High level expression, partial purification, and functional reconstitution of the human AE1 anion exchanger in Saccharomyces cerevisiae.

Human erythroid anion exchanger AE1 (Band 3) was expressed in the yeast Saccharomyces cerevisiae under the control of the constitutive promoter and transcriptional terminator of the yeast phosphoglycerate kinase gene. AE1 expression in stable yeast transformants was estimated to be approximately 0.7 mg AE1 per liter. Density gradient sedimentation analysis indicated that the AE1 protein was associated with a membrane fraction distinct from plasma membrane, most likely the endoplasmic reticulum. AE1 protein was solubilized from yeast membranes with lysophosphatidyl choline, and the protein, tagged with six histidines at its amino terminus, was purified to 35% homogeneity by metal chelation affinity chromatography. Size-exclusion chromatography in the presence of octaethylene glycol monododecyl ether indicated that the solubilized yeast-expressed AE1, like endogenous erythroid AE1, eluted at a stokes radius of 77 Å, consistent with a dimeric oligomeric state. Binding of partially purified yeast-expressed AE1 to 4-acetamido-4′-isothiocyanostilbene2,2′-disulfonate resin was competitive with the transportable substrate chloride but not the nontransported anion citrate, suggesting that the structure of the anion binding site is preserved. The specific activity of sulfate transport by partially purified yeast AE1 was determined in proteoliposomes to be similar to that of authentic AE1 purified from erythrocyte membranes. These data show that this expression system has the capacity to produce functional mammalian plasma membrane anion exchangers at levels sufficient for biochemical and biophysical analysis.

AE1 (Band 3) belongs to a family of anion exchange proteins that facilitate the movement of Cl Ϫ and HCO 3 Ϫ , across the plasma membrane (1). Plasma membrane anion-exchange proteins are widely expressed among mammalian tissues where they participate in the regulation of intracellular pH and volume (1,2). Three anion-exchanger isoforms have been identified, cloned, and sequenced: AE1, found in erythrocytes and kidney (3); AE2, found in kidney, stomach, and lymphocytes (4); AE3, found in the brain, retina, and heart (5). All of these anion-exchange proteins contain two domains. The highly conserved (70% identity) membrane domain of approximately 55 kDa spans the bilayer 12-14 times (6) and is responsible for anion-exchange activity (7). The cytoplasmic domain of  kDa is more divergent. In erythrocyte AE1, the cytoplasmic domain anchors the cytoskeleton to the plasma membranes through interactions with ankyrin (8).
AE1 has served as a model for understanding the structure and function of membrane proteins because of its high abundance in the erythrocyte membrane, where it constitutes nearly 50% of the total integral membrane protein (9). This allowed the early identification of the protein's role in the erythrocyte (10). Subsequently the protein has been extensively studied using a wide range of methods including nuclear magnetic resonance spectroscopy (11), chemical modification of specific residues (12), and electron diffraction (13). The ability to generate specific point mutant and chimeric proteins has provided important insight into the function of many proteins. However, to study recombinant proteins biochemically and biophysically requires protein expression systems that produce sufficient amounts of functional protein. The availability of suitable overexpression systems is a limiting factor for many mammalian membrane proteins, including AE1. Soluble proteins are readily expressed in bacterial cells, but eukaryotic membrane proteins are not functionally expressed in bacteria because bacteria do not usually target eukaryotic membrane proteins for insertion into the membrane. Eukaryotic membrane proteins expressed in bacteria are often toxic to the cell, resulting in low levels of expression (14,15). Furthermore, the thickness of the bacterial inner membrane (25 Å) (16) does not match the hydrophobic region of eukaryotic membrane proteins (30 Å) (17).
Recombinant plasma membrane anion exchange proteins have previously been expressed by transient transfection of human embryonic kidney 293 cells (HEK293) 1 (18), COS cells (19), and by cRNA injection of Xenopus laevis oocytes (20,21). AE2, expressed in insect cells using the baculovirus expression system, yielded an undetermined amount of protein that had a low level of anion-exchange activity (22). Insufficient amounts of protein are produced by these expression systems to permit protein purification and characterization. In this study, we report the establishment of a yeast expression system that produces high levels of human AE1 protein. The yeast-expressed AE1 protein was solubilized in detergent, partially purified, and reconstituted into proteoliposomes. The reconstituted protein is structurally and functionally indistinguishable from the native erythroid protein. A preliminary version of this work has been published as an abstract (23).
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Plasmid Constructions-The yeast expression construct was assembled from the vectors pMA230 (27) and pMA91 (28), kindly provided by Dr. S. Kingsman (Oxford University). pMA230 contains the yeast phosphoglycerate kinase 5Ј-untranslated region followed by the first 11 codons of phosphoglycerate kinase coding sequence and a BamHI site. A synthetic linker (5Ј GAT CGC ATC ATC ATC ATC ATC ATC CGC GG 3Ј/5Ј GAT CCC GCG GAT GAT GAT GAT GAT GAT GC 3Ј) was ligated into the BamHI site. This linker was designed to destroy the 5Ј BamHI site and to insert six histidine codons into the coding sequence, downstream of the phosphoglycerate kinase codons. The BamHI fragment of the cDNA for human AE1 (29) was then inserted into the linkermodified pMA230, yielding pJRC15. pMA230 does not encode any of the 3Ј-untranslated region of the phosphoglycerate kinase gene; this region is essential for message stability. The vector pMA91 contains the phosphoglycerate kinase 5Ј-and 3Ј-untranslated regions surrounding a BglII site. Polymerase chain reaction, using pJRC15 as template and the synthetic oligomers 5Ј GAA GAT CTA TGT CTT TAT CTT CAA AG 3Ј (corresponding to the fusion protein coding sequence, with a BglII site at the 5Ј end) and 5Ј CTC TCT GAC ATG AGG GTG GC 3Ј (from the human AE1 sequence) produced a DNA fragment with a 5Ј BglII site, followed by the first 11 codons of yeast phosphoglycerate kinase, linker sequence, and human AE1. The PCR product was cut at the BglII site at the 5Ј end and at a ClaI site at the 3Ј end, within the region of human AE1. This BglII/ClaI-digested PCR product was then mixed with the ClaI/BamHI fragment of AE1 cDNA, containing the remaining coding sequence of AE1 and ligated into the BglII site of pMA91 to yield pJRC16. DNA sequences generated with polymerase chain reaction were sequenced in their entirety to ensure that no sequence errors were introduced.
Yeast Transformation-Yeast were transformed in the presence of lithium acetate and polyethylene glycol, based on previous protocols (30 -32). Overnight cultures of YPH499 or BJ1991 (1 ml) were centrifuged for 5 min at 5,000 rpm. Cells were washed in water and then with LiTE buffer (0.1 M lithium acetate, 10 mM Tris, 1 mM EDTA, pH 7.5). The cell pellet was then resuspended in 50 l of LiTE buffer. 10 l of freshly boiled sheared salmon sperm DNA (10 mg/ml) and 1 g of plasmid DNA were then added, and the sample was mixed. LiTE buffer (0.3 ml) containing 35% (w/v) polyethylene glycol 4000 was then added. Samples were incubated for 1 h at 30°C and then heat shocked by incubation for 15 min at 42°C. Cells were pelleted as above and resuspended in 0.5 ml of water. Cells were plated on Leu Ϫ -selective plates, and colonies took 5-7 days to appear. When several independent colonies of BJ1991(pJRC16) yeast were picked and membrane preparations were prepared from each, it was found that the slowest growing cultures expressed the highest levels of AE1. Some rapidly growing cultures did not express AE1 at all. Therefore, it was essential to monitor the growth rate and not to use any starter cultures that had a doubling time shorter than 4 h.
Isolation of Yeast Membranes-Yeast cultures grown to A 600 ϭ 1-1.5 were centrifuged at 4,000 ϫ g for 5 min and washed once with water. Cell pellets were resuspended in four pellet volumes of homogenization buffer (10 mM Tris, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 0.1% (v/v) 2-mercaptoethanol, pH 7.4). Cells were disrupted by vigorous vortexing in the presence of glass beads for 15 min at 4°C. Cell debris was removed by centrifugation for 5 min at 500 ϫ g. Membranes were resuspended in 40 ml of homogenization buffer and centrifuged for 30 min at 13,000 rpm in an SS34 rotor. Membranes were resuspended to approximately 10 pellet volumes in homogenization buffer.
Sucrose Gradient Centrifugation-Membranes were isolated as described above and made to 10% (w/v) sucrose. A 0.3-ml sample of membranes was layered onto 11 ml of a 20 -53% (w/v) sucrose gradient made up in 1 mM EDTA, 10 mM Tris, pH 8.0. The sample was centrifuged 18 h at 4°C, 30,000 rpm in a Beckman SW41 rotor. Fractions (0.5 ml) were removed from the top of the gradient.
Protein Purification-To remove EDTA and Tris, membranes were washed in 10 mM sodium phosphate, pH 8.0, by centrifugation for 30 min at 13,000 rpm in an SS34 rotor. Membranes were resuspended in 5 pellet volumes of nickel column buffer (10% glycerol, 100 mM sodium chloride, 5 mM imidazole, 10 mM sodium phosphate, pH 8.0). Membranes were solubilized by the addition of LPC to a final concentration of 2% (w/v). After incubation for 20 min on ice, the solution was centrifuged at 80,000 ϫ g for 10 min. The supernatant (20 -40 mg of protein/ml of resin) was applied at 0.25 ml/min to a His-Bind resin column. The column was washed at a flow rate of 1 ml/min with nickel column buffer containing 0.1% (v/v) C 12 E 8 until the A 280 of the eluent dropped below 0.02. Nickel chromatography was performed in the presence of 0.1% (w/v) LPC instead of C 12 E 8 , in samples prepared for reconstitution. Protein was eluted with nickel column buffer containing 0.1% detergent and 250 mM imidazole. In some cases, fractions containing AE1 were pooled and dialyzed for 2 h against nickel column buffer supplemented with 0.1% (w/v) LPC. Dialyzed AE1 was reapplied to a 1.5-ml nickel column. The column was washed with 3 volumes of nickel column buffer containing 0.1% (v/v) LPC, followed by the same buffer supplemented with 30 mM imidazole. AE1 was eluted with nickel column buffer, containing 0.1% (v/v) C 12 E 8 and 250 mM imidazole. AE1 was purified from erythrocyte membranes that were stripped with 2 mM EDTA, pH 12, solubilized in LPC detergent and chromatographed using aminoethyl-Sepharose resin (33).
Size Exclusion Chromatography-Size exclusion chromatography and immunodetection of the eluate were performed as described previously (34). AE1 purified by nickel-affinity chromatography was subjected to size exclusion HPLC on a 0.75 ϫ 30-cm TSK 4000SW column, eluted with 0.1 M sodium chloride, 0.1% (v/v) C 12 E 8 , 5 mM sodium phosphate, pH 7.0, as described previously (35). Flow rate was 0.5 ml/min, using a Beckman 114 M pump. The column was calibrated with protein standards (Pharmacia Biotech Inc.) that were shown not to bind detergent (36). To assay the elution position of the anion-exchanger protein, 30-s fractions of column eluate were collected, made to 1% (w/v) SDS, dot-blotted onto nitrocellulose membrane, and processed as immunoblots.
SITS⅐Affi-Gel Binding-SITS⅐Affi-Gel 102 was prepared as described previously (37), and binding of yeast AE1 to SITS⅐Affi-Gel was performed essentially as described previously (38). Purified yeast AE1 (50 l of 0.3 mg/ml) was dialyzed against 0.1% (v/v) C 12 E 8 , containing either 200 mM sodium citrate, pH 8.0 or 100 mM sodium citrate, 100 mM sodium chloride, pH 8.0. This protein was incubated with 20 l of SITS⅐Affi-Gel resin for 20 min at 4°C. The resin was sedimented by centrifugation for 10 s in a microcentrifuge and washed 3 times with 200 l of dialysis buffer. The resin was then incubated for 20 min at 20°C with 50 l of 1 mM DNDS in either of the above buffers. The resin was sedimented, and the supernatant was removed. The resin was then incubated with 50 l of SDS gel sample buffer (39) at 60°C for 3 min, and the supernatant was collected.
Deglycosylation-Purified AE1 protein from erythrocytes and partially purified AE1 protein from BJ1991(pJRC16) yeast (2 g) in 0.1% LPC were denatured by incubation with 0.5% (w/v) SDS, 1% (w/w) 2-mercaptoethanol for 10 min at 50°C. The samples were made to 0.5% (w/w) Nonidet P-40, and 13,000 units of N-glycosidase-F (New England Biolabs) or water (to negative control tubes) was added. Samples were incubated for 19 h at 37°C and then subjected to SDS-polyacrylamide gel electrophoresis and immunoblotting, as described below.
Protein Reconstitution and Anion Transport Assay-Yeast membranes (200 g of protein) were reconstituted with 10 mg of soybean asolectin phospholipid and assayed for anion-exchange activity as described previously for human embryonic kidney cell membranes (40). The method for reconstitution of purified AE1 made use of a novel protocol in which the LPC detergent was not removed but rather was incorporated into the lipid vesicles. Experiments showed that proteoliposomes prepared by dilution of LPC into lipid were well sealed, as shown by a very small leak of sulfate, chloride, and nitrate, in final LPC concentrations of up to 0.2% in the reconstitution mixture. 2 Since we noted that many preparations of phosphatidylcholine contain up to 0.4% LPC (41), we reasoned that moderate amounts of LPC would not compromise the integrity of the vesicles. The integrity of phospholipid vesicles resulting from reconstitution in the presence of LPC indicates that this is an effective protocol that may be useful for other proteins that bind to SM2 Bio-Beads. For purified AE1 in 0.1% (w/v) LPC solution, the protein (50 -200 g) was dialyzed against reaction buffer (40 mM Na 2 SO 4 , 4 mM MgSO 4 , 20 mM MES/Tris, pH 6.0) supplemented with 0.1% (w/v) LPC. Approximately 10 g of the purified yeast AE1 and 5 g of purified erythroid AE1 were made to 0.5 ml with reaction buffer. Reconstitution mixtures contained the diluted protein, 0.45 ml of soybean asolectin (22 mg/ml in water), [ 35 S]SO 4 2Ϫ (50 Ci) and for inhibition studies, H 2 DIDS to a final concentration of 200 M. Samples were made to a total volume of 1.0 ml by addition of water. Reconstitution proceeded by freezing the mixture in liquid nitrogen followed by thawing at room temperature and sonication (40). In anion-exchange assays, exchange of intravesicular [ 35  2Ϫ was removed, and sulfate efflux was determined as described previously (40).
Electrophoresis and Immunoblotting-SDS-polyacrylamide gel electrophoresis (39) was performed, and proteins were transferred to nitrocellulose as described previously (42). Western blots were blocked by incubation for 30 min in antibody buffer (5% (w/v) nonfat dry milk, 137 mM NaCl, 20 mM Tris/HCl, pH 7.6). Blots were incubated with anti-AE1 antibody 5-297, an anti-peptide antibody that was raised against a synthetic peptide corresponding to the carboxyl-terminal 12 amino acids of mouse band 3 protein (43). Conditions were 10 ml/blot of 1:2500diluted antibody in antibody buffer, for 2 h at room temperature followed by 10 ml/blot of 1:2500-diluted donkey anti-rabbit IgG conjugated to horse radish peroxidase (Amersham Corp.), incubated for 1 h at room temperature. Blots were visualized using Renaissance chemiluminescent reagent (Dupont) and Hyperfilm (Amersham Corp.). Protein concentrations were determined using bicinchoninic acid reagent (Sigma) (44), in the presence of 0.1% final concentration of SDS. Immunoblots were quantified using a Scanjet Plus scanner (Hewlett Packard) and Image 1.44 software (National Institutes of Health, Bethesda, MD).

RESULTS
Expression and Characterization of Human AE1 Expressed in S. cerevisiae-Initial attempts to express mammalian anion exchangers AE1 and AE2 in yeast using the vector pYES2, containing an inducible galactose promoter failed to produce detectable protein. Consequently, we constructed a new vector, pJRC16 (Fig. 1), which incorporates several features that have been previously shown to be important for high level protein expression in yeast (45,46). These include the 5Ј-and 3Јuntranslated regions from a highly expressed yeast gene, in this case yeast phosphoglycerate kinase, and the first 10 codons of yeast phosphoglycerate kinase. Codons 183-911 of human AE1 (on a BamHI fragment), preceded by a sequence of six histidine codons (an affinity purification tag) were inserted immediately downstream of the strong, constitutive phosphoglycerate kinase promoter. The AE1 protein encoded by this construct, therefore, lacks the NH 2 -terminal 182-amino-acid residues from the cytoplasmic domain but contains the entire membrane domain that is both necessary and sufficient (7,47) to conduct anion exchange. The plasmid was used to transform protease-deficient S. cerevisiae strain, BJ1991. 3 Individual colonies were assayed for expression by immunoblotting of membrane fractions. Several individual colonies expressing high amounts of AE1 were isolated for further analysis. The growth rate (10 h doubling time) of these colonies was up to 5 times slower than strains transformed with vector alone, suggesting that expression of YAE1 interferes with growth.
No immunoreactive material was observed in membranes prepared from yeast transformed with vector pMA91 alone (Fig. 2, lane 13). The AE1 protein encoded by pJRC16 migrated on SDS-PAGE with an apparent molecular mass of 75 kDa, which is in good agreement with the predicted molecular mass of 80 kDa. The expression level of AE1 in BJ1991(pJRC16) membranes was determined by densitometric comparison of immunoblots of AE1 from yeast and erythrocyte membranes (Fig. 2). From the known abundance of AE1 (25% of erythrocyte membrane protein) (9), yeast-expressed protein (YAE1) constitutes approximately 1.5% of the membrane protein in the yeast. Since 1 liter of yeast grown to A 600 ϭ 1.0 yields approximately 44 mg of membrane protein, the expression level is about 0.7 mg of AE1/liter of culture.
Erythrocyte AE1 is heterogeneously glycosylated with up to 10 kDa of carbohydrate (48); however, this carbohydrate is not required for anion-exchange activity of the protein (49). Some mammalian membrane proteins functionally expressed in yeast are not glycosylated (50,51). To determine whether the yeast-expressed AE1 protein was glycosylated, BJ1991-(pJRC16) membranes and human erythrocyte membranes were treated extensively with the enzyme N-glycosidase F, which cleaves the entire carbohydrate structure at the asparagine linkage of N-linked carbohydrates (52) (data not shown). The electrophoretic mobility of YAE1 was unaltered by N-glycosidase F treatment under conditions that resulted in complete deglycosylation of a parallel sample of erythrocyte AE1, suggesting that AE1 expressed in yeast is not glycosylated.
The subcellular location of yeast-expressed AE1 was analyzed by density gradient sedimentation analysis on linear 20 -53% sucrose gradients. Fractions from the gradient were analyzed by immunoblotting for AE1 and the yeast plasma membrane marker PMA1, using specific antisera (Fig. 3). YAE1 and PMA1 were separated on the gradient into two distinct, well defined peaks at 48% sucrose and 51% sucrose, respectively. Although this analysis does not exclude the possibility that some AE1 may be present in the plasma membrane, the data suggest that most AE1 is present in other membranes, most likely endoplasmic reticulum.
To assess the capacity for yeast-expressed AE1 to conduct anion exchange, yeast membranes were reconstituted into vesicles with exogenous lipid (40). The vesicles were loaded with 20 mM [ 35 S]SO 4 2Ϫ , and assayed for anion exchange by measuring the efflux of radioactive sulfate in exchange for extravesicular sulfate (40) (Fig. 4). Since AE1 was reconstituted with a large excess of phospholipid, approximately 80% of sealed vesicles lack an anion exchanger. Consequently, a large fraction of [ 35 (53). We conclude that AE1 protein expressed in yeast is able to carry out anion exchange.
Solubilization and Characterization of YAE1-To determine the optimal solubilization conditions for yeast AE1, membranes from BJ1991(pJRC16) yeast were exposed to a panel of eight different conditions (Fig. 5). AE1 is nearly completely solubilized from erythrocyte membranes by C 12 E 8 and has been extensively characterized in this detergent (33). By contrast, only 38% of AE1 was solubilized from yeast membranes by C 12 E 8 . However, LPC solubilized 93% of the amount solubilized by SDS, while no AE1 was found in the supernatant fraction in the absence of detergent. LPC, which has previously been used to solubilize the plant H ϩ -ATPase expressed in yeast (46), is zwitterionic and is considered to be nondenaturing. LPC was therefore used to solubilize AE1 from yeast membranes.
To purify AE1 from yeast, LPC-solubilized yeast membranes were bound to a Ni 2ϩ -loaded metal chelating resin and eluted with 0.25 M imidazole. Immobilized nickel has an affinity for a sequence of six histidine residues (54), such as those introduced near the amino terminus of YAE1. Approximately 100-fold purification was achieved following a single cycle of binding and elution from the column; this was increased to 145-fold after a second cycle ( Table I). Analysis of the fractions on a Coomassie Blue-stained SDS-polyacrylamide gel (Fig. 6A) indicates that AE1 is the major protein, constituting 35% of the total protein. This band is strongly reactive in immunoblots with an AE1 antibody, confirming its identity as AE1 (Fig. 6B). The major high molecular weight band observed in these immunoblots is probably dimeric AE1 as it copurified with YAE1 and, like erythrocyte AE1, increased in amount as the samples aged (35).
The oligomeric state of partially purified, LPC-solublilized YAE1 was determined by size exclusion HPLC in 0.1% (v/v) C 12 E 8 and immunological detection (34). The elution profile of purified erythrocyte AE1 (Fig. 7, upper panel) has four peaks, which have been previously identified as highly associated protein, eluting at the void volume (5.0 ml), tetrameric AE1 (6.8 ml), dimeric AE1 (7.9 ml), and detergent micelles (10.4 ml) (35). The majority of yeast AE1 (Fig. 7, lower panel) eluted in two major peaks, at 6.6 and 8.1 ml, which correspond almost precisely to the tetramer and dimers of erythroid AE1, respectively. Moreover, since the NH 2 -terminal domain of YAE1 lacks the first 182 amino acid residues of the erythrocyte protein, these results suggest that the truncated region contributes little to the hydrodynamic behavior of full-length AE1 in detergent solution.
Binding of detergent-solubilized, partially purified yeast AE1 to immobilized disulfonic stilbene inhibitor, SITS (38) was used to assess the structural integrity of the anion binding site (Fig. 8). Since binding of transportable anions to AE1 in erythrocytes is competitive with disulfonic stilbenes (53), the SITS⅐Affi-Gel resin was incubated with partially purified YAE1 in the presence of either citrate, a nontransportable anion, or chloride, a transportable anion. In the presence of citrate, almost all of the AE1 bound to the SITS-resin and could be eluted with the structurally related inhibitor, DNDS. By contrast, the presence of chloride significantly attenuated the binding of AE1 to the resin. These results suggest that the stilbene disulfonate and anion binding sites of yeast-expressed AE1 are preserved, supporting the conclusion that the polypeptide is correctly folded.
Sulfate Transport by Reconstituted Yeast AE1-The anionexchange activity of purified YAE1 was directly compared with that of AE1 purified to homogeneity from erythrocytes. Both proteins in LPC solution were reconstituted into phosphatidylcholine vesicles and anion-exchange activity was assayed by measuring the efflux of intravesicular [ 35 (Fig. 9). The data show that AE1, from erythrocytes and yeast, mediated significantly enhanced anion-exchange fluxes compared with background transport in vesicles reconstituted without protein (data not shown). Anion transport mediated by both exchangers was completely inhibited by DIDS (Fig. 9). The initial rates of anion exchange were 6.6 ϫ 10 6 and 2.7 ϫ 10 6 cpm⅐mg of protein Ϫ1 ⅐min Ϫ1 for erythrocyte and yeast AE1, respectively. Since the erythrocyte AE1 was purified to homogeneity while YAE1 was only approximately 35% pure, the similarity of the specific transport activities of the two protein preparations indicates that purified YAE1 is fully functional. In addition, the specific activity of purified YAE1 is approximately 50-fold higher than for crude YAE1expressing yeast membranes (Fig. 4). These data indicate that AE1 retains functionality after purification in detergent solution and reconstitution. DISCUSSION This study demonstrates that the human erythrocyte anion exchanger AE1 can be expressed at high levels in the yeast S. cerevisiae. Several parameters indicate that the yeast-expressed AE1 protein is similar to the endogenous erythrocyte anion exchanger in native membranes. Yeast AE1, like erythroid AE1, had a Stokes radius in C 12 E 8 solution consistent with a dimeric structure and showed anion-specific binding to inhibitor resin. Most significantly, partially purified AE1 from yeast, reconstituted into proteoliposomes, mediated sulfate anion exchange with specific activity similar to erythroid AE1. To-  6. Purification of yeast-expressed AE1. Membranes from BJ1991(pJRC16) yeast were solubilized and centrifuged, and the supernatant was collected. The supernatant was applied to His-Bind resin; the column was washed with nickel column buffer (10% glycerol, 100 mM sodium chloride, 5 mM imidazole, 0.2% (w/v) LPC, 10 mM sodium phosphate, pH 8.0) and eluted with this buffer containing 250 mM imidazole. The pooled peak of eluting protein was dialyzed against nickel column buffer, rechromatographed on the nickel column, and eluted with nickel column buffer containing 300 mM imidazole. Protein samples were resolved on 7.5% acrylamide gels that were either stained with Coomassie Blue (A) or transferred to nitrocellulose and probed with anti-AE1 antibody 5 gether, these data indicate that AE1 has been functionally expressed in S. cerevisiae and that this protein has the structural and functional characteristics of native AE1 from erythrocytes. To our knowledge, this is the first example of purification and functional reconstitution of a mammalian plasma membrane transport protein overexpressed in yeast.
The production of sufficient amounts of purified recombinant membrane proteins is a limiting factor in the study of mem-brane protein structure in biochemistry. Other popular expression systems include stable or transient expression in mammalian cells, Xenopus oocytes, and expression in baculovirusinfected insect cells. However, none of these systems assures high levels of expression in the correct membrane, proper posttranslational modification, or, most importantly, functionality. For example, expression of AE1 by transient transfection of human embryonic 293 cells yields about 170 g of membrane protein/100 cm 2 of tissue culture dish surface of which AE1 constitutes 0.4%, or 6.8 g of AE1 protein. 4 To express the amount of AE1 found in 1 liter of yeast (700 g) in HEK293 cells would therefore require about 1 m 2 of tissue culture dish surface, an impractically large amount. Furthermore, this protein is retained in the endoplasmic reticulum, where it is glycosylated only with core carbohydrate (18). Similarly, AE2 was found at levels comparable with those in HEK293 cells when expressed in Sf9 insect cells using baculovirus (22).
The unicellular eukaryote S. cerevisiae has been used for the expression of heterologous membrane transporters with mixed success. Plant membrane proteins are much more readily ex-4 J. R. Casey, unpublished results. . Radioactivity remaining associated with the vesicles eluted from the Dowex 1 columns was measured by scintillation counting. Data represent the mean of three independent experiments each performed in triplicate. pressed in S. cerevisiae than are mammalian membrane proteins, as exemplified by the sheep Na ϩ ,K ϩ -ATPase, expressed as 0.1% of yeast membrane protein (50) and the plant H ϩ -ATPase found as 40 -50% of endoplasmic reticulum membrane protein (46). Relative to other mammalian membrane proteins expressed in yeast, our observation that AE1 constitutes 1.5% of yeast membrane protein represents a high level of expression. One elegant yeast expression system makes use of yeast sec mutants, which accumulate membrane proteins in uniformsized, sealed vesicles of defined orientation. These sec mutants have been used to express the H ϩ -ATPase (55) and P-glycoprotein (51), which proved useful for transport assays. Unfortunately, the amount of protein accumulated in this expression system is well below the mg of protein/liter range reported here, making it less suitable for biochemical or structural studies.
In conclusion, the expression system described here provides milligram quantities of functional, recombinant AE1 protein, which will be useful for future biophysical and biochemical characterization. The expression system described here should also be useful for the expression and characterization of other mammalian membrane proteins.