Mapping the Ankyrin-binding Site of the Human Erythrocyte Anion ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 16, Issue of June 5, pp. 9665-9672, 1989 Printed in U. S. A .

Mapping the Ankyrin-binding Site of the Human Erythrocyte Anion Exchanger*

(Received for publication, January 18, 1989)

Lydia Davis$, Samuel E. Lux$, and Vann BennettSV From the $Howard Hughes Medical Institute and Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710 and §the Division of HematologylOncology, Children’s Hospital and Dana Farber Cancer Institute, Boston, Massachusetts 02115

This report describes initial efforts to map the ankyrin-binding site of the cytoplasmic domain of the human erythrocyte anion exchanger. The conclusions are that this site is likely to involve a fairly extended sequence in the midregion of the cytoplasmic domain and requires interactions that are not provided by isolated peptides. The region of the sequence involving residues 174-186 is likely to participate in the ankyrin-binding site based on several experiments. Lim- ited tryptic cleavage in the midregion of the cytoplasmic domain (residues 174 and/or 181) nearly abolished the ability of the cytoplasmic domain to inhibit binding of ankyrin to the anion exchanger. Ankyrin protected the cytoplasmic domain from tryptic digestion. Finally, peptide-specific antibodies against the sequence encompassing the site(s) of tryptic cleavage (residues 174-186) blocked binding of ankyrin to the anion exchanger. However, the sequence comprising the tryptic site is not sufficient for high affinity binding of ankyrin. A 39-amino acid peptide (residues 161- 200) that includes the tryptic cleavage site(s) was inactive in inhibiting binding of ankyrin to the anion exchanger. Further evidence for a complex ankyrin- binding site is that peptide-specific antibodies against two different, noncontiguous regions (residues 118- 162 and 174-186) both inhibited binding of ankyrin to the anion exchanger and were only 10-2070 as effective as antibody against the entire cytoplasmic domain. Finally, the ankyrin-binding site of the anion exchanger did not renature following sodium dodecyl sulfate electrophoresis and transfer to nitrocellulose paper even though spectrin did recover ability to bind ankyrin under the same conditions. Thus, the ankyrin- binding site is not defined by a short continuous sequence. A simple consensus sequence for ankyrin-binding regions in other proteins is not likely.

The anion exchanger of erythrocytes is a chimeric molecule with a carboxyl-terminal domain embedded in the membrane bilayer that mediates exchange of bicarbonate and chloride and an amino-terminal cytoplasmic domain that is water- soluble and interacts with a variety of intracellular proteins (reviewed by Steck, 1978; Low, 1986; Jay and Cantley, 1986). The cytoplasmic domain of the anion exchanger is attached

*This research was supported in part by National Institutes of Health Grants DK29808 (to V. B.) and DK34083, HL32262, and HL15157 (to S. E. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

ll To whom correspondence should be addressed.

to a spectrin network on the inner surface of the membrane through a ternary complex involving ankyrin which in turn is linked to spectrin (Bennett and Stenbuck, 1979, 1980; Har- greaves et al., 1980; Bennett, 1982). In erythrocytes, the consequence of the anion exchanger-ankyrin-spectrin linkage is to provide a high affinity attachment of the spectrin skel- eton to the membrane. The attachment of the spectrin skel- eton to the membrane is likely to be essential for mechanical stability of erythrocytes in the circulation (reviewed by Ben- nett, 1985; Marchesi, 1985; Becker and Lux, 1985). The cytoplasmic domain also interacts with other membrane struc- tural proteins including protein 4.2 (Korsgren and Cohen, 1988) and protein 4.1 (Pasternack et al., 1985).

Association between isoforms of the anion exchanger and ankyrin may occur in other tissues such as the intercalated cells of the kidney collecting duct where these proteins are colocalized (Drenckhahn et al., 1985; Schuster et al., 1986; Wagner et al., 1987; Drenckhahn and Merte, 1987). The kidney anion exchanger and ankyrin exhibit a polarized dis- tribution and are confined to the basolateral domains of the intercalated cells. One consequence of association between the kidney anion exchanger and ankyrin could be mainte- nance or initial placement of the anion exchanger in a basolateral localization. Isoforms of ankyrin associate with other ion channels with polarized distributions including the Na/ K-ATPase (Nelson and Veshnock, 1987; Koob et al., 1987) which is confined to the basolateral domains of kidney distal tubule cells, and the voltage-dependent sodium channel (Sri- nivasan et al., 1988) which is localized at the nodes of Ranvier.

Association of the erythrocyte anion exchanger with erythrocyte ankyrin provides a well characterized model system with relevance to other ion channels in kidney and brain, as well as other tissues. The cytoplasmic domain of the erythrocyte anion exchanger is easily isolated and the sequence is known for the proteins of mice (Kopito and Lodish, 1985), humans (Kaul et al., 1983),’ and chickens (Cox and Lazarides, 1988; Kim et al., 1988). The cytoplasmic domain of the human anion exchanger is comprised of a polypeptide of M , = 43,000 which is an asymmetric dimer in solution (Appell and Low, 1981; Low et al., 1984). Binding sites of the cytoplasmic domain for glycolytic enzymes (Murthy et al., 1981) and for hemoglobin (Walder et al., 1984) are located at the amino terminus. The location of the ankyrin-binding site of the cytoplasmic domain is not known, although an amino-terminal site is unlikely because glycolytic enzymes do not compete for binding of ankyrin (Bennett and Stenbuck, 1980). The purpose of this study was to explore the ankyrin-binding site of the human anion exchanger using proteases, peptides, and peptide-specific antibodies.

S. E. Lux, manuscript in preparation.

9665

9666 Ankyrin-binding Site of the Anion Exchanger

EXPERIMENTAL PROCEDURES

Materials-Carrier-free Na"'1 was from Amersham Cop. and '*'I- labeled Bolton-Hunter reagent was from ICN Radiochemicals. Diiso- propylfluorophosphate, leupeptin, pepstatin A, dithiothreitol, phenyl- methylsulfonyl fluoride, EDTA, and Triton X-100 were from Sigma. Cyanogen bromide-activated Sepharose CL-4B and protein A-Seph- arose were from Pharmacia LKB Biotechnology Inc. Nitrocellulose paper and electrophoresis reagents were from Bio-Rad.

Human erythrocyte spectrin, ankyrin-depleted inside-out vesicles, and the cytoplasmic domain of the erythrocyte anion exchanger were purified as described (Bennett, 1983). Erythrocyte ankyrin was purified from human erythrocyte ghost membranes as described (Bennett, 1983; Hall and Bennett, 1987). Peptides were used without further purification following preparation by solid phase synthesis using t- butoxycarbonyl-protected amino acids and cleavage from the resin with HF.

Procedures-Protein was determined by the method of Lowry et al. (1951), using bovine serum albumin as a standard. SDS-polyacrylamide gel electrophoresis was performed using 0.2% SDS' with the buffers of Fairbanks et al. (1971) and 3.5-17% exponential gradient slab gels. Immunoblot analysis using 1251-labeled protein A to visualize antibodies was performed as described (Davis and Bennett, 1983).

Radiolabeling of Proteins-Ankyrin and the cytoplasmic domain of the anion exchanger were radiolabeled using 'ZsI-Bolton-Hunter reagent as described (Bennett and Stenbuck, 1980; Bennett, 1983). The average stoichiometry of label for ankyrin to be used in membrane binding assays was maintained at 1 molecule of label or less per molecule of ankyrin. Reactions were typically performed with incor- poration of 0.1-0.4 nmol radiolabel into 0.5 nmol of ankyrin. Ankyrin radiolabeled with Bolton-Hunter reagent in this manner was unal- tered in association with erythrocyte membranes (Bennett and Sten- buck, 1980). Protein A was radiolabeled with Na'''1 using chloramine T as an oxidant (Hunter and Greenwood, 1962).

Assay of Ankyrin Binding-Ankyrin-depleted inside-out vesicles (8-15 pg/ml) were incubated with 6-7 nM '251-labeled ankyrin (3-5 X 10' cpm/pmol) for 90 min at 4 or 24 "C in a final volume of 0.2 ml of assay buffer containing NaCl (100 mM), sodium phosphate (10 mM), NaEDTA (1 mM), dithiothreitol (1 mM), pH 7.5. Membrane-bound and free ankyrin were separated by centrifugation (30,000 X g for 30 min) of membranes through 0.2 ml of 20% sucrose dissolved in assay buffer in 0.4-ml Eppendorf polyethylene microtest tubes. The tubes were frozen in dry ice and the tips containing the membranes cut off and assayed for '"I to quantitate membrane-bound ankyrin. The control for nonspecific binding was binding of heat-denatured ankyrin (10 min at 62 "C) and these values, representing less than 8% of the total radioactivity, were routinely subtracted. Determinations were in duplicate. The error range after subtracting values for nonspecific binding was +8-12%.

Affinity-purified Antibodies-Antisera against the cytoplasmic domain of the human erythrocyte anion exchanger were prepared in rabbits as described (Bennett and Stenbuck, 1980). Peptides were dissolved at 1-2 mg/ml in a buffer containing 50 mM sodium phosphate, 0.1% sodium dodecyl sulfate, 1 mM NaN3, pH 8, and coupled to CNBr-activated Sepharose for 12 h a t 4 "C. The Sepharose incor- porated approximately 0.5-1 mg/ml peptide, based on loss of absorb- ance at 220 nm following coupling. One to two ml of peptide-Sepha- rose was used to isolate antibodies from antisera as described (Ben- nett and Stenbuck, 1980) with the modification that Ig was eluted with 4 M MgCI,. Antibodies were dialysed against 20% sucrose, 0.15 M NaCl, 10 mM sodium phosphate, 1 mM NaEDTA, 1 mM NaN,, and stored frozen at -70 "C. Nonimmune Ig was isolated by affinity chromatography on protein A-Sepharose and elution with 1 M acetic acid.

RESULTS

Evaluation of Renaturation of the Ankyrin-binding Site following SDS Electrophoresis-The goal at the beginning of this study was to define the ankyrin-binding site known to be located in the M , = 43,000 cytoplasmic domain of the human erythrocyte anion exchanger. An initial question was whether the binding site would renature following SDS electrophoresis and transfer of polypeptides to nitrocellulose paper. If ankyrin binding could be detected following such a procedure, it would,

'The abbreviation used is: SDS, sodium dodecyl sulfate.

in principle, be possible to prepare sub-fragments of the anion exchanger and identify a minimal size for activity. The ankyrin-binding site of the @ subunit of erythrocyte spectrin apparently does renature sufficiently to associate with radiolabeled ankyrin following transfer to nitrocellulose (Fig. 1; Davis and Bennett, 1984). However, the cytoplasmic domain of the anion exchanger does not associate with ankyrin under these conditions (Fig. 1). The cytoplasmic domain was effi- ciently transferred and adsorbed to nitrocellulose based on immunoblots and direct staining of the nitrocellulose with Poinceau S (not shown). I t is conceivable that the cytoplasmic domain binds to ankyrin only when associated as a dimer, the native conformation of this polypeptide (Nigg and Cherry, 1979; Appell and Low, 1981). The cytoplasmic domain cross- linked to a dimer using o-phenanthroline and Cu2+ (Steck et al., 1976; Appell and Low, 1981) was still inactive following transfer to nitrocellulose (not shown). It is conceivable that modification of thiols may have reduced ability of the cytoplasmic domain to associate with ankyrin. The conclusion of these experiments was that the ankyrin-binding site of the cytoplasmic domain does not renature to an extent that would be useful for further mapping. A corollary of this result is that the binding site may require tertiary structure and is not defined by a simple linear sequence of amino acids.

The cytoplasmic domain typically contains a mixture of polypeptides of M , = 41,000 and 43,000 (Fig. 1; Bennett and Stenbuck, 1980; Appell and Low, 1981). The M, = 41,000 polypeptide most likely represents a subfragment of the M , = 43,000 polypeptide because these polypeptides cross-react with affinity-purified antibodies and because the amino acid sequence predicts several potential sites for chymotryptic cleavage.

Limited Proteolytic Cleavage of the Cytoplasmic Domain- As an initial approach to resolving the binding site, trypsin and chymotrypsin were used to prepare sub-fragments of the cytoplasmic domain. The activity of these proteolytic frag-

M r x 10.3

3%

43- 32-

20-

1 2

m -

3 4

Y w -beta spectrin

FIG. 1. Renaturation of the ankyrin-binding site of spectrin @ subunit but not the ankyrin-binding site of the anion exchanger following SDS electrophoresis and transfer of polypeptides to nitrocellulose paper. Samples of the cytoplasmic domain of the anion exchanger (lane l ) , M, = 32,000 (lane 2) and M , = 20,000 ( l a n e 3) fragments of the anion exchanger cytoplasmic domain, and erythrocyte spectrin ( l a n e 4 ) were electrophoresed on a SDS- polyacrylamide gel and electrophoretically transferred to nitrocellulose paper (see "Experimental Procedures"). The nitrocellulose paper was incubated for 15 h at 4 "C with 12sII-labeled ankyrin (10 nM, 350,000 cpm/pmol) in a buffer containing 150 mM NaCI, 10 mM sodium phosphate, 1 mM NaEDTA, 1 mM NaN3, 0.2% Triton X-100, 20 mg/ml bovine serum albumin, pH 7.4. The paper was washed five times with incubation buffer lacking bovine serum albumin, dried, and analyzed by autoradiography.

Ankyrin-binding Site of the Anion Exchanger 9667

ments was evaluated by measuring their ability to inhibit binding of radiolabeled ankyrin to ankyrin-depleted inside- out vesicles. The association of ankyrin with inside-out vesicles has been characterized and demonstrated to result from an interaction of ankyrin with the cytoplasmic domain of the anion exchanger (Bennett and Stenbuck, 1980). The cytoplasmic domain is an effective inhibitor of ankyrin binding to inside-out vesicles as noted previously (Fig. 2) (Bennett and Stenbuck, 1980).

Cleavage of the cytoplasmic domain with chymotrypsin under mild conditions results in a fragment of M, = 32,000 (Fig. 1) that is equally active as undigested cytoplasmic domain in inhibiting binding of ankyrin (Fig. 2). The M, = 32,000 fragment represents the amino-terminal portion of the cytoplasmic domain because subsequent digestion of this fragment with trypsin yields an amino-terminal fragment of M , = 20,000 that is the same size as the tryptic fragment obtained with intact cytoplasmic domain (not shown). The M , = 32,000 fragment copurifies during ion exchange chromatography on Mono Q with lower molecular weight polypeptide(s) of M , about 10,000 (Fig. 1). The activity of the M, = 32,000 fragment in inhibiting binding of ankyrin thus could result either from the fragment itself, lower molecular weight fragments, or a noncovalent complex of the M , = 32,000 fragment with the lower molecular weight fragment. The principal conclusion from these experiments is that cleavage of the cytoplasmic domain at a position approximately 290 residues from the amino terminus does not alter the ankyrin-binding site. The

I I

10 20 Anion Exchanger Fragments, pglml

FIG. 2. Effect of chymotryptic cleavage of the cytoplasmic domain of the anion exchanger to a M. = 32,000 fragment on its ability to inhibit binding of 1261-labeled ankyrin to ankyrin-depleted inside-out vesicles. Association of '251-labeled erythrocyte ankyrin with ankyrin-depleted inside-out vesicles was measured (see "Experimental Procedures") in the presence of increasing concentrations of either the intact cytoplasmic domain of the anion exchanger (0) or a chymotryptic digest of the cytoplasmic domain (0). The chymotryptic digest was prepared by incubation of the cytoplasmic domain (0.9 mg/ml) for 45 min a t 24 "C with 5 pg/ ml chymotrypsin in a buffer containing 10 mM sodium phosphate, 1.0 mM NaEDTA, 1 mM NaN,, 1.0 mM dithiothreitol, pH 7.4. The digestion was stopped by addition of diisopropylfluorophosphate, and the digest was applied to an anion exchange column (Mono Q ) . M, = 32,000 chymotryptic fragment eluted together with polypeptide(s) of M, = 10,000 (see Fig. 1, lane 2). REC, red blood cell.

sequence of the human erythrocyte anion exchanger includes aromatic residues at positions 278, 294, and 299 that are potential cleavage sites for chymotrypsin. The presence of a favorable aromatic-basic dipeptide at Phe2"-Are makes cleavage at this site especially likely.

Mild tryptic digestion of the cytoplasmic domain results in a fragment of M , = 20,000 as well as smaller fragments. The M , = 20,000 fragment is derived from the amino-terminal domain based on work by other investigators (reviewed by Low, 1986; Jay and Cantley, 1986) and immunoprecipitation results with peptide-specific antibodies (see below, Fig. 6). The sequence of the human erythrocyte anion exchanger contains two likely tryptic cleavage sites between L y ~ " ~ - P r o " ~ and ArglBO-SerlR1 that would produce fragments of appropriate molecular weight (Kaul et al., 1983).

Tryptic cleavage of the cytoplasmic domain reduced by over 20-fold its activity in inhibiting binding of ankyrin (Fig. 3). The partially purified M , = 20,000 fragment was even less active than the unfractionated digest (Fig. 3). The striking loss of activity following tryptic cleavage precluded use of this protease to prepare smaller active fragments of the cytoplasmic domain. Several interpretations are possible for the loss of binding activity following tryptic cleavage: 1) The ankyrin-binding site is located a t or near to the site of tryptic cleavage. 2) Contacts occur between ankyrin and the cyto-

1 2 3 100

d

2 60 >

c .- X 2 a 0 40 m a

25 50 75 100 125 Anion Exchanger Fragments, pglml

FIG. 3. Effect of limited tryptic cleavage of the cytoplasmic domain of anion exchanger on its ability to inhibit binding of 1251-labeled ankyrin to ankyrin-depleted inside-out vesicles. Association of '2511-labeled erythrocyte ankyrin with ankyrin-depleted inside-out vesicles was measured at 4 "C (see "Experimental Proce- dures") in the presence of increasing concentrations of either the intact cytoplasmic domain of the anion exchanger (O), an unfractionated tryptic digest of the cytoplasmic domain (O), or with partially purified amino-terminal M, = 20,000 tryptic fragment of the cytoplasmic domain (0). The tryptic digest was prepared by incubation of the cytoplasmic domain (0.2 mg/ml) for 60 min a t 24 "C with 2 pg/ ml trypsin in a buffer containing 10 mM sodium phosphate, 1 mM NaEDTA, 1 mM NaN3, 1 mM dithiothreitol, pH 7.4. The digestion was stopped by addition of diisopropylfluorophosphate, and the digest was applied to an anion exchange column of either DE53 cellulose (unfractionated digest) or Mono Q. The DE53 column was washed with 10 column volumes of digest buffer to remove trypsin and eluted with 0.4 M NaCl in the same buffer. Partially purified M, = 20,000 tryptic fragment was isolated by gradient elution of the Mono Q column. The left panel is a Coomassie Blue-stained SDS gel of the intact cytoplasmic domain (lane I ) , the total digest ( l a n e 2), and the partially purified M, = 20,000 fragment ( l a n e 3) . REC, red blood cell.


plasmic domain on either side of the tryptic cleavage site. 3) Cleavage by trypsin has an indirect effect on the binding site by altering the conformation of the cytoplasmic domain or preventing association of the domain into dimers.

Evidence in support of direct participation in the ankyrin- binding site of the region affected by tryptic cleavage is that ankyrin protects the cytoplasmic domain from digestion by trypsin (Fig. 4). The cytoplasmic domain alone is almost completely digested to the M , = 20,000 amino-terminal fragment by 1 pg/ml trypsin, while the cytoplasmic domain incubated with ankyrin remained intact a t up to 4 pg/ml trypsin (Fig. 4). Spectrin, in contrast to ankyrin did not protect the cytoplasmic domain from digestion (not shown). Thus, ability of ankyrin to protect the cytoplasmic domain from proteolysis is likely to result from association of ankyrin with the cytoplasmic domain and not from simple competition as an alter- native substrate for trypsin.

Activity of Synthetic Peptides and Peptide-specific Antibod- ies Corresponding to Sequence Flanking the Tryptic Cleavage Site-A simple possibility was that the ankyrin-binding site would be expressed in a relatively small peptide including the tryptic cleavage sites. A 39-amino acid peptide corresponding to residues 161-200 of the human anion exchanger was synthesized to test this hypothesis. The peptide was soluble under mild conditions and was tested for ability to inhibit binding of ankyrin to inside-out vesicles (Fig. 5). The synthetic peptide was almost completely ineffective in this assay even at concentrations as high as 60 p ~ , and exhibited less than 0.1% of the activity of the cytoplasmic domain. These results suggested that the ankyrin-binding site may involve interactions that are provided in the context of the intact protein but are reduced or absent from an isolated segment of the cytoplasmic domain.

Antibodies against specific regions of a polypeptide provide

Coomassie blue lmrnunoblot anti-a.exchanger

1 2 3 4 5 6 7 8 1 3 4 5 6 7 8

m m .. I .

I

-

Y

U I U

V

FIG. 4. Ankyrin protects the cytoplasmic domain of the anion exchanger from cleavage by trypsin. The cytoplasmic domain of the anion exchanger (4 p ~ ) was incubated in the presence and or absence of ankyrin (4 p ~ ) for 60 min at 4 "C in a buffer containing 0.1 M NaCI, 10 mM sodium phosphate, 1 mM NaEDTA, 1 mM NaN,, 1 mM dithiothreitol, 10% sucrose, pH 7.4. Samples were then digested for 30 min at 4 "C with trypsin at final concentrations of 1 pg/ml (lanes 3 and 4), 2 pg/ml (lanes 5 and 6) , and 4 pglml (lanes 7 and 8). The samples were analyzed by SDS electrophoresis and gels were either stained for protein with Coomassie Blue (left panel) or the polypeptides transferred to nitrocellulose paper and immunoblot- ted with antibody against the anion (anti-a.) exchanger (rightpanel). The cytoplasmic domain of the anion exchanger is in lane 1, and ankyrin is in lane 2. Arrows point to the positions of the intact cytoplasmic domain and to tryptic fragments of the cytoplasmic domain.

I I

I I

100 200 Anion Exchanger Polypeptides, pglml

FIG. 5. Effect on binding of 12"I-labeled ankyrin to ankyrin- depleted inside-out vesicles of a 39-amino acid synthetic peptide encompassing the tryptic cleavage site of the cytoplasmic domain. Binding of '251-labeled ankyrin to ankyrin-depleted inside- out vesicles was measured at 4 "C ("Experimental Procedures") in the presence of increasing concentrations of either the cytoplasmic domain (0) or a synthetic peptide (0). The synthetic peptide included amino acids from 161 to 200 of the human anion exchanger sequence (see Fig. 6). RBC, red blood cell.

potential probes to evaluate possible participation of an area within the sequence that may not be active in an isolated state. At this point in the study the complete sequence was available for mouse erythrocyte anion exchanger cDNA (Ko- pito and Lodish, 1985) and for mouse genomic DNA including intron sites (Kopito et al., 1987). Mouse and human erythrocyte anion exchangers exhibit substantial homology in the region of tryptic cleavage, and human erythrocyte ankyrin associates with inside-out vesicles from rodent erythrocytes (data not shown). These considerations suggested that peptides derived from the mouse sequence would be useful for isolation of peptide-specific antibodies that could inhibit binding of ankyrin.

Peptides were synthesized corresponding to three exons of the mouse anion exchanger sequence that flanked the tryptic cleavage site (see Fig. 6). These peptides were then coupled to agarose and used to isolate peptide-specific antibodies from antisera raised against the human cytoplasmic domain. The rationale for preparing antibody from a common pool of antisera raised against the intact human cytoplasmic domain rather than immunizing with individual peptides was: 1) The antibody isolated from this antisera would be guaranteed to recognize intact human sequence, while only some of the antibody prepared with individual peptides would have this feature. 2) I t would be possible to compare antibodies against different peptides more meaningfully if these were derived from the same immunized animals. 3) The antisera used to isolate peptide-specific antibody was known to contain antibody that inhibited binding of ankyrin to the anion exchanger (Fig. 8) and thus was likely to include antibodies that recognize the ankyrin-binding site.

Specificity of the peptide-specific antibodies was evaluated by immunoprecipitation of a tryptic digest of the cytoplasmic domain which contained the amino-terminal M , = 20,000

Ankyrin-binding Site HUMAN 1 1 8

TVLLDLOETSLAGVANOLLDRFlFEDOlRPODREELLRALLLKHS 162

HOUSE 1 3 2 iFiiGiAEiSiiGviiHiiDcFi~EDoiuPoDREEiikAiiiKRS

P E P T I D E A 176

HUMAN 163 I I 204 tr

HAGELEALGGVKPAVLTRSGDPSOPLLPOHSSLETOLFCEOG

HAEDiGNiEG~KPAViikSGGASEPiLPHOPSLETOLYCGOG MOUSE 177 2 1 8

P E P T I D E B

HUMAN 2 0 5 DGGTEGHSPSGILEKIPPDSEATLVLVGR

233

EGGs~GPSTsGTi.KiiiDiETiivivG~ MOUSE 2 1 9 24 7

P E P T I D E C

HUMAN 161 173 HSHAGELEALGGV P E P T I D E 1 I74 1 8 6

KPAVLTRSGDPSO P E P T I D E 2 194 219 SLETOLFCEOGDGGTEGHSPSGI LEK P E P T I D E 3

N tr (174 or 180)

C ’

FIG. 6. Peptides selected from sequences of mouse and human erythrocyte anion exchangers for isolation of peptide- specific antibodies from antisera raised against the human anion exchanger. Peptides A, B, and C were derived from mouse sequence while peptides 1, 2, and 3 were from the human sequence. Amino acids that are similar or identical between the two sequences are marked by a dot. The sequences were aligned using the Wisconsin computer software. A schematic model for the cytoplasmic domain with location of peptides and tryptic ( tr) cleavage site(s) is shown at the bottom.

fragment as well as two fragments that presumably are derived from regions of the sequence carboxyl-terminal to the tryptic cleavage site (Fig. 7). The expected specificity of the antibodies was as follows: 1) antibodies against sequence amino- terminal to the tryptic site should recognize the amino-terminal fragment but not other polypeptides; 2) antibodies against the region containing the tryptic cleavage site(s) should react with both amino- and carboxyl-terminal fragments; 3) antibody against the region carboxyl-terminal to the cleavage sites should not recognize the amino-terminal fragment. Experimental results support these predictions. Antibody against peptide A (homologous to human amino acids 118-162) reacted with the amino-terminal M, = 20,000 polypeptide but not with other polypeptides. Antibody against peptide B which contains the tryptic cleavage sites (homologous to human amino acids 163-204) reacted with the amino- terminal fragment as well as two other presumed carboxyl- terminal polypeptides. Antibody against peptide C which is carboxyl-terminal to the tryptic cleavage sites (homologous to human sequence 205-233) did not react with the amino- terminal M , = 20,000 polypeptide, but did react with the two presumed carboxyl-terminal polypeptides.

These peptide-specific antibodies were evaluated for ability to inhibit association of radiolabeled ankyrin with inside-out vesicles (Figs. 8 and 9). Affinity-purified Ig against the intact cytoplasmic domain inhibited 50% of ankyrin binding at 1.5 pg/ml while nonimmune Ig had no effect. Antibodies against peptide A and peptide B were both active in inhibiting bind-

of the Anion Exchanger 9669

1 2 3 4 5 6 A B

20- 16-

13-

A B A 0 A 0

FIG. 7. Characterization of domain specificity of antibodies against peptides A, B, and C isolated from antisera raised against the cytoplasmic domain of the human anion exchanger. Affinity-purified Ig against peptides A, R, and C (see Fig. 6) were isolated from antisera raised against the cytoplasmic domain of the human anion exchanger (see “Experimental Procedures”). The ability of these peptide-specific antibodies to immunoprecipitate the cytoplasmic domain of the anion exchanger (lane A ) in each panel) or polypeptides in a tryptic digest of the cytoplasmic domain (lane B in each panel) was examined. Antibodies against either the entire cytoplasmic domain (panel 2) , peptide A (panel 3) , peptide B (panel 4) , peptide C (panel 5), or nonimmune Ig (panel 6) were incubated in a 0.14-ml volume for 90 min at 4 “C at a concentration of 0.4 pg/ ml Ig with ’“I-labeled cytoplasmic domain (22 pg/ml, 780,000 cpm/ pg) or tryptic digest of the cytoplasmic domain (27 pg/ml, 640,000 cpm/pg) in a buffer containing 150 mM NaC1, 10 mM sodium phosphate, 1 mM NaEDTA, 1 mM NaNs, 0.2% Triton X-100, 10 mg/ml bovine serum albumin, pH 7.4. The radiolabeled proteins were pread- sorbed with protein A-bearing staphylococci prior to assay in order to minimize nonspecific binding. Protein A-bearing staphylococci (0.5 pl packed cell volume) were added to each tube to adsorb Ig, and the incubation continued for an additional 60 min at 4 “C. The samples were diluted with 3.5 ml of a buffer containing 2 M urea, 1% Triton X-100,O.l M glycine, pH 7.0, and the staphylococci with adsorbed Ig collected by centrifugation for 10 min at 2,500 X g. The staphylococci were washed once more in this buffer and analyzed by SDS electrophoresis followed by autoradiography. The cytoplasmic domain and tryptic digest before inmunoprecipitation are in panel 1. The tryptic digest of the cytoplasmic domain was prepared as described in Fig. 4.

ing, with half-maximal inhibition at about 15 pg/ml. Antibody against peptide C was much less active, with less than 15% inhibition at 24 pg/ml. The relatively low activity of peptide- specific antibodies compared with antibody against the intact domain could be explained if the intact domain has antigenic sites encompassed by residues 118-204 that are not repro- duced in the isolated peptides. Alternatively, the binding site may include regions of the sequence that are not represented by residues 118-204. These alternatives are not mutually exclusive and were not resolved in this study. Another consideration is that antibodies against multiple sites in the intact domain may have a cooperative effect that cannot be repro- duced by antibodies with a narrow specificity. The possibility of cooperativity between antibodies against peptides A and B was evaluated by determining the effects of combining the antibodies (Fig. 9). No evidence was found for cooperativity because inhibition by the combination of antibodies was approximately the sum of inhibition obtained with isolated antibodies.

I t is important, in comparing the activity of peptide-specific antibodies, to also know how well these antibodies recognize the native anion exchanger in membranes. The association of antibodies against peptides A, B, and C with inside-out vesi-


[+L"-L - - A

L

10 20

lg, pg/ml

FIG. 8. Effect on binding of '261-labeled ankyrin to ankyrin- depleted inside-out vesicles of antibodies against the cytoplasmic domain of human erythrocyte anion exchanger, and against peptides A, B, or C. Binding of 1251-labeled ankyrin to ankyrin-depleted inside-out vesicles was measured at 24 "C (see "Ex- perimental Procedures") in the presence of various concentrations of affinity-purified antibodies with the following specificities: entire cytoplasmic domain (O), peptide A (W), peptide B (O), peptide C (O), and nonimmune Ig isolated from a protein A affinity column (A). Antibodies were preincubated with membrane vesicles for 60 min at 24 "C before addition of ankyrin. RBC, red blood cell.

10 20

lg, pglml FIG. 9. Effect of a combination of antibodies against pep-

tides A and B on binding of 'Z61-labeled ankyrin to ankyrin- depleted inside-out vesicles. Binding of lZ5I-labeled ankyrin to ankyrin-depleted inside-out vesicles was measured at 24 "C (see "Ex- perimental Procedures") in the presence of various concentrations of affinity-purified antibodies against peptide A (0), peptide B (O), or a 1:l mixture of antibodies against peptide A and peptide B (0). Antibodies were incubated with membrane vesicles for 60 min at 24 "C before addition of radiolabeled ankyrin. RBC, red blood cell.

cles was compared by incubating vesicles with Ig under assay conditions followed by addition of Iz5I-labeled protein A and isolation of vesicles to detect membrane-associated Ig (Table I). The three antibody preparations were approximately equal in extent of association with vesicles. Antibody against peptide C associated as least as well if not better than the other

TABLE I Comparison of ability of peptide-specific antibodies against the anion

exchanger to associate with inside-out vesicles

Ig specificity '"I-Labeled protein A bound"

cpm X 10" Experiment 1 Cytoplasmic domain 2.2 * 0.3

Peptide A 1.9 f 0.02 Peptide B 0.6 f 0.05 Peptide C 2.4 f 0.03

1.7 f 0.03 1.8 f 0.05 2.1 f 0.10 0.3 f 0.04 0.2 f 0.05 0.7 f 0.01

Experiment 2 Peptide A Peptide B Peptide C Peptide 1 Peptide 2 Peptide 3

" Affinity-purified antibodies (2 pglml) against the intact cytoplasmic domain and various peptides were incubated with ankyrin- depleted inside-out vesicles (15 pglml) for 60 min at 4 "C under the same buffer conditions as employed in ankyrin binding assays (see "Experimental Procedures"). lZ5I-Labeled protein A (0.5 pglml; 2 X lo7 cpm/pg) was then added for an additional 90 min. Vesicles with adsorbed Ig and radiolabeled protein A were collected by sedimenta- tion through 20% sucrose in Microfuge tubes as described for the ankyrin binding assay. The amount of radiolabeled protein A adsorbed to vesicles was measured and was corrected for the amount of protein A adsorbed with nonimmune Ig. Data are expressed as the mean * the half range of duplicate (experiment 1) or triplicate (experiment 2) determinations.

100 I I

I I 1

10 20

1% Clglml

FIG. 10. Effect on binding of '2SI-labeled ankyrin to ankyrin-depleted inside-out vesicles of peptide-specific antibodies against peptides 1, 2, or 3 derived from the sequence of the human anion exchanger. Binding of lZ5I-labeled ankyrin to ankyrin-depleted inside-out vesicles was measured at 24 "C (see "Ex- perimental Procedures") in the presence of various concentrations of affinity-purified antibodies with the following specificities: peptide 1 (W), peptide 2 (0), peptide 3 (a). Antibodies were incubated with membrane vesicles for 60 min at 24 "C before addition of radiolabeled ankyrin. RBC, red blood cell.

Ankyrin-binding Site of the Anion Exchanger 9671

antibodies in four separate experiments. The weak activity of antibody against peptide C thus suggests that these antibodies can associate with the cytoplasmic domain but have no effect on binding of ankyrin. Antibodies to peptide C provide an important control for the possibility that any antibody that binds to the cytoplasmic domain will nonselectively inhibit binding of ankyrin to vesicles.

Further mapping in the region of the tryptic cleavage site was approached using peptides derived from the human sequence and encompassing residues 161-173 (peptide l), 174- 186 (peptide 2), and 194-219 (peptide 3) (see Fig. 6). These peptides were used to isolate peptide-specific Ig from the same antisera used for peptides A-C. Antibodies against peptides 1 and 3 were relatively inactive in inhibiting binding of ankyrin to inside-out vesicles. In contrast, antibodies against peptide 2 were the most active of all 6 peptides with 50% inhibition a t 8 pg/ml (Fig. 10). Antibodies against peptides 1 and 2 were approximately equal to each other in ability to bind to inside- out vesicles, while antibodies to peptide 3 bound about twice as well as antibodies against peptides 1 or 2 (Table I). Anti- bodies against peptide 2 associated with inside-out vesicles to about 25% of the extent of antibodies against peptides A-C. The amount of binding of the antibodies was closely related to the size of the peptides, and presumably is proportional to the number of antigenic sites available within the polypeptide sequence. Antibody against peptide 2, when normalized for the amount of antibody bound to vesicles, was by far the most active with at least 10-fold higher specific activity than antibody against peptides A or B. These results implicate a rather small region of 12 amino acids from positions 174-186 as performing an important function in the ankyrin-binding site of the anion exchanger.

DISCUSSION

This report describes initial efforts to map the ankyrin- binding site of the cytoplasmic domain of the erythrocyte anion exchanger. The conclusions are that this site is likely to involve a fairly extended sequence in the midregion of the cytoplasmic domain and requires interactions that are not provided by isolated peptides. The sequence of amino acids in the region of residues 174-186 was implicated as partici- pating in the binding site by several experiments. Limited tryptic cleavage in the midregion of the cytoplasmic domain at residues 174 or 181 nearly abolished ability of the cytoplasmic domain to inhibit binding of ankyrin to the anion exchanger. Moreover, ankyrin protected the cytoplasmic domain from tryptic digestion. However, the sequence comprising the tryptic site is not sufficient for high affinity binding of ankyrin. A 39-amino acid peptide (residues 161-200) that includes the tryptic cleavage sites did not inhibit binding of ankyrin to the anion exchanger. Further evidence for a complex ankyrin-binding site is that peptide-specific antibodies against two different, noncontiguous regions (residues 118- 162 and 174-186) both inhibited binding of ankyrin and were only 10-20% as effective as antibody against the entire cytoplasmic domain. Finally, the ankyrin-binding site of the anion exchanger did not renature following SDS electrophoresis and transfer to nitrocellulose paper even though spectrin did recover ability to bind ankyrin under the same conditions.

A potential complicating factor in elucidating the ankyrin- binding site of the anion exchanger is the possibility that the exchanger is active only as a dimer. Thus, any perturbation that interfered with dimerization would indirectly reduce ability to associate with ankyrin. The anion exchanger can associate with ankyrin as a monomer, however, because a complex has been isolated from erythrocyte membrane skeletons con-

taining 1 molecule of ankyrin and 1 molecule of the exchanger (Bennett, 1982). Another consideration is the possibility that antibodies or protease cleavage indirectly inhibit binding due to induction of a conformational change in the anion exchanger. In support of such a possibility is the observation that the cytoplasmic domain of the anion exchanger under- goes a major pH-dependent conformational change (Low et al., 1984). I t will be important in future work to obtain a region of the anion exchanger that actually expresses an ankyrin-binding site in order to determine the minimum number of residues required for this protein-protein interaction.

Localization of the ankyrin-binding site close to the midregion of the cytoplasmic domain was predicted earlier (Low et al., 1984). A highly conserved region has been noted in the sequence at residues 157-177 (numbering based on mouse erythrocyte sequence) of mouse (Kopito and Lodish, 1985), human (Kaul et al., 1983), and avian (Cox and Lazarides, 1988; Kim et al., 1988) erythrocyte anion exchangers as well as nonerythroid anion exchangers (Demuth et al., 1986; Alper et al., 1988). This conserved sequence has been suggested to correspond to the ankyrin-binding site (Kopito and Lodish; 1985; Demuth et al., 1986; Cox and Lazarides, 1988; Alper et al., 1988). The results of this study in general support these predictions, although with some refinements. I t is likely that the conserved region participates in the binding site, because antibodies against a peptide encompassing this region inhibit binding of ankyrin to vesicles (Fig. 8). The conserved region alone cannot function as an ankyrin-binding site since the amino-terminal tryptic fragment containing this sequence is only 5% as active as the intact cytoplasmic domain in inhibiting binding of ankyrin to inside-out vesicles (Fig. 4). The fact that the sequence of diverse anion exchangers is highly conserved in the region 157-177 may reflect another interaction common to anion exchangers such as self-association to form homodimers. Caution should therefore be exercised before assuming that the nonerythroid forms of the anion exchanger actually bind to ankyrin simply based on this region of homology.

The midregion of the anion exchanger cytoplasmic domain has properties including sensitivity to proteases and predicted conformation of a ,6 turn suggesting that this area represents the junction between two subdomains (Low, 1986; Kopito et al., 1987). The ankyrin-binding site may involve several contacts which are mediated by each of the subdomains of the cytoplasmic domain as well as the hinge sequence. Such a multipoint attachment would explain the loss of activity following tryptic cleavage, because individual contacts may be orders of magnitude weaker than simultaneous interactions.

A motivation in performing this study was the potential to discover a relatively short continuous sequence that would provide a “signature” for identification of ankyrin-binding regions in other proteins. As discussed above, such a small, well-defined sequence is not likely. Ankyrin-binding proteins include the Na/K-ATPase (Nelson and Veshnock, 1987) and the voltage-dependent sodium channel (Srinivasan et al., 1988). Mouse anion exchanger residues 133-221 have some similarity to the a subunit of sheep ATPase (residues 867- 959) and to the sodium channel of electric eel (residues 1587- 1673) (not shown). The homologies of these proteins with the anion exchanger are not striking, however. The diversity in sequence of this group of ankyrin-binding proteins suggests that their association with ankyrin may have developed in- dependently through convergent evolution. A prediction based on independent evolution of ankyrin-binding proteins is that


these proteins may bind to different sites on the ankyrin molecule. A similar situation may occur with the LDL recep- tor and the multiple other membrane receptors that participate in clathrin-mediated endocytosis but have no homology to each other beyond the common feature of a cytoplasmic domain (Goldstein et al., 1985).

The approach of using antibodies to map protein recogni- tion sites has certain inherent limitations which have been noted above. These results do suggest specific areas of the anion exchanger sequence that will be of interest to explore using expression of appropriate regions of the cDNA. The areas of the human sequence from 174 to 186 and 118 to 162 are promising targets to study effects of deletions and muta- tions on binding of ankyrin. Ultimately the most satisfying and complete understanding of the ankyrin-binding site will come from solving by x-ray crystallography the structure of an ankyrin-cytoplasmic domain complex, as has been possible for hemoglobin and the cytoplasmic domain (Walder et al., 1984).

REFERENCES

Alper, S., Kopito, R., Libresco, S., and Lodish, H. (1988) J. Biol.

Appell, K. C., and Low, P. S. (1981) J. Biol. Chem. 256,11104-11111 Becker, P. S., and Lux, S. E. (1985) Clin. Hematol. 14, 15-43 Bennett, V. (1982) Bwchim. Biophys. Acta 689,475-484 Bennett, V. (1983) Methods Enzymol. 9 6 , 313-324 Bennett, V. (1985) Annu. Reu. Biochem. 54,273-304 Bennett, V., and Stenbuck, P. J. (1979) Nature 280,468-473 Bennett, V., and Stenbuck, P. J. (1980) J. Biol. Chem. 2 5 5 , 6424-

Cox, J., and Lazarides, E. (1988) Mol. Cell. Biol. 8, 1327-1335 Davis, J., and Bennett, V. (1983) J. Bid. Chem. 2 5 8 , 7757-7766 Davis, J., and Bennett, V. (1984) J. Biol. Chem. 2 5 9 , 13550-13559 Demuth, D. R., Showe, L. C., Ballantine, M., Palumbo, A,, Fraser, P.

J., Cioe, L., Rovera, G., and Curtis, P. J. (1986) EMBO J. 5,1205- 1214

Drenckhahn, D., and Merte, C. (1987) Eur. J. Cell Biol. 45 , 107-115

Chem. 263 , 17092-17099

6432

Drenckhahn, D., Schluter, K., Allen, D., and Bennett, V. (1985)

Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971) Biochemistry

Goldstein, J., Brown, M., Anderson, R., Russel, P., and Schneider,

Hall, T. G., and Bennett, V. (1987) J. Biol. Chem. 2 6 2 , 10537-10545 Hargreaves, W. R., Giedd, K. N., Verkleij, A,, and Branton, D. (1980)

Hunter, W., and Greenwood, F. (1962) Nature 194,495-496 Jay, D., and Cantley, L. (1986) Annu. Reu. Biochem. 55,511-538 Kaul, R., Murthy, P., Reddy, G., Steck, T., and Kohler, H. (1983) J.

Biol. Chem. 2 5 8 , 7981-7990 Kim, H., Yew, N., Ansorge, W., Voss, H., Schwagner, C., Vennstrom,

B., Zenke, M., and Engel, J. (1988) Mol. Cell. Biology 8, 4416-4424 Koob, R., Zimmermann, M., Schoner, W., and Drenckhahn, D. (1987)

Eur. J. Cell Biol. 4 5 , 230-237 Kopito, R., and Lodish, H. (1985) Nature 3 1 6 , 234-238 Kopito, R., Anderson, M., and Lodish, H. (1987) J. Biol. Chem. 262,

Korsgren, C., and Cohen, C. (1988) J. Biol. Chem. 263 , 10212-10218 Low, P. (1986) Biochim. Biophys. Acta 8 6 4 , 145-167 Low, P., Westfall, M., Allen, D., and Appell, K. (1984) J. Biol. Chem.

Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951)

Marchesi, V. T. (1985) Annu. Reu. Cell Bid. 1, 531-561 Murthy, S. N., Liu, T., Kaul, R. K., Kohler, H., and Steck, T. L.

Nelson, W. J., and Veshnock, P. J. (1987) Nature 328,533-535 Nigg, E., and Cherry, R. J. (1979) Nature 277,493-494 Pasternack, G. R., Anderson, R. A., Leto, T. L., and Marchesi, V. T.

Schuster, V., Bonsib, S., and Jennings, M. (1986) Am. J. Physiol.

Srinivasan, Y., Elmer, L., Davis, J., Bennett, V., and Angelides, K.

Steck, T. L. (1978) J. Supramolec. Struct. 8, 311-324 Steck, T. L., Ramos, B., and Strapazon, E. (1976) Biochemistry 15,

Wagner, S., Vogel, R., Lietzke, R., Koob, R., and Drenckhahn, D. (1987) Am. J . Physiol. 2 5 3 , F213-F221

Walder, J. A., Chatterjee, R., Steck, T. L., Low, P. S., Musso, G. F., Kaiser, E. T., Rogers, P. H., and Arnone, A. (1984) J. Bid. Chem.

Science 2 3 0 , 1287-1289

10,2606-2617

W. (1985) Annu. Rev. Cell Biol. 1 , 1-40

J. Bid. Chem. 2 5 5 , 11965-11972

8035-8040

259 , 13070-13076

J. Biol. Chem. 193 , 265-275

(1981) J. BWl. C k m . 256 , 11203-11208

(1985) J. Bid. Chem. 260,3676-3683

251,347-355

(1988) Nature 3 3 3 , 177-180

1154-1161

2 5 9 , 10238-10246

Mapping the Ankyrin-binding Site of the Human Erythrocyte Anion ...

Documents

Transcript of Mapping the Ankyrin-binding Site of the Human Erythrocyte Anion ...