Localization of the Protein 4.1-binding Site on the Cytoplasmic

5 downloads 0 Views 4MB Size Report
May 15, 2016 - Antibodies-Antibodies to the cytoplasmic domain of band 3 were ... reduced nonspecific binding and improved the precision of the assays.
THEJOURNALOF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 267, No. 14,Issue of May 15,pp. 9540-9546,1992 Printed in U.S.A.

Localization of the Protein4.1-binding Site on the Cytoplasmic Domain of Erythrocyte Membrane Band 3” (Received for publication, October 28, 1991)

Christian R. Lombardol, Barry M. Willardson, and PhilipS. Low5 From the Department of Chemistry, Purdue University, West Lafayette, Indiana 47907

Of the several proteins that bind along the cytoplasmic domain of erythrocyte membrane band 3,only the sites of interaction of proteins 4.1 and4.2 remain to be at least partially localized. Using five independent techniques, we have undertaken tomap and characterize the binding site of band 4.1 on band 3. First, transfer of a radioactive cross-linker(1261-2-(p-azidosalicylamido)ethyl-1-3-dithiopropionate) from purified band 4.1 to its binding sites on stripped inside-out erythrocyte membrane vesicles (stripped IOVs) revealed major labeling of band 3, glycophorin C, and glycophorin A. Proteolytic mapping of the stripped IOVs then demonstrated that the label on band 3 was confined largely to a fragment comprising residues 1201. Second, competitive binding experiments with Fab fragments of monoclonal and peptide-specific polyclonal antibodies to numerous epitopes along the cytoplasmic domain of band 3 displayed stoichiometric competition only with Fabs to epitopes between residues 1 and 91 of band 3. Weak competition was also observed with Fabs to a sequence of the cytoplasmic domain directly adjacent to the membrane-spanning domain, but only at 50-100-fold excess of Fab. Third, band 4.1 protected band 3 from chymotryptic hydrolysis at tyrosine 46 and to amuch lesser extent at a site withinthe junctional peptide connecting the membrane-spanning and cytoplasmic domains of band 3. Fourth, ankyrin, which has been previously shown to hinge interact with band 3 both near a putative central and at the N terminus competed with band 4.1 for band 3 in stripped IOVs. Since band 4.1 does not associate with band 3 near the flexible central hinge, the competition withankyrin can be assumed to derive from a mutual association withthe N terminus. Finally, a synthetic peptide corresponding to residues 1-15 of band 3 was found to mildly inhibit band 4.1 binding to stripped IOVs. Taken together,these data suggest that band 4.1 binds band 3 predominantly near the N terminus, with apossible secondary site near the junction of the cytoplasmic domain and themembrane.

The lipid bilayer of the erythrocyte membraneis connected to the membrane skeleton by two distinct proteins. Ankyrin, present at-100,000 copies/cell, links the cytoplasmic domain of the anion transporter, band 3, to the p strand of spectrin

(Bennett andStenbuck, 1980; Bennett, 1990; Coleman, et al., 1989). Defects in this junction, whether natural or manually induced, lead to abnormalities of cell shape, deformability, or fragility (Liu et al., 1990; Kay et al., 1988; Conboyet al., 1986; Coetzer et al., 1988; Low et al., 1991; Jinbu et al., 1984). Band 4.1, a protein present at 200,000 copies/cell, links the spectrinactin junctional complex to the cytoplasmic tail of a glycophorin, probably glycophorin C (Mueller and Morrison, 1981; Reid et al., 1990; reviewed in Bennett, 1989). When either band 4.1 or glycophorin C is missing or altered,a fragile elliptocytic cell may result (Conboy et al., 1986; Anstee et al., 1984; Alloisio et al., 1985; McGuire et al., 1988; Reid et al., 1987). Although all copies of ankyrin could conceivably participate in a band 3-spectrinlinkage, not all members of the band 4.1 population can connect a junctional complex to glycophorin C, even if complete saturation of sites were desirable. Thus, there are only -IO5 junctional complexes/cell and even fewer ( i e . 50,000) copies of glycophorin C (Bennett, 1989). The -150,000 molecules of band 4.l/cell not attached to glycophorin C are probably not associated with unanchored junctional complexes, since complete extraction of the spectrinactin complexes from the membrane removes very little of the protein 4.1. Thus, most of band 4.1 may associate with the membrane in vivo at a cytoplasmic site on glycophorin A or band 3. While the former interaction may be expressed only under stimulated conditions (Anderson and Marchesi, 1985; Chasis et al., 1985, 1988), thelatter interaction is potentially available under all conditions, except when specifically inhibited by protein kinase C (Danilov et al., 1990). In view of the relatively high affinity of band 4.1 for band 3 (Pasternack et al., 1985; Danilov et al., 1990), it wouldbe surprising if a substantial fraction of band 4.1were not associated with band 3 in vivo. Evaluations of the morphology of the cytoplasmic domain of band 3 indicate that itis an elongated molecule with distinct regions of concentrated charge and hydrophobicity (Wein1986). In stein et al., 1978;Appell and Low,1981;Low, addition to providing an anchoringsite for band 4.1 and ankyrin, cdb3l also links band 4.2 (Korsgren and Cohen, 1988), several glycolytic enzymes (Murthy et al., 1981; Tsai et al., 1982; Jenkins et al., 1984), hemoglobin (Walder et al., 1984), and hemichromes (Waugh and Low, 1985) to themembrane. While the binding sites of the enzymes, hemoglobin, hemichromes, and ankyrinon cdb3 have been at least partially

* This work was supported by National Institutes of Health Grant GM 24417. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with18 U.S.C. Section 1734 solely to indicate this fact. $ Recipient of a Herman and Margaret Sokol Graduate Fellowship. § T o whom correspondence and reprint requests should be addressed.

The abbreviationsused are: cdb3, the cytoplasmic domain of band 3; ASD, 2-(p-azidosalicylamido)ethyl-l-3-dithiopropionate;SASD, sulfosuccinimidyl-2-(p-azidosalicylamido)ethyl-l-3-dithiopropionate; DFP, diisopropylfluorophosphate; NTCB, 2-nitro-5-thiocyanobenzoate, SDS, sodiumdodecyl sulfate; IOVs, inside outvesicles; APT, azophenylthioether; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; PMSF, phenylmethylsulfonyl fluoride; BSA, bovine serum albumin; DTT, ditbiothreitol.

9540

Localization of Protein the characterized(Low,1986;Waugh et al., 1987; Davis et al., 1989; Willardson et al., 1989), the sites on band 3 for bands 4.1 and 4.2 have not been examined. The objective of this study has beento locate the binding site of band 4.1 on band 3 and to determinewhether it overlaps with sites of other skeletal proteins. By five independent techniques, wedemonstrate that the predominant band 4.1 site on band 3is near the N terminus of cdb3; however, some evidence for binding near the extreme C terminus is also obtained. As predicted, the N-terminal location was found to lead to a competition with ankyrinforband3. Portions of this workhave been presented previously in abstract form (Lombard0 et al., 1990). EXPERIMENTALPROCEDURES

Materials-Human blood waspurchased from the Central Indiana Regional Blood Center and used within 1 week of the drawing date. DFP, bovine serum albumin, and DEAE-Sepharose were obtained from Sigma. DEAE-Bio-Gel, Affi-Gels 10 and 15, Tween-20, and electrophoresis reagents were obtained from Bio-Rad. Other reagents and their suppliers wereAccellQMA anion-exchange resin from Millipore, Inc.; Sephacryl S-300-HR and Sepharose CL-GB from Pharmacia LKB Biotechnology Inc.; PMSF and DTTfrom Research Organics, Inc.; sucrose from Schwarz-Mann; prestained molecular weight markers from Bethesda Research Laboratories; chymotrypsin from Worthington Biochemicak protein A-Sepharose, SASD, and Iodogen from Pierce Chemical Co.; Bolton-Hunter reagent and Na'T from ICN; and TritonX-100 from Boehringer Mannheim. Protein Purification and Handling-Protein 4.1 was purified from Triton X-100 shells dissociated with 2.0 M Tris by sequential chromatography on Sepharose CL-GB and DEAE-Bio-Gel (Ohanian and Gratzer, 1984; Korsgren and Cohen, 1988)or from a 1.0 M KC1 extract of IOVs as described (Tyler et al., 1979), employing an Accell QMA anion-exchange column. Protein 4.1 purified by either procedure migrated in SDS-polyacrylamide gel electrophoresis predominantly as a doublet of78 and 80 kDa, but minor amounts of protein 4.1 isoforms (Tang et al., 1990; Conboy et al., 1991) above and below the main bands were also detectable. These minor isoforms were also recognized by anti-protein 4.1 antibodies. Similar results were obtained with both band 4.1 preparations. Ankyrin was purified from a 1.0 M KC1 extract of Triton X-100 shells using modifications of published procedures (Bennett, 1983; Pinder et al., 1989). Briefly, membrane skeletons were extracted with 1.0 M KC1 (Bennett, 1983). The extract was concentrated by ammonium sulfate precipitation and then fractionated on a 2.8 X 120-cm column of Sephacryl S-300 in 0.6 M NaBr, as described (Pinder et al., 1989). Ankyrin (band 2.1) eluted in the first half of the second peak. The cytoplasmic domain of band 3 (cdb3) was isolated from a chymotryptic digest of acetic acid-stripped IOVs by DEAE-Sepharose chromatography (Bennett, 1983). All proteinpreparations were treated with 10 mM DFP prior to storage on ice in the presence of 20 mM DTT. Ankyrin was stored as above in the Sepharose column buffer supplemented with 10% sucrose and was used within 2 weeks of its purification. KI strippedIOVs (Bennett, 1983)or pH 11stripped IOVs (Danilov et al., 1990) were prepared by standard methods. Differences in the binding of protein 4.1 to either vesicle preparation were not discernible. Bolton-Hunter reagent was used to radioiodinate protein 4.1 for membrane binding studies (Bennett, 1983). Specific activities ranged from 3 to 6 X lo6 cpmlpg. Radiolabeling was performed in 100 mM HEPES, pH 7.5. Prior to use, proteins, peptides, and vesicles were didyzed against buffer A (130 mM KC1, 20 mM NaCI, 10 mM HEPES, 1 mM EDTA, 0.5 mM DTT, 1 mM NaN3, pH 7.5). Antibodies-Antibodies to the cytoplasmic domain of band 3 were epitope mapped and characterized in aprevious publication (Willardson et al., 1989). Fab fragments were prepared from IgGs using immobilized papain (PierceChemical Co.) and thenpurified utilizing either protein A-Sepharose (rabbit IgGs) or DEAE-Bio-Gel A (mouse monoclonal IgGs) as described (Willardson et al., 1989). Binding Assays-Competitive inhibition of protein 4.1 binding to membranes by Fabs, ankyrin, and synthetic peptides was examined in buffer Acontaining 0.5 mM DFP, as previously described for ankyrin (Willardson et al., 1989). Fab fragments were incubated at room temperature with membrane vesicles for 3 h prior to band 4.1 addition, while ankyrin and synthetic peptides were incubated for 90 min prior to addition. BSA was added to maintain the total protein

4.1-binding Site

on Band 3

9541

concentration at 6.5 mg/ml. lz51-Labeledband 4.1 (5 pg/ml)was incubated with the stripped IOVs (35 pg/ml) for 20 min, after which the bound and free protein 4.1 were separated by pelleting the IOVs through buffer A containing20% sucrose and 2 mg/ml BSA. The tips of the microfuge tubes were frozen in liquid nitrogen, cut off, and counted for radioactivity in the gamma counter. Assays of cdb3 peptide and ankyrin inhibition were performed similarly except that all assay and centrifuge tubes were coated with 10 mg/ml BSA in buffer A for 20 min and thenrinsed with buffer A lacking BSA prior to use (Joy1 and Purich, 1990). This coating procedure significantly reduced nonspecific binding and improved the precision of the assays. Nonspecific binding was assayed using heat denatured (60 "C, 20 min) 'Z51-protein4.1 and the values were routinely subtracted. These values ranged from 3 to 10% of the total binding in the absence of any competing Fab, peptide, or ankyrin. Binding assays were always performed in duplicate or triplicate. Radiolabel Transfer of 125Z-ASD-Protein4.1 to Membranes-Protein 4.1 was derivatized with '*'I-SASD, a radioiodinatable, photoactivatable, and cleavable cross-linker using modifications of published protocols (Shepard et al., 1988; Petrucci and Morrow, 1991). Prior to photoactivation, all procedures were performed in the dark with the aid of a red safety light. SASD (5 mg) was dissolved in 200 pl of dry dimethyl sulfoxide and stored at -20 "C. Ten pl of this SASD stock was added to 170 p1 of 100 mM HEPES, pH 7.5, containing 20 p1 of Na"'I (100 pCi/pl, 2 mCi total), and thesolution was transferred to a tube containing 2.0 mg of Iodogen previously dried under Nz from chloroform. Iodination of SASD was allowed to continue for 4 min and then terminated by removal of the solution from the Iodogencoated tube. '"I-SASD was then reacted with protein 4.1 at a molar ratio of 6.67 to 1 in 100 mM NaC1,20 mM HEPES, 1 mM EDTA, 1 mM NaN3, 10% sucrose, pH 8.0 for 30 min at room temperature. Preliminary experiments showed that one molecule of '"I-SASD or less was incorporated/molecule of protein 4.1 under these conditions, assuming 25% incorporation of lZ5Iinto SASD (Shepard etal., 1988). The '251-ASD-protein4.1 derivative was separated from free iodine and unreacted "'I-SASD on a PD-10 desalting column (Bio-Rad). After determining the specific activity of the conjugate, the Iz5IASD-protein 4.1 derivative was incubated at room temperature for 90 min with stripped IOVs at a ratio of 300 pg of conjugate/mg IOV (saturatingprotein 4.1 concentration). Following incubation, the sample was photolyzed with a Mineralite UV lamp (254 nm) for 10 min approximately 2-4 cm above the sample in an open glass tube. The sample was immediately incubated at 25 "C in 50 mM dithiothreitol, 0.17 M acetic acid for 1 h to simultaneously cleave the crosslink and strip the released or unbound protein 4.1 from the membrane. To remove any residual unreacted lZ5I-ASD-protein4.1 from the IOVs, the above sample was subsequently pelleted and washed with 0.10 N NaOH and then neutralized in lysis buffer containing 1 mM DTT. Controls for this experiment included the use of heat-denatured 4.1 as well as preincubation of the (62 "C, 20 min) 1251-ASD-protein conjugate with 50 mM dithiothreitol for 60 min at 25 'C to cleave the cross-linker prior to addition of membranes. Both control samples were treated similarly to test samples with respect to photolysis, cross-linker cleavage, and stripping. When desired, the above labeled IOVs were cleavedwith NTCB in order to localize the lZ5I-ASDlabel to a specific site on cdb3. For this purpose, the labeled IOVs were washed with 200 mM Tris-HC1, 0.1 mM EDTA, pH 8.0, and suspended in the same buffer containing 7.5 M guanidine HCl. The unfolded polypeptides were then cleaved with NTCB, as described (Petrucci and Morrow, 1991). Miscellaneous Procedures-Electrophoresis was performed with the buffer system of Laemmli (Laemmli, 1970). Protein concentrations were measured using either the bicinchoninic acid or Coomassie Blue assays, according to the manufacturer's instructions(Pierce Chemical Co.), using BSA as a standard. Transfer of proteins from SDS gels to nitrocellulose was performed by an established protocol (Towbin et al., 1979). Transfer to azophenylthioether paper was carried out according to the manufacturer's instructions (Schleicher & Schuell). After protein transfer, blots were blocked with 4% BSA for 15 min and then incubated overnight at 4 "C with appropriate antibodies in 500 mM NaCl, 10 mM Tris-OH, 0.05% Tween-20, pH 7.5 (TBS-Tween). The blots were washed three times with TBSTween, once with 2.0 M urea, 100 mM glycine, 1.0% Triton X-100, and then once more with TBS-Tween. Blots were then incubated in 1:500 dilutions of either goat-anti-mouseIgG-horseradish peroxidase conjugate or protein A-horseradish peroxidase conjugate for 2 h, washed again as before, and rinsed three times briefly with distilled water. Bands were visualized with 4-chloro-1-naphthol. Autoradiog-

Localization of Protein the

9542

4.1-binding Site on Band 3

raphy was performed at -80 "C using a Cronex intensifying screen. Limited N-terminal sequence analysis was performed after transfer of proteins to ProBlott paper following the instructions of the manufacturer (Applied Biosystems).

migrate at theiroriginal molecular weights on the polyacrylamide gels (see species a t M , -88,000 and -30,000 in lune 8, respectively; neither glycophorin species stains with Coomassie Blue in lune 7). In contrast, band3 was partially digested at cysteines 201,317, and 479 (lane 7), yielding RESULTS predominant fragments of 23 kDa (1-201), 36 kDa (1-317), Identification of the Site of Photolytic Cross-linking of Pro- 17 kDa (317-479), and 12.5 kDa (201-317).' Importantly, tein 4.1 to Bund 3-Purified protein 4.1 was reacted with lZ5I- fragments of band 3 containing the N terminus (i.e. the 23SASD, a photoactivatable cross-linker withcentral a disulfide and 36-kDa species) were heavily labeled, while peptides bridge and then incubated with stripped IOVs, as described lacking this region were weaklylabeled (e.g. 12.5- and 17-kDa under "Experimental Procedures." After photolysis to cross- fragments). To more precisely quantitate thedegree of band 3 fragment link the bound band 4.1 to its membrane sites, both crosslinked and free band 4.1 were removed from the membranes labeling, a portion of the NTCB digest was transferred to by a rigorous stripping and DTT-reducing protocol. This nitrocellulose or azophenylthioether paper following electrotreatment left the '"1 label exclusively on membrane proteins phoresis and immunostained with antibodies specific for the with which band 4.1 had been interacting at the time of different fragments of band 3. The location of each fragment photolysis. None of the original band 4.1 remained on the on the blot was then used to accurately excise the correspondmembranes as evidenced by an inability to immunoblot any ing region of an adjacent (nonblotted) lane in the polyacrylprotein 4.1 in the samples (data not shown). As seen in Fig. amide gel. The excised gel bands were then counted in the 1,lune 6, membrane proteins labeled by this protocol include gamma counter and compared for radioactivity. The most band 3, glycophorin A, and glycophorin C. While glycophorin heavily labeled fragments were, as anticipated, the 23- and A was also partially labeled in control experimentswith heat- 36-kDa N-terminalfragments, with the 12.5- and 17-kDa denatured '251-ASD-protein 4.1 (lane 4 ) , band 3 and glyco- fragments containing only 20 and 14%, respectively, of the phorin C were devoid of radioactivity in these control lanes, radioactivity of the 23-kDa fragment. These data suggest that suggesting the latter interactions require nativeband 4.1. the major site of band 4.1 binding to band 3 lies in the 23Curiously, minor labeling of unidentified integral membrane kDa N-terminal fragment of the anion transporter. Identification of the Site of Inhibition of Bund 3 Proteolysis proteins of M, 40,000-60,000 was also detected with native "'I-ASD-protein 4.1, suggesting heretofore uncharacterized by Protein 4.1-A second independent method of mapping the protein 4.1-binding site on band 3 is to examine where minorprotein 4.1-binding sites may existonerythrocyte protein 4.1 protects band 3from proteolysis. For this purpose, membranes. In order to identify the subsite on band 3 involved in the stripped IOVs were digested with chymotrypsin either in the band 4.1 interaction, the radioiodinated stripped IOVs were presence or absence of 0.44 mg/ml band 4.1 (a chymotrypsin digested with NTCB, separated electrophoretically in SDS concentration too low to cleave the glycophorins (Danilov et gels, and autoradiographed (Fig. 1, lane 8 ) . Because glyco- ul., 1990)) and then analyzed for proteolysis products using a phorins A and C lack cysteine residues, they continued to polyclonal IgG to thecytoplasmic domain of band 3. Asshown inthe immunostained samples of Fig. 2 A , chymotrypsin cleaves band 3 in the absence of protein 4.1 into major CB AR CB AR cytoplasmic fragments of -43 and 34 kDa (lane B ) . Limited cleavage in the middle of the cytoplasmic domain yielding fragments of M , -20,000-24,000 is also observed as is a band at -62,000 corresponding to a natural breakdown product of band3 band 3 (Morrison et ul., 1985). In the presence of band 4.1, A dimer however, the anion transporter is totally protected from proteolysis at the34-kDa site (lane C ) . Some reduction in cleavage at the43-kDa site is also detected in theimmunoblots. Although the 43-kDa site has been unambiguously assigned C dimer to a junctional peptide connecting the membrane-spanning 23 and cytoplasmic domains of band 3, the 34-kDa site hasnever been carefully mapped. For this purpose, purified cytoplasmic domain of band 3 was digested in the absence of protein 4.1, 17 only this time the chymotrypsin concentration was elevated 12 &fold to maximize yield of the 34-kDa fragment. To identify which end of the 43-kDa cytoplasmic domain of band 3 was FIG. 1. Identification of protein 4.1-binding sites on missing in the34-kDa fragment, the proteolysis products were stripped IOVs by ""I-SASD radiolabel transfer cross-linking. immunostained with antibodies to the extreme C terminus pH 11 stripped IOVs were incubated with '251-ASD-protein4.1 and (m41-43) or extreme N terminus (m00-10;p00-01)of the then photolyzed to cross-link the protein 4.1 to its membrane sites. cytoplasmic domain. As seen in Fig. 2B, lune 3, the monoclonal After cleaving the cross-link and stripping to remove both unreacted and released protein 4.1, the stripped IOVs were either examined IgG to the junctional peptide connecting the membranedirectly by SDS-polyacrylamide gel electrophoresis on 10-15% gra- spanning and cytoplasmic domains of band 3 readily recogdient gels or cleaved first with NTCB and then analyzed as above. nizes the 34-kDa fragment. In contrast, the monoclonal IgG Lunes 1-3 and 7 were stained with Coomassie Blue, while lanes 4-6 mapped to the N-terminal 10-kDa fragment of band 3 (moo-and 8 represent the corresponding autoradiographs. The content of 10) does not stain the34-kDa peptide (lune 4). Similarly, the the lanes are: lunes 1 and 4, stripped IOVs labeled with heat-denatured weaker polyclonal IgG to residues 1-15of band 3 does not '""IASD-protein 4.1; lunes 2 and 5, stripped IOVs labeled with 1251-

-

-

-

-

-

ASD-protein 4.1 that was treated with 50 mM DTT to cleave the cross-linker prior to photolysis on the membranes; lane 3 and 6, stripped IOVs labeled with native lZ5I-ASD-protein4.1;lanes 7 and 8, the sample of lunes 3 and 6 after digestion with NTCB.

The identities of the NTCB fragments were determined by immunoblotting with the anti-cdb3 antibodies shown in Figs. 3 and 4, as described elsewhere (Lombardo, 1992).

Localization of the Protein 4.1-binding Site

A

A

95 iris

43



3 4 ,

B

1

2

3

4

5

FIG. 2. A , protection of cdb3 from chymotryptic cleavage by protein 4.1. pH 11, stripped IOVs (1.17 mg/ml) were either left unmodified or incubated with protein 4.1 (0.44 mg/ml) for 60 min in 150 mM NaCI, 10 mM HEPES, 1 mM EDTA, 0.5 mM DTT, pH 7.5 on ice. Chymotrypsin was added a t a band 3 to chymotrypsin ratio (by weight) of 875 and thesample was digested on ice for 30 min. Under these conditions, the glycophorins are not cleaved. Proteolysis was then terminated with 5 mM DFP. Samples were separated electrophoretically on a 10-15% SDS-gel, transferred to nitrocellulose, and immunostained with a polyclonal IgG raised against the entire cytoplasmic domain of band 3, as described under “Experimental Procedures.” Lane A, stripped IOVs, untreated. Lune B, stripped IOVs digested with chymotrypsin. Lane C, stripped IOVs incubated with protein 4.1 and then digested with chymotrypsin. B, Mapping of the 34-kDa chymotryptic cleavage site on cdb3 with N- and C-terminal IgGs. Purified cdb3 was digested with chymotrypsin a t a cdb3/ chymotrypsin ratio (by weight) of 180 in 100 mM NaCI, 10 mM HEPES, 1 mM EDTA, 1 mM NaN3, 0.5 mM DTT, pH 7.5 for 45 min at room temperature. Digestions were terminated with 5 mM DFP. Samples were then separated electrophoretically on a 12.5% SDS-gel and either stained directly or transferred to azophenylthioether paper and probed with antibodies specific for the N and C terminusof band 3. Lanes 1 and 2 are stainedwith Coomassie Blue while lanes 3-5 are immunostained with antibodies to cdb3 (see “Experimental Procedures’’). The contents of the lanes are: 1, intact 43-kDa cdb3; 2, cdb3 digested with chymotrypsin; 3, chymotrypsin-digested cdb3 stained with amonoclonal IgG (1.5 pg/ml) that recognizes an epitope between residues 359 and 379; 4, chymotrypsin-digested cdb3 stained with a monoclonal IgG(1.5 pg/ml) that recognizes an epitope between residues 1 and 91; 5, chymotrypsin-digested cdb3 stained with a polyclonal IgG (1.5 pg/ml) raised against residues 1-15 of band 3. It should be noted that localization of the chymotrypsin cleavage site by sequence analysis to the Y46-H47 peptide bond (see below) may further limit the epitope of the monoclonal antibody used in lane 4 (m00-10) to a sitebetween residues 1and 46, since m00-10 recognizes intact cdb3 (residues 1-379) but not the 34-kDa chymotryptic fragment (residues 47-379).

recognize the protected fragment ( l a n e 5 ) . These datasuggest the chymotryptic site responsible for the 34-kDa peptide of band 3 is -8 kDa from its N terminus. To more accurately locate this chymotryptic cleavage site, the 34-kDa chymotryptic fragment was electrophoretically transferred to ProBlott paper and partially sequenced (see “ExperimentalProcedures”). The first 6 amino acids were HTTSHP corresponding to residues 47-52 of the humansequence (Tanner et al., 1988; Lux et al., 1989). Thus, chymotrypsin must cleave the cytoplasmic domain of band 3 a t tyrosine 46, and this site must be protected by band 4.1 binding. Localization of the Protein 4.1 Site on Band 3 by Competition

on Band 3

9543

with Epitope-mapped Fabs-A third method of localizing the protein 4.1 site on the elongated cytoplasmic domain of band 3 is to analyze competition between band 4.1 and epitopemapped antibodies for binding to different loci along the cytoplasmic domain. The antibodies used for this purpose were mapped previously and characterized withregard to their relative affinities for cdb3 (Willardson, 1990; Willardson et al., 1989). The relative affinities of the monoclonal antibodies differed among themselves by less than a factor of 3, while the affinities of the antipeptideantibodies varied by no more than a factor of 5. No exhaustive comparison of the affinities of the monoclonal with the polyclonal antibodies was conducted, but the immunoblots in Fig. 2B obtained at thesame IgG concentration suggest they lie within an order of magnitude of each other, with the polyclonal Fabs binding weaker. For competition studies, these IgG were converted to their Fab fragments in order to reduce their radii of steric interference and also to avoid long range disturbances due to bivalent cross-linking. M41-43, a monoclonal Fab to thepeptide joining the membrane-spanning and cytoplasmic domains of band 3 (residues 359-379) was found to be an inhibitor of band 4.1 binding, but only at extremely high concentrations (Fig. 3a). Thus, nearly quantitative inhibition of band 4.1 binding was only achieved at more than 100-fold molar excess of Fab to band 3. Since easily accessible epitopes should be opsonized at essentially 1 to 1 molar ratios, and since the observed inhibition was highly cooperative (the derived Hill coefficient was 2.4), we speculate that the inhibition mechanism may involve a long range conformational change in the band 3 oligomer or some other indirect perturbation of the band 4.1 site. A similar but weaker inhibition of protein 4.1 binding was

-

b

loo 10041

80 80

-

60 40

-

20

-

0

. ...

.

11b.1

(

. . . . . . a

1

.

.......I

100

10

mole Fablband

2

..

1.31.1

1000

3

C

\

-1

I

1 :L r L l 40

20

1

10

100

mol* Flblband I

10

100

molo Fablband 3

FIG. 3. Inhibition of protein 4.1 binding to stripped IOVs by Fab fragments of monoclonal antibodies with epitopes on cdb3 near its junction with the membrane. KI-stripped IOVs were incubated with increasing concentrations of m41-43, m36-41A, and m36-40B for 3 h a t 25 “C in buffer A. Binding of protein 4.1 was then quantitated as described under “Experimental Procedures.” Panel a, m41-43; panel b, m36-414 panel c, m36-41B. The locations of the antibodies epitopes are described in the text anddocumented elsewhere (Willardson et al., 1989).

Localization of Protein the

9544

observed with two non-overlapping monoclonal Fabs mapped to regions 36-41 kDa from the N terminus of band 3 (Fig. 3, b and c ) . M36-41A, which maps slightly closer to the membrane thanM36-41B, achieved 50% inhibition only at an80fold molar excess of Fab, while the latter monoclonal Fab required a 200-fold excess for the same degree of competition. The antipeptide polyclonal Fab p22-23 raised against residues 189-203 on the C-terminal side of the putative prolinerich hinge was a very weak competitor of protein 4.1 binding (Fig. 4a). Although this same Fab was previously shown to aggressively block ankyrin binding at a 1:l molar ratio (Willardson et al., 1989),even at 100-fold molar excess it had little influence on band 4.1 association. This lack of participation of the central hinge in protein 4.1 binding was also confirmed by the absence of any major effect of N-ethylmaleimide on band 4.1 association. Although this reagent inhibited the ankyrin linkage via derivatization of cysteine 201 (Thevenin et al., 1989), it had only a minor effect on the association of protein 4.1 with IOVs (data not shown). P16-17, a polyclonal Fab raised against a highly conserved sequence comprising residues 142-154 of band 3, was also unable to compete with protein 4.1 for band 3 (Fig. 4a). Finally, Fabs to the N terminus were examined for their abilities to block protein 4.1 binding to IOVs (Fig. 46). In contrast to all previous Fabs, m00-10 potently inhibited band 4.1 association, achieving 50% reduction at less than 1:l molar ratio. At high concentrations of m00-10, nearly complete obstruction of protein 4.1 binding to stripped IOVs was observed. Since this monoclonal Fab has been mapped to an epitope between residues 1 and 91 of band 3 (Willardson, et al., 1989), and since the inhibition data describe a Hill coefficient of essentially unity, the dataof Fig. 4b strongly imply

l o 80o-

40 6o

1

4.1-binding Site on

en 8

Band 3

““1 20

1



.

‘....I

1

- . - .....’ 10

. . ‘...‘I 100

mole ankyrinlband 3

FIG. 5. Inhibition of protein 4.1 binding to stripped IOVs by ankyrin. pH 11 stripped IOVs were incubated with increasing concentrations of erythrocyte ankyrin for 90 min at 25 “C in buffer A. 1251-Protein4.1 binding was then quantitated as described under “Experimental Procedures.”

a one to one competition between m00-10 and protein 4.1 for the Nterminus of band 3. A second Fab (p00-01) raised against residues 1-15 of band 3was also able to compete with band 4.1 for the anion transporter(Fig. 46). Probably because of its much lower affinity for band 3 (compare lanes 4 and 5 of Fig. 2 B ) , it was also found to be a much weaker inhibitor of protein 4.1 binding. Influence of Ankyrin on the Band 4.1-Band 3 InteructionIn previous work, ankyrin was found to bind band 3 at two nonadjacent sites: (i) near the putative central hinge and (ii) near the N terminus (Davis et al., 1989; Willardson et al., 1989). In fact, ankyrin association was shown to inhibit phosphorylation of tyrosines 8 and 21 of band 3 (Willardson et al., 1989). In viewof this evidence for an N-terminal interaction, the question naturally arises as to whether band 4.1 and ankyrin might compete for an overlapping site in this region. To test thispossibility, stripped IOVs were incubated with increasing concentrations of ankyrin and thenexamined for their ability to bind protein 4.1. As displayed in Fig. 5, ankyrin did block association of band 4.1 with band 3, achieving 50% inhibition at a 5:l ankyrin to band 3 ratio. Because veryhigh ankyrinconcentrationsare required to saturate band 3-binding sites(Theveninand Low, 1990), complete inhibition was never achieved. Nevertheless, these data do suggest that quantitative competition may be possible if sufficient ankyrin is forced onto the IOVs. Whether this inhibition results from competition for overlapping sites on cdb3 or whether the decrease in band 4.1 binding derives from steric interference due to binding of a bulky ankyrin molecule to an adjacent independent site requires further scrutiny. Competitive Displacement of Protein 4.1 by the N-terminal Pentudecameric Peptide of Band 3-Because of the above evidence for association of protein 4.1 with a region near the N terminus of band 3, we decided to determine whether a synthetic peptide comprising of residues 1-15 of band 3might at least partially compete with intact band 3 for protein 4.1. As shown in Fig. 6, the pentadecameric peptide was able to weakly inhibit the association of protein 4.1 with IOVs, whereas a control peptide comprising residues 189-203of band 3 was not. While the observed weak competition does not argue in favor of a prominent role for this acidic peptide in the protein 4.1 interaction, the clear superiority of the Nterminal peptide over the control peptide suggests that the acidic N terminus may be involved.

2 0

1

10

100

1000

mole Fab/band 3

\ ; -

loo 80

0 :

*

.

.......I

1

. ....



.......I

10

T

100

1000

mole Fabkdb3

FIG. 4. Inhibition of protein 4.1 binding to stripped IOVs by Fab fragments of antibodies with epitopes on cdb3 near the middle and near the N terminus of the cytoplasmic domain. See textfor the location of the epitope of each antibody. Experiments were carried out as described in Fig. 3. Panel a, p22-23 (O), p16-17 ( 0 )panel ; b, m00-10 (O), p00-01 (A).

DISCUSSION

As with many other erythrocyte membrane protein interactions characterized in vitro, evidence for an in vivo associ-

Localization of Protein the

4.1 -binding on Site

ation of band3 with band 4.1 is lacking. The strongest arguments in favor of this association in uiuo are that (i) the interaction occurs in uitro under physiological solution conditions, (ii) there are insufficient binding sites to accommodate all copies of band 4.1 elsewhere on the membrane (see Introduction),(iii)theband 3-band 4.1 interaction is of roughly equal affinity to theglycophorin-band 4.1 interaction (Danilov et al., 1990), (iv) there are more band 3 sites than any other class of protein 4.1-binding sites on the membrane, and (v) the band 3-band4.1 interaction is obliterated by band 4.1 denaturation (Fig. 1). Although, final resolution of this issue must await analysis of the molecular distribution of band 4.1 among membrane components in uiuo, the in uitro data suggest that a prominent band4.1-band 3 interaction in uiuo warrants serious consideration. The lZ5I-SASDlabel transfer studies, the proteolysis protection experiments, and the Fab, ankyrin, and peptide competition studies all point to an interaction between protein 4.1 and band 3 nearthe latter protein’s N terminus. Because band 3also binds glyceraldehyde-3-phosphate dehydrogenase, aldolase, phosphofructokinase, hemoglobin, hemichromes, and ankyrin within this region (Murthy et al., 1981; Tsai et al., 1982; Jenkins et al., 1984; Low, 1986; Waugh and Low, 1985; Davis et al., 1989; Willardson et al., 1989), the question naturally arises as to whether there is sufficient band 3 to bind all potential peripheral protein ligands. Beginning with the most avid interaction, one canestimate that ankyrin would occupy approximately IO5of 1.2 x lo6band 3 sites/cell (Bennett and Stenbuck, 1980). Band 4.1, which may exhibit the next highest affinity for band 3, could then consume an additional 150,000 copies, assuming one allows 50,000 molecules of 4.1 to reside on glycophorin C (Mueller and Morrison, 1981; Reid et al., 1987). The glycolytic enzymes can then be estimated to control approximately 230,000 band 3 sites/cell, assuming two-thirds of the enzyme population is bound at a 1:l stoichiometry in vivo (Klimanand Steck, 1980). This would finally leave -7 x lo5 band 3 sitesfor hemoglobin and hemichromes. Although the affinity of the N terminusof band 3 for hemichromes is fairly high (Waugh et al., 1987), their appearance only in senescent (Sears et al., 1975; Kannan et al., 1991) or abnormal erythrocytes (Kannan et al., 1988; Rachmilewitz et al., 1985) dismisses them from consideration in the architecture of a healthy cell. However, native hemoglobin, with a band 3association constant of -IO4 M-‘ and an intracellular concentration of -5 mM (Walder et al., 1984), could occupy most of the remaining sites in the absence of other unidentified interactions. Thus, there is currently no 5

110

I

I

0

0 0



!

o

o

l 90

EO

70

! . ......., . . . ...., . ......., . ......, . ......., . . ....., . ...A . . ...” 100 10 mole peptidelmole cdB3

1000

10000

FIG.6. Partial inhibition of protein 4.1 binding to stripped IOVs by an N-terminal peptide of cdb3. Increasing concentrations of synthetic peptides representing residues 1-15 (U)or 189-203 (0)of cdb3 were incubated with pH 11 stripped IOVs for 90 min at 25 “C in buffer A. T - P r o t e i n 4.1 binding was then quantitated as described under “Experimental Procedures.”

Band 3

9545

reason to suspect significant competition for the N terminus of band 3 in situ. With so many proteins binding to theN terminus of band 3, it may seem initially surprising that this region would be least conserved among all regions of band 3(Lux et al., 1989). However, it must not be overlooked that residues 57-110 of the human, bovine, and murine erythrocyte band 3 sequences are highly homologous (Lux et al., 1989; Kopito and Lodish, 1985; Moriyama et al., 1989), and that regions N-terminal to valine 57 are in all three cases strongly acidic even though the positions of Glu and Asp do not align. It is, therefore, conceivable that for certain N-terminal interactions, conservation of charge is sufficient to assure at least part of a high affinity interaction. For band 3 homologues that completely lack this region (e.g. human kidney band 3 which begins at the equivalent of Met-66 in thehuman sequence (Alper et al., 1988)), there is currently no evidence that any of the above peripheral protein interactions can occur. In fact, we have recently examined the affinity of a bacterially expressed cytoplasmic domain of kidney band 3for band 4.1, ankyrin, and glyceraldehyde-3-phosphate dehydrogenase and havebeen unable to detect anya~sociation.~ Thus, itmay be that simple retention of charge adjacent to regions of strong sequence homologyis both necessary and sufficient to maintain an affinity for several of the band 3 N-terminal ligands. Although most of the data strongly suggest an association between band 4.1 and the N terminus of band 3, several of the studies also implicate involvement of regions of band 3 directly adjacent to the membrane. Thus, antibodies to band 3 epitopes near the membrane exhibit weaker but measurable competition with protein 4.1, the lZ5I-ASD-protein4.1 conjugate weakly labels a peptide (17 kDa) spanning this membrane-juxtaposed region, and saturation of IOVs with protein 4.1 reduces cleavage of the junctional peptide connecting the 43-kDa cytoplasmic domain to themembrane. Assuming these data are meaningful, unless protein 4.1 can independently occupy two widely separated sites on band 3, the simplest interpretation is that the N terminus of those copies of band 3 that carry a protein 4.1 lies next to bilayer.Given the flexibility of the elongated cytoplasmic domain of band 3 (Low et al., 1984), several geometric arrangements that satisfy this criterion are clearly possible. Unfortunately, little information is available on the tertiary structure of band 3 and it is not even known whether cdb3 subunits align parallel or antiparallel to each other. Still, the possibility that the band 4.1-associatedpopulationof band 3lies against the membrane is attractive, since considerable data indicate that partof the protein 4.1 population is, in fact, situated there. Thus, band 4.1 is the predominant membrane protein labeled in situ by photoactivatable phospholipid probes (Pradhan et al., 1991), protein 4.1 displays high affinity for anionic phospholipids that are present in the inner leaflet of the red cell membrane (Cohen et al., 1988), and energy transfer measurements place band 4.1 only -75 A from the lipid surface (Shahrokh et al., 1991). Finally, as argued in the Introduction, up to 25% of the protein 4.1 could reside on high affinity sites at the cytoplasmic domain of glycophorin C. Still othercopies of protein 4.1 could associate with the intracellular pole of glycophorin A, especially if the proximal lipid poolwere enriched in phosphorylated phosphatidylinositols (Anderson andMarchesi, 1985). Why then are Fabs to the cytoplasmic domain of band 3 able to quantitatively inhibit band 4.1 association with stripped IOVs (see Figs. 3and 4). Although we still



C. R. Lombardo, C. C. Wang, R. Moriyama, andP. manuscript in preparation.

S. Low,

Localization ofProtein the

9546

4.1-biding Site on Band 3

Kay, M. M. B., Bosman, G. J. C. G. M., and Lawrence, C. (1988) Pmc. Natl. A d . Sci. U. S. A. 85,492-496 Kliman, H. J., and Steck, T. L. (1980) J. Biol. Chem. 256,6314-6321 Kopito, R. R., and Lodish, H.F. (1985) Nature 316,234-238 Korsmen. C.. and Cohen. C. M. (1988) J. Biol. Chem. 263.10212-10218 Lae~mli,'U.'K~~(l970) Natu~~227,&3&685 Liu, S. C., Zhai, S., Palek, J., Golan, D. E., Amato, D., Hassan, K., Nurse, G., Babona. D.. Coetzer. T.. Jarolin.. P... Zaik., M... and Bonvein.. S. (1990) . . New Engl. J.'Med. 323,153d-1538 Lombardo, C. R. (1992) Protein 4.1: Static and Dynamic Aspects of Its Association with the Human Erythrocyte Membrane, Ph.D. dissertation, Purdue University Lombardo, C. R., Willardson, B. M., and Low, P. S. (1990) J. Cell Biol. 1 1 1 , 46 (abstr.) Low, P. S. (1986) Biochim. Biophys. Acta 864,145-167 Low, P. S., Westfall, M. A,, Allen, D. P., and Appell, K. C. (1984) J.Bwl. Chem. 2m13070-1~076 "_, Low,P. S., Willardson, B. M., Mohandas, N., Rossi, M., and Shohet, S. B. (1991) Blood 77,1581-1586 Lux. S. E.. John. K.M..KoDito. R. R.. and Lodish. H. F. (1989) . , Proc. Natl. A d . Sii. U.S A. 86,'908619093 ' McGuire, M., Smith, B. L., and Agre, P. (1988) B h d 72,287-293 Morivama. R.. Kawamatsu. S.. Kondo. Y.. Tomida.. M... and Makino. S. (1989) A&. Bwchkm. Biophys. 274,130-137 M.. Grant. W.. Smith. H. T..Mueller. T. J.. and Hsu. L. (1985) ~~~~~. ~~.~ Acknowledgments-We thank Dr. Merril Benson and William Kus- Morrison. Biochehist; 24,6311-k315-~ ' ter for raising the monoclonal antibodies employed in this study. Dr. Mueller, T. J., and Morrison, M. (1981) in Erythrocyte Membranes 2: Recent Jon S. Morrow is gratefully acknowledged for suggesting the use of Clinical and Exwrimental Advances (Kruckebure. W.. Eaton.. J... and Brewer, G., eds) pp. 95-112, Alan R. Liss, Inc., NewYo& . SASD as a cross-linker and for providing a manuscript in advance of S. N. P., Liu, T., Kaul, R. K., Kohler, H., and Steck, T. L. (1981) J. its publication. Mary Woenker is thanked for performing the se- Murthy, Biol. Chem. 266,11203-11208 quence analysis. Drs. Paul0 Arese and Ryuichi Moriyama are thanked Nigg, E. A., Bron, C., Girardet, M., and Cherry, R. J. (1980) Biochemistry 19, 1887-1893 for critically reviewing the manuscript. Ohanian, U., and Gratzer, W. (1984) Eur. J. Biochem. 144,375-379 Pasternack, G. R., Anderson, R. A., Leto, T. L., and Marchesi, V. T. (1985) J. REFERENCES Biol. Chem. 260,3676-3683 Alloisio, N., Morlb, L., Dorleac, E., Gentilhomme, O., Bachir, D., Guetarni, D., Petrucci, T., and Morrow, J. S. (1991) Biochemistry 3 0 , 413-422 Colonna, P., Bost, M., Zouaui, Z., Rods, L., Roussed, D., and Deluaney, J. Pinder, J. C., and Gratzer, W. B. (1989) Bwchem. J. 264,423-428 (1985) Blood 65,46-51 Pradhan, D., Williamson, P., and Schlegel,R. A. (1991) Biochemistry 30,7754Alper, S. L., Kopito, R. R., Libresco, S. M., and Lodish, H. F. (1988) J. Biol. 7758 Chem. 263,17092-17099 Rachmilewitz,E. A., Shinar, E., Shalev, O., Galili, U., and Schrier, S. L. (1985) Anderson, R. A., and Marchesi, V. T. (1985) Nature 318,295-298 Clin. Haemotol. 1 4 , 163-182 Anstee, D.J., Parsons, S. F., Ridgwell, K., Tanner, M. J. A., Merry, A.H., Reid, M., Chasis, J., and Mohandas, N. (1987) Blood 6 9 , 10&1072 Thomson. E.E.. Judson. P. A.. Johnson. P.. Bates.,S..,and Fraser. I. D. (1984) Reid, M. E., Takakuwa, Y., Conboy J., Tchernia, G., and Mohandas, N. (1990) Biochem. 2. 216,615- 619 ' ' ' Blood 75,2229-2234 Appell, K. C., and Low, P. S. (1981) J. Biol. Chem. 256,11104-11111 Sears, D. A., Friedman, J. M., and White, D. R. (1975) J. Lab. Clin. Med. 86, Bennett. V. (1983) Methods Enrvmol. 96.313-324 722-732 Bennett: V. (1989) Biochim. Biuihys. Acta 988,107-121 Shahrokh. Z.. Verkman. A. S.. and Shohet. S. B. (1991) . . J. Biol. Chem. 266. Bennett, V. (1990) Physiol. Reu. 70,1029-1065 12082-12089 Bennett, V., and Stenbuck, P. J. (1980) J.Biol. Chem. 256,6424-6432 Shepard, E. G., DeBeer, F. C., Von Holt, C., and Hapgood, J. P. (1988) Anal. Chasis, J. A., Mohandas, N., and Shohet,S. B. (1985)J.Clin. Inuest. 76,1919Biochem. 168,306-313 1926 Tang, T. K., gin, Z., Leto, T., Marchesi, V. T., and Benz, E. J. (1990) J. Cell. Chasis, J. A., Reid, M. E., Jensen, R. H., and Mohandas, N.(1988) J. Cell Biol. Biol. 110,617-624 107,1351-1357 Tanner, M. J. A,, Martin, P. G., and High, S. (1988) Biochem. J. 266,703-712 Coetzer, T. L., Lawler, J., Liu, S. C., Prchal, J. T., Gualpieri, R. J., Brain, M. Thevenin, B. J.-M., and Low, P. S. (1990) J. Biol. Chem. 2 6 5 , 16166-16172 C., Dacie, J. V., and Palek, J. (1988) N. Engl. J. Med. 3 1 8 , 230-234 Thevenin, B. J. M.. Willardson, B. M., Low, P. S. (1989) J. Biol. Chem. 2 6 4 , Cohen. A. M.. Lui. S. C.. Lawler..J... Derick.. L... and Palek.. J. (1988) 15886-i5892 . . Blochemistrv 27,614-619 ' ' Towbin, H., Stachelin, T., and Gordon, J. (1979) Proc. Natl. Acud. Sci. U. S. A. Coleman, T. R., Fishkind, D. J., Mooseker, M. S., and Morrow, J. S. (1989) 76,4350-4354 Cell. Motil. Cytoskeleton 1 2 , 225-247 Tsai, I. H., Murthy, S. N. P., and Steck, T. L. (1982) J.Biol. Chem. 2 5 7 , 1438Conboy, J., Mohandas, N., Tchernia, G., and Kan, Y. W. (1986) N . Engl. J. 1442 Med. 316,680-685 Tyler, J. M., Hargreaves, W. R., and Branton, D. (1979) Proc. Natl. Acud. Sci. Conboy, J., Chan, J. Y., Chasis, J., Kan, Y. W., and Mohandas, N. (1991) J. U. S. A. 76,5192-5196 Biol. Chem. 266,8273-8280 Walder, J. A., Chatterjee, R., Steck, T. L., Low, P. S., Musso, G. F.,Kaiser, E. Danilov, Y. N., Fennell, R., Ling, E., and Cohen, C. M. (1990) J. Biol. Chem. T., Ro ers, P. H., and Arnone, A. (1984) J. Biol. Chem. 269,10238-10246 265,2556-2562 Waugh, M., and Low, P. S. (1985) Blochemutry 24,34-39 Davis, L., Lux, S. E., and Bennett, V. (1989) J. Biol. Chem. 264,9665-9672 Wpph. S. M., Walder, J. A., and Low, P. S. (1987) Biochemistry 2 6 , 1777Jenkins, J. D., Madden, D. P., and Steck, T. L. (1984) J. Biol. Chem. 2 6 9 , 110.3 9374-9278 Weinstein, R. S., Khodadad, J. K., and Steck, T. L. (1978) J. Supramol. Struc. ... - - - . . Jinbu. Y.. Sato. S.. Nakao. T.. Tsukita. S.. Tsukita, S., and Ishikawa, H. (1894) 8,325-335 Willardson, B. M. (1990) @calizution of Peripheral Membrane Protein Binding B&hem. Biiphys. Acta.773,237-245 Sites on the Cytoplasmic Domum of Erythrocyte Band 3, Ph.D. dissertatlon, Joyl, J. C., and Purich, D. L. (1990) Biochemistry 29,8916-8920 Purdue Universit K";;III~ R., Lobotka, R., and Low, P. S. (1988) J. Biol. Chem. 2 6 3 , 13766Willardson, B. M., Jhevenin, B. J.-M., Harrison, M. L., Kuster, W. L., Benson, Id I Id M. D., and Low, P. S. (1989) J. Bzol. Chem. 264.15893-15899 Kannan, R., Yuan, J., and Low, P. S. (1991) Biochem. J. 278,57-62

consider this observation very perplexing, we can offer two possible explanations. First, binding of protein 4.1 to the glycophorins could require simultaneous association with band 3. Obstruction of the latter interaction would then not only eliminate the glycophorin-independent interactions with band 3, but also the glycophorin-dependent ones. Second, saturation of band 3 sites with certain monoclonal Fabs could nonspecifically obstruct access of protein 4.1 to the glycophorins, especially if the glycophorins were associated with band 3, as suggested by some (Nigg et al., 1980). However, why some anti-band 3 Fabs could nonspecifically occlude all glycophorin tails while other Fabs have absolutely no effect (see Figs. 3 and 4)would remain difficult to explain. Clearly, much remains to be learned about the interactions of protein 4.1with the red cell membrane.

~~~~

"" ~

~

'

~

~

~~

~

~~~~~~~~~

~~

8.

~

I

~~

I

.

.