Lysine 539 of human band 3 is not essential for ion transport or inhibition by stilbene disulfonates.

The anion transporter from human red blood cells, band 3, has been expressed in Xenopus laevis frog oocytes microinjected with mRNA prepared from the cDNA clone. About 10% of the protein is present at the plasma membrane as determined by immunoprecipitation of covalently bound 4,4'-diisothiocyano-2,2'-disulfonic acid stilbene (DIDS) with anti-DIDS antibody. The expressed band 3 transport chloride at a rate comparable to that in erythrocytes. Transport of chloride is inhibited by stilbene disulfonates, niflumic acid, and dipyridamole at concentrations similar to those that inhibit transport in red blood cells: DIDS and 4,4'-dinitro-2,2'-stilbene disulfonate inhibit chloride uptake with Kiapp of 34 nM and 2.5 microM, respectively. Lysine 539 has been tentatively identified as the site of stilbene disulfonate binding. Site-directed mutagenesis of this lysine to five different amino acids has no effect on transport. Inhibition by stilbene disulfonates or their covalent binding was not affected when Lys-539 was substituted by Gln, Pro, Leu, or His. However, substitution by Ala resulted in weaker inhibition and covalent binding. These results indicate that lysine 539 is not part of the anion transport site and that it is not essential for stilbene disulfonate binding and inhibition.

type of binding, inhibition by stilbene disulfonates is competitive with substrate, suggesting interaction at or near the transport site. The presence of an isothiocyanate group in DIDS,' H2-DIDS, and SITS, allows covalent reaction with the amino group of a lysine side chain which is accessible only from the outside of the red blood cell membrane and only when the transport site is facing outward. These two pieces of evidence, competitive inhibition and orientation of the transport site for covalent reaction with DIDS, have suggested that this lysine might be part of the anion transport site. However, modification of presumably the same lysine by reductive methylation (B), BSSS (9), or pyridoxal phosphate (10) does not affect the anion transport properties of band 3. Reductive methylation and BSSS prevent covalent but not reversible binding of stilbene disulfonates. Although it remains to be shown whether the same lysine which reacts with DIDS and other stilbene disulfonates is modified in the three cases mentioned above, these results strongly suggest that this lysine may not be directly involved in anion transport. Nevertheless, DIDS and other stilbene disulfonates are also effective inhibitors of a number of other anion transporters, including chloride channels (11). Thus, the structure of the DIDS binding site and by extension that of the transport site could be conserved among different anion transporters. For this reason, DIDS has been used in attempts to identify, purify, and clone putative anion transporters (12)(13)(14). It is thus important to identify the DIDS-binding lysine and to determine its role in anion transport.
Recently, the human (15), murine (16), and chicken (17,18) band 3 cDNAs have been cloned. The deduced amino acid sequences show a very high degree of identity in the membrane-associated domain. The homology is less in the cytoplasmic domain, but regions such as the putative ankyrin binding site are highly conserved in all three proteins (15)(16)(17)(18). Several sites of chemical modification or proteolytic cleavage have been assigned to the primary sequence of HB3 (15). Thus, the site of stilbene disulfonate covalent binding, localized near the carboxyl terminus of the 17-kDa tryptic fragment (see reviews, Refs. [1][2][3][4], has been identified as residues 539 or 542 of human band 3. These 2 lysines are conserved in human and murine band 3, as well as in a related anion transporter cloned from kidney (19). However, only one of these, Lys-539, is present in avian band 3 (17)(18), making it the most likely candidate for the DIDS-binding site.

19607
Previous reports using mouse spleen mRNA (20,21) and our own preliminary studies with a murine band 3 clone (22) have shown that band 3 can be functionally expressed in frog oocytes. Here we show that human band 3 is expressed in oocytes injected with mRNA prepared in vitro and that the properties of the expressed protein are similar to those of native band 3 in red blood cell membranes. Furthermore, we mutated Lys-539 to five different residues and found that anion transport is not affected. The reversible and irreversible binding of DIDS was not affected when Lys-539 was substituted by Ala, Glu, Leu, or Pro, although when Lys was substituted by Ala, the inhibitory potency of stilbene disulfonates was markedly decreased. We conclude that Lys-539 is not an essential part of the anion transport site and that it is not the primary site of stilbene disulfonate binding.

EXPERIMENTAL PROCEDURES
Materials-cDNA clones encoding the full coding sequence of human band 3 were isolated by S. Lux in our laboratory: the sequence is identical to that reported by Tanner et al. (15). The following inhibitors of band 3 transport were used DIDS (Pierce Chemical Co.); Hz-DIDS, and SITS (Molecular Probes); DNDS (Pfaltz and Bauer); niflumic acid and dipyridamole (Sigma). Radioisotopes were from Amersham Corp. All reagents were of the highest grade available.
In Vitro Transcription-A 3.4-kilobase band 3 clone (pHB3) comprising the full length coding sequence plus 5'-and 3'-noncoding sequences of 30 and 630 base pairs, respectively, was subcloned into the AccI and Sac1 sites of Bluescribe (Stratagene). It was linearized with Hind111 and mRNA was prepared according to the manufacturers instructions, using T7 RNA polymerase (Stratagene) and 0.5 mM 5'-GpppG (Pharmacia LKB Biotechnology Inc.) as the mRNA 5'cap (23). After 2 h at 37 "C, the reaction was stopped by the addition of RNase-free DNase (Cappel) for 10 min at 37 "C. The mRNA was purified free of unincorporated nucleotides by centrifugation through a 1-ml Sephadex G-50 column equilibrated with water. The mRNA (200 ~1 ) was concentrated by centrifugation for 15 min in a Centricon-30 concentrator (Amicon), separated into 10-pl aliquots, and stored frozen at -70 "C.
Site-directed Mutagenesis-The internal 1.6-kilobase KpnI fragment of HB3 was subcloned into Bluescribe. From this, a 300-base pair XbaI fragment (residues 1561-1855) was digested and subcloned into the XbaI site of M13mp18. The following mutagenic 20-mer oligonucleotide was synthesized (Research Genetics): GATCTTGAT-CAG(G,T)(A,C,T,G)(A,C,G)GGAGA. It contains multiple substitutions at the site of Lys-539, allowing for 14 different codons. This sequence was chosen in order to minimize the possibility of obtaining wild type HB3. For the mutagenesis, the procedure of Kunkel(24) as described in the Bio-Rad mutagenesis kit was followed. Approximately 600 clones were obtained, of which 24 were sequenced by the dideoxy chain termination method (25) as modified (26). Of these, 5 were wild type HB3 and the rest corresponded to 10 different mutations. The mutant XbaI fragments were subcloned back into pHB3 and mRNA was prepared as described above.
Antibodies-An antibody against the cytoplasmic, amino-terminal domain of band 3 was a generous gift of Dr. P. Low, Purdue University.
To raise antibodies against DIDS, 5 mg of KLH in 150 mM NaCl, 10 mM phosphate buffer, pH 7.0, was reacted with 2.4 mM DIDS at 37 "C for 40 min. The sample was centrifuged for 30 s in an Eppendorf centrifuge to remove large aggregates, dialyzed overnight against phosphate-buffered saline at 4 "C, and stored frozen at -20 "C. The resulting complex had a molar ratio of 30 DIDS per KLH. Rabbits were immunized with 0.4 mg of KLH-DIDS in Freund's complete adjuvant, followed by two boosts in Freund's incomplete adjuvant. After the second boost, the rabbits were rested for 3 months, boosted again with 0.1 mg of KLH-DIDS in phosphate-buffered saline directly in the ear vein, three times at 3-4-day intervals. Blood was collected 5 days after the last boost and then four more times at 3-day intervals.
Microinjection of Xenopus laeuis Oocytes-Mature X. laevis were obtained from Nasco (Fort Atkinson, WI). Pieces of ovary were surgically removed and incubated in 10 mg/ml collagenase (Sigma, type 1.4) in ND-96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaC12, 1 mM MgC12, 5 mM HEPES, pH 7.6) for 1 h at room temperature. Individual stage V and VI oocytes were dissected without removing the remain-ing follicle cells and were injected with 50 nl of mRNA (200-300 ng/ pl). The microinjected oocytes were incubated in ND-96 at room temperature for the times indicated in the text. The remaining follicle cells fell off by themselves after 2 or 3 days.
Chloride Transport-At least 2 days after microinjection, groups of 10 oocytes were placed in disposable culture tubes and incubated at 20 "C in 0.15 ml of ND-96 containing 5-8 pCi/ml 36Cl and inhibitors as indicated in the figure legends. Alternatively, prior to the addition of radioactive medium, oocytes were incubated in ND-96 containing 10 or 100 p~ DIDS for 1 h at 20 "C. The DIDS-containing medium was removed, the oocytes washed two times with ND-96, followed by three washes in ND-96 plus 0.5% bovine serum albumin and two more washes in ND-96. These oocytes were assayed for chloride uptake within 1 b. At the times indicated, the radioactive medium was removed and the oocytes were quickly washed in about 30 ml of 96 mM sodium gluconate, 2 mM potassium gluconate, 1 mM MgSO,, and 5 mM HEPES, pH 7.6. They were placed in individual scintillation vials and 0.1 ml of 1% SDS was added, followed by vortexing to release the radioactivity. The radioactivity was determined by scintillation counting in a Packard p counter with ACS (Amersham Corp.) as scintillant. Chloride uptake is expressed as nanomoles of chloride/ oocyte & S.E.
Labeling and Immunoprecipitation-Immediately after microinjection, groups of 20 oocytes were incubated for 16 h in ND-96 containing 1-1.5 mCi/ml ["Slmethionine (Amersham Corp.). If the oocytes were labeled longer than 16 h, the medium contained, in addition, 30 p~ unlabeled methionine to ensure linear incorporation of radioisotope (27). After removing the labeling medium, membranes were prepared according to Colman (27). Briefly, oocytes were homogenized in 0.5 ml of 10% (w/v) sucrose in 150 mM NaCl, 10 mM magnesium acetate, and 20 mM Tris.HC1, pH 7.6, containing 0.1 mM phenylmethylsulfonyl fluoride. The homogenate was layered on top of a discontinuous sucrose gradient (0.4 ml each of 50 and 20% (w/v) sucrose in 150 mM NaCl, 10 mM magnesium acetate, 20 mM Tris.HC1, pH 7.6) and centrifuged for 30 min at 15,000 X g in a Beckman TLS.55 rotor. Membranes were removed from the 20-50% sucrose interface, washed once in homogenization buffer without sucrose, and resuspended in 0.5 ml of 0.5 M NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 10 mM Tris. C1, pH 7.4 (IPP buffer). The samples were precleared by addition of 10 pl of preimmune serum (30 min at room temperature) followed by the addition of 50 pl of Affi-Gelprotein A (Bio-Rad) for 30 min. After removal of the Affi-Gel-protein A by centrifugation, 5-10 pl of immune serum were added and the samples were incubated overnight at 4 "C or for 2 h at room temperature. The antigen-antibody complex was removed by addition of Affi-Gel-protein A as above, washed with IPP buffer, and dissociated in Laemmli gel loading buffer (28). After removing the Affi-Gelprotein A beads, the sample was analyzed by gel electrophoresis through 8% polyacrylamide (28) followed by autoradiography.
To identify surface molecules, [35S]methionine-labeled oocytes were reacted with 1 mM DIDS in ND-96 for 1 h at 20 "C in the dark. Unreacted DIDS was removed by three or four washes in a large volume of ND-96. Oocyte membranes were prepared and immunoprecipitated as above, except that anti-band 3 antibody covalently coupled to protein A-Sepharose (29) was used. The immune complex was dissociated in 2% sodium dodecyl sulfate for 5 min at 45 'C. The supernatant containing HB3 was removed, and the SDS concentration was adjusted to 0.1% by the addition of IPP buffer and then subjected to two sequential precipitations: anti-DIDS antibody followed, after removal of the anti-DIDS. HB3 complex, by anti-band 3 antibody. The two antigen-antibody complexes were analyzed as above.

RESULTS
Expression of Human Band 3-Total membranes prepared from oocytes microinjected with water (control) or mRNA prepared in vitro from the pHB3 clone coding for human band 3 (HB3) were immunoprecipitated with an antibody against the cytoplasmic, amino-terminal domain of HB3. After autoradiography of proteins separated by gel electrophoresis, a major protein of molecular mass 90-95 kDa was observed in mRNA-injected oocytes (Fig. 1, lane 2) that was absent in control oocytes (lane 1 ). A large number of nonspecific polypeptides are present in both control and band 3 injected oocytes; the background varies from experiment to experi- Each lane corresponds to 20 microinjected oocytes, labeled with radioactive methionine for 16 h. ment and seems to reflect individual frog differences (see Fig.  4). However, the 90-95-kDa polypeptide immunoprecipitated by the anti-band 3 antibody is always the predominant species in HB3-injected oocytes.
In order to determine the relative amount of newly synthesized band 3 present at the plasma membrane, intact oocytes were covalently labeled with DIDS. After washing the oocytes to remove unreacted DIDS, band 3 was isolated by immunoprecipitation of oocyte membranes. This material was reimmunoprecipitated with antibodies against DIDS and band 3 as described under "Experimental Procedures." About 10% of band 3 was precipitated by anti-DIDS antibody (Fig. 1,  lane 3 ) as compared to anti-HB3 antibody (lane 4 ) . DIDS will not react with any internal band 3 molecules: free DIDS is quantitatively removed during the washes and membrane preparation. Also, in several experiments using mRNA encoding a mutant murine band 3 clone which remains in internal oocyte membranes and does not appear at the plasma membrane, no protein was immunoprecipitated by anti-DIDS antibody (data not shown). Assuming that the reaction with DIDS was complete (as the labeling was done under saturating concentration of DIDS) and that the immunoprecipitation was quantitative (adding more anti-DIDS or anti-HB3 antibody did not increase the amount of HB3 immunoprecipitated), this experiment indicates that 10% of band 3 is present at the plasma membrane. This figure probably represents a lower estimate of the surface molecules. But the most important point shown by this experiment is that HB3 expressed in frog oocytes can be covalently labeled by DIDS. This property of band 3 has been used to identify the anion transport protein in red blood cells (7), and it has recently been used to identify, purify, and clone other putative anion transporters (12)(13)(14).
Properties of Human Band 3 in Frog Oocytes-In red blood cells, the physiological role of band 3 is to catalyze a one-forone exchange of chloride for bicarbonate. However, a great deal of kinetic information has been obtained by studying chloride self exchange (for reviews, see Refs. [1][2][3]. In order to study anion transport properties of human band 3 expressed in frog oocytes, we measured uptake of 36Cl under steady state conditions, i.e. oocytes in ND-96 at room temperature (Fig.  2). Control, mock-injected oocytes show two phases of chloride accumulation: a rapid uptake during the first half-hour (our first time point), and a slower accumulation which continues for up to 8 h at a constant rate of 0.1-0.3 nmol of Cl-/oocyte/ h. In contrast, the amount of chloride taken up by band 3injected oocytes increases rapidly for up to 4 h and then continues at the same rate as control oocytes. During the initial phase of transport, the total amount of chloride taken up by mRNA-injected oocytes is about 5 times higher than the control. In most experiments, the rate of band 3-mediated chloride uptake is constant for the first 1 or 2 h. However, in some experiments (like that shown here) uptake can be linear for up to 4 h. Both the rate and total amount of chloride accumulated are variable, and this seems to be related to individual frog differences, seasonal variations, and other factors out of our control. In all subsequent experiments the rate of chloride uptake was determined within the first hour.
As indicated, the initial rate of band 3-mediated chloride accumulation varied among experiments, ranging from 0.5 to 2.5 nmol of chloride/oocyte/h and occasionally higher. However, in most experiments the rate of transport was between 1 and 1.5 nmol of Cl-/oocyte/h. For purposes of calculation, we will use 1 nmol of Cl-/oocyte/h (1.7 X 10" ions/s/oocyte) as the average rate. To compare this rate to the rate of chloride self-exchange by HB3 in erythrocyte membranes, we determined the number of surface band 3 molecules per oocyte. The amount of band 3 immunoprecipitated by the specific antibody was determined by cutting the slice of gel after autoradiography and measuring its radioactivity by scintillation counting. The specific activity of methionine was calculated from the total radioactivity of the oocyte homogenate and a methionine pool size of 30 pmol/oocyte (27). In one particular experiment, the methionine specific activity was 3.2 x lo3 cpm/pmol and the amount of HB3 immunoprecipitated from 10 oocytes was 300 cpm. Since each band 3 mole-0.0 I . cule has 24 methionines, we calculated that 2.3 X lo8 molecules of band 3 were synthesized per oocyte. Assuming that 10% is at the plasma membrane (Fig. 1, lanes 3 and 4, we estimated a turnover number of 7.4 X lo3 Cl-/s for band 3mediated anion transport in frog oocytes. From Brahm (30), the rate of chloride transport for HB3 in red blood cells in 150 mM KC1 and 20 "c is 7.7 X lo3 ions/s and at 38 "C is 4.8 x lo4 ions/s. Thus, although our calculation is approximate, we conclude that HB3 in the plasma membrane of an oocyte transport chloride at a rate similar to that in red blood cell membranes.
Effect of Inhibitors of Transport-To analyze in more detail the properties of HB3 expressed in frog oocytes, different inhibitors of chloride transport were studied and their effects compared to those on HB3 in erythrocytes. The inhibitors used can be divided in two groups: competitive and noncompetitive. Among the competitive inhibitors, DIDS (7), Hz-DIDS (31), and SITS (6) are stilbene disulfonates which covalently react with band 3, although covalent reaction is not necessary for inhibition. Another stilbene &sulfonate, DNDS (32), does not react covalently. Two other inhibitors, dipyridamole (33) and niflumic acid (34), are reversible inhibitors whose effect does not depend on chloride concentration. Thus, these probably do not bind to the transport site and, therefore, have a different mechanism of inhibition from that of stilbene disulfonates.
Results of two experiments are summarized in Table I. The concentration of each inhibitor used corresponds to Epp values determined for erythrocytes (1). At the concentrations used, none of the inhibitors had a significant effect on the basal anion transport by the oocytes, indicating that the endogenous frog oocyte anion transport systems have properties different from those of band 3. In contrast, all six drugs inhibited band 3-mediated chloride transport. At the concentrations tested, DIDS, Hz-DIDS, DNDS, and dipyridamole reduced the initial rate of chloride uptake by approximately 50%. SITS inhibits oocyte-expressed HB3 more than expected from the Kpp determined in red blood cells. However, available data for SITS inhibition refer exclusively to inhibition of transport by covalently bound SITS (6). The covalent reaction with SITS is slower and less efficient than that of DIDS (6). Clearly, inhibition of transport occurs at a concentration lower than that needed for covalent reaction. Our results show that SITS inhibits chloride accumulation by HB3 in frog oocytes with an estimated Kpp of 100 nM (data not shown). Niflumic acid inhibition is less than expected from the Rpp measured on red blood cells (34). However, the reported epp was determined for oxalate self-exchange (34).
Oxalate seems to be transported by band 3, but it is not clear whether the mechanism of oxalate transport is comparable to that of chloride. Nevertheless, our data indicate that niflumic acid inhibits band 3-mediated chloride transport, although with an estimated Kpp of 4 p M at 20 "C.
We studied in detail the inhibition of chloride transport by two stilbene disulfonates, DIDS, which reacts covalently with band 3 in red blood cells (7), and DNDS, a reversible inhibitor (32). Fig. 3 shows that in agreement with previous data (20), uptake of chloride by control oocytes is not affected by concentrations of DIDS up to 1 WM (Fig. 3, open circles). In contrast, increasing concentrations of DIDS decreases the initial rate of chloride uptake by mRNA-injected oocytes to the same value as control oocytes: 1 p~ DIDS inhibits 99% of the band %mediated chloride transport. A plot of the reciprocal of the inhibitor concentration versus the reciprocal of the fractional inhibition generates a straight line (inset, Fig. 3), indicating that under the conditions of our experiments DIDS acts at a single site (1,31). Expression and Properties of Lys-539 Mutants- Fig. 4 shows that oocytes injected with mRNA coding for HB3 with five different substitutions of Lys-539 ( K / A , K/H, KIL, KIP, and KIQ lanes), make comparable amounts of a polypeptide that is indistinguishable from wild type HB3 (HB3 lane), both in terms of its mobility as well as its reactivity toward the anti-band 3 antibody. These mutants are all capable of transporting chloride at a rate similar to wild type band 3, i.e. 2.8-3.8 nmol of Cl-/oocyte/h (Fig. 5, -DIDS). In a different experiment, the rate of transport was normalized for the amount of protein made: all the mutants show the same rate of chloride accumulation as wild type (data not shown).
In order to characterize these mutants further, the effect of inhibitors of transport was examined. Fig. 5 (+DIDS), shows the effect of 100 nM DIDS on the initial rate of chloride uptake. As determined from the Kpp this concentration of DIDS was expected to inhibit the rate of transport about 70%.     The accumulation of chloride by wild type HB3 as well as by the K539Q, K539L and K539P mutants was inhibited 90%, higher than expected but within experimental variation. In contrast, chloride transport by the K539A mutant was inhib- ited only 30%. Table I1 shows a summary of two experiments (different from the one shown in Fig. 5) in which several inhibitors were tested. Despite some variability in the absolute extent of inhibition of chloride transport, K539A mutant consistently shows less inhibition of chloride uptake by all these stilbene disulfonates, while the K539Q mutant is inhibited to the same extent as is the wild type. However, the noncompetitive inhibitor dipyridamole inhibits anion transport by all the mutants to the same extent as wild type band 3 (Table 11).
Further characterization of ion transport by K539A generates a Kpp of 42.5 nM for DIDS inhibition and 6.1 p M for DNDS (Fig. 6, A and B ) . In contrast, in this particular experiment wild type HB3 is inhibited with a Kpp of 18.4 nM for DIDS and 2.5 p~ for DNDS (Fig. 6, A and B ) . Thus, the K539A mutant is inhibited by these stilbene disulfonates 2fold less than the wild type, regardless of whether the inhibitor binds reversibly or irreversibly.
Since only reversible binding of stilbene disulfonates is necessary for inhibition of anion transport, we needed to  determine whether the mutants are capable of binding DIDS covalently. To this end, 2 days after injection the oocytes were treated with 10 or 100 p~ DIDS for 1 h at 20 "C, conditions that irreversibly inhibit anion exchange by erythrocytes. After thoroughly washing to remove unreacted DIDS, the oocytes were assayed for their ability to accumulate 36Cl- (Table 111). Uptake of chloride was almost completely inhibited in oocytes injected with wild type HB3 or the K539Q mutant, indicating that even after washing the oocytes three times with a solution containing 0.5% bovine serum albumin, DIDS remains bound, presumably covalently. In contrast, K539A is inhibited much less, probably reflecting a lower binding affinity for DIDS. It remains to be shown whether longer incubations or higher DIDS concentration would promote complete inhibition by DIDS.

DISCUSSION
The work presented here clearly shows that frog oocytes injected with in uitro prepared human band 3 mRNA synthesize a functional protein which is recognized by anti-band 3 antibodies. All properties of the protein we tested, its rate of chloride transport as well as the effect of several transport inhibitors, are similar to those of band 3 in erythrocyte membranes. In particular, the ability of oocyte-synthesized band 3 to react covalently with DIDS and most likely with other stilbene disulfonates indicates that it is in a native conformation. Thus, within its limits and despite its variability, frog oocytes seem to be an appropriate system in which to test the effect of mutations on the transport properties of band 3.
Our aim was to identify unequivocally the putative DIDSbinding site and to determine its role in chloride transport. Chemical modification data together with knowledge of the primary sequence have localized this site to lysines 539 or 542 of the human band 3 sequence. However, since only Lys-539 is found in all three band 3 proteins cloned so far, it was considered the most likely candidate for being the DIDSbinding site. Our site-directed mutagenesis strategy provided several mutants from which five were randomly chosen and tested. The rate of chloride accumulation was not affected, regardless of whether Lys-539 was substituted for a smaller, uncharged residue (Ala or Leu) or a bulkier one (His, Pro, Glu) (Fig. 5). This indicates clearly that Lys-539 is not part of the anion transport site or at least that it does not participate in a limiting step in the transport of anions, corroborating previous information (8-10).
The results of inhibition by stilbene disulfonates were more surprising. For the K539Q, K539L, and K539P mutants, inhibition of chloride transport by the reversible inhibitor DNDS or by the irreversible inhibitors DIDS and SITS was not affected (Fig. 5 and Table 11). Furthermore, inhibition by DIDS was accompanied by irreversible binding (Table 111), suggesting that Lys-539 is not important for binding or inhibition of anion transport by stilbene disulfonates. These results per se are not surprising, in view of previous reports on chemical modification (8, 9). However, substituting Lys-539 by Ala resulted in a %fold weaker inhibition by DIDS or DNDS (Fig. 6), an effect which was also reflected in a poorer capacity to bind DIDS covalently (Table 111). These results might be explained by a conformational change induced by the Lys to Ala substitution. As a test of the overall conformation of the binding site, we used a different transport inhibitor, dipyridamole. The binding site for dipyridamole is different from the DIDS-binding site, but it seems to overlap with it (1). Our results show that both K539Q and K539A respond normally to inhibition by dipyridamole (Table 11). Thus, if the observed decrease in DIDS inhibition in K539A is due to a conformational change, this must be small since it does not affect binding of dipyridamole.
The above results can be interpreted in either of two ways: ( a ) Lys-539 is not the DIDS-binding lysine, in which case Lys-542 is the most likely candidate for it; or ( b ) DIDS can bind to either Lys-539 or Lys-542. The lack of effect on stilbene disulfonate inhibition observed with mutants K539Q, K539L, and K539P favors the first possibility, although if both lysines are equally reactive, DIDS could randomly bind to either Lys-539 or Lys-542. The fact that the K539A mutant shows an effect on stilbene disulfonate inhibition indicates that Lys-539 is near the DIDS-binding site. It is possible that the presence of Ala instead of Lys induces a conformational change which renders the DIDS-binding lysine less accessible. In fact, according to the Chou-Fasman prediction program (35), the stretch of 10 amino acids around Lys-539 in HB3 as well as in four of the five mutants has a higher probability of being in a @-sheet conformation. The only exception is the K539A substitution, for which the probability of the same stretch of folding in an a-helix configuration is higher. It is possible that in an a-helix conformation the binding of DIDS to a lysine residue such as Lys-542 is constrained. However, since there is very little information about the secondary and tertiary structure of band 3, such prediction values are speculative.
In summary, our results indicate that Lys-539 has no specific role in anion transport. According to secondary structure prediction models, Lys-539 is in an external loop between transmembrane helices 5 and 6. There is a cluster of about 20 positive charges lining the mouth of the transport pathway on both sides of the membrane. These charges are part of the loops between transmembrane segments and their role seems to be to attract anions to the transport site. Thus, the role of Lys-539 in transport could be mainly to contribute a positive charge. Also, Lys-539 does not seem to be the primary DIDSbinding residue. However, it must be kept in mind that the DIDS-binding site is localized in an area of the protein which, after inward translocation of an anion, becomes inaccessible to external DIDS. Lys-539 may be located in a flexible structure, possibly &sheet, which can accomodate a small conformational change, either a rotation or a packing of the peptide chain. This conformational change could result in the inaccessibility of a lysine side chain for binding to stilbene disulfonates. Thus, it will be important to mutate Lys-542 alone and in combination with Lys-539 and study the resultant transport and inhibition properties.