| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
,
* Abteilung Vegetative Physiologie, Medizinische Hochschule Hannover, Hannover, Germany;
Institut National de la Santé et de la Recherche Médicale, Paris, France; and
Institut National de la Transfusion Sanguine, Paris, France
1Correspondence: Abt. Vegetative Physiologie Medizinische Hochschule Hannover, 30623-Hannover, Germany. E-mail: gros.gerolf{at}mh-hannover.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
0.15 cm/s, we find in red blood cells (RBCs) lacking the Rh protein complex (Rhnull) a significantly reduced PCO2 of 0.07 cm/s ±0.02 cm/s (P<0.02). This value is similar to the value we have reported previously for RBCs lacking aquaporin-1 protein (AQP-1null), suggesting that each of the Rh and AQP-1 proteins is responsible for 1/2 of the normal CO2 permeability of the RBC membrane. Four other blood group deficiencies tested lack diverse membrane proteins but exhibit normal CO2 permeability. The CO2 pathway constituted by Rh proteins was inhibitable at pHe= 7.4 by NH4Cl with an I50 of 10 mM corresponding to an I50 for NH3 of 0.3 mM. The pathway independent of Rh proteins, presumably that constituted by AQP-1, was not inhibitable by NH4Cl/NH3. However, both pathways were strongly inhibited by DIDS, which accounts for the marked inhibitory effect of DIDS on normal PCO2, while in contrast another AE1 inhibitor, DiBAC, does not inhibit PCO2, although it markedly reduces PHCO3-. We conclude that Rh protein, presumably the Rh-associated glycoprotein RhAG, possesses a gas channel that allows passage of CO2 in addition to NH3.—Endeward, V., Cartron, J.-P., Ripoche, P., and Gros, G. RhAG protein of the Rhesus complex is a CO2 channel in the human red cell membrane.
Key Words: Rhesus-associated glycoprotein • membrane CO2 permeability • ammonia • aquaporin-1
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
; ii) the red blood cell (RBC) membrane, for example, possesses an extremely high CO2 permeability, about half of which is due to the water channel aquaporin-1 (AQP-1) acting as a CO2 channel (4
, 5)
; and iii) this latter conclusion is supported by the previous observations that AQP-1 or its analog in the tobacco plant increase the CO2 permeability of Xenopus oocytes when expressed in these cells (5
6
7
8)
. The view that membrane proteins are responsible for a high rate of gas permeation across some membranes has been supported by evidence that MEP proteins (methylammonium permeases) mediate the flux of NH3 across the membrane of Saccharomyces cerevisiae (9)
. This view has greatly been strengthened by reports on the detailed molecular architecture of the NH3 channel of the bacterial ammonium-conducting protein AmtB (10
, 11)
and by the demonstration that the Rhesus-associated glycoprotein (RhAG) of the RBC membrane, a protein with significant homology to AmtB, also conducts NH3 rather than NH4+ (12)
. Here, we demonstrate for the first time by direct assay that a) the Rh proteins in the RBC membrane act as a channel for gaseous CO2; b) that this channel may not be selective for one gas, but probably allows passage of NH3 as well as CO2; and c) that the high CO2 permeability of the human RBC membrane essentially is due to the two blood group membrane proteins AQP-1 and RhAG. The function of Rhesus protein as a CO2 transporting protein is in excellent agreement with previous reports that (1)
the expression of a Rhesus protein in a green alga is regulated by the CO2 pressure and that (2)
lack of this Rhesus protein impairs the growth of this alga under conditions of high CO2 partial pressure (13
, 14)
. Part of the results of this paper have been presented in short preliminary form in a symposium report (15)
. The mass spectrometric method employed here to measure CO2 and HCO3– permeability of isolated cells by observing the exchange of 18O between CO2, HCO3– and H2O, as illustrated in Fig. 1
A, has been described previously (3
, 5
, 16)
.
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
illustrates the principle of the mass spectrometric technique used to quantitate PCO2 of RBCs. RBCs are introduced into a solution of 18O-labeled CO2-HCO3– solution. In a rapid first phase, labeled CO2 enters the RBCs where, due to the presence of a high activity of carbonic anhydrase, it is rapidly converted to labeled HCO3–. This does not yet result in isotopic equilibrium, because in a relatively slow process most of the 18O-label is transferred into water, whose concentration (55 M) is orders-of-magnitude greater than that of CO2 and HCO3– (1.2 and 24 mM, respectively). Thus, some 18O is transferred into the water pool with each dehydration reaction step, resulting in a continuing decay of the concentration of labeled CO2 in the system, as seen in the initial portion of the record of Fig. 1B
. After the addition of red cells, the approach to isotopic equilibrium is greatly accelerated, as seen in the subsequent phase of the record. After the first phase of rapid entry of labeled CO2 into the RBCs, a slower phase of C16O18O decay follows, as apparent in Fig. 1B
, which is characterized by the much slower entry of labeled HCO3– into the cells as indicated in Fig. 1A
. Again, the label entering the RBC by this pathway is transferred eventually into the water pool and then lost from the CO2-HCO3– pool. For the continuous measurement of C16O18O in the cell suspension an inlet system originally developed by Itada and Forster (16)
was employed. To derive PCO2 and PHCO3– from the experimental curves, knowledge of the intracellular carbonic anhydrase activity of the RBCs was necessary and was determined independently in RBC lysates under conditions of pH and [Cl–] as they prevail in the RBC interior. All mass spectrometric measurements were performed at 37°C.
For this process, originally utilized by Itada and Forster (16)
to measure CA activity and PHCO3–, we have reported a system of six differential equations that can be used to determine values of PCO2 and PHCO3– from experimental curves of the decay of 18O-labeled CO2 as shown in Fig. 1B
(3)
. These equations describe the reaction rates and fluxes of the labeled CO2, labeled HCO3– and labeled water species in and between the extra- and intracellular compartments of the dilute red cell suspensions (0.03% hematocrit) as studied in the mass spectrometric measurement. The constants inserted into this system of equations were: i) ku, the velocity constant of CO2 hydration at 37°C as calculated from the first linear segment of the decrease of 18O-labeled CO2 with time (see Fig. 1B
), 0.13 ± 0.01 s–1 (SD, n=90), a figure that agrees well with data in the literature (17)
; ii) the permeability of the human RBC membrane to water, 2 x 10–3 cm/s (18)
; iii) the surface-to-volume ratio of normal human red cells (a) of 20,000 cm–1 (16)
; this latter figure was used for normal as well as for Rhnull RBC, as no precise quantitative information on a is available for Rhnull cells. However, as discussed below, the surface-to-volume ratio of Rhnull RBCs may be 
10% below the normal value. In the case of RBCs treated with ammonium, red cell volume was determined experimentally for each NH4Cl concentration and used to correct a under the assumption of an unaltered surface area (values of a calculated under this assumption varied between 20,000 and 16,000 cm–1); iv) red cell volume v in the final dilution was determined from hematocrit and intraerythrocytic water space. The extracellular pH, pHe, was adjusted in each experiment to 7.40, and intracellular pH, pHi, was taken to be 7.20, a number checked in several instances on blood not older than 2 days and found to be 7.18 ± 0.02 (SD, n=6). In one case, a Rhnull sample had traveled for 5 days before the day of the measurement, giving a somewhat lower pHi as seen below (
Fig. 4A
). In this case the control RBCs were adjusted to the same pHi. The intracellular acceleration of CO2 hydration by carbonic anhydrase in the cells, Ai, equal to the intracellular reaction velocity constant/ku, was determined independently for each blood sample by assessing the CA activity of a RBC lysate at suitable dilution by mass spectrometry. For normal and for Rhnull RBCs we obtained Ai = 21,000 ± 2400 (SD, n=86) and Ai = 20,400 ± 1900 (n=20), respectively. It may be noted that SD falls to about ±900 when measurements from a given blood sample are considered only. The transport inhibitors employed (see below) at the concentrations used did not affect CA activity. The red cells used to study the effect of NH3/NH4+ exhibited a decrease in Ai from 20,900 at [NH4Cl] = 0 to 20,300 at [NH4Cl] = 50 mM, due to opposing effects on CA activity of increasing pHi on one hand and increasing cell volume and [Cl–]i on the other hand. For the calculation, we used the actual measured values of Ai and pHi (see below). In all calculations of this paper, PCO2 and PHCO3– were the only parameters derived from the biphasic records obtained after the addition of RBCs (example given in Fig. 1B
).
|
|
|
Sensitivity analysis of calculated permeabilities to parameters used in calculations
Figure 2
shows a sensitivity analysis of calculated PHCO3– and PCO2 for the major parameters just discussed and inserted into the fitting procedure applied to the experimental curves of C18O16O decay. The results show that a change of a of the RBC causes changes in both PHCO3– and PCO2 of equal size and opposite sign. The first apparent dissociation constant of carbonic acid K'1 is critical but is entirely dependent on the well-controlled parameters temperature and ionic strength. The intraerythrocytic CA activity Ai is quite critical, especially for PCO2 as discussed before (3)
and has, therefore, been separately and independently measured for each blood sample studied. Similarly, extracellular pHe is important for PCO2 but is carefully controlled in each experiment within 0.01 U. Nevertheless, minimal changes in pHe might be a significant source of scatter in the present technique. pHi is quite uncritical for PCO2 (but not for PHCO3–), and membrane water permeability PH2O is so large in the RBC membrane that it is far in excess of what could become limiting for the entire process of 18O exchange, as discussed before (3)
. This shows that special care has to be taken to closely control a, Ai, pHe, and temperature, as we have attempted to do here.
Blood samples and reagents
The present studies were conducted using normal human RBCs from members of the laboratory in Hannover, and samples typed as RhD-positive and RhD-negative samples in the laboratory in Paris, and Rhnull RBCs (from 4 unrelated blood samples) obtained through the collections of the INTS in Paris. Rhnull RBCs have a severe defect of the Rh blood group antigens, which normally form a complex that includes the Rh and RhAG proteins, as well as the accessory chains LW, CD47, and glycophorin B (19
, 20)
. Rhnull RBCs, however, have a normal content of AQP-1 (12)
. The molecular defects of Rhnull RBCs are well characterized and result from diverse mutations of either one of the genes encoding the RhAG or Rh protein, such that when one of these proteins is lacking, the whole Rh complex does not reach the cell surface (21)
. Transport studies in RBC ghosts deficient in RhAG from a variety of human and mouse blood specimen have shown that it is RhAG, but not the other proteins of the complex, that is responsible for ammonia transport across the erythrocyte membrane (12)
.
There are a few other parameters of Rhnull cells whose possible influence on the determination of PCO2 should be considered. Besides their lack of Rh proteins, Rhnull erythrocytes may have a slightly reduced a due to mild stomatocytosis (22
, 23)
. In the extreme case of a complete spherocytosis, a would be reduced by 20% (24)
, resulting in a maximal underestimation of PCO2 by 20% (see Fig. 2
). The actual effect due to the mild stomatocytosis of Rhnull blood, far from a complete spherocytosis, is expected to be 10%, a value that agrees roughly with the moderate morphological alterations of these cells (23
, 24
, and J.-P. Cartron, unpublished results). Because the product of a x PCO2 is what determines the experimental decay of C16O18O, this finding would raise the true PCO2 of Rhnull RBCs from 0.066 to 0.072 cm/s, implying a still significant 50% rather than 55% reduction of PCO2 compared to normal RBCs. This argument is supported by the finding shown below (see Fig. 3
A, B) that PHCO3– in Rhnull RBCs clearly is not decreased as expected if the parameter a was seriously overestimated, because a enters identically into the calculation of both permeabilities. Red cell volume in Rhnull blood has been reported to be about normal (23
, and J.-P.C., unpublished). Although Rhnull RBC membranes have been reported to possess a decreased phospholipid asymmetry (25)
, it has been shown that overall membrane lipid composition and membrane fluidity at different depths in the membrane are unaltered (22
, 26)
. In addition, it has been reported that CO2 permeability of lipid bilayers does not depend on bilayer fluidity (27)
. Therefore, we do not expect major alterations in CO2 permeability of the Rhnull membrane bilayer.
Other blood samples with various blood group deficiencies analyzed here were obtained through INTS, including Jk (a-b-)/Jknull (2 samples; Jk protein is the carrier of the Jk/Kidd antigens and a well-characterized urea transporter of human RBCs), Fy (a-b-)/Fynull (3 samples; Fy is the carrier of the Fy blood group antigens and receptor for P. vivax merozoites and for chemokines of the IL8 family), AQP-1null (1 sample; AQP-1 deficiency; AQP-1 carrier of Colton blood group antigens), K0/Kellnull (1 sample; the Kell protein is a Zn-metalloprotease) and a McLeod sample (1 sample; Kx protein, a polytopic membrane protein with a putative transport function, is missing in McLeod RBCs together with part of the Kell protein). The molecular defects of these rare bloods are also well characterized (28
, 29)
. All RBCs were used within at least 4 days, usually much less, after blood withdrawal.
Loading of RBCs with various concentrations of NH4Cl (at pH 7.4) was done with solutions in which an equimolar part of the NaCl concentration of the medium was omitted. Exposure of RBCs to NH4Cl was started with 10 mM and was increased over a total time period of 4 h in steps of 10 mM to the final NH4Cl concentration.
pHi of red cells was measured after equilibrating 40% RBC suspensions at 37°C with 5%CO2/95%O2, either in the absence or presence of NH4Cl. The suspensions were sucked into glass capillaries, centrifuged, frozen at –20°C, and thawed, all steps being performed under anaerobic conditions. From the portion of the capillary containing the lysed packed RBCs, a sample was transferred anaerobically into a blood gas analyzer (ABL, Radiometer, Copenhagen) and pH was determined at 37°C.
DIDS (4,4'-diisothiocyanato-stilbene-2.2'-disulfonate; Sigma-Aldrich, Seelze, Germany) was used as a previously described inhibitor of RBC permeability for CO2 as well as for HCO3– (4
, 5
, 30)
. DiBAC (or diBA; bis(1,3-dibutylbarbituric acid)pentamethine oxonol; Invitrogen GmbH, Karlsruhe, Germany) was used as an established inhibitor of HCO3––Cl– transfer by AE1 (31)
. All mass spectrometric experiments with RBCs were performed in the presence of the impermeable extracellular carbonic anhydrase (CA) sulfonamide inhibitor 1-[5-sulfamoyl-1,3,4-thiadiazol-2-yl-(aminosulfonyl-4-phenyl)]-2,4,6-trimethyl-pyridinium perchlorate (STAPTPP) as described by Casey et al. (32)
, at a final concentration of 1 x 10–5 M. All other chemicals were reagent grade.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
), the time course of C18O16O decay as measured by mass spectrometry is consistently slower with Rhnull than with normal human RBCs (Fig. 1B
). Since the first rapid phase of these decay curves is determined to a large extent by the CO2 permeability PCO2 (in addition to intraerythrocytic CA activity Ai), while the second phase is strongly affected by PHCO3– (3)
, the calculations on the basis of the published model of C18O16O exchange (3)
yield a PCO2 of Rhnull RBCs that is significantly lower [values of Fig. 3
B: PCO2=0.070±0.011 (SD) cm/s, number of bloods studied n=4, number of mass spectrometer experiments n=31] than that of normal RBCs [again data of Fig. 3B
: PCO2=0.16±0.06 (SD) cm/s, n=7; n=50] (P<0.05), while the bicarbonate permeabilities are similar (Fig. 3A
). It should be noted that the data given in Table 1
comprise a different group of control RBCs, and the statistical treatment of the Rhnull RBC results is different as explained in the legend to Table 1
, but nevertheless the results are almost identical to those of Fig. 3B
. It is discussed below that PCO2 of Rhnull RBCs may be slightly (10%) underestimated because the a of these cells may be somewhat lower than that of normal RBCs. Nevertheless, the results show that about half of the CO2 permeability of human RBCs is lost when the proteins that make up the Rh complex (19
20
21)
are lacking although AQP-1 is expressed normally (12)
. We conclude that the observed reduction of PCO2 in Rhnull cells compared to normal RBCs is due to the absence of Rh proteins in these cells. In Materials and Methods we present arguments that altered RBC shape and the possibly altered lipid properties of Rhnull cells do not contribute to this effect (22
23
24
25
26
27)
. We note that it has not been possible to demonstrate such an effect of Rh proteins (and neither one of DIDS) on CO2 permeation across RBC ghost membranes in stopped flow studies reported by Ripoche et al. (33)
. This finding may be due to the extremely fast kinetics of CO2 permeation in RBC ghosts with a considerably lower intracellular buffer capacity than in intact RBCs. The stopped-flow technique is very difficult to apply to such a fast reaction, which may be the cause of the controversial results obtained with this method regarding the contribution of AQP-1 to CO2 permeation across membranes (34
, 35)
. In principle, the chances of picking up an effect of a CO2 channel should be improvable by slowing down the process of CO2 uptake through establishing very high intracellular H+ buffer capacities.
|
Figure 3C, D
shows that DIDS (4,4'-diisothiocyanato-stilbene-2,2'-disulfonate), besides inhibiting the Cl––HCO3– exchanger and thus PHCO3–, is an efficient inhibitor of RBC PCO2, as reported previously (30)
, and acts in a dose-dependent manner with maximal inhibition reached with 10 µM DIDS (data not shown). In normal RBCs, 10 µM DIDS reduces PCO2 to 1/3, from 0.15 or 0.16 cm/s to 0.05cm/s, i.e., it causes a reduction in PCO2 by 0.10–0.11 cm/s (Fig. 3D
, Table 1
). In Rhnull RBCs, DIDS is still effective, although less so, and causes PCO2 to fall from 0.066–0.070 to 0.029 cm/s, i.e., by 0.04 cm/s (Fig. 3B
, Table 1
). Thus, 60% of the inhibitory effect of DIDS, or 0.065 cm/s, is lost when the Rh protein complex is missing. We conclude that Rh proteins are responsible for 60% of the decrease in PCO2 produced by DIDS. We might note that the effect of 10 µM DIDS on PHCO3– seen here is somewhat less than expected from the literature, a phenomenon that we have discussed previously in more detail and attribute to some moderate red cell lysis, which will raise the apparent PHCO3– but will hardly affect PCO2 (5)
.
Figure 3E
shows that another inhibitor of Cl––HCO3– exchange in red cells, DiBAC [bis(1,3-dibutylbarbituric acid)pentamethine oxonol; (31)
], produces a significant reduction of PHCO3– in a dose-dependent fashion, but, in contrast to DIDS, does not affect CO2 permeability at all concentrations studied. This shows that transfer of CO2 is not inhibited by DiBAC concentrations that significantly inhibit HCO3– transfer and suggests that CO2 transfer does not occur via the HCO3– transport path of AE1. As discussed below, the effect of DIDS on PCO2 seems to result from its actions on Rh and AQP-1 proteins rather than from its action on AE1.
By using RBC ghosts from normal and Rhnull blood it has been shown that RhAG, a member of the Rh protein complex, acts as a channel for NH3 (12)
. We have, therefore, asked whether CO2 and NH3 pass the RhAG protein via the same channel. If the gas channel of the RhAG protein is similar to the NH3 channel of the bacterial AmtB, it is expected in view of the molecular dimensions of the channel and of the two gases that either gas can permeate the channel only in single-file manner. This implies that the two gases compete with each other for entry into and passage through the channel. We have, therefore, tested whether the presence of NH3 affects the apparent permeability of RhAG to CO2. Figure 1B
shows that the decay of C18O16O occurs more slowly in the presence of NH4Cl than under control conditions, similar to what is seen with Rhnull RBCs. Figure 4
illustrates that the presence of NH3/NH4+ indeed reduces PCO2. Figure 4
A shows that loading Rhnull and normal RBCs with NH4Cl in an identical manner by the procedure described in Materials and Methods causes identical increases in intraerythrocytic pH at given concentrations of NH4Cl. All experiments of Figs. 4A, B
were conducted at an extracellular pH of 7.40 and at variable intracellular pH values as seen in Fig. 4A
. At a nominal extracellular NH4Cl concentration of 50 mM, the extracellular concentration of NH3 is calculated to be 1.6 mM (pKa 8.90, 37°C) and may be assumed to be identical in the intracellular space. In comparison, the intra- and extracellular concentration of unlabeled plus labeled dissolved CO2 is 1.2 mM in all mass spectrometer experiments. The results of PCO2 determinations at different NH4Cl concentrations are given in Fig. 4B
for normal and Rhnull RBCs. The calculations were done using the experimentally determined pHi values and mean erythrocytic volumes as obtained from RBC counts and hematocrits of RBC suspensions. Figure 4B
shows that PCO2 of normal RBCs is inhibited by increasing concentrations of NH4+/NH3 in a dose-dependent manner. In addition, NH4+/NH3 clearly has no effect on PCO2 of Rhnull RBCs at all concentrations studied. NH4+/NH3 reduces PCO2 of normal RBCs just to the value of Rhnull RBCs but not further. Figure 4C
shows the differences 
in PCO2 between the values at the different NH4Cl/NH3 concentrations and the minimal value obtained at NH4Cl concentrations of 40–50 mM, plotted vs. NH3 concentration. If the inhibitory effect is attributed to NH3, this curve allows us to estimate an I50 of NH3 of 0.3 mM. The open triangles in Fig. 4C
represent experiments conducted with various NH4Cl concentrations under conditions of a constant pHi of 7.2 but variable extracellular pH (ranging from 7.40 to 7.26). The data show that both experimental conditions yield near-identical results and indicate that variation of pHi is not responsible for the inhibitory effect of NH4+/NH3 seen in Fig. 4B, C
. It may be noted that NH4+/NH3 had no effect on PHCO3–, which in the experiments of Fig. 4B
varied unsystematically between values of 0.0012 and 0.0015 cm/s (data not shown). In conclusion, these data indicate that NH4+/NH3 can fully inhibit the RhAG-protein-mediated pathway of CO2 but does not affect other pathways for CO2 such as AQP-1.
Since the CO2-transporting membrane proteins RhAG and AQP-1 are both blood group proteins, we have done an extensive study of several blood group deficiencies, which is shown in Table 1
. We draw four conclusions from the results: i) Lack of either RhAG or AQP-1 reduces PCO2 to roughly 
of the control value, suggesting that most of the CO2 permeability of the membrane is due to these two proteins. ii) Consistent with this conclusion, PCO2 of AQP-1null RBCs in the presence of DIDS is as low as 0.015 cm/s (Table 1)
. In this condition the CO2 channel of AQP-1 is lacking and most of the CO2 channel of the RhAG protein is inhibited by DIDS, as discussed above. iii) No other blood group protein investigated contributes to membrane CO2 permeability. This holds for the Fy, Jk, Kell, and Kx proteins (28
, 29
, see also Materials and Methods). iv) In addition, RhD-positive and RhD-negative RBCs exhibit identical PCO2s, indicating that the RhD protein is not involved in CO2 permeation.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, last two lines) shows that a) DIDS still inhibits PCO2 of either cell type; however, b) its inhibitory effect in both cases is markedly lower than it is in normal RBCs. While in normal RBCs DIDS reduces PCO2 by 0.10 cm/s (data of Fig. 3
and Table 1
), it reduces PCO2 in Rhnull RBCs by 0.04 cm/s and in AQP-1null RBCs by 0.05 cm/s. The latter two values add up to 0.09 cm/s, a figure that corresponds nicely to the total inhibitory effect of DIDS in normal RBCs. The simplest explanation of this finding is that DIDS inhibits the CO2 pathways of both RhAG and AQP-1 and that the sum of these inhibitory effects represents the DIDS effect on PCO2 of normal RBCs. In the case of AQP-1 expressed in Xenopus oocytes, we have previously confirmed independently that DIDS indeed exerts a direct inhibitory effect on CO2 permeation through this protein, even in the absence of other RBC membrane proteins (5)
. In AQP-1null RBCs in the presence of 10 µM DIDS (Table 1
, last line) the lowest PCO2 of this study with a value of 0.015 cm/s is observed. Under this condition, AQP-1 is absent and most of the CO2 permeation through RhAG is inhibited. Thus, if no other proteins besides AQP-1 and RhAG are involved in CO2 permeation, this figure may be close to the CO2 permeability constituted by the membrane lipids of the RBC. This would suggest that only 1/10 of the PCO2 of normal RBCs is mediated by membrane lipids and that membrane proteins are responsible for most of the CO2 transported across the membrane. We have previously reported (3)
that the PCO2 of the apical membrane of colon epithelium is 0.001 cm/s, which is even 10 times lower than the above figure and supports earlier qualitative findings. For artificial membranes, Blank and Roughton (36)
have observed low PCO2 values of 0.002 to 0.01 cm/s in various monolayers, although Gutknecht et al. (37)
have reported a rather high PCO2 of 0.35 cm/s in a lipid bilayer. We conclude that the CO2 permeabilities of the lipid phase of the RBC membrane, as well as of some other biological and artificial membranes, are rather low. The picture emerging appears to be that a) the basal CO2 permeability of biological membranes in the absence of CO2-conducting proteins may be rather low, 0.001 to 0.01 cm/s, although probably variable between different cell types (3
, 38)
, and b) when a high PCO2 is functionally required such as in the case of the RBC, this is accomplished by the presence of protein CO2 channels. This concept is a major deviation from the paradigm that all gases pass easily through the lipid portion of all biological membranes [e.g., Alberts et al. (39)
].
Molecular basis of the high CO2 permeability of the red cell membrane
We identify here two membrane proteins as CO2 channels in the RBC membrane, RhAG and AQP-1. The lack of an effect of DiBAC on PCO2 suggests that AE1 is not involved in CO2 transfer across the RBC membrane, and this is supported by our finding that the contributions to PCO2 of Rh and AQP-1 roughly add up to the total PCO2 of the RBC membrane. RhAG is a member of a large protein complex that includes the Rh proteins (D and CcEe proteins), which form hetero-oligomers in the membrane and are associated with accessory proteins such as LW, CD47, and GPB (20)
. Furthermore, in the RBC the Rh complex is linked to the membrane skeleton through ankyrin R (40)
, an adaptor protein that also connects the AE1 protein (anion exchanger) to the membrane skeleton, thus defining a macrocomplex with a putative gas transport function (41)
. In the Rhnull RBCs studied here Rh and RhAG are absent or drastically reduced, while AQP-1 as well as band 3 are expressed normally (12)
. Conversely, in AQP-1null RBCs, AQP-1 is completely lacking, but band 3, Rh, and RhAG proteins and several other membrane proteins are expressed normally (12
, 42)
. This shows that the effects on PCO2 of the deficiencies of AQP-1 and RhAG observed here are specific for each of the two proteins. It may be asked, however, whether each of the two proteins independently provides a channel for CO2 or whether the channel is constituted by a protein complex of which these proteins are members. That AQP-1 conducts CO2 in the absence of other proteins of the RBC membrane has been shown with various methods for human AQP-1 expressed in Xenopus laevis oocytes (5
6
7)
. The effect of pCMBS on the contribution of AQP-1 to CO2 permeation suggests that CO2 passes at least partly through the water pore of the aquaporin monomer. Furthermore, in the case of AQP-1 the majority of the available evidence is in favor of an independent mobility of the protein in the membrane and no association with the Rh membrane complex (41
, 43)
, although some interaction between these proteins may not be excluded (33)
. Thus, we propose that AQP-1 constitutes an independent channel for CO2 as discussed previously, possibly through both the water pores of the AQP-1 monomers and the central pore of the tetramer (5)
. With regard to the role of the Rh complex, the finding shown in Fig. 4
of an inhibitory effect of NH3 on CO2 permeation mediated by Rh suggests that CO2 is likely to pass through the NH3 channel of the RhAG protein: i) it is well established that it is not the Rh proteins (D or CcEe) but RhAG that conducts ammonium (12
, 44
, 45)
; ii) as it has been shown that RhAG as well as the bacterial AmtB/Mep proteins conduct ammonia rather than ammonium ions (9
10
11
12)
, it is likely that the effect on PCO2 seen in Fig. 4B, C
is exerted by NH3 competing with CO2 for entrance and passage of the same gas channel. Alternatively, it is conceivable that the NH4+ that is bound in the vestibule at the entrance to the gas channel and thought to serve the recruitment of NH3 (10)
impedes access of CO2. In either case, the present observation is compatible with CO2 passing through the NH3 channel of RhAG. In line with this, it may noted that the molecular diameter of CO2 [2.9 to 3.3 Å; (46
, 47)
] is quite similar to that of NH3 [3.0 to 3.15 Å; (47
, 48)
], which should allow either molecule to pass this channel. The hydrophobic environment that prevails inside the NH3 channel (10
, 11)
should also be suitable for CO2, which has a high solubility in lipids and nonpolar solvents (17)
. Similar considerations hold for the water channel of AQP-1 (5)
. It should be noted that in principle there may be alternative pathways for CO2 across the Rh oligomer. Callebaut et al. (49)
have speculated that either Rh protein (CcEe or D) might be a candidate for a CO2 channel (while the present finding of identical PCO2 in RhD-negative and positive RBCs argues against the latter possibility), or the central hole of a putative trimer with a diameter of 5 Å might be a pathway for CO2. These alternative pathways would not explain the competition between CO2 and NH3 as simply as the RhAG pathway.
Regarding the relative affinities of CO2 and NH3 for the gas channel, it may be noted that the IC50 of 0.3 mM for inhibition of PCO2 by NH3, as apparent in Fig. 4C
, was obtained in the presence of 1.2 mM CO2. Thus, an ammonia concentration of 
of the CO2 concentration is sufficient to halve CO2 permeability, which would suggest that the affinity of the channel is somewhat greater for NH3 than for CO2.
CO2 fluxes through RhAG and AQP-1 gas channels
We have determined PCO2 values at a concentration of dissolved CO2 (unlabeled plus labeled) of 1.2 mM. With a normal PCO2 of 0.15 cm/s, and PCO2 values in the absence of either RhAG or AQP-1 of 0.07 cm/s, one obtains a rough estimate of the contribution to CO2 permeability by each, RhAG and AQP-1, of 0.08 cm/s. Using the latter value and a RBC surface area of 145 µm2, one calculates a unidirectional flux of CO2 across the membrane of one RBC of 1.4 x 10–13 mol/s, or 0.84 x 1011 gas molecules per second. For the RhAG channel, we use the number of 81 x 103 copies of RhAG per RBC (12)
to obtain a single channel transport rate of 1.0 x 106 CO2 molecules per channel and per second. The same calculation for the AQP-1 channel, using a number of AQP-1 copies per RBC of 2 x 105 (50
, 51)
, gives a single channel transport rate of 0.42 x 106 molecules per channel per second. These numbers are similar and it is of interest to note that both numbers are quite high, and much higher than estimated by Zheng et al. (11)
for the transfer of ammonia though the RhAG channel, but similar to the transfer rate of 2 x 106 for NH3 across human RBC RhAG estimated by Ripoche et al. (12)
under different conditions. The diffusion limit of the transport of substrates through open channels is in the order of 108 molecules/s. Because the present CO2 transport rates do not represent maximal transport rates (a saturation curve of CO2 transport has not been established and may be difficult to measure), it appears possible that the maximal rate of transfer of CO2 across the channels is in fact close to the diffusion limit. This would support the idea that gas transfer indeed occurs through an open channel and does not involve major conformational changes of the transport protein, which would typically reduce the speed of transfer to <104 molecules/s.
Physiological significance of protein-mediated CO2 permeability
We have previously described a way of calculating the time required by RBC to release CO2 as it passes through the lung capillary and as its CO2 partial pressure falls from the venous value of 46 mmHg to the arterial value of 40 mmHg (52)
. This calculation considered diffusion and chemical reaction inside the RBC and neglected any effect by the RBC membrane. We have recently extended this model to take into account the diffusion resistance exerted by the RBC membrane in addition to the simultaneous diffusion and reaction process in the intracellular space of the RBC (5)
. Table 2
summarizes some results. In the absence of any membrane resistance, 50 ms are required for the process of CO2 release in the lung to reach 95% completion (t95%) and with normal membrane resistance (CO2 permeability of 0.15 cm/s) this time increases to 100 ms. Both times are far below the capillary transit time in the lung of 700 ms under resting conditions. For Rhnull or AQP-1null RBCs with a PCO2 of 0.07 cm/s, t95% increases to 180 ms, which is still considerably less than the available transit time. However, when PCO2 falls to a basal value of 0.01 cm/s in the absence of both CO2 channel proteins, t95% rises to 1000 ms. This time is longer than the capillary transit time under resting and even more so under conditions of exercise, when transit time may decrease to 300 ms. This implies that lack of the protein channels will require a rise in venous CO2 partial pressure to maintain CO2 output of the body. The high CO2 permeability of human RBCs may thus be seen as an optimization of the conditions of CO2 exchange in lung and tissues. If the two channels can transport O2 with similar efficiency as CO2, they may be of even greater significance for O2 uptake in the lung and O2 release in the tissues.
|
|
ACKNOWLEDGMENTS |
|---|
Received for publication June 1, 2007. Accepted for publication July 19, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. C. N. Brown, D. Hallouane, W. J. Mawby, F. E. Karet, M. A. Saleem, A. J. Howie, and A. M. Toye RhCG is the major putative ammonia transporter expressed in the human kidney, and RhBG is not expressed at detectable levels Am J Physiol Renal Physiol, June 1, 2009; 296(6): F1279 - F1290. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Musa-Aziz, L.-M. Chen, M. F. Pelletier, and W. F. Boron Relative CO2/NH3 selectivities of AQP1, AQP4, AQP5, AmtB, and RhAG PNAS, March 31, 2009; 106(13): 5406 - 5411. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Endeward and G. Gros Extra- and intracellular unstirred layer effects in measurements of CO2 diffusion across membranes - a novel approach applied to the mass spectrometric 18O technique for red blood cells J. Physiol., March 15, 2009; 587(6): 1153 - 1167. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. J. Bruce, H. Guizouarn, N. M. Burton, N. Gabillat, J. Poole, J. F. Flatt, R. L. Brady, F. Borgese, J. Delaunay, and G. W. Stewart The monovalent cation leak in overhydrated stomatocytic red blood cells results from amino acid substitutions in the Rh-associated glycoprotein Blood, February 5, 2009; 113(6): 1350 - 1357. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Significant difference from the left neighboring column (P<0.05). All mass spectrometric experiments with RBCs reported here were performed in the presence of the extracellular CA inhibitor STAPTPP (32)
) indicate the results of measurements conducted with a constant intracellular pH of 7.20 and the extracellular pH varying with NH4Cl concentration between 7.40 ([NH4Cl]=0) and 7.26 ([NH4Cl]=50 mM). Value of n for closed circles is as indicated in B for control cells, each triangle is the average of two measurements. *Significant differences between Rhnull and control RBCs (P=0.022 for [NH4Cl]=0 and P<0.002 for [NH4Cl]=10 mM).