Distinct Regions of Human Glycophorin A Enhance Human Red Cell Anion Exchanger (Band 3; AE1) Transport Function and Surface Trafficking*

Human red cell glycophorin A (GPA) enhances the expression of band 3 anion transport activity at the cell surface of Xenopus oocytes. This effect of GPA could occur in two ways, enhancement of band 3 anion transport function or enhancement of band 3 trafficking to the cell surface. We have examined the GPA effect using GPA mutants. We compared the sequences of GPA and its homolog glycophorin B (GPB; which does not facilitate band 3 cell-surface activity or trafficking) to identify candidate regions of GPA for study. We constructed several GPA or GPB mutants, including naturally occurring GPA/GPB hybrid molecules and insertion, deletion, and substitution mutants. We analyzed the effects of the mutant proteins on band 3-specific chloride transport and surface presentation using co-expression in Xenopus oocytes. We find that the C-terminal cytoplasmic tail of GPA enhances trafficking of band 3 to the cell surface, whereas the extracellular residues 68–70 increase the specific anion transport activity of band 3. In addition, examination of the oligomerization of GPA mutants showed that single amino acid substitutions N-terminal to the transmembrane domain greatly reduce SDS-stable GPA dimer formation, implying that regions outside the transmembrane domain of GPA are important for GPA dimer formation.

tial evidence for interactions between GPA and band 3 in the red cell membrane. Anti-GPA antibodies decrease the rotational mobility of band 3 in red cell membranes (9 -13). Anti-Wr b antibodies require the presence of both GPA and band 3 for their reactivity with red cells and co-immunoprecipitate both GPA and band 3 (10). GPA facilitates the movement of band 3 to the cell surface in Xenopus oocytes. Co-expression of cRNAs corresponding to wild-type band 3 and GPA induces an increased level of band 3-mediated chloride influx in the oocytes by facilitating the trafficking of band 3 to the oocyte surface and potentially also by enhancing the anion transport function of band 3 (14,15). The enhancing effect of GPA on band 3 transport activity at the oocyte surface is independent of the oligomeric state of GPA; GPA mutants that are unable to form stable dimers act upon band 3 in the same fashion as wild-type GPA (16).
Other evidence also suggests an interaction between band 3 and GPA at an early stage in the biosynthesis and intracellular processing of the two proteins (1,9,17). The N-glycan chain of band 3 is increased in size in red cells that lack GPA (18 -20) but is decreased in size in red cells that effectively have more GPA than normal (21). The sulfate transport activity of band 3 is diminished in GPA-deficient human red cells, although the amount of band 3 remains unchanged (20). The red cells of band 3 gene knock-out mice do not express either band 3 or GPA, although the erythroid precursors contain GPA mRNA, indicating that band 3 is also involved in the movement of GPA to the cell surface (22). In contrast to these interactions between GPA and band 3, there is no evidence that the closely related red cell glycophorin B (GPB) interacts directly with band 3. Co-expression of GPB with band 3 in Xenopus oocytes has no effect on band 3 anion transport activity (14).
In the present study we have examined the role of different regions of the GPA molecule in its interaction with band 3 by performing an extensive mutational analysis of the GPA molecule. Detailed examination of the sequences of GPA and GPB, a molecule with a close sequence relationship to GPA that does not enhance band 3-mediated chloride transport (14), allowed identification of target regions for mutagenesis. We examined the effects of the GPA mutations on both band 3 function in Xenopus oocytes and GPA dimerization. We find that the cytoplasmic C-terminal tail of GPA is responsible for enhancing trafficking of band 3 to the cell surface, whereas the extracellular residues 68 -70 of GPA (N-terminal to the transmembrane domain) have a role in increasing the specific activity of band 3 for anion transport. Our data demonstrate that the presence of GPA results in the enhancement of band 3 anion transport function as well as band 3 trafficking to the cell surface. In addition, we find that single amino acid substitutions N-terminal to the transmembrane domain of GPA greatly reduce SDS-stable GPA dimer formation, implying that regions outside the transmembrane domain of GPA are important for GPA dimer formation.

EXPERIMENTAL PROCEDURES
Construction of GPA Mutants-The constructs pBSXG1.B3, pBSXG. GPA, and pBSXG.GPB have been described previously (14). They contain the cDNAs encoding human intact red cell band 3, GPA, and GPB, respectively, flanked by the 5Ј-and 3Ј-non-coding regions of Xenopus ␤-globin. The three GPA/GPB hybrid molecules MiV (GPA-(1-58)-GPB (27-72)), Sta (GPB-(1-26)-GPA-(59 -131)), and Dantu (GPB-(1-39)-GPA-(72-131)) were prepared as follows. The respective regions of GPA and GPB present in each hybrid were amplified by PCR using a 5Ј or 3Ј BSXG-specific primer and a hybrid-specific primer to give two fragments with an overlap of 20 bases at the hybrid crossover position. The two fragments were mixed and allowed to anneal and extend in 10 cycles of PCR without primers (95°C for 30s, 60°C for 40s, 72°C for 1 min). 5Ј and 3Ј BSXG-specific primers were added, and 30 cycles of PCR (95°C for 30s, 52°C for 40s, 72°C for 1 min) were performed. The resulting PCR products were digested with BstXI and BstEII and ligated into BstXI-and BstEII-digested pBSXG vector. The double substitution mutant GPA F78L/G79C was prepared in the same manner as the GPA/GPB hybrids except that both fragments were amplified from pBSXG.GPA before annealing. A six-amino acid insertion in GPA (insertion of VPAPVV between Glu-72 and Ile-73; BintoA) was constructed by amplification of the entire pBSXG.GPA plasmid using primers to generate a linear fragment with the inserted bases and NarI sites at the 5Ј and 3Ј ends. The PCR product was restricted using NarI and ligated to generate the insertion mutant. Two C-terminal deletion mutants of GPA were also used. GPA-(1-97) has been described previously (CtDelA (14)). GPA-(1-101) was constructed by PCR amplification of the GPAcoding region with a 3Ј primer-containing sequence annealing to the region of GPA from Arg-97 to Lys-101 including a stop codon and a BstEII site. The PCR product was restricted with BstXI and BstEII and ligated into BstXI-and BstEII-digested pBSXG vector. The GPA mutants H66Q, H67Q, F68C, S69C, E70C, P71C, E72C, and R97M were prepared using the Seamless Cloning kit (Stratagene) (31,32). The Seamless Cloning kit was also used to generate the deletion constructs GPA⌬Phe-68 -Glu-70, GPA⌬Ala-65-Glu-70, GPA⌬Arg-61-Glu-70, and GPB⌬Pro-41-Val-43.
In Vitro Transcription, Cell-free Translation, and Expression in Xenopus Oocytes-Transcription of cRNA in vitro was performed with the mMessage mMachine kit (Ambion). The methods used for cell-free translation in the rabbit reticulocyte system with canine pancreatic microsomes, purification of microsomes, isolation of Xenopus oocytes, injection with cRNA, Cl Ϫ transport assay, and chymotrypsin treatment of oocytes have been described previously (14,15,33,34). Where possible chloride transport studies were carried out in parallel on the same batch of oocytes as those used for the cell surface expression assays using chymotrypsin, but different amounts of the cRNAs were injected for the transport and chymotrypsin assays as indicated in the legends to Figs. 2 and 3. For the oocyte chymotrypsin assay, samples were separated by SDS-PAGE (35) and subjected to Western blotting using the ECL method as described in the figure legends. The chymotrypsin assays were done in parallel on matched groups of oocytes from the same batch (injected with band 3 cRNA alone, band 3 plus normal GPA cRNAs, band 3 plus GPA mutant cRNAs and uninjected oocytes). The quantitative data shown in Table I were obtained by scanning films exposed so that the band density was within the linear range of the film. Band 3 Interactions- Fig. 1A shows a "bubble diagram" representation of the human GPA molecule in the membrane, with the C terminus at the cytosolic face. Fig. 1B shows a sequence alignment of GPA and GPB, its close homolog. Because GPB does not enhance band 3-mediated anion transport in oocytes (14), this alignment was used to examine regions of sequence difference that could be important for the action of GPA on band 3. Analysis of the two sequences indicates four main differences. First, GPA has a 32-amino acid insertion within its extracellular N terminus (amino acids 27-58). Second, there is considerable sequence difference in the extracellular region N-terminal to the transmembrane domain (Arg-61-Glu-72 in GPA; Thr-29 -Val-43 in GPB). GPB has a 3-amino acid insertion (Pro-41-Val-43) N-terminal to the transmembrane domain. Third, there are three sequence differences within the transmembrane spans of GPA and GPB (F78L, G79C, and V84I from GPA to GPB). Fourth, GPA has an additional 30-amino acid C terminus (amino acids 102-131).

Analysis of GPA and GPB Sequences for Regions Involved in
Many naturally occurring hybrids of GPA and GPB in human red cells have been reported (36), and some of these molecules were also used in analysis of regions of interest on the GPA molecule.

Effect of GPA and GPB Mutants on Band 3-specific Anion Transport and Band 3 Surface Presentation in Xenopus
Oocytes-In vitro transcribed mRNA was co-injected into Xenopus oocytes with mRNA for human erythrocyte band 3 and either wild-type GPA, wild-type GPB, GPA mutant, or GPB mutant mRNA, and the band 3-specific anion transport was estimated from the 4,4Ј-dinitrostilbene-2,2Ј-disulfonate (DNDS)-sensitive chloride uptake induced in the oocytes (Fig. 2). Chloride transport of all the mutants was measured in individual oocytes using a total of ϳ15 oocytes per experimental sample and in duplicate sets of experiments using different batches of oocytes. In each experiment control samples comprising oocytes injected with band 3 mRNA alone and oocytes co-injected with both band 3 and wild-type GPA mRNAs were analyzed to allow normalization of data sets. The graphs in Fig. 2 show the percentage enhancement of band 3-mediated chloride transport induced in the oocytes by co-injection of band 3 and the GPA mutant mRNAs (over basal band 3 chloride transport activity induced in the oocytes by band 3 mRNA alone) with transport due to band 3 mRNA alone set at 0% and transport due to band 3 mRNA co-injected with wild-type GPA mRNA set at 100%. Data from the transport assays are shown in Table I. Because GPA is known to increase the amount of band 3 at the oocyte surface, we also measured the proportion of band 3 protein that moved to the oocyte surface using a protease accessibility assay (33). Band 3 has a single extracellular site susceptible to extracellular chymotrypsin cleavage, which is located between transmembrane spans 5 and 6. Intact oocytes expressing band 3 and glycophorin mutants were treated with chymotrypsin, and the 35-kDa C-terminal band 3 fragment generated by extracellular cleavage of band 3 at the oocyte surface was separated from the uncleaved band 3 located within the cells using SDS-PAGE. Previous studies of the time course of chymotrypsin cleavage show that extracellular band 3 cleavage is complete under the conditions used in these experiments whether GPA is present or absent (14). Immunoblotting with a monoclonal antibody was used to detect and estimate the relative proportions of the 35-kDa fragment (derived from surface band 3) and intact band 3 (internal band 3) in the oocytes (Fig. 3). Films exposed so that the band density was in the linear range of the film were scanned, and the quantitative data are shown in Table I (the images shown in Fig. 3 are from FIG. 2. Chloride transport of band 3 co-expressed with glycophorin constructs. Xenopus oocytes were injected with 0.5 ng of cRNA/oocyte band 3 with or without 0.15 ng of cRNA/oocyte of GPA, GPB, or glycophorin mutant and allowed to express protein for 24 h at 18°C. 36 Cl Ϫ influx over a 1-h period was measured with groups of 12-15 oocytes in ND96 buffer (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 ⅐2H 2 O, 1 mM MgCl 2 ⅐6H 2 O, 5 mM HEPES, pH 7.5) in the same fashion as described previously (14). The bars represent the glycophorin mutantspecific percentage enhancement of band 3-specific chloride influx. The error bars show the S.E. of the influx measurements. For oocytes expressing band 3 alone, this percentage is set to 0, and for oocytes expressing band 3 plus GPA this percentage is set to 100. The controls for band 3 and band 3 plus GPA are shown in each panel. Panel A, chloride influx enhancement due to GPB, GPA/GPB, and GPB/GPA hybrids, GPB deletion GPB⌬Pro-41-Val-43, GPA C-terminal deletion mutants, and GPA deletion series N-terminal to the transmembrane domain. Panel B, chloride influx enhancement due to F78L/G79C (FGLC), BintoA, and GPA single substitution mutants. film exposed for optimum reproduction rather than linearity of band density).
The percentage increase of surface band 3 protein induced by each mutant GPA relative to that induced by wild-type GPA was calculated and is shown in Table I. The values shown in Table I represent the increase in band 3 protein expressed at the cell surface after subtraction of basal band 3 surface expression (when no GPA is present). Each lane from the blots in Fig. 3 represents approximately half the protein from 10 oocytes of the same batch, and where possible the protease accessibility assays for band 3 co-expressed with each glycophorin mutant were repeated at least once on a different batch of oocytes to allow for batch variability. It should be noted that although each transport experiment and the corresponding protease accessibility assays were carried out in parallel on the same batches of oocytes, 10-fold higher concentrations of cRNAs and a longer expression time in the oocytes were employed for the protease accessibility assays compared with the transport assays to enable the expressed band 3 to be reliably detected by immunoblotting. As previously reported (14,15), under the conditions used for the chloride uptake transport assay it is necessary to inject small amounts of band 3 cRNA. This ensures that the basal level of chloride uptake resulting from the expression of band 3 alone is low enough to maximize the enhancement in chloride uptake resulting from co-expression of GPA that is observed in the assay. However, because of the different cRNA levels and expression times employed for the transport and protease accessibility assays, there is a qualitative rather than a precise quantitative correlation of the effects of GPA and the GPA mutants in the two assays.
Co-expression with GPA increased the amount of band 3 protein at the cell surface (38% compared with 9% for band 3 alone; Fig. 3A and Table I legend) as has been described previously (14,15). Co-expression of band 3 with GPB had little effect on either the band 3-specific anion transport or the surface presentation of band 3 protein, as has been previously co-expressed with GPA mutants Values for chloride transport are expressed as percentage enhancement of the basal level of band 3 chloride influx (% enhancement of chloride transport), with transport due to band 3 alone set at 0% and transport due to band 3 co-expressed with wild-type GPA set at 100%. Values for surface presentation of band 3 co-expressed with GPA mutants (% enhancement of surface band 3) are expressed as percentage enhancement of the basal amount of band 3 at the oocyte surface, with surface presentation of band 3 alone (9 Ϯ 8% from 10 experiments) set at 0% and surface presentation of band 3 co-expressed with wild-type GPA (  Oocytes were co-injected with 5 ng of cRNA/oocyte full-length band 3 with 1.5 ng of cRNA/oocyte of GPA, GPB, or glycophorin mutant and allowed to express protein for 48 h at 18°C. Groups of 30 oocytes were treated with chymotrypsin for 1 h at 4°C as described previously (14). Groups of 10 oocytes were homogenized with 30 l of radioimmune precipitation assay buffer (10 mM Tris-HCl, pH7.2, 150 mM NaCl, 1% (w/v) Triton X-100, 0.1% (w/v) SDS) and centrifuged twice at 13000 rpm for 5 min to remove cell debris and yolk granules. The homogenate was separated on 12% polyacrylamide gels according to the method of Laemmli (35). Band 3 and band 3 fragments were visualized by Western blotting using the monoclonal antibody BRIC 155 (against the C terminus of band 3 (B3)). Each lane represents approximately half the protein from 10 oocytes. The proportion of band 3 that was cleaved by chymotrypsin at the surface of the oocytes to yield the 35-kDa Cterminal portion detected by BRIC 155 was determined using scanning densitometry and is shown in Table I  reported (14). The GPA/GPB hybrid MiV had a slightly inhibitory effect on band 3-specific anion transport and no effect on band 3 surface expression, but both the GPB/GPA hybrids (Dantu and Sta) enhanced both band 3-specific anion transport and the surface presentation of band 3 protein in a parallel fashion ( Figs. 2A and 3A; Table I). The Dantu hybrid appeared to be particularly effective in enhancing band 3 surface movement, whereas the Sta hybrid was slightly less effective than normal GPA. The GPB deletion mutant GPB⌬Pro-41-Val-43 behaved in the same fashion as GPB ( Figs. 2A and 3B; Table I). Double substitution of Phe-78 and Gly-79 in the GPA transmembrane span (F78L/G79C) did not alter the effect of GPA on band 3 (Fig. 2B and 3A; Table I). Both insertion of six residues N-terminal to the GPA transmembrane span (BintoA) and deletion of 3, 6, or 10 residues within the region of GPA Nterminal to the transmembrane domain (GPA⌬Phe-68 -Glu-70, GPA⌬Ala-65-Glu-70, and GPA⌬Arg-61-Glu-70) gave rise to constructs produced either no enhancement or little enhancement of band 3-specific chloride transport in the oocytes (Ϫ3, 20, 18 and 19%, respectively Fig. 2, Table I). However, these constructs retained the ability to enhance the cell surface trafficking of band 3 protein at least as efficiently as wild-type GPA (Fig. 3, A and B, Table I). Although the enhancement of chloride transport and increase in cell surface band 3 are not quantitatively comparable, the increased cell surface band 3 induced by these GPA mutants might be expected to have resulted in a similar large increase in chloride transport, as is observed with the other GPA mutants. This suggests that alterations in the Phe-68 -Glu-70 region of GPA may result in inhibition of the chloride transport activity of band 3.
Deletion of the C-terminal portion of GPA (GPA-(1-97) and GPA-(1-101)) resulted in impaired enhancement of band 3-specific anion transport (25 and 51%, respectively; Fig. 2A, Table  I). However, these constructs were also unable to enhance band 3 protein movement to the cell surface (Ϫ11 and Ϫ33% for-GPA (1-97) and 5% for GPA-(1-101); Fig. 3B, Table I). The point mutation GPA R97M (adjacent to the cytoplasmic end of the transmembrane span) had impaired ability to move band 3 to the cell surface but retained normal GPA enhancement of band 3-specific anion transport. These results imply that enhancement of band 3 anion transport function and enhancement of band 3 protein movement to the cell surface are two distinct effects that are mediated by different portions of the GPA molecule. Finally, most of the single substitution mutants in residues 66 -72 in the extracellular region GPA were able to enhance both band 3-specific chloride transport and band 3 protein movement to the cell surface (Figs. 2B and 3C; Table I).
Cell-free Translation of GPA Mutants-All the GPA and GPB mutants were expressed as full-length constructs using cellfree translation into canine pancreatic microsomes to confirm their ability to insert into endoplasmic reticulum membranes and to analyze their propensity to form SDS-stable dimers. SDS-stable dimer formation is not an absolute indicator of whether or not GPA mutants may form dimers in cell membranes; however, in previous analyses of GPA mutants, dimer formation in bacterial membranes and in SDS has usually given good agreement (23)(24)(25)27). After cell-free translation, the microsomes were purified and analyzed by SDS-PAGE (Fig.  4). All the GPA mutants were recovered in the microsomes, and those mutants that contained extracellular N-glycan addition sites were N-glycosylated as efficiently as wild-type GPA, demonstrating that they could integrate into the endoplasmic reticulum in the correct orientation and as effectively as wildtype GPA. As previously reported (16), GPA showed a distinct pattern of bands on SDS-PAGE (Fig. 4A, first lane). Two monomer bands were observed, the lower band, representing non-glycosylated monomer, and the upper band, representing the more abundant N-glycosylated monomer. Three SDS-stable dimer bands were visible, representing non-glycosylated dimer, mono-glycosylated dimer, and di-glycosylated dimer. The top band (di-glycosylated dimer) predominated. GPB is known to form homodimers and GPB-GPA heterodimers in red cells (9). However, only one band corresponding to the GPB monomer was observed in our cell-free translation experiments, (Fig. 4B,  seventh lane). This indicates that GPB expressed in cell-free translation could not form SDS-stable dimmers under the conditions of the experiment. A single monomer band was observed for GPB because the molecule does not contain an Nglycosylation site. The GPA/GPB hybrid MiV showed two bands corresponding to non-glycosylated and N-glycosylated monomer (Fig 4A, second lane) but no dimer bands. MiV did not form an SDS-stable dimer in cell-free translation because it FIG. 4. Cell-free translation of GPA, GPB, and glycophorin mutants. RNA corresponding to the glycophorin constructs indicated was translated in a cell-free system in the presence of canine pancreatic microsomes, as described previously (15). Purified microsomes were dissolved in SDS sample buffer containing 10% (v/v) 2-mercaptoethanol and separated by SDS-PAGE on 12% (w/v) gels (40). The positions of the non-glycosylated monomer (monomer), N-glycosylated monomer (glyc. mon.), and partially resolved mixture of glycosylated and non-glycosylated dimers (dimer) for full-length GPA constructs are indicated by arrows. Panel A, cell-free translation of GPA, the GPA/GPB and GPB/ GPA hybrids, F78L/G79C (FGLC), BintoA, and the C-terminal deletions. Panel B, cell-free translation of the three GPA substitution mutants H66Q, H67Q, and R97M, the deletion series at the N-terminal side of the GPA transmembrane span, and GPB. Panel C, cell-free translation of the GPA single cysteine substitutions and the GPB deletion GPB⌬Pro-41-Val-43. contains the GPB transmembrane span. The two GPB/GPA hybrids Sta and Dantu showed two bands corresponding to monomer and dimer (Fig. 4A, third and fourth lanes). These constructs lack the extracellular N-glycosylation addition site since they possess the GPB N terminus but contain the GPA transmembrane span and, thus, can form SDS-stable dimers. The double substitution mutant F78L/G79C showed partial SDS-stable dimer formation (Fig. 4A, fifth lane). The GPA insertion mutant BintoA, the two GPA C-terminal deletion mutants GPA-(1-97) and GPA- (1-101), and the three substitution mutants GPA H66Q, GPA H67Q, and GPA R97M all showed wild-type levels of dimer formation (Fig. 4, A and B). SDS-stable dimer formation was markedly reduced after 6 or 10 amino acids were deleted from the region of GPA N-terminal to the transmembrane span (GPA⌬Ala-65-Glu-70 and GPA⌬Arg-61-Glu-70; Fig. 4B, fifth and sixth lanes). This result was of interest as it clearly shows that regions outside the GPA transmembrane span may affect SDS-stable GPA dimer formation. Dimer formation may have been affected either by loss of important interacting residues or, for example, by the charged cluster of three glutamic acid sequences (Glu-55-Glu-57) in the N-terminal region of each GPA monomer being brought too close together in the dimer. Deletion of the three-amino acid insertion in GPB (GPB⌬Pro-41-Val-43; Fig 4C, eighth lane) did not give rise to any detectable SDS-stable dimer formation. The GPA single substitution mutants F68C, S69C, P71C, and E72C all exhibited substantially reduced SDS-stable dimer formation compared with wild-type GPA. It appears that single substitution of these residues for cysteine may affect SDS-stable dimer formation, implying that this region of the GPA molecule Nterminal to the transmembrane span is involved in dimer formation. In contrast GPA E70C formed a similar proportion of SDS-stable dimers as wild-type GPA (Fig. 4C). Because single cysteine mutants potentially could be capable of forming a disulfide bridge in a dimer, the samples were also separated by SDS-PAGE under non-reducing conditions, but no increase in dimer formation was observed (results not shown). DISCUSSION GPA enhances both the anion transport function and the movement of band 3 to the cell surface. We have carried out an extensive mutational study of GPA to understand which regions of the molecule are involved in the above functions by co-expression of GPA mutants with band 3 in Xenopus oocytes and analysis of the effects of the mutants on band 3 anion transport and surface presentation. GPA also forms an SDSstable dimer, and we have expressed GPA mutants using cellfree translation into canine pancreatic microsomes to establish that they insert efficiently into the endoplasmic reticulum with the correct orientation and examine their ability to form SDSstable dimers.

Different Regions of the GPA Molecule Have Different Effects on Band 3 Anion Transport Functional Activity and Band 3
Protein Movement to the Cell Surface-There are four main regions of sequence difference between GPA and GPB (see "Results"). Because GPB appears to have no significant effect on either band 3 anion transport function or surface expression, some or all of the four main regions of sequence difference in GPA must be involved in the effects of GPA on band 3.
The Cytoplasmic C Terminus of GPA Is Involved in the Enhancement of Band 3 Protein Movement on the Cell Surface-Deletion of the cytoplasmic C terminus of GPA led to the loss of ability to increase band 3 movement to the cell surface, but these mutants retained the ability to enhance band 3-specific chloride transport. The single point mutation GPA R97M (located on the cytoplasmic side of the TM domain) also exhibited an impaired ability to increase the surface presentation of band 3 but efficiently enhanced band 3 anion transport activity. These results imply that the cytoplasmic C terminus of GPA, including the region close to the TM domain, is responsible for enhanced trafficking of band 3 protein to the cell surface. The GPA-mediated enhancement of band 3 surface presentation could be achieved either by direct interaction of the GPA C terminus with band 3 or by interaction of the GPA C terminus with other proteins that may facilitate band 3 trafficking. Because the C-terminal GPA deletions retain the ability to enhance band 3 anion transport functional activity, it is clear this activity of GPA resides in the remaining part of the GPA molecule.
Extracellular GPA Residues 61-70 Are Involved in the Enhancement of Band 3 Anion Transport Activity-The region of GPA encompassed by the extracellular amino acid residues Arg-61-Glu-70 is thought to form a shared antigen epitope with residue 658 on band 3 (the Wr b antigen; (13)). Insertion or deletion of residues within this region severely reduced the ability of the GPA mutants to enhance the anion transport function of band 3 without affecting their ability to increase the movement of band 3 to the cell surface. In fact, deletion of Phe-68 -Glu-70 reduced enhancement of band 3-specific chloride transport by 80%, whereas the insertion of the 5 residues of GPB in this region (BintoA, Fig. 1) showed normal GPA enhancement of band 3 movement to the cell surface without a similar large enhancement of band 3 anion transport activity above the basal level. Although the transport and cell surface expression assays are not directly comparable, these results suggest that these GPA mutations inhibit the transport activity of band 3 relative to the situation where GPA is absent. Examination of the sequence in this region of the GPB/GPA hybrids Sta and Dantu (which both behave in the same way as wild-type GPA on band 3) shows that in Sta this sequence is FSE, and in Dantu this sequence is FTV, with residue Phe-68 in common. The mutant GPA F68C is somewhat impaired in its ability to enhance band 3 anion transport (61% enhancement). These observations suggest that residue Phe-68 of GPA may be important in the interaction between GPA and band 3, which results in enhanced band 3 anion transport activity. We conclude that the portion of GPA responsible for mediating the enhancement of band 3 anion transport activity (independently of effects on band 3 movement to the surface) is within the Phe-68 -Glu-70 extracellular region N-terminal to the GPA transmembrane span. This is located in the region of the Wr b antigen, which is implicated by other data (10 -13) to be involved in interactions between GPA and band 3.
Our observations that the presence of GPA results in the enhancement of the anion transport functional activity of band 3 confirm earlier results that showed that human red cells that lack GPA (and GPB) contain the normal number of band 3 molecules but have only 60% normal anion (sulfate) transport activity (20). The lower transport activity resulted from an increase in the apparent K m of the band 3 for sulfate. The band 3 in these cells has an abnormally long N-glycan chain, suggesting that there is an intracellular effect of GPA on band 3 during the biosynthetic pathway of the two proteins (20). GPA may facilitate the maturation of newly synthesized band 3 into a fully transport-active structure either directly or by recruiting other components that aid this process (16,20). It is not clear at present whether GPA also has a direct effect in increasing the transport activity of mature band 3 already present at red cell surface. The region around Phe-68 -Glu-70 of GPA clearly has a major role in these effects on the transport activity of band 3.
The N Terminus and Transmembrane Domain of GPA-The GPA/GPB hybrid MiV (GPA-(1-58)-GPB-(27-72)) behaved in the same fashion as GPB, indicating that the region of GPA from amino acids 1-58 is not involved in the band 3-GPA interaction. The two GPB/GPA hybrids Sta (GPB-(1-26)-GPA-(59 -131)) and Dantu (GPB-(1-39)-GPA-(72-131)) both enhanced band 3 presentation at the cell surface. There is a high degree of homology between amino acids 26 and 39 of GPB and 59 and 72 of GPA (Fig. 1B) and therefore portions of this region of GPA may still be important for the band 3-GPA interaction. Double substitution in the GPA transmembrane span (F78G793 LC) had no effect on the ability of GPA to interact with band 3, and we conclude from this that the membranespanning region of GPA is unlikely to be involved in the GPAband 3 interaction.
The Site of Interaction between Band 3 and GPA on the Band 3 Molecule-The above results show that the extracellular portion of the GPA molecule next to the plasma membrane is involved in enhancing band 3 anion transport function independently of effects on band 3 trafficking to the cell surface. Because this region is within the Wr b region of the protein, the site of interaction between band 3 and GPA on the band 3 molecule is likely to be close to the region of band 3 involved in the Wr b epitope, band 3 residue Glu-658 (15). According to recent topology models of band 3, Glu-658 is positioned at the extracellular end of transmembrane span 8 (4). It has also been observed that GPA enhances the surface movement of C-terminal fragments of band 3 but not the N-terminal fragments of band 3 (37). Further evidence suggests that the minimum fragment of band 3 on which GPA may act to enhance band 3 surface presentation is the portion encompassing putative TM 9 -12 (residues Met-696 -Tyr-824; (37)). The band 3 mutant G701D is totally dependent upon GPA for its cell-surface expression in oocytes (38,39). Band 3 Gly-701 is located at the cytoplasmic face of TM 9. It is conceivable that the band 3 mutation G701D leads to local misfolding of band 3 around TM 9, which GPA may refold by a chaperone-like effect. However, it is also possible that band 3 G701D is more generally misfolded and that GPA is able to rectify this problem in some other fashion.
Regions of the GPA Molecule N-terminal to the Transmembrane Span Influence GPA Dimer Formation-Cell-free translation of the GPA mutants showed that deletion of the extracellular six-amino acid region Ala-65-Glu-70 led to a large reduction in SDS-stable GPA dimer formation. This reduction could be explained in terms of the deletion bringing the Nterminal domains of the two GPA monomers too close together, causing steric hindrance and dissociation of the dimer. The fact that significant reduction in SDS-stable dimer formation was also observed for the four GPA substitution mutants F68C, S69C, P71C, and E72C suggests that for these mutants at least steric hindrance was not the reason for reduced dimer formation. A previous study using the GPA transmembrane domain fused at its N terminus to staphylococcal nuclease investigated the effect of truncating residues N-terminal to the transmembrane domain on SDS-stable dimer formation (30). The study employed the portion of GPA including residues Glu-60 -Ile-99 in the fusion protein with staphylococcal nuclease. SDS-stable dimer formation was significantly reduced upon truncation of the N-terminal 9 residues (Glu-60 -Phe-68) or 11 residues (Glu-60 -Glu-70). The addition of two alanine residues N-terminal to the nine-amino acid truncation did not significantly restore dimer formation, but the addition of four alanine residues N-terminal to the 11-amino acid truncation did. The authors concluded that the region His-67-Glu-70 was acting simply as a flexible linker between the GPA transmembrane span and the staphylococcal nuclease fusion (30). The effect of residues Pro-71 and Glu-72 upon GPA dimer formation has not been tested previously. Because our data indicate that both deletions and single substitutions within the region of Ala-65-Glu-72 of GPA affect SDS-stable dimer formation of the full-length protein, we infer that this region of the GPA molecule is also important for GPA dimer formation. The GPA transmembrane span is capable of forming dimers without the presence of the adjacent extracellular region of the protein, but in the fulllength protein it appears that the amino acid sequence of this region is critical for dimer formation. This result has implications for in vitro studies of GPA dimer formation.
It is also interesting that although the SDS-stable dimer formation of the double substitution mutant F78L/G79C was impaired, it was not completely abolished. Mutants of GPA where Gly-79 alone is changed to Leu, Val, or Thr exhibit essentially no dimer formation (16,23). It is possible that G79C is a more stable substitution or that F78G793 LC partially compensates for the deleterious G79 mutation. The GPB transmembrane span, which exhibits no SDS-stable dimer formation in our assay but has been shown to form dimers when extracted from red cell membranes (9), has two further amino acid differences compared with the F78L/G79C mutant (Thr-74I and Val-84I). This suggests that GPB dimer formation is significantly weaker than GPA dimer formation when expressed in endoplasmic reticulum microsomes because of the dual effect of the FG3 LC double mutation and the two further sequence differences in GPB. Our results also provide further confirmation of other data indicating that the oligomeric state of GPA has no bearing on its ability to interact with band 3 (16), as the GPA mutants S69C, P71C, and E72C all exhibit markedly reduced SDS-stable dimer formation but still enhance the expression of band 3-specific chloride transport in oocytes in a similar manner to normal GPA.
In summary, we have used site-directed mutagenesis to analyze the role of GPA in band 3 anion transport activity and cell surface expression, and our results lead to two main conclusions; first, that the role of the cytoplasmic C terminus of GPA is to enhance the cell surface movement of band 3 either by direct interaction with band 3 itself or by interaction with other proteins; second, that the extracellular region including residues 68 -70 of GPA close to the TM span is at least partially responsible for enhancement of band 3 anion transport functional activity, and this result leads to the conclusion that the Wr b antigen is a structurally and functionally important area of the GPA-band 3 interaction. Our results provide further evidence that GPA increases band 3 anion transport functional activity, consistent with previous observations that band 3-specific anion transport was reduced in glycophorin-deficient cells (20). These data also indicate that when expressed in oocytes, GPA acts upon band 3 in the same fashion as when physiologically expressed in red cells to both increase the anion transport activity of band 3 and band 3 trafficking to the cell surface. Finally, using cell-free translation experiments we provide evidence that the extracellular portion of GPA adjacent to the TM span (encompassing residues 61-72) is involved in dimer formation of intact GPA in addition to the transmembrane region.