We have examined the mechanism by which glycophorin A (GPA) facilitates the movement of the human red-cell anion exchanger (band 3, AE1) to the cell surface. GPA itself forms stable dimers in membranes and detergent solution. Four mutants of human GPA with impaired dimerization were prepared (L75I, I76A, G79L and G83L). All four GPA mutants enhanced band 3 translocation to the Xenopus oocyte plasma membrane in the same way as wild-type GPA, showing that the GPA monomer is sufficient to mediate this process. Cell-surface expression of the natural band 3 mutant G701D has an absolute requirement for GPA. GPA monomers also rescued the cell-surface expression of this mutant band 3. Taking into account other evidence, we infer that the site of GPA interaction with band 3 is located outside the GPA dimerization interface but within the GPA transmembrane span. The results of examination of the cell-surface expression of GPA and band 3 in different K562 erythroleukaemia cell clones stably transfected with band 3 are consistent with the movement of GPA and band 3 to the cell surface together. We discuss the pathways by which band 3 moves to the cell surface in the presence and the absence of GPA, concluding that GPA has a role in enhancing the folding and maturation of band 3. We propose that GPA functions in erythroid cells to assist with the incorporation of large amounts of properly folded band 3 into the membrane within a limited time span during erythroid maturation.

We describe three mutations of the red-cell anion exchanger band 3 ( AE1 , SLC4A1 ) gene associated with distal renal tubular acidosis (dRTA) in families from Malaysia and Papua New Guinea: Gly 701 → Asp (G701D), Ala 858 → Asp (A858D) and deletion of Val 850 (δV850). The mutations A858D and ∆V850 are novel; all three mutations seem to be restricted to South-East Asian populations. South-East Asian ovalocytosis (SAO), resulting from the band 3 deletion of residues 400–408, occurred in many of the families but did not itself result in dRTA. Compound heterozygotes of each of the dRTA mutations with SAO all had dRTA, evidence of haemolytic anaemia and abnormal red-cell properties. The A858D mutation showed dominant inheritance and the recessive ∆V850 and G701D mutations showed a pseudo-dominant phenotype when the transport-inactive SAO allele was also present. Red-cell and Xenopus oocyte expression studies showed that the ∆V850 and A858D mutant proteins have greatly decreased anion transport when present as compound heterozygotes (∆V850/A858D, ∆V850/SAO or A858D/SAO). Red cells with A858D/SAO had only 3% of the SO 4 2- efflux of normal cells, the lowest anion transport activity so far reported for human red cells. The results suggest dRTA might arise by a different mechanism for each mutation. We confirm that the G701D mutant protein has an absolute requirement for glycophorin A for movement to the cell surface. We suggest that the dominant A858D mutant protein is possibly mis-targeted to an inappropriate plasma membrane domain in the renal tubular cell, and that the recessive ∆V850 mutation might give dRTA because of its decreased anion transport activity.

The red-cell anion exchanger (band 3; AE1) is a multispanning membrane protein that traverses the bilayer up to 14 times and is N-glycosylated at Asn-642. We have shown that the integrity of six different loops are not essential for stilbene disulphonate-sensitive chloride uptake in Xenopus oocytes. We used an N-glycosylation mutagenesis approach to examine the orientation of the N-terminus and the endogenous glycosylation site of each C-terminal fragment by cell-free translation. The fragments initiating in the loops preceding spans 2, 9 and 11 did not insert into the membrane with the expected orientation. Furthermore, N-glycosylation of Asn-642 might facilitate the membrane integration of span 7. The correct integration of spans 2-3 required the presence of the region containing span 4 and that the luminal exposure of the C-terminus of span 7 is increased in the presence of the region including span 6 or span 8. The results suggest the span 8 region is required for the correct folding of spans 9-10, at least in the presence of the span 11-12 region. Our results suggest that there are intramolecular interactions between the regions of transmembrane spans 1 and 2, 2 and 4, 4 and 5, 7 and 8, 8 and 9-10, and 9-10 and 11-12. Spans 1, 4, 5, 6 and 8 might act as a scaffold for the assembly of spans 2-3, 7 and 9-10. This approach might provide a general method for dissecting the interactions between membrane-spanning regions of polytopic membrane proteins.

We have examined the functional co-assembly of non-complementary pairs of N- and C-terminal polypeptide fragments of the anion transport domain (b3mem) of human red-cell band 3. cDNA clones encoding non-contiguous pairs of fragments with one transmembrane (TM) region omitted, or overlapping pairs of fragments with between one and ten TM regions duplicated, were co-expressed in Xenopus oocytes and a cell-free translation system. Stilbene disulphonate-sensitive chloride uptake assays in oocytes revealed that the omission of any single TM region of b3mem except spans 6 and 7 caused a complete loss of functional expression. In contrast, co-expressed pairs of fragments overlapping a single TM region 5, 6, 7, 8, 9-10 or 11-12 retained a high level of functionality, whereas fragments overlapping the clusters of TM regions 2-5, 4-5, 5-8 and 8-10 also mediated some stilbene disulphonate-sensitive uptake. The co-assembly of N- or C-terminal fragments with intact band 3, b3mem or other fragments was examined by co-immunoprecipitation in non-denaturing detergent solutions by using monoclonal antibodies against the termini of b3mem. All the fragments, except for TM spans 13-14, co-immunoprecipitated with b3mem. The medium-sized N-terminal fragments comprising spans 1-6, 1-7 or 1-8 co-immunoprecipitated particularly strongly with the C-terminal fragments containing spans 8-14 or 9-14. The fragments comprising spans 1-4 or 1-12 co-immunoprecipitated less extensively than the other N-terminal fragments with either b3mem or C-terminal fragments. There is sufficient flexibility in the structure of b3mem to allow the inclusion of at least one duplicated TM span without a loss of function. We propose a working model for the organization of TM spans of dimeric band 3 based on current evidence.

We examined the assembly of the membrane domain of the human red cell anion transporter (band 3; AE1) by co-expression of recombinant N- and C-terminal fragments in Xenopus oocytes and in cell-free translation with canine pancreatic microsomes. Co-immunoprecipitation was performed in non-denaturing detergent solutions using antibodies directed against the N- and C-termini of the membrane domain. Eleven of the twelve fragments were expressed stably in oocytes in the presence or absence of their respective partners. However, the fragment containing from putative span nine to the C-terminus could be detected in oocytes only when co-expressed with its complementary partner containing the first eight spans. Co-expression of pairs of fragments divided in the first, second, third and fourth exofacial loops and in the fourth cytoplasmic loop resulted in a concentration-dependent association, but a pair of fragments divided in the sixth cytoplasmic loop did not co-immunoprecipitate. When two complementary fragments were translated separately in the cell-free system and the purified microsomes were then mixed, co-immunoprecipitation was observed only if the membranes were first fused using polyethylene glycol. This shows that co-immunoprecipitation results from specific interactions within the membrane and is not an artefact of detergent solubilization or immunoprecipitation. We demonstrate that band 3 assembly can occur within the membrane after translation, insertion and initial folding of the individual fragments have been completed. We conclude that most band 3 fragments contain the necessary information to fold in the membrane and adopt a structure that is sufficiently similar to the native protein that it permits correct assembly with its complementary partner.

1. A new human erythrocyte variant (Ph) is described. The variant contains an unusual sialic acid-rich glycoprotein in addition to the blood-group-MN([unk])- and blood-group-Ss(δ)-active sialoglycoproteins found in normal erythrocytes. 2. The unusual component Ph has an apparent mol.wt. of 32000 on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis. The Ph component is not degraded during trypsin treatment of intact erythrocytes. 3. The Ph component was labelled by lacto-peroxidase-mediated radioiodination of intact erythrocytes and was found to be present in amounts approximately equimolar to α-sialoglycoprotein in the variant erythrocytes. 4. The Ph component had receptors for the lectins from Maclura aurantiaca (osage orange) and Triticum vulgaris (wheat-germ), but lacked a receptor for the Phaseolus vulgaris (red kidney bean) lectin, suggesting that it carries only O -linked oligosaccharides. 5. The presence of the Ph component in these erythrocytes does not correspond to any of the known blood-group-MNSs-related antigens examined. 6. We suggest that this component may be a hybrid polypeptide containing the N -terminal portion derived from normal δ-sialoglycoprotein, and the C -terminal portion from normal α-sialoglycoprotein, in a manner similar to the anti-Lepore haemoglobin.