D9160CCAd01

8/8/2019 D9160CCAd01

1/5

T H E J O U R N A LF BIOLOGICAL CHEMISTRYQ 1986 by The Am erican Society of Biological Chemists, Inc Vol. 261, No. 28. Issue of October 5, pp. 13362-13366,1986Printed in U .S.A.

Structure of the Spectrin-Actin Binding Site of ErythrocyteProtein 4.1*(Received for publication, May 8, 1986)

Isabel Correas, David W. Speicher, and Vincent T. MarchesiSFrom the Yale University School of Medicine, Department of Pathology, New Haven, Connecticut06510

The complete primary structure of the functional siteof erythrocyte protein 4.1 involved in spectrin-actinassociations has been determined. The sequence of thisdomain, which contains 67 amino acids and has a mo-lecular mass of 8045 daltons, has been obtained byNH2-terminal sequence analysis of an 8-kDa chymo-tryptic peptide, three endoproteinase lysine C-cleavedpeptides and two peptides obtained by Staphylococcusaureus protease V 8 cleavage. All peptides includingthe 8-kDa domain peptide were purified by reverse-phase high performance liquid chromatography. Anti-bodies against two different synthetic peptides of the8-kDa domain are able to inhibit the association be-tween protein4. , spectrin, and F-actin, corroboratingthat the 8-kDa domain is responsible for the formationof a ternary complex.A computer search of the 8-kDa sequence with theNational Biomedical Research Foundation databasedid not detect any significant homologies to knownsequences. Protein 4.1 is not related to any knownproteins and may represent a new protein superfamily.

The red cell membrane has been chosen as a model forstudying membrane-cytoskeleton interactions because of itsapparent simplicity relative to other cells. The membraneskeleton is composed of a two-dimensional protein networkof which spectrin, a long, rod-like molecule composed of twonon-identical ubunits, is the major component. Spectrindimers further self-associate into tetramers and ligomers (1)and interact with protein 4.1 and actin (2-5) to form thesubmembranous lattice. This anastomosing network is linkedto the membrane by two proteins. One, designated band 4.1based on its position on SDS gels, binds to the nd of spectrindistal to the self-association site but proximal to thebindingsite for filamentous actin ( 5 , 6); this protein attaches hespectrin-actin network to the cytoplasmic face of the mem-brane through specific associations with at least two differenttransmembrane glycoproteins (7, 8). The other membranelinking protein, protein 2.1 (ankyrin) (9), binds to a part ofthe @ subunit near the self-association end of the spectrindimer (10) and links it to the cytoplasmic domain of band 3(11).In addition to linking the spectrin-actin complex to themembrane, protein 4.1 greatly enhances the affinity of spec-

* This research was supported by National Institutes of HealthGrants GM21714-12 and AM27932-06. The costs of publication ofthis article were defrayed in part by the payment of page charges.This article must therefore be hereby marked advertisement inaccordance with 18U.S.C. Section 1734 solely o indicate this fact.$ To whom all correspondence should be addressed.The abbreviations used are: SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; EL, endoproteinase lysine-C;SV8,S. aureus protease V8; KLH, keyhole limpet hemocyanine; HPLC,high performance liquid chromatography.

trin for F-actin, forming stable ternary complexes as demon-strated by several different methods (2-5). Although the pre-cise nature of the ternary complex is not yet well understood,it is likely to be of general significance since proteins immu-nologically related to erythrocyte spectrin and protein 4.1have been described in numerous non-erythroid cells (forreviews, see Refs. 12 and 13).

Previous work in this laboratory has identified a peptide oferythrocyte protein 4.1 that mimics the capacity of the 4.1molecule to enhance the association between spectrin andactin. This fragment is incorporated into the ernary complexin approximately stoichiometric amounts and its activity iscomparable to that f the intact4.1 molecule on a molar basis(14).This 8-kDa peptide has been located within the 10-kDaregion of erythrocyte protein 4.1 which had previously beenshown to contain a CAMP-dependent phosphorylation site(15, 16).

This report describes the further characterization of the 8-kDa fragment including the purification and analysis of pep-tides derived from two protease cleavages of the 8-kDa peptideand thecomplete sequence of this segment. Antibodies raisedagainst two synthetic peptides whose sequences derived fromthe 8-kDa peptide sequence are able to inhibit the interactionbetween protein 4.1, spectrin, and actin.

MATERIALS ANDM E T H O D SIsohtion of a Complex-promoting Peptide of Protein4.1Twodifferent methods have been followed for the isolation of the 8-kDadomain of erythrocyteprotein 4.1. In the first, the 8-kDa activepeptide was sedimented as a complex with spectrin and F-actin.Protein 4.1 (2 mg) wasdigested with a-chymotrypsin at 1:260 enzymeto substrate ratio and incubated with spectrin (1 2 mg) and F-actin (8mg) for 90 min at 0 C. After centrifugation at 150,000X g for 30 minon a SW50.1 rotor, the 8-kDa peptide obtained in the pellet fractionwas chromatographed on gel filtration columns (two 30-cm Bio-Si1TSK-400 columns, two 30-cm Bio-Si1TSK-250 columns, and one 30-cm Bio-Si1 TSK-125 column, all in tandem, Bio-Rad) equilibrated in8 M urea, 0.2 M Tris, 10mM 2-mercaptoethanol, pH 7.0. Effluent wasmonitored at 280 nm and flow rates of 0.3 ml/min were used. The 8-kDa fraction was further purified by reverse-phase chromatography

on a 4.6 X 250-mm, C-4, RP-304 column (Bio-Rad) equilibrated with0.1% trifluoroacetic acid. Elution of the peptide was achieved using alinear gradient of0-60% acetonitrile in the same buffer. Peptideswere detected by absorbance at 215 nm and by tryptophan fluores-cence (excitation 280 nm, emission filter 370 nm).A second method for isolation of the 8-kDa peptide involved aninitial size separation of the 4.1-derived peptides (8 mg) produced bymild chymotrypsin digestion by HPLC gel filtration, followed byfurther purification of the resulting 8-kDa peptide by reverse-phasehigh performance liquid chromatography. The columns and bufferconditions have been described above.Lysine-specificProteolytic Cleavage-The 8-kDa peptide (10 nmol)was incubated for 17 h at 37 C in 200 mM Tris-HC1, 1 mM EDTA,0.02% sodium azide, pH 7.8, with endoproteinase lysine-C (Boehrin-ger Mannheim) using an enzyme to substrate ratio of 1: lOO . Thereaction was terminated by titrating the samples to pH 2.0 withtrifluoroacetic acid. The sample was analyzed by reverse-phase chro-13362

8/8/2019 D9160CCAd01

2/5

Spectrin-Actin Binding Site f Erythrocyte Protein4.1matography under the conditions described above.Staphylococcus aureus Protease V8 (SVS) Digestion-The 8-kDapeptide (17 nmol) was incubated for 16 h a t 37 "C in 300 mMNH,HCO,, 50mM Tris, 5 mM EDTA, pH 7.8, with protease V8 usingan enzyme to substrate atio of 1:30. The reaction was terminated bytitrating the sample to pH 2.0 with trifluoroacetic acid. The samplewas then treated as described for the lysine-C digest.Amino Acid Composition and Sequence Analysis-Peptides werehydrolyzed at 100 "C for 20 h in uacuo in 6 N HCl containing 0.2%phenol. Amino acid analysis was performed on aDionex D-500 aminoacid analyzer using a ninhidrin detection system.Automated sequence analysis of the peptides was performed on aBeckman 890C Sequencer using th e 0.1 M Quadrol program 030176.3 mg of Polybrene were added to thespinning cup and a blank cyclewas run before addition of peptide samples. Conversion of the thia-zolinone-derivatives to phenylthiohydantoin derivatives was carriedout in a Sequemat P-6 autoconverter using 1 N methanolic HCl.Phenylthiohydantoin derivatives were identified by HPLC as previ-ously described (17).Nomenclature of Peptides-Peptides were named based on the typeof cleavage used endoprotease lysine-C (EL); S. aureus protease V8(SV8). Each of these two sets of peptides have been numbered in theorder which they occur in the complete sequence starting at th eamino-terminal.Preparation of Antibodies-Based on the 8-kDa peptide sequencethat we obtained, two synthetic peptides were made with additionalcysteine residues in their COOH-terminal ends (Peninsula Labora-tories). Peptide A included amino acids 1-15 and peptide B aminoacids 46-59 (Fig. 4). Both peptides were independently coupled toKLH (keyhole limpet hemocyanine) using the bifunctional reagentrn-maleimidobenzoylsulfosuccinimide estern 50 mM phosphatebuffer, 1 mM EDTA, pH 7.0.New Zealand White rabbits wereimmunized by subcutaneous and intramuscular injections of 1mg ofKLH-peptide in complete Freund's adjuvant. They were boosted twiceat 4-week intervals using 0.5 mg of KLH-peptide with incompleteadjuvant. IgG-enriched fractions were obtained from 40% saturationammonium sulfate precipitates. Antibodies against KLH were ad-sorbed to Sepharose CL-4B beads to which KLH was linked viacyanogen bromide (18), and he resulting unbound fraction waspurified further on a Sepharose CL-4B column to which hexanedi-amine was linked via cyanogen bromide (18) and activated with m-maleimidobenzoylsulfosuccinimideo get the synthetic peptide cou-pled. Antibodies were eluted from the column with 1M acetic acid in0.15 M NaCl, pH 4.0, dialyzed against phosphate-buffered saline, andconcentrated.

Monovalent Fab antibodies were produced by cleavage of antibod-ies with papain by the method described by Porter (19).Sedimentation Assay-Protein 4.1, spectrin, and F-actin were in-cubated and sedimented as earlier described (14). In the cases wherethe monovalent Fab antibodies were present, protein 4.1 (8 pg) andthe Fabragment (15pg) were reincubated overnight on ice. Sampleswere processed as described elsewhere (14).PAGE and Immumblotting-SDS-PAGE was carried out as de-scribed (20).SDS-PAGE proteinswere transferred onto nitrocellulosepaper and immunoblotting was performed according to Towbin et al.(21).

RESULTS AND DISCUSSIONFig. 1 represents typical profiles of the reverse-phase puri-fication of the 8-kDa domain of protein 4.1 following wo

different approaches. In he first, the active peptide wassedimented in a complex with spectrin and F-actin. An a-chymotryptic digest of protein 4.1 was ncubated with spectrinand F-actinas described under "Materials and Methods," t oform astable ernary complex. The pellet fraction whichcontained spectrin, F-actin, and the8-kDa peptide was sepa-rated by gel filtration. The 8-kDa peak was further purifiedby HPLC reverse-phase chromatography and three peptidecomponents were identified, as indicated in Fig. lA. Thesepeptides are very similar in theiramino acid composition(data not shown) and probably represent differing sites ofcleavage by chymotrypsin. The presence of all three peptidesin the sedimentable complex indicates that all three formsretain complex-promoting activity. Peak 3 was used for the

T

B

a3

I

BEwk0.2

0.1

0 -

-r

FIG.1. Reverse-phase purificationof the 8-kDa domain ofprotein 4.1.Samples were eluted from a 4.6 X 250-mm, C-4, R P-304 column (Bio-Rad) equilibrated with 0.1% trifluoroacetic acid, pH2.0 using a linear gradient of 0-60% acetonitrile in 0.1% trifluoroac-etic acid, pH 2.0. The gradient shape is indicated. The differencesbetween the earliest steps of the purification procedure for panels Aand B are explained under "Materials and Methods." Peak 3 wasused for sequence analysis.sequencing studies presentedbelow.

The second method for isolation of the 8-kDa peptideinvolved an initial size separation of the 4.1-derived peptidesproduced by mild a-chymotrypsin digestion by HPLC gelfiltration in the presence of 8 M urea, as previously described(14). The resulting 8-kDa fraction was further purified byreverse-phase chromatography, as indicated in Fig. 1B. Thepeaks designated 1,2, nd 3 correspond to theimilarly labeledpeaks isolated by the complex-formation method. The identitywith peaks in Fig. lA is based on similar amino acid compo-

8/8/2019 D9160CCAd01

3/5

13364 Spectrin-Actinindingite of Erythrocyterotein 4.1

-IG. 2. Reverse-phase separation of 8-kDa peptides aftercleavage with endoproteinaseysine-C. 10nmol of 8-kDa peptide(peak 3, in Fig. 1)were digested as described under Materials andMethods. The column and buffer conditions are similar to thoseshown in Fig. 1. Peptides EL-1,EL-2, and EL-3 were used forsequence analysis. The figure shown is a representative chromato-gram.TABLE

Amino acid compositionof 8-kDa peptides cleaved withendoproteinase lysine-CAmino acid compositions were determined as described underMaterials and Methods using 20 h hydrolysis. Numbers in paren.theses are number of residues indicated by the complete sequence.

CYsThrASPSerGluProGlYAlaValMetIleLeuTYrPheHisLYsArgTr pTotal

EL-1

2.49 (2 )1.38 (1)

1.00 (1)

0.76 (1)0.89 (1)1.97 (2)2.20 (3)

11

EL-22.00 (2 )1.99 (2)3.16 (3)2.93 (3)

1.51 (1)0.78 (1)

0.95 (1)0.88 ( I )0.96 (1)NA (1)

16

EL-3

1.01 (1)2.08 (2)0.88 (1)

1.00 (1)1.01 (1)1.01 (1)0.97 (1)

8

P e a k 15.00 (5)1.41 (1)3.30 (3)1.33 (1)

0.51 (1)1.95 (2)3.48 (4)0.82 (1)1.06 (1)3.37 (4)2.52 (3)

26

3.14 (3)3.45 (4)

1.07 (1)

1.24 (2)0.93 (1)2.47 (2)3.99 (4)

17

MHmsFIG. 3. Reverse-phase separation of 8-kDa peptides aftercleavage with S. aureus protease V 8 . Approximately 17 nmol of8-kDa peptide (peak 3, in Fig. 1) were digested as described underMaterials and Methods and separated under the conditions shownin Fig. 1.Peptides SV8-1and SV8-2 were used for sequence analysis.

TABLE1Amino acid compositionof 8-kDa peptides cleaved with S. aurew V8protease

Amino acid compositions were determined as described underMaterials and Methods using 20 h hydrolysis. Numbers in paren-theses are number of residues indicated by the complete sequence.SV8-1 SV8-2 Peak 1 Peak 2 Peak 3 Peak 4

CYSAsp 1.12 (1) 1.19 (1) 5.12 (5) 5.10 (5) 2.13 (2)Thr 0.99 (1)Ser 4.56 (4) 2.47 (2 ) 4.05 (4) 1.92 (2 ) 3.87 (4)1.03 (1)Glu 1.30 (1) 2.45 (2) 6.58 (6) 7.35 (7 ) 1.07 (1) 3.60 (3)Pro 3.74 (4)G ~ Y

3.76 (4)Ala 1.20 (1) 1.07 (1)Val 2.31 (1)Met 1.00 (1) 1.05 (1)0.90 (1) 0.93 (1)Ile 1.48 (2 ) 2.83 (4) 1.45 (2) 0.82 (1)Leu 1.03 (1) 1.18 (1) 4.00 (4) 3.96 (4) 2.00 (2 )TYr 0.80 (1) 0.70 (1)Phe 0.92 (1) 0.94 (1)His 1.77 (2 )1.00 (1) 1.08 (1) 2.96 (3) 2.00 (2) 0.89 (1)Lys 1.87 (2 ) 1.18 (1) 3.76 (4) 5.28 (6) 1.68 (2) 2.60 (3)

1.32 (1) 1.39 (1)

-4% 1.67 (2 ) 2.58 (3) 2.55 (3) 1.75 (2 )Trp NA (1) NA (1)Total 70 30 40 10 27

Position 33-43 44-590-67-267-43 Position 41-478-64-30-401-401-67NA, not analyzed. a NA, not analyzed.

sitionsand etention ime on the reverse-phase column.Amino acid compositions of the rest of the peaks in Fig. 1Bindicated tha t they were fragments derived from the basic 30-kDa domain of protein 4.1 (da ta not given). The absence ofthese 30-kDa derived peptides in Fig. L4 further demonstratedthe specificity of the sedimentation assay. Peak 3 (Fig. 1B)was also used for the sequencing studies since i t was identicalto the orresponding peptide isolated by the alternatemethoddescribed above.The intact 8-kDa domain (peak 3, Fig. L 4 ) was subjectedto automated Edman degradation. These results are summa-rized in Fig. 4.Approximately half of the total sequence ofthe domain was determined by sequence analysis of the intactdomain. The sequence of this region of the domain wasverified by sequence analysis of peak 3 in Fig. 1B (data notshown). The NHz-terminal sequence of these two fractions(peaks 3 in Fig. 1, A and B ) further confirmed their identityas suggested above based on amino acid composition andretention time on the reverse-phase column.

The reverse-phase separation of peptides obtained from the8-kDa peptide following cleavagewith endoproteinase lysine-C is shown in Fig. 2 and amino acid compositions are listedin Table I. Peptides whose amino acid compositions fit theknown sequence (NHz-terminal sequence of the intac t do-main) were not treated further while those whose structureswere needed to document the complete sequence of the 8-kDapeptide were sequenced to the COOH-terminal amino acid.Only three of the peaks (EL-1, EL-2, and EL-3) were neededfor further sequencing analysis and they esolved the completesequence of the 8-kDa peptide (Fig. 4). It was possible todeduce the alignment of these peptides since EL-1 overlappedthe NHz-terminal sequence of the intact domain and EL-3did not contain a lysine indicating tha t it derived from theCOOH-terminal end of the 8-kDa domain. Amino acid anal-ysis showed that peak 1 in Fig. 2 corresponded to a peptidecontaining residues 1-26 and peak 2 residues 27-43 of thedomain sequence (Table I and Fig. 4).The reverse-phase chromatogram of peptides obtained from

8/8/2019 D9160CCAd01

4/5

Spectrin-Actin Binding Site of Erythrocyterotein 4.1 1336510 20Lys-Lys-Lys-Arg-Glu-Arg-Leu-AspGly-Glu-AIle-Tyr-Ile-Arg-His-Ser-~n-~t-

FIG. 4. The complete amino acidsequence and peptide overlaps of 67residues from the 8-kDa domain ofprotein 4.1. Peptidesbtained bycleavage with endoproteinase lysine-C(EL) and S. aureus protease V8 (SV8)are shown. Residues identified by auto-mated Edman degradation are indicatedas follows: solid l i n e , unambiguous iden-tification of phenylthiohydantoin deriv-ative; dashed line, tentative identificz-tion. Residues are numbered consecu-tively from the NH2-terminal end of the8-kDa peptide.

30 40~ 1 1 1 - A s p L e u - A s p L y s - S e r - G l ~ l ~ l ~ I l e - L y s - L y s - H i s - ~ s - ~ a - S e r - I l e - S e r - G l ~"_"50 60L e u - L y s - L y s - A s n - P h e - M e t - G l u - S e r - V a l - ~ ~ l ~ ~ ~ A r g - ~ ~ r ~ l ~ ~ ~ p L y s - A r

5 2 --ILeu-Ser-?hr-His-Ser-PFhe- 5 3 *

1 2 3 4 5 r;3 R A

3 ^

RA C A C A C A

Spectrin= - - -80565034 4.I28

-8

cT/4.1 o 1/400 Moo *A00 1/50FIG. 5. Characterization of antibodies raised against syn-thetic peptides of the 8-kDa domain. Protein 4.1 was digestedwith a-chymotrypsin at 0 "C for 30 min at enzyme to substrate atios( C T / 4 . I ) anging from 1:400 to 1:50. Proteins were run on a 10-15%SDS-PAGE and electrophoretically transferred onto nitrocellulosepaper as described by Towbin (21). A mixture of antibody A andantibody B was used to immunoblot the nitrocellulose. C, CoomassieBlue stained proteins. A , autoradiogram of immunoblots. Numbersdesignate the molecular weights of a-chymotryptic ragments nkiladaltons. Red blood cell ghostmembranes (lunes I ) were runwithout digestion.

the 8-kDa peptide following cleavage with S . aureus proteaseV8 is given in Fig. 3. This alternative cleavage method wasused to verify the alignment of the lysine-cleaved peptidesand confirm that these sequences were contiguous. Aminoacid compositions of the peaks labeled in Fig. 3 are summa-rized inTable 11. Two peptides (SV8-1 and SV8-2) weresequenced to establish the overlaps of the lysine-cleaved pep-tides. The results of these sequence determinations are sum-marized in Fig. 4. The amino acid analysis of theotherpeptides from V8 cleavage indicated that:peak 1 correspondedto residues 1-30; peak 2, residues 1-40; peak 3, residues 31-40; and peak 4, residues 41-67.

The sequence summarized in Fig. 4 represents the entire8-kDa complex-promoting domain of erythrocyte protein 4.1and includes several interesting features. The peptide has ahighly charged NH2-terminal region; the first 6 residues and8 of the first 10 are charged amino acids. Another cluster ofcharged amino acids involves residues 29-35 which includes

FIG.6. Inhibition of the ternary complex formation by an-tibodies A and B. Protein 4.1 (8 pg ) was incubated overnight at0 "C with monovalent Fab antibody (15 pg) or intact antibodies (24pg of each antibody, A and B ) . Spectrin (33 pg) and F-actin (30 pg)were added and incubated for an additional 45 min at 0 "C and themixture was sedimented as previously described (14). Equivalentsamples of supernatant (S ) and pellets (P) ere electrophoresed on a7-15% acrylamide gradient SDS gel. Activity was assessed based onthe distribution of spectrin between the supernatant and the ellet.Spectrin and actin alone (lanes I); protein 4.1, spectrin and actin(lanes2) ;monovalent Fab-A antibody, protein 4.1, spectrin and actin(lanes3) ;monovalent Fab-B antibody, protein 4.1, spectrin and actin(lanes 4) ; monovalent Fab-A and Fab-B antibodies, protein 4.1,spectrin and actin (lanes 5); ntact antibodies A and B, protein 4.1,spectrin and actin (lanes 6) .The band migrating above actin in lane6s is IgG-heavy chain.6 charges in 7 positions. Overall nearly half of the residues inthis domain are charged residues (3 1 of 67) suggesting thatthis entire segment may be exposed on the surface of theintact molecule. Such an exposed configuration wouldbeconsistent with the role of the 8-kDa peptide as a promoterof interaction between spectrinand F-actin. This domainmust bind t o at least one other polypeptide chain n hecomplex and could conceivably bind to asmany as three therpolypeptides simultaneously (both spectrin subunits and ac-tin).

Although we have no direct evidence that the -kDa peptideis phosphorylated, this domain has been located within the10-kDa region of protein 4.1 which has previously been shownto contain one or more sites for CAMP-dependent phospho-rylation (15, 16). Previous work by Krebs and his associates(for a eview, see Ref. 22) has shown tha t the atalytic subunit

8/8/2019 D9160CCAd01

5/5

13366 Spectrin-Actin Binding Site of Erythrocyte Protein 4.1of cyclic AMP-dependent protein kinase recognizes sites ofphosphorylation on substrates that contain at least one posi-tively charged residue followedby one or two interveningresidues between a serine or threonine, both of which can bephosphorylated. Based on this criterion, serines in positions17, 55, and 62 arepotentialsites for phosphorylation byCAMP-dependent protein kinase. Serine in position 27 mightalso be a candidate (23).

The 8-kDa peptide sequence has been compared to theNational Biomedical Research Foundation database of 3182protein sequences (May 1985 version) and no significantrelationships to other known proteins have been detected.Protein 4.1 may represent a new protein class or superfamilyand this report presents the first sequence information forprotein 4.1.

In order to further nvestigate the natureof the interactionof the 8-kDa peptide with spectrin and F-actin, antibodiesagainst two synthetic peptides whose sequences correspondedto residues 1-15 (antibody A) and residues 46-59 (antibodyB) of the 8-kDa peptide (Fig. 4) have been raised. The speci-ficity of these antibodies s demonstrated in Fig. 5. It has beenestablished that mild proteolysis of protein 4.1 by a-chymo-trypsin cleaves the molecule mainly in three central regionsgenerating fragments of 30-, 16-, l o - , and 22/24-kDa, respec-tively (15). The 8-kDa domain is located within the 10-kDaregion (14). Antibody A and antibody B (used as a mixture)recognized the intact protein 4.1 and those a-chymotrypticfragments of protein 4.1 which contained the 8-kDa peptide,as 56-kDa (30 + 16 + 10-kDa), 50-kDa (16 + 10 + 22/24-kDa), 32/34-kDa (10 + 22/24-kDa), 26-kDa (16 + 10-kDa),and the8-kDa peptide itself (Fig. 5).

Monovalent Fab antibodieswere produced and their abilityto inhibit the interaction of protein 4.1 with spectrin andactin investigated. In the experiment shown in Fig. 6,spectrinand F-actin were incubated alone (lanes 1 )or with protein 4.1(lanes 2) or with 4.1 in he presence of monovalent Fabantibodies (lunes 3-5) or with 4.1 in the presence of intactantibodies ( lunes 6) .Either Fab-A ( lanes 3)) or Fab-B (lanes4 ) fragments or both fragments added together (lanes 5) arecapable of inhibiting the association of protein 4.1 with spec-trin andF-actin. A mixture f intact antibody A and antibodyB also inhibits the formation of the ternary complex (lanes6) .The fact that either Fab-A or Fab-B fragments are ble toprevent the interaction between protein 4.1, spectrin, andactin may suggest that the entire 8-kDa peptide is requiredfor such interaction.Alternatively, one or both Fab fragmentsmight sterically block a binding site that is only part of the8-kDa piece.

Analogs of protein 4.1 have been identified in a variety ofnon-erythroid cells (12, 13),but hey have not yet beencharacterized functionally. Avian erythrocytes contain mul-

tiple isoforms of protein 4.1 not found in red cell precursorsof chick embryos at different stages of development (24). Thissuggests that different isoforms may play different functionsrelated to their role during cell development. It would be ofinterest t o determine which of these forms contain the 8-kDadomain and o what degree this peptide is conserved indifferent cells and tissues.Acknowledgments-We thank Dr. William C. Horne for his helpfuldiscussions, Drs. Barbara-Jean Bormann and TakashiTobe for theiradvice in the production of antibodies and Drs. Thomas L. Let0 andRichard A. Anderson for their comments. We gratefully acknowledgethe expert technical assistance of Gary Davis, Raymond DeAngelis,and Anthony Lanzetti. We thank Vicki Rosenweig for assistance intyping this manuscript.

REFERENCES1. Morrow, J. S., and Marchesi, V. T. (1981) J . Cell Bwl. 88, 63-2. Ungewickell, E., Bennett, P. M., Calvert, R., Ohanian, V., and3. Fowler, V. , and Taylor, D. L. (1980) J . Cell Biol. 85 , 361-3764. Cohen, C.M., and Korsgren, C . (1980) Biochern. Biophys. Res.5. Cohen, C. M., Tyler, J. M., and Branton,D. (1980) Cell 21,875-

468Gratzer, W. B. (1979) Nature 28 0, 811-814Cornrnun. 9 7 , 1429-14358R36. Ty-l&, J. M. , Reinhardt, B. N., and Branton, D. (1980) J. Biol.Chem. 255,7034-70397. Anderson, R. A., and Lovrien, R. E. (1984) Nature 307,655-6588. Pasternack, G. R., Anderson, R. A., Leto, T. L., and Marchesi,V. T. 1985) J. Bwl . Chem. 260,3676-36839. Bennett, V., and Stenbuck, P.J. (1979) J . Biol. Chem. 254,2533-254110. Morrow, J. S., Speicher, D. W., Knowles, W. J., Hsu, C. J., andMarchesi, V. T. (1980) Proc. Natl. Acad. Sci . U. S. A. 77,6592-659611. Bennett, V. , and Stenbuck, P. J. (1980)J. iol. Chern. 255,6424-643212. Bennett, V. (1985) Annu. Reu. Biochern. 54,273-30413. Marchesi, V. T. (1985) Annu. Rev. Cell Biol. 1, 531-56114. Correas, I. , Leto, T. L., Speicher, D. W., and Marchesi, V. T.

15. Leto, T. L., and Marchesi, V. T. (1984)J . Biol. Chern. 259,4603-16. Horne, W. C., Leto, T. L., and Marchesi, V. T. (1985) J . Biol.17. Speicher, D. W., Davis, G., Yurchenko, P. D., and Marchesi, V.18. March, S. C., Parikh, I., and Cuatrecasas, P. (1974) Anal.19. Porter, R. R. (1959) Biochern. J. 7 3 , 119-12620. Laemmli, U. K. (1970) Nature 227, 680-68521. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad.22. Krebs, E. G., and Beavo, J. A. (1979) Annu. Reu. Biochern. 48,23. Kishimoto, A., Nishiyama K., Nakanishi, H., Uratsuji, Y., No-mura, H., Takeyama, Y. , and Nishizuka, Y. (1985) J . Biol.Chern. 260,12492-12499

(1986)J. B i d . Chern. 261, 3310-33154608Chem. 260,9073-9076T. 1983) J . B i d . Chem. 258,14931-14937Biochem. 60 , 149-152

Sci. U. S. A . 76,4350-4354923-959

24. Granger, B. L., and Lazarides, E. (1985) Nature 313,238-241