
of red blood cell membranes using this
method is affected by extracellular unstirred layers (25). In
these previous experiments, and in all experiments done here
with cells as well as vesicles, the speed of stirring was identical.
As described previously (25), at given values of fluid viscosity
and convection in the suspension, the unstirred layer thick-
ness ␦is proportional to the square root of the characteristic
length ᐉof the cell or vesicle considered. The human red
blood cell, with its diameter of ᐉ⫽7.5 m, exhibited an
unstirred layer thickness of ␦⫽0.5 m. Accordingly, vesicles
with a diameter of 0.15 m are predicted to exhibit an
unstirred layer thickness of 0.5 m⫻(0.15/7.5)
1/2
⫽0.07
m. A water layer of 0.07 m possesses a CO
2
permeability of
P
CO2
⫽D
CO2
/␦⫽(2.4⫻10
⫺5
cm
2
/s)/(0.07⫻10
⫺4
cm) ⫽3.4
cm/s, where D
CO2
is the CO
2
diffusion coefficient of CO
2
in
water (25). In view of a vesicle CO
2
permeability range of
0.001 to 0.1 cm/s, as reported below, the diffusion resistance
5183CO
2
PERMEABILITY OF CELL MEMBRANES
of the unstirred layer is therefore negligible. We apply an
analogous consideration to the MDCK and transformed hu-
man embryonic kidney 293 (tsA201) cells studied in this
work. Both cells have a diameter of ⬃15 m, which leads to a
prediction of the unstirred layer thickness of 0.7 m. The
P
CO2
of this water layer is then 0.34 cm/s, which again
constitutes an insignificant diffusion resistance for CO
2
in
view of the actually measured P
CO2
values of these cells of 0.02
cm/s (see below). We conclude that unstirred layer effects
are not expected to markedly affect the measurements of
P
CO2
as reported here.
Cell culture and cell manipulations
MDCK and tsA201 cells were obtained from the European
Collection of Cell Cultures (ECACC; Porton Down, UK).
MDCK cells were kept in Eagle’s minimal essential medium
(EMEM) with 1% nonessential amino acids (NEAA; both
from Sigma-Aldrich, St. Louis, MO, USA), 10% fetal bovine
serum (FBS; Gold; PAA, Pasching, Austria), 2 mM l-glu-
tamine and penicillin-streptomycin (both from Life Technol-
ogies, Inc.; Invitrogen, Carlsbad, CA, USA) at 37°C and 5%
CO
2
. Cells were grown until no more than 80% confluency
was reached, i.e., the development of polarization of the
MDCK cells along with full confluency was avoided (26).
tsA201 cells were grown in Dulbecco’s modified Eagle me-
dium (DMEM; Sigma-Aldrich) with 2 mM l-glutamine (Invit-
rogen), penicillin-streptomycin, and 10% FBS. Cells were
harvested by enzymatic detachment using Accutase (PAA), then
centrifuged at 2000 gfor 10 min and washed twice in PBS.
Stable transfection of MDCK cells was achieved with a
pCB6-hAQP-1 vector. To generate pCB6-hAQP-1, a PCR was
performed on the pXgev1hAQP-1 vector (kindly provided
by Peter Agre, Johns Hopkins University School of Medicine,
Baltimore, MD, USA; ref. 27) using the following primers:
sense, GCTCTAGAGCCAGCATGGCCAGCGAG, and anti-
sense, GCTCTAGACTATTTGGGCTTCATCTCC; the frag-
ment was cut with XbaI and ligated into the XbaI site of the
pCB6 vector (28). Orientation and proper sequence were
checked using DNA sequence analysis. For the control plas-
mid, the hAQP-1 sequence was deleted. Cells were transfected
using the calcium phosphate method (29). Positive clones
were selected during 2 wk of incubation in EMEM containing
0.6 mg G418/ml (Carl Roth, Karlsruhe, Germany). Expres-
sion of hAQP-1 in MDCK cells was verified by Western
blotting using AQP-1 antibody (affinity-purified polyclonal
antibody against the AQP-1 C-terminal region; RA3391/
2353AP; Nejsum et al., ref. 30) and, as secondary antibody,
anti-rabbit-IgG IRDye800CW (Li-Cor Biosciences, Lincoln,
NE, USA). A Trans-Blott SD semidry transfer cell (Bio-Rad,
Richmond, CA, USA) was used with a nitrocellulose mem-
brane. The Odyssey Infrared Imaging System (Li-Cor Biosci-
Figure 1. Effects of cholesterol on CO
2
permeability of artificial phospholipid vesicles with l-␣-phosphatidylcholine:
l-␣-phosphatidylserine (PC:PS) at a molar ratio of 8:2. Cholesterol content of the vesicle membranes varied between 0 and 70
mol% cholesterol per total lipids. Vesicle diameter for all concentrations of cholesterol was 150 nm. All vesicles had an
intravesicular carbonic anhydrase activity of ⬃10.000 (acceleration factor of the kinetics of CO
2
hydration). Mass spectrometric
measurement was conducted at 37°C. A) Original records of mass spectrometric experiments with 70% cholesterol and 30%
cholesterol in the vesicle membranes, respectively. On a logarithmic scale (yaxis) the concentration of
18
O-labeled CO
2
in the
fluid of the mass spectrometric measuring chamber (14, 16) minus its final value at isotopic equilibrium, is plotted vs. time (x
axis). The first phase of both records represents the slow linear decay of C
18
O
16
O in the absence of any vesicles. The subsequent
accelerated phases are initiated by the addition of vesicles into the chamber. P
CO2
representing a given curve of C
18
O
16
O decay
is determined by 3 parameters: amplitude and slope of the second (fast) phase and slope of the third phase of the record. P
CO2
values obtained from the curves shown are given within the figure, and were calculated by the procedures described previously
(10, 14, 16, 25). It should be appreciated that, as published previously (14, 16), these curves are extremely reproducible, and
the curves calculated with the fitted vales of P
CO2
(and bicarbonate permeability) are perfectly superimposed on and practically
indistinguishable from the experimental curves. The reproducibility with which P
CO2
values are derived from curves such as
those shown becomes apparent in the sd values indicated in panel B.B) Dependency of P
CO2
on membrane cholesterol content.
Dashed line indicates a range of cholesterol concentrations, for which no or no reliable P
CO2
values were obtained; solid lines
connect data points representing measured mean P
CO2
values; dotted line represents the linear regression of the data points
between 30 and 70% cholesterol. The first 3 data points at 0, 5, and 17 mol% cholesterol are greater than or equal to than the
upper limit of detectability of P
CO2
by the mass spectrometric method, in this case 0.16 cm/s. It is possible that the P
CO2
values
at 5 and 17 mol% cholesterol are actually higher than the value at 0% cholesterol, as it has been reported that lipid bilayer
permeability for small molecules may increase with increasing cholesterol up to ⬃20 mol%, before it decreases with further
increases in cholesterol content (49, 50). Between 17 and 30 mol% cholesterol P
CO2
is seen to fall by ⱖ1 order of magnitude,
and between 30 and 70 mol% it decreases by a further order of magnitude. Error bars ⫽sd;n⫽7–18.
5184 Vol. 26 December 2012 ITEL ET AL.The FASEB Journal 䡠www.fasebj.org
ences) was used for visualization of the antibody-labeled
protein bands. SDS-PAGE was performed on a 1-mm-thick 9%
acrylamide gel in a Mini-Protean 3 SDS-PAGE chamber
(Bio-Rad). Samples were mixed 1:2 with sample buffer (130
mM Tris-HCl, 20% glycerol, 4.6% SDS, 0.02% bromphenol
blue, and 2% DTT) and heated to 95°C for 5 min. This
mixture was loaded onto the gel (15 l/lane).
Membrane cholesterol of isolated MDCK cells was reduced
by resuspending the cells from 8 culture dishes of 10 cm
diameter, after washing them once, in 40 ml EMEM contain-
ing 20 mM methyl--cyclodextrin (MCD; Sigma-Aldrich)
without FBS, and then incubating them under mild stirring at
37°C and 5% CO
2
for 1 h. The protocol essentially follows the
procedure described by Francis et al. (31). After incubation,
cells were centrifuged for 10 min at 2000 gand washed once
with PBS. The cell pellet was resuspended in 4 vol of PBS, and
the cell number and vitality were determined with an auto-
matic cell counter (Countess; Invitrogen). Controls were
subjected to the identical protocol but in the absence of
MCD in EMEM. Vitality of controls was 90%, that of
MCD-treated cells 85%. We note that leaky cells do not
contribute to the C
18
O
16
O mass spectrometric signal, because
any carbonic anhydrase activity of these cells is inhibited by
the extracellular carbonic anhydrase inhibitor that is present
during the experiment (10, 14).
Membrane cholesterol of MDCK cell membranes was in-
creased by suspending them in 80 ml EMEM with 5 mM
cholesterol-loaded MCD (32), after cells had been washed
once. This suspension was incubated under mild stirring
overnight at 37°C and 5% CO
2
. Then, cells were centrifuged
at 2000 gand washed once with PBS. The cell pellet was
resuspended in 4 vol PBS, and cell number and vitality were
determined. Control cells were treated identically, except for
MCD being absent. Vitality was 80% in both control and
MCD-treated cells. The cholesterol content of untreated
MDCK cells (33) is compared with that of human red blood
cells (34, 35) in Table 1.
Liposomes
Control liposomes and proteoliposomes were prepared by
standard methods (36, 37). Briefly, phospholipids l-␣-phos-
phatidylcholine (PC; chicken egg) and l-␣-phosphatidylser-
ine (PS; porcine brain; sodium salt; both Avanti Polar Lipids,
Alabaster, AL, USA) and cholesterol (ⱖ99%, Sigma Grade;
Sigma-Aldrich) dissolved in chloroform were mixed at the
desired molar ratios (PC:PS 8:2 and variable amounts of
cholesterol). The lipid mixture was dried under a gentle
nitrogen stream, forming a smooth lipid film on the inside of
a small glass vial, followed by high-vacuum drying for ⱖ6h.
The lipid film was hydrated to a concentration of 3 mg/ml in
a Hepes buffer (20 mM Hepes and 100 mM NaCl) containing
1.25 mg/ml CAII (Sigma-Aldrich) and the detergent n-octyl-
-d-glucopyranoside (-OG; Affymetrix, Santa Clara, CA,
USA) was added to a concentration of 4% (w/v). For proteo-
liposomes, a certain volume of AQP stock solution was added
to give the desired lipid to protein ratio (LPR). The resulting
mixture was incubated with intermittent agitation for1hat
room temperature, loaded into dialysis tubing (Spectra/Por;
Spectrum Labs, Rancho Dominguez, CA, USA) with a molec-
ular cutoff of 3500 Da and dialyzed against 5.0 L of Hepes
buffer for 24–48 h at room temperature. The resulting
suspension was extruded 15 times through a 0.2-m track-
etched filter (Nucleopore; Whatman, Maidstone, UK). Ex-
travesicular material was removed on a Superdex 200 column
(GE Healthcare, Little Chalfont, UK). Liposomes and proteo-
liposomes were stored at 4°C, and permeability measure-
ments were performed within 3–4 d. Average vesicle size was
determined by transmission electron microscopy on a Philips
FEI Morgagni 268D instrument (Philips, Amsterdam, The
Netherlands) and found to be 150 nm, independent of
cholesterol concentration.
To count the number of aquaporin molecules of the proteo-
liposomes, freeze-fracture preparations were performed as de-
scribed previously (38). Proteoliposomes were centrifuged at
300,000 gin a TL100 ultracentrifuge (Beckman Coulter, Fuller-
ton, CA, USA) for 30 min at 4°C. The resulting pellet was
cryoprotected with glycerol (30% v/v). A small droplet of the
sample was placed on the copper holder and quenched in liquid
propane (39). The frozen sample was fractured at ⫺125°C in
high vacuum of ⬃10
⫺5
Pa with a liquid nitrogen-cooled knife in
a Balzers 400 freeze-etching unit (Balzers AG, Balzers, Liechten-
stein). The fractured sample was replicated with a 1- to 1.5-nm
deposit of platinum-carbon, and coated with 20 nm carbon film.
The Pt/C replica was cleaned with 2% SDS, washed with pure
water, transferred onto a copper EM grid, and observed with a
Philips CM10 electron microscope. Particle densities were deter-
mined by counting the total number of visible particles (AQP
tetramers) divided by the total vesicle area calculated from the
diameter of each vesicle visible on the freeze-fracture images.
Aquaporins
Human AQP-1 was produced in Pichia pastoris and character-
ized as described previously (40), with some fine tuning of the
growth and purification protocols. Mainly, X-33/pPICZB-
hAQP-1-Myc-His
6
was cultivated in 1 L BMMY (Invitrogen)
shake flask cultures, and expression was induced for 48 h,
with additional methanol (0.5%) added after 24 h. Cells were
collected and resuspended in breaking buffer (50 mM KPi,
5% glycerol, pH 7.5) and broken using X-press. Urea- and
NaOH-washed membranes were solubilized [20 mM Tris-HCl,
TABLE 1. P
CO2
of cell membranes vs. cholesterol content
Membrane P
CO2
(cm/s) Source
Cholesterol in membrane
lipids (mol%)
P
CO2
predicted from
cholesterol content (cm/s)
MDCK 0.017 ⫾0.004 Present work 37
a
0.015
tsA201 0.007 ⫾0.003 Present work
Red cell: AAQP1, Afunctional Rh 0.015 ⫾0.003 Ref. 16 45
b
0.010
Basolateral membrane of PCE ⬃0.022 Ref. 10 42
c
0.011
Apical membrane of PCE 0.0015 ⫾0.0006 Ref. 10 77
c
0.0016
P
CO2
values (means⫾sd) of the listed cell membranes may be compared with the P
CO2
values reported for planar lipid bilayers as 0.35 and
3.2 cm/s (refs. 3, 4). All membranes considered in the table lack AQP-1: MDCK cells, see Fig. 4Band Missner et al. (4); tsA201 cells, this work
(not shown); both apical and basolateral membranes of the proximal colon epithelium (PCE) possess no AQP-1, although the basolateral
membrane expresses AQP3, which is has a very low CO
2
conductance even though it conducts water very well (60). Thus, both colon membranes
are presumably free of major protein CO
2
channels.
a
Ref. 33.
b
Refs. 34, 35.
c
Ref. 61.
5185CO
2
PERMEABILITY OF CELL MEMBRANES
100 mM NaCl, 10% glycerol, 5% -OG (Anatrace, Maumee,
OH, USA), 2 mM -mercaptoethanol (-MeOH), and protease
inhibitor cocktail (Complete EDTA-free; Roche Diagnostics,
Basel, Switzerland), pH 8.0] at a concentration of 2 mg total
protein/ml for1hatRT.Imidazole (10 mM) was added to the
solubilized material (140,000 g, 30 min, 4°C), which was incu-
bated with equilibrated (20 mM Tris-HCl, 300 mM NaCl, 10%
glycerol, 2 mM -MeOH, 1% -OG, and 10 mM imidazole, pH
8.0) Ni-NTA agarose for 16 h at 4°C. The matrix was washed in
30 mM imidazole, and the protein was eluted in 300 mM
imidazole, using the same buffer. The protein-containing
samples were concentrated using a 30,000 MW cutoff filter
(VivaSpinn; Vivaproducts, Littleton, MA, USA), and protein
buffer (20 mM Tris-HCl, 100 mM NaCl, 10% glycerol, and 1%
-OG, pH 8.0) was exchanged on a Superdex 200 column
(GE Healthcare). Purity and concentration of hAQP-1 after
concentration (12 mg/ml) were ascertained by SDS-PAGE
electrophoresis and by UV spectrophotometry (Nanodrop;
ThermoScientific, Waltham, MA, USA), respectively.
Aquaporin Z (AqpZ) was purified as described by Borgnia
et al. (37), with slight adjustments. Briefly, the plasmid for
overexpression of histidine-tagged AqpZ (pTrc10HisAqpZ)
was transfected into the Escherichia coli strain JM109 by elec-
troporation (41). Luria Broth (LB) cultures (10 ml, 50 g/ml
ampicillin) were incubated for 17 h, diluted into1Loffresh
LB medium, and propagated to an A
600
of ⬃2.5. Production
of 10-His-AqpZ was induced by addition of 1 mM isopropyl-
-d-thiogalactoside (IPTG) for 3 h, then cultures were cen-
trifuged (10,000 g, 20 min, 4°C). Harvested cells were resus-
pended in 10 ml of ice-cold lysis buffer [100 mM K
2
HPO
4
,1
mM MgSO
4
, 1 mM phenylmethylsulfonylfluoride (PMSF), 0.1
mg/ml deoxyribonuclease I (DNase I), and 10 mM imidazole,
pH 7.0]. Cells were lysed by ultrasonication on ice for 30–40
min with intermittent cooling steps. Unbroken cells were
separated by 20 min centrifugation at 10,000 gat 4°C;
subsequently, the supernatant was centrifuged at 140,000 gat
4°C for 60 min. The resulting pellet was solubilized in 10 ml
of solubilization buffer (5% -OG, 100 mM K
2
HPO
4
, 10%
glycerol, 200 mM NaCl, and 15 mM imidazole, pH 8.0) and
incubated with agitation on ice overnight. The suspension
was then filtered through a 0.45-m pore size filter (What-
man), and 1 ml of freshly washed Ni-NTA agarose beads
(Qiagen, Valencia, CA, USA) was added to the filtrate and
incubated on ice for 2 h. The resin was then packed on a
column, washed with 10 ml of wash buffer (1% -OG, 100 mM
K
2
HPO
4
, 10% glycerol, 200 mM NaCl, and 100 mM imid-
azole, pH 7.0). Bound material was first incubated for 15 min
with 1 ml of elution buffer (1% -OG, 100 mM K
2
HPO
4
, 10%
glycerol, 200 mM NaCl, and 400 mM imidazole, pH 7.0) at
room temperature and collected in Eppendorf tubes. For
more protein recovery, repeated 1-ml elution steps were
performed. After AqpZ purification, protein buffer was
exchanged (1% -OG, 20 mM Hepes, and 100 mM NaCl,
pH 7.4) by size-exclusion chromatography through two
coupled HiTrap desalting columns (Sephadex G-25 Super-
fine; GE Healthcare). Protein purity and concentrations
(typically 3-10 g/ml of culture) were determined by gel
electrophoresis and UV/VIS spectroscopy (NanoDrop
2000c; ThermoScientific).
Statistics
Statistical significance was tested by Student’s ttest where
applicable, or by ANOVA followed by a suitable posttest
(Bonferoni, comparing selected pairs of data groups, or
Tukey, multiple comparisons).
RESULTS AND DISCUSSION
Intrinsic P
CO2
of different types of cell membranes
CO
2
permeability values of phospholipid bilayers have
been determined as 0.35 cm/s (3), 3.2 cm/s (4) and
⬎0.16 cm/s (present study). Here, we report the P
CO2
values of the membranes of different types of mamma-
lian cells, which are all devoid of functional gas chan-
nels (Table 1). As determined by Western blotting (see
Fig. 4Band ref. 4) and shown in Table 1, MDCK and
the tsA201 cell line do not express AQP-1. The human
red blood cells also included in Table 1 were AQP-1
deficient, while the Rh CO
2
channel was present but
largely inhibited by 4,4=-diisothiocyano-2,2=-stilbenedi-
sulfonic acid (DIDS; ref. 16). These three cell types
were studied at 37°C in suspensions of the isolated cells
by the mass spectrometric
18
O-exchange technique
reported earlier (10). Their P
CO2
values are all of the
order of ⬃0.01 cm/s; i.e., they are ⬎1–2 orders of
magnitude lower than P
CO2
of artificial phospholipid
membranes. The fourth type of membrane given in
Table 1, the basolateral membrane measured in intact
proximal colon epithelium layers, exhibits a P
CO2
sim-
ilar to that of these three cell types, ⬃0.02 cm/s (10).
The fifth cell membrane in Table 1, the apical mem-
brane, again determined in intact epithelial layers of
the proximal colon, also lacks gas channels and repre-
sents a special case by exhibiting a P
CO2
of 0.001 cm/s
(10), an order of magnitude below the P
CO2
of the first
four membranes. Such a difference in P
CO2
between the
apical and basolateral membranes in the colon is qualita-
tively in agreement with previous evidence reported by
Hasselblatt et al. (9). We observe that all cell membranes
considered here have (in the absence of gas channels)
P
CO2
values that are much more than one order of
magnitude lower than P
CO2
of phospholipid bilayers.
Which component of the biological membranes is
responsible for this property? Two possibilities have
recently been discussed in a state-of-the-field letter (5):
a special component of the lipid phase of cell mem-
branes that would increase substantially its resistance to
CO
2
, or the proteins of the membranes that would have
to render ⬎90% of the membrane area inaccessible to
CO
2
. Even with a membrane protein content of 50%,
this would require that the ectodomains of membrane
proteins interact tightly with the majority of the surface
lipids of the membrane (42); such a situation would,
however, hinder access of all other solutes to the
membrane lipid phase, as well.
Drastic effect of cholesterol enrichment on P
CO2
of
artificial lipid vesicles
Since it is well known that cholesterol not only en-
hances the mechanical stability of membranes and
increases their microviscosity (43) but can also signifi-
cantly reduce membrane permeability to uncharged
solutes, such as water (44–46) and NH
3
(45, 47), we
investigated the effect of incorporation of cholesterol
into phospholipid vesicles. The phospholipid composi-
tion in all measurements was POPC (chicken egg) and
POPS (porcine brain) at a molar ratio of 8:2. In many
5186 Vol. 26 December 2012 ITEL ET AL.The FASEB Journal 䡠www.fasebj.org
cell membranes, PC is a major component of the
membrane lipids (33, 48), especially in the outer
membrane leaflet, and is important for membrane
structure and permeability barrier, while PS is typically
present in the inner leaflet (45). To these phospholip-
ids increasing amounts of cholesterol were added,
accounting for 5 to 70 mol% of total membrane lipids.
Vesicles were produced in the presence of bovine
carbonic anhydrase II, attaining an intravesicular CA
activity of ⬃10,000 (i.e., 10,000-fold acceleration of the
CO
2
hydration reaction), as necessary for the
18
O-
exchange measurement (10). Vesicle diameter by elec-
tron microscopy was 150 nm.
Mass spectrometric determinations of P
CO2
in sus-
pensions of the vesicles at 37°C yielded the two exam-
ples of original mass spectrometric recordings shown in
Fig. 1A. These types of recordings were used to calcu-
late the results shown in Fig. 1B. The central message
from this latter figure is a drastic fall in P
CO2
,byⱖ1
order of magnitude, between 0 and 30% cholesterol
(Fig. 1B, dashed line), and the continuation of this fall
between 30 and 70% cholesterol (Fig. 1B, solid line and
dotted regression line), resulting in a decrease in P
CO2
by a further order of magnitude. Thus, 70% cholesterol
causes a decrease in P
CO2
by ⱖ2 orders of magnitude.
A second aspect of Fig. 1Bis the finding that we
cannot determine the correct P
CO2
of the pure phos-
pholipid vesicles, but can only give a lower limit indi-
cating that P
CO2
in the absence of cholesterol is ⬎0.16
cm/s. The mass spectrometric method is sensitive to
changes in P
CO2
only up to an upper limit, whose value
depends on the surface to volume ratio of the vesicles
or cells (see equations in Endeward and Gros; ref. 10).
It may be noted here that this upper limit value per se
indicates that the thickness of the unstirred layer
around the vesicles must be ⱕ1m[D
CO2
/P
CO2
ⱕ
(1.7⫻10
⫺5
cm/s)/(0.16 cm/s) ⫽1⫻10
⫺4
cm, an esti-
mate based on the consideration that the diffusion
resistance due to the unstirred layer must be less than
or equal to the observed total diffusion resistance of the
vesicle]. As discussed above, previous studies suggest
that the unstirred layer is in fact ⬍0.1 m (25) and thus
exerts no significant effect on the vesicle P
CO2
values.
Third, it may be noted in Fig. 1Bthat cholesterol
concentrations of 5 and 17% have no detectable effect on
P
CO2
, whose values also exceed the upper limit of detec-
tion. This result is compatible with findings that lipid
bilayer permeability for small solutes may actually increase
with increasing cholesterol content up to ⬃20 mol%, and
begin to decrease in a concentration-dependent manner
only at cholesterol concentrations ⬎20% (49, 50). Such
increases would then, of course, also exceed the present
detection limit, and therefore become not visible in Fig.
1A, which could explain the apparent sharp bend at 17%
cholesterol in the curve of Fig. 1B.
We note that most previous investigators have not
studied the effect of cholesterol on vesicle and planar
phospholipid bilayer P
CO2
, with the exception of Miss-
ner et al. (4), who, in contrast to the present results,
report that presence vs. absence of cholesterol did not
affect P
CO2
of planar lipid bilayers, although they do
not present complete statistics for this observation. It
should be considered here that planar lipid bilayers,
produced by using decane as a solvent, may retain
considerable amounts of decane in the final bilayer
(51), which is expected to raise the P
CO2
of the bilayer
due to the high solubility of CO
2
in decane (52) and its
likely effect of increasing membrane fluidity (3). It is of
interest to mention here that Hub et al. (53), from
molecular dynamic simulations of lipid bilayers, have
predicted a drastic increase of the intramembrane
energy barrier for CO
2
with increasing membrane
cholesterol, a result qualitatively in excellent agree-
ment with the findings of Fig. 1B.
With regard to the absolute value of the P
CO2
of pure
phospholipid vesicles observed here (⬎0.16 cm/s), it
should be mentioned that previous studies using rapid
reaction stopped-flow spectrophotometry (54, 55) have
reported P
CO2
values of phospholipid vesicles to be ex-
tremely low, ⬃10
⫺3
cm/s. This is 3 orders of magnitude
lower than the above-mentioned values for planar phos-
pholipid bilayers. As discussed previously (25), the enor-
mous discrepancy between these two groups of measure-
ments is likely to be due to an underestimation in the
stopped-flow apparatus experiments caused by severe
unstirred layers and poor mixing in the face of the
extremely fast process of CO
2
uptake by phospholipid
vesicles.
We conclude that the present study with P
CO2
ⱖ0.16
cm/s tends to support the high estimates of P
CO2
derived
from pure planar phospholipid bilayers. However, the
membrane lipid composition representative of a cell
membrane, containing 40–50 mol% cholesterol, by virtue
of the drastic effect of cholesterol, results in a P
CO2
that is
1–2 orders of magnitude below these figures.
Effect of cholesterol depletion and enrichment on
P
CO2
of MDCK cells
Nonpolarized MDCK cells in suspension were treated with
20 mM MCD largely as described by Francis et al. (31) to
reduce membrane cholesterol, or were treated with
MCD preloaded with cholesterol at a molar ratio of 1:8
(32) to increase membrane cholesterol. These treatments
did not alter cell diameter, and cholesterol enrichment
had no effect on cell vitality, while cholesterol depletion
decreased vitality moderately by 0 to 11%. It should be
noted that nonvital cells do not contribute to the mass
spectrometric signal and thus do not affect P
CO2
.Figure 2
shows the results of P
CO2
determinations at 37°C. Control
MDCK cells, whose membrane lipids have a cholesterol
content of ⬃37% (33), exhibit a P
CO2
of 0.017 cm/s at
37°C (Fig. 2 and Table 1). Cholesterol depletion increases
P
CO2
dramatically to ⬎0.75 cm/s, which is the upper
detection limit in the case of the MDCK cells. Cholesterol
enrichment, on the other hand, decreases P
CO2
to 0.0065
cm/s, i.e.,⬃1/3 of control. The data clearly show that
cholesterol content controls cell membrane P
CO2
in a
massive way. This is qualitatively in line with the vesicle
results of Fig. 1B.
Cholesterol content quantitatively determines P
CO2
of
cell membranes
The cholesterol content of the membrane lipids of 4 of
the 5 cell membranes of Table 1 is known and given in
5187CO
2
PERMEABILITY OF CELL MEMBRANES
the fourth column. We have used these figures to
predict P
CO2
on the basis of the regression line of Fig.
1B, which describes the vesicle data between 30 and 70
mol% cholesterol. The predicted P
CO2
values are given
in the fifth column of Table 1. Comparison with the
measured P
CO2
values (Table 1, second column) shows
a marked agreement within ⱕ50% (factor of 2 in the
case of the basolateral colon membrane) between what
has been measured in a biological membrane and what
has been observed in vesicles with an identical choles-
terol content. We conclude that in the range of choles-
terol concentrations as they occur in cell membranes,
and in the absence of gas channels, cholesterol is the
major determinant of the membrane permeability for
CO
2
. This leads to intrinsic P
CO2
values of cell mem-
branes with a “normal” cholesterol content of 40–50
mol% of ⬃0.01 cm/s, and to a 10⫻lower P
CO2
of
⬃0.001 cm/s when cholesterol amounts to as much as
77 mol%. This has not been recognized before and
revises drastically the view that the P
CO2
of cell mem-
branes is similar to that of pure phospholipid mem-
branes (4, 5). We note that the same view has hitherto
been held for other lipophilic gases, such as O
2
, CO,
and NO, but experimental evidence is presently lacking
in the case of these gases.
Effect of AQP-1 reconstitution on P
CO2
of lipid
vesicles
The data presented so far show that mammalian cell
membranes exhibit a rather low “intrinsic” CO
2
perme-
ability. We have already reported previously that P
CO2
of the red blood cell membrane increases 10-fold, to
0.15 cm/s, when functional AQP-1 and Rh protein are
present (14, 16). Here, we demonstrate in lipid vesicles
containing 50 mol% cholesterol that P
CO2
of this
relatively gastight membrane can be raised 9-fold from
0.003 to 0.027 cm/s (Fig. 3B) by incorporating human
AQP-1 into the vesicle membrane. Vesicles were pre-
pared in the presence of LPRs of 400, 230, 200, and
140. These theoretical LPR values were established in
the solution from which vesicles were generated, and
need not be identical to those present in the mem-
branes of the final vesicles. Nevertheless, the different
LPR values yielded graded increases in vesicle P
CO2
(⌬P
CO2
) as seen in Fig. 3A. The actual incorporation of
AQP-1 into the final vesicles was obtained from freeze-
fracture electron microscopy, an example of which is
shown in Fig. 3C. With LPR 400, vesicles exhibited 54
aquaporin monomers/m
2
membrane area, while with
LPR 230 and 140, these numbers were 63 and 80/m
2
,
respectively. The ratios of these AQP-1 densities corre-
spond roughly to the ratios of increases in P
CO2
(Fig.
3A), although the P
CO2
at LPR 140 or 80 AQP-1/m
2
stands out somewhat compared to the other P
CO2
values. We note that on average, the ratios of ⌬P
CO2
over AQP-1 density in the membrane are similar to the
value observed previously for human red blood cells
(14, 16). In contrast, bacterial AqpZ, reconstituted at
an LPR of 200 (70 AqpZ/m
2
), did not increase P
CO2
,
AqpZ thus seeming to conduct no or little CO
2
. Figure
3Bshows that 10
⫺5
M DIDS, while having no significant
effect on P
CO2
of vesicles without AQP-1, markedly
inhibits the increase in P
CO2
caused by AQP-1. A strong
inhibiting effect of DIDS on CO
2
conductivity of AQP-1
is consistent with previous observations (14–16) and,
together with the lack of effect of AqpZ, confirms the
specificity of the AQP-1 effect on P
CO2
. It should be
noted that DIDS at the present concentration has no
effect on the activity of carbonic anhydrase, the only
other protein present in these vesicles.
Effect of AQP-1 expression on P
CO2
of MDCK cells
Since the intrinsic membrane CO
2
permeabilities of
vesicles with 40% cholesterol and of MDCK cell mem-
branes are of a similar order of magnitude, it should be
possible to discern the CO
2
conduction by AQP-1 in
MDCK cells, as well as in vesicles. We stably transfected
nonpolarized MDCK cells with the human AQP-1 ex-
pression construct pCB6-hAQP-1. Figure 4Bconfirms
that this resulted in a significant expression of AQP-1.
Mass spectrometric measurements with MDCK cells in
suspension (Fig. 4A) show that expression of AQP-1
results in an ⬃50% increase of P
CO2
. AQP-1-mediated
CO
2
conduction is, as in the case of reconstituted
vesicles, inhibited by DIDS. These measurements
clearly confirm that AQP-1 is significant in CO
2
-chan-
neling across cell membranes with an average content
of cholesterol.
This effect of AQP-1 in MDCK cells has not been
observed by Missner et al. (4). We have previously
presented evidence that the setup used by these au-
thors, due to thick water layers associated with their
layer of MDCK cells, does not allow detection of an
increase in membrane CO
2
permeability of the size
seen in Fig. 4A(32). In contrast, the isolated MDCK
cells used in the present mass spectrometric setup are
Figure 2. Effects of cholesterol on CO
2
permeability of MDCK
cell membranes. Compared to the P
CO2
of control MDCK
cells (center bar), MDCK cells depleted of cholesterol by
exposure to MCD exhibit a drastically increased P
CO2
(⬎0.75 cm/s; left bar) and MDCK cells loaded with choles-
terol (right bar) show a P
CO2
⬃3 times lower than controls.
Error bars ⫽se;n⫽12–20. Since the elevated P
CO2
of the left
bar is not a defined value but just a lower limit, statistical
analysis was not possible. *P⬍0.05 vs. control.
5188 Vol. 26 December 2012 ITEL ET AL.The FASEB Journal 䡠www.fasebj.org
expected to have an unstirred layer of ⬍1m, which
will not affect the measured P
CO2
values to a major
extent (see Materials and Methods).
CONCLUSIONS
Cholesterol can decrease membrane CO
2
permeability
by ⬎2 orders of magnitude in phospholipid vesicles
and intact cells. While cell membranes with normal
cholesterol contents of 30–50 mol% exhibit a relatively
low P
CO2
of ⬃0.01 cm/s, very low P
CO2
values of 0.001
cm/s result from cholesterol contents ⬎70%. By diffu-
sion calculations, we have previously illustrated that a
P
CO2
of 0.01 cm/s implies a significant impairment of
CO
2
exchange of red blood cells, especially under
conditions of exercise (14). Likewise, we have demon-
strated that across the apical membrane of colon epi-
thelium, the P
CO2
of 0.001 cm/s is associated with a
large gradient of pCO
2
across this membrane (10).
Higher CO
2
permeabilities than both these low values
appear to be achieved by inserting membrane protein
gas channels, as shown here with hAQP-1, but may also
result from extremely low cholesterol contents, as they
are found in mitochondrial membranes. With these two
variables, depending on functional requirements, ei-
ther highly gas-permeable (red blood cells, lung epi-
thelium) or nearly gas-tight membranes (apical mem-
branes of colon, stomach and kidney epithelia) can be
established. The latter property is functionally impor-
tant when epithelial cells are to be protected against a
severe acid load by the extremely high CO
2
pressure as
can occur, for example, in the colonic lumen (10). The
conclusions of this work may hold similarly for physio-
logically important gases other than CO
2
, such as O
2
,
N
2
, NO, and CO, which have membrane lipid-water
partition coefficients just slightly higher than CO
2
(56–58). It should be clear that the behavior of these
diatomic gases remains speculative as long as it has not
been possible to measure the effect of cholesterol on
their membrane transfer. It may be noted, nevertheless,
that a gas permeability of 0.01 cm/s would be expected
to constitute a much greater problem for O
2
than for
CO
2
, due to the 24 times lower O
2
solubility in water
AqpZ
hAQP1
hAQP1
Figure 3. Effects on P
CO2
of hAQP-1 and AqpZ reconstitution in phospholipid vesicles containing 50 mol% cholesterol, and
effect of DIDS. A) Increase in CO
2
permeability over control, ⌬P
CO2
, at various LPRs, i.e., moles of lipid per mole of aquaporin.
For hAQP-1, a graded increase in ⌬P
CO2
with increasing AQP concentration is seen. All ⌬P
CO2
values between LPR 400 and 140
are significantly different from 0; AqpZ at an LPR of 200 is not significantly different from 0. Error bars ⫽se;n⫽9, 6, 8, 7, 12.
ns, not significant. **P⬍0.02; ***P⬍0.01. B) Effect of 10
⫺5
M DIDS on P
CO2
of liposomes and proteoliposomes. While DIDS
has no effect on control vesicles, it significantly inhibits the CO
2
permeability of hAQP-1. Error bars ⫽se;n⫽6–9. *P⬍0.05.
C) Visualization of reconstituted aquaporins in proteoliposomes by freeze-fracture electron microscopy. hAQP-1 at LPR 140 (a)
and AqpZ at LPR 200 (b). Arrows indicate aquaporin tetramers. Number of vesicles analyzed per condition for determination
of AQP density in the various vesicle membranes was 600–900. Scale bar ⫽100 nm.
5189CO
2
PERMEABILITY OF CELL MEMBRANES
compared to that of CO
2
. Then channels such as AQP1,
which has been shown by molecular dynamics simula-
tions to posses a conductance for O
2
similar to that for
CO
2
(59), would be even more crucial for gas exchange
across membranes. Permeabilities of cell membranes
for CO
2
, and possibly for other gases as well, thus
appear to be regulated over a wide range spanning ⬃2
to 3 orders of magnitude by means of the two param-
eters membrane cholesterol content and incorporation
of protein gas channels into the membrane.
The authors thank Prof. Ernst Ungewickell (Medizinische
Hochschule Hannover) and Dawid Krenc (Universität Kiel,
Kiel, Germany) for helpful discussions, and Mr. Mark Inglin
(University of Basel) for language-editing the manuscript.
The authors are indebted to Dr. Aleksandra Rojek (University
of Aarhus, Aarhus, Denmark) for the kind gift of the AQP-1
antibody RA3391/2353AP. This work was supported by the
Deutsche Forschungsgemeinschaft through grant EN 908/1-1
to V.E., S.A.-S., and G.G.. F.I., W.M., and M.C. gratefully
acknowledge the National Center of Competence in Nano-
scale Science for financial support and Prof. Henning Stahl-
berg (University of Basel) for access to the electron micros-
copy facility. P.M.T.D. is a recipient of a VICI grant of the
Netherlands Organization for Scientific research (NWO) and
acknowledges VICI grant 865.07.002. Author contributions:
V.E. and G.G. conceived the study; V.E. designed the research
and performed all mass spectrometric
18
O exchange experi-
ments; F.I. and W.M. prepared liposomes and proteolipo-
somes, as well as AqpZ; S.A.-S. performed all experiments with
MDCK and tsA201 cells; F.Ö. and K.H. contributed hAQP-1
protein; M.C. performed the freeze-fracture studies; P.T.M.D.
contributed the pCB6-hAQP-1 expression construct; C.T.S.
contributed the extracellular carbonic anhydrase inhibitor;
M.K. contributed to lipo- and proteoliposome preparation;
G.G. and V.E. wrote the paper; all authors critically com-
mented on the manuscript. The authors declare no conflicts
of interest.
REFERENCES
1. Overton, E. (1901) Studien über die Narkose, Gustav Fischer
Verlag, Jena
2. Forster, R.E. (1969) The rate of CO
2
equilibration between red
cells and plasma. In CO
2
: Chemical, Biochemical, and Physiological
Aspects (Forster, R. E., Edsall, J. T., Otis, A. B., and Roughton,
F. J. W., eds) NASA SP-188, pp. 275–284, National Technical
Information Service, Washington, DC
3. Gutknecht, J., Bisson, M. A., and Tosteson, F. C. (1977) Diffu-
sion of carbon dioxide through lipid bilayer membranes: effects
of carbonic anhydrase, bicarbonate, and unstirred layers. J. Gen.
Physiol. 69, 779–794
4. Missner, A., Kügler, P., Saparov, S. M., Sommer, K., Mathai, J. C.,
Zeidel, M. L., and Pohl, P. (2008) Carbon dioxide transport
through membranes. J. Biol. Chem. 283, 25340–25347
5. Boron, W. F., Endeward, V., Gros, G., Musa-Aziz, R., and Pohl, P.
(2011) Intrinsic CO
2
permeability of cell membranes and
potential biological relevance of CO
2
channels. Chemphyschem
12, 1017–1019
6. Geers, C., and Gros, G. (2000) Carbon dioxide transport and
carbonic anhydrase in blood and muscle. Physiol. Rev. 80,
681–715
7. Kikeri, D., Sun, A., Zeidel, M. L., and Hebert, S. C. (1989) Cell
membranes impermeable to NH
3
.Nature 339, 478–480
8. Waisbren, S. J., Geibel, J. P., Modlin, I. M., and Boron, W. F.
(1994) Unusual permeability properties of gastric gland cells.
Nature 368, 332–335
9. Hasselblatt, P., Warth, R., Schulz-Baldes, A., Greger, R., and
Bleich, M. (2000) pH regulation in isolated in vitro perfused rat
colonic crypts. Pflügers Arch. 441, 118–124
10. Endeward, V., and Gros, G. (2005) Low carbon dioxide perme-
ability of the apical epithelial membrane of guinea-pig colon. J.
Physiol. 567, 253–265
11. Nakhoul, N. L., Davis, B. A., Romero, M. F., and Boron, W. F.
(1998) Effect of expressing the water channel aquaporin-1 on
the CO
2
permeability of Xenopus oocytes. Am. J. Physiol. 274,
C543–C548
12. Uehlein, N., Lovisolo, C., Siefritz, F., and Kaldenhoff, R. (2003)
The tobacco aquaporin NtAQP-1 is a membrane CO
2
pore with
physiological functions. Nature 425, 734–737
13. Blank, M. E., and Ehmke, H. (2003) Aquaporin-1 and HCO
3⫺
-
Cl
-
transporter-mediated transport of CO
2
across the human
erythrocyte membrane. J. Physiol. 550, 419–429
14. Endeward, V., Musa-Aziz, R., Cooper, G. J., Chen, L. M.,
Pelletier, M. F., Virkki, L. V., Supuran, C. T., King, L. S., Boron,
W. F., and Gros, G. (2006) Evidence that aquaporin 1 is a major
pathway for CO
2
transport across the human erythrocyte mem-
brane. FASEB J. 20, 1974–1981
15. Musa-Aziz, R., Chen, L. M., Pelletier, M. F., and Boron, W. F.
(2009) Relative CO
2
/NH
3
selectivities of AQP-1, AQP4, AQP5,
AmtB, and RhAG. Proc. Natl. Acad. Sci. U. S. A. 106, 5406–5411
16. Endeward, V., Cartron, J. P., Ripoche, P., and Gros, G. (2008)
RhAG protein of the Rhesus complex is a CO
2
channel in the
human red cell membrane. FASEB J. 22, 64–73
17. Verkman, A. S. (2002) Does aquaporin-1 pass gas? An opposing
view. J. Physiol. 542, 31
18. Hub, J. S., and de Groot. B. L. (2006) Does CO
2
permeate
through aquaporin-1? Biophys. J. 91, 842–848
Figure 4. Expression of hAQP-1 in MDCK cells increases P
CO2
.
A) MDCK cells expressing hAQP-1 show an increased P
CO2
.
DIDS (10
⫺5
M) has no effect on control MDCK cell P
CO2
, but
significantly decreases P
CO2
in AQP-1-expressing MDCK cells.
ns, not significant. Error bars ⫽se;nⱖ7. **P⬍0.02 vs. control;
$$
P⬍0.02 vs. hAQP1. B) Western blot showing that MDCK cells
transfected with pCB6-hAQP-1 express hAQP-1 protein (arrow).
Lane 1: lysed suspension of control cells transfected with empty
vector pCB6, applied at a final dilution of cell pellet of 1:16.
Lane 2: same sample at a dilution of 1:8. Lane 3: MDCK cells
transfected with pCB6-hAQP-1, diluted 1:32. Lane 4: same
sample diluted 1:16. Lane 5: same sample diluted 1:8. Lane 6:
lysed human red blood cells, diluted 1:800, also showing
hAQP-1. Right hand side: molecular mass scale (kDa).
5190 Vol. 26 December 2012 ITEL ET AL.The FASEB Journal 䡠www.fasebj.org
19. Missner, A., and Pohl, P. (2009) 110 years of the Meyer-Overton
rule: predicting membrane permeability of gases and other
small compounds. Chemphyschem 10, 1405–1414
20. Mathai, J. C., Missner, A., Kügler, P., Saparov, S. M., Zeidel,
M. L., Lee, J. K., and Pohl, P. (2009) No facilitator required for
membrane transport of hydrogen sulfide. Proc. Natl. Acad. Sci.
U. S. A. 106, 16633–16638
21. Verkman, A. S. (2011) Aquaporins at a glance. J. Cell Sci. 124,
2107–2112
22. De Groot, B. L., and Hub, J. S. (2011) A decade of debate:
significance of CO
2
permeation through membrane channels
still controversial. Chemphyschem 12, 1021–1022
23. Itada, N., and Forster, R. E. (1977) Carbonic anhydrase activity
in intact red blood cells measured with
18
O exchange. J. Biol.
Chem. 252, 3881–3890
24. Supuran, C. T. (2008) Carbonic anhydrases: novel therapeutic
applications for inhibitors and activators. Nat. Rev. Drug Discov.
7, 168–181
25. Endeward, V., and Gros, G. (2009) Extra- and intracellular
unstirred layer effects in measurements of CO
2
diffusion across
membranes - a novel approach applied to the mass spectromet-
ric
18
O technique for red blood cells. J. Physiol. 587, 1153–1167
26. Van Beek, M., Robben, J. H., Savelkoul, P. J. M., Hendriks, G.,
Devonald, M. A. J., Konings, I. B. M., Lagendijk, A. K., Karet, F.,
and Deen, P. M. T. (2006) Polarization, key to good localization.
Biochim. Biophys. Acta 1758, 1126–1133
27. Jung, J. S., Preston, G. M., Smith, B. L., Guggino, W. B., and
Agre, P. (1994) Molecular structure of the water channel
through aquaporin CHIP. The hourglass model. J. Biol. Chem.
269, 14648–14654
28. Brewer, C. B. (1994) Cytomegalovirus plasmid vectors for per-
manent lines of polarized epithelial cells. Methods Cell Biol. 43,
233–245
29. Wigler, M., Pellicer, A., Silverstein, S., Axel, R., Urlaub, G., and
Chasin, L. (1979) DNA-mediated transfer of the adenine phos-
phoribosyltransferase locus into mammalian cells. Proc. Natl.
Acad. Sci. U. S. A. 76, 1373–1376
30. Nejsum, L. N., Kwon, T.-H., Jensen, U. B., Fumagalli, O.,
Frøkiaer, J., Krane, C. M., Menon, A. G., King, L. S., Agre, P. C.,
and Nielsen, S. (2002) Functional requirement of aquaporin-5
in plasma membranes of sweat glands. Proc. Natl. Acad. Sci.
U. S. A. 99, 511–516
31. Francis, S. A., Kelly, J. M., McCormack, J., Rogers, R. A., Lai, J.,
Schneeberger, E. E., and Lynch, R. D. (1999) Rapid reduction
of MDCK cell cholesterol by methyl-beta-cyclodextrin alters
steady state transepithelial electrical resistance. Eur. J. Cell Biol.
78, 473–484
32. Christian, A. E., Haynes, M. P., Phillips, M. C., and Rothblat,
G. H. (1997) Use of cyclodextrins for manipulating cellular
cholesterol content. J. Lipid Res. 38, 2264–2272
33. Hansson, G. C., Simons, K., and van Meer, G. (1986) Two strains
of the Madin-Darby canine kidney (MDCK) cell line have
distinct glycosphingolipid compositions. EMBO J. 5, 483–489
34. O’Brien, J. S. (1967) Cell membranes - composition: structure:
function. J. Theor. Biol. 15, 307–324
35. Korn, E. D. (1969) Cell membranes: structure and synthesis.
Annu. Rev. Biochem. 38, 263–288
36. Smith, S. A., and Morrissey, J. H. (2004) Rapid and efficient
incorporation of tissue factor into liposomes. J. Thromb. Haemost.
2, 1155–1162
37. Borgnia, M. J., Kozono, D., Calamita, G., Maloney, P. C., and
Agre, P. (1999) Functional reconstitution and characterization
of AqpZ, the E. coli water channel protein. J. Mol. Biol. 291,
1169–1179
38. Plançon, L., Chami, M., and Letellier, L. (1997) Reconstitution
of FhuA, an Escheria coli outer membrane protein, into lipo-
somes. J. Biol. Chem. 272, 16868–16872
39. Aggerbeck, L. P., and Gulik-Krzywicki, T. (1986) Studies of
lipoproteins by freeze-fracture and etching electron microscopy.
Methods Enzymol. 128, 457–472
40. Nyblom, M., Oberg, F., Lindkvist-Petersson, K., Hallgren, K.,
Findlay, H., Wikström, J., Karlsson, A., Hansson, O., Booth, P. J.,
Bill, R. M., Neutze, R., and Hedfalk, K. (2007) Exceptional
overproduction of a functional human membrane protein.
Protein. Expr. Purif. 56, 110–120
41. Kumar, M., Grzelakowski, M., Zilles, J., Clark, M., and Meier, W.
(2007) Highly permeable polymeric membranes based on the
incorporation of the functional water channel protein Aqua-
porin Z. Proc. Natl. Acad. Sci. U. S. A. 104, 20719–20724
42. Engel, D. M. (2005) Membranes are more mosaic than fluid.
Nature 438, 578–580
43. Shinitzky, M., and Inbar, M. (1976) Microviscosity parameters
and protein mobility in biological membranes. Biochim. Biophys.
Acta 433, 133–149
44. Lande, M. B., Donovan, J. M., and Zeidel, M. L. (1995) The
relationship between membrane fluidity and permeabilities to
water, solutes, ammonia, and protons. J. Gen. Physiol. 106, 67–84
45. Hill, W. G., and Zeidel, M. L. (2000) Reconstituting the barrier
properties of a water-tight epithelial membrane by design of
leaflet-specific liposomes. J. Biol. Chem. 275, 30176–30185
46. Finkelstein, A. (1976) Water and nonelectrolyte permeability of
lipid bilayer membranes. J. Gen. Physiol. 68, 127–135
47. Antonenko, Y. N., Pohl, P., and Denisov, G. A. (1997) Perme-
ation of ammonia across bilayer lipid membranes studied by
ammonium ion selective microelectrodes. Biophys. J. 72, 2187–
2195
48. Hsiao, L. L., Howard, R. J., Aikawa, M., and Taraschi, T. F.
(1991) Modification of host cell membrane lipid composition
by the intra-erythrocytic human malaria parasite Plasmodium
falciparum. Biochem. J. 274, 121–132
49. Corvera, E., Mouritsen, O. G., Singer, M. A., and Zuckermann,
M. J. (1992) The permeability and the effect of acyl-chain length
for phospholipid bilayers containing cholesterol: theory and
experiment. Biochim. Biophys. Acta 1107, 261–270
50. Trandum, C., Westh, P., Jorgensen, K., and Mouritsen, O. G.
(2000) A thermodynamic study of the effects of cholesterol on
the interaction between liposomes and ethanol. Biophys. J. 78,
2486–2492
51. Bunce, A. S., and Hider, R. C. (1974) The composition of black
lipid membranes formed from egg-yolk lecithin, cholesterol and
n-decane. Biochim. Biophys. Acta 363, 423–427
52. Wilcock, R. J., Battino, R., Danforth, W. F., and Wilhelm, E.
(1978) Solubilities of gases in liquids II. The solubilities of He,
Ne, Ar, Kr, O
2
,N
2
, CO, CO
2
,CH
4
,CF
4
, and SF
6
in n-octane,
1-octanol, n-decane, and 1-decanol. J. Chem. Thermodynamics 10,
817–822
53. Hub, J. S., Winkler, F. K., Merrick, M., and de Groot, B. L. (2010)
Potentials of mean force and permeabilities for carbon dioxide,
ammonia, and water flux across a Rhesus protein channel and
lipid membranes. J. Am. Chem. Soc. 132, 13251–13263
54. Prasad, G. V., Coury, L. A., Finn, F., and Zeidel, M. L. (1998)
Reconstituted aquaporin 1 water channels transport CO
2
across
membranes. J. Biol. Chem. 273, 33123–33126
55. Yang, B., Fukuda, N., van Hoek, A., Matthay, M. A., Ma, T., and
Verkman, A. S. (2000) Carbon dioxide permeability of aqua-
porin-1 measured in erythrocytes and lung of aquaporin-1 null
mice and in reconstituted proteoliposomes. J. Biol. Chem. 275,
2686–2692
56. Simon, S. A., and Gutknecht, J. (1980) Solubility of carbon
dioxide in lipid bilayer membranes and organic solvents.
Biochim. Biophys. Acta 596, 352–358
57. Möller, M., Botti, H., Batthyany, C., Rubbo, H., Radi, R., and
Denicola, A. (2005) Direct measurement of nitric oxide and
oxygen partitioning into liposomes and low density lipoprotein.
J. Biol. Chem. 280, 8850–8854
58. Battino, R., Rettich, T. R., and Tominaga, T. (1984) The
solubility of nitrogen and air in liquids. J. Phys. Chem. Ref. Data
13, 563–600
59. Wang, Y., Cohen, J., Boron, W. F., Schulten, K., and Tajkhor-
shid, E. (2007) Exploring gas permeability of cellular mem-
branes and membrane channels with molecular dynamics. J.
Struct. Biol. 157, 534–544
60. Musa-Aziz, R., Geyer, R. R., and Boron, W. F. (2011) Relative
CO
2
/NH
3
permeabilities of several members of the mammalian
aquaporin familiy: bAQP0, hAQP1, hAQP3, rAQP4-M1, rAQP4-
M23, and hAQP8. Presented at Experimental Biology 2011,
Washington, DC. FASEB J. 25(Suppl.), 1040.5 (abstr.)
61. Meyer zu Düttingdorf, H., Sallmann, H., Glockenthör, U., von
Engelhardt, W., and Busche, R. (1999) Isolation and lipid
composition of apical and basolateral membranes of colonic
segments of guinea pig. Anal. Biochem. 269, 45–53
Received for publication June 27, 2012.
Accepted for publication August 20, 2012.
5191CO
2
PERMEABILITY OF CELL MEMBRANES
