evidence for the presence aqp3 in human red blood cells

7/30/2019 Evidence for the Presence AQP3 in Human Red Blood Cells

1/6

Evidence for the Presence of Aquaporin-3 in Human Red Blood Cells*

(Received for publication, November 14, 1997, and in revised form, January 24, 1998)

Nathalie Roudier, Jean-Marc Verbavatz, Christophe Maurel, Pierre Ripoche, and

Frederique Tacnet

From the Service de Biologie Cellulaire, Departement de Biologie Cellulaire et Moleculaire, CEA/Saclay, 91191 Gif-sur-

Yvette Cedex, France, and the Institut des Sciences Vegetales, CNRS, 91198 Gif-sur-Yvette Cedex, France

A facilitated diffusion for glycerol is present in humanerythrocytes. Glycerol transporters identified to datebelong to the Major Intrinsic Protein (MIP) family ofintegral membrane proteins, and one of them, aqua-porin-3 (AQP3), has been characterized in mammals. Us-ing an antibody raised against a peptide correspondingto the rat AQP3 carboxyl terminus, we examined thepresence of AQP3 in normal and Colton-null (aqua-porin-1 (AQP1)-deficient) human erythrocytes. Threeimmunoreactive bands were detected on immunoblotsof both normal and Colton-null red cells, very similar tothe bands revealed in rat kidney, a material in whichAQP3 has been extensively studied. By immunofluores-cence, anti-AQP3 antibodies stained the plasma mem-branes of both normal and Colton-null erythrocytes.Glycerol transport was measured on intact erythrocytesby stopped-flow light scattering and on one-step pinkghosts by a rapid filtration technique. Glycerol perme-ability values, similar in both cell types, suggest thatAQP1 does not represent the major path for glycerolmovement across red blood cell membranes. Further-more, pharmacological studies showed that Colton-nullred cells remain sensitive to water and glycerol fluxinhibitors, supporting the idea that another proteina-ceous path, probably AQP3, mediates most of the glyc-erol movements across red cell membranes and repre-sents part of the residual water transport activity found

in AQP1-deficient red cells.

Human erythrocytes are highly permeable to water, urea,

and glycerol (13). The existence of membrane proteins that

facilitate water and solute movements in these cells has been

postulated, but most of these proteins remain to be character-

ized. In 1992, Agre and co-workers identified a very abundant

protein of the human red blood cell as the first water-selective

channel that was named aquaporin-1 (AQP1)1 (4). Colton-null

erythrocytes, lacking AQP1, exhibit a reduced osmotic water

permeability (5). However, these cells are found to be residually

permeable to water, suggesting the presence of additional, non-

AQP1, water channels.

The existence of a protein-mediated glycerol transport inerythrocytes has relied mostly on pharmacological evidence. In

particular, inhibition of glycerol transport by sulfhydryl re-

agents, phloretin, and copper ions has been reported (1, 3, 6).

Yet, the identity of the erythrocyte glycerol carrier remains

unknown. The glycerol facilitators identified to date all belong

to the Major Intrinsic Protein (MIP) family (7). They have been

characterized in several organisms: GlpF, a bacterial glycerol

permease facilitator (8); Fps1p, a yeast glycerol exporter (9);

aquaporin-3 (AQP3), initially characterized in mammalian kid-

ney (1012); and AQP7, recently identified in rat testis (13).

Compared with other mammalian aquaporins, which are selec-

tive mostly for water, AQP3 is moderately permeable to water,

but highly permeable to glycerol and possibly to urea (11, 12,

14). AQP3 expression has been reported in several mammaliantissues, including kidney, intestine, stomach, spleen, and eye

(11, 15, 16).

The aim of this study was to investigate the nature of the red

cell glycerol transporter. Due to the high permeability of AQP3

to glycerol, we hypothesized that AQP3, if expressed in human

erythrocytes, could account for their facilitated glycerol perme-

ability. Since AQP1 itself was previously suggested to trans-

port glycerol (17), in our study we used red cells from an

individual who does not express AQP1, associated to the Colton

group antigens (18). We found no difference in glycerol trans-

port between the normal and Colton-null (Co(a-b-)) red cells,

suggesting that AQP1 does not constitute a major pathway for

glycerol. By contrast, our immunological data demonstrate that

AQP3 is expressed in both normal and Colton-null red cells.

Pharmacological evidence strongly supports the hypothesis

that AQP3 could account for the high glycerol permeability of

these cells and that, in Colton-null erythrocytes, AQP3 could

constitute a residual, mercury-sensitive pathway for water.

EXPERIMENTAL PROCEDURES

MaterialsControl and Co(a-b-) red cells (18) were obtained from theInstitut National de Transfusion Sanguine (Paris, France). Rabbit poly-clonal antibodies against purified human AQP1 were raised as de-scribed previously (19). Polyclonal antibodies against a 26-amino acidsynthetic peptide corresponding to the carboxyl terminus of rat kidney

AQP3 were raised in rabbits (Neosystem, France). The anti-AQP3 se-rum was affinity purified by chromatography (immunobilization Kit 2,Pierce). The secondary antibodies used in Western blot experiments(goat anti-rabbit-peroxidase) were from Promega (Madison, WI); those

used in immunofluorescence (mouse anti-rabbit-CY3) were from Jack-son Immunoresearch. 14C-Glycerol (120 mCi/mM) was synthesized bythe Service des Molecules Marquees (CEA/Saclay, France). Phenyl-methylsulfonyl fluoride, pepstatin, DIDS, and phloretin were fromSigma Immunochemicals. Reagents used in electrophoresis and immu-noblotting were from Pharmacia LKB Biotechnology (Uppsala, Sweden)and from Amersham Corporation (Buckinghamshire, United Kingdom).

All other reagents were at least analytical grade.Red Cell PreparationThawed normal and Co(a-b-) red cells were

washed three times in a Carlsen buffer (3) (in mM): NaCl, 154; D-glucose, 5; KH2PO4, 0.25; Na2HPO4, 0.25; pH 7.4, 300 mosm/kg H2O, bycentrifugations at 2,000 g for 10 min at 8 C. The cells were thenresuspended in the Carlsen buffer at a hematocrit of 1.5%, correspond-ing to 107 cells/ml and directly used for stopped-flow experiments.

Stopped-flow ExperimentsKinetics of red cell volume changes werefollowed at 26 C, by 90 light scattering (exc 600 nm) using a

* The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely toindicate this fact. To whom correspondence should be addressed. Tel.: 33-1-69-08-74-

00; Fax: 33-1-69-08-80-46; E-mail: [email protected] The abbreviations used are: AQP1, aquaporin-1; AQP3, aqua-

porin-3; MIP, Major Intrinsic Protein; Pf, osmotic water permeabilitycoefficient; Pgly, glycerol permeability; DIDS, 4,4-diisothiocyanato-stil-bene-2,2-disulfonic acid; PBS, phosphate-buffered saline; Co(a-b-), Col-ton-null; BSA, bovine serum albumin.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 14, Issue of April 3, pp. 84078412, 1998 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org 8407


2/6

stopped-flow spectrophotometer (SFM3, Biologic, Claix, France). Anemission wavelength 500 nm was obtained by using a cut-on filter(Specivex, J526a). Cell osmotic water permeability was measured bymixing 100 l of cells with an equal volume of a hyperosmotic solutionof sucrose to produce a 100 mosm/kg H2O inwardly-directed osmoticgradient. Data from at least 10 time-courses were averaged and fitted to

single exponential functions (k is the rate constant, in s1) by using theSimplex procedure of the BIOKINE software (Biologic, France). Theosmotic water permeability coefficient, Pf , in cm/s, was determinedusing the following equation (5),

dV t /dt Pf SAV MVW Cin/V t Cout (Eq. 1)

where V(t) is the relative red cell volume as a function of time, [SAV] isthe red cell surface area to initial volume ratio, MVW is the molar

volume of water (18 cm3/mol) and Cin and Cout (mol/cm3) are the initial

concentrations of intracellular and extracellular solute. SAV was takenequal to 13,500 cm1 (20). Theoretical kinetics were simulated for

various arbitrary values of Pf and fitted by the BIOKINE software assingle exponentials. The obtained calibration curve (Pf versus k) was

used to determine Pf from the experimental k values.Time-courses of glycerol influx were performed by mixing the red

cells with an equal volume of a hyperosmotic solution of glycerol to givean inwardly directed gradient of 100 mM. Glycerol uptake rates wereestimated by single exponential fits as reported by Dix et al. (21) forurea transport in red cells.

To determine pharmacological features of glycerol transport in red

blood cells, glycerol effluxes were performed in isoosmotic conditions, toavoid any initial water movement. A red cell suspension, equilibrated inCarlsen buffer complemented with 200 mM glycerol, was mixed in thestopped-flow apparatus to an equal volume of 200 mM sucrose in

Carlsen buffer. After mixing, an outwardly-directed glycerol gradient of100 mM was generated and glycerol efflux rates were estimated bysingle exponential fits.

In inhibition studies, red cells were preincubated 10 min at roomtemperature in the presence of 0.1 mM CuSO4, 0.1 or 0.5 mM phloretin,or 0.2 mM DIDS, or 5 min with 0.5 mM HgCl2. The inhibitor was presentin both the red cell suspension and the sucrose solution. Osmolalitieswere controlled using a Roebling osmometer. Results are means S.E.Statistics are calculated by Students t test.

One-step Pink Ghost Preparation and Glycerol Efflux Measure-mentsThe hemoglobin-free erythrocyte ghosts were prepared by the

method described by Ojcius (22). Briefly, washed normal and Colton-null red cells were lysed in a hypotonic medium, 5 m M Na2HPO4, pH

8.0, containing 20 g/ml phenylmethylsulfonyl fluoride, and 2 g/mlpepstatin, and centrifuged at 25,000 g for 20 min at 4 C. Themembranes were resuspended in 5 mM Na2HPO4, 100 mM NaCl, and100 mM glycerol, pH 8.0, at 108 ghosts/ml, and sealed by incubation at37 C for 1 h. Glycerol effluxes were measured at 20 C by a rapidfiltration method (23). Samples of 6 107 ghosts were loaded for 1 hwith 14C-glycerol (5 Ci/ml) and 3H-mannitol (5 Ci/ml). Mannitol wasused as an impermeant intracellular marker, to quantify the ghostsretained on glass-fiber filters (Whatman GF/A). The filters were rinsed

over a preset period of 500 ms to 10 s at a flow rate of 1 ml/s. The coldwashout solution contained 100 mM raffinose instead of glycerol. Theretained radioactivity (14C/3H) was quantified by liquid scintillation.Glycerol permeability coefficient (Pgly, in cm/s) was calculated from therate constant of the exponential time course of glycerol efflux and fromghost surface area to initial volume ratio.

Electrophoresis and ImmunoblottingHuman and rat hemoglobin-free ghosts were prepared as described above. Purified human AQP1was obtained as described previously (17). Bloodless rat kidney outermedulla was prepared according to Ecelbarger et al. (15). Protein con-centration of the membrane fractions was measured using the Pierce

BCA Protein Assay reagent kit. Five g of membrane proteins and 0.6g of pure AQP1 were denatured at 65 C for 10 min in Laemmli samplebuffer (24), and separated by 12.5% SDS-polyacrylamide gel electro-phoresis. Proteins were transferred to PVDF membranes (NEN LifeScience Products) and probed with the affinity-purified anti-AQP3 atapproximately 0.25 g/ml or the anti-AQP1 whole serum diluted 1:500.Controls were carried out by using the affinity-purified anti-AQP3preabsorbed for 10 min with 0.2 mg/ml of immunizing peptide. Immu-noreactive proteins were revealed by the ECL Western blotting tech-nique (Enhanced ChemiLuminescence, Amersham Pharmacia Biotech).

Indirect ImmunofluorescenceErythrocytes washed in PBS were

fixed for 1 h in 4% paraformaldehyde in PBS, followed by PBS washes.After a 3-min centrifugation at 2000 g, the cell pellet was infiltrated

in PBS containing 2.3 M sucrose. The samples were frozen in liquid

nitrogen, and 0.75 m sections were cut on an ultracryomicrotome at

70 C (Reichert Ultracut, Leica, Wien, Austria) and collected on Su-

perfrost Plus glass slides. Adult Sprague-Dawley male rats were anes-thetized (pentobarbital, 5 mg/100 g of body weight, intraperitoneally)

and perfused with a fixative containing 7.1 mM Na2HPO4, 30.4 mMNaH2PO4, 75 mM lysine, 10 mM sodium periodate, and 2% paraformal-dehyde (pH 7.4). Kidneys were removed, sliced, and kept in fixativeovernight at 4 C, followed by extensive PBS washes. Tissue slices wereinfiltrated overnightin PBScontaining 30% sucrose andfrozen in liquidN2. Cryosections of 5 m were cut on a cryostat (Leica) and collected onSuperfrost Plus glass slides. The slides were incubated for 5 min in PBScontaining 1% bovine serum albumin (PBS/BSA), followed by 1-h incu-bation in primary antibody (1:300 dilution of anti-AQP3 antiserum with

or without preincubation of the serum with the peptide used for immu-nization, or 1:100 dilution of anti-AQP1 anti-serum) in PBS/BSA. Thesections were then washed 3 10 min in PBS, followed by a 45-minincubation in CY3-conjugated mouse anti-rabbit antibodies (6 g/ml) inPBS/BSA. The sections were washed 2 10 min in PBS and mountedfor observation under a fluorescence microscope (Olympus Vanox-T).

RESULTS

Immunoblot Analysis of AQP3 Expression in Various Tis-

suesFig. 1A shows results from immunoblots probed with

affinity-purified antibodies raised against a rat AQP3 carboxyl-

terminus peptide. In rat renal outer medulla (Fig. 1A, lane 4),

a characteristic profile of three stained bands was observed:

one band at 25 kDa and two broader bands at 3340 and 67

kDa. None of these bands was detected when the anti-AQP3

antibody was preabsorbed with the immunizing peptide (Fig.

1A, lane 1). A similar profile with three stained bands was also

observed in rat red cell membranes probed with the anti-AQP3

antibody (Fig. 1A, lane 5), suggesting that, in addition to kid-

ney, AQP3 was also expressed in rat erythrocytes. Again, no

signal was detected when the anti-AQP3 was preincubated

with the immunizing peptide (Fig. 1A, lane 2). Although the 26

amino acids of human AQP3 carboxyl terminus differ fromthose of rat by three residues, the anti-AQP3 antibody also

recognized three bands in human normal ghosts (Fig. 1A, lane

7): a faint 25-kDa band and two broad 3748-kDa and 70-kDa

bands. An identical signal was observed in Co(a-b-) ghosts (Fig.

1A, lane 6). Similar to controls with rat membrane samples, no

signal was detected in human red cells when the anti-AQP3

was first blocked by the immunizing peptide (Fig. 1A, lane 3).

The anti-AQP3 antibody failed to recognize human AQP1 pu-

rified from red blood cells (Fig. 1A, lane 8), thus confirming the

specificity of anti-AQP3 antibodies. Most importantly, our re-

sults indicate that AQP3 is present in normal and Colton-null

human red cells.

Immunoblots were also carried out with an anti-human

AQP1 whole serum (Fig. 1B). A typical AQP1 profile was ob-

FIG. 1. Immunoblots of AQP3 (A) and AQP1 (B) in varioustissues. Five g of membrane proteins and 0.6 g of pure AQP1 wereloaded in individual lanes and were probed with either the affinity-purified anti-AQP3 at a concentration of 0.25 g/ml (A) or the anti-

AQP1 whole serum diluted 1:500 (B). A, lanes 13, the anti-AQP3antiserum was preabsorbed with 0.2 mg of immunizing peptide; lanes 1and 4, rat renal outer medulla; lanes 2 and 5, rat ghosts; lanes 3 and 7,normal human ghosts; lane 6, Colton-null ghosts; lane 8, purified hu-man AQP1. B, lane 1, purified human AQP1; lane 2, Colton-null ghosts;lane 3, normal human ghosts.

AQP3 in Red Blood Cells8408


3/6

served with purified AQP1 (lane 1) and normal human ghosts

(lane 3). The lower band in the human AQP1 lane may corre-

spond to a degraded form due to successive freezing/thawing of

the sample. No signal was detected in Colton-null ghosts (lane

2), confirming the absence of AQP1 in these cells. These results

indicate that the anti-AQP1 and anti-AQP3 antibodies do not

cross-react in the various tissues tested.

Indirect ImmunofluorescenceIn experiments on human red

blood cells, the anti-AQP3 serum strongly stained the plasma

membranes of both normal (Fig. 2a) and Colton-null cells (Figs.

2, c and d). No staining was observed when the antiserum was

preincubated with the peptide used for immunization (Fig. 2b,

normal cells). In contrast, the polyclonal antiserum against

purified AQP1 stained the plasma membranes of normal hu-

man red blood cells (Fig. 2e) but not those of Colton-null cells

(Fig. 2f). In rat kidney, the anti-AQP3 antibodies recognized

the basolateral plasma membranes of collecting duct principal

cells (Fig. 2g) but not those of adjacent intercalated cells (Fig.

2g, arrowheads). Although no staining was observed in other

cell types of the rat kidney nephron, as previously reported

(16), we also found staining of rat red blood cells (Fig. 2h).

Altogether, these results confirm the presence of AQP3 in both

human (normal and Colton-null) and rat red blood cellmembranes.

Water and Glycerol Permeabilities of Normal and Colton-null

Red CellsFig. 3A shows a typical time course of osmotic

shrinkage induced by a 100 mosm/kg H2O sucrose gradient,

carried out with thawed normal and Colton-null red cells at

26 C. Averaged Pf was (1.32 0.09) 102 cm/s and (2.16

0.16) 103 cm/s (n 6) for normal and Co(a-b-), respectively.

As previously reported (5), the Pf values in Co(a-b-) cells were

significantly lower than in normal cells.

The stopped-flow technique also allows studying solute

transport (21). Normal and Colton-null erythrocytes were sub-

mitted to an inward 100 mM glycerol gradient. In these condi-

tions, the time course of red cell volume change was biphasic

(Fig. 3B). After the initial cell shrinkage due to water efflux,

FIG. 2. Indirect immunofluorescence staining of red bloodcells and rat kidney. On cryosections, both anti-AQP3 (a) and anti-

AQP1 antiserum (e) stained the normal human red blood cell plasmamembranes. At two different magnifications, Colton-null human redblood cells were also stained by anti-AQP3 antibodies (c and d) but notby anti-AQP1 antibodies (f). No staining was observed when the anti-

AQP3 antiserum was preabsorbed with the peptide used for immuni-zation (b, normal red blood cells shown). In rat kidney nephrons, theanti-AQP3 only stained the basolateral plasma membrane of collectingduct principal cells (g), whereas intercalated cells were unstained (ar-rowheads). In addition, rat red blood cells were also stained by theanti-AQP3 antibodies (h). a-f, bars 5 m; g and h, bar 15 m.

FIG. 3. Measurements of water (A) and glycerol (B) transportin human normal (thick line) and Colton-null (thin line) eryth-rocytes by stopped-flow light scattering. A, a suspension of eryth-rocytes at 107 cells/ml was submitted to a 100 mosm/kg H2O sucrosegradient. The increase in 90 scattered light intensity, corresponding towater efflux, was recorded at 600 nm as a function of time. The curveswere fitted as single exponentials and Pf were calculated from rateconstants as described under Experimental Procedures. B, the dilute

suspension of erythrocytes was submitted to an inwardly directed 100mM glycerol gradient. The biphasic curves show the osmotically inducedcell shrinkage consecutive to initial water efflux, followed by concomi-tant glycerol influx and cell swelling. The rates of glycerol transportwere estimated by single exponential fits on the second part of thecurves.

AQP3 in Red Blood Cells 8409


4/6

cells progressively recovered their initial volume as glycerol

entered. Apparent rate constants of glycerol influxes at 26 C

were 0.12 0.02 s1

and 0.11 0.01 s1

(n 7) for normal andColton-null, respectively. Thus, no significant difference in

glycerol uptake rates was observed between the two cell types.

However, this approach relies on concomitant glycerol and

water flow, with the latter being possibly limiting in Colton-

null cells. This prompted us to measure directly the red cell

glycerol permeability (Pgly) using a tracer technique. Fig. 4

illustrates 14C-glycerol effluxes from normal and Colton-null

hemoglobin-free ghosts at 20 C. Calculated Pgly was 1.4

106 cm/s for control and 1.6 106 cm/s for Co(a-b-) ghosts.

These values are consistent with data in the literature (3, 25).

They further indicate that the permeability to glycerol is not

altered in Colton-null cells. Taken together, these results show

that both normal and Colton-null red cells exhibit a glycerol

transport capacity, which obviously, does not follow from AQP1

activity.

Inhibition StudiesThe effects of transport inhibitors were

tested on normal and Colton-null red cells using the stopped-

flow technique. Glycerol effluxes were followed in isoosmotic

conditions, to avoid initial water movements. Fig. 5 shows the

effects of 0.2 mM DIDS, 0.1 or 0.5 mM phloretin, 0.1 mM CuSO4,

or 0.5 mM HgCl2 on glycerol (Fig. 5A, normal; Fig. 5B, Co(a-b-))

and water (Fig. 5C, normal; Fig. 5D, Co(a-b-)) efflux rates.

DIDS inhibited the glycerol outflow from normal and Co(a-b-)

cells by approximately 45%. It also reduced the rate of water

efflux of both cell types, with a greater efficiency in Co(a-b-)

erythrocytes. A large inhibition of glycerol transport in normal

and Co(a-b-) red cells was observed in the presence of 0.1 mM

(80 and 67% inhibition, respectively) and 0.5 mM phloretin (93

and 73% inhibition, respectively). In addition, phloretin mark-edly inhibited water transport in Colton-null erythrocytes, at

the 2 concentrations tested (53% inhibition). CuSO4 signifi-

cantly reduced the glycerol permeability of normal (69% inhi-

bition) and Co(a-b-) (33% inhibition) red cells, and also strongly

reduced the water permeability of Colton-null erythrocytes

(60% inhibition). HgCl2 dramatically inhibited both glycerol

and water transport in normal and Co(a-b-) erythrocytes. Per-

centages of inhibition of glycerol efflux were 75 and 72% for

normal and Colton-null cells, respectively, and the water efflux

was inhibited by 90% in normal and by 50% in Colton-null

erythrocytes. This indicates the presence, in the latter, of a

residual mercury-sensitive path for water. In conclusion, all

the reagents inhibited glycerol transport in human normal and

Colton-null red blood cells in a similar manner. In contrast,

these molecules had differential effects on water transport in

both cell genotypes. These data support the idea that the major

path for water transport was disrupted in Colton-null cells

while that for glycerol was not.

DISCUSSION

AQP1 is the major water channel of human red blood cells.

However, as we showed here and as previously reported (5), a

significant mercury-sensitive part of osmotic water permeabil-

ity remains in AQP1-knocked out (Colton-null) red cells. Theexistence of additional, non-AQP1, protein pathways for water

transport has been discussed by Toon and Solomon (26). Red

blood cells also exhibit a high glycerol transport capacity,

whose molecular bases have remained unknown. AQP3 is a

mammalian aquaporin which, in addition to being moderately

permeable to water, is highly permeable to glycerol and, to a

lesser extent, to urea (11, 12, 14). In the present work, we

provide evidence that AQP3 is expressed in normal and Colton-

null red cells and likely mediates not only the residual water

transport, but also most of glycerol transport in human

erythrocytes.

Human AQP3 was detected by Western blotting in both

normal and Colton-null red cells and exhibited a migration

profile very similar to that of rat AQP3 in renal outer medullaand erythrocytes. The broad high molecular weight bands re-

vealed in human tissues appeared slightly larger than those of

rat tissues and may correspond to different glycosylation pat-

terns. Indirect immunofluorescence confirmed the presence of

AQP3 in the plasma membranes of normal and Colton-null

cells. Interestingly, the presence in red blood cells of AQP1 and

AQP3 constitutes a new example of aquaporin colocalization on

the same plasma membrane. AQP3 and AQP4 were colocalized

on the basolateral membrane of kidney collecting duct princi-

pal cells (16). AQP1 and AQP7, another aquaporin permeable

to glycerol (13), are colocalized in the kidney proximal tubule

brush-border membrane (27). AQP7 (13) and AQP8 (28) were

both detected in rat testis.

To investigate the relevance of AQP3-mediated transport in

erythrocytes, we characterized glycerol transport in both nor-

mal and Colton-null red cells. For this purpose, two distinct

methods were used. The rates of red cell volume changes were

estimated by light-scattering stopped-flow experiments on in-

tact cells submitted to an osmotic glycerol gradient. This tech-

nique, which requires small amounts of material, was particu-

larly useful for the study of Colton-null cells, which are not

easily available. However, stopped-flow light scattering meas-

urements do not allow accurate calculations of solute perme-

ability coefficients, because of uncertainties in red cell surface

area changes and refractive index effects (2, 29). Consequently,

we considered in these experiments only the shrinking or swell-

ing rate constants related to glycerol movements. In a second

type of experiment, the glycerol permeability was directly

measured by a rapid filtration method using one-step pinkghosts equilibrated in 14C-glycerol. In this case, the Pgly can be

calculated from the rate of radioactive glycerol efflux and was

1.4 and 1.6 106 cm/s in normal and Colton-null ghosts,

respectively, i.e. in the range of already published values for

human red blood cells (3, 25). These values are higher than the

Pgly of bovine red cells, lacking the facilitated diffusion mech-

anism for glycerol transport (30). These values remain also 2

orders of magnitude lower than the Pf of Colton-null cells,

indicating that, even in these cells, the rate of water flow was

not limiting (29). Thus, both stopped-flow and tracer methods

indicate a similar glycerol transport capacity in normal and

AQP1-deficient red blood cells. This means that, in its native

environment, AQP1 does not constitute a major pathway for

glycerol. We previously reported that AQP1 itself had a small

FIG. 4. Time courses of glycerol efflux from 100 mM 14C-glycer-ol-loaded resealed ghosts (filled squares, normal; open squares,Colton-null). Results are presented as the ratio of the apparent vol-ume of 14C-glycerol remaining in the cells (Vt) over the initial volume(V0) as a function of time. Pgly were calculated as indicated underExperimental Procedures.



5/6

glycerol permeability (17). However, the glycerol transport ca-

pacity of AQP1, estimated after expression in Xenopus oocytes,

is much lower than that of AQP3,2 and may therefore not be

detectable in the intact red cell membrane. The unitary Pgly ofthe bacterial GlpF is 2 105 molecules/s (31). If we assume

that the unitary Pgly of AQP3 is in the same range, the number

of AQP3 copies that account for Pgly in erythrocytes would be

around 15,000, i.e. at least ten times lower than the number of

AQP1 in these cells (32). Conversely, AQP1 appears to have a

higher unitary Pf than AQP3. This, with its higher abundance,

explains why AQP1 dominates water transport in the red blood

cell membrane.

A pharmacological characterization of glycerol and water

transport was performed in normal and Colton-null red cells.

All the reagents tested exerted similar inhibitory effects on

glycerol transport in normal and Colton-null red cells, confirm-

ing the idea that a protein, distinct of AQP1, mediates most of

glycerol transport in the red cell membrane. Some of the inhib-itors tested can have unspecific membrane effects (33), and

indeed, their use may be controversial. In particular, Ma et al.

(12) did not observe any inhibitory effect of 0.2 m M DIDS on

glycerol transport in human erythrocytes. In contrast, an inhi-

bition of water transport by DIDS has been reported by Toon

and Solomon (34) in human red blood cells. In our experiments,

DIDS was inhibitory both on glycerol and water effluxes. Cop-

per was found to affect the glycerol transport, as already de-

scribed (3), and also the water transport in an unexplained

fashion. Both mercury and phloretin have been shown to alter

water transport in AQP3-injected oocytes (11, 14). HgCl2 had

the strongest inhibitory effects, suggesting that this reagent

blocked the protein-mediated pathways for both water and

glycerol transport. Phloretin markedly inhibited the residual

water transport of the Colton-null cells, further supporting the

hypothesis that AQP3 could be the phloretin-sensitive waterchannel of these AQP1-deficient cells.

Taken together, our results demonstrate the presence of

functional AQP3 in the human red cell membrane, which can

account for the glycerol permeability of these cells and the

residual water permeability of AQP1-deficient erythrocytes.

Our findings may help explain the lack of clinical defects in

Colton-null patients (18). While AQP1 transports mostly water,

the significance of AQP3 expression in red blood cells may

reside in its ability to transport solutes. However, the role

devoted to a glycerol transporter in erythrocytes is not eluci-

dated. In contrast to testis or liver, where an important metab-

olism of glycerol has been described (13, 35), glycerol appears

not to be metabolized by erythrocytes. The presence of a glyc-

erol facilitator in red cells could make them less susceptible toosmotic stress upon local exposure to high glycerol concentra-

tions. A similar hypothesis has been advanced by Macey (36),

who suggested that the high urea permeability of human red

blood cells could protect these cells when they reach the deeper

regions of the renal medulla. AQP3 can also transport urea to

some extent and yet to be discovered solutes that can be im-

portant for the red cell osmoregulation. Also, the possibility

that a glycerol facilitator serves physiologically for another

purpose should not be dismissed.

AcknowledgmentsWe are very grateful to Dr. Pascal Bailly (Insti-tut National de Transfusion Sanguine, Paris) for generously providingus with the Colton-null phenotype human red blood cells and for helpfuldiscussions. We also thank Dr. Jean Labarre (Service de Biochimie et

Genetique Moleculaire, CEA/Saclay) for providing radiolabeled14

C-2

N. Roudier, unpublished data.

FIG. 5. Inhibition of glycerol and water transport in normal (A and C) and Colton-null (B and D) erythrocytes. The cells weresubmitted in a stopped-flow apparatus to a 100 mM outwardly directed glycerol gradient (A and B) or to a 100 mosmol/kg H2O inwardly directedsucrose gradient (C and D). Inhibitors (0.2 mM DIDS, 0.1 and 0.5 mM phloretin, 0.1 mM CuSO4, 0.5 mM HgCl2) were added 10 min (or 5 min forHgCl2) before the stopped-flow assay. Results are expressed as percentages of the maximal (no inhibitor) glycerol and water efflux rate constants.Results are means S.E. of the indicated number of individual experiments (0.001 p 0.030).

AQP3 in Red Blood Cells 8411


6/6

glycerol. The discussions concerning Pf calculations with Dr. JeanThiery (CEA/Cadarache) are greatly acknowledged. We also thank ourcolleague Dr. Germain Rousselet for helpful discussions and sugges-tions and Dr. Florent Guillain for advice in stopped-flow experiments.

REFERENCES

1. Macey, R. I., and Farmer, R. E. L. (1970) Biochim. Biophys. Acta 211, 1041062. Levitt, D. G., and Mlekoday, H. J. (1983) J. Gen. Physiol. 81, 2392533. Carlsen, A., and Wieth, J. O. (1976) Acta Physiol. Scand. 97, 5015134. Preston, G. M., Carroll, T. P., Guggino, W. B., and Agre, P. (1992) Science 256,

385387

5. Mathai, J. C., Mori, S., Smith, B. L., Preston, G. M., Mohandas, N., Collins, M.,van Zijl, P. C. M., Zeidel, M. L., and Agre, P. (1996) J. Biol. Chem. 271,13091313

6. Jacobs, M. H., Glassman, H. N., and Parpart, A. K. (1950) J. Exp. Zool. 113,277300

7. Reizer, J., Reizer, A., and Saier, M. H., Jr. (1993) Crit. Rev. Biochem. Mol. Biol.28, 235257

8. Maurel, C., Reizer, J., Schroeder, J. I., Chrispeels, M. J., and Saier, M. H., Jr.(1994) J. Biol. Chem. 269, 1186911872

9. Luyten, K., Albertyn, J., Skibbe, W. F., Prior, B. A., Ramos, J., Thevelein,J. M., and Hohmann, S. (1995) EMBO J. 14, 13601371

10. Echevarria, M., Windhager, E. E., Tate, S. S., and Frindt, G. (1994) Proc. Natl.Acad. Sci. U. S. A. 91, 1099711001

11. Ishibashi, K., Sasaki, S., Fushimi, K., Uchida, S., Kuwahara, M., Saito, H.,Furukawa, T., Nakajima, K., Yamaguchi, Y., Gojobori, T., and Marumo, F.(1994) Proc. Natl. Acad. Sci. U. S. A. 91, 62696273

12. Ma, T., Frigeri, A., Hasegawa, H., and Verkman, A. S. (1994) J. Biol. Chem.269, 2184521849

13. Ishibashi, K., Kuwahara, M., Gu, Y., Kageyama, Y., Tohsaka, A., Suzuki, F.,Marumo, F., and Sasaki, S. (1997) J. Biol. Chem. 272, 2078220786

14. Echevarria, M., Windhager, E. E., and Frindt, G. (1996) J. Biol. Chem. 271,2507925082

15. Ecelbarger, C. A., Terris, J., Frindt, G., Echevarria, M., Marples, D., Nielsen,S., and Knepper, M. A. (1995) Am. J. Physiol. 269, F663F672

16. Frigeri, A., Gropper, M. A., Turck, C. W., and Verkman, A. S. (1995) Proc. Natl.Acad. Sci. U. S. A. 92, 43284331

17. Abrami, L., Tacnet, F., and Ripoche, P. (1995) Pflugers Arch. Eu r. J. Physiol.430, 447458

18. Preston, G. M., Smith, B. L., Zeidel, M. L., Moulds, J. J., and Agre, P. (1994)Science 265, 15851587

19. Laize, V., Rousselet, G., Verbavatz, J. M., Berthonaud, V., Gobin, R., Roudier,N., Abrami, L., Ripoche, P., and Tacnet, F. (1995) FEBS Lett. 373, 269274

20. Jay, A. W. (1975) Biophys. J. 15, 20522221. Dix, J., Ausiello, D., Jung, C., and Verkman, A. (1985) Biochim. Biophys. Acta

821, 24325222. Ojcius, D. M., Toon, M. R., and Solomon, A. K. (1988) Biochim. Biophys. Acta

944, 192823. Martial, S., and Ripoche, P. (1991) Anal. Biochem. 197, 29630424. Laemmli, U. K. (1970) Nature 227, 68068525. Du, J., Kleinhans, F. W., Mazur, P., and Critser, J. K. (1994) Biochim. Biophys.

Acta 1194, 11126. Toon, M. R., and Solomon, A. K. (1996) J. Membr. Biol. 153, 13714627. Ishibashi, K., Kuriyama, R., Gu, Y., Kuwahara, M., Sasaki, S., and Marumo, F.

(1997) J. Am. Soc. Nephrol. 8, 18 (abstr.)28. Ishibashi, K., Kuwahara, M., Kageyama, Y., Tohsaka, A., Marumo, F., and

Sasaki, S. (1997) Biochem. Biophys. Res. Commun. 237, 71471829. Meyer, M. M., and Verkman, A. S. (1986) Am. J. Physiol. 251, C549C55730. Mazur, P., Leibo, S. P., and Miller, R. H. (1974) J. Membr. Biol. 15, 10713631. Heller, K., Lin, E., and Wilson, T. (1980) J. Bacteriol. 144, 27427832. Denker, B. M., Smith, B. L., Kuhadja, F. P., and Agre, P. (1988) J. Biol. Chem.

263, 156341564233. Jennings, M., and Solomon, A. K. (1976) J. Gen. Physiol. 67, 38139734. Toon, M. R., and Solomon, A. K. (1987) J. Membr. Biol. 99, 15716435. vom Dahl, S., and Haussinger, D. (1997) Am. J. Physiol. 272, G563G57436. Macey, R. (1984) Am. J. Physiol. 246, C195C203


evidence for the presence aqp3 in human red blood cells

Documents