The monovalent cation leak in overhydrated stomatocytic red blood cells results from amino acid substitutions in the Rh-associated glycoprotein

Overhydrated hereditary stomatocytosis (OHSt) was first reported in 1961¹  and has been extensively studied. The red cells have an invaginated cell shape suggestive of inward membrane bending. The red cells leak cations at a rate 20 to 40 times greater than normal and show a monotonic temperature dependence of the cation leak.²  The massively increased permeability of these cells to the monovalent cations Na⁺ and K⁺ elicits a major increase in Na⁺K⁺-ATPase activity that in turn mandates an increase in glycolytic activity to provide ATP.²  The cells have multiple abnormalities in the trafficking of membrane proteins, notably that of stomatin, a 32-kDa raft protein. Stomatin is lost from these cells as they mature,³  and was named after this disease. Some patients with OHSt show subtle differences in red cell behavior. OHSt (type 1) ghost membranes show a defect in a process of ATP-dependent endocytic vesiculation that can be corrected by the cross-linker dimethyl adipimidate (DMA), which also corrects the cation leak in these cells.⁴  OHSt (type 2) ghost membranes do not have the defect in ATP-dependent endocytic vesiculation, and the cation leak is not corrected by DMA.⁴ 

We were prompted to study the Rh-associated glycoprotein (RhAG) because we found that a related, less leaky condition known as cryohydrocytosis was caused by mutations in SLC4A1, which codes for the band 3 anion exchanger.⁵  Band 3 is the focus of a macrocomplex of proteins that includes RhAG, and RhAG can be coimmunoprecipitated with anti–band 3 antibodies and thus associates directly with band 3.⁶  Band 3 is involved in efficient red cell CO₂ transport, and the macrocomplex probably forms an integrated gas exchange metabolon⁶ : however, the role of RhAG within the complex is not known. The function of RhAG is disputed, but it has been proposed that it may act as a gas channel or ammonium transporter because the Rh proteins form part of the ammonia transport (Amt) protein family. The crystallographic structures have been solved for certain Amt-like proteins, giving insight into the mechanism of ammonia transport. The first to be solved was for AmtB,^7,8  and more recently for the Rh50 protein from Nitrosomonas europaea (NeRh50),^9,10  which is a closer homolog of human RhAG (RhAG shares 20% sequence identity with AmtB and 36% with NeRh50). The structures of AmtB and NeRh50 are similar, with a root mean squared deviation of 1.4 Å over 268 aligned Cα atoms.⁹  Both form trimers, with a hydrophobic pore through the center of each monomer. An unusual conserved twin-histidine motif protrudes into the center of the pore, which is otherwise entirely lined by hydrophobic residues. Entry to the pore from the extracellular vestibule in both AmtB and NeRh50 is restricted by the side chains of 2 conserved phenylalanine residues (the “phe-gate”).

We have studied 7 OHSt kindreds (Table 1) and identified 2 novel heterozygous mutations (t182g, t194c) in RHAG leading to substitutions of 2 highly conserved amino acids (Ile61Arg, Phe65Ser). Expression in Xenopus laevis oocytes confirmed that the mutant RhAG proteins induced a large monovalent cation leak, and modeling of RhAG based on the homologous N europaea Rh50 protein structure suggested that these mutations lead to a more open pore structure.

All patients had classical OHSt with very leaky red cells (Na⁺ and K⁺ transport rates of about 40 times that of normal at 37°C) resulting in moderate hemolytic anemia.²  Mean red cell volumes were abnormally large (approximately 140 fL), and blood smears showed the presence of stomatocytes, echinocytes, target cells, macrocytes, and (if splenectomized) punctate red cells. Osmotic ektacytometry displayed a shifted curve typical in OHSt (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Patients are designated by their place of origin and their position in the family (Table 1). Some pedigrees have been reported previously.^1,11-14  The male patient “OHSt Nancy,” who was born in 1987, is described in Figure S1. Patient blood samples were collected and informed consent was obtained in accordance with the Declaration of Helsinki. The red cell protein and DNA analysis are diagnostic, both extending and refining the diagnosis of the disease phenotype.

Plasma was tested for antibody specificity using standard serologic indirect antiglobulin test (IAT) tests using both tube and gel techniques and untreated and papain-treated cells. Red cells were tested for selected high-incidence antigens.

Preparation of red cell membranes, SDS-PAGE, Coomassie blue staining, and Western blotting of membrane proteins were carried out as previously described.⁶  Protein concentration was estimated using the Bradford assay and equal amounts (typically 10 μg) of ghosts loaded per track of each gel. The rabbit antipeptide antibodies anti-RhAG, anti-Rh, anti-CD47, and anti-stomatin were used as previously described.⁶ 

Genomic DNA was isolated from blood samples. The coding regions of exons 1 to 10 of human RHAG^15,16  were analyzed for single-strand conformation polymorphisms (SSCPs) using exon-specific primers. All altered or anomalous banding patterns were further investigated by DNA sequencing as described previously.⁵  DNA sequencing was used to analyze exon 2 of RHAG from all available family members and 56 unrelated controls.

The cDNA clones in X laevis expression vector BSXG1 encoding human RhAG (BSXG1.RHAG) and glycophorin B (BSXG.GPB) have been described.¹⁷  They contain the cDNAs encoding human intact red cell RhAG and GPB, respectively, flanked by the 5′ and 3′ noncoding regions of Xenopus β-globin. Amino acid substitutions Phe65Ser and Ile61Arg were made using the Quikchange II mutagenesis kit (Strategene, Amsterdam Zuidoost, The Netherlands) using BSXG1.RHAG as a template and the following primers: I61R sense ccaagatgtacatgttatgagatttgttgggtttggcttcc, I61R antisense ggaagccaaacccaacaaatctcataacatgtacatcttgg, F65S sense agatgtacatgttatgatatttgttgggtctggcttcctcatgac, and F65S antisense gtcatgaggaagccagacccaacaaatatcataacatgtacatct.

The sequences of the final constructs were confirmed by automated sequencing (ABI PRISM 3100 Genetic Analyzer Automatic Sequencer; Applied Biosystems, Courtaboeuf, France). The methods used for the preparation of cRNA, expression in oocytes, and measurement of Na⁺ and K⁺ content by flame photometry have been described.⁵  Glycophorin B is thought to increase the expression of RhAG at the plasma membrane,¹⁸  and was coinjected (2.5 ng) with the wild-type (10 ng) or mutant (10 ng) RhAG. After injection, oocytes were kept at 19°C in modified Barth saline (MBS) composed of 85 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 10 mM HEPES, and 4.5 mM NaOH (pH 7.4) supplemented with penicillin (10 U/mL) and streptomycin (10 μg/mL). Oocytes dedicated to Na⁺ and K⁺ content measurements were incubated following injection at 19°C for 2 or 3 days in MBS with ouabain (0.5 mM) and bumetanide (5 μM) to prevent any Na⁺ or K⁺ recycling or movement through the Na⁺/K⁺ pump or Na⁺-K⁺-2Cl⁻ cotransporter. After incubation 5 oocytes were quickly rinsed twice in milliQH₂O and dried overnight at 80°C. Each condition and experiment was done in triplicate. Intracellular cations were extracted from dried oocytes by overnight incubation in 4 mL milliQH₂O. Na⁺ and K⁺ were quantified by flame photometry (Eppendorf, Le Pecq, France).

Li⁺ was used as a substitute for Na⁺ to measure oocyte cation permeability. NO₃⁻ containing medium was used to avoid any involvement of cation movements through the KCl cotransporter (Li⁺ influx was not significantly different when measured in MBS where NaCl was substituted with LiNO₃ or with LiCl). Oocytes (7 per condition) were incubated for 2 hours at 19°C in MBS, where NaCl was substituted by LiNO₃. Composition was as follows: 85 mM LiNO₃, 1 mM KNO₃, 2.4 mM KHCO₃, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 10 mM HEPES, and 4.5 mM NaOH (pH 7.4). In addition, ouabain (0.5 mM) and bumetanide (5 μM) were added to block the Na⁺/K⁺ pump and the Na⁺-K⁺-2Cl⁻ cotransporter, respectively. Incubation for 2 hours was within the linear phase of Li influx. Oocytes were washed 3 times in milliQH₂O and placed one by one in a tube heated at 95°C to desiccate. Intracellular Li⁺ was extracted by addition of 50 μL 0.1 N NaOH and further diluted by addition of 250 μL milliQH₂O. The Li⁺ content in each oocyte extract was measured by atomic absorption spectrometry with a PerkinElmer 3110 (PerkinElmer SAS, Courtaboeuf, France).

Oocyte membranes were prepared as described previously,¹⁹  resuspended in 20 mM Tris HCl buffer (pH 7.4), 250 mM sucrose, and 8 M urea with protease inhibitor Pefabloc (Roche, Meylan, France) and heated for 15 minutes at 37°C. The membranes were mixed with loading buffer containing DTT and run on an 8% SDS-PAGE polyacrylamide gel. Western blots were incubated with a rabbit antipeptide RhAG antibody (as in “Red cell membrane protein in analysis”, 1:1000) in Tris-buffered saline, 0.1% Tween-20 with 1% fat-free dry milk powder (wt/vol). The secondary antibody was a goat IgG-peroxydase (Sigma, Lyons, France) used at 1:80 000 dilution. The signal was visualized by enhanced chemiluminescence (PerkinElmer SAS) and Kodak Biomax film (Sigma).

The crystallographically determined structure of NeRh50 (Protein Data Bank [PDB] ID 3B9W)⁹ ) was used as a template to build homology models of RhAG and the Rh polypeptides using the MODELLER software.²⁰  Models were optimized using a thorough variable target function optimization and 300 cycles of simulated annealing molecular dynamics. The process was repeated to generate 20 models for each of the wild-type and OHSt mutant RhAG proteins. For each protein, the model with the lowest objective function was selected for further analysis. The quality of each model was assessed with PROCHECK²¹  and MOLPROBITY.²²  For all the models, no less than 95% of residues lay in the most favored regions of the Ramachandran plot, and all other stereochemical parameters were within expected ranges.

SDS-PAGE and Coomassie staining of OHSt red cell membranes (Table 1; Stockport A-II-1) revealed the reduction in stomatin, as described previously,^1,11  but no other obvious changes in the major membrane proteins (Figure 1). However, immunoblotting analysis showed the RhAG protein from OHSt red cells was expressed at slightly lower levels and displayed altered electrophoretic mobility (Figure 1). The Rh polypeptides were likewise reduced, reflecting their dependence on RhAG for stable expression in the red cell membrane; CD47, however, was not noticeably altered (Figure 1).

These alterations in the RhAG protein prompted us to examine the RHAG gene in patients with OHSt. SSCP analysis showed 2 different band shift patterns in exon 2, and DNA sequencing identified 2 heterozygous mutations (t182g, t194c) in RHAG leading to substitutions of amino acids Ile61Arg and Phe65Ser, respectively. The Phe65Ser mutation was associated with OHSt in 6 kindreds (OHSt type 1) and Ile61Arg in 1 (OHSt type 2; Table 1; Figure 2). To check that these novel mutations were not simply single-nucleotide polymorphisms, we sequenced the DNA from 56 unrelated control DNAs (112 alleles), all coded for the normal RhAG protein. Amino acids 61 and 65 are within a highly conserved region of the RhAG protein. Ile61 is conserved and Phe65 is identical across 37 RhAG or RhAG-like protein sequences²⁶  (Figure 3) and across 39 RhBG or RhCG protein sequences, the nonerythroid homologs of RhAG.²⁶ 

To investigate the cation transport properties of the mutant RhAG proteins when expressed in a heterologous system, we injected X laevis oocytes with cRNA coding for wild-type RhAG, Ser65-RhAG, or Arg61-RhAG. Both wild-type and mutant RhAG proteins were substantially expressed at the oocyte membrane (Figure 4A). We measured both Li⁺ uptake (Figure 4B) and the levels of intracellular Na⁺ and K⁺ after 2 (Figure 4C) and 3 (Figure 4D) days. In the presence of a monovalent cation leak, internal [Na⁺] will rise and [K⁺] will fall.⁵  Unexpectedly, expression of wild-type RhAG led to an increase in Li⁺ uptake, a rise in intracellular [Na⁺], and a fall in intracellular [K⁺], such that after 3 days, intracellular [Na⁺] exceeded intracellular [K⁺] (Figure 4D). When the mutants Ser65-RhAG or Arg61-RhAG were expressed, Li⁺ uptake was about 6 times that seen with wild-type, and the intracellular Na⁺ and K⁺ levels were very markedly abnormal (Figure 4B,C); after 3 days of incubation, intracellular [Na⁺] hugely exceeded intracellular [K⁺], which was barely detectable (Figure 4D). These changes exactly mimic those seen on OHSt red cells,²  and strongly support the contention that it is the mutated RhAG that mediates the cation leak in these cells.

We used the NeRh50 structure as a template to construct models of the wild-type and mutant RhAG proteins (Figure 5; Document S1 and Figure S2 show a more detailed description of the RhAG model). Our model of wild-type human RhAG shows no major structural changes compared with NeRh50. In the central pore region, both the twin-histidine motif (His175 and His334 in RhAG) and the phe-gate (Phe120 and Phe225 in RhAG) are conserved (Figure 5A-C,F). At the cytoplasmic end of the pore there is a constriction formed by His334 of the conserved pair and by Phe65. Mutation of Phe65 to serine, as found in the patients with type 1 OHSt, alters this constriction, opening the pore from the cytoplasmic surface to the phe-gate (Figure 5D,G). The modeling predicts a minimum pore diameter of approximately 3.6 Å for Ser65-RhAG, which is larger than the hydrated radii of Na⁺ (3.58 Å), K⁺ (3.31 Å), and NH₄⁺ (3.31 Å).²⁷  Our model of the Ile61Arg mutant of RhAG predicts small, local, structural changes, which again result in the opening up of the central pore region (Figure 5E,H). In this case, the minimum pore diameter observed in the model is 1.5 Å. However, incorporation of the positively charged guanidinium group from the arginine into a hydrophobic region (Figure 5E,H) may lead to more pronounced structural changes. Both models predict that the OHSt mutations will result in the phe-gate being the major barrier to the passage of substrate through the central pore. Opening of the phe-gate appears to be a prerequisite for the passage of substrate through all AmtB and Rh50 proteins.^7,9  The phe-gate could be opened by a simple rotation of the side-chain of the C-terminal phenylalanine without any major structural rearrangement (Document S1).

One patient (Table 1; Stockport A-II-2) had received transfusions multiple times, and it became increasingly difficult to find compatible blood. Anti-c was identified in the plasma and an additional, weaker antibody that reacted with all normal red cell samples. The patient's red cells were positive with a range of antibodies to high-incidence antigens. The only cells that were compatible with the plasma were Rh_null cells (3 examples) and those of the patient's affected brother (Table 1; Stockport A-II-1). D- - cells were weakly reactive. Compatibility with the cells of the patient's brother suggested that their red cells lacked the same epitope, although no antibody was detectable in his plasma. The compatibility with Rh_null cells, that lack not only RhAG but also the Rh polypeptides and CD47, suggested that the epitope was formed against one of the proteins of the Rh complex. Both the patient's and the brother's red cells had normal Rh antigens (R1R1); therefore, the Rh polypeptides were present. Immunoblotting showed normal amounts of CD47 (Figure 1). D- - cells contain reduced amounts of RhAG (J.F.F., L.J.B., unpublished results, July 2008), and thus these data suggested that this weak antibody was formed against an epitope on RhAG.

We have shown in this paper that mutations in RHAG cause OHSt, and that the large cation leak in OHSt red cells is almost certainly carried by the mutant RhAG proteins. Interestingly, we have also found that wild-type RhAG leaks monovalent cations. The physiologic function of RhAG is disputed. Its homology to the Amt proteins suggests that it may act as a cation channel transporting NH₄⁺, or as a gas channel moving NH₃ or CO₂. However, early studies of the transport properties of human RhAG gave conflicting results, variously describing RhAG as a NH₃/NH₄⁺ exchanger or ammonium transporter,^28-31  or as an ammonia channel,^32-34  or suggesting that RhAG transports CO₂,³⁵  or both CO₂ and NH₃.³⁶  Recently, even the gas channel hypothesis has been disputed.³⁷  This is the first study to show that wild-type RhAG can act as a pore for other monovalent cations (Na⁺, K⁺, Li⁺), at least when expressed in X laevis oocytes. Interestingly, a recent study has shown that AmtB allows accumulation of methylammonium cations in response to variation in the electrical potential across the bacterial membrane.³⁸  These authors also showed that the rate of flux of methylammonium through an AmtB mutant (Trp148Leu), predicted to enlarge the AmtB pore, was about 10 times faster than through wild-type AmtB.³⁸  Modifying their own previous conclusions, these authors inferred that AmtB can act as an ion channel.³⁸  Similarly, we find that RhAG can mediate a basal cation flux, the rate of which is increased by mutations that are predicted to open the pore structure.

Two observations suggested that the Phe65Ser mutation not only alters the pore structure but also causes long-range structural alterations in RhAG. The first involved a patient (Table 1; Stockport A-II-2) who had made an antibody against a RhAG epitope. Since in this heterozygous condition both the wild-type and Ser65-RhAG proteins should be present, the altered epitope was probably formed by changes in the oligomeric interactions of RhAG, and suggests that WT-RhAG forms hetero-oligomers with Ser65-RhAG. Another observation indicating long-range structural alterations was the inhibitory effect of DMA on these Phe65Ser RhAG cells,⁴  suggesting that the position or orientation of extracellular lysine residue(s) are altered in the Phe65Ser RhAG, creating a DMA binding site that is not found in wild-type or I61R RhAG.

OHSt cells are deficient in the raft protein stomatin, and stomatin expression is also reduced in a subtype of cryohydrocytosis.¹²  However, no mutation was found in either SLC4A1 or RHAG for this condition (Table 1). Both this subtype and OHSt have exceptionally leaky red cells, and we have previously suggested that the loss of stomatin in these cells may be linked to glycolytic activity.¹²  It has more recently been shown that stomatin may act as a molecular switch that, when present, converts the glucose transporter Glut1 (SLC2A1) into a transporter of L-dehydroascorbic acid, an oxidized form of ascorbic acid.³⁹  If stomatin is absent, Glut1 acts as a glucose transporter. Thus, the loss of stomatin from these cells may well be triggered by the need for increased glycolysis required to fuel the increased Na⁺K⁺-ATPase activity⁴⁰  demanded by the cation leak. The mechanism of this regulation has yet to be defined.

The data here show unequivocally that OHSt is caused by mutations in RHAG. We have shown that wild-type RhAG protein induces a cation flux when expressed in X laevis oocytes, while the mutants induce a much larger flux. Modeling studies indicate that both mutations dilate a cytoplasmic constriction in the RhAG pore. Together, these data suggest that wild-type RhAG has the properties of a cation pathway. It is possible that in its natural environment, as part of the band 3 macrocomplex, RhAG forms a regulated cation channel that is only active under certain conditions. This is the first report of mutant RhAG proteins with any measurable physiologic activity: all previous RHAG mutations have been recessive and have resulted in the Rh_null phenotype. Further study of the present mutants may help reveal the true function of RhAG.

We thank Peter Martin for DNA sequencing; Dr Thérèse Cynober for performing osmotic gradient ektacytometry; Dr Corinne Armari-Alla, Dr Alain Robert, and Prof Danièle Sommelet for providing patient blood samples; and Dr Stuart Winter and Dr Cecil Reid for permission to report their patients. We thank all of the patients for their cooperation.

This work was supported by the UK National Health Service R&D Directorate (L.J.B., J.F.F.). N.M.B. was supported by a studentship from the UK Medical Research Council. G.W.S. thanks Advocacy for Neuroacanthocytosis for generous funding.

Contribution: L.J.B. conceived the original hypothesis, performed research design and protein and DNA analysis, and prepared the manuscript; H.G., F.B., and N.G. performed expression studies in X laevis oocytes; N.M.B. and R.L.B. performed modeling studies and prepared the manuscript; J.P. performed serology; J.F.F. performed immunoblotting; and J.D. and G.W.S. provided patient samples and prepared the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lesley Bruce, Bristol Institute for Transfusion Sciences, NHS Blood and Transplant, North Bristol Park, Filton, Bristol, BS34 7QH, United Kingdom; e-mail: lesley.bruce@nhsbt.nhs.uk.