A structural study of the membrane domain of band 3 by tryptic digestion. Conformational change of band 3 in situ induced by alkali treatment.

Nine peptides derived from the transmembrane domain of band 3 were purified and sequenced. All of the sequences agreed completely with deduced sequences from cDNA of human erythroid band 3. Five peptides, KS-1 to KS-5, were released from the band 3 molecule when alkali-stripped membranes were digested with trypsin, while four other peptides, KM-6 to KM-9, were obtained following subsequent urea treatment. This indicates that at least 13 new in situ cleavage sites were demonstrable by these procedures, that the released peptides are parts of hydrophilic connector loops, and that the other peptide portions constitute membrane-spanning helices. The topological designations are consistent with the hydropathy prediction of murine band 3 according to Passow ((1986) Rev. Physiol. Biochem. Pharmacol. 103, 61-203). One mol of histidine residue was found/mole of KS-1, KS-2, KS-4, and KM-6. The conformation of band 3 in situ was apparently changed by alkali treatment of erythrocyte membranes, i.e. the amount of KS-1, KS-2, and KS-4 peptides released by trypsin treatment increased as NaOH concentration was raised from 10 to 100 mM. Similarly, [3H]dihydro-4,4'-diisothiocyanostilbene-2,2'-disulfonic acid was found to bind to band 3 in membranes treated with 10 mM NaOH as well as to band 3 in white ghosts, but not to membranes treated with 100 mM NaOH. In addition, alkali treatment of membranes tended to increase the amount of band 3 cross-linked by 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid (DIDS). The conformational change in band 3 by alkali treatment was also supported by the interaction of antibodies against peptides released by trypsin. The release of KS-1, KS-2, and KS-4 from the membrane was strongly inhibited by pretreating the erythrocyte membrane with DIDS, suggesting that the DIDS-band 3 complex which is in the outward facing form, is more compact and becomes resistant to trypsin compared to band 3 without DIDS.

1982; Passow, 1986;Jennings and Anderson, 1987). Although the complete sequence of the human erythrocyte band 3 protein has been deduced from the cDNA sequence (Tanner et al., 1988;Lux et al., 1989), little is known about the nature and location of the functionally important amino acids. The site of inhibition of anion-exchange has been investigated by amino acid modification with pyridoxal phosphate on suitable proteolytic fragments of band 3 in combination with flux measurements (Hamasaki et al., 1983;Kawano and Hamasaki, 1986). The inhibitory site was identified as Lys-851 in the deduced sequence (Kawano et al., 1988), and this amino acid is the only one which is well characterized in anion transport.
In addition to the above amino acids, we have shown that an intracellular histidine residue of band 3 also participates in anion transport (Matsuyama et al., 1986). The intracellular histidine residue was modified by DEPC' and extracellular binding of DNDS to band 3 prevented this DEPC modification , suggesting that this histidine residue is hidden from the intracellular surface of the membrane in its outward conformation and is thereby protected from DEPC modification by extracellular DNDS binding .
In order to elucidate the topology of the functionally important amino acid residues within band 3, we have tried to assess the hydropathy predictions of band 3 in the present paper by digesting cell membranes with trypsin and relating the trypsin susceptibility of band 3 to its conformational changes.

EXPERIMENTAL PROCEDURES
Preparation of Leaky Human Erythrocyte Membranes (White Ghosts)-Human blood which was stored for less than 2 weeks was used in these studies. The stored blood was obtained from the Fukuoka Red Cross Blood Center where it had been maintained in a citrate/phosphate/dextrose solution at 4 "C. Erythrocytes were pretreated with bovine trypsin to digest the glycophorins. After washing, red blood cells were lysed osmotically with more than 20 volumes of low ionic strength solution, i.e. 5 mM sodium phosphate, pH 8 (5P8). Membranes (white ghosts) were centrifuged at 24,000 X g for 30 min at 4 "C in a Hitachi RR 19A centrifuge and washed with the same buffer. The packed ghosts were stored at -40 "C until they were used.

Isolation of Hydrophilic Connector Loops
of Band 3 trypsin treatment with high salt concentration); the cytosolic 40,000dalton domain of band 3 was thereby cleaved. After addition of antipain (10 pg/ml), the treated membranes were separated by centrifugation (45,600 X g for 20 min at 4 "C) and washed extensively with the same buffer. Peripheral membrane proteins in the washed membranes were stripped with 5 volumes of 0.1 M NaOH at 4 "C for 30 min and centrifuged at 45,600 X g for 20 min at 4 "C. The alkalistripped membranes were washed with 5P8 three times and resuspended in 5P8 at a protein concentration of 1.5 mg/ml. The membrane suspension was redigested with 15 pg/ml of trypsin at 37 "C for 30 min (i.e. trypsin treatment with low salt concentration). Peptides released into the supernatant were collected by centrifugation at 27,200 X g for 30 min at 4 "C. Under these conditions, supernatants were clear and had no membranes. The filtered supernatant was analyzed by high performance liquid chromatography (HPLC) equipped with a reversed phase column (Cosmosil C-18, 4.6 X 250 mm), using a linear gradient of 0-100% acetonitrile containing 0.1% trifluoroacetic acid. This was the standard HPLC procedure for chromatographing peptides released into the supernatant.
Analysis of Peptides Remaining in the Membranes after Tryptic Digestion-White ghosts were treated with 5 volumes of 0.1 M NaOH at 4 "C for 30 min to remove peripheral proteins. The alkali-stripped membranes in 5P8 (1 mg/ml) were digested by 200 pg/ml of trypsin at 37 "C for 1 h. After centrifugation at 27,200 X g for 30 min at 4 "C, the membranes were resuspended in 5P8 containing 10 pg/ml of antipain, and the released peptides were washed out from the membranes with 5P8. The digested membranes were dissolved in an equal volume of 8 M urea and 4% SDS containing 0.05 M dithiothreitol and 10% (v/v) formic acid, and warmed at 60 "C for 5 min. Peptides in the dissolved membrane solution were separated on a Cosmosil C-8 column (4.6 X 300 mm) using a linear gradient of 40-100% acetonitrile containing 0.1% trifluoroacetic acid.

Treatment of White Ghosts with Various Concentrations of NaOH-
White ghosts in cold water were mixed with 6 volumes of ice-cold NaOH solution varying its concentration from 10 to 100 mM. The mixture of white ghosts and NaOH was immediately centrifuged at 45,600 X g for 20 min at 4 "C. The pellet was washed three times with 40 volumes of 5P8 by centrifugation at 45,600 X g for 20 min at 4 "C. The alkali-treated membranes were suspended in 5P8 at a protein concentration of 1.5 mg/ml and digested with 15 pg/ml of trypsin at 37 "C for 30 min as described above. Peptides released into the supernatant were collected by centrifugation a t 27,200 X g for 30 min a t 4 "C and analyzed by the standard HPLC procedure. Peripheral membrane proteins were completely extracted from white ghosts when the membranes were treated with 10 mM NaOH as shown in Fig. 3  H,DIDS just as in the case of intact cells. The reaction was stopped by addition of one-tenth volume of 5% bovine serum albumin, and the DIDS-pretreated membranes were washed with 40 volumes of 5P8 three times to remove nonreacted free DIDS. The pretreated white ghosts were suspended in ice-cold water with a protein concentration of 4 mg/ml. The DIDS-pretreated ghosts were mixed with 6 volumes of ice-cold NaOH solution varying in concentration from 10 to 100 mM. The mixture of ghosts and NaOH was immediately centrifuged at 45,600 X g for 20 min at 4 "C. The membranes were washed three times with 40 volumes of 5P8 by centrifugation at 45,600 X g for 20 min at 4 "C. The alkali-treated membranes were suspended in 5P8 with a protein concentration of 1.5 mg/ml and digested with 15 pg/ml of trypsin at 37 "C for 30 min as described above. Peptides released into the supernatant were collected by centrifugation and analyzed by the standard HPLC procedure.
Radioactive Dihydro-DZDS Binding to Band 3-White ghosts in 5P8 (1 mg/ml), which had been treated with 6 volumes of various concentrations of NaOH as described above, were incubated with 25 p~ of [3H]H,DIDS with a specific radioactivity of not less than 200 mCi/mmol for 30 min at 37 "C. Free [3H]H2DIDS was removed by with albumin-free 5P8 three times. The quantity of [3H]H2DIDS washing with 0.5% bovine serum albumin-containing 5P8 once and bound to band 3 was analyzed by SDS-PAGE in sliced gels containing a constant amount of band 3 (15 pg).
Cross-linking of Band 3 by DZDS-Erythrocytes were pretreated with chymotrypsin to cleave band 3 into two fragments of 38,000and 60,000-dalton domains as described previously (Matsuyama et al., 19831, and white ghosts were prepared from the digested erythrocytes as described above. The digested ghosts (1 mg/ml) were pretreated with 100 pM DIDS in 5P8 at 37 "C for 1 h and washed with 5P8 containing 0.5% bovine serum albumin once and with albumin-free 5P8 three times to remove free DIDS. The DIDSpretreated ghosts (4 mg/ml) in cold water were suspended in 6 volumes of ice-cold NaOH varying its concentration from 10 to 100 mM. The NaOH-ghosts mixture was immediately centrifuged a t 45,600 X g for 20 min at 4 "C. Further washing of the ghosts was done as mentioned above. The amount of cross-linked band 3 in the ghosts was determined by SDS-PAGE. Peptide Synthesis and Antibody Production-The multiple antigen peptides of KS-1 and KS-4 (which correspond to amino acid residues from Lys-817 t o Arg-827 and from Arg-646 to Arg-656, respectively, in the deduced sequence of human erythroid band 3) were synthesized using a peptide synthesizer (431 A, Applied Biosystems, Foster City, CA) in accordance with the method of Tam (Tam, 1988;Tam and Lu, 1989). The synthesized crude peptides were solubilized by 50% acetic acid and separated by HPLC equipped with a reversed-phase column (Chemcosorb 300-7C4,4.6 X 300, Chemcopak, Osaka, Japan) using a linear gradient of 0-100% acetonitrile containing 0.1% trifluoroacetic acid. The molecular weight and amino acid composition of the purified peptides examined were as anticipated (results not shown).
Except for the initial 200-pg dose, 100 pg/rabbit of each multiple antigen were used with Freund's adjuvant to immunize rabbits five times every 2 weeks. The binding of each antiserum to erythrocyte membranes was measured as follows. Ten pg of membranes/well were attached to plates (Sumitomo multiplates MS-3596F/H, Tokyo, Japan). The indicated volume of serum in 100 p1 of PBS containing 0.5% bovine serum albumin was added per well and allowed to stand for 1 h at 37 "C. After washing with PBS three times, 100 pl of PBS containing horseradish peroxidase-conjugated goat anti rabbit IgG (E-Y Laboratories, San Mateo, CA) and 0.5% of bovine serum albumin were added, and the mixture was incubated at 37 "C for 1 h. The content of bound IgG to membranes was assayed by a peroxidation reaction using o-phenylenediamine as a substrate at 492 nm. There was no cross-reaction between the site-specific antibodies.
Analytical Procedures-SDS-PAGE for protein analysis was carried out according to the method of Laemmli (1970). Protein was determined by the method of Lowry et al. (1951) using bovine serum albumin as the standard. Radioactivity was determined in sliced gels with a liquid scintillation spectrometer. Peptides were sequenced on a gas-phase sequencer (Applied Biosystems, model 470A), and the phenylthiohydantoins were identified by an Applied Biosystems 120 A phenylthiohydantoin analyzer on-line system.
Materials-[3H]H2DIDS were purchased from HSC Research Development Corp. (Toronto, Canada). DIDS was obtained from Calbiochem. Other reagents were of analytical grade.

Peptides Released from Erythrocyte Membranes by Tryptic
Digestion-To simplify the analytical procedure, the cytosolic 40,000-dalton domain of band 3 was nicked and removed by treating white ghosts with trypsin at 4 "C as described under "Experimental Procedures." Peripheral membrane proteins were also removed with NaOH at 4 "C. Thus, the transmembranous 55,000-dalton domain of band 3 is the major component remaining (Fig. 1). The treated membranes were further digested with the same concentration of trypsin (trypsin with low salt concentration) at 37 "C, and the peptides released were analyzed by the standard HPLC procedure. Fig. 2A shows a typical elution profile of a HPLC analysis. The major peaks were collected, rechromatographed with the same HPLC system, and the purified peptides were sequenced by a gas-phase sequencer. At least five peptides of band 3 origin, designated as KS-1-KS-5, were released into the supernatant from the alkali-stripped membranes by trypsin treatment, and the amino acid sequences were the same as the amino acid sequences deduced from cDNA ( Table I) (Tanner et al., 1988;Lux e t al., 1989). This indicates that the released peptides are included in the hydrophilic connector loops. Among Erythrocyte membranes (1 mg/ml) were treated with a low concentration of trypsin (15 pg/ml) a t 0 "C for 30 min in 0.14 M NaCl (pH 8). The trypsinized membranes were extracted with 100 mM NaOH. The various membrane preparations were analyzed by SDS-PAGE according to Laemmli (1970) with 9% acrylamide and stained with Coomassie Blue. 55KDa, the transmembranous 55,000-dalton domain of band 3. A , white ghosts; B, trypsinized membranes; C, alkali-extracted trypsinized membranes. these peptides, KS-1, KS-2, and KS-4 contained 1 mol of histidine residue/mol of peptide.
Analysis of Peptides Remaining in the Membranes after Tryptic Digestion-Peptides remaining in the membranes after the trypsin treatment were also analyzed. Washed trypsinized membranes were dissolved in an equal volume of 8 M urea and 4% SDS containing 0.05 M dithiothreitol and 10% (v/v) formic acid, and this mixture was applied to a Cosmosil C-8 column. Four major peaks, KM-6-KM-9, were obtained ( Fig. 2B). Following purification and analysis of their primary structures, all of the amino acid sequences were also found to be in complete agreement with the deduced amino acid sequences, and KM-6 contained a histidine residue (Table I).
The 8,500-dalton peptide which comprises the pyridoxal phosphate-binding site of Lys-851, also contains a histidine residue (Kawano et al., 1988). Thus, we were able to isolate five out of six histidine-containing peptides in the transmembranous 55,000-dalton domain of band 3.
Effect of NaOH Concentration on Tryptic Digestion-In the course of these experiments, we realized that the susceptibility of band 3 in situ to trypsin was modified by treatment of the membranes with NaOH. When erythrocyte membranes, which had not been trypsinized at 4 "C to remove the cytosolic 40,000-dalton domain, were treated with 100 mM NaOH and digested with trypsin at 37 "C in the presence of a low salt concentration (see "Experimental Procedures"), many more peptides including KS-1-KS-5 were released from the membranes (Fig. 3A), in contrast to Fig. 2A. On the other hand, when these membranes were treated with 10 instead of 100 mM NaOH, KS-1, KS-2, and KS-4 peptides were not released from the membranes by tryptic digestion, and the amounts of KS-3 and KS-5 released were smaller (Fig. 3B). The primary structure of a peptide appearing at about 25 min of elution time in the HPLC profile was YQSSPAKPDSSFYK corresponding to the deduced sequence associated with Tyr-347 to Lys-360 of the cytosolic 40,000-dalton domain and was designated as peak 25 (Fig. 3A). In Fig. 3, A and B, peak 25 was excised and released to the supernatant equally from both membranes irrespective of alkali concentration. Judging from the SDS-PAGE results, all of the band 3 molecules in the membranes were digested with trypsin (data not shown). We selected peak 25, therefore, as an internal standard and quan- alkali-treated membranes (1.5 mg/ml), from which the cytosolic domain of band 3 had been removed by trypsin treatment a t 4 "C, were suspended in 5P8 and redigested with trypsin (15 pg/ml) a t 37 "C as described under "Experimental Procedures." Peptides released into the supernatant were collected and analyzed with a reversed-phase column (Cosmosil C-18, 4.6 X 250 mm) using a linear gradient of 0-100% acetonitrile containing 0.1% trifluoroacetic acid. 1, KS-1; 2, KS-2; 3, KS-3; 4, KS-4; 5, KS-5. Panel B, after alkali-stripped membranes were redigested with trypsin and washed as described in panel A , the peptides which remained in the membranes were dissolved with an equal volume of 8 M urea and 4% SDS containing 0.05 M dithiothreitol and 10% (v/v) formic acid. After warming a t 60 "C for 5 min, peptides in this solution were separated on a Cosmosil C-8 column (4.6 X 300 mm) using a linear gradient of 40-100% acetonitrile containing 0.1% trifluoroacetic acid. 6, KM-6; 7, KM-7; 8, KM-8; 9, KM-9.
titatively analyzed trypsin susceptibility of band 3 by using peak 25 and KS-4. The amount of KS-4 released into the supernatant increased as the alkali concentration was raised, and all of the band 3 molecules at the KS-4 region were digested with trypsin when the membranes were pretreated with 100 mM NaOH (Fig. 3C). With the KS-1 and KS-2 peptides, only about 30-40% of the band 3 molecules appeared to be digested with trypsin (Fig. 3C).
All peripheral membrane proteins appeared to be extracted when white ghosts were pretreated with 10 mM as well as with 100 mM NaOH-treatment (Fig. 30), indicating that the altered susceptibility to trypsin is not due to steric hindrance caused by peripheral proteins but may be due to conformational changes of band 3. DIDS Stabilization of Band 3 Conformation-When membranes instead of intact erythrocytes were used for the [3H] H2DIDS labeling experiment, spectrin and band 3 in the membrane proteins were labeled with [3H]H2DIDS. This is consistent with the previous observation of Cabantchik and Rothstein (1974) that a part of DIDS binding is not relevant to the anion transport process. However, the binding of [3H] H2DIDS to band 3 was completely inhibited in membranes prepared from DIDS-pretreated erythrocytes (data not shown), indicating that the nature of [3H]H2DIDS binding to band 3 is not different from the case of labeling intact cells with ['HH]H2DIDS. Therefore, we used erythrocyte membranes for the DIDS binding experiments instead of intact cells.
Cell membranes were pretreated with 100 /IM DIDS a t 37 "C for 1 h to convert all band 3 molecules in the membranes to the DIDS-bound form. After washing out free DIDS, the DIDS-pretreated membranes were exposed to various concentrations of NaOH and digested with trypsin. As shown in Fig.  4B, the release of KS-1-KS-5 from the membranes was strongly inhibited by pretreating membranes with DIDS, suggesting that portions of the KS-1-KS-5 peptides in the DIDSband 3 complex were not exposed to the membrane surface even when the membranes were treated with NaOH. Peptides released into the supernatant by digesting with trypsin were quantitatively analyzed based on comparisons with peak 25 as described above. Stabilization of the band 3 conformation in the outward facing form with DIDS caused band 3 to become more resistant to alkali treatment (Fig. 3C).
Effects of NaOH Concentrations on Dihydro-DIDS Binding to Band 3-Alkali treatment of membranes also affected the binding of H,DIDS to band 3. The [3H]H2DIDS binding activity of band 3 was reduced to about 50% of control at 100 mM NaOH (Fig. 5) as the susceptibility to trypsin increased (Fig. 3C). This result also suggests that a conformational change of band 3 in situ is induced by alkali treatment. This view is also supported by DIDS cross-linking experiments. When intact erythrocytes are digested with chymotrypsin, band 3 splits into the 60,000-and 38,000-dalton domains. The split domains can be cross-linked by incubating the cells with H2DIDS at pH 9.5 (Jennings and Passow, 1979). Although both DIDS and H2DIDS are specific inhibitors of anion transport mediated by band 3, the cleaved band 3 domains were cross-linked by incubating cells with H2DIDS but not by DIDS. In our study, however, the cross-linking of band 3 with DIDS was also observed by treating chymotrypsinized membranes with ice-cold NaOH. The amount of band 3 crosslinked by DIDS (Fig. 6) as well as the increase in trypsin susceptibility of band 3 (Fig. 3C) were both increased when NaOH concentration was raised from 10 to 100 mM. Thus, the DIDS cross-linking with band 3 increased by alkali treatment (Fig. 6) was inversely proportional to the [3H]H2DIDS binding ability to band 3 (Fig. 5).
Reactivity of Antibody to Alkali-treated Membranes-The conformational change of band 3 was also confirmed in experiments with antibodies against band 3. Site-specific antibodies were developed with the multiple antigen peptides of KS-1 and KS-4 in accordance with the method of Tam et al. (Tam, 1988;Tam and Lu, 1989). The antibodies against KS-1 (IgGKS.l) and KS-4 (IgGKS.J specifically reacted to band 3 after immunoblotting of erythrocyte membranes, and there was no cross-reaction between the two antibodies. The reactivity of the antibodies, IgGKS.I and IgGKs.a, to the membranes increased as NaOH concentration was raised (Figs. 7, A and  B), indicating that the conformation of band 3 in situ is changed by treating the cell membranes with NaOH.

DISCUSSION
Nine peptides derived from the transmembranous 55,000dalton domain of band 3 were purified and sequenced in this study. All of the sequences agreed completely with deduced sequences from cDNA of human erythroid band 3. The cleavage sites involved the amino groups of Gly-361, Tyr-390, Val-604, Leu-632, Gly-647, Gly-699, Ser-731, Ile-761, Tyr-818, and the carboxyl groups of Lys-639, Arg-656, Lys-743, and Lys-826 (Table I) water were mixed with 6 volumes of ice-cold NaOH solution varying in concentration from 10 mM to 100 mM as described under "Experimental Procedures." The alkali-treated membranes were suspended in 5P8 at a protein concentration of 1.5 mg/ml and digested with 15 pg/ml of trypsin at 37 "C for 30 min. Peptides released into the supernatant were collected by centrifugation and analyzed by HPLC as described in Fig. 2.4 proteinase when band 3 was not treated with higher concen-part of the model of human band 3 (Lux et al., 1989). The trations of NaOH (Fig. 3C). The other peptides, KM-6 (Gly-portion of KS-2 from Ser-731 to Lys-743 in the deduced 699-), KM-7 (Tyr-390-), , and KM-9 (Ile-sequence was predicted as a membrane-spanning region by 761-), constituted membrane-spanning helices. These topo-Lux et al. (1989), but this peptide was released from the logical sites corresponded closely with the hydropathy predic-membrane by tryptic digestion, signifying that this peptide is tion of Passow (1986) on murine band 3 but did not fit in a part of a hydrophilic connector loop. It seems that their model (1 mg/ml) were mixed with 6 volumes of ice-cold 40 mM NaOH solution and subjected to trypsin treatment and analyzed by HPLC as described in Fig. 3A. Panel B, white ghosts in 5P8 (1 mg/ml) were incubated with 100 p~ of DIDS a t 37 "C for 1 h as described under "Experimental Procedures." The DIDS-pretreated ghosts were mixed with 6 volumes of ice-cold 40 mM NaOH solution.
After washing, the alkali-treated membranes were digested with 15 pg/ml of trypsin a t 37 "C for 30 min as described above. Peptides released into the supernatant were analyzed by HPLC as described in Fig. 2A. 1, KS-1; 2, KS-2; 3, KS-3; 4, 5, of band 3 should be reviewed with respect to the region around the KS-2 peptide. The sidedness of these hydrophilic connector loops remains to be resolved because we used unsealed membranes in the present investigation. The transport of phosphate and phosphoenolpyruvate across the erythrocyte membrane is mediated by band 3 (Hamasaki et al., 1983;Hamasaki and Kawano, 1987), and transport rates were inhibited when intracellular pH was lowered from pH 6.8 to 6.0, as if inhibition depended on protonation of groups with a pK of approximately 6.6 inside the cell membrane (Matsuyama et al., 1986). Diethyl pyrocar- FIG. 6. Effects of NaOH concentration on the cross-linking of band 3 by DIDS. Erythrocytes were treated with chymotrypsin to cleave band 3 into two fragments of 38,000-and 60,000-dalton domains, and white ghosts were prepared from the digested erythrocytes as described under "Experimental Procedures." The digested ghosts (1 mg/ml) were pretreated with 100 p~ DIDS in 5P8 at 37 "C for 1 h followed by washing to remove free DIDS as outlined under "Experimental Procedures." The DIDS-pretreated ghosts (4 mg/ml) in cold water were suspended in 6 volumes of ice-cold NaOH solution in order to vary its final concentration from 10 to 100 mM. The NaOH-ghosts mixture was immediately centrifuged and the amount of cross-linked band 3 in the ghosts analyzed by SDS-PAGE (9% acrylamide). Chymotryptic membranes were treated with I , 10 mM; 2, 20 mM; 3, 30 mM; 4, 40 mM; 5, 50 mM; 6, 100 mM; and 7, 0 mM NaOH. bonate (DEPC), a histidine-oriented reagent, inhibited phosphate transport across the cell membrane only when membranes were modified with the reagent from the cytosolic surface of membranes . In addition, extracellular DNDS protected the intracellular amino acid from DEPC modification and, reciprocally, DEPC modification inhibited extracellular H,DIDS binding to band 3 (Izuhara et al., 1989). Thus, it appears that the essential histidine residue(s) for anion transport is located on the cytosolic surface of band 3 and that extracellular binding of DNDS to band 3 induces a conformational change in the intracellular portion of band 3 such that the histidine residue(s) is hidden from the cytosolic surface of the cell membrane . In order to identify the essential histidine residue within the band 3 molecule, we attempted to isolate peptides containing histidine residues. According to the cDNA sequence of human erythroid band 3 (Tanner et al., 1988Lux et al., 1989, 6 histidine residues are expected in the 55,000-dalton domain of band 3. In the present paper, we were able to isolate nine peptides from the transmembrane domain of band 3. Among them, four peptides, KS-1, KS-2, KS-4, and KM-6, contained a histidine residue (Table I). Another histidine residue was isolated with the acylated 8,500-dalton peptide which contains Lys-851, the pyridoxal phosphate-binding site (Kawano et al., 1988;Okubo et al., 1991). As a consequence, we have been able to isolate histidine-containing peptides associated with 5 of the 6 histidine residues in the 55,000dalton domain of band 3.
As shown in Fig. 3C, the amount of the KS-1, KS-2, and KS-4 peptide released by trypsin treatment increased as NaOH concentration was raised during alkaline treatment of the erythrocyte membranes. The change in trypsin susceptibility was not due to steric hindrance by peripheral membrane proteins because all of these proteins had been extracted when the membranes were treated with 10 mM NaOH; at this concentration none of the band 3 was digested by trypsin (Fig. 3C). Thus, the change in trypsin susceptibility appears to be due to the conformational change in band 3 induced by alkali treatment. Susceptibility of band 3 to carboxypeptidase Y is altered when cell membranes are treated with alkali (Lieberman et al., 1987), thereby suggesting also that the conformation of band 3 in situ may be changed by alkali treatment of cell membranes.
The DIDS-bound form of band 3 was resistant to conformational change induced by alkali treatment. The release of peptides by trypsin was strongly inhibited by the DIDS pretreatment (Figs. 3C and Fig. 4), indicating that DIDS binding to band 3 stabilizes the band 3 conformation which is resistant to alkali treatment. The DIDS-complexed form of band 3 is resistant to thermal denaturation (Appell and Low, 1982) as well as to papain digestion (Jennings et al., 1984), indicating that DIDS-stabilized band 3 conformation also resists these other influences.
The present paper indicates that the hydrophilic connector loops of band 3 containing histidine residues tend to be resistant to alkali-trypsin treatment when DIDS bound to band 3 and converted the configuration to the outward facing form. One of the mobile histidine residues included in the peptides of KS-1 (His-819), KS-2 (His-734), and KS-4 (His-651) could be the essential histidine residue for anion transport because this residue was hidden from the cytosolic surface of the cell membrane in the outward facing form and was thereby protected from the DEPC modification Hamasaki et al., 1989). However, there was no direct evidence in this paper to indicate that the conformational change observed is related to the anion transport process.