Identification of ubiquitinated repeats in human erythroid α-spectrin

Identification of ubiquitinated repeats in human erythroid a-spectrin

Luca Galluzzi1, GaeÈ l Nicolas2, Mirko Paiardini1, Mauro Magnani1 and Marie-Christine Lecomte2

1Institute of Biological Chemistry `G. Fornaini', University of Urbino, Italy; 2INSERM U409, Faculte de MeÂdecine `X. Bichat', Paris, France

The spectrin role(s) is (are) very important for the shape and the physical properties of red cells, such as

deformability and resistance to mechanical stresses. Moreover a variety of spectrin diseases are known. We have

previously demonstrated [Corsi, D., Galluzzi, L., Crinelli, R. & Magnani, M. (1995) J. Biol. Chem. 270,

8928±8935] that human erythroid a-spectrin is ubiquitinated in vitro and in vivo. In order to define the

ubiquitinated repeats of this long protein and find out a possible function, we have produced recombinant

peptides encompassing the aIII-, aIV-, aV- and EF hand domains of a-spectrin chain. These peptides were tested

in in vitro ubiquitin conjugation assays and two regions susceptibles to ubiquitination were found. The first one,

in the aIV-domain, includes the repeat 17 and the second one, in the aV-domain, includes the repeat 20 and a part

of repeat 21. We also demonstrated that the susceptibility to ubiquitination of the aV-domain is reduced by

interaction with the corresponding portion of b-spectrin chain (bIV-domain). Thus, at least ubiquitination of

aV-domain is susceptible to cytoskeleton assembly and spectrin dimerization.

Keywords: a-spectrin; ubiquitin; recombinant peptides.

Erythrocyte spectrin is the main component of the red cellmembrane skeleton, a two-dimensional multiprotein networkunderlying the lipid bilayer and responsible for the shape andthe physical properties of red cells, such as deformability andresistance to mechanical stresses [1]. Spectrin is composed oftwo elongated subunits, a and b chains (280 and 246 kDa,respectively). The erythroid a-spectrin chain consists of 22repeating segments of about 106 amino acids in length whereasthe b-spectrin chain consists of only 17 segments. Therepeating segments show a low sequence identity (about20%) with some highly conserved residues (notably leucineand tryptophan at positions 26 and 45 of the repeats,respectively) [2,3]. These segments are composed of threehelices (A, B and C) which are organized in a stable triplehelical coiled-coiled structure [4,5]. The tryptic digestion ofspectrin revealed the presence of several trypsin-resistantdomains separated by zones much more sensible to digestion.Five domains were defined on the a chain (aI to aV fromthe N-terminus) and four on the b chain (bI to bIV fromthe C-terminus) [6]. The a and b chains are noncovalentlyassociated side-to-side in an antiparallel orientation to formheterodimers. Spectrin heterodimer assembly is initiated bythe association of a specific region near the C-terminal ofa-spectrin, with a complementary region near the N-terminalof b-spectrin [7], involving the 18±21 repeats of the a-chainand the first four repeats of the b-chain. These regions arecalled nucleation site. Initial association of these dimerizationsites is rapidly followed by lateral pairing of additional a and bmotifs along the length of the molecule. Spectrin heterodimersassociate head-to-head to form tetramers, which constitute thelong flexible filaments of the network. Spectrin tetramers

interact with other proteins of the erythrocyte skeleton, such asankyrin, band 4.1 [2], actin, adducin and tropomyosin [3,8].Mutations affecting these interactions result in increasedmembrane fragility and are associated with hereditary typesof anemia such as elliptocytosis and spherocytosis [9]. Spectrin,like many of these proteins which were first identified in redcells, has isoforms expressed in nonerythroid cells. Inmammals, two genes coding for an a-chain (aI, the erythroidisoform and aII, the nonerythroid isoform) and three b geneshave been identified. However, the composition, organizationand regulatory processes of the membrane skeleton in none-rythroid cells are less well known. This nonerythroid spectrin-based skeleton might have different functions, as suggested bythe pattern of the expression of spectrin isoforms. Theseisoforms display a nonhomogeneous spatial and temporaldistribution and may not be associated with the peripheralmembrane. For instance, in neurons and skeletal muscle,different isoforms of spectrin are confined to distinct regionsand may undergo to rearrangements in response to variousstimuli. The compartmentalization of the erythroid isoform intocell bodies and dendrites and the location of the nonerythroidisoform in axon presynaptic terminals [10,11] suggest that brainspectrin isoforms may perform related but distinct functions inneuronal cells.

Previously, we demonstrated that red blood cell a-spectrin isa specific substrate for ubiquitination [12] and that ubiquiti-nation occurs within the aIII, aIV and aV trypsin-resistantdomains, corresponding roughly to repeats a11±14, a15±18and a19±21, respectively [13].

Ubiquitin is a highly conserved 76-amino acid polypeptide,found both free and covalently bound to target proteins.Ubiquitination of a protein requires a multienzymatic systemconsisting of three classes of enzymes: a ubiquitin-activatingenzyme (E1), a ubiquitin-conjugating enzyme (E2) and aubiquitin protein ligase (E3) that catalyze the bond between thecarboxyl terminus of ubiquitin and the lysine residue on thetarget protein. Multiubiquitin chains can be formed throughthe attachment of additional ubiquitin molecules to lysineresidues 48 or 63 of ubiquitin itself [14,15]. Multiubiquitination

Eur. J. Biochem. 267, 2812±2819 (2000) q FEBS 2000

Correspondence to M. Magnani, Institute of Biological Chemistry `G.

Fornaini', University of Urbino, via Saffi 2, 61029 Urbino (PU), Italy.

Fax: 1 39 07223 20188, Tel.: 1 39 07223 05211,

E-mail: [email protected]

Abbreviations: AEBSF, 4-(2-aminoethyl)benzenesulfonyl fluoride;

GST, glutathione-S-transferase; TB, Terrific broth.

(Received 3 March 2000, accepted 16 March 2000)

of some proteins determines their degradation by the 26S-protease complex [16]. In other cases ubiquitination serves as amodification resulting in functional regulation [17]. In general,ubiquitination has important functions in cellular regulatorymechanisms such as the cell cycling and tumor growth, antigenprocessing, signal transduction pathways, receptor-mediatedendocytosis, etc. [18].

In this study we searched for the ubiquitination sites withinthe a-spectrin chain in an attempt to gain insight into thephysiological significance of this process. Using recombinantpeptides we identified two sites of ubiquitination: the firstin a zone including the repeating segment 17; the secondincluding the repeating segment 20 and part of 21. Theseregions appear to contain the amino acid involved inrecognition of the substrate and lysines potentially involvedin ubiquitin-conjugation.

E X P E R I M E N T A L P R O C E D U R E S

Construction of the spectrin peptide expression plasmids

Specific regions of the human a-spectrin molecule wereobtained by PCR amplification of reticulocyte cDNA usingPfu DNA Polymerase (Stratagene) and subcloned into thepGEX-KG plasmid [19]. Oligonucleotide primers weredesigned as shown in Table 1. For constructs a11±13,a13±14, a11±14, a19±20, a20±21, a19±20±21 and a22,the primers contained restriction enzyme sites for XbaI andXhoI, at the 5 0 and 3 0 ends, respectively. For constructsa15±16, a16±17, a17 and a15±16±17, the primerscontained restriction enzyme sites for EcoRI and NcoI, at the5 0 and 3 0 ends, respectively. For construct a14±15, the primerscontained restriction enzyme sites for BamHI and NcoI, at the5 0 and 3 0 ends, respectively. All the restriction sites at 3 0 endswere preceded by a stop codon. The purified PCR-amplifiedDNAs were cut with the appropriate enzymes (Gibco-BRL) anddirectionally cloned into the restricted vector using standardtechniques. Transformation was performed in E. coli strainJM 109. The clones expressing the human cDNA corres-ponding to the a18±21 [20], a14±17 and b1±41 [20]spectrin repeats, kindly provided by D.W. Speicher (WistarInstitute, PA, USA), were subcloned into the pGEX-2Tplasmid (Pharmacia Biotech). The peptides were expressed asglutathione S-transferase fusion proteins.

Expression and purification of recombinant peptides

Overnight cultures were diluted 1 : 15 in Terrific broth (TB)medium containing 100 mg´mL21 ampicillin and were grown at37 8C or 30 8C to an optical density of 0.5±0.7 at 600 nm.Cultures were diluted again 4 : 5 and incubated for a further15 min before induction with 0.5 mm IPTG. Cells were inducedfor 2 h. After centrifugation, the bacterial pellets were frozenand stored at 220 8C. Glutathione-S-transferase (GST) fusionproteins were purified as described previously [21]. Briefly, cellpellets from 1 L of culture were resuspended in 30 mL of20 mm phosphate buffer pH 7.3, 150 mm NaCl, 1% TritonX-100, containing 1 mm EDTA, 0.5 mm 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF) or phenylmethanesulfonylfluoride, 2 mg´mL21 leupeptin, 2 mg´mL21 pepstatin,100 U mL21 aprotinin, 1 mm 2-mercaptoethanol and lysed bysonication. After centrifugation, the cleared lysates were loadedonto a reduced glutathione±Sepharose 4B column (PharmaciaBiotech). GST fusion proteins were eluted with 10 mmglutathione in 50 mm Tris, pH 8.0.

Ubiquitin conjugation assays

Ubiquitin conjugation assays were performed using reticulocyteor erythrocyte lysate (fraction II) essentially as described in[12] with some modifications. In a final volume of 30 mL, thereaction mixture contained 80 mm Tris, pH 7.6, 4 mm MgCl2,1 mm dithiothreitol, 3.5 mm ATP, 14 mm creatine phos-phate, 2.2 mg of creatine phosphokinase, 13 mg of reticu-locyte or erythrocyte fraction II [12,14], 5 mm 125I-labelledubiquitin, 1 mm phenylmethanesulfonyl fluoride, 1 mmAEBSF, 1 mm leupeptin, 1 mm pepstatin and 50 pmol ofpurified GST fusion protein. In some experiments 15 mg ofhuman red blood cell membranes (containing about 6.8 pmolof a-spectrin) were used as substrate and different amounts ofGSTa16±17 peptide were added as competitor (6.8, 20.4, 34and 47.6 pmol). After 1 h of incubation at 37 8C the mixtureswere boiled for 3 min in Laemmli sample buffer containing 2%2-mercaptoethanol and electrophoresed in SDS/PAGE. Gelswere stained, dried and autoradiographed.

Binding assay using GSTa18±21 and GSTb1±41

In some experiments, 80 pmol of GSTa18±21 and 80 pmol ofGSTb1±41 were mixed in a final volume of 70 mL in 50 mm

Table 1. Design of recombinant a-spectrin peptides.

Peptide Amino acids Codons 5 0 Primer 3 0 Primer

a11±13 R1077-Q1393 3415±4365 GGGGGTCTAGATCGTCTATTGCAACGTTATAATGAA GGGGCTCGAGTTACTGGTCTAGGATCTTCTTGCG

a13±14 T1278-D1498 4018±4680 GGGGGTCTAGATACAAAGGATCGTAAGGAGAGCC GGGGCTCGAGTTAGTCTCCAAGCTTTGTCCGC

a11±14 R1077-D1498 3415±4680 GGGGGTCTAGATCGTCTATTGCAACGTTATAATGAA GGGGCTCGAGTTAGTCTCCAAGCTTTGTCCGC

a14±15 L1395-N1603 4368±4995 GCGTGGATCCCTGGAGTTG GACGCCATGGTTAATTGAGCTTCTTCCCTTTGT

a15±16 R1493-A1711 4663±5319 GGGGGAATTCGGACAAAGCTTGGAGACTATG GGTGCCATGGTTAGGCCTCTTTCAATTTTTCGTG

a16±17 K1601-L1818 4987±5640 GGGGGAATTCTAAAGCTCAATGAGGCCAGTCG GGGGCCATGGTTATAGGGATTCTTCCAACTTAAG

a17 K1707-L1818 5305±5640 GGGGGAATTCTAAAATTGAAAGAGGCCTATGCC GGGGCCATGGTTATAGGGATTCTTCCAACTTAA

a15±16±17 R1493-L1818 4663±5640 GGGGGAATTCGGACAAAGCTTGGAGACTATG GGGGCCATGGTTATAGGGATTCTTCCAACTTAAG

a19±20 L1921-R2150 5947±6636 GGGGTCTAGACTTGCAATTGGAAGACGATTATG GGGGCTCGAGTTATCTTGCCTCTTCCTTTTG

a20±21 R2027-Q2257 6265±6957 GGGGGTCTAGACAGACAGAAATTGCTGGAG GGGGCTCGAGTTATTGCTCCAGGTTGTGTTG

a19±20±21 L1921-Q2257 5947±6957 GGGGTCTAGACTTGCAATTGGAAGACGATTATG GGGGCTCGAGTTATTGCTCCAGGTTGTGTTG

a22 M2251-N2418 6937±7439 GGGGGTCTAGAGATGCAACACAACCTGGAGC GGAGCTCGAGTTCCCACGACACTAAGATTTT

q FEBS 2000 The human a-spectrin ubiquitinated repeats (Eur. J. Biochem. 267) 2813

Tris, pH 8.0, 0.1% BSA. Dimer complexes were allowed toreach equilibrium by incubation on ice for 45 min and thenwere utilized in ubiquitination assays.

Other procedures

SDS/PAGE was performed as described by Laemmli [22]. Gelswere stained with Coomassie Brilliant Blue R-250. Proteinconcentration was determined by absorbency at 280 nmusing extinction coefficients calculated from sequences.Alternatively, protein concentration was determined by themethod of Bradford [23]. Quantitative determinations ofprotein bands in Coomassie Blue-stained gels were performedusing an LKB Ultroscan XL laser densitometer. After stainingand drying, the gels were also analyzed with a GS-250Molecular Imager (Bio-Rad) to evaluate the radioactivityincorporated into the a-spectrin peptides.

R E S U L T S

Expression and purification of recombinant peptides

To identify the repeats of a-spectrin as targets ofubiquitination, several recombinant peptides encompassingthe aIII-, aIV-, aV- and EF hands domains were expressedas glutathione-S-transferase fusion protein in E. coli. To obtainstable recombinant peptides that serve as a reliable model of theentire protein, two factors were considered: the phase and thelength of the fragments. The appropriate phase is veryimportant for the maintenance of structure and stability.Winograd et al. [24] provided experimental evidence fromDrosophila a-spectrin fragments suggesting a downstreamphase-shift of 26 amino acids for the structural domain suchthat the first amino acid in a folded unit is residue 26 inthe 106-residue sequence motif. This conformationally definedmotif phasing was confirmed by determination of the crystalstructure of the 14th segment of Drosophila a-spectrin [5].Assuming that human a-spectrin has the same phase-shift asDrosophila a-spectrin, the first amino acid of everystructural domain would be the highly conserved leucine26. Subsequently, Kotula et al. [25], through proteaseprotection analysis, suggest that structural domains of humana-spectrin begin and end around residue 30 of the sequencemotif. Finally, Lusitani et al. [26] identified the first amino acidof the first domain in human erythrocyte a-spectrin as serine 52(residue 29 in the 106-residue sequence motif ). All the peptidesproduced for this study began and stopped around theconserved leucine 26 of the repeating segments. This residuewas always included in the N- or C-terminus of recombinantpeptides. Menhart et al. [27] showed that a single 106-aminoacid unit of a-spectrin folds into helical bundles as a stablestructural unit but that peptides composed of more than onehomologous repeat are substantially more stable and exhibit abetter secondary structure than peptides composed of a singlerepeat. All of the peptides used for this work were at least twoconformational units long, with the exception of peptide a17(which contained a single 106-amino acid unit) and peptide a22(which included the nonhomologous EF hand domain). Allclones were sequenced to verify the fidelity of the recombinantproteins. Sequencing of clone a22 gave different results incomparison with the sequence of human erythroid a-spectrinpublished by K. E. Sahr et al. [28]. We noted that the stretch ofG residues at position 7379 of the published sequence consistsof four G residues instead of three G residues. The presence ofan additional G causes a change in reading frame, which

changes the last 31 residues and the C-terminus of this proteinappears 11 amino acids shorter than the original sequence [29].Therefore, we can affirm that a-spectrin is 2418 amino acidsinstead of 2429, and its calculated molecular mass is about279 900 Da.

All the peptides were highly expressed in bacteria, butthey differed in solubility and stability. The fusion proteinsGSTa18±21, GSTa19±20±21, GSTa20±21 and GSTa22were less soluble than the other peptides and were primarily ininclusion bodies. This means that the C-terminus of a-spectrin,consistent in repeats 21 and 22, gives hydrophobic features tothis molecule. In particular GSTa22, containing EF hands, wasvery insoluble and susceptible to degradation. The amount ofthese peptides in the soluble fraction and the yield of purifiedfusion proteins were improved by growing the bacterial culturesat 30 8C instead of 37 8C. In this way we avoided denaturationand possible incorrect refolding of the recombinant peptides.Yields of the purified fusion proteins ranged from 2 to20 mg´L21 of bacterial culture. After elution by glutathione±Sepharose 4B affinity column, the peptides had an apparentmolecular mass in SDS/PAGE in agreement with the calculatedvalue. Gel scan analysis of band intensities indicated that thepurity of the samples was between 50% and 85%. The peptideswere utilized in ubiquitin conjugation assays without otherpurification step.

Identification of ubiquitinated a-spectrin repeats

Ubiquitin conjugation experiments were performed in vitro asdescribed in Experimental procedures utilizing 50 pmol of eachrecombinant peptide. In a first test, ubiquitination assays wereperformed on GST alone, on the entire fusion proteins and onthe purified peptides, free of GST. The fusion proteins exhibitthe same ubiquitination pathways as purified peptides and GSTis not a substrate for ubiquitination (data not shown). On thebasis of these data, the entire fusion proteins representedschematically in Fig. 1 were used instead of free peptides inubiquitination assays. The conjugation mixtures were electro-phoresed in SDS/PAGE and the gels were stained, dried andautoradiographed. Initially we tested peptides GSTa11±13 andGSTa13±14 (encompassing aIII-domain), GSTa14±17(encompassing aIV-domain) and GSTa18±21 (encompassingaV-domain). In this experiment only peptides encompassingaIV- and aV-domains were ubiquitinated. We subsequentlyextended our study utilizing a peptide encompassing the entireaIII-domain (GSTa11±14 peptide) and shorter and overlappingpeptides in aIV- and aV-domains (Fig. 2A).

Peptides GSTa14±17, GSTa15±17, GSTa16±17, but notGSTa15±16 and GSTa17, were highly ubiquitinated,indicating that the initial portion of repeat 17 may contain thesequence involved in the ubiquitination process. In fact itappears that the a helix structure CA of repeat 17 must be intactto permit ubiquitin conjugation. Moreover peptides GSTa18±21, GSTa19±21, GSTa19±20 and GSTa20±21 wereubiquitinated, indicating that the common portion from arginine23 of repeat 20 to arginine 32 of repeat 21 may also beubiquitinated. Ubiquitin conjugation causes a shift in themolecular mass of the protein target of about 9 kDa consistentwith the attachment of one ubiquitin molecule. Monoubiquiti-nated adducts of a-spectrin peptides were detectable inSDS/PAGE as bands with a slowen electrophoretic mobility.Those bands were not present in fraction II (Fig. 2A, lane 13)and corresponded to the autoradiographyc signal of the gel(Fig. 2A, autoradiography). Laser scan densitometry of thecoomassie stained bands corresponding to unconjugated and

2814 L. Galluzzi et al. (Eur. J. Biochem. 267) q FEBS 2000

ubiquitin-conjugated peptides showed that 15% of GSTa16±17(7.5 pmol; 0.4 mg) was conjugated to ubiquitin under theexperimental conditions used, whereas only 8% of GSTa14±17was ubiquitinated. Analysis of the GSTa15±17 ubiquitinatedband was not possible because of comigration with a fraction IIminor band. Gel scan analysis of ubiquitinated peptidesencompassing the aV domain was not possible becausethey were not detectable with Coomassie staining. Quantitativeanalysis performed by molecular imager of 125I-labelledubiquitin a-spectrin peptides showed a great differencebetween peptides encompassing aIV-domain and peptidesencompassing aV-domain (Fig. 2B). In fact, the ubiquitinationof GSTa19±20 was about 20 times less efficient than that ofGSTa16±17. Moreover, ubiquitination appears to increasewhen the number of repetitive segments is lower, either for aIVor aV peptides. This fact may be explained considering thatpeptides with more repetitive units may be better folded [27],thus making the lysines susceptible to ubiquitination lessaccessible. Peptide GSTa20±21 appears to be only slightlyubiquitinated even after normalization of the gel band withGSTa19±20 through a densitometer. This means that theentire structure of repeat 20 may be necessary for a higherubiquitination affinity. In the conjugation assays we utilizedalternatively wild-type 125I-labelled ubiquitin or K48R

mutant 125I-labelled ubiquitin which was unable to producepolyubiquitin chains.

The patterns of peptide ubiquitination were the same,indicating that all of the peptides were monoubiquitinated.

Competition assay

In order to evaluate the specificity of ubiquitination on a-spectrinrecombinant peptides, ubiquitin-conjugation experiments onwhole a-spectrin were performed in the presence of increasingamounts of the GSTa16±17 peptide as competitor. This peptidewas chosen because, among all of the peptides investigated, itwas the best substrate in the ubiquitin conjugation assay underour standard conditions. Ubiquitin conjugation assays wereperformed on human red blood cell membranes (15 mg ofmembrane corresponding to 1.9 mg or 6.8 pmol of a-spectrinassembled in the cytoskeleton) in the presence of differentamounts of GSTa16±17 peptide (from 6.8 mp to 47.6 pmol).The analysis on a molecular imager of 125I-labelled ubiquitinincorporation revealed a decreased signal in a-spectrinlabelling in the presence of GSTa16±17 as competitor. Thisdecrease reached approximately 49%, compared to a-spectrinalone, when the molar ratio of a-spectrin/GSTa16±17 was 1/7(Fig. 3). The same experiments were performed in the presence

Fig. 1. Schematic representation of a-spectrin

recombinant peptides tested in ubiquitin-

conjugation assays. (A) Correlation of

recombinant peptides with the repeating segment

of a-spectrin. The alignment of the spectrin

amino acid sequence (numbers at top of the

figure) is correlated with the three helices A, B

and C (helix C in following repeating segment)

forming the triple helical conformational unit.

Spaces between helices represent turn regions.

(B) The a-spectrin recombinant peptides

encompassing domains aIII, aIV and aV tested

in ubiquitin conjugation assay were GST fusion

proteins. Dotted lines represent ubiquitinated

peptides. The purified GSTSH3 peptide (kindly

provided by C. Fournier) was from leucine 26

of repeat 9 to arginine 29 of repeat 11. EF is the

EF hand domain.

q FEBS 2000 The human a-spectrin ubiquitinated repeats (Eur. J. Biochem. 267) 2815

of the same concentrations of lactalbumin instead of theGSTa16±17 peptide. Lactalbumin is a protein commonlyused as a model substrate for in vitro ubiquitination anddegradation assays [30]. The results did not show a decreasein incorporation of 125I-labelled ubiquitin in a-spectrin in thepresence of lactalbumin as competitor. A significant reductionin a-spectrin ubiquitination (35%) was observed only in thepresence of 140 pmol of lactalbumin per pmol of a-spectrin.Taken together, these data demonstrate that a-spectrin and theGSTa16±17 peptide compete for the same enzymatic pathwayand that the recombinant fusion protein is a reliable copy of theentire protein.

Ubiquitin conjugation on the aV domain: inhibition bydimerization with the bIV domain

Spectrin dimerization requires a specific region at the tail endof the a and b subunits: the aV and bIV trypsin-resistantdomains (nucleation site), which include the repetitivesegments a19±21 and b1±4, respectively [7]. It has beendemonstrated that the recombinant peptides corresponding tothe a18±21 and b1±41 repetitive segments exhibit the highestaffinity to reconstitute the nucleation site [20]. The recombinantfusion proteins GSTa18±21 or GSTb1±41 can also be utilizedin dimerization affinity assays. As the peptide GSTa18±21 wasubiquitinated in vitro, we investigated the ubiquitination of thispeptide when coupled with the GSTb1±41 peptide. In fact, thisis the case in the spectrin dimer, which contributes to form the

Fig. 2. a-spectrin peptides ubiquitination.

(A) 125I-Labelled ubiquitin was conjugated to

recombinant peptides as described in

Experimental procedures. Samples were then

electrophoresed in an 8% SDS/PAGE. The gel

was stained, dried and autoradiographed. Lane 1

contained GSTa11±14 (aIII domain); lanes 2

through 7 contained GSTa14±17, GSTa14±15,

GSTa15±16±17, GSTa15±16, GSTa16±17,

GSTa17, respectively (aIV-domain); lanes 8

through 12 contained GSTa18±21,

GSTa19±20±21, GSTa19±20, GSTa20±21

(aV-domain) and GSTa22 (EF hand domain),

respectively; lane 13 contained only fraction II as

a control. As GSTa22 peptide comigrates with

creatine phosphokinase, it is not clearly visible in

the gel. Molecular mass standards (St) are

indicated in kDa. The minor radiolabelled bands

at lower molecular masses with respect to

GSTa14±17, GSTa15±16±17 and GSTa16±17

ubiquitinated peptides are probably due to

degradation products. (B) quantitative analysis on

a molecular imager. The amount of ubiquitin

present in GSTa16±17 was taken as 100% and

the relative amounts for the other a-spectrin

peptides calculated.

Fig. 3. Competition assay between a-spectrin and GSTa16±17 peptide.

Ubiquitin conjugation assays were performed as described in Experimental

procedures utilizing 15 mg of erythroid membranes and an increasing

amount of GSTa16±17 peptide, as indicated in the graph. After 1 h at

37 8C the conjugation mixtures were boiled and analyzed by SDS/PAGE.

Molecular imager analysis of the ubiquitinated a-spectrin band shows a

marked decrease in ubiquitination of this molecule in the presence of

GSTa16±17 as competitor. Considering that ubiquitin is not a limiting

factor because it is present at concentrations as high as 5 mm, we can affirm

that the entire a-spectrin and GSTa16±17 peptide compete for the same

enzymatic pathway.

2816 L. Galluzzi et al. (Eur. J. Biochem. 267) q FEBS 2000

network of the erythrocyte skeleton under physiologicalconditions.

The peptides were first allowed to form heterodimercomplexes as described under `experimental procedures'.Then, 40 pmol were migrated in a nondenaturing gel toevaluate dimer assembly and 40 pmol were submitted to aubiquitin conjugation assay. To maintain the same BSAconcentrations, all conjugation mixtures contained 0.08%(w/v) BSA (final concentration). It is worth noting that, underthese experimental conditions, dimer formation accounted forabout 90% of the peptides used. In vitro ubiquitination ofGSTa18±21 results in single autoradiographyc signal corre-sponding to 125I-labelled ubiquitin GSTa18±21 adducts. Thisband is specific because not present in incubation containingonly fraction II (Fig. 4, lane 3) and show a shift in the

molecular mass of about 9 kDa with respect to the Coomassie-stained band, in agreement with the conjugation of one125I-labelled ubiquitin molecule. However when GSTa18±21 isassociated to the GSTb1±41 the same radioactive band is lesspronounced (Fig. 4, lane 2) showing that when GSTa18±21associate in dimeric form with GSTb1±41 it became lesssusceptible to ubiquitination (roughly 45%).

D I S C U S S I O N

In a previous study [13] utilizing limited proteolysis oferythroid a-spectrin and western blotting techniques, weidentified the ubiquitination regions of assembled humanerythroid a-spectrin on domains aIII, aIV and aV of themolecule. In an attempt to localize the ubiquitination sites on a-spectrin more precisely, we have now generated and utilizedseveral recombinant peptides encompassing, entirely or in part,the aIII, aIV and aV domains of a-spectrin. Utilizing thisapproach we have essentially confirmed the existence of aubiquitination site on the aV domain (including repeats 19±21)and the aIV domain (including repeats 15±18). We found that agroup of ubiquitinated peptides contained a common regionlocated between the first residue (isoleucine) of repeat 17 andleucine 30 of repeat 18, and another group contained a commonregion located between arginine 23 of repeat 20 andarginine 32 of repeat 21. Thus, at least two ubiquitinationsites on a-spectrin are located in these two parts of themolecule. However, the peptides encompassing the aIII domain(including repeats 11±14) were not found to be ubiquitinatedunder the experimental conditions used. One possible explana-tion for this discrepancy may be found in the status of the a-spectrin used in the previous study: a-spectrin was alwaysassembled in the membrane or at least in dimer form withb-spectrin. This fact, as documented for the nucleation site(see above), may influence aIII ubiquitination. In other words,the susceptibility of a protein to ubiquitination requires at leasttwo conditions: an available lysine residue within a consensussequence and a tertiary structure recognized by the ubiquitin-conjugating enzymes (E2s/E3s). Although the primary andsecondary structures of aIII are respected in the recombinantpeptides and in the assembled cytoskeleton, the E2s/E3srecognition site may not be maintained.

The ubiquitination regions found in a-spectrin presentdifferent susceptibility to ubiquitin-conjugation. In fact, theaIV region is ubiquitinated about 20 times more efficiently thanthe aV region. Moreover, the peptides encompassing eachdomain have different susceptibility to ubiquitin conjugationbetween them. In particular, the GSTa20±21 peptide was foundto be approximately three times less ubiquitinated thanGSTa19±20 (Fig. 2) even though both peptides are made upof two repetitive units. This may mean that the beginning ofrepeat 20 and/or part of repeat 19 must be present to have atertiary structure efficiently recognized by the ubiquitinatingenzymes. This is also evident in the case of the peptidesencompassing repeat 17. In fact, only the peptides that containthe entire repeat 17 are ubiquitinated, indicating that thesequence involved in recognition of the substrate is probablylocated in the a-helix structure CA of this repeat. The structuraland folding characteristics of the entire a-spectrin andrecombinant a-spectrin fragments were amply investigated,demonstrating that the recombinant spectrin fragments canserve as a model for native spectrin [20,31±33].

Interestingly, we noted a marked decrease in ubiquitinationaffinity for the peptide GSTa18±21 when coupled withGSTb1±41. Together these peptides constitute the nucleation

Fig. 4. Interaction between GSTa18±21 and GSTb1±41: involvement

in ubiquitination affinity. (A) 125I-labelled ubiquitin was conjugated to the

GSTa18±21 peptide and to the dimer formed by GSTa18±21 and

GSTb1±41 as described in Experimental procedures. The samples were

electrophoresed in a 6.5% SDS/PAGE and the gel was stained, dried, and

autoradiographed. Lane 1 contains GSTa18±21 (a), lane 2 contains

GSTa18±21 (a) and GSTb1±41 (b), lane 3 is only fraction II. The large

band in the Coomassie-stained gel is BSA. The autoradiography shows a

ubiquitinated band, corresponding to 125I-labelled ubiquitin GSTa18±21

(c) that is much less marked in lane 2. The radioactive band under that

corresponding to 125I-labelled ubiquitin GSTa18±21 is also present in lanes

1 and 3 and likely corresponds to a nonspecific ubiquitinated protein

present in the incubation mixture. (B) evaluation on molecular imager of125I-labelled ubiquitin GSTa18±21 in the presence of GSTb1±41

compared to 125I-labelled ubiquitin GSTa18±21 alone. The value considers

a dimer formation of about 90% between the peptides.

q FEBS 2000 The human a-spectrin ubiquitinated repeats (Eur. J. Biochem. 267) 2817

site essential for dimerization of the a- and b-spectrinmonomers. The lysine(s) or other residues involved inubiquitin conjugation or in E2s/E3s recognition in thisregion of a-spectrin may be partially masked by interactionswith the b monomer. This fact appears to be relevant in vivoprobably influencing the stability of the entire cytoskeleton byaffecting nucleation site ubiquitination. The good folding ofthe entire a-spectrin molecule and its association with theb-spectrin monomer in vivo, as well as interactions withother cytoskeletal proteins, could explain the low amount(about 3%) of ubiquitinated a-spectrin in erythroid membranes.

In mammalian red cell precursors, the newly synthesizeda-spectrin that is present in the cytosol is in much greateramounts than b-spectrin. In contrast, the membrane skeletoncontains equimolar amounts of a- and b-spectrin, suggestingthat the association of a-spectrin with the membrane skeletonmay be limited by the amount of b-spectrin synthesized [34].The limited ubiquitination of a-spectrin observed [13] does notappear to be a degradation signal for the proteasome [35,36]. Inthis respect a-spectrin is more similar to several cell surfaceproteins that are ubiquitinated on their cytoplasmic tails but areprobably not degraded by the proteasome as a consequence. Forexample, the yeast receptors Ste6p and Ste2p are ubiquitinated,internalized by endocytosis and degraded in the lysosome-likevacuole [37]. Although unassembled a-spectrin is degradedmore rapidly than b-spectrin during erythroid maturation, themechanism(s) is likely different than the one investigated in thispaper. In fact ubiquitination of a protein substrate forsubsequent degradation usually occurs as polyubiquitin chainformation and on well-defined signal sequences (PESTsequences, the cyclin `destruction box', N-terminal residues,etc.). Thus, a-spectrin ubiquitination may play a different rolewhen occurring in the erythroid precursor (degradation) ormature erythrocyte. Furthermore, in nonerythroid cells, theerythroid isoform a-spectrin may not be organized as it is in themembrane skeleton and its ubiquitination may be involvedin some rearrangements in response to various stimuli.Alternatively, because the erythroid a-spectrin does not containbinding sites for calpains, enzymes involved in the degradationof spectrin isoform aII (fodrin), ubiquitination of thisisoform in nonerythroid cells may be involved in dynamicturnover processes. The identification of lysines susceptible toubiquitination in the regions of a-spectrin here localized willgive more precise information on the sequence involved inubiquitin conjugation and will produce further information toexplain the physiological significance of this post-translationalmodification.

A C K N O W L E D G E M E N T S

We thank Dr Dario Corsi and Dr Catherine Fournier for their technical help.

This work was partially supported by PRIM founds from the Ministry for

University and Research, Italy.

R E F E R E N C E S

1. Bennett, V. & Lambert, S. (1991) The spectrin skeleton: from red cells

to brain. J. Clin. Invest. 87, 1483±1489.

2. Gallagher, P.G. & Forget, B.G. (1993) Spectrin genes in health and

disease. Semin. Hematol. 30, 4±20.

3. Bennett, V. & Gilligan, D.M. (1993) The spectrin-based membrane

skeleton and micron-scale organization of the plasma membrane.

Annu. Rev. Cell Biol. 9, 27±66.

4. Speicher, D.W., DeSilva, T.M., Speicher, K.D., Ursitti, J.A., Hembach,

P. & Werglarz, L. (1993) Location of the human red cell spectrin

tetramer binding site and detection of a related `closed' hairpin loop

dimer using proteolytic footprinting. J. Biol. Chem. 268, 4227±4235.

5. Yan, Y., Winograd, W., Viel, A., Cronin, T., Harrison, S.C. & Branton,

D. (1993) Crystal structure of the repetitive segments of spectrin.

Science 262, 2027±2030.

6. Speicher, D.W., Morrow, J.S., Knowles, W.J. & Marchesi, V.T. (1982)

A structural model of human erythrocyte spectrin. Alignement of

chemical and functional domains. J. Biol. Chem. 257, 9093±9101.

7. Speicher, D.W., Werglarz, L. & DeSilva, T.M. (1992) Properties of

human red cell spectrin heterodimer (side-to-side) assembly and

identification of an essential nucleation site. J. Biol. Chem. 267,

14775±14782.

8. Cohen, C.M. & Gascard, P. (1992) Regulation and post-translational

modification of erythrocyte membrane and membrane-skeletal

proteins. Semin. Hematol. 29, 244±292.

9. Tse, W.T. & Lux, S.E. (1999) Red blood cell membrane disorders.

Br. J. Haematol. 104, 2±13.

10. Riederer, B.M., Zagon, I.S. & Goodman, S.R. (1986) Brain spectrin

(240/235) and brain spectrin (240/235E): two distinct spectrin

subtypes with different locations within mammalian neural cells.

J. Cell Biol. 102, 2088±2097.

11. Clark, M.B., Ma, Y., Bloom, M.L., Barker, J.E., Zagon, I.S., Zimmer,

W.E. & Goodman, S.R. (1994) Brain alpha erythroid spectrin:

identification, compartmentalization and beta spectrin associations.

Brain Res. 663, 223±236.

12. Corsi, D., Galluzzi, L., Crinelli, R. & Magnani, M. (1995) Ubiquitin is

conjugated to the cytoskeletal protein alpha-spectrin in mature

erythrocytes. J. Biol. Chem. 270, 8928±8935.

13. Corsi, D., Galluzzi, L., Lecomte, M.C. & Magnani, M. (1997)

Identification of alpha-spectrin domains susceptible to ubiquitination.

J. Biol. Chem. 272, 2977±2983.

14. Chau, V., Tobias, J.W., Bachmair, A., Marriott, D., Ecker, D.J., Gonda,

D.K. & Varshavsky, A. (1989) A multiubiquitin chain is confined

to specific lysine in a targeted short-lived protein. Science 243,

1576±1583.

15. Arnason, T. & Ellison, M.J. (1994) Stress resistance in Saccharomyces

cerevisiae is strongly correlated with assembly of a novel type of

multiubiquitin chain. Mol. Cell Biol. 14, 7876±7883.

16. Laney, J.D. & Hochstrasser, M. (1999) Substrate targeting in the

ubiquitin system. Cell 97, 427±430.

17. Hochstrasser, M. (1996) Protein degradation or regulation: Ub the

judge. Cell 84, 813±815.

18. Schwartz, A.L. & Ciechanover, A. (1999) The ubiquitin-proteasome

patway and pathogenesis of human diseases. Annu. Rev. Med. 50,

57±74.

19. Guan, K.L. & Dixon, J.E. (1991) Eukaryotic proteins expressed in

Escherichia coli: an improved thrombin cleavage and purification

procedure of fusion proteins with glutathione S-transferase. Anal.

Biochem. 192, 262±267.

20. Ursitti, J.A., Kotula, L., DeSilva, T.M., Curtis, P.J. & Speicher, D.W.

(1996) Mapping the human erythrocyte beta-spectrin dimer initiation

site using recombinant peptides and correlation of its phasing with

the alpha-actinin dimer site. J. Biol. Chem. 271, 6636±6644.

21. Nicolas, G., Pedroni, S., Fournier, C., Gautero, H., Craescu, G.,

Dhermy, D. & Lecomte, M.C. (1998) Spectrin self-association site:

characterization and study of beta-spectrin mutations associated with

hereditary elliptocytosis. Biochem. J. 332, 81±89.

22. Laemmli, U.K. (1970) Cleavage of structural proteins during the

assembly of the head of bacteriophage T4. Nature 227, 680±685.

23. Bradford, M.M. (1976) A rapid and sensitive method for the

quantitation of microgram quantities of protein utilizing the principle

of protein-dye binding. Anal. Biochem. 72, 248±254.

24. Winograd, E., Hume, D. & Branton, D. (1991) Phasing the

conformational unit of spectrin. Proc. Natl Acad. Sci. USA 88,

10788±10791.

25. Kotula, L., DeSilva, T.M., Speicher, D.W. & Curtis, P.J. (1993)

Functional characterization of recombinant human red cell alpha-

spectrin polypeptides containing the tetramer binding site. J. Biol.

Chem. 268, 14788±14793.

2818 L. Galluzzi et al. (Eur. J. Biochem. 267) q FEBS 2000

26. Lusitani, D.M., Qtaishat, N., LaBrake, C.C., Yu, R.N., Davis, J.,

Kelley, M.R. & Fung, L.W.-M. (1994) The first human alpha-

spectrin structural domain begins with serine. J. Biol. Chem. 269,

25955±25958.

27. Menhart, N., Mitchell, T., Lusitani, D., Topouzian, N. & Fung, L.W.-M.

(1996) Peptides with more than one 106-amino acid sequence motif

are needed to mimic the structural stability of spectrin. J. Biol. Chem.

271, 30410±30416.

28. Sahr, K.E., Laurila, P., Kotula, L., Scarpa, A.L., Coupal, E., Leto, T.L.,

Linnenbach, A.J., Winkelmann, J.C., Speicher, D.W., Marchesi, V.T.,

Curtis, P.J. & Forget, B.G. (1990) The complete cDNA and

polypeptide sequences of human erythroid alpha-spectrin. J. Biol.

Chem. 265, 4434±4443.

29. Galluzzi, L., Paiardini, M., Magnani, M., Nicolas, G., Lecomte, M.C.,

Harper, S. & Speicher, D.W. (1999) cDNA sequence of the human

erythroid alpha-spectrin: identification of a base deletion in the

sequence database. Blood 93, 2421±2422.

30. Beal, R.E., Toscano-Cantaffa, D., Young, P., Rechsteiner, M. & Pickart,

C.M. (1998) The hydrophobic effect contributes to polyubiquitin

chain recognition. Biochemistry 37, 2925±2934.

31. Bjùrkùy, A., Mikkelsen, A. & Elgsaeter, A. (1999) Electric

birefringence of recombinant spectrin segments 14, 14±15, 14±16,

and 14±17 from Drosophila alpha-spectrin. Biochim. Biophys. Acta

1430, 323±340.

32. Lusitani, D., Menhart, N., Keiderling, T.A. & Fung, L.W.-M. (1998)

Ionic strength effect on the thermal unfolding of alpha-spectrin

peptides. Biochemistry 37, 16546±16554.

33. Viel, A., Gee, M.S., Tomooka, D. & Branton, D. (1998) Motifs

involved in interchain binding at the tail-end of spectrin. Biochim.

Biophys. Acta 1384, 396±404.

34. Hanspal, M. & Palek, J. (1987) Synthesis and assembly of membrane

skeletal proteins in mammalian red cell precursors. J. Cell Biol. 105,

1417±1424.

35. Woods, C.M. & Lazarides, E. (1985) Degradation of unassembled

alpha- and beta-spectrin by distinct intracellular patways: regulation

of spectrin topogenesis by beta-spectrin degradation. Cell 40,

959±969.

36. Ciechanover, A. (1994) The ubiquitin-proteasome proteolytic patway.

Cell 79, 13±21.

37. Hicke, L. & Riezman, H. (1996) Ubiquitination of a yeast plasma

membrane receptor signals its ligand-stimulated endocytosis. Cell 84,

277±287.38.

q FEBS 2000 The human a-spectrin ubiquitinated repeats (Eur. J. Biochem. 267) 2819